diff --git a/.gitignore b/.gitignore index 6fa0865df7..8a3d338b50 100644 --- a/.gitignore +++ b/.gitignore @@ -52,6 +52,7 @@ training/wordlist2dawg *.patch # ignore compilation files +build/* */.deps/* */.libs/* *.lo @@ -74,4 +75,8 @@ kernel*.bin # build dirs /build* /cppan -/win* \ No newline at end of file +/.cppan +/win* +*.dll +*.exe +*.lnk diff --git a/AUTHORS b/AUTHORS index 4252027d51..4d9c75c4aa 100644 --- a/AUTHORS +++ b/AUTHORS @@ -2,12 +2,14 @@ Ray Smith (lead developer) Ahmad Abdulkader Rika Antonova Nicholas Beato +Jeff Breidenbach Samuel Charron Phil Cheatle Simon Crouch David Eger Sheelagh Huddleston Dan Johnson +Rajesh Katikam Thomas Kielbus Dar-Shyang Lee Zongyi (Joe) Liu @@ -26,3 +28,15 @@ Joern Wanke Ping Ping Xiu Andrew Ziem Oscar Zuniga + +Community Contributors: +Zdenko Podobný (Maintainer) +Jim Regan (Maintainer) +James R Barlow +Amit Dovev +Martin Ettl +Tom Morris +Tobias Müller +Egor Pugin +Sundar M. Vaidya +Stefan Weil diff --git a/CMakeLists.txt b/CMakeLists.txt index e4bc5cbb9e..68473fb4ca 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -46,19 +46,25 @@ set(VERSION_PLAIN ${VERSION_MAJOR}.${VERSION_MINOR}) set(MINIMUM_LEPTONICA_VERSION 1.71) -if(NOT EXISTS ${PROJECT_SOURCE_DIR}/cppan) +if(NOT EXISTS ${PROJECT_SOURCE_DIR}/.cppan) if (NOT Leptonica_DIR AND NOT MSVC) find_package(PkgConfig REQUIRED) - pkg_check_modules(Leptonica REQUIRED lept) + pkg_check_modules(Leptonica REQUIRED lept>=${MINIMUM_LEPTONICA_VERSION}) else() find_package(Leptonica ${MINIMUM_LEPTONICA_VERSION} REQUIRED CONFIG) endif() else() - add_subdirectory(cppan) + if (STATIC) + set(CPPAN_BUILD_SHARED_LIBS 0) + else() + set(CPPAN_BUILD_SHARED_LIBS 1) + endif() + add_subdirectory(.cppan) endif() find_package(OpenCL QUIET) -find_package(PkgConfig) + +option(BUILD_TRAINING_TOOLS "Build training tools" ON) ############################################################################### # @@ -76,6 +82,9 @@ if (WIN32) add_definitions(-D_CRT_SECURE_NO_WARNINGS) set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /MP") + if (APPVEYOR) + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /W0") + endif() endif() set(LIB_Ws2_32 Ws2_32) @@ -201,25 +210,29 @@ set(tesseract_src ${tesseract_src} api/pdfrenderer.cpp ) -add_library (tesseract ${LIBRARY_TYPE} ${tesseract_src} ${tesseract_hdr}) +add_library (libtesseract ${LIBRARY_TYPE} ${tesseract_src} ${tesseract_hdr}) if (NOT STATIC) -target_compile_definitions (tesseract PUBLIC -DTESS_EXPORTS) +target_compile_definitions (libtesseract + PRIVATE -DTESS_EXPORTS + INTERFACE -DTESS_IMPORTS +) +set_target_properties (libtesseract PROPERTIES WINDOWS_EXPORT_ALL_SYMBOLS True) endif() -target_link_libraries (tesseract ${LIB_Ws2_32} ${LIB_pthread}) -set_target_properties (tesseract PROPERTIES VERSION ${VERSION_MAJOR}.${VERSION_MINOR_0}.${VERSION_MINOR_1}) -set_target_properties (tesseract PROPERTIES SOVERSION ${VERSION_MAJOR}.${VERSION_MINOR_0}.${VERSION_MINOR_1}) +target_link_libraries (libtesseract ${LIB_Ws2_32} ${LIB_pthread}) +set_target_properties (libtesseract PROPERTIES VERSION ${VERSION_MAJOR}.${VERSION_MINOR_0}.${VERSION_MINOR_1}) +set_target_properties (libtesseract PROPERTIES SOVERSION ${VERSION_MAJOR}.${VERSION_MINOR_0}.${VERSION_MINOR_1}) if (WIN32) -set_target_properties (tesseract PROPERTIES OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR}) -set_target_properties (tesseract PROPERTIES DEBUG_OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR}d) +set_target_properties (libtesseract PROPERTIES OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR}) +set_target_properties (libtesseract PROPERTIES DEBUG_OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR}d) endif() if (NOT CPPAN_BUILD) - target_link_libraries (tesseract ${Leptonica_LIBRARIES}) - export(TARGETS tesseract FILE ${CMAKE_BINARY_DIR}/TesseractTargets.cmake) + target_link_libraries (libtesseract ${Leptonica_LIBRARIES}) + export(TARGETS libtesseract FILE ${CMAKE_BINARY_DIR}/TesseractTargets.cmake) else() - target_link_libraries (tesseract cppan) + target_link_libraries (libtesseract pvt.cppan.demo.danbloomberg.leptonica) file(WRITE ${CMAKE_BINARY_DIR}/TesseractTargets.cmake "include(${CMAKE_BINARY_DIR}/cppan.cmake)\n") - export(TARGETS tesseract APPEND FILE ${CMAKE_BINARY_DIR}/TesseractTargets.cmake) + export(TARGETS libtesseract APPEND FILE ${CMAKE_BINARY_DIR}/TesseractTargets.cmake) endif() ######################################## @@ -231,12 +244,13 @@ set(tesseractmain_src vs2010/tesseract/resource.h vs2010/tesseract/tesseract.rc ) -add_executable (tesseractmain ${tesseractmain_src}) -target_link_libraries (tesseractmain tesseract) -set_target_properties (tesseractmain PROPERTIES OUTPUT_NAME tesseract) +add_executable (tesseract ${tesseractmain_src}) +target_link_libraries (tesseract libtesseract) ######################################## +if (BUILD_TRAINING_TOOLS) add_subdirectory(training) +endif() ############################################################################### diff --git a/ChangeLog b/ChangeLog index 492d6984c9..4836aeca73 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,25 @@ +2017-02-16 - V3.05.00 + * Made some fine tuning to the hOCR output. + * Added TSV as another optional output format. + * Fixed ABI break introduced in 3.04.00 with the AnalyseLayout() method. + * text2image tool - Enable all OpenType ligatures available in a font. This feature requires Pango 1.38 or newer. + * Training tools - Replaced asserts with tprintf() and exit(1). + * Fixed Cygwin compatibility. + * Improved multipage tiff processing. + * Improved the embedded pdf font (pdf.ttf). + * Enable selection of OCR engine mode from command line. + * Changed tesseract command line parameter '-psm' to '--psm'. + * Write output of tesseract --help, --version and --list-langs to stdout instead of stderr. + * Added new C API for orientation and script detection, removed the old one. + * Increased minimum autoconf version to 2.59. + * Removed dead code. + * Fixed many compiler warning. + * Fixed memory and resource leaks. + * Fixed some issues with the 'Cube' OCR engine. + * Fixed some openCL issues. + * Added option to build Tesseract with CMake build system. + * Implemented CPPAN support for easy Windows building. + 2016-02-17 - V3.04.01 * Added OSD renderer for psm 0. Works for single page and multi-page images. * Improve tesstrain.sh script. diff --git a/INSTALL.GIT.md b/INSTALL.GIT.md index 07acbb0b01..31277e15fb 100644 --- a/INSTALL.GIT.md +++ b/INSTALL.GIT.md @@ -24,7 +24,7 @@ So, the steps for making Tesseract are: You need to install at least English language and OSD data files to TESSDATA_PREFIX directory. You can retrieve single file with tools like [wget](https://www.gnu.org/software/wget/), [curl](https://curl.haxx.se/), [GithubDownloader](https://github.com/intezer/GithubDownloader) or browser. -All language data files can be retrieved from git repository (usefull only for packagers!): +All language data files can be retrieved from git repository (useful only for packagers!): $ git clone https://github.com/tesseract-ocr/tessdata.git tesseract-ocr.tessdata diff --git a/LICENSE b/LICENSE new file mode 100644 index 0000000000..d645695673 --- /dev/null +++ b/LICENSE @@ -0,0 +1,202 @@ + + Apache License + Version 2.0, January 2004 + http://www.apache.org/licenses/ + + TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION + + 1. Definitions. + + "License" shall mean the terms and conditions for use, reproduction, + and distribution as defined by Sections 1 through 9 of this document. + + "Licensor" shall mean the copyright owner or entity authorized by + the copyright owner that is granting the License. + + "Legal Entity" shall mean the union of the acting entity and all + other entities that control, are controlled by, or are under common + control with that entity. For the purposes of this definition, + "control" means (i) the power, direct or indirect, to cause the + direction or management of such entity, whether by contract or + otherwise, or (ii) ownership of fifty percent (50%) or more of the + outstanding shares, or (iii) beneficial ownership of such entity. + + "You" (or "Your") shall mean an individual or Legal Entity + exercising permissions granted by this License. + + "Source" form shall mean the preferred form for making modifications, + including but not limited to software source code, documentation + source, and configuration files. + + "Object" form shall mean any form resulting from mechanical + transformation or translation of a Source form, including but + not limited to compiled object code, generated documentation, + and conversions to other media types. + + "Work" shall mean the work of authorship, whether in Source or + Object form, made available under the License, as indicated by a + copyright notice that is included in or attached to the work + (an example is provided in the Appendix below). + + "Derivative Works" shall mean any work, whether in Source or Object + form, that is based on (or derived from) the Work and for which the + editorial revisions, annotations, elaborations, or other modifications + represent, as a whole, an original work of authorship. For the purposes + of this License, Derivative Works shall not include works that remain + separable from, or merely link (or bind by name) to the interfaces of, + the Work and Derivative Works thereof. + + "Contribution" shall mean any work of authorship, including + the original version of the Work and any modifications or additions + to that Work or Derivative Works thereof, that is intentionally + submitted to Licensor for inclusion in the Work by the copyright owner + or by an individual or Legal Entity authorized to submit on behalf of + the copyright owner. For the purposes of this definition, "submitted" + means any form of electronic, verbal, or written communication sent + to the Licensor or its representatives, including but not limited to + communication on electronic mailing lists, source code control systems, + and issue tracking systems that are managed by, or on behalf of, the + Licensor for the purpose of discussing and improving the Work, but + excluding communication that is conspicuously marked or otherwise + designated in writing by the copyright owner as "Not a Contribution." + + "Contributor" shall mean Licensor and any individual or Legal Entity + on behalf of whom a Contribution has been received by Licensor and + subsequently incorporated within the Work. + + 2. Grant of Copyright License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + copyright license to reproduce, prepare Derivative Works of, + publicly display, publicly perform, sublicense, and distribute the + Work and such Derivative Works in Source or Object form. + + 3. Grant of Patent License. Subject to the terms and conditions of + this License, each Contributor hereby grants to You a perpetual, + worldwide, non-exclusive, no-charge, royalty-free, irrevocable + (except as stated in this section) patent license to make, have made, + use, offer to sell, sell, import, and otherwise transfer the Work, + where such license applies only to those patent claims licensable + by such Contributor that are necessarily infringed by their + Contribution(s) alone or by combination of their Contribution(s) + with the Work to which such Contribution(s) was submitted. If You + institute patent litigation against any entity (including a + cross-claim or counterclaim in a lawsuit) alleging that the Work + or a Contribution incorporated within the Work constitutes direct + or contributory patent infringement, then any patent licenses + granted to You under this License for that Work shall terminate + as of the date such litigation is filed. + + 4. Redistribution. You may reproduce and distribute copies of the + Work or Derivative Works thereof in any medium, with or without + modifications, and in Source or Object form, provided that You + meet the following conditions: + + (a) You must give any other recipients of the Work or + Derivative Works a copy of this License; and + + (b) You must cause any modified files to carry prominent notices + stating that You changed the files; and + + (c) You must retain, in the Source form of any Derivative Works + that You distribute, all copyright, patent, trademark, and + attribution notices from the Source form of the Work, + excluding those notices that do not pertain to any part of + the Derivative Works; and + + (d) If the Work includes a "NOTICE" text file as part of its + distribution, then any Derivative Works that You distribute must + include a readable copy of the attribution notices contained + within such NOTICE file, excluding those notices that do not + pertain to any part of the Derivative Works, in at least one + of the following places: within a NOTICE text file distributed + as part of the Derivative Works; within the Source form or + documentation, if provided along with the Derivative Works; or, + within a display generated by the Derivative Works, if and + wherever such third-party notices normally appear. The contents + of the NOTICE file are for informational purposes only and + do not modify the License. You may add Your own attribution + notices within Derivative Works that You distribute, alongside + or as an addendum to the NOTICE text from the Work, provided + that such additional attribution notices cannot be construed + as modifying the License. + + You may add Your own copyright statement to Your modifications and + may provide additional or different license terms and conditions + for use, reproduction, or distribution of Your modifications, or + for any such Derivative Works as a whole, provided Your use, + reproduction, and distribution of the Work otherwise complies with + the conditions stated in this License. + + 5. Submission of Contributions. Unless You explicitly state otherwise, + any Contribution intentionally submitted for inclusion in the Work + by You to the Licensor shall be under the terms and conditions of + this License, without any additional terms or conditions. + Notwithstanding the above, nothing herein shall supersede or modify + the terms of any separate license agreement you may have executed + with Licensor regarding such Contributions. + + 6. Trademarks. This License does not grant permission to use the trade + names, trademarks, service marks, or product names of the Licensor, + except as required for reasonable and customary use in describing the + origin of the Work and reproducing the content of the NOTICE file. + + 7. Disclaimer of Warranty. Unless required by applicable law or + agreed to in writing, Licensor provides the Work (and each + Contributor provides its Contributions) on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or + implied, including, without limitation, any warranties or conditions + of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A + PARTICULAR PURPOSE. You are solely responsible for determining the + appropriateness of using or redistributing the Work and assume any + risks associated with Your exercise of permissions under this License. + + 8. Limitation of Liability. In no event and under no legal theory, + whether in tort (including negligence), contract, or otherwise, + unless required by applicable law (such as deliberate and grossly + negligent acts) or agreed to in writing, shall any Contributor be + liable to You for damages, including any direct, indirect, special, + incidental, or consequential damages of any character arising as a + result of this License or out of the use or inability to use the + Work (including but not limited to damages for loss of goodwill, + work stoppage, computer failure or malfunction, or any and all + other commercial damages or losses), even if such Contributor + has been advised of the possibility of such damages. + + 9. Accepting Warranty or Additional Liability. While redistributing + the Work or Derivative Works thereof, You may choose to offer, + and charge a fee for, acceptance of support, warranty, indemnity, + or other liability obligations and/or rights consistent with this + License. However, in accepting such obligations, You may act only + on Your own behalf and on Your sole responsibility, not on behalf + of any other Contributor, and only if You agree to indemnify, + defend, and hold each Contributor harmless for any liability + incurred by, or claims asserted against, such Contributor by reason + of your accepting any such warranty or additional liability. + + END OF TERMS AND CONDITIONS + + APPENDIX: How to apply the Apache License to your work. + + To apply the Apache License to your work, attach the following + boilerplate notice, with the fields enclosed by brackets "[]" + replaced with your own identifying information. (Don't include + the brackets!) The text should be enclosed in the appropriate + comment syntax for the file format. We also recommend that a + file or class name and description of purpose be included on the + same "printed page" as the copyright notice for easier + identification within third-party archives. + + Copyright [yyyy] [name of copyright owner] + + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. diff --git a/Makefile.am b/Makefile.am index a4aa1dd915..8e2dbcf42f 100644 --- a/Makefile.am +++ b/Makefile.am @@ -4,9 +4,12 @@ ACLOCAL_AMFLAGS = -I m4 if ENABLE_TRAINING TRAINING_SUBDIR = training training: + $(MAKE) @cd "$(top_builddir)/training" && $(MAKE) training-install: @cd "$(top_builddir)/training" && $(MAKE) install +training-uninstall: + @cd "$(top_builddir)/training" && $(MAKE) uninstall clean-local: @cd "$(top_builddir)/training" && $(MAKE) clean else diff --git a/README.md b/README.md index fd64016d78..9dbdd8d2aa 100644 --- a/README.md +++ b/README.md @@ -5,25 +5,26 @@ For the latest online version of the README.md see: https://github.com/tesseract-ocr/tesseract/blob/master/README.md -#About +# About This package contains an OCR engine - `libtesseract` and a command line program - `tesseract`. The lead developer is Ray Smith. The maintainer is Zdenko Podobny. -For a list of contributors see [AUTHORS](https://github.com/tesseract-ocr/tesseract/blob/master/AUTHORS) and github's log of [contributors](https://github.com/tesseract-ocr/tesseract/graphs/contributors). +For a list of contributors see [AUTHORS](https://github.com/tesseract-ocr/tesseract/blob/master/AUTHORS) +and GitHub's log of [contributors](https://github.com/tesseract-ocr/tesseract/graphs/contributors). Tesseract has unicode (UTF-8) support, and can recognize more than 100 languages "out of the box". It can be trained to recognize other languages. See [Tesseract Training](https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract) for more information. Tesseract supports various output formats: plain-text, hocr(html), pdf. -This project does not include a GUI application. If you need one, please see the [3rdParty](https://github.com/tesseract-ocr/tesseract/wiki/3rdParty) wiki page. +This project does not include a GUI application. If you need one, please see the [3rdParty](https://github.com/tesseract-ocr/tesseract/wiki/User-Projects-%E2%80%93-3rdParty) wiki page. You should note that in many cases, in order to get better OCR results, you'll need to [improve the quality](https://github.com/tesseract-ocr/tesseract/wiki/ImproveQuality) of the image you are giving Tesseract. -The latest stable version is 3.04.01, released in February 2016. +The latest stable version is 3.05.00, released in February 2017. -#Brief history +# Brief history Tesseract was originally developed at Hewlett-Packard Laboratories Bristol and at Hewlett-Packard Co, Greeley Colorado between 1985 and 1994, with some @@ -33,13 +34,13 @@ In 2005 Tesseract was open sourced by HP. Since 2006 it is developed by Google. [Release Notes](https://github.com/tesseract-ocr/tesseract/wiki/ReleaseNotes) -#For developers +# For developers Developers can use `libtesseract` [C](https://github.com/tesseract-ocr/tesseract/blob/master/api/capi.h) or [C++](https://github.com/tesseract-ocr/tesseract/blob/master/api/baseapi.h) API to build their own application. If you need bindings to `libtesseract` for other programming languages, please see the [wrapper](https://github.com/tesseract-ocr/tesseract/wiki/AddOns#tesseract-wrappers) section on AddOns wiki page. Documentation of Tesseract generated from source code by doxygen can be found on [tesseract-ocr.github.io](http://tesseract-ocr.github.io/). -#License +# License The code in this repository is licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. @@ -55,19 +56,19 @@ Documentation of Tesseract generated from source code by doxygen can be found on **NOTE**: This software depends on other packages that may be licensed under different open source licenses. -#Installing Tesseract +# Installing Tesseract You can either [Install Tesseract via pre-built binary package](https://github.com/tesseract-ocr/tesseract/wiki) or [build it from source](https://github.com/tesseract-ocr/tesseract/wiki/Compiling). -#Running Tesseract +# Running Tesseract Basic command line usage: - tesseract imagename outputbase [-l lang] [-psm pagesegmode] [configfiles...] + tesseract imagename outputbase [-l lang] [--psm pagesegmode] [configfiles...] For more information about the various command line options use `tesseract --help` or `man tesseract`. -#Support +# Support Mailing-lists: * [tesseract-ocr](https://groups.google.com/d/forum/tesseract-ocr) - For tesseract users. diff --git a/android/jni/Android.mk b/android/jni/Android.mk index 4624801469..fd0e5112f0 100644 --- a/android/jni/Android.mk +++ b/android/jni/Android.mk @@ -4,7 +4,7 @@ include $(CLEAR_VARS) LOCAL_MODULE := tesseract-$(APP_ABI) LOCAL_STATIC_LIBRARIES := \ - mobile_base \ + base \ leptonica-$(APP_ABI) LOCAL_C_INCLUDES := $(APP_C_INCLUDES) @@ -51,7 +51,6 @@ LOCAL_CFLAGS := -DANDROID_BUILD -DNO_CUBE_BUILD -DGRAPHICS_DISABLED include $(BUILD_SHARED_LIBRARY) -$(call import-module,mobile/base) -$(call import-module,mobile/base) +$(call import-module,base/port) $(call import-module,mobile/util/hash) $(call import-module,third_party/leptonica/android/jni) diff --git a/api/Makefile.am b/api/Makefile.am index 9d20919b2e..df4bcd6368 100644 --- a/api/Makefile.am +++ b/api/Makefile.am @@ -81,15 +81,10 @@ tesseract_LDADD = libtesseract.la tesseract_LDFLAGS = $(OPENCL_LDFLAGS) -if OPENMP -tesseract_LDADD += $(OPENMP_CFLAGS) -endif - if T_WIN -tesseract_LDADD += -lws2_32 +tesseract_LDADD += -lws2_32 -ltiff libtesseract_la_LDFLAGS += -no-undefined -Wl,--as-needed -lws2_32 endif if ADD_RT tesseract_LDADD += -lrt endif - diff --git a/api/baseapi.cpp b/api/baseapi.cpp index a0689978e5..d552b8d1c3 100644 --- a/api/baseapi.cpp +++ b/api/baseapi.cpp @@ -34,8 +34,6 @@ // workaround for stdlib.h with -std=c++11 for _splitpath and _MAX_FNAME #undef __STRICT_ANSI__ #endif // _MSC_VER -#include -#include #include #include #else @@ -121,7 +119,6 @@ TessBaseAPI::TessBaseAPI() block_list_(NULL), page_res_(NULL), input_file_(NULL), - input_image_(NULL), output_file_(NULL), datapath_(NULL), language_(NULL), @@ -130,6 +127,7 @@ TessBaseAPI::TessBaseAPI() truth_cb_(NULL), rect_left_(0), rect_top_(0), rect_width_(0), rect_height_(0), image_width_(0), image_height_(0) { + unknown_title_ = ""; } TessBaseAPI::~TessBaseAPI() { @@ -515,9 +513,7 @@ void TessBaseAPI::ClearAdaptiveClassifier() { /** * Provide an image for Tesseract to recognize. Format is as - * TesseractRect above. Does not copy the image buffer, or take - * ownership. The source image may be destroyed after Recognize is called, - * either explicitly or implicitly via one of the Get*Text functions. + * TesseractRect above. Copies the image buffer and converts to Pix. * SetImage clears all recognition results, and sets the rectangle to the * full image, so it may be followed immediately by a GetUTF8Text, and it * will automatically perform recognition. @@ -525,9 +521,11 @@ void TessBaseAPI::ClearAdaptiveClassifier() { void TessBaseAPI::SetImage(const unsigned char* imagedata, int width, int height, int bytes_per_pixel, int bytes_per_line) { - if (InternalSetImage()) + if (InternalSetImage()) { thresholder_->SetImage(imagedata, width, height, bytes_per_pixel, bytes_per_line); + SetInputImage(thresholder_->GetPixRect()); + } } void TessBaseAPI::SetSourceResolution(int ppi) { @@ -539,18 +537,17 @@ void TessBaseAPI::SetSourceResolution(int ppi) { /** * Provide an image for Tesseract to recognize. As with SetImage above, - * Tesseract doesn't take a copy or ownership or pixDestroy the image, so - * it must persist until after Recognize. + * Tesseract takes its own copy of the image, so it need not persist until + * after Recognize. * Pix vs raw, which to use? - * Use Pix where possible. A future version of Tesseract may choose to use Pix - * as its internal representation and discard IMAGE altogether. - * Because of that, an implementation that sources and targets Pix may end up - * with less copies than an implementation that does not. + * Use Pix where possible. Tesseract uses Pix as its internal representation + * and it is therefore more efficient to provide a Pix directly. */ void TessBaseAPI::SetImage(Pix* pix) { - if (InternalSetImage()) + if (InternalSetImage()) { thresholder_->SetImage(pix); - SetInputImage(pix); + SetInputImage(thresholder_->GetPixRect()); + } } /** @@ -693,8 +690,8 @@ Boxa* TessBaseAPI::GetComponentImages(PageIteratorLevel level, if (pixa != NULL) { Pix* pix = NULL; if (raw_image) { - pix = page_it->GetImage(level, raw_padding, input_image_, - &left, &top); + pix = page_it->GetImage(level, raw_padding, GetInputImage(), &left, + &top); } else { pix = page_it->GetBinaryImage(level); } @@ -809,9 +806,7 @@ int CubeAPITest(Boxa* boxa_blocks, Pixa* pixa_blocks, * has not been subjected to a call of Init, SetImage, Recognize, Clear, End * DetectOS, or anything else that changes the internal PAGE_RES. */ -PageIterator* TessBaseAPI::AnalyseLayout() { - return AnalyseLayout(false); -} +PageIterator* TessBaseAPI::AnalyseLayout() { return AnalyseLayout(false); } PageIterator* TessBaseAPI::AnalyseLayout(bool merge_similar_words) { if (FindLines() == 0) { @@ -836,8 +831,7 @@ int TessBaseAPI::Recognize(ETEXT_DESC* monitor) { return -1; if (FindLines() != 0) return -1; - if (page_res_ != NULL) - delete page_res_; + delete page_res_; if (block_list_->empty()) { page_res_ = new PAGE_RES(false, block_list_, &tesseract_->prev_word_best_choice_); @@ -940,17 +934,10 @@ int TessBaseAPI::RecognizeForChopTest(ETEXT_DESC* monitor) { return 0; } -void TessBaseAPI::SetInputImage(Pix *pix) { - if (input_image_) - pixDestroy(&input_image_); - input_image_ = NULL; - if (pix) - input_image_ = pixCopy(NULL, pix); -} +// Takes ownership of the input pix. +void TessBaseAPI::SetInputImage(Pix* pix) { tesseract_->set_pix_original(pix); } -Pix* TessBaseAPI::GetInputImage() { - return input_image_; -} +Pix* TessBaseAPI::GetInputImage() { return tesseract_->pix_original(); } const char * TessBaseAPI::GetInputName() { if (input_file_) @@ -994,8 +981,7 @@ bool TessBaseAPI::ProcessPagesFileList(FILE *flist, } // Begin producing output - const char* kUnknownTitle = ""; - if (renderer && !renderer->BeginDocument(kUnknownTitle)) { + if (renderer && !renderer->BeginDocument(unknown_title_)) { return false; } @@ -1042,6 +1028,7 @@ bool TessBaseAPI::ProcessPagesMultipageTiff(const l_uint8 *data, OpenclDevice od; #endif // USE_OPENCL int page = (tessedit_page_number >= 0) ? tessedit_page_number : 0; + size_t offset = 0; for (; ; ++page) { if (tessedit_page_number >= 0) page = tessedit_page_number; @@ -1052,9 +1039,8 @@ bool TessBaseAPI::ProcessPagesMultipageTiff(const l_uint8 *data, od.pixReadTiffCl(filename, page); } else { #endif // USE_OPENCL - pix = (data) ? - pixReadMemTiff(data, size, page) : - pixReadTiff(filename, page); + pix = (data) ? pixReadMemFromMultipageTiff(data, size, &offset) + : pixReadFromMultipageTiff(filename, &offset); #ifdef USE_OPENCL } #endif // USE_OPENCL @@ -1068,6 +1054,7 @@ bool TessBaseAPI::ProcessPagesMultipageTiff(const l_uint8 *data, pixDestroy(&pix); if (!r) return false; if (tessedit_page_number >= 0) break; + if (!offset) break; } return true; #else @@ -1107,7 +1094,6 @@ bool TessBaseAPI::ProcessPagesInternal(const char* filename, const char* retry_config, int timeout_millisec, TessResultRenderer* renderer) { -#ifndef ANDROID_BUILD PERF_COUNT_START("ProcessPages") bool stdInput = !strcmp(filename, "stdin") || !strcmp(filename, "-"); if (stdInput) { @@ -1142,7 +1128,15 @@ bool TessBaseAPI::ProcessPagesInternal(const char* filename, // Maybe we have a filelist if (r != 0 || format == IFF_UNKNOWN) { - STRING s(buf.c_str()); + STRING s; + if (stdInput) { + s = buf.c_str(); + } else { + std::ifstream t(filename); + std::string u((std::istreambuf_iterator(t)), + std::istreambuf_iterator()); + s = u.c_str(); + } return ProcessPagesFileList(NULL, &s, retry_config, timeout_millisec, renderer, tesseract_->tessedit_page_number); @@ -1164,8 +1158,7 @@ bool TessBaseAPI::ProcessPagesInternal(const char* filename, } // Begin the output - const char* kUnknownTitle = ""; - if (renderer && !renderer->BeginDocument(kUnknownTitle)) { + if (renderer && !renderer->BeginDocument(unknown_title_)) { pixDestroy(&pix); return false; } @@ -1187,9 +1180,6 @@ bool TessBaseAPI::ProcessPagesInternal(const char* filename, } PERF_COUNT_END return true; -#else - return false; -#endif } bool TessBaseAPI::ProcessPage(Pix* pix, int page_index, const char* filename, @@ -1379,8 +1369,9 @@ static void AddBaselineCoordsTohOCR(const PageIterator *it, hocr_str->add_str_double(" ", round(p0 * 1000.0) / 1000.0); } -static void AddIdTohOCR(STRING* hocr_str, const std::string base, int num1, int num2) { - const unsigned long BUFSIZE = 64; +static void AddIdTohOCR(STRING* hocr_str, const std::string base, int num1, + int num2) { + const size_t BUFSIZE = 64; char id_buffer[BUFSIZE]; if (num2 >= 0) { snprintf(id_buffer, BUFSIZE - 1, "%s_%d_%d", base.c_str(), num1, num2); @@ -1393,8 +1384,7 @@ static void AddIdTohOCR(STRING* hocr_str, const std::string base, int num1, int *hocr_str += "'"; } -static void AddBoxTohOCR(const ResultIterator *it, - PageIteratorLevel level, +static void AddBoxTohOCR(const ResultIterator* it, PageIteratorLevel level, STRING* hocr_str) { int left, top, right, bottom; it->BoundingBox(level, &left, &top, &right, &bottom); @@ -1410,7 +1400,7 @@ static void AddBoxTohOCR(const ResultIterator *it, // add custom height measures float row_height, descenders, ascenders; // row attributes it->RowAttributes(&row_height, &descenders, &ascenders); - // TODO: Do we want to limit these to a single decimal place? + // TODO(rays): Do we want to limit these to a single decimal place? hocr_str->add_str_double("; x_size ", row_height); hocr_str->add_str_double("; x_descenders ", descenders * -1); hocr_str->add_str_double("; x_ascenders ", ascenders); @@ -1418,9 +1408,8 @@ static void AddBoxTohOCR(const ResultIterator *it, *hocr_str += "\">"; } -static void AddBoxToTSV(const PageIterator *it, - PageIteratorLevel level, - STRING* hocr_str) { +static void AddBoxToTSV(const PageIterator* it, PageIteratorLevel level, + STRING* hocr_str) { int left, top, right, bottom; it->BoundingBox(level, &left, &top, &right, &bottom); hocr_str->add_str_int("\t", left); @@ -1429,8 +1418,6 @@ static void AddBoxToTSV(const PageIterator *it, hocr_str->add_str_int("\t", bottom - top); } - - /** * Make a HTML-formatted string with hOCR markup from the internal * data structures. @@ -1440,7 +1427,7 @@ static void AddBoxToTSV(const PageIterator *it, * STL removed from original patch submission and refactored by rays. */ char* TessBaseAPI::GetHOCRText(int page_number) { - return GetHOCRText(NULL,page_number); + return GetHOCRText(NULL, page_number); } /** @@ -1452,13 +1439,12 @@ char* TessBaseAPI::GetHOCRText(int page_number) { * STL removed from original patch submission and refactored by rays. */ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { - if (tesseract_ == NULL || - (page_res_ == NULL && Recognize(monitor) < 0)) + if (tesseract_ == NULL || (page_res_ == NULL && Recognize(monitor) < 0)) return NULL; int lcnt = 1, bcnt = 1, pcnt = 1, wcnt = 1; int page_id = page_number + 1; // hOCR uses 1-based page numbers. - bool para_is_ltr = true; // Default direction is LTR + bool para_is_ltr = true; // Default direction is LTR const char* paragraph_lang = NULL; bool font_info = false; GetBoolVariable("hocr_font_info", &font_info); @@ -1470,13 +1456,13 @@ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { #ifdef _WIN32 // convert input name from ANSI encoding to utf-8 - int str16_len = MultiByteToWideChar(CP_ACP, 0, input_file_->string(), -1, - NULL, 0); + int str16_len = + MultiByteToWideChar(CP_ACP, 0, input_file_->string(), -1, NULL, 0); wchar_t *uni16_str = new WCHAR[str16_len]; str16_len = MultiByteToWideChar(CP_ACP, 0, input_file_->string(), -1, uni16_str, str16_len); - int utf8_len = WideCharToMultiByte(CP_UTF8, 0, uni16_str, str16_len, NULL, - 0, NULL, NULL); + int utf8_len = WideCharToMultiByte(CP_UTF8, 0, uni16_str, str16_len, NULL, 0, + NULL, NULL); char *utf8_str = new char[utf8_len]; WideCharToMultiByte(CP_UTF8, 0, uni16_str, str16_len, utf8_str, utf8_len, NULL, NULL); @@ -1509,7 +1495,7 @@ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { // Open any new block/paragraph/textline. if (res_it->IsAtBeginningOf(RIL_BLOCK)) { - para_is_ltr = true; // reset to default direction + para_is_ltr = true; // reset to default direction hocr_str += "
WordRecognitionLanguage(); if (paragraph_lang) { - hocr_str += " lang='"; - hocr_str += paragraph_lang; - hocr_str += "'"; + hocr_str += " lang='"; + hocr_str += paragraph_lang; + hocr_str += "'"; } AddBoxTohOCR(res_it, RIL_PARA, &hocr_str); } @@ -1567,8 +1553,12 @@ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { } switch (res_it->WordDirection()) { // Only emit direction if different from current paragraph direction - case DIR_LEFT_TO_RIGHT: if (!para_is_ltr) hocr_str += " dir='ltr'"; break; - case DIR_RIGHT_TO_LEFT: if (para_is_ltr) hocr_str += " dir='rtl'"; break; + case DIR_LEFT_TO_RIGHT: + if (!para_is_ltr) hocr_str += " dir='ltr'"; + break; + case DIR_RIGHT_TO_LEFT: + if (para_is_ltr) hocr_str += " dir='rtl'"; + break; case DIR_MIX: case DIR_NEUTRAL: default: // Do nothing. @@ -1600,7 +1590,7 @@ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { if (last_word_in_para) { hocr_str += "\n

\n"; pcnt++; - para_is_ltr = true; // back to default direction + para_is_ltr = true; // back to default direction } if (last_word_in_block) { hocr_str += "
\n"; @@ -1620,8 +1610,7 @@ char* TessBaseAPI::GetHOCRText(ETEXT_DESC* monitor, int page_number) { * page_number is 0-based but will appear in the output as 1-based. */ char* TessBaseAPI::GetTSVText(int page_number) { - if (tesseract_ == NULL || - (page_res_ == NULL && Recognize(NULL) < 0)) + if (tesseract_ == NULL || (page_res_ == NULL && Recognize(NULL) < 0)) return NULL; int lcnt = 1, bcnt = 1, pcnt = 1, wcnt = 1; @@ -1629,9 +1618,10 @@ char* TessBaseAPI::GetTSVText(int page_number) { STRING tsv_str(""); - int page_num = page_id, block_num = 0, par_num = 0, line_num = 0, word_num = 0; + int page_num = page_id, block_num = 0, par_num = 0, line_num = 0, + word_num = 0; - tsv_str.add_str_int("1\t", page_num); // level 1 - page + tsv_str.add_str_int("1\t", page_num); // level 1 - page tsv_str.add_str_int("\t", block_num); tsv_str.add_str_int("\t", par_num); tsv_str.add_str_int("\t", line_num); @@ -1642,7 +1632,7 @@ char* TessBaseAPI::GetTSVText(int page_number) { tsv_str.add_str_int("\t", rect_height_); tsv_str += "\t-1\t\n"; - ResultIterator *res_it = GetIterator(); + ResultIterator* res_it = GetIterator(); while (!res_it->Empty(RIL_BLOCK)) { if (res_it->Empty(RIL_WORD)) { res_it->Next(RIL_WORD); @@ -1652,46 +1642,46 @@ char* TessBaseAPI::GetTSVText(int page_number) { // Add rows for any new block/paragraph/textline. if (res_it->IsAtBeginningOf(RIL_BLOCK)) { block_num++, par_num = 0, line_num = 0, word_num = 0; - tsv_str.add_str_int("2\t", page_num); // level 2 - block + tsv_str.add_str_int("2\t", page_num); // level 2 - block tsv_str.add_str_int("\t", block_num); tsv_str.add_str_int("\t", par_num); tsv_str.add_str_int("\t", line_num); tsv_str.add_str_int("\t", word_num); AddBoxToTSV(res_it, RIL_BLOCK, &tsv_str); - tsv_str += "\t-1\t\n"; // end of row for block + tsv_str += "\t-1\t\n"; // end of row for block } if (res_it->IsAtBeginningOf(RIL_PARA)) { par_num++, line_num = 0, word_num = 0; - tsv_str.add_str_int("3\t", page_num); // level 3 - paragraph + tsv_str.add_str_int("3\t", page_num); // level 3 - paragraph tsv_str.add_str_int("\t", block_num); tsv_str.add_str_int("\t", par_num); tsv_str.add_str_int("\t", line_num); tsv_str.add_str_int("\t", word_num); AddBoxToTSV(res_it, RIL_PARA, &tsv_str); - tsv_str += "\t-1\t\n"; // end of row for para + tsv_str += "\t-1\t\n"; // end of row for para } if (res_it->IsAtBeginningOf(RIL_TEXTLINE)) { line_num++, word_num = 0; - tsv_str.add_str_int("4\t", page_num); // level 4 - line + tsv_str.add_str_int("4\t", page_num); // level 4 - line tsv_str.add_str_int("\t", block_num); tsv_str.add_str_int("\t", par_num); tsv_str.add_str_int("\t", line_num); tsv_str.add_str_int("\t", word_num); AddBoxToTSV(res_it, RIL_TEXTLINE, &tsv_str); - tsv_str += "\t-1\t\n"; // end of row for line + tsv_str += "\t-1\t\n"; // end of row for line } // Now, process the word... int left, top, right, bottom; bool bold, italic, underlined, monospace, serif, smallcaps; int pointsize, font_id; - const char *font_name; + const char* font_name; res_it->BoundingBox(RIL_WORD, &left, &top, &right, &bottom); - font_name = res_it->WordFontAttributes(&bold, &italic, &underlined, - &monospace, &serif, &smallcaps, - &pointsize, &font_id); + font_name = + res_it->WordFontAttributes(&bold, &italic, &underlined, &monospace, + &serif, &smallcaps, &pointsize, &font_id); word_num++; - tsv_str.add_str_int("5\t", page_num); // level 5 - word + tsv_str.add_str_int("5\t", page_num); // level 5 - word tsv_str.add_str_int("\t", block_num); tsv_str.add_str_int("\t", par_num); tsv_str.add_str_int("\t", line_num); @@ -1712,11 +1702,11 @@ char* TessBaseAPI::GetTSVText(int page_number) { tsv_str += res_it->GetUTF8Text(RIL_SYMBOL); res_it->Next(RIL_SYMBOL); } while (!res_it->Empty(RIL_BLOCK) && !res_it->IsAtBeginningOf(RIL_WORD)); - tsv_str += "\n"; // end of row + tsv_str += "\n"; // end of row wcnt++; } - char *ret = new char[tsv_str.length() + 1]; + char* ret = new char[tsv_str.length() + 1]; strcpy(ret, tsv_str.string()); delete res_it; return ret; @@ -1760,7 +1750,7 @@ char* TessBaseAPI::GetBoxText(int page_number) { int total_length = blob_count * kBytesPerBoxFileLine + utf8_length + kMaxBytesPerLine; char* result = new char[total_length]; - strcpy(result, "\0"); + result[0] = '\0'; int output_length = 0; LTRResultIterator* it = GetLTRIterator(); do { @@ -1907,43 +1897,70 @@ char* TessBaseAPI::GetUNLVText() { return result; } - /** - * The recognized text is returned as a char* which is coded - * as UTF8 and must be freed with the delete [] operator. - * page_number is a 0-based page index that will appear in the osd file. - */ -char* TessBaseAPI::GetOsdText(int page_number) { +/** + * Detect the orientation of the input image and apparent script (alphabet). + * orient_deg is the detected clockwise rotation of the input image in degrees (0, 90, 180, 270) + * orient_conf is the confidence (15.0 is reasonably confident) + * script_name is an ASCII string, the name of the script, e.g. "Latin" + * script_conf is confidence level in the script + * Returns true on success and writes values to each parameter as an output + */ +bool TessBaseAPI::DetectOrientationScript(int* orient_deg, float* orient_conf, const char** script_name, float* script_conf) { OSResults osr; bool osd = DetectOS(&osr); if (!osd) { - return NULL; + return false; } int orient_id = osr.best_result.orientation_id; int script_id = osr.get_best_script(orient_id); - float orient_conf = osr.best_result.oconfidence; - float script_conf = osr.best_result.sconfidence; - const char* script_name = + if (orient_conf) + *orient_conf = osr.best_result.oconfidence; + if (orient_deg) + *orient_deg = orient_id * 90; // convert quadrant to degrees + + if (script_name) { + const char* script = osr.unicharset->get_script_from_script_id(script_id); - // clockwise orientation of the input image, in degrees - int orient_deg = orient_id * 90; + *script_name = script; + } + + if (script_conf) + *script_conf = osr.best_result.sconfidence; + + return true; +} + +/** + * The recognized text is returned as a char* which is coded + * as UTF8 and must be freed with the delete [] operator. + * page_number is a 0-based page index that will appear in the osd file. + */ +char* TessBaseAPI::GetOsdText(int page_number) { + int orient_deg; + float orient_conf; + const char* script_name; + float script_conf; + + if (!DetectOrientationScript(&orient_deg, &orient_conf, &script_name, &script_conf)) + return NULL; // clockwise rotation needed to make the page upright - int rotate = OrientationIdToValue(orient_id); - - char* osd_buf = new char[255]; - snprintf(osd_buf, 255, - "Page number: %d\n" - "Orientation in degrees: %d\n" - "Rotate: %d\n" - "Orientation confidence: %.2f\n" - "Script: %s\n" - "Script confidence: %.2f\n", - page_number, - orient_deg, rotate, orient_conf, - script_name, script_conf); + int rotate = OrientationIdToValue(orient_deg / 90); + + const int kOsdBufsize = 255; + char* osd_buf = new char[kOsdBufsize]; + snprintf(osd_buf, kOsdBufsize, + "Page number: %d\n" + "Orientation in degrees: %d\n" + "Rotate: %d\n" + "Orientation confidence: %.2f\n" + "Script: %s\n" + "Script confidence: %.2f\n", + page_number, orient_deg, rotate, orient_conf, script_name, + script_conf); return osd_buf; } @@ -2063,7 +2080,7 @@ void TessBaseAPI::Clear() { if (thresholder_ != NULL) thresholder_->Clear(); ClearResults(); - SetInputImage(NULL); + if (tesseract_ != NULL) SetInputImage(NULL); } /** @@ -2109,10 +2126,6 @@ void TessBaseAPI::End() { delete input_file_; input_file_ = NULL; } - if (input_image_ != NULL) { - pixDestroy(&input_image_); - input_image_ = NULL; - } if (output_file_ != NULL) { delete output_file_; output_file_ = NULL; @@ -2243,6 +2256,8 @@ void TessBaseAPI::Threshold(Pix** pix) { if (y_res < kMinCredibleResolution || y_res > kMaxCredibleResolution) { // Use the minimum default resolution, as it is safer to under-estimate // than over-estimate resolution. + tprintf("Warning. Invalid resolution %d dpi. Using %d instead.\n", + y_res, kMinCredibleResolution); thresholder_->SetSourceYResolution(kMinCredibleResolution); } PageSegMode pageseg_mode = @@ -2767,7 +2782,7 @@ void TessBaseAPI::GetFeaturesForBlob(TBLOB* blob, INT_FX_RESULT_STRUCT fx_info; tesseract_->ExtractFeatures(*blob, false, &bl_features, &cn_features, &fx_info, &outline_counts); - if (cn_features.size() == 0 || cn_features.size() > MAX_NUM_INT_FEATURES) { + if (cn_features.empty() || cn_features.size() > MAX_NUM_INT_FEATURES) { *num_features = 0; return; // Feature extraction failed. } diff --git a/api/baseapi.h b/api/baseapi.h index 3b0d3f67ce..ab51ef20d6 100644 --- a/api/baseapi.h +++ b/api/baseapi.h @@ -20,7 +20,7 @@ #ifndef TESSERACT_API_BASEAPI_H__ #define TESSERACT_API_BASEAPI_H__ -#define TESSERACT_VERSION_STR "3.05.00dev" +#define TESSERACT_VERSION_STR "3.05.00" #define TESSERACT_VERSION 0x030500 #define MAKE_VERSION(major, minor, patch) (((major) << 16) | ((minor) << 8) | \ (patch)) @@ -142,6 +142,7 @@ class TESS_API TessBaseAPI { * is stored in the PDF so we need that as well. */ const char* GetInputName(); + // Takes ownership of the input pix. void SetInputImage(Pix *pix); Pix* GetInputImage(); int GetSourceYResolution(); @@ -333,9 +334,7 @@ class TESS_API TessBaseAPI { /** * Provide an image for Tesseract to recognize. Format is as - * TesseractRect above. Does not copy the image buffer, or take - * ownership. The source image may be destroyed after Recognize is called, - * either explicitly or implicitly via one of the Get*Text functions. + * TesseractRect above. Copies the image buffer and converts to Pix. * SetImage clears all recognition results, and sets the rectangle to the * full image, so it may be followed immediately by a GetUTF8Text, and it * will automatically perform recognition. @@ -345,13 +344,11 @@ class TESS_API TessBaseAPI { /** * Provide an image for Tesseract to recognize. As with SetImage above, - * Tesseract doesn't take a copy or ownership or pixDestroy the image, so - * it must persist until after Recognize. + * Tesseract takes its own copy of the image, so it need not persist until + * after Recognize. * Pix vs raw, which to use? - * Use Pix where possible. A future version of Tesseract may choose to use Pix - * as its internal representation and discard IMAGE altogether. - * Because of that, an implementation that sources and targets Pix may end up - * with less copies than an implementation that does not. + * Use Pix where possible. Tesseract uses Pix as its internal representation + * and it is therefore more efficient to provide a Pix directly. */ void SetImage(Pix* pix); @@ -376,8 +373,7 @@ class TESS_API TessBaseAPI { * delete it when it it is replaced or the API is destructed. */ void SetThresholder(ImageThresholder* thresholder) { - if (thresholder_ != NULL) - delete thresholder_; + delete thresholder_; thresholder_ = thresholder; ClearResults(); } @@ -588,8 +584,8 @@ class TESS_API TessBaseAPI { * data structures. * page_number is 0-based but will appear in the output as 1-based. * monitor can be used to - * cancel the recognition - * receive progress callbacks + * cancel the recognition + * receive progress callbacks */ char* GetHOCRText(ETEXT_DESC* monitor, int page_number); @@ -622,6 +618,16 @@ class TESS_API TessBaseAPI { */ char* GetUNLVText(); + /** + * Detect the orientation of the input image and apparent script (alphabet). + * orient_deg is the detected clockwise rotation of the input image in degrees (0, 90, 180, 270) + * orient_conf is the confidence (15.0 is reasonably confident) + * script_name is an ASCII string, the name of the script, e.g. "Latin" + * script_conf is confidence level in the script + * Returns true on success and writes values to each parameter as an output + */ + bool DetectOrientationScript(int* orient_deg, float* orient_conf, const char** script_name, float* script_conf); + /** * The recognized text is returned as a char* which is coded * as UTF8 and must be freed with the delete [] operator. @@ -750,13 +756,9 @@ class TESS_API TessBaseAPI { */ static void NormalizeTBLOB(TBLOB *tblob, ROW *row, bool numeric_mode); - Tesseract* tesseract() const { - return tesseract_; - } + Tesseract* tesseract() const { return tesseract_; } - OcrEngineMode oem() const { - return last_oem_requested_; - } + OcrEngineMode oem() const { return last_oem_requested_; } void InitTruthCallback(TruthCallback *cb) { truth_cb_ = cb; } @@ -855,9 +857,7 @@ class TESS_API TessBaseAPI { int** y1, PAGE_RES* page_res); - TESS_LOCAL const PAGE_RES* GetPageRes() const { - return page_res_; - }; + TESS_LOCAL const PAGE_RES* GetPageRes() const { return page_res_; } /* @} */ @@ -870,7 +870,6 @@ class TESS_API TessBaseAPI { BLOCK_LIST* block_list_; ///< The page layout. PAGE_RES* page_res_; ///< The page-level data. STRING* input_file_; ///< Name used by training code. - Pix* input_image_; ///< Image used for searchable PDF STRING* output_file_; ///< Name used by debug code. STRING* datapath_; ///< Current location of tessdata. STRING* language_; ///< Last initialized language. @@ -898,7 +897,7 @@ class TESS_API TessBaseAPI { const char* retry_config, int timeout_millisec, TessResultRenderer* renderer, int tessedit_page_number); - // TIFF supports multipage so gets special consideration + // TIFF supports multipage so gets special consideration. bool ProcessPagesMultipageTiff(const unsigned char *data, size_t size, const char* filename, @@ -906,6 +905,12 @@ class TESS_API TessBaseAPI { int timeout_millisec, TessResultRenderer* renderer, int tessedit_page_number); + // There's currently no way to pass a document title from the + // Tesseract command line, and we have multiple places that choose + // to set the title to an empty string. Using a single named + // variable will hopefully reduce confusion if the situation changes + // in the future. + const char *unknown_title_; }; // class TessBaseAPI. /** Escape a char string - remove &<>"' with HTML codes. */ diff --git a/api/capi.cpp b/api/capi.cpp index 849d296104..57bed872df 100644 --- a/api/capi.cpp +++ b/api/capi.cpp @@ -538,9 +538,18 @@ TESS_API void TESS_CALL TessBaseAPISetProbabilityInContextFunc(TessBaseAPI* hand TESS_API BOOL TESS_CALL TessBaseAPIDetectOS(TessBaseAPI* handle, OSResults* results) { - return handle->DetectOS(results) ? TRUE : FALSE; + return FALSE; // Unsafe ABI, return FALSE always } +TESS_API BOOL TESS_CALL TessBaseAPIDetectOrientationScript(TessBaseAPI* handle, + int* orient_deg, float* orient_conf, const char** script_name, float* script_conf) +{ + bool success; + success = handle->DetectOrientationScript(orient_deg, orient_conf, script_name, script_conf); + return (BOOL)success; +} + + TESS_API void TESS_CALL TessBaseAPIGetFeaturesForBlob(TessBaseAPI* handle, TBLOB* blob, INT_FEATURE_STRUCT* int_features, int* num_features, int* FeatureOutlineIndex) { diff --git a/api/capi.h b/api/capi.h index a0c54a20e4..93d43f9483 100644 --- a/api/capi.h +++ b/api/capi.h @@ -285,8 +285,14 @@ TESS_API void TESS_CALL TessBaseAPIClearPersistentCache(TessBaseAPI* handle); TESS_API void TESS_CALL TessBaseAPISetProbabilityInContextFunc(TessBaseAPI* handle, TessProbabilityInContextFunc f); TESS_API void TESS_CALL TessBaseAPISetFillLatticeFunc(TessBaseAPI* handle, TessFillLatticeFunc f); + +// Deprecated, no longer working TESS_API BOOL TESS_CALL TessBaseAPIDetectOS(TessBaseAPI* handle, OSResults* results); +// Call TessDeleteText(*best_script_name) to free memory allocated by this function +TESS_API BOOL TESS_CALL TessBaseAPIDetectOrientationScript(TessBaseAPI* handle, + int* orient_deg, float* orient_conf, const char **script_name, float* script_conf); + TESS_API void TESS_CALL TessBaseAPIGetFeaturesForBlob(TessBaseAPI* handle, TBLOB* blob, INT_FEATURE_STRUCT* int_features, int* num_features, int* FeatureOutlineIndex); diff --git a/api/pdfrenderer.cpp b/api/pdfrenderer.cpp index 4708300492..001c86ce41 100644 --- a/api/pdfrenderer.cpp +++ b/api/pdfrenderer.cpp @@ -20,12 +20,12 @@ #include "config_auto.h" #endif +#include "allheaders.h" #include "baseapi.h" -#include "renderer.h" #include "math.h" +#include "renderer.h" #include "strngs.h" #include "tprintf.h" -#include "allheaders.h" #ifdef _MSC_VER #include "mathfix.h" @@ -282,7 +282,7 @@ void AffineMatrix(int writing_direction, } } -// There are some really stupid PDF viewers in the wild, such as +// There are some really awkward PDF viewers in the wild, such as // 'Preview' which ships with the Mac. They do a better job with text // selection and highlighting when given perfectly flat baseline // instead of very slightly tilted. We clip small tilts to appease @@ -441,8 +441,8 @@ char* TessPDFRenderer::GetPDFTextObjects(TessBaseAPI* api, int code = unicodes[i]; // Convert to UTF-16BE https://en.wikipedia.org/wiki/UTF-16 if ((code > 0xD7FF && code < 0xE000) || code > 0x10FFFF) { - tprintf("Dropping invalid codepoint %d\n", code); - continue; + tprintf("Dropping invalid codepoint %d\n", code); + continue; } if (code < 0x10000) { snprintf(utf16, sizeof(utf16), "<%04X>", code); @@ -567,7 +567,8 @@ bool TessPDFRenderer::BeginDocumentHandler() { "<<\n" " /Length %lu /Filter /FlateDecode\n" ">>\n" - "stream\n", (unsigned long)len); + "stream\n", + (unsigned long)len); if (n >= sizeof(buf)) { lept_free(comp); return false; @@ -619,7 +620,6 @@ bool TessPDFRenderer::BeginDocumentHandler() { AppendPDFObject(buf); // FONT DESCRIPTOR - const int kCharHeight = 2; // Effect: highlights are half height n = snprintf(buf, sizeof(buf), "7 0 obj\n" "<<\n" @@ -635,10 +635,10 @@ bool TessPDFRenderer::BeginDocumentHandler() { " /Type /FontDescriptor\n" ">>\n" "endobj\n", - 1000 / kCharHeight, - 1000 / kCharHeight, + 1000, + 1000, 1000 / kCharWidth, - 1000 / kCharHeight, + 1000, 8L // Font data ); if (n >= sizeof(buf)) return false; @@ -819,10 +819,6 @@ bool TessPDFRenderer::imageToPDFObj(Pix *pix, *pdf_object_size = b1_len + colorspace_len + b2_len + cid->nbytescomp + b3_len; *pdf_object = new char[*pdf_object_size]; - if (!pdf_object) { - l_CIDataDestroy(&cid); - return false; - } char *p = *pdf_object; memcpy(p, b1, b1_len); diff --git a/api/renderer.cpp b/api/renderer.cpp index 4a88a24608..e683149381 100644 --- a/api/renderer.cpp +++ b/api/renderer.cpp @@ -155,11 +155,11 @@ TessHOcrRenderer::TessHOcrRenderer(const char *outputbase, bool font_info) bool TessHOcrRenderer::BeginDocumentHandler() { AppendString( - "\n" - "\n" - "\n \n "); + "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" + "<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.0 Transitional//EN\"\n" + " \"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd\">\n" + "<html xmlns=\"http://www.w3.org/1999/xhtml\" xml:lang=\"en\" " + "lang=\"en\">\n <head>\n <title>"); AppendString(title()); AppendString( "\n" @@ -198,25 +198,25 @@ bool TessHOcrRenderer::AddImageHandler(TessBaseAPI* api) { /********************************************************************** * TSV Text Renderer interface implementation **********************************************************************/ -TessTsvRenderer::TessTsvRenderer(const char *outputbase) +TessTsvRenderer::TessTsvRenderer(const char* outputbase) : TessResultRenderer(outputbase, "tsv") { - font_info_ = false; + font_info_ = false; } -TessTsvRenderer::TessTsvRenderer(const char *outputbase, bool font_info) +TessTsvRenderer::TessTsvRenderer(const char* outputbase, bool font_info) : TessResultRenderer(outputbase, "tsv") { - font_info_ = font_info; + font_info_ = font_info; } bool TessTsvRenderer::BeginDocumentHandler() { // Output TSV column headings - AppendString("level\tpage_num\tblock_num\tpar_num\tline_num\tword_num\tleft\ttop\twidth\theight\tconf\ttext\n"); + AppendString( + "level\tpage_num\tblock_num\tpar_num\tline_num\tword_" + "num\tleft\ttop\twidth\theight\tconf\ttext\n"); return true; } -bool TessTsvRenderer::EndDocumentHandler() { - return true; -} +bool TessTsvRenderer::EndDocumentHandler() { return true; } bool TessTsvRenderer::AddImageHandler(TessBaseAPI* api) { char* tsv = api->GetTSVText(imagenum()); @@ -266,8 +266,7 @@ bool TessBoxTextRenderer::AddImageHandler(TessBaseAPI* api) { * Osd Text Renderer interface implementation **********************************************************************/ TessOsdRenderer::TessOsdRenderer(const char* outputbase) - : TessResultRenderer(outputbase, "osd") { -} + : TessResultRenderer(outputbase, "osd") {} bool TessOsdRenderer::AddImageHandler(TessBaseAPI* api) { char* osd = api->GetOsdText(imagenum()); diff --git a/api/renderer.h b/api/renderer.h index 6b47813f7b..d868f267fa 100644 --- a/api/renderer.h +++ b/api/renderer.h @@ -77,7 +77,7 @@ class TESS_API TessResultRenderer { bool EndDocument(); const char* file_extension() const { return file_extension_; } - const char* title() const { return title_; } + const char* title() const { return title_.c_str(); } /** * Returns the index of the last image given to AddImage @@ -126,7 +126,7 @@ class TESS_API TessResultRenderer { private: const char* file_extension_; // standard extension for generated output - const char* title_; // title of document being renderered + STRING title_; // title of document being renderered int imagenum_; // index of last image added FILE* fout_; // output file pointer @@ -153,13 +153,13 @@ class TESS_API TessHOcrRenderer : public TessResultRenderer { explicit TessHOcrRenderer(const char *outputbase, bool font_info); explicit TessHOcrRenderer(const char *outputbase); -protected: + protected: virtual bool BeginDocumentHandler(); virtual bool AddImageHandler(TessBaseAPI* api); virtual bool EndDocumentHandler(); -private: - bool font_info_; // whether to print font information + private: + bool font_info_; // whether to print font information }; /** @@ -167,15 +167,15 @@ class TESS_API TessHOcrRenderer : public TessResultRenderer { */ class TESS_API TessTsvRenderer : public TessResultRenderer { public: - explicit TessTsvRenderer(const char *outputbase, bool font_info); - explicit TessTsvRenderer(const char *outputbase); + explicit TessTsvRenderer(const char* outputbase, bool font_info); + explicit TessTsvRenderer(const char* outputbase); -protected: + protected: virtual bool BeginDocumentHandler(); virtual bool AddImageHandler(TessBaseAPI* api); virtual bool EndDocumentHandler(); -private: + private: bool font_info_; // whether to print font information }; @@ -188,12 +188,12 @@ class TESS_API TessPDFRenderer : public TessResultRenderer { // we load a custom PDF font from this location. TessPDFRenderer(const char *outputbase, const char *datadir); -protected: + protected: virtual bool BeginDocumentHandler(); virtual bool AddImageHandler(TessBaseAPI* api); virtual bool EndDocumentHandler(); -private: + private: // We don't want to have every image in memory at once, // so we store some metadata as we go along producing // PDFs one page at a time. At the end that metadata is diff --git a/api/tesseractmain.cpp b/api/tesseractmain.cpp index 3fe8dc8c53..7cd4021fbe 100644 --- a/api/tesseractmain.cpp +++ b/api/tesseractmain.cpp @@ -36,32 +36,31 @@ #if defined(HAVE_TIFFIO_H) && defined(_WIN32) #include -#include static void Win32WarningHandler(const char* module, const char* fmt, va_list ap) { - if (module != NULL) { - fprintf(stderr, "%s: ", module); - } - fprintf(stderr, "Warning, "); - vfprintf(stderr, fmt, ap); - fprintf(stderr, ".\n"); + if (module != NULL) { + fprintf(stderr, "%s: ", module); + } + fprintf(stderr, "Warning, "); + vfprintf(stderr, fmt, ap); + fprintf(stderr, ".\n"); } #endif /* HAVE_TIFFIO_H && _WIN32 */ void PrintVersionInfo() { - char *versionStrP; + char* versionStrP; - printf("tesseract %s\n", tesseract::TessBaseAPI::Version()); + printf("tesseract %s\n", tesseract::TessBaseAPI::Version()); - versionStrP = getLeptonicaVersion(); - printf(" %s\n", versionStrP); - lept_free(versionStrP); + versionStrP = getLeptonicaVersion(); + printf(" %s\n", versionStrP); + lept_free(versionStrP); - versionStrP = getImagelibVersions(); - printf(" %s\n", versionStrP); - lept_free(versionStrP); + versionStrP = getImagelibVersions(); + printf(" %s\n", versionStrP); + lept_free(versionStrP); #ifdef USE_OPENCL cl_platform_id platform; @@ -82,7 +81,7 @@ void PrintVersionInfo() { printf(" Found %d devices.\n", num_devices); for (i = 0; i < num_devices; ++i) { clGetDeviceInfo(devices[i], CL_DEVICE_NAME, 256, info, 0); - printf(" Device %d name: %s.\n", i+1, info); + printf(" Device %d name: %s.\n", i + 1, info); } #endif } @@ -90,7 +89,7 @@ void PrintVersionInfo() { void PrintUsage(const char* program) { printf( "Usage:\n" - " %s --help | --help-psm | --version\n" + " %s --help | --help-psm | --help-oem | --version\n" " %s --list-langs [--tessdata-dir PATH]\n" " %s --print-parameters [options...] [configfile...]\n" " %s imagename|stdin outputbase|stdout [options...] [configfile...]\n", @@ -100,27 +99,33 @@ void PrintUsage(const char* program) { void PrintHelpForPSM() { const char* msg = "Page segmentation modes:\n" - " 0 Orientation and script detection (OSD) only.\n" - " 1 Automatic page segmentation with OSD.\n" - " 2 Automatic page segmentation, but no OSD, or OCR.\n" - " 3 Fully automatic page segmentation, but no OSD. (Default)\n" - " 4 Assume a single column of text of variable sizes.\n" - " 5 Assume a single uniform block of vertically aligned text.\n" - " 6 Assume a single uniform block of text.\n" - " 7 Treat the image as a single text line.\n" - " 8 Treat the image as a single word.\n" - " 9 Treat the image as a single word in a circle.\n" - " 10 Treat the image as a single character.\n" - - //TODO: Consider publishing these modes. - #if 0 - " 11 Sparse text. Find as much text as possible in no" - " particular order.\n" - " 12 Sparse text with OSD.\n" - " 13 Raw line. Treat the image as a single text line,\n" - "\t\t\tbypassing hacks that are Tesseract-specific.\n" - #endif - ; + " 0 Orientation and script detection (OSD) only.\n" + " 1 Automatic page segmentation with OSD.\n" + " 2 Automatic page segmentation, but no OSD, or OCR.\n" + " 3 Fully automatic page segmentation, but no OSD. (Default)\n" + " 4 Assume a single column of text of variable sizes.\n" + " 5 Assume a single uniform block of vertically aligned text.\n" + " 6 Assume a single uniform block of text.\n" + " 7 Treat the image as a single text line.\n" + " 8 Treat the image as a single word.\n" + " 9 Treat the image as a single word in a circle.\n" + " 10 Treat the image as a single character.\n" + " 11 Sparse text. Find as much text as possible in no" + " particular order.\n" + " 12 Sparse text with OSD.\n" + " 13 Raw line. Treat the image as a single text line,\n" + "\t\t\tbypassing hacks that are Tesseract-specific.\n"; + + printf("%s", msg); +} + +void PrintHelpForOEM() { + const char* msg = + "OCR Engine modes:\n" + " 0 Original Tesseract only.\n" + " 1 Cube only.\n" + " 2 Tesseract + cube.\n" + " 3 Default, based on what is available.\n"; printf("%s", msg); } @@ -136,32 +141,34 @@ void PrintHelpMessage(const char* program) { " -l LANG[+LANG] Specify language(s) used for OCR.\n" " -c VAR=VALUE Set value for config variables.\n" " Multiple -c arguments are allowed.\n" - " -psm NUM Specify page segmentation mode.\n" - "NOTE: These options must occur before any configfile.\n" - ; + " --psm NUM Specify page segmentation mode.\n" + " --oem NUM Specify OCR Engine mode.\n" + "NOTE: These options must occur before any configfile.\n"; printf("\n%s\n", ocr_options); PrintHelpForPSM(); + PrintHelpForOEM(); - const char *single_options = + const char* single_options = "Single options:\n" " -h, --help Show this help message.\n" " --help-psm Show page segmentation modes.\n" + " --help-oem Show OCR Engine modes.\n" " -v, --version Show version information.\n" " --list-langs List available languages for tesseract engine.\n" - " --print-parameters Print tesseract parameters to stdout.\n" - ; + " --print-parameters Print tesseract parameters to stdout.\n"; printf("\n%s", single_options); } -void SetVariablesFromCLArgs(tesseract::TessBaseAPI* api, int argc, char** argv) { +void SetVariablesFromCLArgs(tesseract::TessBaseAPI* api, int argc, + char** argv) { char opt1[256], opt2[255]; for (int i = 0; i < argc; i++) { if (strcmp(argv[i], "-c") == 0 && i + 1 < argc) { strncpy(opt1, argv[i + 1], 255); opt1[255] = '\0'; - char *p = strchr(opt1, '='); + char* p = strchr(opt1, '='); if (!p) { fprintf(stderr, "Missing = in configvar assignment\n"); exit(1); @@ -190,8 +197,8 @@ void PrintLangsList(tesseract::TessBaseAPI* api) { } void PrintBanner() { - tprintf("Tesseract Open Source OCR Engine v%s with Leptonica\n", - tesseract::TessBaseAPI::Version()); + tprintf("Tesseract Open Source OCR Engine v%s with Leptonica\n", + tesseract::TessBaseAPI::Version()); } /** @@ -209,31 +216,26 @@ void PrintBanner() { * but that doesn't work. */ void FixPageSegMode(tesseract::TessBaseAPI* api, - tesseract::PageSegMode pagesegmode) { + tesseract::PageSegMode pagesegmode) { if (api->GetPageSegMode() == tesseract::PSM_SINGLE_BLOCK) - api->SetPageSegMode(pagesegmode); + api->SetPageSegMode(pagesegmode); } // NOTE: arg_i is used here to avoid ugly *i so many times in this function -void ParseArgs(const int argc, char** argv, - const char** lang, - const char** image, - const char** outputbase, - const char** datapath, - bool* list_langs, - bool* print_parameters, - GenericVector* vars_vec, - GenericVector* vars_values, - int* arg_i, - tesseract::PageSegMode* pagesegmode) { +void ParseArgs(const int argc, char** argv, const char** lang, + const char** image, const char** outputbase, + const char** datapath, bool* list_langs, bool* print_parameters, + GenericVector* vars_vec, + GenericVector* vars_values, int* arg_i, + tesseract::PageSegMode* pagesegmode, + tesseract::OcrEngineMode* enginemode) { if (argc == 1) { PrintHelpMessage(argv[0]); exit(0); } if (argc == 2) { - if ((strcmp(argv[1], "-h") == 0) || - (strcmp(argv[1], "--help") == 0)) { + if ((strcmp(argv[1], "-h") == 0) || (strcmp(argv[1], "--help") == 0)) { PrintHelpMessage(argv[0]); exit(0); } @@ -241,8 +243,11 @@ void ParseArgs(const int argc, char** argv, PrintHelpForPSM(); exit(0); } - if ((strcmp(argv[1], "-v") == 0) || - (strcmp(argv[1], "--version") == 0)) { + if ((strcmp(argv[1], "--help-oem") == 0)) { + PrintHelpForOEM(); + exit(0); + } + if ((strcmp(argv[1], "-v") == 0) || (strcmp(argv[1], "--version") == 0)) { PrintVersionInfo(); exit(0); } @@ -269,8 +274,16 @@ void ParseArgs(const int argc, char** argv, noocr = true; *list_langs = true; } else if (strcmp(argv[i], "-psm") == 0 && i + 1 < argc) { + // The parameter -psm is deprecated and was replaced by --psm. + // It is still supported for compatibility reasons. + *pagesegmode = static_cast(atoi(argv[i + 1])); + ++i; + } else if (strcmp(argv[i], "--psm") == 0 && i + 1 < argc) { *pagesegmode = static_cast(atoi(argv[i + 1])); ++i; + } else if (strcmp(argv[i], "--oem") == 0 && i + 1 < argc) { + *enginemode = static_cast(atoi(argv[i + 1])); + ++i; } else if (strcmp(argv[i], "--print-parameters") == 0) { noocr = true; *print_parameters = true; @@ -298,10 +311,10 @@ void ParseArgs(const int argc, char** argv, } } -void PreloadRenderers(tesseract::TessBaseAPI* api, - tesseract::PointerVector* renderers, - tesseract::PageSegMode pagesegmode, - const char* outputbase) { +void PreloadRenderers( + tesseract::TessBaseAPI* api, + tesseract::PointerVector* renderers, + tesseract::PageSegMode pagesegmode, const char* outputbase) { if (pagesegmode == tesseract::PSM_OSD_ONLY) { renderers->push_back(new tesseract::TessOsdRenderer(outputbase)); } else { @@ -311,7 +324,7 @@ void PreloadRenderers(tesseract::TessBaseAPI* api, bool font_info; api->GetBoolVariable("hocr_font_info", &font_info); renderers->push_back( - new tesseract::TessHOcrRenderer(outputbase, font_info)); + new tesseract::TessHOcrRenderer(outputbase, font_info)); } api->GetBoolVariable("tessedit_create_tsv", &b); @@ -324,8 +337,8 @@ void PreloadRenderers(tesseract::TessBaseAPI* api, api->GetBoolVariable("tessedit_create_pdf", &b); if (b) { - renderers->push_back(new tesseract::TessPDFRenderer(outputbase, - api->GetDatapath())); + renderers->push_back( + new tesseract::TessPDFRenderer(outputbase, api->GetDatapath())); } api->GetBoolVariable("tessedit_write_unlv", &b); @@ -359,8 +372,7 @@ void PreloadRenderers(tesseract::TessBaseAPI* api, * **********************************************************************/ - -int main(int argc, char **argv) { +int main(int argc, char** argv) { const char* lang = "eng"; const char* image = NULL; const char* outputbase = NULL; @@ -369,21 +381,26 @@ int main(int argc, char **argv) { bool print_parameters = false; int arg_i = 1; tesseract::PageSegMode pagesegmode = tesseract::PSM_AUTO; + tesseract::OcrEngineMode enginemode = tesseract::OEM_DEFAULT; /* main() calls functions like ParseArgs which call exit(). * This results in memory leaks if vars_vec and vars_values are * declared as auto variables (destructor is not called then). */ static GenericVector vars_vec; static GenericVector vars_values; +#if !defined(DEBUG) + // Disable debugging and informational messages from Leptonica. + setMsgSeverity(L_SEVERITY_WARNING); +#endif + #if defined(HAVE_TIFFIO_H) && defined(_WIN32) /* Show libtiff warnings on console (not in GUI). */ TIFFSetWarningHandler(Win32WarningHandler); #endif /* HAVE_TIFFIO_H && _WIN32 */ - ParseArgs(argc, argv, - &lang, &image, &outputbase, &datapath, - &list_langs, &print_parameters, - &vars_vec, &vars_values, &arg_i, &pagesegmode); + ParseArgs(argc, argv, &lang, &image, &outputbase, &datapath, &list_langs, + &print_parameters, &vars_vec, &vars_values, &arg_i, &pagesegmode, + &enginemode); bool banner = false; if (outputbase != NULL && strcmp(outputbase, "-") && @@ -396,8 +413,8 @@ int main(int argc, char **argv) { api.SetOutputName(outputbase); - int init_failed = api.Init(datapath, lang, tesseract::OEM_DEFAULT, - &(argv[arg_i]), argc - arg_i, &vars_vec, &vars_values, false); + int init_failed = api.Init(datapath, lang, enginemode, &(argv[arg_i]), + argc - arg_i, &vars_vec, &vars_values, false); if (init_failed) { fprintf(stderr, "Could not initialize tesseract.\n"); exit(1); @@ -406,8 +423,8 @@ int main(int argc, char **argv) { SetVariablesFromCLArgs(&api, argc, argv); if (list_langs) { - PrintLangsList(&api); - exit(0); + PrintLangsList(&api); + exit(0); } if (print_parameters) { @@ -436,12 +453,13 @@ int main(int argc, char **argv) { tesseract::TextlineOrder order; float deskew_angle; - tesseract::PageIterator* it = api.AnalyseLayout(); + tesseract::PageIterator* it = api.AnalyseLayout(); if (it) { it->Orientation(&orientation, &direction, &order, &deskew_angle); - tprintf("Orientation: %d\nWritingDirection: %d\nTextlineOrder: %d\n" \ - "Deskew angle: %.4f\n", - orientation, direction, order, deskew_angle); + tprintf( + "Orientation: %d\nWritingDirection: %d\nTextlineOrder: %d\n" + "Deskew angle: %.4f\n", + orientation, direction, order, deskew_angle); } else { ret_val = 1; } @@ -456,14 +474,12 @@ int main(int argc, char **argv) { // ambigs.train, box.train, box.train.stderr, linebox, rebox bool b = false; bool in_training_mode = - (api.GetBoolVariable("tessedit_ambigs_training", &b) && b) || - (api.GetBoolVariable("tessedit_resegment_from_boxes", &b) && b) || - (api.GetBoolVariable("tessedit_make_boxes_from_boxes", &b) && b); + (api.GetBoolVariable("tessedit_ambigs_training", &b) && b) || + (api.GetBoolVariable("tessedit_resegment_from_boxes", &b) && b) || + (api.GetBoolVariable("tessedit_make_boxes_from_boxes", &b) && b); tesseract::PointerVector renderers; - - if (in_training_mode) { renderers.push_back(NULL); } else { diff --git a/appveyor.yml b/appveyor.yml index 020331c461..c6aeee9221 100644 --- a/appveyor.yml +++ b/appveyor.yml @@ -1,25 +1,45 @@ -os: Visual Studio 2015 +os: Visual Studio 2017 platform: - Win32 - Win64 configuration: - - Release + - Release + + # for curl + install: + - set PATH=C:\Program Files\Git\mingw64\bin;%PATH% before_build: - - if %platform%==Win32 set generator=Visual Studio 14 - - if %platform%==Win64 set generator=Visual Studio 14 Win64 + - if %platform%==Win32 set generator=Visual Studio 15 2017 + - if %platform%==Win64 set generator=Visual Studio 15 2017 Win64 - if %platform%==Win32 set vcplatform=Win32 - if %platform%==Win64 set vcplatform=x64 - - - curl -fsS -o cppan.zip https://cppan.org/client/cppan-master-Windows-client.zip + + - curl -fsS -L -o cppan.zip https://cppan.org/client/cppan-master-Windows-client.zip - 7z x cppan.zip - set PATH=%PATH%;%cd% - + + - cppan # dummy run to create %USERPROFILE%\.cppan\cppan.yml + - ps: 'Add-Content $env:USERPROFILE\.cppan\cppan.yml "`n`nvar_check_jobs: 1`n"' + - ps: 'Add-Content $env:USERPROFILE\.cppan\cppan.yml "`n`nbuild_warning_level: 0`n"' + - ps: 'Add-Content $env:USERPROFILE\.cppan\cppan.yml "`n`nbuild_system_verbose: false`n"' + build_script: - - cppan - mkdir build + - mkdir build\bin + - mkdir build\bin\Release - cd build - - cmake .. -G "%generator%" -DSTATIC=1 - - msbuild tesseract.sln /p:Platform=%vcplatform% /logger:"C:\Program Files\AppVeyor\BuildAgent\Appveyor.MSBuildLogger.dll" + #- cmd: 'echo local_settings: > cppan.yml' + #- cmd: 'echo generator: %generator% >> cppan.yml' + #- cmd: 'echo use_shared_libs: true >> cppan.yml' + #- cppan --build .. + - cmake .. -G "%generator%" -DBUILD_TRAINING_TOOLS=Off -DAPPVEYOR=1 + - cmake --build . --config Release > bin\Release\log.txt 2>&1 + +artifacts: + - path: build\bin\Release + #- path: build + name: tesseract-$(APPVEYOR_BUILD_VERSION) + diff --git a/autogen.sh b/autogen.sh index ac44d35770..5319afbc86 100755 --- a/autogen.sh +++ b/autogen.sh @@ -1,4 +1,13 @@ #!/bin/sh +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. # This is a simple script which is meant to help developers # better deal with the GNU autotools, specifically: @@ -37,6 +46,19 @@ if [ "$1" = "clean" ]; then find . -iname "Makefile.in" -type f -exec rm '{}' + fi +# Prevent any errors that might result from failing to properly invoke +# `libtoolize` or `glibtoolize,` whichever is present on your system, +# from occurring by testing for its existence and capturing the absolute path to +# its location for caching purposes prior to using it later on in 'Step 2:' +if command -v libtoolize >/dev/null 2>&1; then + LIBTOOLIZE="$(command -v libtoolize)" +elif command -v glibtoolize >/dev/null 2>&1; then + LIBTOOLIZE="$(command -v glibtoolize)" +else + echo "Unable to find a valid copy of libtoolize or glibtoolize in your PATH!" + bail_out +fi + # create m4 directory if it not exists if [ ! -d m4 ]; then mkdir m4 @@ -61,9 +83,9 @@ aclocal -I config || bail_out # --- Step 2: -echo "Running libtoolize" -libtoolize -f -c || glibtoolize -f -c || bail_out -libtoolize --automake || glibtoolize --automake || bail_out +echo "Running $LIBTOOLIZE" +$LIBTOOLIZE -f -c || bail_out +$LIBTOOLIZE --automake || bail_out # --- Step 3: Generate config.h.in from: # . configure.ac (look for AM_CONFIG_HEADER tag or AC_CONFIG_HEADER tag) diff --git a/ccmain/control.cpp b/ccmain/control.cpp index 3970c5429e..cdc2bb5a6e 100644 --- a/ccmain/control.cpp +++ b/ccmain/control.cpp @@ -1,8 +1,8 @@ /****************************************************************** * File: control.cpp (Formerly control.c) * Description: Module-independent matcher controller. - * Author: Ray Smith - * Created: Thu Apr 23 11:09:58 BST 1992 + * Author: Ray Smith + * Created: Thu Apr 23 11:09:58 BST 1992 * ReHacked: Tue Sep 22 08:42:49 BST 1992 Phil Cheatle * * (C) Copyright 1992, Hewlett-Packard Ltd. @@ -73,7 +73,6 @@ void Tesseract::recog_pseudo_word(PAGE_RES* page_res, } } - /** * Recognize a single word in interactive mode. * @@ -219,16 +218,14 @@ bool Tesseract::RecogAllWordsPassN(int pass_n, ETEXT_DESC* monitor, if (pass_n == 1) { monitor->progress = 70 * w / words->size(); if (monitor->progress_callback != NULL) { - TBOX box = pr_it->word()->word->bounding_box(); - (*monitor->progress_callback)(monitor->progress, - box.left(), box.right(), - box.top(), box.bottom()); + TBOX box = pr_it->word()->word->bounding_box(); + (*monitor->progress_callback)(monitor->progress, box.left(), + box.right(), box.top(), box.bottom()); } } else { monitor->progress = 70 + 30 * w / words->size(); - if (monitor->progress_callback!=NULL) { - (*monitor->progress_callback)(monitor->progress, - 0, 0, 0, 0); + if (monitor->progress_callback != NULL) { + (*monitor->progress_callback)(monitor->progress, 0, 0, 0, 0); } } if (monitor->deadline_exceeded() || @@ -539,7 +536,7 @@ void Tesseract::bigram_correction_pass(PAGE_RES *page_res) { } } } - if (overrides_word1.size() >= 1) { + if (!overrides_word1.empty()) { // Excellent, we have some bigram matches. if (EqualIgnoringCaseAndTerminalPunct(*w_prev->best_choice, *overrides_word1[best_idx]) && diff --git a/ccmain/cube_control.cpp b/ccmain/cube_control.cpp index 1430debc5b..50f7512dd1 100644 --- a/ccmain/cube_control.cpp +++ b/ccmain/cube_control.cpp @@ -5,6 +5,16 @@ * Author: Raquel Romano * Created: September 2009 * + * (C) Copyright 2009, Google Inc. + ** Licensed under the Apache License, Version 2.0 (the "License"); + ** you may not use this file except in compliance with the License. + ** You may obtain a copy of the License at + ** http://www.apache.org/licenses/LICENSE-2.0 + ** Unless required by applicable law or agreed to in writing, software + ** distributed under the License is distributed on an "AS IS" BASIS, + ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + ** See the License for the specific language governing permissions and + ** limitations under the License. **********************************************************************/ // Include automatically generated configuration file if running autoconf. @@ -169,13 +179,11 @@ bool Tesseract::init_cube_objects(bool load_combiner, // Create the combiner object and load the combiner net for target languages. if (load_combiner) { tess_cube_combiner_ = new tesseract::TesseractCubeCombiner(cube_cntxt_); - if (!tess_cube_combiner_ || !tess_cube_combiner_->LoadCombinerNet()) { + if (!tess_cube_combiner_->LoadCombinerNet()) { delete cube_cntxt_; cube_cntxt_ = NULL; - if (tess_cube_combiner_ != NULL) { - delete tess_cube_combiner_; - tess_cube_combiner_ = NULL; - } + delete tess_cube_combiner_; + tess_cube_combiner_ = NULL; if (cube_debug_level > 0) tprintf("Cube ERROR (Failed to instantiate TesseractCubeCombiner\n"); return false; diff --git a/ccmain/cube_reco_context.cpp b/ccmain/cube_reco_context.cpp index fed53f0954..dadc0624a4 100644 --- a/ccmain/cube_reco_context.cpp +++ b/ccmain/cube_reco_context.cpp @@ -55,40 +55,26 @@ CubeRecoContext::CubeRecoContext(Tesseract *tess_obj) { } CubeRecoContext::~CubeRecoContext() { - if (char_classifier_ != NULL) { - delete char_classifier_; - char_classifier_ = NULL; - } + delete char_classifier_; + char_classifier_ = NULL; - if (word_size_model_ != NULL) { - delete word_size_model_; - word_size_model_ = NULL; - } + delete word_size_model_; + word_size_model_ = NULL; - if (char_set_ != NULL) { - delete char_set_; - char_set_ = NULL; - } + delete char_set_; + char_set_ = NULL; - if (char_bigrams_ != NULL) { - delete char_bigrams_; - char_bigrams_ = NULL; - } + delete char_bigrams_; + char_bigrams_ = NULL; - if (word_unigrams_ != NULL) { - delete word_unigrams_; - word_unigrams_ = NULL; - } + delete word_unigrams_; + word_unigrams_ = NULL; - if (lang_mod_ != NULL) { - delete lang_mod_; - lang_mod_ = NULL; - } + delete lang_mod_; + lang_mod_ = NULL; - if (params_ != NULL) { - delete params_; - params_ = NULL; - } + delete params_; + params_ = NULL; } /** @@ -145,11 +131,6 @@ bool CubeRecoContext::Load(TessdataManager *tessdata_manager, lang_mod_ = new TessLangModel(lm_params, data_file_path, tess_obj_->getDict().load_system_dawg, tessdata_manager, this); - if (lang_mod_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeRecoContext::Load): unable to create " - "TessLangModel\n"); - return false; - } // Create the optional char bigrams object. char_bigrams_ = CharBigrams::Create(data_file_path, lang_); @@ -190,11 +171,6 @@ CubeRecoContext * CubeRecoContext::Create(Tesseract *tess_obj, UNICHARSET *tess_unicharset) { // create the object CubeRecoContext *cntxt = new CubeRecoContext(tess_obj); - if (cntxt == NULL) { - fprintf(stderr, "Cube ERROR (CubeRecoContext::Create): unable to create " - "CubeRecoContext object\n"); - return NULL; - } // load the necessary components if (cntxt->Load(tessdata_manager, tess_unicharset) == false) { fprintf(stderr, "Cube ERROR (CubeRecoContext::Create): unable to init " diff --git a/ccmain/cubeclassifier.h b/ccmain/cubeclassifier.h index 98bdb5cf00..3ae00f7974 100644 --- a/ccmain/cubeclassifier.h +++ b/ccmain/cubeclassifier.h @@ -23,6 +23,7 @@ #define THIRD_PARTY_TESSERACT_CCMAIN_CUBECLASSIFIER_H_ #include "shapeclassifier.h" +#include "platform.h" namespace tesseract { @@ -35,7 +36,7 @@ class TrainingSample; struct UnicharRating; // Cube implementation of a ShapeClassifier. -class CubeClassifier : public ShapeClassifier { +class TESS_API CubeClassifier : public ShapeClassifier { public: explicit CubeClassifier(Tesseract* tesseract); virtual ~CubeClassifier(); @@ -55,7 +56,7 @@ class CubeClassifier : public ShapeClassifier { }; // Combination of Tesseract class pruner with scoring by cube. -class CubeTessClassifier : public ShapeClassifier { +class TESS_API CubeTessClassifier : public ShapeClassifier { public: explicit CubeTessClassifier(Tesseract* tesseract); virtual ~CubeTessClassifier(); diff --git a/ccmain/docqual.cpp b/ccmain/docqual.cpp index c6e7f17e0f..4706fb3b26 100644 --- a/ccmain/docqual.cpp +++ b/ccmain/docqual.cpp @@ -1,8 +1,8 @@ /****************************************************************** * File: docqual.cpp (Formerly docqual.c) * Description: Document Quality Metrics - * Author: Phil Cheatle - * Created: Mon May 9 11:27:28 BST 1994 + * Author: Phil Cheatle + * Created: Mon May 9 11:27:28 BST 1994 * * (C) Copyright 1994, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -98,8 +98,8 @@ void Tesseract::word_char_quality(WERD_RES *word, ROW *row, inT16 *match_count, inT16 *accepted_match_count) { - if (word->bln_boxes == NULL || - word->rebuild_word == NULL || word->rebuild_word->blobs.empty()) { + if (word->bln_boxes == NULL || word->rebuild_word == NULL || + word->rebuild_word->blobs.empty()) { *match_count = 0; *accepted_match_count = 0; return; @@ -132,7 +132,7 @@ inT16 Tesseract::count_outline_errs(char c, inT16 outline_count) { int expected_outline_count; if (STRING (outlines_odd).contains (c)) - return 0; //Don't use this char + return 0; // Don't use this char else if (STRING (outlines_2).contains (c)) expected_outline_count = 2; else @@ -151,17 +151,16 @@ void Tesseract::quality_based_rejection(PAGE_RES_IT &page_res_it, } } - /************************************************************************* * unrej_good_quality_words() * Accept potential rejects in words which pass the following checks: * - Contains a potential reject * - Word looks like a sensible alpha word. * - Word segmentation is the same as the original image - * - All characters have the expected number of outlines + * - All characters have the expected number of outlines * NOTE - the rejection counts are recalculated after unrejection * - CAN'T do it in a single pass without a bit of fiddling - * - keep it simple but inefficient + * - keep it simple but inefficient *************************************************************************/ void Tesseract::unrej_good_quality_words( //unreject potential PAGE_RES_IT &page_res_it) { @@ -403,7 +402,6 @@ void Tesseract::doc_and_block_rejection( //reject big chunks } // namespace tesseract - /************************************************************************* * reject_whole_page() * Don't believe any of it - set the reject map to 00..00 in all words diff --git a/ccmain/equationdetect.cpp b/ccmain/equationdetect.cpp index 06aab24923..5fa955e432 100644 --- a/ccmain/equationdetect.cpp +++ b/ccmain/equationdetect.cpp @@ -20,6 +20,7 @@ #ifdef _MSC_VER #pragma warning(disable:4244) // Conversion warnings #include +#include #endif #ifdef __MINGW32__ diff --git a/ccmain/fixspace.cpp b/ccmain/fixspace.cpp index f58c9610fa..5fbe8c9a4e 100644 --- a/ccmain/fixspace.cpp +++ b/ccmain/fixspace.cpp @@ -3,8 +3,8 @@ * Description: Implements a pass over the page res, exploring the alternative * spacing possibilities, trying to use context to improve the * word spacing -* Author: Phil Cheatle -* Created: Thu Oct 21 11:38:43 BST 1993 +* Author: Phil Cheatle +* Created: Thu Oct 21 11:38:43 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -211,7 +211,6 @@ void Tesseract::match_current_words(WERD_RES_LIST &words, ROW *row, } } - /** * @name eval_word_spacing() * The basic measure is the number of characters in contextually confirmed diff --git a/ccmain/ltrresultiterator.cpp b/ccmain/ltrresultiterator.cpp index d5b8594667..f80e594518 100644 --- a/ccmain/ltrresultiterator.cpp +++ b/ccmain/ltrresultiterator.cpp @@ -145,13 +145,12 @@ float LTRResultIterator::Confidence(PageIteratorLevel level) const { return 0.0f; } -void LTRResultIterator::RowAttributes(float* row_height, - float* descenders, +void LTRResultIterator::RowAttributes(float* row_height, float* descenders, float* ascenders) const { - *row_height = it_->row()->row->x_height() + it_->row()-> row->ascenders() - - it_->row()->row->descenders(); - *descenders = it_->row()->row->descenders(); - *ascenders = it_->row()->row->ascenders(); + *row_height = it_->row()->row->x_height() + it_->row()->row->ascenders() - + it_->row()->row->descenders(); + *descenders = it_->row()->row->descenders(); + *ascenders = it_->row()->row->ascenders(); } // Returns the font attributes of the current word. If iterating at a higher diff --git a/ccmain/ltrresultiterator.h b/ccmain/ltrresultiterator.h index 8819c2a0ee..f2605b52d2 100644 --- a/ccmain/ltrresultiterator.h +++ b/ccmain/ltrresultiterator.h @@ -92,8 +92,7 @@ class TESS_API LTRResultIterator : public PageIterator { float Confidence(PageIteratorLevel level) const; // Returns the attributes of the current row. - void RowAttributes(float* row_height, - float* descenders, + void RowAttributes(float* row_height, float* descenders, float* ascenders) const; // ============= Functions that refer to words only ============. diff --git a/ccmain/osdetect.cpp b/ccmain/osdetect.cpp index f2fe94a1c3..69e626222f 100644 --- a/ccmain/osdetect.cpp +++ b/ccmain/osdetect.cpp @@ -164,8 +164,14 @@ void remove_nontext_regions(tesseract::Tesseract *tess, BLOCK_LIST *blocks, int vertical_y = 1; tesseract::TabVector_LIST v_lines; tesseract::TabVector_LIST h_lines; - int resolution = (kMinCredibleResolution > pixGetXRes(pix)) ? - kMinCredibleResolution : pixGetXRes(pix); + int resolution; + if (kMinCredibleResolution > pixGetXRes(pix)) { + resolution = kMinCredibleResolution; + tprintf("Warning. Invalid resolution %d dpi. Using %d instead.\n", + pixGetXRes(pix), resolution); + } else { + resolution = pixGetXRes(pix); + } tesseract::LineFinder::FindAndRemoveLines(resolution, false, pix, &vertical_x, &vertical_y, diff --git a/ccmain/output.cpp b/ccmain/output.cpp index ddfcfc54b6..6fca63e420 100644 --- a/ccmain/output.cpp +++ b/ccmain/output.cpp @@ -1,8 +1,8 @@ /****************************************************************** * File: output.cpp (Formerly output.c) * Description: Output pass - * Author: Phil Cheatle - * Created: Thu Aug 4 10:56:08 BST 1994 + * Author: Phil Cheatle + * Created: Thu Aug 4 10:56:08 BST 1994 * * (C) Copyright 1994, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -78,18 +78,16 @@ void Tesseract::output_pass( //Tess output pass //send to api while (page_res_it.word () != NULL) { check_debug_pt (page_res_it.word (), 120); - if (target_word_box) - { - - TBOX current_word_box=page_res_it.word ()->word->bounding_box(); - FCOORD center_pt((current_word_box.right()+current_word_box.left())/2,(current_word_box.bottom()+current_word_box.top())/2); - if (!target_word_box->contains(center_pt)) - { - page_res_it.forward (); - continue; - } - - } + if (target_word_box) { + TBOX current_word_box = page_res_it.word()->word->bounding_box(); + FCOORD center_pt( + (current_word_box.right() + current_word_box.left()) / 2, + (current_word_box.bottom() + current_word_box.top()) / 2); + if (!target_word_box->contains(center_pt)) { + page_res_it.forward(); + continue; + } + } if (tessedit_write_block_separators && block_of_last_word != page_res_it.block ()) { block_of_last_word = page_res_it.block (); @@ -337,7 +335,7 @@ void Tesseract::set_unlv_suspects(WERD_RES *word_res) { rating_per_ch = word.rating() / word_res->reject_map.length(); if (rating_per_ch >= suspect_rating_per_ch) - return; //Don't touch bad ratings + return; // Don't touch bad ratings if ((word_res->tess_accepted) || (rating_per_ch < suspect_accept_rating)) { /* Unreject any Tess Acceptable word - but NOT tess reject chs*/ diff --git a/ccmain/pageiterator.cpp b/ccmain/pageiterator.cpp index fc15840c44..7d7865ae30 100644 --- a/ccmain/pageiterator.cpp +++ b/ccmain/pageiterator.cpp @@ -87,7 +87,7 @@ const PageIterator& PageIterator::operator=(const PageIterator& src) { rect_top_ = src.rect_top_; rect_width_ = src.rect_width_; rect_height_ = src.rect_height_; - if (it_ != NULL) delete it_; + delete it_; it_ = new PAGE_RES_IT(*src.it_); BeginWord(src.blob_index_); return *this; @@ -597,10 +597,8 @@ void PageIterator::BeginWord(int offset) { } word_ = NULL; // We will be iterating the box_word. - if (cblob_it_ != NULL) { - delete cblob_it_; - cblob_it_ = NULL; - } + delete cblob_it_; + cblob_it_ = NULL; } else { // No recognition yet, so a "symbol" is a cblob. word_ = word_res->word; diff --git a/ccmain/pagesegmain.cpp b/ccmain/pagesegmain.cpp index 4e3c342070..d815365137 100644 --- a/ccmain/pagesegmain.cpp +++ b/ccmain/pagesegmain.cpp @@ -18,9 +18,6 @@ **********************************************************************/ #ifdef _WIN32 -#ifndef __GNUC__ -#include -#endif // __GNUC__ #ifndef unlink #include #endif @@ -412,9 +409,10 @@ ColumnFinder* Tesseract::SetupPageSegAndDetectOrientation( "Don't rotate.\n", osd_margin); osd_orientation = 0; } else { - tprintf("OSD: Weak margin (%.2f) for %d blob text block, " - "but using orientation anyway: %d\n", - osd_margin, osd_blobs.length(), osd_orientation); + tprintf( + "OSD: Weak margin (%.2f) for %d blob text block, " + "but using orientation anyway: %d\n", + osd_margin, osd_blobs.length(), osd_orientation); } } } diff --git a/ccmain/par_control.cpp b/ccmain/par_control.cpp index 7a7d0415d6..6797a5a2a7 100644 --- a/ccmain/par_control.cpp +++ b/ccmain/par_control.cpp @@ -18,9 +18,6 @@ /////////////////////////////////////////////////////////////////////// #include "tesseractclass.h" -#ifdef OPENMP -#include -#endif // OPENMP namespace tesseract { @@ -53,7 +50,6 @@ void Tesseract::PrerecAllWordsPar(const GenericVector& words) { } // Pre-classify all the blobs. if (tessedit_parallelize > 1) { - #pragma omp parallel for num_threads(10) for (int b = 0; b < blobs.size(); ++b) { *blobs[b].choices = blobs[b].tesseract->classify_blob(blobs[b].blob, "par", White, NULL); diff --git a/ccmain/paragraphs.cpp b/ccmain/paragraphs.cpp index 7459940dd0..c7d21a9192 100644 --- a/ccmain/paragraphs.cpp +++ b/ccmain/paragraphs.cpp @@ -2052,7 +2052,7 @@ void ConvertHypothesizedModelRunsToParagraphs( bool single_line_paragraph = false; SetOfModels models; rows[start].NonNullHypotheses(&models); - if (models.size() > 0) { + if (!models.empty()) { model = models[0]; if (rows[start].GetLineType(model) != LT_BODY) single_line_paragraph = true; @@ -2190,17 +2190,17 @@ void LeftoverSegments(const GenericVector &rows, SetOfModels models_w_crowns; rows[i].StrongHypotheses(&models); rows[i].NonNullHypotheses(&models_w_crowns); - if (models.empty() && models_w_crowns.size() > 0) { + if (models.empty() && !models_w_crowns.empty()) { // Crown paragraph. Is it followed by a modeled line? for (int end = i + 1; end < rows.size(); end++) { SetOfModels end_models; SetOfModels strong_end_models; rows[end].NonNullHypotheses(&end_models); rows[end].StrongHypotheses(&strong_end_models); - if (end_models.size() == 0) { + if (end_models.empty()) { needs_fixing = true; break; - } else if (strong_end_models.size() > 0) { + } else if (!strong_end_models.empty()) { needs_fixing = false; break; } @@ -2485,7 +2485,7 @@ void InitializeRowInfo(bool after_recognition, info->ltr = ltr >= rtl; info->has_leaders = num_leaders > 3; info->num_words = werds.size(); - if (werds.size() > 0) { + if (!werds.empty()) { WERD_RES *lword = werds[0], *rword = werds[werds.size() - 1]; info->lword_text = lword->best_choice->unichar_string().string(); info->rword_text = rword->best_choice->unichar_string().string(); @@ -2538,7 +2538,7 @@ void DetectParagraphs(int debug_level, // If we're called before text recognition, we might not have // tight block bounding boxes, so trim by the minimum on each side. - if (row_infos.size() > 0) { + if (!row_infos.empty()) { int min_lmargin = row_infos[0].pix_ldistance; int min_rmargin = row_infos[0].pix_rdistance; for (int i = 1; i < row_infos.size(); i++) { diff --git a/ccmain/paramsd.cpp b/ccmain/paramsd.cpp index 7784f85361..e0e60539f6 100644 --- a/ccmain/paramsd.cpp +++ b/ccmain/paramsd.cpp @@ -329,13 +329,19 @@ void ParamsEditor::WriteParams(char *filename, fclose(fp); sprintf (msg_str, "Overwrite file " "%s" "? (Y/N)", filename); int a = sv_window_->ShowYesNoDialog(msg_str); - if (a == 'n') { return; } // don't write + if (a == 'n') { + return; + } // don't write } fp = fopen (filename, "wb"); // can we write to it? if (fp == NULL) { - sv_window_->AddMessage("Can't write to file " "%s" "", filename); + sv_window_->AddMessage( + "Can't write to file " + "%s" + "", + filename); return; } diff --git a/ccmain/pgedit.cpp b/ccmain/pgedit.cpp index d78c0dacc0..5e23595422 100644 --- a/ccmain/pgedit.cpp +++ b/ccmain/pgedit.cpp @@ -191,7 +191,7 @@ ScrollView* bln_word_window_handle() { // return handle */ void build_image_window(int width, int height) { - if (image_win != NULL) { delete image_win; } + delete image_win; image_win = new ScrollView(editor_image_win_name.string(), editor_image_xpos, editor_image_ypos, width + 1, diff --git a/ccmain/reject.cpp b/ccmain/reject.cpp index aacc80dd6e..72f9d873d1 100644 --- a/ccmain/reject.cpp +++ b/ccmain/reject.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: reject.cpp (Formerly reject.c) * Description: Rejection functions used in tessedit - * Author: Phil Cheatle - * Created: Wed Sep 23 16:50:21 BST 1992 + * Author: Phil Cheatle + * Created: Wed Sep 23 16:50:21 BST 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); diff --git a/ccmain/tessedit.cpp b/ccmain/tessedit.cpp index dd96ba0ebd..b2fa02196a 100644 --- a/ccmain/tessedit.cpp +++ b/ccmain/tessedit.cpp @@ -44,7 +44,7 @@ #include "params.h" #define VARDIR "configs/" /*variables files */ - //config under api + // config under api #define API_CONFIG "configs/api_config" ETEXT_DESC *global_monitor = NULL; // progress monitor @@ -468,7 +468,9 @@ int Tesseract::init_tesseract_lm(const char *arg0, if (!init_tesseract_lang_data(arg0, textbase, language, OEM_TESSERACT_ONLY, NULL, 0, NULL, NULL, false)) return -1; - getDict().Load(Dict::GlobalDawgCache()); + getDict().SetupForLoad(Dict::GlobalDawgCache()); + getDict().Load(tessdata_manager.GetDataFileName().string(), lang); + getDict().FinishLoad(); tessdata_manager.End(); return 0; } diff --git a/ccmain/tesseract_cube_combiner.cpp b/ccmain/tesseract_cube_combiner.cpp index e17bd04c2a..2b91f5259d 100644 --- a/ccmain/tesseract_cube_combiner.cpp +++ b/ccmain/tesseract_cube_combiner.cpp @@ -21,8 +21,6 @@ // the recognition results of Tesseract and Cube at the word level #include -#include -#include #include #include "tesseract_cube_combiner.h" diff --git a/ccmain/tesseract_cube_combiner.h b/ccmain/tesseract_cube_combiner.h index 49a0e2f4b0..9a6eb35302 100644 --- a/ccmain/tesseract_cube_combiner.h +++ b/ccmain/tesseract_cube_combiner.h @@ -27,16 +27,6 @@ #include #include "pageres.h" -#ifdef _WIN32 -#include -using namespace std; -#endif - -#ifdef USE_STD_NAMESPACE -using std::string; -using std::vector; -#endif - namespace tesseract { class CubeObject; @@ -72,15 +62,15 @@ class TesseractCubeCombiner { // output parameter will be true if both answers are identical, // false otherwise. Modifies the cube_alt_list, so no assumptions // should be made about its state upon return. - bool ComputeCombinerFeatures(const string &tess_res, + bool ComputeCombinerFeatures(const std::string &tess_res, int tess_confidence, CubeObject *cube_obj, WordAltList *cube_alt_list, - vector *features, + std::vector *features, bool *agreement); // Is the word valid according to Tesseract's language model - bool ValidWord(const string &str); + bool ValidWord(const std::string &str); // Loads the combiner neural network from file, using cube_cntxt_ // to find path. @@ -88,11 +78,11 @@ class TesseractCubeCombiner { private: // Normalize a UTF-8 string. Converts the UTF-8 string to UTF32 and optionally // strips punc and/or normalizes case and then converts back - string NormalizeString(const string &str, bool remove_punc, bool norm_case); + std::string NormalizeString(const std::string &str, bool remove_punc, bool norm_case); // Compares 2 strings after optionally normalizing them and or stripping // punctuation - int CompareStrings(const string &str1, const string &str2, bool ignore_punc, + int CompareStrings(const std::string &str1, const std::string &str2, bool ignore_punc, bool norm_case); NeuralNet *combiner_net_; // pointer to the combiner NeuralNet object diff --git a/ccmain/tesseractclass.cpp b/ccmain/tesseractclass.cpp index 8db50fbd54..5cbf70c8c2 100644 --- a/ccmain/tesseractclass.cpp +++ b/ccmain/tesseractclass.cpp @@ -398,8 +398,8 @@ Tesseract::Tesseract() "Don't suspect dict wds longer than this", this->params()), BOOL_MEMBER(suspect_constrain_1Il, false, "UNLV keep 1Il chars rejected", this->params()), - double_MEMBER(suspect_rating_per_ch, 999.9, "Don't touch bad rating limit", - this->params()), + double_MEMBER(suspect_rating_per_ch, 999.9, + "Don't touch bad rating limit", this->params()), double_MEMBER(suspect_accept_rating, -999.9, "Accept good rating limit", this->params()), BOOL_MEMBER(tessedit_minimal_rejection, false, @@ -512,7 +512,6 @@ Tesseract::Tesseract() "Page separator (default is form feed control character)", this->params()), - // The following parameters were deprecated and removed from their // original // locations. The parameters are temporarily kept here to give Tesseract @@ -606,6 +605,7 @@ Tesseract::Tesseract() pix_binary_(NULL), cube_binary_(NULL), pix_grey_(NULL), + pix_original_(NULL), pix_thresholds_(NULL), source_resolution_(0), textord_(this), @@ -625,6 +625,7 @@ Tesseract::Tesseract() Tesseract::~Tesseract() { Clear(); + pixDestroy(&pix_original_); end_tesseract(); sub_langs_.delete_data_pointers(); #ifndef NO_CUBE_BUILD diff --git a/ccmain/tesseractclass.h b/ccmain/tesseractclass.h index 91d25bc8ae..e01625e354 100644 --- a/ccmain/tesseractclass.h +++ b/ccmain/tesseractclass.h @@ -38,7 +38,6 @@ class BLOB_CHOICE_LIST_CLIST; class BLOCK_LIST; -class CharSamp; struct OSResults; class PAGE_RES; class PAGE_RES_IT; @@ -98,6 +97,7 @@ namespace tesseract { class ColumnFinder; #ifndef NO_CUBE_BUILD +class CharSamp; class CubeLineObject; class CubeObject; class CubeRecoContext; @@ -189,7 +189,7 @@ class Tesseract : public Wordrec { } // Destroy any existing pix and return a pointer to the pointer. Pix** mutable_pix_binary() { - Clear(); + pixDestroy(&pix_binary_); return &pix_binary_; } Pix* pix_binary() const { @@ -202,16 +202,20 @@ class Tesseract : public Wordrec { pixDestroy(&pix_grey_); pix_grey_ = grey_pix; } - // Returns a pointer to a Pix representing the best available image of the - // page. The image will be 8-bit grey if the input was grey or color. Note - // that in grey 0 is black and 255 is white. If the input was binary, then - // the returned Pix will be binary. Note that here black is 1 and white is 0. - // To tell the difference pixGetDepth() will return 8 or 1. - // In either case, the return value is a borrowed Pix, and should not be - // deleted or pixDestroyed. - Pix* BestPix() const { - return pix_grey_ != NULL ? pix_grey_ : pix_binary_; + Pix* pix_original() const { return pix_original_; } + // Takes ownership of the given original_pix. + void set_pix_original(Pix* original_pix) { + pixDestroy(&pix_original_); + pix_original_ = original_pix; } + // Returns a pointer to a Pix representing the best available (original) image + // of the page. Can be of any bit depth, but never color-mapped, as that has + // always been dealt with. Note that in grey and color, 0 is black and 255 is + // white. If the input was binary, then black is 1 and white is 0. + // To tell the difference pixGetDepth() will return 32, 8 or 1. + // In any case, the return value is a borrowed Pix, and should not be + // deleted or pixDestroyed. + Pix* BestPix() const { return pix_original_; } void set_pix_thresholds(Pix* thresholds) { pixDestroy(&pix_thresholds_); pix_thresholds_ = thresholds; @@ -1010,8 +1014,7 @@ class Tesseract : public Wordrec { INT_VAR_H(suspect_level, 99, "Suspect marker level"); INT_VAR_H(suspect_space_level, 100, "Min suspect level for rejecting spaces"); - INT_VAR_H(suspect_short_words, 2, - "Don't Suspect dict wds longer than this"); + INT_VAR_H(suspect_short_words, 2, "Don't Suspect dict wds longer than this"); BOOL_VAR_H(suspect_constrain_1Il, false, "UNLV keep 1Il chars rejected"); double_VAR_H(suspect_rating_per_ch, 999.9, "Don't touch bad rating limit"); double_VAR_H(suspect_accept_rating, -999.9, "Accept good rating limit"); @@ -1175,6 +1178,8 @@ class Tesseract : public Wordrec { Pix* cube_binary_; // Grey-level input image if the input was not binary, otherwise NULL. Pix* pix_grey_; + // Original input image. Color if the input was color. + Pix* pix_original_; // Thresholds that were used to generate the thresholded image from grey. Pix* pix_thresholds_; // Input image resolution after any scaling. The resolution is not well diff --git a/ccmain/thresholder.cpp b/ccmain/thresholder.cpp index df6abd01eb..4208c65ba2 100644 --- a/ccmain/thresholder.cpp +++ b/ccmain/thresholder.cpp @@ -152,19 +152,27 @@ void ImageThresholder::SetImage(const Pix* pix) { int depth; pixGetDimensions(src, &image_width_, &image_height_, &depth); // Convert the image as necessary so it is one of binary, plain RGB, or - // 8 bit with no colormap. - if (depth > 1 && depth < 8) { + // 8 bit with no colormap. Guarantee that we always end up with our own copy, + // not just a clone of the input. + if (pixGetColormap(src)) { + Pix* tmp = pixRemoveColormap(src, REMOVE_CMAP_BASED_ON_SRC); + depth = pixGetDepth(tmp); + if (depth > 1 && depth < 8) { + pix_ = pixConvertTo8(tmp, false); + pixDestroy(&tmp); + } else { + pix_ = tmp; + } + } else if (depth > 1 && depth < 8) { pix_ = pixConvertTo8(src, false); - } else if (pixGetColormap(src)) { - pix_ = pixRemoveColormap(src, REMOVE_CMAP_BASED_ON_SRC); } else { - pix_ = pixClone(src); + pix_ = pixCopy(NULL, src); } depth = pixGetDepth(pix_); pix_channels_ = depth / 8; pix_wpl_ = pixGetWpl(pix_); scale_ = 1; - estimated_res_ = yres_ = pixGetYRes(src); + estimated_res_ = yres_ = pixGetYRes(pix_); Init(); } @@ -173,8 +181,11 @@ void ImageThresholder::SetImage(const Pix* pix) { // Caller must use pixDestroy to free the created Pix. void ImageThresholder::ThresholdToPix(PageSegMode pageseg_mode, Pix** pix) { if (pix_channels_ == 0) { - // We have a binary image, so it just has to be cloned. - *pix = GetPixRect(); + // We have a binary image, but it still has to be copied, as this API + // allows the caller to modify the output. + Pix* original = GetPixRect(); + *pix = pixCopy(NULL, original); + pixDestroy(&original); } else { OtsuThresholdRectToPix(pix_, pix); } @@ -257,10 +268,10 @@ void ImageThresholder::OtsuThresholdRectToPix(Pix* src_pix, OpenclDevice od; if ((num_channels == 4 || num_channels == 1) && od.selectedDeviceIsOpenCL() && rect_top_ == 0 && rect_left_ == 0 ) { - od.ThresholdRectToPixOCL((const unsigned char*)pixGetData(src_pix), - num_channels, pixGetWpl(src_pix) * 4, - thresholds, hi_values, out_pix /*pix_OCL*/, - rect_height_, rect_width_, rect_top_, rect_left_); + od.ThresholdRectToPixOCL((unsigned char*)pixGetData(src_pix), num_channels, + pixGetWpl(src_pix) * 4, thresholds, hi_values, + out_pix /*pix_OCL*/, rect_height_, rect_width_, + rect_top_, rect_left_); } else { #endif ThresholdRectToPix(src_pix, num_channels, thresholds, hi_values, out_pix); diff --git a/ccstruct/blamer.cpp b/ccstruct/blamer.cpp index 5d2837d084..4573e9b3f0 100644 --- a/ccstruct/blamer.cpp +++ b/ccstruct/blamer.cpp @@ -317,7 +317,7 @@ void BlamerBundle::SetChopperBlame(const WERD_RES* word, bool debug) { int num_blobs = word->chopped_word->blobs.size(); int box_index = 0; int blob_index = 0; - inT16 truth_x; + inT16 truth_x = -1; while (box_index < truth_word_.length() && blob_index < num_blobs) { truth_x = norm_truth_word_.BlobBox(box_index).right(); TBLOB * curr_blob = word->chopped_word->blobs[blob_index]; diff --git a/ccstruct/blobbox.cpp b/ccstruct/blobbox.cpp index 280096b5d3..3ffb9dc930 100644 --- a/ccstruct/blobbox.cpp +++ b/ccstruct/blobbox.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: blobbox.cpp (Formerly blobnbox.c) * Description: Code for the textord blob class. - * Author: Ray Smith - * Created: Thu Jul 30 09:08:51 BST 1992 + * Author: Ray Smith + * Created: Thu Jul 30 09:08:51 BST 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -31,7 +31,9 @@ #define PROJECTION_MARGIN 10 //arbitrary #define EXTERN -ELISTIZE (BLOBNBOX) ELIST2IZE (TO_ROW) ELISTIZE (TO_BLOCK) +ELISTIZE(BLOBNBOX) +ELIST2IZE(TO_ROW) +ELISTIZE(TO_BLOCK) // Up to 30 degrees is allowed for rotations of diacritic blobs. const double kCosSmallAngle = 0.866; @@ -176,7 +178,7 @@ void BLOBNBOX::NeighbourGaps(int gaps[BND_COUNT]) const { gaps[dir] = MAX_INT16; BLOBNBOX* neighbour = neighbours_[dir]; if (neighbour != NULL) { - TBOX n_box = neighbour->bounding_box(); + const TBOX& n_box = neighbour->bounding_box(); if (dir == BND_LEFT || dir == BND_RIGHT) { gaps[dir] = box.x_gap(n_box); } else { diff --git a/ccstruct/boxread.cpp b/ccstruct/boxread.cpp index f4aedca5b3..fee0aa9aef 100644 --- a/ccstruct/boxread.cpp +++ b/ccstruct/boxread.cpp @@ -34,8 +34,7 @@ FILE* OpenBoxFile(const STRING& fname) { STRING filename = BoxFileName(fname); FILE* box_file = NULL; if (!(box_file = fopen(filename.string(), "rb"))) { - CANTOPENFILE.error("read_next_box", TESSEXIT, - "Can't open box file %s", + CANTOPENFILE.error("read_next_box", TESSEXIT, "Can't open box file %s", filename.string()); } return box_file; diff --git a/ccstruct/boxword.h b/ccstruct/boxword.h index 742bbb8e4d..c1fab068bb 100644 --- a/ccstruct/boxword.h +++ b/ccstruct/boxword.h @@ -82,9 +82,7 @@ class BoxWord { const TBOX& bounding_box() const { return bbox_; } - int length() const { - return length_; - } + int length() const { return length_; } const TBOX& BlobBox(int index) const { return boxes_[index]; } diff --git a/ccstruct/coutln.cpp b/ccstruct/coutln.cpp index bc2b119d8c..238272d2c6 100644 --- a/ccstruct/coutln.cpp +++ b/ccstruct/coutln.cpp @@ -48,9 +48,9 @@ ICOORD C_OUTLINE::step_coords[4] = { * @param length length of loop */ -C_OUTLINE::C_OUTLINE (CRACKEDGE * startpt, ICOORD bot_left, - ICOORD top_right, inT16 length) - : box (bot_left, top_right), start (startpt->pos), offsets(NULL) { +C_OUTLINE::C_OUTLINE(CRACKEDGE* startpt, ICOORD bot_left, ICOORD top_right, + inT16 length) + : box(bot_left, top_right), start(startpt->pos), offsets(NULL) { inT16 stepindex; //index to step CRACKEDGE *edgept; //current point @@ -71,7 +71,6 @@ C_OUTLINE::C_OUTLINE (CRACKEDGE * startpt, ICOORD bot_left, } } - /** * @name C_OUTLINE::C_OUTLINE * @@ -139,7 +138,7 @@ inT16 length //length of loop * @param rotation rotate to coord */ -C_OUTLINE::C_OUTLINE(C_OUTLINE *srcline, FCOORD rotation) : offsets(NULL) { +C_OUTLINE::C_OUTLINE(C_OUTLINE* srcline, FCOORD rotation) : offsets(NULL) { TBOX new_box; //easy bounding inT16 stepindex; //index to step inT16 dirdiff; //direction change @@ -300,7 +299,6 @@ inT32 C_OUTLINE::perimeter() const { return total_steps; } - /** * @name C_OUTLINE::outer_area * @@ -332,7 +330,6 @@ inT32 C_OUTLINE::outer_area() const { return total; } - /** * @name C_OUTLINE::count_transitions * @@ -459,7 +456,6 @@ inT32 C_OUTLINE::count_transitions(inT32 threshold) { return total; } - /** * @name C_OUTLINE::operator< * @@ -468,8 +464,7 @@ inT32 C_OUTLINE::count_transitions(inT32 threshold) { */ BOOL8 -C_OUTLINE::operator< (const C_OUTLINE & other) const -{ +C_OUTLINE::operator<(const C_OUTLINE& other) const { inT16 count = 0; //winding count ICOORD pos; //position of point inT32 stepindex; //index to cstep @@ -495,7 +490,6 @@ C_OUTLINE::operator< (const C_OUTLINE & other) const return count != 0; } - /** * @name C_OUTLINE::winding_number * @@ -534,7 +528,6 @@ inT16 C_OUTLINE::winding_number(ICOORD point) const { return count; //winding number } - /** * C_OUTLINE::turn_direction * @@ -563,7 +556,6 @@ inT16 C_OUTLINE::turn_direction() const { //winding number return count; //winding number } - /** * @name C_OUTLINE::reverse * @@ -586,7 +578,6 @@ void C_OUTLINE::reverse() { //reverse drection } } - /** * @name C_OUTLINE::move * @@ -661,14 +652,27 @@ static void ComputeGradient(const l_uint32* data, int wpl, int x, int y, int width, int height, ICOORD* gradient) { const l_uint32* line = data + y * wpl; - int pix_x_y = x < width && y < height ? - GET_DATA_BYTE(const_cast (reinterpret_cast(line)), x) : 255; - int pix_x_prevy = x < width && y > 0 ? - GET_DATA_BYTE(const_cast (reinterpret_cast(line - wpl)), x) : 255; - int pix_prevx_prevy = x > 0 && y > 0 ? - GET_DATA_BYTE(const_cast (reinterpret_cast(line - wpl)), x - 1) : 255; - int pix_prevx_y = x > 0 && y < height ? - GET_DATA_BYTE(const_cast (reinterpret_cast(line)), x - 1) : 255; + int pix_x_y = + x < width && y < height + ? GET_DATA_BYTE( + const_cast(reinterpret_cast(line)), x) + : 255; + int pix_x_prevy = + x < width && y > 0 + ? GET_DATA_BYTE( + const_cast(reinterpret_cast(line - wpl)), x) + : 255; + int pix_prevx_prevy = + x > 0 && y > 0 + ? GET_DATA_BYTE( + const_cast(reinterpret_cast(line - wpl)), + x - 1) + : 255; + int pix_prevx_y = + x > 0 && y < height + ? GET_DATA_BYTE( + const_cast(reinterpret_cast(line)), x - 1) + : 255; gradient->set_x(pix_x_y + pix_x_prevy - (pix_prevx_y + pix_prevx_prevy)); gradient->set_y(pix_x_prevy + pix_prevx_prevy - (pix_x_y + pix_prevx_y)); } @@ -684,8 +688,10 @@ static bool EvaluateVerticalDiff(const l_uint32* data, int wpl, int diff_sign, if (y <= 0 || y >= height) return false; const l_uint32* line = data + y * wpl; - int pixel1 = GET_DATA_BYTE(const_cast (reinterpret_cast(line - wpl)), x); - int pixel2 = GET_DATA_BYTE(const_cast (reinterpret_cast(line)), x); + int pixel1 = GET_DATA_BYTE( + const_cast(reinterpret_cast(line - wpl)), x); + int pixel2 = + GET_DATA_BYTE(const_cast(reinterpret_cast(line)), x); int diff = (pixel2 - pixel1) * diff_sign; if (diff > *best_diff) { *best_diff = diff; @@ -705,8 +711,10 @@ static bool EvaluateHorizontalDiff(const l_uint32* line, int diff_sign, int* best_diff, int* best_sum, int* best_x) { if (x <= 0 || x >= width) return false; - int pixel1 = GET_DATA_BYTE(const_cast (reinterpret_cast(line)), x - 1); - int pixel2 = GET_DATA_BYTE(const_cast (reinterpret_cast(line)), x); + int pixel1 = GET_DATA_BYTE( + const_cast(reinterpret_cast(line)), x - 1); + int pixel2 = + GET_DATA_BYTE(const_cast(reinterpret_cast(line)), x); int diff = (pixel2 - pixel1) * diff_sign; if (diff > *best_diff) { *best_diff = diff; @@ -954,8 +962,7 @@ void C_OUTLINE::render_outline(int left, int top, Pix* pix) const { */ #ifndef GRAPHICS_DISABLED -void C_OUTLINE::plot(ScrollView* window, - ScrollView::Color colour) const { +void C_OUTLINE::plot(ScrollView* window, ScrollView::Color colour) const { inT16 stepindex; // index to cstep ICOORD pos; // current position DIR128 stepdir; // direction of step @@ -1016,7 +1023,6 @@ void C_OUTLINE::plot_normed(const DENORM& denorm, ScrollView::Color colour, } #endif - /** * @name C_OUTLINE::operator= * @@ -1024,7 +1030,7 @@ void C_OUTLINE::plot_normed(const DENORM& denorm, ScrollView::Color colour, * @param source assign from this */ -C_OUTLINE & C_OUTLINE::operator= (const C_OUTLINE & source) { +C_OUTLINE& C_OUTLINE::operator=(const C_OUTLINE& source) { box = source.box; start = source.start; if (steps != NULL) diff --git a/ccstruct/fontinfo.cpp b/ccstruct/fontinfo.cpp index d3e6f3756e..536ac280de 100644 --- a/ccstruct/fontinfo.cpp +++ b/ccstruct/fontinfo.cpp @@ -241,7 +241,7 @@ bool read_set(FILE* f, FontSet* fs, bool swap) { if (fread(&fs->size, sizeof(fs->size), 1, f) != 1) return false; if (swap) Reverse32(&fs->size); - fs->configs = new int[fs->size]; + fs->configs = new int32_t[fs->size]; for (int i = 0; i < fs->size; ++i) { if (fread(&fs->configs[i], sizeof(fs->configs[i]), 1, f) != 1) return false; if (swap) diff --git a/ccstruct/fontinfo.h b/ccstruct/fontinfo.h index 5f2d420852..597a179c97 100644 --- a/ccstruct/fontinfo.h +++ b/ccstruct/fontinfo.h @@ -25,6 +25,8 @@ #include "host.h" #include "unichar.h" +#include + template class UnicityTable; namespace tesseract { @@ -135,8 +137,8 @@ struct FontInfo { // the FontInfo in the FontSet structure, it's better to share FontInfos among // FontSets (Classify::fontinfo_table_). struct FontSet { - int size; - int* configs; // FontInfo ids + int32_t size; + int32_t* configs; // FontInfo ids }; // Class that adds a bit of functionality on top of GenericVector to diff --git a/ccstruct/hpdsizes.h b/ccstruct/hpdsizes.h index 2670e21b07..f4d886a0b4 100644 --- a/ccstruct/hpdsizes.h +++ b/ccstruct/hpdsizes.h @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef HPDSIZES_H #define HPDSIZES_H diff --git a/ccstruct/imagedata.cpp b/ccstruct/imagedata.cpp index 3c244c7724..86a2d74deb 100644 --- a/ccstruct/imagedata.cpp +++ b/ccstruct/imagedata.cpp @@ -30,6 +30,16 @@ #include "helpers.h" #include "tprintf.h" +#if defined(__MINGW32__) +# include +#elif __cplusplus > 199711L // in C++11 +# include +#endif + +// Number of documents to read ahead while training. Doesn't need to be very +// large. +const int kMaxReadAhead = 8; + namespace tesseract { WordFeature::WordFeature() : x_(0), y_(0), dir_(0) { @@ -182,6 +192,19 @@ bool ImageData::DeSerialize(bool swap, TFile* fp) { return true; } +// As DeSerialize, but only seeks past the data - hence a static method. +bool ImageData::SkipDeSerialize(bool swap, TFile* fp) { + if (!STRING::SkipDeSerialize(swap, fp)) return false; + inT32 page_number; + if (fp->FRead(&page_number, sizeof(page_number), 1) != 1) return false; + if (!GenericVector::SkipDeSerialize(swap, fp)) return false; + if (!STRING::SkipDeSerialize(swap, fp)) return false; + if (!GenericVector::SkipDeSerialize(swap, fp)) return false; + if (!GenericVector::SkipDeSerializeClasses(swap, fp)) return false; + inT8 vertical = 0; + return fp->FRead(&vertical, sizeof(vertical), 1) == 1; +} + // Saves the given Pix as a PNG-encoded string and destroys it. void ImageData::SetPix(Pix* pix) { SetPixInternal(pix, &image_data_); @@ -195,37 +218,34 @@ Pix* ImageData::GetPix() const { // Gets anything and everything with a non-NULL pointer, prescaled to a // given target_height (if 0, then the original image height), and aligned. // Also returns (if not NULL) the width and height of the scaled image. -// The return value is the scale factor that was applied to the image to -// achieve the target_height. -float ImageData::PreScale(int target_height, Pix** pix, - int* scaled_width, int* scaled_height, - GenericVector* boxes) const { +// The return value is the scaled Pix, which must be pixDestroyed after use, +// and scale_factor (if not NULL) is set to the scale factor that was applied +// to the image to achieve the target_height. +Pix* ImageData::PreScale(int target_height, int max_height, float* scale_factor, + int* scaled_width, int* scaled_height, + GenericVector* boxes) const { int input_width = 0; int input_height = 0; Pix* src_pix = GetPix(); ASSERT_HOST(src_pix != NULL); input_width = pixGetWidth(src_pix); input_height = pixGetHeight(src_pix); - if (target_height == 0) - target_height = input_height; + if (target_height == 0) { + target_height = MIN(input_height, max_height); + } float im_factor = static_cast(target_height) / input_height; if (scaled_width != NULL) *scaled_width = IntCastRounded(im_factor * input_width); if (scaled_height != NULL) *scaled_height = target_height; - if (pix != NULL) { - // Get the scaled image. - pixDestroy(pix); - *pix = pixScale(src_pix, im_factor, im_factor); - if (*pix == NULL) { - tprintf("Scaling pix of size %d, %d by factor %g made null pix!!\n", - input_width, input_height, im_factor); - } - if (scaled_width != NULL) - *scaled_width = pixGetWidth(*pix); - if (scaled_height != NULL) - *scaled_height = pixGetHeight(*pix); + // Get the scaled image. + Pix* pix = pixScale(src_pix, im_factor, im_factor); + if (pix == NULL) { + tprintf("Scaling pix of size %d, %d by factor %g made null pix!!\n", + input_width, input_height, im_factor); } + if (scaled_width != NULL) *scaled_width = pixGetWidth(pix); + if (scaled_height != NULL) *scaled_height = pixGetHeight(pix); pixDestroy(&src_pix); if (boxes != NULL) { // Get the boxes. @@ -241,7 +261,8 @@ float ImageData::PreScale(int target_height, Pix** pix, boxes->push_back(box); } } - return im_factor; + if (scale_factor != NULL) *scale_factor = im_factor; + return pix; } int ImageData::MemoryUsed() const { @@ -266,19 +287,20 @@ void ImageData::Display() const { // Draw the boxes. win->Pen(ScrollView::RED); win->Brush(ScrollView::NONE); - win->TextAttributes("Arial", kTextSize, false, false, false); - for (int b = 0; b < boxes_.size(); ++b) { - boxes_[b].plot(win); - win->Text(boxes_[b].left(), height + kTextSize, box_texts_[b].string()); - TBOX scaled(boxes_[b]); - scaled.scale(256.0 / height); - scaled.plot(win); + int text_size = kTextSize; + if (!boxes_.empty() && boxes_[0].height() * 2 < text_size) + text_size = boxes_[0].height() * 2; + win->TextAttributes("Arial", text_size, false, false, false); + if (!boxes_.empty()) { + for (int b = 0; b < boxes_.size(); ++b) { + boxes_[b].plot(win); + win->Text(boxes_[b].left(), height + kTextSize, box_texts_[b].string()); + } + } else { + // The full transcription. + win->Pen(ScrollView::CYAN); + win->Text(0, height + kTextSize * 2, transcription_.string()); } - // The full transcription. - win->Pen(ScrollView::CYAN); - win->Text(0, height + kTextSize * 2, transcription_.string()); - // Add the features. - win->Pen(ScrollView::GREEN); win->Update(); window_wait(win); #endif @@ -340,27 +362,51 @@ bool ImageData::AddBoxes(const char* box_text) { return false; } +// Thread function to call ReCachePages. +void* ReCachePagesFunc(void* data) { + DocumentData* document_data = reinterpret_cast(data); + document_data->ReCachePages(); + return NULL; +} + DocumentData::DocumentData(const STRING& name) - : document_name_(name), pages_offset_(0), total_pages_(0), - memory_used_(0), max_memory_(0), reader_(NULL) {} + : document_name_(name), + pages_offset_(-1), + total_pages_(-1), + memory_used_(0), + max_memory_(0), + reader_(NULL) {} -DocumentData::~DocumentData() {} +DocumentData::~DocumentData() { + SVAutoLock lock_p(&pages_mutex_); + SVAutoLock lock_g(&general_mutex_); +} // Reads all the pages in the given lstmf filename to the cache. The reader // is used to read the file. bool DocumentData::LoadDocument(const char* filename, const char* lang, int start_page, inT64 max_memory, FileReader reader) { + SetDocument(filename, lang, max_memory, reader); + pages_offset_ = start_page; + return ReCachePages(); +} + +// Sets up the document, without actually loading it. +void DocumentData::SetDocument(const char* filename, const char* lang, + inT64 max_memory, FileReader reader) { + SVAutoLock lock_p(&pages_mutex_); + SVAutoLock lock(&general_mutex_); document_name_ = filename; lang_ = lang; - pages_offset_ = start_page; + pages_offset_ = -1; max_memory_ = max_memory; reader_ = reader; - return ReCachePages(); } // Writes all the pages to the given filename. Returns false on error. bool DocumentData::SaveDocument(const char* filename, FileWriter writer) { + SVAutoLock lock(&pages_mutex_); TFile fp; fp.OpenWrite(NULL); if (!pages_.Serialize(&fp) || !fp.CloseWrite(filename, writer)) { @@ -370,112 +416,174 @@ bool DocumentData::SaveDocument(const char* filename, FileWriter writer) { return true; } bool DocumentData::SaveToBuffer(GenericVector* buffer) { + SVAutoLock lock(&pages_mutex_); TFile fp; fp.OpenWrite(buffer); return pages_.Serialize(&fp); } +// Adds the given page data to this document, counting up memory. +void DocumentData::AddPageToDocument(ImageData* page) { + SVAutoLock lock(&pages_mutex_); + pages_.push_back(page); + set_memory_used(memory_used() + page->MemoryUsed()); +} + +// If the given index is not currently loaded, loads it using a separate +// thread. +void DocumentData::LoadPageInBackground(int index) { + ImageData* page = NULL; + if (IsPageAvailable(index, &page)) return; + SVAutoLock lock(&pages_mutex_); + if (pages_offset_ == index) return; + pages_offset_ = index; + pages_.clear(); + #ifndef GRAPHICS_DISABLED + SVSync::StartThread(ReCachePagesFunc, this); + #endif // GRAPHICS_DISABLED +} + // Returns a pointer to the page with the given index, modulo the total -// number of pages, recaching if needed. +// number of pages. Blocks until the background load is completed. const ImageData* DocumentData::GetPage(int index) { - index = Modulo(index, total_pages_); - if (index < pages_offset_ || index >= pages_offset_ + pages_.size()) { - pages_offset_ = index; - if (!ReCachePages()) return NULL; + ImageData* page = NULL; + while (!IsPageAvailable(index, &page)) { + // If there is no background load scheduled, schedule one now. + pages_mutex_.Lock(); + bool needs_loading = pages_offset_ != index; + pages_mutex_.Unlock(); + if (needs_loading) LoadPageInBackground(index); + // We can't directly load the page, or the background load will delete it + // while the caller is using it, so give it a chance to work. +#if __cplusplus > 199711L + std::this_thread::sleep_for(std::chrono::seconds(1)); +#elif _WIN32 // MSVS + Sleep(1000); +#else + sleep(1); +#endif + } + return page; +} + +// Returns true if the requested page is available, and provides a pointer, +// which may be NULL if the document is empty. May block, even though it +// doesn't guarantee to return true. +bool DocumentData::IsPageAvailable(int index, ImageData** page) { + SVAutoLock lock(&pages_mutex_); + int num_pages = NumPages(); + if (num_pages == 0 || index < 0) { + *page = NULL; // Empty Document. + return true; + } + if (num_pages > 0) { + index = Modulo(index, num_pages); + if (pages_offset_ <= index && index < pages_offset_ + pages_.size()) { + *page = pages_[index - pages_offset_]; // Page is available already. + return true; + } } - return pages_[index - pages_offset_]; + return false; +} + +// Removes all pages from memory and frees the memory, but does not forget +// the document metadata. +inT64 DocumentData::UnCache() { + SVAutoLock lock(&pages_mutex_); + inT64 memory_saved = memory_used(); + pages_.clear(); + pages_offset_ = -1; + set_total_pages(-1); + set_memory_used(0); + tprintf("Unloaded document %s, saving %d memory\n", document_name_.string(), + memory_saved); + return memory_saved; } -// Loads as many pages can fit in max_memory_ starting at index pages_offset_. +// Locks the pages_mutex_ and Loads as many pages can fit in max_memory_ +// starting at index pages_offset_. bool DocumentData::ReCachePages() { + SVAutoLock lock(&pages_mutex_); // Read the file. + set_total_pages(0); + set_memory_used(0); + int loaded_pages = 0; + pages_.truncate(0); TFile fp; - if (!fp.Open(document_name_, reader_)) return false; - memory_used_ = 0; - if (!pages_.DeSerialize(false, &fp)) { - tprintf("Deserialize failed: %s\n", document_name_.string()); - pages_.truncate(0); + if (!fp.Open(document_name_, reader_) || + !PointerVector::DeSerializeSize(false, &fp, &loaded_pages) || + loaded_pages <= 0) { + tprintf("Deserialize header failed: %s\n", document_name_.string()); return false; } - total_pages_ = pages_.size(); - pages_offset_ %= total_pages_; - // Delete pages before the first one we want, and relocate the rest. + pages_offset_ %= loaded_pages; + // Skip pages before the first one we want, and load the rest until max + // memory and skip the rest after that. int page; - for (page = 0; page < pages_.size(); ++page) { - if (page < pages_offset_) { - delete pages_[page]; - pages_[page] = NULL; + for (page = 0; page < loaded_pages; ++page) { + if (page < pages_offset_ || + (max_memory_ > 0 && memory_used() > max_memory_)) { + if (!PointerVector::DeSerializeSkip(false, &fp)) break; } else { - ImageData* image_data = pages_[page]; - if (max_memory_ > 0 && page > pages_offset_ && - memory_used_ + image_data->MemoryUsed() > max_memory_) - break; // Don't go over memory quota unless the first image. + if (!pages_.DeSerializeElement(false, &fp)) break; + ImageData* image_data = pages_.back(); if (image_data->imagefilename().length() == 0) { image_data->set_imagefilename(document_name_); image_data->set_page_number(page); } image_data->set_language(lang_); - memory_used_ += image_data->MemoryUsed(); - if (pages_offset_ != 0) { - pages_[page - pages_offset_] = image_data; - pages_[page] = NULL; - } + set_memory_used(memory_used() + image_data->MemoryUsed()); } } - pages_.truncate(page - pages_offset_); - tprintf("Loaded %d/%d pages (%d-%d) of document %s\n", - pages_.size(), total_pages_, pages_offset_, - pages_offset_ + pages_.size(), document_name_.string()); + if (page < loaded_pages) { + tprintf("Deserialize failed: %s read %d/%d pages\n", + document_name_.string(), page, loaded_pages); + pages_.truncate(0); + } else { + tprintf("Loaded %d/%d pages (%d-%d) of document %s\n", pages_.size(), + loaded_pages, pages_offset_, pages_offset_ + pages_.size(), + document_name_.string()); + } + set_total_pages(loaded_pages); return !pages_.empty(); } -// Adds the given page data to this document, counting up memory. -void DocumentData::AddPageToDocument(ImageData* page) { - pages_.push_back(page); - memory_used_ += page->MemoryUsed(); -} - // A collection of DocumentData that knows roughly how much memory it is using. DocumentCache::DocumentCache(inT64 max_memory) - : total_pages_(0), memory_used_(0), max_memory_(max_memory) {} + : num_pages_per_doc_(0), max_memory_(max_memory) {} DocumentCache::~DocumentCache() {} // Adds all the documents in the list of filenames, counting memory. // The reader is used to read the files. bool DocumentCache::LoadDocuments(const GenericVector& filenames, - const char* lang, FileReader reader) { - inT64 fair_share_memory = max_memory_ / filenames.size(); + const char* lang, + CachingStrategy cache_strategy, + FileReader reader) { + cache_strategy_ = cache_strategy; + inT64 fair_share_memory = 0; + // In the round-robin case, each DocumentData handles restricting its content + // to its fair share of memory. In the sequential case, DocumentCache + // determines which DocumentDatas are held entirely in memory. + if (cache_strategy_ == CS_ROUND_ROBIN) + fair_share_memory = max_memory_ / filenames.size(); for (int arg = 0; arg < filenames.size(); ++arg) { STRING filename = filenames[arg]; DocumentData* document = new DocumentData(filename); - if (document->LoadDocument(filename.string(), lang, 0, - fair_share_memory, reader)) { - AddToCache(document); - } else { - tprintf("Failed to load image %s!\n", filename.string()); - delete document; - } + document->SetDocument(filename.string(), lang, fair_share_memory, reader); + AddToCache(document); } - tprintf("Loaded %d pages, total %gMB\n", - total_pages_, memory_used_ / 1048576.0); - return total_pages_ > 0; + if (!documents_.empty()) { + // Try to get the first page now to verify the list of filenames. + if (GetPageBySerial(0) != NULL) return true; + tprintf("Load of page 0 failed!\n"); + } + return false; } -// Adds document to the cache, throwing out other documents if needed. +// Adds document to the cache. bool DocumentCache::AddToCache(DocumentData* data) { inT64 new_memory = data->memory_used(); - memory_used_ += new_memory; documents_.push_back(data); - total_pages_ += data->NumPages(); - // Delete the first item in the array, and other pages of the same name - // while memory is full. - while (memory_used_ >= max_memory_ && max_memory_ > 0) { - tprintf("Memory used=%lld vs max=%lld, discarding doc of size %lld\n", - memory_used_ , max_memory_, documents_[0]->memory_used()); - memory_used_ -= documents_[0]->memory_used(); - total_pages_ -= documents_[0]->NumPages(); - documents_.remove(0); - } return true; } @@ -488,11 +596,104 @@ DocumentData* DocumentCache::FindDocument(const STRING& document_name) const { return NULL; } +// Returns the total number of pages in an epoch. For CS_ROUND_ROBIN cache +// strategy, could take a long time. +int DocumentCache::TotalPages() { + if (cache_strategy_ == CS_SEQUENTIAL) { + // In sequential mode, we assume each doc has the same number of pages + // whether it is true or not. + if (num_pages_per_doc_ == 0) GetPageSequential(0); + return num_pages_per_doc_ * documents_.size(); + } + int total_pages = 0; + int num_docs = documents_.size(); + for (int d = 0; d < num_docs; ++d) { + // We have to load a page to make NumPages() valid. + documents_[d]->GetPage(0); + total_pages += documents_[d]->NumPages(); + } + return total_pages; +} + // Returns a page by serial number, selecting them in a round-robin fashion -// from all the documents. -const ImageData* DocumentCache::GetPageBySerial(int serial) { - int document_index = serial % documents_.size(); - return documents_[document_index]->GetPage(serial / documents_.size()); +// from all the documents. Highly disk-intensive, but doesn't need samples +// to be shuffled between files to begin with. +const ImageData* DocumentCache::GetPageRoundRobin(int serial) { + int num_docs = documents_.size(); + int doc_index = serial % num_docs; + const ImageData* doc = documents_[doc_index]->GetPage(serial / num_docs); + for (int offset = 1; offset <= kMaxReadAhead && offset < num_docs; ++offset) { + doc_index = (serial + offset) % num_docs; + int page = (serial + offset) / num_docs; + documents_[doc_index]->LoadPageInBackground(page); + } + return doc; +} + +// Returns a page by serial number, selecting them in sequence from each file. +// Requires the samples to be shuffled between the files to give a random or +// uniform distribution of data. Less disk-intensive than GetPageRoundRobin. +const ImageData* DocumentCache::GetPageSequential(int serial) { + int num_docs = documents_.size(); + ASSERT_HOST(num_docs > 0); + if (num_pages_per_doc_ == 0) { + // Use the pages in the first doc as the number of pages in each doc. + documents_[0]->GetPage(0); + num_pages_per_doc_ = documents_[0]->NumPages(); + if (num_pages_per_doc_ == 0) { + tprintf("First document cannot be empty!!\n"); + ASSERT_HOST(num_pages_per_doc_ > 0); + } + // Get rid of zero now if we don't need it. + if (serial / num_pages_per_doc_ % num_docs > 0) documents_[0]->UnCache(); + } + int doc_index = serial / num_pages_per_doc_ % num_docs; + const ImageData* doc = + documents_[doc_index]->GetPage(serial % num_pages_per_doc_); + // Count up total memory. Background loading makes it more complicated to + // keep a running count. + inT64 total_memory = 0; + for (int d = 0; d < num_docs; ++d) { + total_memory += documents_[d]->memory_used(); + } + if (total_memory >= max_memory_) { + // Find something to un-cache. + // If there are more than 3 in front, then serial is from the back reader + // of a pair of readers. If we un-cache from in-front-2 to 2-ahead, then + // we create a hole between them and then un-caching the backmost occupied + // will work for both. + int num_in_front = CountNeighbourDocs(doc_index, 1); + for (int offset = num_in_front - 2; + offset > 1 && total_memory >= max_memory_; --offset) { + int next_index = (doc_index + offset) % num_docs; + total_memory -= documents_[next_index]->UnCache(); + } + // If that didn't work, the best solution is to un-cache from the back. If + // we take away the document that a 2nd reader is using, it will put it + // back and make a hole between. + int num_behind = CountNeighbourDocs(doc_index, -1); + for (int offset = num_behind; offset < 0 && total_memory >= max_memory_; + ++offset) { + int next_index = (doc_index + offset + num_docs) % num_docs; + total_memory -= documents_[next_index]->UnCache(); + } + } + int next_index = (doc_index + 1) % num_docs; + if (!documents_[next_index]->IsCached() && total_memory < max_memory_) { + documents_[next_index]->LoadPageInBackground(0); + } + return doc; +} + +// Helper counts the number of adjacent cached neighbours of index looking in +// direction dir, ie index+dir, index+2*dir etc. +int DocumentCache::CountNeighbourDocs(int index, int dir) { + int num_docs = documents_.size(); + for (int offset = dir; abs(offset) < num_docs; offset += dir) { + int offset_index = (index + offset + num_docs) % num_docs; + if (!documents_[offset_index]->IsCached()) return offset - dir; + } + return num_docs; } } // namespace tesseract. diff --git a/ccstruct/imagedata.h b/ccstruct/imagedata.h index 6321f121b1..ae6722934e 100644 --- a/ccstruct/imagedata.h +++ b/ccstruct/imagedata.h @@ -25,6 +25,7 @@ #include "normalis.h" #include "rect.h" #include "strngs.h" +#include "svutil.h" struct Pix; @@ -34,8 +35,22 @@ namespace tesseract { const int kFeaturePadding = 2; // Number of pixels to pad around text boxes. const int kImagePadding = 4; -// Number of training images to combine into a mini-batch for training. -const int kNumPagesPerMiniBatch = 100; + +// Enum to determine the caching and data sequencing strategy. +enum CachingStrategy { + // Reads all of one file before moving on to the next. Requires samples to be + // shuffled across files. Uses the count of samples in the first file as + // the count in all the files to achieve high-speed random access. As a + // consequence, if subsequent files are smaller, they get entries used more + // than once, and if subsequent files are larger, some entries are not used. + // Best for larger data sets that don't fit in memory. + CS_SEQUENTIAL, + // Reads one sample from each file in rotation. Does not require shuffled + // samples, but is extremely disk-intensive. Samples in smaller files also + // get used more often than samples in larger files. + // Best for smaller data sets that mostly fit in memory. + CS_ROUND_ROBIN, +}; class WordFeature { public: @@ -103,6 +118,8 @@ class ImageData { // Reads from the given file. Returns false in case of error. // If swap is true, assumes a big/little-endian swap is needed. bool DeSerialize(bool swap, TFile* fp); + // As DeSerialize, but only seeks past the data - hence a static method. + static bool SkipDeSerialize(bool swap, tesseract::TFile* fp); // Other accessors. const STRING& imagefilename() const { @@ -145,11 +162,12 @@ class ImageData { // Gets anything and everything with a non-NULL pointer, prescaled to a // given target_height (if 0, then the original image height), and aligned. // Also returns (if not NULL) the width and height of the scaled image. - // The return value is the scale factor that was applied to the image to - // achieve the target_height. - float PreScale(int target_height, Pix** pix, - int* scaled_width, int* scaled_height, - GenericVector* boxes) const; + // The return value is the scaled Pix, which must be pixDestroyed after use, + // and scale_factor (if not NULL) is set to the scale factor that was applied + // to the image to achieve the target_height. + Pix* PreScale(int target_height, int max_height, float* scale_factor, + int* scaled_width, int* scaled_height, + GenericVector* boxes) const; int MemoryUsed() const; @@ -184,6 +202,8 @@ class ImageData { // A collection of ImageData that knows roughly how much memory it is using. class DocumentData { + friend void* ReCachePagesFunc(void* data); + public: explicit DocumentData(const STRING& name); ~DocumentData(); @@ -192,6 +212,9 @@ class DocumentData { // is used to read the file. bool LoadDocument(const char* filename, const char* lang, int start_page, inT64 max_memory, FileReader reader); + // Sets up the document, without actually loading it. + void SetDocument(const char* filename, const char* lang, inT64 max_memory, + FileReader reader); // Writes all the pages to the given filename. Returns false on error. bool SaveDocument(const char* filename, FileWriter writer); bool SaveToBuffer(GenericVector* buffer); @@ -200,26 +223,62 @@ class DocumentData { void AddPageToDocument(ImageData* page); const STRING& document_name() const { + SVAutoLock lock(&general_mutex_); return document_name_; } int NumPages() const { + SVAutoLock lock(&general_mutex_); return total_pages_; } inT64 memory_used() const { + SVAutoLock lock(&general_mutex_); return memory_used_; } + // If the given index is not currently loaded, loads it using a separate + // thread. Note: there are 4 cases: + // Document uncached: IsCached() returns false, total_pages_ < 0. + // Required page is available: IsPageAvailable returns true. In this case, + // total_pages_ > 0 and + // pages_offset_ <= index%total_pages_ <= pages_offset_+pages_.size() + // Pages are loaded, but the required one is not. + // The requested page is being loaded by LoadPageInBackground. In this case, + // index == pages_offset_. Once the loading starts, the pages lock is held + // until it completes, at which point IsPageAvailable will unblock and return + // true. + void LoadPageInBackground(int index); // Returns a pointer to the page with the given index, modulo the total - // number of pages, recaching if needed. + // number of pages. Blocks until the background load is completed. const ImageData* GetPage(int index); + // Returns true if the requested page is available, and provides a pointer, + // which may be NULL if the document is empty. May block, even though it + // doesn't guarantee to return true. + bool IsPageAvailable(int index, ImageData** page); // Takes ownership of the given page index. The page is made NULL in *this. ImageData* TakePage(int index) { + SVAutoLock lock(&pages_mutex_); ImageData* page = pages_[index]; pages_[index] = NULL; return page; } + // Returns true if the document is currently loaded or in the process of + // loading. + bool IsCached() const { return NumPages() >= 0; } + // Removes all pages from memory and frees the memory, but does not forget + // the document metadata. Returns the memory saved. + inT64 UnCache(); private: - // Loads as many pages can fit in max_memory_ starting at index pages_offset_. + // Sets the value of total_pages_ behind a mutex. + void set_total_pages(int total) { + SVAutoLock lock(&general_mutex_); + total_pages_ = total; + } + void set_memory_used(inT64 memory_used) { + SVAutoLock lock(&general_mutex_); + memory_used_ = memory_used; + } + // Locks the pages_mutex_ and Loads as many pages can fit in max_memory_ + // starting at index pages_offset_. bool ReCachePages(); private: @@ -239,43 +298,77 @@ class DocumentData { inT64 max_memory_; // Saved reader from LoadDocument to allow re-caching. FileReader reader_; + // Mutex that protects pages_ and pages_offset_ against multiple parallel + // loads, and provides a wait for page. + SVMutex pages_mutex_; + // Mutex that protects other data members that callers want to access without + // waiting for a load operation. + mutable SVMutex general_mutex_; }; // A collection of DocumentData that knows roughly how much memory it is using. +// Note that while it supports background read-ahead, it assumes that a single +// thread is accessing documents, ie it is not safe for multiple threads to +// access different documents in parallel, as one may de-cache the other's +// content. class DocumentCache { public: explicit DocumentCache(inT64 max_memory); ~DocumentCache(); + // Deletes all existing documents from the cache. + void Clear() { + documents_.clear(); + num_pages_per_doc_ = 0; + } // Adds all the documents in the list of filenames, counting memory. // The reader is used to read the files. bool LoadDocuments(const GenericVector& filenames, const char* lang, - FileReader reader); + CachingStrategy cache_strategy, FileReader reader); - // Adds document to the cache, throwing out other documents if needed. + // Adds document to the cache. bool AddToCache(DocumentData* data); // Finds and returns a document by name. DocumentData* FindDocument(const STRING& document_name) const; - // Returns a page by serial number, selecting them in a round-robin fashion - // from all the documents. - const ImageData* GetPageBySerial(int serial); + // Returns a page by serial number using the current cache_strategy_ to + // determine the mapping from serial number to page. + const ImageData* GetPageBySerial(int serial) { + if (cache_strategy_ == CS_SEQUENTIAL) + return GetPageSequential(serial); + else + return GetPageRoundRobin(serial); + } const PointerVector& documents() const { return documents_; } - int total_pages() const { - return total_pages_; - } + // Returns the total number of pages in an epoch. For CS_ROUND_ROBIN cache + // strategy, could take a long time. + int TotalPages(); private: + // Returns a page by serial number, selecting them in a round-robin fashion + // from all the documents. Highly disk-intensive, but doesn't need samples + // to be shuffled between files to begin with. + const ImageData* GetPageRoundRobin(int serial); + // Returns a page by serial number, selecting them in sequence from each file. + // Requires the samples to be shuffled between the files to give a random or + // uniform distribution of data. Less disk-intensive than GetPageRoundRobin. + const ImageData* GetPageSequential(int serial); + + // Helper counts the number of adjacent cached neighbour documents_ of index + // looking in direction dir, ie index+dir, index+2*dir etc. + int CountNeighbourDocs(int index, int dir); + // A group of pages that corresponds in some loose way to a document. PointerVector documents_; - // Total of all pages. - int total_pages_; - // Total of all memory used by the cache. - inT64 memory_used_; + // Strategy to use for caching and serializing data samples. + CachingStrategy cache_strategy_; + // Number of pages in the first document, used as a divisor in + // GetPageSequential to determine the document index. + int num_pages_per_doc_; // Max memory allowed in this cache. inT64 max_memory_; }; diff --git a/ccstruct/matrix.h b/ccstruct/matrix.h index e13ef31899..4b5b242a43 100644 --- a/ccstruct/matrix.h +++ b/ccstruct/matrix.h @@ -1,8 +1,12 @@ /* -*-C-*- ****************************************************************************** + * File: matrix.h (Formerly matrix.h) + * Description: Generic 2-d array/matrix and banded triangular matrix class. + * Author: Ray Smith + * TODO(rays) Separate from ratings matrix, which it also contains: * - * File: matrix.h (Formerly matrix.h) - * Description: Ratings matrix code. (Used by associator) + * Descrition: Ratings matrix class (specialization of banded matrix). + * Segmentation search matrix of lists of BLOB_CHOICE. * Author: Mark Seaman, OCR Technology * Created: Wed May 16 13:22:06 1990 * Modified: Tue Mar 19 16:00:20 1991 (Mark Seaman) marks@hpgrlt @@ -25,9 +29,13 @@ #ifndef TESSERACT_CCSTRUCT_MATRIX_H__ #define TESSERACT_CCSTRUCT_MATRIX_H__ +#include #include "kdpair.h" +#include "points.h" +#include "serialis.h" #include "unicharset.h" +class BLOB_CHOICE; class BLOB_CHOICE_LIST; #define NOT_CLASSIFIED reinterpret_cast(0) @@ -44,34 +52,60 @@ class GENERIC_2D_ARRAY { // either pass the memory in, or allocate after by calling Resize(). GENERIC_2D_ARRAY(int dim1, int dim2, const T& empty, T* array) : empty_(empty), dim1_(dim1), dim2_(dim2), array_(array) { + size_allocated_ = dim1 * dim2; } // Original constructor for a full rectangular matrix DOES allocate memory // and initialize it to empty. GENERIC_2D_ARRAY(int dim1, int dim2, const T& empty) : empty_(empty), dim1_(dim1), dim2_(dim2) { - array_ = new T[dim1_ * dim2_]; - for (int x = 0; x < dim1_; x++) - for (int y = 0; y < dim2_; y++) - this->put(x, y, empty_); + int new_size = dim1 * dim2; + array_ = new T[new_size]; + size_allocated_ = new_size; + for (int i = 0; i < size_allocated_; ++i) + array_[i] = empty_; + } + // Default constructor for array allocation. Use Resize to set the size. + GENERIC_2D_ARRAY() + : array_(NULL), empty_(static_cast(0)), dim1_(0), dim2_(0), + size_allocated_(0) { + } + GENERIC_2D_ARRAY(const GENERIC_2D_ARRAY& src) + : array_(NULL), empty_(static_cast(0)), dim1_(0), dim2_(0), + size_allocated_(0) { + *this = src; } virtual ~GENERIC_2D_ARRAY() { delete[] array_; } + void operator=(const GENERIC_2D_ARRAY& src) { + ResizeNoInit(src.dim1(), src.dim2()); + memcpy(array_, src.array_, num_elements() * sizeof(array_[0])); + } + + // Reallocate the array to the given size. Does not keep old data, but does + // not initialize the array either. + void ResizeNoInit(int size1, int size2) { + int new_size = size1 * size2; + if (new_size > size_allocated_) { + delete [] array_; + array_ = new T[new_size]; + size_allocated_ = new_size; + } + dim1_ = size1; + dim2_ = size2; + } + // Reallocate the array to the given size. Does not keep old data. void Resize(int size1, int size2, const T& empty) { empty_ = empty; - if (size1 != dim1_ || size2 != dim2_) { - dim1_ = size1; - dim2_ = size2; - delete [] array_; - array_ = new T[dim1_ * dim2_]; - } + ResizeNoInit(size1, size2); Clear(); } // Reallocate the array to the given size, keeping old data. void ResizeWithCopy(int size1, int size2) { if (size1 != dim1_ || size2 != dim2_) { - T* new_array = new T[size1 * size2]; + int new_size = size1 * size2; + T* new_array = new T[new_size]; for (int col = 0; col < size1; ++col) { for (int row = 0; row < size2; ++row) { int old_index = col * dim2() + row; @@ -87,6 +121,7 @@ class GENERIC_2D_ARRAY { array_ = new_array; dim1_ = size1; dim2_ = size2; + size_allocated_ = new_size; } } @@ -106,9 +141,16 @@ class GENERIC_2D_ARRAY { if (fwrite(array_, sizeof(*array_), size, fp) != size) return false; return true; } + bool Serialize(tesseract::TFile* fp) const { + if (!SerializeSize(fp)) return false; + if (fp->FWrite(&empty_, sizeof(empty_), 1) != 1) return false; + int size = num_elements(); + if (fp->FWrite(array_, sizeof(*array_), size) != size) return false; + return true; + } // Reads from the given file. Returns false in case of error. - // Only works with bitwise-serializeable typ + // Only works with bitwise-serializeable types! // If swap is true, assumes a big/little-endian swap is needed. bool DeSerialize(bool swap, FILE* fp) { if (!DeSerializeSize(swap, fp)) return false; @@ -122,6 +164,18 @@ class GENERIC_2D_ARRAY { } return true; } + bool DeSerialize(bool swap, tesseract::TFile* fp) { + if (!DeSerializeSize(swap, fp)) return false; + if (fp->FRead(&empty_, sizeof(empty_), 1) != 1) return false; + if (swap) ReverseN(&empty_, sizeof(empty_)); + int size = num_elements(); + if (fp->FRead(array_, sizeof(*array_), size) != size) return false; + if (swap) { + for (int i = 0; i < size; ++i) + ReverseN(&array_[i], sizeof(array_[i])); + } + return true; + } // Writes to the given file. Returns false in case of error. // Assumes a T::Serialize(FILE*) const function. @@ -163,11 +217,17 @@ class GENERIC_2D_ARRAY { } // Put a list element into the matrix at a specific location. + void put(ICOORD pos, const T& thing) { + array_[this->index(pos.x(), pos.y())] = thing; + } void put(int column, int row, const T& thing) { array_[this->index(column, row)] = thing; } // Get the item at a specified location from the matrix. + T get(ICOORD pos) const { + return array_[this->index(pos.x(), pos.y())]; + } T get(int column, int row) const { return array_[this->index(column, row)]; } @@ -187,6 +247,207 @@ class GENERIC_2D_ARRAY { return &array_[this->index(column, 0)]; } + // Adds addend to *this, element-by-element. + void operator+=(const GENERIC_2D_ARRAY& addend) { + if (dim2_ == addend.dim2_) { + // Faster if equal size in the major dimension. + int size = MIN(num_elements(), addend.num_elements()); + for (int i = 0; i < size; ++i) { + array_[i] += addend.array_[i]; + } + } else { + for (int x = 0; x < dim1_; x++) { + for (int y = 0; y < dim2_; y++) { + (*this)(x, y) += addend(x, y); + } + } + } + } + // Subtracts minuend from *this, element-by-element. + void operator-=(const GENERIC_2D_ARRAY& minuend) { + if (dim2_ == minuend.dim2_) { + // Faster if equal size in the major dimension. + int size = MIN(num_elements(), minuend.num_elements()); + for (int i = 0; i < size; ++i) { + array_[i] -= minuend.array_[i]; + } + } else { + for (int x = 0; x < dim1_; x++) { + for (int y = 0; y < dim2_; y++) { + (*this)(x, y) -= minuend(x, y); + } + } + } + } + // Adds addend to all elements. + void operator+=(const T& addend) { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + array_[i] += addend; + } + } + // Multiplies *this by factor, element-by-element. + void operator*=(const T& factor) { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + array_[i] *= factor; + } + } + // Clips *this to the given range. + void Clip(const T& rangemin, const T& rangemax) { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + array_[i] = ClipToRange(array_[i], rangemin, rangemax); + } + } + // Returns true if all elements of *this are within the given range. + // Only uses operator< + bool WithinBounds(const T& rangemin, const T& rangemax) const { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + const T& value = array_[i]; + if (value < rangemin || rangemax < value) + return false; + } + return true; + } + // Normalize the whole array. + double Normalize() { + int size = num_elements(); + if (size <= 0) return 0.0; + // Compute the mean. + double mean = 0.0; + for (int i = 0; i < size; ++i) { + mean += array_[i]; + } + mean /= size; + // Subtract the mean and compute the standard deviation. + double sd = 0.0; + for (int i = 0; i < size; ++i) { + double normed = array_[i] - mean; + array_[i] = normed; + sd += normed * normed; + } + sd = sqrt(sd / size); + if (sd > 0.0) { + // Divide by the sd. + for (int i = 0; i < size; ++i) { + array_[i] /= sd; + } + } + return sd; + } + + // Returns the maximum value of the array. + T Max() const { + int size = num_elements(); + if (size <= 0) return empty_; + // Compute the max. + T max_value = array_[0]; + for (int i = 1; i < size; ++i) { + const T& value = array_[i]; + if (value > max_value) max_value = value; + } + return max_value; + } + + // Returns the maximum absolute value of the array. + T MaxAbs() const { + int size = num_elements(); + if (size <= 0) return empty_; + // Compute the max. + T max_abs = static_cast(0); + for (int i = 0; i < size; ++i) { + T value = static_cast(fabs(array_[i])); + if (value > max_abs) max_abs = value; + } + return max_abs; + } + + // Accumulates the element-wise sums of squares of src into *this. + void SumSquares(const GENERIC_2D_ARRAY& src) { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + array_[i] += src.array_[i] * src.array_[i]; + } + } + + // Scales each element using the ada-grad algorithm, ie array_[i] by + // sqrt(num_samples/max(1,sqsum[i])). + void AdaGradScaling(const GENERIC_2D_ARRAY& sqsum, int num_samples) { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + array_[i] *= sqrt(num_samples / MAX(1.0, sqsum.array_[i])); + } + } + + void AssertFinite() const { + int size = num_elements(); + for (int i = 0; i < size; ++i) { + ASSERT_HOST(isfinite(array_[i])); + } + } + + // REGARDLESS OF THE CURRENT DIMENSIONS, treats the data as a + // num_dims-dimensional array/tensor with dimensions given by dims, (ordered + // from most significant to least significant, the same as standard C arrays) + // and moves src_dim to dest_dim, with the initial dest_dim and any dimensions + // in between shifted towards the hole left by src_dim. Example: + // Current data content: array_=[0, 1, 2, ....119] + // perhaps *this may be of dim[40, 3], with values [[0, 1, 2][3, 4, 5]... + // but the current dimensions are irrelevant. + // num_dims = 4, dims=[5, 4, 3, 2] + // src_dim=3, dest_dim=1 + // tensor=[[[[0, 1][2, 3][4, 5]] + // [[6, 7][8, 9][10, 11]] + // [[12, 13][14, 15][16, 17]] + // [[18, 19][20, 21][22, 23]]] + // [[[24, 25]... + // output dims =[5, 2, 4, 3] + // output tensor=[[[[0, 2, 4][6, 8, 10][12, 14, 16][18, 20, 22]] + // [[1, 3, 5][7, 9, 11][13, 15, 17][19, 21, 23]]] + // [[[24, 26, 28]... + // which is stored in the array_ as: + // [0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 1, 3, 5, 7, 9, 11, 13...] + // NOTE: the 2 stored matrix dimensions are simply copied from *this. To + // change the dimensions after the transpose, use ResizeNoInit. + // Higher dimensions above 2 are strictly the responsibility of the caller. + void RotatingTranspose(const int* dims, int num_dims, int src_dim, + int dest_dim, GENERIC_2D_ARRAY* result) const { + int max_d = MAX(src_dim, dest_dim); + int min_d = MIN(src_dim, dest_dim); + // In a tensor of shape [d0, d1... min_d, ... max_d, ... dn-2, dn-1], the + // ends outside of min_d and max_d are unaffected, with [max_d +1, dn-1] + // being contiguous blocks of data that will move together, and + // [d0, min_d -1] being replicas of the transpose operation. + // num_replicas represents the large dimensions unchanged by the operation. + // move_size represents the small dimensions unchanged by the operation. + // src_step represents the stride in the src between each adjacent group + // in the destination. + int num_replicas = 1, move_size = 1, src_step = 1; + for (int d = 0; d < min_d; ++d) num_replicas *= dims[d]; + for (int d = max_d + 1; d < num_dims; ++d) move_size *= dims[d]; + for (int d = src_dim + 1; d < num_dims; ++d) src_step *= dims[d]; + if (src_dim > dest_dim) src_step *= dims[src_dim]; + // wrap_size is the size of a single replica, being the amount that is + // handled num_replicas times. + int wrap_size = move_size; + for (int d = min_d; d <= max_d; ++d) wrap_size *= dims[d]; + result->ResizeNoInit(dim1_, dim2_); + result->empty_ = empty_; + const T* src = array_; + T* dest = result->array_; + for (int replica = 0; replica < num_replicas; ++replica) { + for (int start = 0; start < src_step; start += move_size) { + for (int pos = start; pos < wrap_size; pos += src_step) { + memcpy(dest, src + pos, sizeof(*dest) * move_size); + dest += move_size; + } + } + src += wrap_size; + } + } + // Delete objects pointed to by array_[i]. void delete_matrix_pointers() { int size = num_elements(); @@ -206,6 +467,13 @@ class GENERIC_2D_ARRAY { if (fwrite(&size, sizeof(size), 1, fp) != 1) return false; return true; } + bool SerializeSize(tesseract::TFile* fp) const { + inT32 size = dim1_; + if (fp->FWrite(&size, sizeof(size), 1) != 1) return false; + size = dim2_; + if (fp->FWrite(&size, sizeof(size), 1) != 1) return false; + return true; + } // Factored helper to deserialize the size. // If swap is true, assumes a big/little-endian swap is needed. bool DeSerializeSize(bool swap, FILE* fp) { @@ -219,11 +487,26 @@ class GENERIC_2D_ARRAY { Resize(size1, size2, empty_); return true; } + bool DeSerializeSize(bool swap, tesseract::TFile* fp) { + inT32 size1, size2; + if (fp->FRead(&size1, sizeof(size1), 1) != 1) return false; + if (fp->FRead(&size2, sizeof(size2), 1) != 1) return false; + if (swap) { + ReverseN(&size1, sizeof(size1)); + ReverseN(&size2, sizeof(size2)); + } + Resize(size1, size2, empty_); + return true; + } T* array_; T empty_; // The unused cell. int dim1_; // Size of the 1st dimension in indexing functions. int dim2_; // Size of the 2nd dimension in indexing functions. + // The total size to which the array can be expanded before a realloc is + // needed. If Resize is used, memory is retained so it can be re-expanded + // without a further alloc, and this stores the allocated size. + int size_allocated_; }; // A generic class to store a banded triangular matrix with entries of type T. diff --git a/ccstruct/mod128.cpp b/ccstruct/mod128.cpp index ee4aa6c3c6..4e5f4bd325 100644 --- a/ccstruct/mod128.cpp +++ b/ccstruct/mod128.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: mod128.c (Formerly dir128.c) * Description: Code to convert a DIR128 to an ICOORD. - * Author: Ray Smith - * Created: Tue Oct 22 11:56:09 BST 1991 + * Author: Ray Smith + * Created: Tue Oct 22 11:56:09 BST 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -86,16 +86,3 @@ DIR128::DIR128( //from fcoord while (high - low > 1); dir = low; } - - -/********************************************************************** - * dir_to_gradient - * - * Convert a direction to a vector. - **********************************************************************/ - -#if 0 // code is buggy for negative dir and unused -ICOORD DIR128::vector() const { //convert to vector - return dirtab[dir]; //easy really -} -#endif diff --git a/ccstruct/mod128.h b/ccstruct/mod128.h index 592264ba62..c0e71a423c 100644 --- a/ccstruct/mod128.h +++ b/ccstruct/mod128.h @@ -1,8 +1,8 @@ /********************************************************************** * File: mod128.h (Formerly dir128.h) * Description: Header for class which implements modulo arithmetic. - * Author: Ray Smith - * Created: Tue Mar 26 17:48:13 GMT 1991 + * Author: Ray Smith + * Created: Tue Mar 26 17:48:13 GMT 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -77,7 +77,6 @@ class DLLSYM DIR128 inT8 get_dir() const { //access function return dir; } - ICOORD vector() const; //turn to vector private: inT8 dir; //a direction diff --git a/ccstruct/otsuthr.cpp b/ccstruct/otsuthr.cpp index 071b0d2345..8822ce2943 100644 --- a/ccstruct/otsuthr.cpp +++ b/ccstruct/otsuthr.cpp @@ -51,23 +51,16 @@ int OtsuThreshold(Pix* src_pix, int left, int top, int width, int height, // only use opencl if compiled w/ OpenCL and selected device is opencl #ifdef USE_OPENCL - // all of channel 0 then all of channel 1... - int *histogramAllChannels = new int[kHistogramSize * num_channels]; + // all of channel 0 then all of channel 1... + int* histogramAllChannels = new int[kHistogramSize * num_channels]; - // Calculate Histogram on GPU - OpenclDevice od; - if (od.selectedDeviceIsOpenCL() && - (num_channels == 1 || num_channels == 4) && top == 0 && left == 0 ) { - od.HistogramRectOCL( - (const unsigned char*)pixGetData(src_pix), - num_channels, - pixGetWpl(src_pix) * 4, - left, - top, - width, - height, - kHistogramSize, - histogramAllChannels); + // Calculate Histogram on GPU + OpenclDevice od; + if (od.selectedDeviceIsOpenCL() && (num_channels == 1 || num_channels == 4) && + top == 0 && left == 0) { + od.HistogramRectOCL((unsigned char*)pixGetData(src_pix), num_channels, + pixGetWpl(src_pix) * 4, left, top, width, height, + kHistogramSize, histogramAllChannels); // Calculate Threshold from Histogram on cpu for (int ch = 0; ch < num_channels; ++ch) { @@ -143,7 +136,6 @@ int OtsuThreshold(Pix* src_pix, int left, int top, int width, int height, delete[] histogramAllChannels; #endif // USE_OPENCL - if (!any_good_hivalue) { // Use the best of the ones that were not good enough. (*hi_values)[best_hi_index] = best_hi_value; diff --git a/ccstruct/pageres.cpp b/ccstruct/pageres.cpp index b66e5636ff..32635365e4 100644 --- a/ccstruct/pageres.cpp +++ b/ccstruct/pageres.cpp @@ -882,17 +882,17 @@ void WERD_RES::FakeClassifyWord(int blob_count, BLOB_CHOICE** choices) { choice_it.add_after_then_move(choices[c]); ratings->put(c, c, choice_list); } - FakeWordFromRatings(); + FakeWordFromRatings(TOP_CHOICE_PERM); reject_map.initialise(blob_count); done = true; } // Creates a WERD_CHOICE for the word using the top choices from the leading // diagonal of the ratings matrix. -void WERD_RES::FakeWordFromRatings() { +void WERD_RES::FakeWordFromRatings(PermuterType permuter) { int num_blobs = ratings->dimension(); WERD_CHOICE* word_choice = new WERD_CHOICE(uch_set, num_blobs); - word_choice->set_permuter(TOP_CHOICE_PERM); + word_choice->set_permuter(permuter); for (int b = 0; b < num_blobs; ++b) { UNICHAR_ID unichar_id = UNICHAR_SPACE; float rating = MAX_INT32; @@ -1105,6 +1105,7 @@ void WERD_RES::InitNonPointers() { x_height = 0.0; caps_height = 0.0; baseline_shift = 0.0f; + space_certainty = 0.0f; guessed_x_ht = TRUE; guessed_caps_ht = TRUE; combination = FALSE; diff --git a/ccstruct/pageres.h b/ccstruct/pageres.h index 7329bc89ea..33b9f4cb35 100644 --- a/ccstruct/pageres.h +++ b/ccstruct/pageres.h @@ -1,7 +1,7 @@ /********************************************************************** * File: pageres.h (Formerly page_res.h) * Description: Results classes used by control.c - * Author: Phil Cheatle + * Author: Phil Cheatle * Created: Tue Sep 22 08:42:49 BST 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. @@ -295,6 +295,9 @@ class WERD_RES : public ELIST_LINK { float x_height; // post match estimate float caps_height; // post match estimate float baseline_shift; // post match estimate. + // Certainty score for the spaces either side of this word (LSTM mode). + // MIN this value with the actual word certainty. + float space_certainty; /* To deal with fuzzy spaces we need to be able to combine "words" to form @@ -327,7 +330,7 @@ class WERD_RES : public ELIST_LINK { } // Deep copies everything except the ratings MATRIX. // To get that use deep_copy below. - WERD_RES(const WERD_RES &source) : ELIST_LINK(source) { + WERD_RES(const WERD_RES& source) : ELIST_LINK(source) { InitPointers(); *this = source; // see operator= } @@ -590,7 +593,7 @@ class WERD_RES : public ELIST_LINK { // Creates a WERD_CHOICE for the word using the top choices from the leading // diagonal of the ratings matrix. - void FakeWordFromRatings(); + void FakeWordFromRatings(PermuterType permuter); // Copies the best_choice strings to the correct_text for adaption/training. void BestChoiceToCorrectText(); @@ -630,7 +633,7 @@ class WERD_RES : public ELIST_LINK { static WERD_RES* deep_copy(const WERD_RES* src) { WERD_RES* result = new WERD_RES(*src); // That didn't copy the ratings, but we want a copy if there is one to - // begin width. + // begin with. if (src->ratings != NULL) result->ratings = src->ratings->DeepCopy(); return result; diff --git a/ccstruct/params_training_featdef.h b/ccstruct/params_training_featdef.h index ff76480be0..6e021f0b30 100644 --- a/ccstruct/params_training_featdef.h +++ b/ccstruct/params_training_featdef.h @@ -126,7 +126,7 @@ typedef GenericVector ParamsTrainingHypothesisList; // explored on PASS1, PASS2, fix xheight pass, etc). class ParamsTrainingBundle { public: - ParamsTrainingBundle() {}; + ParamsTrainingBundle() {} // Starts a new hypothesis list. // Should be called at the beginning of a new run of the segmentation search. void StartHypothesisList() { diff --git a/ccstruct/pdblock.h b/ccstruct/pdblock.h index b64eff36d0..cf29b782cb 100644 --- a/ccstruct/pdblock.h +++ b/ccstruct/pdblock.h @@ -29,90 +29,74 @@ struct Pix; CLISTIZEH (PDBLK) ///page block -class PDBLK -{ +class PDBLK { friend class BLOCK_RECT_IT; //< block iterator - public: - ///empty constructor - PDBLK() { - hand_poly = NULL; - index_ = 0; - } - ///simple constructor - PDBLK(inT16 xmin, //< bottom left - inT16 ymin, - inT16 xmax, //< top right - inT16 ymax); - - ///set vertex lists - ///@param left list of left vertices - ///@param right list of right vertices - void set_sides(ICOORDELT_LIST *left, - ICOORDELT_LIST *right); - - ///destructor - ~PDBLK () { - if (hand_poly) delete hand_poly; - } - - POLY_BLOCK *poly_block() const { - return hand_poly; - } - ///set the poly block - void set_poly_block(POLY_BLOCK *blk) { - hand_poly = blk; - } - ///get box - void bounding_box(ICOORD &bottom_left, //bottom left - ICOORD &top_right) const { //topright - bottom_left = box.botleft (); - top_right = box.topright (); - } - ///get real box - const TBOX &bounding_box() const { - return box; - } - - int index() const { - return index_; - } - void set_index(int value) { - index_ = value; - } - - ///is pt inside block - BOOL8 contains(ICOORD pt); - - /// reposition block - void move(const ICOORD vec); // by vector - - // Returns a binary Pix mask with a 1 pixel for every pixel within the - // block. Rotates the coordinate system by rerotation prior to rendering. - // If not NULL, mask_box is filled with the position box of the returned - // mask image. - Pix *render_mask(const FCOORD &rerotation, TBOX *mask_box); - - #ifndef GRAPHICS_DISABLED - ///draw histogram - ///@param window window to draw in - ///@param serial serial number - ///@param colour colour to draw in - void plot(ScrollView* window, - inT32 serial, - ScrollView::Color colour); - #endif // GRAPHICS_DISABLED - - ///assignment - ///@param source from this - PDBLK & operator= (const PDBLK & source); - - protected: - POLY_BLOCK *hand_poly; //< weird as well - ICOORDELT_LIST leftside; //< left side vertices - ICOORDELT_LIST rightside; //< right side vertices - TBOX box; //< bounding box - int index_; //< Serial number of this block. + public: + /// empty constructor + PDBLK() { + hand_poly = NULL; + index_ = 0; + } + /// simple constructor + PDBLK(inT16 xmin, //< bottom left + inT16 ymin, + inT16 xmax, //< top right + inT16 ymax); + + /// set vertex lists + ///@param left list of left vertices + ///@param right list of right vertices + void set_sides(ICOORDELT_LIST *left, ICOORDELT_LIST *right); + + /// destructor + ~PDBLK() { delete hand_poly; } + + POLY_BLOCK *poly_block() const { return hand_poly; } + /// set the poly block + void set_poly_block(POLY_BLOCK *blk) { hand_poly = blk; } + /// get box + void bounding_box(ICOORD &bottom_left, // bottom left + ICOORD &top_right) const { // topright + bottom_left = box.botleft(); + top_right = box.topright(); + } + /// get real box + const TBOX &bounding_box() const { return box; } + + int index() const { return index_; } + void set_index(int value) { index_ = value; } + + /// is pt inside block + BOOL8 contains(ICOORD pt); + + /// reposition block + void move(const ICOORD vec); // by vector + + // Returns a binary Pix mask with a 1 pixel for every pixel within the + // block. Rotates the coordinate system by rerotation prior to rendering. + // If not NULL, mask_box is filled with the position box of the returned + // mask image. + Pix *render_mask(const FCOORD &rerotation, TBOX *mask_box); + +#ifndef GRAPHICS_DISABLED + /// draw histogram + ///@param window window to draw in + ///@param serial serial number + ///@param colour colour to draw in + void plot(ScrollView *window, inT32 serial, ScrollView::Color colour); +#endif // GRAPHICS_DISABLED + + /// assignment + ///@param source from this + PDBLK &operator=(const PDBLK &source); + + protected: + POLY_BLOCK *hand_poly; //< weird as well + ICOORDELT_LIST leftside; //< left side vertices + ICOORDELT_LIST rightside; //< right side vertices + TBOX box; //< bounding box + int index_; //< Serial number of this block. }; class DLLSYM BLOCK_RECT_IT //rectangle iterator diff --git a/ccstruct/polyaprx.cpp b/ccstruct/polyaprx.cpp index 81b8500a0e..7597349543 100644 --- a/ccstruct/polyaprx.cpp +++ b/ccstruct/polyaprx.cpp @@ -214,7 +214,7 @@ EDGEPT edgepts[] //output is array void fix2( //polygonal approx EDGEPT *start, /*loop to approimate */ int area) { - EDGEPT *edgept; /*current point */ + EDGEPT *edgept; /*current point */ EDGEPT *edgept1; EDGEPT *loopstart; /*modified start of loop */ EDGEPT *linestart; /*start of line segment */ diff --git a/ccstruct/polyblk.cpp b/ccstruct/polyblk.cpp index e0a455905f..b5ca2e1212 100644 --- a/ccstruct/polyblk.cpp +++ b/ccstruct/polyblk.cpp @@ -1,7 +1,7 @@ /********************************************************************** * File: polyblk.c (Formerly poly_block.c) * Description: Polygonal blocks - * Author: Sheelagh Lloyd? + * Author: Sheelagh Lloyd? * Created: * * (C) Copyright 1993, Hewlett-Packard Ltd. diff --git a/ccstruct/quspline.cpp b/ccstruct/quspline.cpp index f50cfe50ee..82107e1e25 100644 --- a/ccstruct/quspline.cpp +++ b/ccstruct/quspline.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: quspline.cpp (Formerly qspline.c) * Description: Code for the QSPLINE class. - * Author: Ray Smith - * Created: Tue Oct 08 17:16:12 BST 1991 + * Author: Ray Smith + * Created: Tue Oct 08 17:16:12 BST 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); diff --git a/ccstruct/ratngs.h b/ccstruct/ratngs.h index 31b27cfb37..2ee9c94a30 100644 --- a/ccstruct/ratngs.h +++ b/ccstruct/ratngs.h @@ -268,7 +268,7 @@ const char *ScriptPosToString(tesseract::ScriptPos script_pos); } // namespace tesseract. -class WERD_CHOICE : public ELIST_LINK { +class TESS_API WERD_CHOICE : public ELIST_LINK { public: static const float kBadRating; static const char *permuter_name(uinT8 permuter); @@ -288,7 +288,8 @@ class WERD_CHOICE : public ELIST_LINK { src_certainty, src_permuter); } WERD_CHOICE(const char *src_string, const UNICHARSET &unicharset); - WERD_CHOICE(const WERD_CHOICE &word) : ELIST_LINK(word), unicharset_(word.unicharset_) { + WERD_CHOICE(const WERD_CHOICE &word) + : ELIST_LINK(word), unicharset_(word.unicharset_) { this->init(word.length()); this->operator=(word); } diff --git a/ccstruct/rect.cpp b/ccstruct/rect.cpp index 22417485db..4a9fe00b34 100644 --- a/ccstruct/rect.cpp +++ b/ccstruct/rect.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: rect.c (Formerly box.c) * Description: Bounding box class definition. - * Author: Phil Cheatle - * Created: Wed Oct 16 15:18:45 BST 1991 + * Author: Phil Cheatle + * Created: Wed Oct 16 15:18:45 BST 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -29,10 +29,10 @@ * **********************************************************************/ -TBOX::TBOX( //constructor - const ICOORD pt1, //one corner - const ICOORD pt2 //the other corner - ) { +TBOX::TBOX( // constructor + const ICOORD pt1, // one corner + const ICOORD pt2 // the other corner + ) { if (pt1.x () <= pt2.x ()) { if (pt1.y () <= pt2.y ()) { bot_left = pt1; diff --git a/ccstruct/rect.h b/ccstruct/rect.h index d9b90642f4..f31247a1af 100644 --- a/ccstruct/rect.h +++ b/ccstruct/rect.h @@ -1,8 +1,8 @@ /********************************************************************** * File: rect.h (Formerly box.h) * Description: Bounding box class definition. - * Author: Phil Cheatle - * Created: Wed Oct 16 15:18:45 BST 1991 + * Author: Phil Cheatle + * Created: Wed Oct 16 15:18:45 BST 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -307,9 +307,9 @@ class DLLSYM TBOX { // bounding box * **********************************************************************/ -inline TBOX::TBOX( // constructor - const FCOORD pt // floating centre - ) { +inline TBOX::TBOX( // constructor + const FCOORD pt // floating centre + ) { bot_left = ICOORD ((inT16) floor (pt.x ()), (inT16) floor (pt.y ())); top_right = ICOORD ((inT16) ceil (pt.x ()), (inT16) ceil (pt.y ())); } diff --git a/ccstruct/rejctmap.cpp b/ccstruct/rejctmap.cpp index 9c9ff2e503..a2910675a8 100644 --- a/ccstruct/rejctmap.cpp +++ b/ccstruct/rejctmap.cpp @@ -267,7 +267,7 @@ void REJ::full_print(FILE *fp) { //The REJMAP class has been hacked to use alloc_struct instead of new []. //This is to reduce memory fragmentation only as it is rather kludgy. -//alloc_struct by-passes the call to the contsructor of REJ on each +//alloc_struct by-passes the call to the constructor of REJ on each //array element. Although the constructor is empty, the BITS16 members //do have a constructor which sets all the flags to 0. The memset //replaces this functionality. diff --git a/ccstruct/rejctmap.h b/ccstruct/rejctmap.h index d945dda1fa..009ba58a78 100644 --- a/ccstruct/rejctmap.h +++ b/ccstruct/rejctmap.h @@ -1,8 +1,8 @@ /********************************************************************** * File: rejctmap.h (Formerly rejmap.h) * Description: REJ and REJMAP class functions. - * Author: Phil Cheatle - * Created: Thu Jun 9 13:46:38 BST 1994 + * Author: Phil Cheatle + * Created: Thu Jun 9 13:46:38 BST 1994 * * (C) Copyright 1994, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -48,46 +48,45 @@ OF THIS IMPLIED TEMPORAL ORDERING OF THE FLAGS!!!! #include "bits16.h" #include "params.h" -enum REJ_FLAGS -{ +enum REJ_FLAGS { /* Reject modes which are NEVER overridden */ - R_TESS_FAILURE, // PERM Tess didn't classify - R_SMALL_XHT, // PERM Xht too small - R_EDGE_CHAR, // PERM Too close to edge of image - R_1IL_CONFLICT, // PERM 1Il confusion - R_POSTNN_1IL, // PERM 1Il unrejected by NN - R_REJ_CBLOB, // PERM Odd blob - R_MM_REJECT, // PERM Matrix match rejection (m's) - R_BAD_REPETITION, // TEMP Repeated char which doesn't match trend + R_TESS_FAILURE, // PERM Tess didn't classify + R_SMALL_XHT, // PERM Xht too small + R_EDGE_CHAR, // PERM Too close to edge of image + R_1IL_CONFLICT, // PERM 1Il confusion + R_POSTNN_1IL, // PERM 1Il unrejected by NN + R_REJ_CBLOB, // PERM Odd blob + R_MM_REJECT, // PERM Matrix match rejection (m's) + R_BAD_REPETITION, // TEMP Repeated char which doesn't match trend /* Initial reject modes (pre NN_ACCEPT) */ - R_POOR_MATCH, // TEMP Ray's original heuristic (Not used) - R_NOT_TESS_ACCEPTED, // TEMP Tess didn't accept WERD - R_CONTAINS_BLANKS, // TEMP Tess failed on other chs in WERD - R_BAD_PERMUTER, // POTENTIAL Bad permuter for WERD + R_POOR_MATCH, // TEMP Ray's original heuristic (Not used) + R_NOT_TESS_ACCEPTED, // TEMP Tess didn't accept WERD + R_CONTAINS_BLANKS, // TEMP Tess failed on other chs in WERD + R_BAD_PERMUTER, // POTENTIAL Bad permuter for WERD /* Reject modes generated after NN_ACCEPT but before MM_ACCEPT */ - R_HYPHEN, // TEMP Post NN dodgy hyphen or full stop - R_DUBIOUS, // TEMP Post NN dodgy chars - R_NO_ALPHANUMS, // TEMP No alphanumerics in word after NN - R_MOSTLY_REJ, // TEMP Most of word rejected so rej the rest - R_XHT_FIXUP, // TEMP Xht tests unsure + R_HYPHEN, // TEMP Post NN dodgy hyphen or full stop + R_DUBIOUS, // TEMP Post NN dodgy chars + R_NO_ALPHANUMS, // TEMP No alphanumerics in word after NN + R_MOSTLY_REJ, // TEMP Most of word rejected so rej the rest + R_XHT_FIXUP, // TEMP Xht tests unsure /* Reject modes generated after MM_ACCEPT but before QUALITY_ACCEPT */ - R_BAD_QUALITY, // TEMP Quality metrics bad for WERD + R_BAD_QUALITY, // TEMP Quality metrics bad for WERD /* Reject modes generated after QUALITY_ACCEPT but before MINIMAL_REJ accep*/ - R_DOC_REJ, // TEMP Document rejection - R_BLOCK_REJ, // TEMP Block rejection - R_ROW_REJ, // TEMP Row rejection - R_UNLV_REJ, // TEMP ~ turned to - or ^ turned to space + R_DOC_REJ, // TEMP Document rejection + R_BLOCK_REJ, // TEMP Block rejection + R_ROW_REJ, // TEMP Row rejection + R_UNLV_REJ, // TEMP ~ turned to - or ^ turned to space /* Accept modes which occur between the above rejection groups */ - R_NN_ACCEPT, //NN acceptance - R_HYPHEN_ACCEPT, //Hyphen acceptance - R_MM_ACCEPT, //Matrix match acceptance - R_QUALITY_ACCEPT, //Accept word in good quality doc - R_MINIMAL_REJ_ACCEPT //Accept EVERYTHING except tess failures + R_NN_ACCEPT, // NN acceptance + R_HYPHEN_ACCEPT, // Hyphen acceptance + R_MM_ACCEPT, // Matrix match acceptance + R_QUALITY_ACCEPT, // Accept word in good quality doc + R_MINIMAL_REJ_ACCEPT // Accept EVERYTHING except tess failures }; /* REJECT MAP VALUES */ diff --git a/ccstruct/statistc.cpp b/ccstruct/statistc.cpp index 39d5edd180..8b1ba8c9a1 100644 --- a/ccstruct/statistc.cpp +++ b/ccstruct/statistc.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: statistc.c (Formerly stats.c) * Description: Simple statistical package for integer values. - * Author: Ray Smith - * Created: Mon Feb 04 16:56:05 GMT 1991 + * Author: Ray Smith + * Created: Mon Feb 04 16:56:05 GMT 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -215,7 +215,6 @@ inT32 STATS::min_bucket() const { // Find min return rangemin_ + min; } - /********************************************************************** * STATS::max_bucket * diff --git a/ccutil/ambigs.h b/ccutil/ambigs.h index b278f9f39d..faab21989b 100644 --- a/ccutil/ambigs.h +++ b/ccutil/ambigs.h @@ -59,17 +59,18 @@ class UnicharIdArrayUtils { // less than length of array2, if any array1[i] is less than array2[i]. // Returns 0 if the arrays are equal, 1 otherwise. // The function assumes that the arrays are terminated by INVALID_UNICHAR_ID. - static inline int compare(const UNICHAR_ID array1[], - const UNICHAR_ID array2[]) { - const UNICHAR_ID *ptr1 = array1; - const UNICHAR_ID *ptr2 = array2; - while (*ptr1 != INVALID_UNICHAR_ID && *ptr2 != INVALID_UNICHAR_ID) { - if (*ptr1 != *ptr2) return *ptr1 < *ptr2 ? -1 : 1; - ++ptr1; - ++ptr2; + static inline int compare(const UNICHAR_ID *ptr1, const UNICHAR_ID *ptr2) { + for (;;) { + const UNICHAR_ID val1 = *ptr1++; + const UNICHAR_ID val2 = *ptr2++; + if (val1 != val2) { + if (val1 == INVALID_UNICHAR_ID) return -1; + if (val2 == INVALID_UNICHAR_ID) return 1; + if (val1 < val2) return -1; + return 1; + } + if (val1 == INVALID_UNICHAR_ID) return 0; } - if (*ptr1 == INVALID_UNICHAR_ID && *ptr2 == INVALID_UNICHAR_ID) return 0; - return *ptr1 == INVALID_UNICHAR_ID ? -1 : 1; } // Look uid in the vector of uids. If found, the index of the matched diff --git a/ccutil/bits16.h b/ccutil/bits16.h index 352b48bee2..6bbec4c0f1 100644 --- a/ccutil/bits16.h +++ b/ccutil/bits16.h @@ -37,12 +37,12 @@ class DLLSYM BITS16 void turn_on_bit( // flip specified bit uinT8 bit_num) { // bit to flip 0..7 val = val | 01 << bit_num; - }; + } void turn_off_bit( // flip specified bit uinT8 bit_num) { // bit to flip 0..7 val = val & ~(01 << bit_num); - }; + } void set_bit( // flip specified bit uinT8 bit_num, // bit to flip 0..7 @@ -51,11 +51,11 @@ class DLLSYM BITS16 val = val | 01 << bit_num; else val = val & ~(01 << bit_num); - }; + } BOOL8 bit( // access bit uinT8 bit_num) const { // bit to access return (val >> bit_num) & 01; - }; + } }; #endif diff --git a/ccutil/ccutil.cpp b/ccutil/ccutil.cpp index ecf2cb5e36..8f965bb8c1 100644 --- a/ccutil/ccutil.cpp +++ b/ccutil/ccutil.cpp @@ -1,5 +1,14 @@ // Copyright 2008 Google Inc. All Rights Reserved. // Author: scharron@google.com (Samuel Charron) +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include "ccutil.h" diff --git a/ccutil/ccutil.h b/ccutil/ccutil.h index 974ef7d225..faba0aef06 100644 --- a/ccutil/ccutil.h +++ b/ccutil/ccutil.h @@ -26,9 +26,7 @@ #include "params.h" #include "unicharset.h" -#ifdef _WIN32 -#include -#else +#ifndef _WIN32 #include #include #endif @@ -51,7 +49,7 @@ class CCUtilMutex { }; -class CCUtil { +class TESS_API CCUtil { public: CCUtil(); virtual ~CCUtil(); diff --git a/ccutil/clst.cpp b/ccutil/clst.cpp index fbbb561fad..52caadf38f 100644 --- a/ccutil/clst.cpp +++ b/ccutil/clst.cpp @@ -26,7 +26,7 @@ **********************************************************************/ /*********************************************************************** - * CLIST::internal_deep_clear + * CLIST::internal_deep_clear * * Used by the "deep_clear" member function of derived list * classes to destroy all the elements on the list. @@ -56,9 +56,8 @@ void (*zapper) (void *)) { //ptr to zapper functn } } - /*********************************************************************** - * CLIST::shallow_clear + * CLIST::shallow_clear * * Used by the destructor and the "shallow_clear" member function of derived * list classes to destroy the list. @@ -83,7 +82,7 @@ void CLIST::shallow_clear() { //destroy all links } /*********************************************************************** - * CLIST::assign_to_sublist + * CLIST::assign_to_sublist * * The list is set to a sublist of another list. "This" list must be empty * before this function is invoked. The two iterators passed must refer to @@ -107,9 +106,8 @@ void CLIST::assign_to_sublist( //to this list last = start_it->extract_sublist (end_it); } - /*********************************************************************** - * CLIST::length + * CLIST::length * * Return count of elements on list **********************************************************************/ @@ -123,9 +121,8 @@ inT32 CLIST::length() const { //count elements return count; } - /*********************************************************************** - * CLIST::sort + * CLIST::sort * * Sort elements on list **********************************************************************/ @@ -239,7 +236,7 @@ void CLIST::set_subtract(int comparator(const void*, const void*), **********************************************************************/ /*********************************************************************** - * CLIST_ITERATOR::forward + * CLIST_ITERATOR::forward * * Move the iterator to the next element of the list. * REMEMBER: ALL LISTS ARE CIRCULAR. @@ -276,9 +273,8 @@ void *CLIST_ITERATOR::forward() { return current->data; } - /*********************************************************************** - * CLIST_ITERATOR::data_relative + * CLIST_ITERATOR::data_relative * * Return the data pointer to the element "offset" elements from current. * "offset" must not be less than -1. @@ -312,9 +308,8 @@ void *CLIST_ITERATOR::data_relative( //get data + or - ... return ptr->data; } - /*********************************************************************** - * CLIST_ITERATOR::move_to_last() + * CLIST_ITERATOR::move_to_last() * * Move current so that it is set to the end of the list. * Return data just in case anyone wants it. @@ -336,9 +331,8 @@ void *CLIST_ITERATOR::move_to_last() { return current->data; } - /*********************************************************************** - * CLIST_ITERATOR::exchange() + * CLIST_ITERATOR::exchange() * * Given another iterator, whose current element is a different element on * the same list list OR an element of another list, exchange the two current @@ -434,9 +428,8 @@ void CLIST_ITERATOR::exchange( //positions of 2 link other_it->current = old_current; } - /*********************************************************************** - * CLIST_ITERATOR::extract_sublist() + * CLIST_ITERATOR::extract_sublist() * * This is a private member, used only by CLIST::assign_to_sublist. * Given another iterator for the same list, extract the links from THIS to @@ -478,7 +471,7 @@ CLIST_LINK *CLIST_ITERATOR::extract_sublist( //from temp_it.mark_cycle_pt (); do { //walk sublist - if (temp_it.cycled_list ()) //can't find end pt + if (temp_it.cycled_list()) // can't find end pt BAD_SUBLIST.error ("CLIST_ITERATOR.extract_sublist", ABORT, NULL); if (temp_it.at_last ()) { diff --git a/ccutil/clst.h b/ccutil/clst.h index a5a42a6e40..f93d75afcb 100644 --- a/ccutil/clst.h +++ b/ccutil/clst.h @@ -28,9 +28,9 @@ class CLIST_ITERATOR; /********************************************************************** - * CLASS - CLIST_LINK + * CLASS - CLIST_LINK * - * Generic link class for singly linked CONS cell lists + * Generic link class for singly linked CONS cell lists * * Note: No destructor - elements are assumed to be destroyed EITHER after * they have been extracted from a list OR by the CLIST destructor which @@ -50,13 +50,13 @@ class DLLSYM CLIST_LINK data = next = NULL; } - CLIST_LINK( //copy constructor - const CLIST_LINK &) { //don't copy link + CLIST_LINK( // copy constructor + const CLIST_LINK &) { // don't copy link data = next = NULL; } - void operator= ( //don't copy links - const CLIST_LINK &) { + void operator=( // don't copy links + const CLIST_LINK &) { data = next = NULL; } }; @@ -89,8 +89,8 @@ class DLLSYM CLIST void internal_deep_clear ( //destroy all links void (*zapper) (void *)); //ptr to zapper functn - void shallow_clear(); //clear list but don't - //delete data elements + void shallow_clear(); // clear list but don't + // delete data elements bool empty() const { //is list empty? return !last; @@ -136,9 +136,10 @@ class DLLSYM CLIST }; /*********************************************************************** - * CLASS - CLIST_ITERATOR + * CLASS - CLIST_ITERATOR * - * Generic iterator class for singly linked lists with embedded links + * Generic iterator class for singly linked lists with embedded + *links **********************************************************************/ class DLLSYM CLIST_ITERATOR @@ -231,8 +232,8 @@ class DLLSYM CLIST_ITERATOR BOOL8 cycled_list(); //Completed a cycle? - void add_to_end( //add at end & - void *new_data); //don't move + void add_to_end( // add at end & + void *new_data); // don't move void exchange( //positions of 2 links CLIST_ITERATOR *other_it); //other iterator @@ -246,7 +247,7 @@ class DLLSYM CLIST_ITERATOR }; /*********************************************************************** - * CLIST_ITERATOR::set_to_list + * CLIST_ITERATOR::set_to_list * * (Re-)initialise the iterator to point to the start of the list_to_iterate * over. @@ -270,9 +271,8 @@ inline void CLIST_ITERATOR::set_to_list( //change list ex_current_was_cycle_pt = FALSE; } - /*********************************************************************** - * CLIST_ITERATOR::CLIST_ITERATOR + * CLIST_ITERATOR::CLIST_ITERATOR * * CONSTRUCTOR - set iterator to specified list; **********************************************************************/ @@ -281,9 +281,8 @@ inline CLIST_ITERATOR::CLIST_ITERATOR(CLIST *list_to_iterate) { set_to_list(list_to_iterate); } - /*********************************************************************** - * CLIST_ITERATOR::add_after_then_move + * CLIST_ITERATOR::add_after_then_move * * Add a new element to the list after the current element and move the * iterator to the new element. @@ -329,9 +328,8 @@ inline void CLIST_ITERATOR::add_after_then_move( // element to add current = new_element; } - /*********************************************************************** - * CLIST_ITERATOR::add_after_stay_put + * CLIST_ITERATOR::add_after_stay_put * * Add a new element to the list after the current element but do not move * the iterator to the new element. @@ -380,9 +378,8 @@ inline void CLIST_ITERATOR::add_after_stay_put( // element to add } } - /*********************************************************************** - * CLIST_ITERATOR::add_before_then_move + * CLIST_ITERATOR::add_before_then_move * * Add a new element to the list before the current element and move the * iterator to the new element. @@ -425,9 +422,8 @@ inline void CLIST_ITERATOR::add_before_then_move( // element to add current = new_element; } - /*********************************************************************** - * CLIST_ITERATOR::add_before_stay_put + * CLIST_ITERATOR::add_before_stay_put * * Add a new element to the list before the current element but don't move the * iterator to the new element. @@ -471,11 +467,11 @@ inline void CLIST_ITERATOR::add_before_stay_put( // element to add } } - /*********************************************************************** - * CLIST_ITERATOR::add_list_after + * CLIST_ITERATOR::add_list_after * - * Insert another list to this list after the current element but don't move the + * Insert another list to this list after the current element but don't move + *the * iterator. **********************************************************************/ @@ -518,9 +514,8 @@ inline void CLIST_ITERATOR::add_list_after(CLIST *list_to_add) { } } - /*********************************************************************** - * CLIST_ITERATOR::add_list_before + * CLIST_ITERATOR::add_list_before * * Insert another list to this list before the current element. Move the * iterator to the start of the inserted elements @@ -563,9 +558,8 @@ inline void CLIST_ITERATOR::add_list_before(CLIST *list_to_add) { } } - /*********************************************************************** - * CLIST_ITERATOR::extract + * CLIST_ITERATOR::extract * * Do extraction by removing current from the list, deleting the cons cell * and returning the data to the caller, but NOT updating the iterator. (So @@ -606,9 +600,8 @@ inline void *CLIST_ITERATOR::extract() { return extracted_data; } - /*********************************************************************** - * CLIST_ITERATOR::move_to_first() + * CLIST_ITERATOR::move_to_first() * * Move current so that it is set to the start of the list. * Return data just in case anyone wants it. @@ -626,9 +619,8 @@ inline void *CLIST_ITERATOR::move_to_first() { return current != NULL ? current->data : NULL; } - /*********************************************************************** - * CLIST_ITERATOR::mark_cycle_pt() + * CLIST_ITERATOR::mark_cycle_pt() * * Remember the current location so that we can tell whether we've returned * to this point later. @@ -651,9 +643,8 @@ inline void CLIST_ITERATOR::mark_cycle_pt() { started_cycling = FALSE; } - /*********************************************************************** - * CLIST_ITERATOR::at_first() + * CLIST_ITERATOR::at_first() * * Are we at the start of the list? * @@ -671,9 +662,8 @@ inline BOOL8 CLIST_ITERATOR::at_first() { !ex_current_was_last)); //first and last } - /*********************************************************************** - * CLIST_ITERATOR::at_last() + * CLIST_ITERATOR::at_last() * * Are we at the end of the list? * @@ -691,9 +681,8 @@ inline BOOL8 CLIST_ITERATOR::at_last() { ex_current_was_last)); //first and last } - /*********************************************************************** - * CLIST_ITERATOR::cycled_list() + * CLIST_ITERATOR::cycled_list() * * Have we returned to the cycle_pt since it was set? * @@ -709,9 +698,8 @@ inline BOOL8 CLIST_ITERATOR::cycled_list() { } - /*********************************************************************** - * CLIST_ITERATOR::length() + * CLIST_ITERATOR::length() * * Return the length of the list * @@ -726,9 +714,8 @@ inline inT32 CLIST_ITERATOR::length() { return list->length (); } - /*********************************************************************** - * CLIST_ITERATOR::sort() + * CLIST_ITERATOR::sort() * * Sort the elements of the list, then reposition at the start. * @@ -747,9 +734,8 @@ const void *, const void *)) { move_to_first(); } - /*********************************************************************** - * CLIST_ITERATOR::add_to_end + * CLIST_ITERATOR::add_to_end * * Add a new element to the end of the list without moving the iterator. * This is provided because a single linked list cannot move to the last as @@ -811,7 +797,7 @@ The macro generates: - An element deletion function: CLASSNAME##_c1_zapper - An element copier function: CLASSNAME##_c1_copier - - A CLIST subclass: CLASSNAME##_CLIST + - A CLIST subclass: CLASSNAME##_CLIST - A CLIST_ITERATOR subclass: CLASSNAME##_C_IT @@ -830,114 +816,116 @@ CLISTIZEH is a concatenation of 3 fragments CLISTIZEH_A, CLISTIZEH_B and CLISTIZEH_C. ***********************************************************************/ -#define CLISTIZEH_A( CLASSNAME ) \ - \ -extern DLLSYM void CLASSNAME##_c1_zapper( /*delete a link*/ \ -void* link); /*link to delete*/ \ - \ -extern DLLSYM void* CLASSNAME##_c1_copier( /*deep copy a link*/ \ -void* old_element); /*source link */ - -#define CLISTIZEH_B( CLASSNAME ) \ - \ -/*********************************************************************** \ -* CLASS - CLASSNAME##_CLIST \ -* \ -* List class for class CLASSNAME \ -* \ -**********************************************************************/ \ - \ -class DLLSYM CLASSNAME##_CLIST : public CLIST \ -{ \ -public: \ - CLASSNAME##_CLIST():CLIST() {} \ - /* constructor */ \ - \ - CLASSNAME##_CLIST( /* don't construct */ \ - const CLASSNAME##_CLIST&) /*by initial assign*/ \ - { DONT_CONSTRUCT_LIST_BY_COPY.error( QUOTE_IT( CLASSNAME##_CLIST ), \ - ABORT, NULL ); } \ - \ -void deep_clear() /* delete elements */ \ - { CLIST::internal_deep_clear( &CLASSNAME##_c1_zapper ); } \ - \ -void operator=( /* prevent assign */ \ - const CLASSNAME##_CLIST&) \ - { DONT_ASSIGN_LISTS.error( QUOTE_IT( CLASSNAME##_CLIST ), \ - ABORT, NULL ); } - -#define CLISTIZEH_C( CLASSNAME ) \ - \ -}; \ - \ - \ - \ -/*********************************************************************** \ -* CLASS - CLASSNAME##_C_IT \ -* \ -* Iterator class for class CLASSNAME##_CLIST \ -* \ -* Note: We don't need to coerce pointers to member functions input \ -* parameters as these are automatically converted to the type of the base \ -* type. ("A ptr to a class may be converted to a pointer to a public base \ -* class of that class") \ -**********************************************************************/ \ - \ -class DLLSYM CLASSNAME##_C_IT : public CLIST_ITERATOR \ -{ \ -public: \ - CLASSNAME##_C_IT():CLIST_ITERATOR(){} \ - \ - CLASSNAME##_C_IT( \ - CLASSNAME##_CLIST* list):CLIST_ITERATOR(list){} \ - \ - CLASSNAME* data() \ - { return (CLASSNAME*) CLIST_ITERATOR::data(); } \ - \ - CLASSNAME* data_relative( \ - inT8 offset) \ - { return (CLASSNAME*) CLIST_ITERATOR::data_relative( offset ); } \ - \ - CLASSNAME* forward() \ - { return (CLASSNAME*) CLIST_ITERATOR::forward(); } \ - \ - CLASSNAME* extract() \ - { return (CLASSNAME*) CLIST_ITERATOR::extract(); } \ - \ - CLASSNAME* move_to_first() \ - { return (CLASSNAME*) CLIST_ITERATOR::move_to_first(); } \ - \ - CLASSNAME* move_to_last() \ - { return (CLASSNAME*) CLIST_ITERATOR::move_to_last(); } \ -}; +#define CLISTIZEH_A(CLASSNAME) \ + \ + extern DLLSYM void CLASSNAME##_c1_zapper( /*delete a link*/ \ + void *link); /*link to delete*/ \ + \ + extern DLLSYM void \ + *CLASSNAME##_c1_copier( /*deep copy a link*/ \ + void *old_element); /*source link */ + +#define CLISTIZEH_B(CLASSNAME) \ + \ + /*********************************************************************** \ + * CLASS - \ + *CLASSNAME##_CLIST \ + * \ + * List class for class \ + *CLASSNAME \ + * \ + **********************************************************************/ \ + \ + class DLLSYM CLASSNAME##_CLIST : public CLIST { \ + public: \ + CLASSNAME##_CLIST() : CLIST() {} \ + /* constructor */ \ + \ + CLASSNAME##_CLIST( /* don't construct */ \ + const CLASSNAME##_CLIST &) /*by initial assign*/ \ + { \ + DONT_CONSTRUCT_LIST_BY_COPY.error(QUOTE_IT(CLASSNAME##_CLIST), ABORT, \ + NULL); \ + } \ + \ + void deep_clear() /* delete elements */ \ + { \ + CLIST::internal_deep_clear(&CLASSNAME##_c1_zapper); \ + } \ + \ + void operator=(/* prevent assign */ \ + const CLASSNAME##_CLIST &) { \ + DONT_ASSIGN_LISTS.error(QUOTE_IT(CLASSNAME##_CLIST), ABORT, NULL); \ + } -#define CLISTIZEH( CLASSNAME ) \ - \ -CLISTIZEH_A( CLASSNAME ) \ - \ -CLISTIZEH_B( CLASSNAME ) \ - \ -CLISTIZEH_C( CLASSNAME ) +#define CLISTIZEH_C(CLASSNAME) \ + } \ + ; \ + \ + /*********************************************************************** \ + * CLASS - CLASSNAME##_C_IT \ + * \ + * Iterator class for class CLASSNAME##_CLIST \ + * \ + * Note: We don't need to coerce pointers to member functions input \ + * parameters as these are automatically converted to the type of the base \ + * type. ("A ptr to a class may be converted to a pointer to a public base \ + * class of that class") \ + **********************************************************************/ \ + \ + class DLLSYM CLASSNAME##_C_IT : public CLIST_ITERATOR { \ + public: \ + CLASSNAME##_C_IT() : CLIST_ITERATOR() {} \ + \ + CLASSNAME##_C_IT(CLASSNAME##_CLIST *list) : CLIST_ITERATOR(list) {} \ + \ + CLASSNAME *data() { return (CLASSNAME *)CLIST_ITERATOR::data(); } \ + \ + CLASSNAME *data_relative(inT8 offset) { \ + return (CLASSNAME *)CLIST_ITERATOR::data_relative(offset); \ + } \ + \ + CLASSNAME *forward() { return (CLASSNAME *)CLIST_ITERATOR::forward(); } \ + \ + CLASSNAME *extract() { return (CLASSNAME *)CLIST_ITERATOR::extract(); } \ + \ + CLASSNAME *move_to_first() { \ + return (CLASSNAME *)CLIST_ITERATOR::move_to_first(); \ + } \ + \ + CLASSNAME *move_to_last() { \ + return (CLASSNAME *)CLIST_ITERATOR::move_to_last(); \ + } \ + }; + +#define CLISTIZEH(CLASSNAME) \ + \ + CLISTIZEH_A(CLASSNAME) \ + \ + CLISTIZEH_B(CLASSNAME) \ + \ + CLISTIZEH_C(CLASSNAME) /*********************************************************************** CLISTIZE( CLASSNAME ) MACRO ***********************************************************************/ -#define CLISTIZE( CLASSNAME ) \ - \ -/*********************************************************************** \ -* CLASSNAME##_c1_zapper \ -* \ -* A function which can delete a CLASSNAME element. This is passed to the \ -* generic deep_clear list member function so that when a list is cleared the \ -* elements on the list are properly destroyed from the base class, even \ -* though we don't use a virtual destructor function. \ -**********************************************************************/ \ - \ -DLLSYM void CLASSNAME##_c1_zapper( /*delete a link*/ \ -void* link) /*link to delete*/ \ -{ \ -delete (CLASSNAME *) link; \ -} \ +#define CLISTIZE(CLASSNAME) \ + \ + /*********************************************************************** \ + * CLASSNAME##_c1_zapper \ + * \ + * A function which can delete a CLASSNAME element. This is passed to the \ + * generic deep_clear list member function so that when a list is cleared \ + *the \ + * elements on the list are properly destroyed from the base class, even \ + * though we don't use a virtual destructor function. \ + **********************************************************************/ \ + \ + DLLSYM void CLASSNAME##_c1_zapper( /*delete a link*/ \ + void *link) /*link to delete*/ \ + { \ + delete (CLASSNAME *)link; \ + } #endif diff --git a/ccutil/elst.cpp b/ccutil/elst.cpp index 8ad999b5ba..2d2c9ad65d 100644 --- a/ccutil/elst.cpp +++ b/ccutil/elst.cpp @@ -26,7 +26,7 @@ **********************************************************************/ /*********************************************************************** - * ELIST::internal_clear + * ELIST::internal_clear * * Used by the destructor and the "clear" member function of derived list * classes to destroy all the elements on the list. @@ -57,7 +57,7 @@ void (*zapper) (ELIST_LINK *)) { } /*********************************************************************** - * ELIST::assign_to_sublist + * ELIST::assign_to_sublist * * The list is set to a sublist of another list. "This" list must be empty * before this function is invoked. The two iterators passed must refer to @@ -81,9 +81,8 @@ void ELIST::assign_to_sublist( //to this list last = start_it->extract_sublist (end_it); } - /*********************************************************************** - * ELIST::length + * ELIST::length * * Return count of elements on list **********************************************************************/ @@ -97,9 +96,8 @@ inT32 ELIST::length() const { // count elements return count; } - /*********************************************************************** - * ELIST::sort + * ELIST::sort * * Sort elements on list * NB If you don't like the const declarations in the comparator, coerce yours: @@ -187,7 +185,7 @@ ELIST_LINK *ELIST::add_sorted_and_find( **********************************************************************/ /*********************************************************************** - * ELIST_ITERATOR::forward + * ELIST_ITERATOR::forward * * Move the iterator to the next element of the list. * REMEMBER: ALL LISTS ARE CIRCULAR. @@ -224,9 +222,8 @@ ELIST_LINK *ELIST_ITERATOR::forward() { return current; } - /*********************************************************************** - * ELIST_ITERATOR::data_relative + * ELIST_ITERATOR::data_relative * * Return the data pointer to the element "offset" elements from current. * "offset" must not be less than -1. @@ -260,9 +257,8 @@ ELIST_LINK *ELIST_ITERATOR::data_relative( //get data + or - ... return ptr; } - /*********************************************************************** - * ELIST_ITERATOR::move_to_last() + * ELIST_ITERATOR::move_to_last() * * Move current so that it is set to the end of the list. * Return data just in case anyone wants it. @@ -281,9 +277,8 @@ ELIST_LINK *ELIST_ITERATOR::move_to_last() { return current; } - /*********************************************************************** - * ELIST_ITERATOR::exchange() + * ELIST_ITERATOR::exchange() * * Given another iterator, whose current element is a different element on * the same list list OR an element of another list, exchange the two current @@ -379,9 +374,8 @@ void ELIST_ITERATOR::exchange( //positions of 2 link other_it->current = old_current; } - /*********************************************************************** - * ELIST_ITERATOR::extract_sublist() + * ELIST_ITERATOR::extract_sublist() * * This is a private member, used only by ELIST::assign_to_sublist. * Given another iterator for the same list, extract the links from THIS to @@ -425,7 +419,7 @@ ELIST_LINK *ELIST_ITERATOR::extract_sublist( //from temp_it.mark_cycle_pt (); do { //walk sublist - if (temp_it.cycled_list ()) //can't find end pt + if (temp_it.cycled_list()) // can't find end pt BAD_SUBLIST.error ("ELIST_ITERATOR.extract_sublist", ABORT, NULL); if (temp_it.at_last ()) { diff --git a/ccutil/elst.h b/ccutil/elst.h index e239577419..d53a7c34f0 100644 --- a/ccutil/elst.h +++ b/ccutil/elst.h @@ -98,8 +98,8 @@ class DLLSYM ELIST_LINK next = NULL; } - void operator= ( //don't copy links - const ELIST_LINK &) { + void operator=( // don't copy links + const ELIST_LINK &) { next = NULL; } }; @@ -273,8 +273,8 @@ class DLLSYM ELIST_ITERATOR bool cycled_list(); //Completed a cycle? - void add_to_end( //add at end & - ELIST_LINK *new_link); //don't move + void add_to_end( // add at end & + ELIST_LINK *new_link); // don't move void exchange( //positions of 2 links ELIST_ITERATOR *other_it); //other iterator @@ -458,7 +458,6 @@ inline void ELIST_ITERATOR::add_before_then_move( // element to add current = new_element; } - /*********************************************************************** * ELIST_ITERATOR::add_before_stay_put * @@ -501,11 +500,11 @@ inline void ELIST_ITERATOR::add_before_stay_put( // element to add } } - /*********************************************************************** * ELIST_ITERATOR::add_list_after * - * Insert another list to this list after the current element but don't move the + * Insert another list to this list after the current element but don't move + *the * iterator. **********************************************************************/ @@ -959,30 +958,29 @@ ELISTIZEH_C( CLASSNAME ) ELISTIZE( CLASSNAME ) MACRO ***********************************************************************/ -#define ELISTIZE(CLASSNAME) \ - \ -/*********************************************************************** \ -* CLASSNAME##_zapper \ -* \ -* A function which can delete a CLASSNAME element. This is passed to the \ -* generic clear list member function so that when a list is cleared the \ -* elements on the list are properly destroyed from the base class, even \ -* though we don't use a virtual destructor function. \ -**********************************************************************/ \ - \ -DLLSYM void CLASSNAME##_zapper(ELIST_LINK* link) { \ - delete reinterpret_cast(link); \ -} \ - \ -/* Become a deep copy of src_list*/ \ -void CLASSNAME##_LIST::deep_copy(const CLASSNAME##_LIST* src_list, \ - CLASSNAME* (*copier)(const CLASSNAME*)) { \ - \ - CLASSNAME##_IT from_it(const_cast(src_list)); \ - CLASSNAME##_IT to_it(this); \ - \ - for (from_it.mark_cycle_pt(); !from_it.cycled_list(); from_it.forward()) \ - to_it.add_after_then_move((*copier)(from_it.data())); \ -} +#define ELISTIZE(CLASSNAME) \ + \ + /*********************************************************************** \ + * CLASSNAME##_zapper \ + * \ + * A function which can delete a CLASSNAME element. This is passed to the \ + * generic clear list member function so that when a list is cleared the \ + * elements on the list are properly destroyed from the base class, even \ + * though we don't use a virtual destructor function. \ + **********************************************************************/ \ + \ + DLLSYM void CLASSNAME##_zapper(ELIST_LINK *link) { \ + delete reinterpret_cast(link); \ + } \ + \ + /* Become a deep copy of src_list*/ \ + void CLASSNAME##_LIST::deep_copy(const CLASSNAME##_LIST *src_list, \ + CLASSNAME *(*copier)(const CLASSNAME *)) { \ + CLASSNAME##_IT from_it(const_cast(src_list)); \ + CLASSNAME##_IT to_it(this); \ + \ + for (from_it.mark_cycle_pt(); !from_it.cycled_list(); from_it.forward()) \ + to_it.add_after_then_move((*copier)(from_it.data())); \ + } #endif diff --git a/ccutil/elst2.cpp b/ccutil/elst2.cpp index 30cedec17b..0d4960ed49 100644 --- a/ccutil/elst2.cpp +++ b/ccutil/elst2.cpp @@ -27,7 +27,7 @@ **********************************************************************/ /*********************************************************************** - * ELIST2::internal_clear + * ELIST2::internal_clear * * Used by the destructor and the "clear" member function of derived list * classes to destroy all the elements on the list. @@ -58,7 +58,7 @@ void (*zapper) (ELIST2_LINK *)) { } /*********************************************************************** - * ELIST2::assign_to_sublist + * ELIST2::assign_to_sublist * * The list is set to a sublist of another list. "This" list must be empty * before this function is invoked. The two iterators passed must refer to @@ -82,9 +82,8 @@ void ELIST2::assign_to_sublist( //to this list last = start_it->extract_sublist (end_it); } - /*********************************************************************** - * ELIST2::length + * ELIST2::length * * Return count of elements on list **********************************************************************/ @@ -98,9 +97,8 @@ inT32 ELIST2::length() const { // count elements return count; } - /*********************************************************************** - * ELIST2::sort + * ELIST2::sort * * Sort elements on list * NB If you don't like the const declarations in the comparator, coerce yours: @@ -180,7 +178,7 @@ void ELIST2::add_sorted(int comparator(const void*, const void*), **********************************************************************/ /*********************************************************************** - * ELIST2_ITERATOR::forward + * ELIST2_ITERATOR::forward * * Move the iterator to the next element of the list. * REMEMBER: ALL LISTS ARE CIRCULAR. @@ -218,9 +216,8 @@ ELIST2_LINK *ELIST2_ITERATOR::forward() { return current; } - /*********************************************************************** - * ELIST2_ITERATOR::backward + * ELIST2_ITERATOR::backward * * Move the iterator to the previous element of the list. * REMEMBER: ALL LISTS ARE CIRCULAR. @@ -257,9 +254,8 @@ ELIST2_LINK *ELIST2_ITERATOR::backward() { return current; } - /*********************************************************************** - * ELIST2_ITERATOR::data_relative + * ELIST2_ITERATOR::data_relative * * Return the data pointer to the element "offset" elements from current. * (This function can't be INLINEd because it contains a loop) @@ -289,9 +285,8 @@ ELIST2_LINK *ELIST2_ITERATOR::data_relative( //get data + or - .. return ptr; } - /*********************************************************************** - * ELIST2_ITERATOR::exchange() + * ELIST2_ITERATOR::exchange() * * Given another iterator, whose current element is a different element on * the same list list OR an element of another list, exchange the two current @@ -399,9 +394,8 @@ void ELIST2_ITERATOR::exchange( //positions of 2 li other_it->current = old_current; } - /*********************************************************************** - * ELIST2_ITERATOR::extract_sublist() + * ELIST2_ITERATOR::extract_sublist() * * This is a private member, used only by ELIST2::assign_to_sublist. * Given another iterator for the same list, extract the links from THIS to @@ -445,7 +439,7 @@ ELIST2_LINK *ELIST2_ITERATOR::extract_sublist( //fr temp_it.mark_cycle_pt (); do { //walk sublist - if (temp_it.cycled_list ()) //can't find end pt + if (temp_it.cycled_list()) // can't find end pt BAD_SUBLIST.error ("ELIST2_ITERATOR.extract_sublist", ABORT, NULL); if (temp_it.at_last ()) { diff --git a/ccutil/elst2.h b/ccutil/elst2.h index 364abd86bc..bf078fbd56 100644 --- a/ccutil/elst2.h +++ b/ccutil/elst2.h @@ -46,9 +46,9 @@ i) The duplication in source does not affect the run time code size - the **********************************************************************/ /********************************************************************** - * CLASS - ELIST2_LINK + * CLASS - ELIST2_LINK * - * Generic link class for doubly linked lists with embedded links + * Generic link class for doubly linked lists with embedded links * * Note: No destructor - elements are assumed to be destroyed EITHER after * they have been extracted from a list OR by the ELIST2 destructor which @@ -68,13 +68,13 @@ class DLLSYM ELIST2_LINK prev = next = NULL; } - ELIST2_LINK( //copy constructor - const ELIST2_LINK &) { //don't copy link + ELIST2_LINK( // copy constructor + const ELIST2_LINK &) { // don't copy link prev = next = NULL; } - void operator= ( //don't copy links - const ELIST2_LINK &) { + void operator=( // don't copy links + const ELIST2_LINK &) { prev = next = NULL; } }; @@ -142,9 +142,10 @@ class DLLSYM ELIST2 }; /*********************************************************************** - * CLASS - ELIST2_ITERATOR + * CLASS - ELIST2_ITERATOR * - * Generic iterator class for doubly linked lists with embedded links + * Generic iterator class for doubly linked lists with embedded + *links **********************************************************************/ class DLLSYM ELIST2_ITERATOR @@ -240,8 +241,8 @@ class DLLSYM ELIST2_ITERATOR BOOL8 cycled_list(); //Completed a cycle? - void add_to_end( //add at end & - ELIST2_LINK *new_link); //don't move + void add_to_end( // add at end & + ELIST2_LINK *new_link); // don't move void exchange( //positions of 2 links ELIST2_ITERATOR *other_it); //other iterator @@ -255,7 +256,7 @@ class DLLSYM ELIST2_ITERATOR }; /*********************************************************************** - * ELIST2_ITERATOR::set_to_list + * ELIST2_ITERATOR::set_to_list * * (Re-)initialise the iterator to point to the start of the list_to_iterate * over. @@ -279,9 +280,8 @@ inline void ELIST2_ITERATOR::set_to_list( //change list ex_current_was_cycle_pt = FALSE; } - /*********************************************************************** - * ELIST2_ITERATOR::ELIST2_ITERATOR + * ELIST2_ITERATOR::ELIST2_ITERATOR * * CONSTRUCTOR - set iterator to specified list; **********************************************************************/ @@ -290,9 +290,8 @@ inline ELIST2_ITERATOR::ELIST2_ITERATOR(ELIST2 *list_to_iterate) { set_to_list(list_to_iterate); } - /*********************************************************************** - * ELIST2_ITERATOR::add_after_then_move + * ELIST2_ITERATOR::add_after_then_move * * Add a new element to the list after the current element and move the * iterator to the new element. @@ -339,9 +338,8 @@ inline void ELIST2_ITERATOR::add_after_then_move( // element to add current = new_element; } - /*********************************************************************** - * ELIST2_ITERATOR::add_after_stay_put + * ELIST2_ITERATOR::add_after_stay_put * * Add a new element to the list after the current element but do not move * the iterator to the new element. @@ -391,9 +389,8 @@ inline void ELIST2_ITERATOR::add_after_stay_put( // element to add } } - /*********************************************************************** - * ELIST2_ITERATOR::add_before_then_move + * ELIST2_ITERATOR::add_before_then_move * * Add a new element to the list before the current element and move the * iterator to the new element. @@ -438,9 +435,8 @@ inline void ELIST2_ITERATOR::add_before_then_move( // element to add current = new_element; } - /*********************************************************************** - * ELIST2_ITERATOR::add_before_stay_put + * ELIST2_ITERATOR::add_before_stay_put * * Add a new element to the list before the current element but don't move the * iterator to the new element. @@ -486,11 +482,11 @@ inline void ELIST2_ITERATOR::add_before_stay_put( // element to add } } - /*********************************************************************** - * ELIST2_ITERATOR::add_list_after + * ELIST2_ITERATOR::add_list_after * - * Insert another list to this list after the current element but don't move the + * Insert another list to this list after the current element but don't move + *the * iterator. **********************************************************************/ @@ -537,9 +533,8 @@ inline void ELIST2_ITERATOR::add_list_after(ELIST2 *list_to_add) { } } - /*********************************************************************** - * ELIST2_ITERATOR::add_list_before + * ELIST2_ITERATOR::add_list_before * * Insert another list to this list before the current element. Move the * iterator to the start of the inserted elements @@ -586,9 +581,8 @@ inline void ELIST2_ITERATOR::add_list_before(ELIST2 *list_to_add) { } } - /*********************************************************************** - * ELIST2_ITERATOR::extract + * ELIST2_ITERATOR::extract * * Do extraction by removing current from the list, returning it to the * caller, but NOT updating the iterator. (So that any calling loop can do @@ -631,9 +625,8 @@ inline ELIST2_LINK *ELIST2_ITERATOR::extract() { return extracted_link; } - /*********************************************************************** - * ELIST2_ITERATOR::move_to_first() + * ELIST2_ITERATOR::move_to_first() * * Move current so that it is set to the start of the list. * Return data just in case anyone wants it. @@ -651,9 +644,8 @@ inline ELIST2_LINK *ELIST2_ITERATOR::move_to_first() { return current; } - /*********************************************************************** - * ELIST2_ITERATOR::move_to_last() + * ELIST2_ITERATOR::move_to_last() * * Move current so that it is set to the end of the list. * Return data just in case anyone wants it. @@ -671,9 +663,8 @@ inline ELIST2_LINK *ELIST2_ITERATOR::move_to_last() { return current; } - /*********************************************************************** - * ELIST2_ITERATOR::mark_cycle_pt() + * ELIST2_ITERATOR::mark_cycle_pt() * * Remember the current location so that we can tell whether we've returned * to this point later. @@ -696,9 +687,8 @@ inline void ELIST2_ITERATOR::mark_cycle_pt() { started_cycling = FALSE; } - /*********************************************************************** - * ELIST2_ITERATOR::at_first() + * ELIST2_ITERATOR::at_first() * * Are we at the start of the list? * @@ -716,9 +706,8 @@ inline BOOL8 ELIST2_ITERATOR::at_first() { !ex_current_was_last)); //first and last } - /*********************************************************************** - * ELIST2_ITERATOR::at_last() + * ELIST2_ITERATOR::at_last() * * Are we at the end of the list? * @@ -736,9 +725,8 @@ inline BOOL8 ELIST2_ITERATOR::at_last() { ex_current_was_last)); //first and last } - /*********************************************************************** - * ELIST2_ITERATOR::cycled_list() + * ELIST2_ITERATOR::cycled_list() * * Have we returned to the cycle_pt since it was set? * @@ -754,9 +742,8 @@ inline BOOL8 ELIST2_ITERATOR::cycled_list() { } - /*********************************************************************** - * ELIST2_ITERATOR::length() + * ELIST2_ITERATOR::length() * * Return the length of the list * @@ -771,9 +758,8 @@ inline inT32 ELIST2_ITERATOR::length() { return list->length (); } - /*********************************************************************** - * ELIST2_ITERATOR::sort() + * ELIST2_ITERATOR::sort() * * Sort the elements of the list, then reposition at the start. * @@ -792,9 +778,8 @@ const void *, const void *)) { move_to_first(); } - /*********************************************************************** - * ELIST2_ITERATOR::add_to_end + * ELIST2_ITERATOR::add_to_end * * Add a new element to the end of the list without moving the iterator. * This is provided because a single linked list cannot move to the last as @@ -854,7 +839,7 @@ will NOT work correctly for classes derived from this. The macro generates: - An element deletion function: CLASSNAME##_zapper - - An E_LIST2 subclass: CLASSNAME##_LIST + - An E_LIST2 subclass: CLASSNAME##_LIST - An E_LIST2_ITERATOR subclass: CLASSNAME##_IT @@ -873,132 +858,132 @@ ELIST2IZEH is a concatenation of 3 fragments ELIST2IZEH_A, ELIST2IZEH_B and ELIST2IZEH_C. ***********************************************************************/ -#define ELIST2IZEH_A( CLASSNAME ) \ - \ -extern DLLSYM void CLASSNAME##_zapper( /*delete a link*/ \ -ELIST2_LINK* link); /*link to delete*/ - -#define ELIST2IZEH_B( CLASSNAME ) \ - \ -/*********************************************************************** \ -* CLASS - CLASSNAME##_LIST \ -* \ -* List class for class CLASSNAME \ -* \ -**********************************************************************/ \ - \ -class DLLSYM CLASSNAME##_LIST : public ELIST2 \ -{ \ -public: \ - CLASSNAME##_LIST():ELIST2() {} \ - /* constructor */ \ - \ - CLASSNAME##_LIST( /* don't construct */ \ - const CLASSNAME##_LIST&) /*by initial assign*/\ - { DONT_CONSTRUCT_LIST_BY_COPY.error( QUOTE_IT( CLASSNAME##_LIST ), \ - ABORT, NULL ); } \ - \ -void clear() /* delete elements */\ - { ELIST2::internal_clear( &CLASSNAME##_zapper ); } \ - \ - ~CLASSNAME##_LIST() /* destructor */ \ - { clear(); } \ -\ -/* Become a deep copy of src_list*/ \ -void deep_copy(const CLASSNAME##_LIST* src_list, \ - CLASSNAME* (*copier)(const CLASSNAME*)); \ -\ -void operator=( /* prevent assign */ \ - const CLASSNAME##_LIST&) \ - { DONT_ASSIGN_LISTS.error( QUOTE_IT( CLASSNAME##_LIST ), \ - ABORT, NULL ); } - -#define ELIST2IZEH_C( CLASSNAME ) \ -}; \ - \ - \ - \ -/*********************************************************************** \ -* CLASS - CLASSNAME##_IT \ -* \ -* Iterator class for class CLASSNAME##_LIST \ -* \ -* Note: We don't need to coerce pointers to member functions input \ -* parameters as these are automatically converted to the type of the base \ -* type. ("A ptr to a class may be converted to a pointer to a public base \ -* class of that class") \ -**********************************************************************/ \ - \ -class DLLSYM CLASSNAME##_IT : public ELIST2_ITERATOR \ -{ \ -public: \ - CLASSNAME##_IT():ELIST2_ITERATOR(){} \ - \ - CLASSNAME##_IT( \ -CLASSNAME##_LIST* list):ELIST2_ITERATOR(list){} \ - \ - CLASSNAME* data() \ - { return (CLASSNAME*) ELIST2_ITERATOR::data(); } \ - \ - CLASSNAME* data_relative( \ - inT8 offset) \ - { return (CLASSNAME*) ELIST2_ITERATOR::data_relative( offset ); } \ - \ - CLASSNAME* forward() \ - { return (CLASSNAME*) ELIST2_ITERATOR::forward(); } \ - \ - CLASSNAME* backward() \ - { return (CLASSNAME*) ELIST2_ITERATOR::backward(); } \ - \ - CLASSNAME* extract() \ - { return (CLASSNAME*) ELIST2_ITERATOR::extract(); } \ - \ - CLASSNAME* move_to_first() \ - { return (CLASSNAME*) ELIST2_ITERATOR::move_to_first(); } \ - \ - CLASSNAME* move_to_last() \ - { return (CLASSNAME*) ELIST2_ITERATOR::move_to_last(); } \ -}; - -#define ELIST2IZEH( CLASSNAME ) \ - \ -ELIST2IZEH_A( CLASSNAME ) \ - \ -ELIST2IZEH_B( CLASSNAME ) \ - \ -ELIST2IZEH_C( CLASSNAME ) +#define ELIST2IZEH_A(CLASSNAME) \ + \ + extern DLLSYM void CLASSNAME##_zapper( /*delete a link*/ \ + ELIST2_LINK *link); /*link to delete*/ + +#define ELIST2IZEH_B(CLASSNAME) \ + \ + /*********************************************************************** \ + * CLASS - \ + *CLASSNAME##_LIST \ + * \ + * List class for class \ + *CLASSNAME \ + * \ + **********************************************************************/ \ + \ + class DLLSYM CLASSNAME##_LIST : public ELIST2 { \ + public: \ + CLASSNAME##_LIST() : ELIST2() {} \ + /* constructor */ \ + \ + CLASSNAME##_LIST( /* don't construct */ \ + const CLASSNAME##_LIST &) /*by initial assign*/ \ + { \ + DONT_CONSTRUCT_LIST_BY_COPY.error(QUOTE_IT(CLASSNAME##_LIST), ABORT, \ + NULL); \ + } \ + \ + void clear() /* delete elements */ \ + { \ + ELIST2::internal_clear(&CLASSNAME##_zapper); \ + } \ + \ + ~CLASSNAME##_LIST() /* destructor */ \ + { \ + clear(); \ + } \ + \ + /* Become a deep copy of src_list*/ \ + void deep_copy(const CLASSNAME##_LIST *src_list, \ + CLASSNAME *(*copier)(const CLASSNAME *)); \ + \ + void operator=(/* prevent assign */ \ + const CLASSNAME##_LIST &) { \ + DONT_ASSIGN_LISTS.error(QUOTE_IT(CLASSNAME##_LIST), ABORT, NULL); \ + } +#define ELIST2IZEH_C(CLASSNAME) \ + } \ + ; \ + \ + /*********************************************************************** \ + * CLASS - CLASSNAME##_IT \ + * \ + * Iterator class for class CLASSNAME##_LIST \ + * \ + * Note: We don't need to coerce pointers to member functions input \ + * parameters as these are automatically converted to the type of the base \ + * type. ("A ptr to a class may be converted to a pointer to a public base \ + * class of that class") \ + **********************************************************************/ \ + \ + class DLLSYM CLASSNAME##_IT : public ELIST2_ITERATOR { \ + public: \ + CLASSNAME##_IT() : ELIST2_ITERATOR() {} \ + \ + CLASSNAME##_IT(CLASSNAME##_LIST *list) : ELIST2_ITERATOR(list) {} \ + \ + CLASSNAME *data() { return (CLASSNAME *)ELIST2_ITERATOR::data(); } \ + \ + CLASSNAME *data_relative(inT8 offset) { \ + return (CLASSNAME *)ELIST2_ITERATOR::data_relative(offset); \ + } \ + \ + CLASSNAME *forward() { return (CLASSNAME *)ELIST2_ITERATOR::forward(); } \ + \ + CLASSNAME *backward() { return (CLASSNAME *)ELIST2_ITERATOR::backward(); } \ + \ + CLASSNAME *extract() { return (CLASSNAME *)ELIST2_ITERATOR::extract(); } \ + \ + CLASSNAME *move_to_first() { \ + return (CLASSNAME *)ELIST2_ITERATOR::move_to_first(); \ + } \ + \ + CLASSNAME *move_to_last() { \ + return (CLASSNAME *)ELIST2_ITERATOR::move_to_last(); \ + } \ + }; + +#define ELIST2IZEH(CLASSNAME) \ + \ + ELIST2IZEH_A(CLASSNAME) \ + \ + ELIST2IZEH_B(CLASSNAME) \ + \ + ELIST2IZEH_C(CLASSNAME) /*********************************************************************** ELIST2IZE( CLASSNAME ) MACRO ***********************************************************************/ -#define ELIST2IZE( CLASSNAME ) \ - \ -/*********************************************************************** \ -* CLASSNAME##_zapper \ -* \ -* A function which can delete a CLASSNAME element. This is passed to the \ -* generic clear list member function so that when a list is cleared the \ -* elements on the list are properly destroyed from the base class, even \ -* though we don't use a virtual destructor function. \ -**********************************************************************/ \ - \ -DLLSYM void CLASSNAME##_zapper( /*delete a link*/ \ -ELIST2_LINK* link) /*link to delete*/ \ -{ \ -delete (CLASSNAME *) link; \ -} \ -\ -/* Become a deep copy of src_list*/ \ -void CLASSNAME##_LIST::deep_copy(const CLASSNAME##_LIST* src_list, \ - CLASSNAME* (*copier)(const CLASSNAME*)) { \ -\ - CLASSNAME##_IT from_it(const_cast(src_list)); \ - CLASSNAME##_IT to_it(this); \ -\ - for (from_it.mark_cycle_pt(); !from_it.cycled_list(); from_it.forward()) \ - to_it.add_after_then_move((*copier)(from_it.data())); \ -} +#define ELIST2IZE(CLASSNAME) \ + \ + /*********************************************************************** \ + * CLASSNAME##_zapper \ + * \ + * A function which can delete a CLASSNAME element. This is passed to the \ + * generic clear list member function so that when a list is cleared the \ + * elements on the list are properly destroyed from the base class, even \ + * though we don't use a virtual destructor function. \ + **********************************************************************/ \ + \ + DLLSYM void CLASSNAME##_zapper( /*delete a link*/ \ + ELIST2_LINK *link) /*link to delete*/ \ + { \ + delete (CLASSNAME *)link; \ + } \ + \ + /* Become a deep copy of src_list*/ \ + void CLASSNAME##_LIST::deep_copy(const CLASSNAME##_LIST *src_list, \ + CLASSNAME *(*copier)(const CLASSNAME *)) { \ + CLASSNAME##_IT from_it(const_cast(src_list)); \ + CLASSNAME##_IT to_it(this); \ + \ + for (from_it.mark_cycle_pt(); !from_it.cycled_list(); from_it.forward()) \ + to_it.add_after_then_move((*copier)(from_it.data())); \ + } #endif diff --git a/ccutil/errcode.h b/ccutil/errcode.h index d690240036..2f31a7b9ae 100644 --- a/ccutil/errcode.h +++ b/ccutil/errcode.h @@ -87,11 +87,10 @@ const ERRCODE ASSERT_FAILED = "Assert failed"; __FILE__, __LINE__); \ } -#define ASSERT_HOST_MSG(x, ...) if (!(x)) \ - { \ - tprintf(__VA_ARGS__); \ - ASSERT_FAILED.error(#x, ABORT, "in file %s, line %d", \ - __FILE__, __LINE__); \ +#define ASSERT_HOST_MSG(x, ...) \ + if (!(x)) { \ + tprintf(__VA_ARGS__); \ + ASSERT_FAILED.error(#x, ABORT, "in file %s, line %d", __FILE__, __LINE__); \ } void signal_exit(int signal_code); diff --git a/ccutil/genericheap.h b/ccutil/genericheap.h index bb5f8ddc79..ccf273b33a 100644 --- a/ccutil/genericheap.h +++ b/ccutil/genericheap.h @@ -108,6 +108,8 @@ class GenericHeap { const Pair& PeekTop() const { return heap_[0]; } + // Get the value of the worst (largest, defined by operator< ) element. + const Pair& PeekWorst() const { return heap_[IndexOfWorst()]; } // Removes the top element of the heap. If entry is not NULL, the element // is copied into *entry, otherwise it is discarded. @@ -136,22 +138,12 @@ class GenericHeap { // not NULL, the element is copied into *entry, otherwise it is discarded. // Time = O(n). Returns false if the heap was already empty. bool PopWorst(Pair* entry) { - int heap_size = heap_.size(); - if (heap_size == 0) return false; // It cannot be empty! - - // Find the maximum element. Its index is guaranteed to be greater than - // the index of the parent of the last element, since by the heap invariant - // the parent must be less than or equal to the children. - int worst_index = heap_size - 1; - int end_parent = ParentNode(worst_index); - for (int i = worst_index - 1; i > end_parent; --i) { - if (heap_[worst_index] < heap_[i]) - worst_index = i; - } + int worst_index = IndexOfWorst(); + if (worst_index < 0) return false; // It cannot be empty! // Extract the worst element from the heap, leaving a hole at worst_index. if (entry != NULL) *entry = heap_[worst_index]; - --heap_size; + int heap_size = heap_.size() - 1; if (heap_size > 0) { // Sift the hole upwards to match the last element of the heap_ Pair hole_pair = heap_[heap_size]; @@ -162,6 +154,22 @@ class GenericHeap { return true; } + // Returns the index of the worst element. Time = O(n/2). + int IndexOfWorst() const { + int heap_size = heap_.size(); + if (heap_size == 0) return -1; // It cannot be empty! + + // Find the maximum element. Its index is guaranteed to be greater than + // the index of the parent of the last element, since by the heap invariant + // the parent must be less than or equal to the children. + int worst_index = heap_size - 1; + int end_parent = ParentNode(worst_index); + for (int i = worst_index - 1; i > end_parent; --i) { + if (heap_[worst_index] < heap_[i]) worst_index = i; + } + return worst_index; + } + // The pointed-to Pair has changed its key value, so the location of pair // is reshuffled to maintain the heap invariant. // Must be a valid pointer to an element of the heap_! diff --git a/ccutil/genericvector.h b/ccutil/genericvector.h index d867d8929b..3a70e21ce0 100644 --- a/ccutil/genericvector.h +++ b/ccutil/genericvector.h @@ -174,6 +174,8 @@ class GenericVector { // If swap is true, assumes a big/little-endian swap is needed. bool DeSerialize(bool swap, FILE* fp); bool DeSerialize(bool swap, tesseract::TFile* fp); + // Skips the deserialization of the vector. + static bool SkipDeSerialize(bool swap, tesseract::TFile* fp); // Writes a vector of classes to the given file. Assumes the existence of // bool T::Serialize(FILE* fp) const that returns false in case of error. // Returns false in case of error. @@ -186,6 +188,8 @@ class GenericVector { // If swap is true, assumes a big/little-endian swap is needed. bool DeSerializeClasses(bool swap, FILE* fp); bool DeSerializeClasses(bool swap, tesseract::TFile* fp); + // Calls SkipDeSerialize on the elements of the vector. + static bool SkipDeSerializeClasses(bool swap, tesseract::TFile* fp); // Allocates a new array of double the current_size, copies over the // information from data to the new location, deletes data and returns @@ -238,14 +242,13 @@ class GenericVector { int binary_search(const T& target) const { int bottom = 0; int top = size_used_; - do { + while (top - bottom > 1) { int middle = (bottom + top) / 2; if (data_[middle] > target) top = middle; else bottom = middle; } - while (top - bottom > 1); return bottom; } @@ -361,7 +364,7 @@ inline bool LoadDataFromFile(const STRING& filename, size_t size = ftell(fp); fseek(fp, 0, SEEK_SET); // Pad with a 0, just in case we treat the result as a string. - data->init_to_size((int)size + 1, 0); + data->init_to_size(static_cast(size) + 1, 0); bool result = fread(&(*data)[0], 1, size, fp) == size; fclose(fp); return result; @@ -556,34 +559,54 @@ class PointerVector : public GenericVector { } bool DeSerialize(bool swap, TFile* fp) { inT32 reserved; - if (fp->FRead(&reserved, sizeof(reserved), 1) != 1) return false; - if (swap) Reverse32(&reserved); + if (!DeSerializeSize(swap, fp, &reserved)) return false; GenericVector::reserve(reserved); truncate(0); for (int i = 0; i < reserved; ++i) { - inT8 non_null; - if (fp->FRead(&non_null, sizeof(non_null), 1) != 1) return false; - T* item = NULL; - if (non_null) { - item = new T; - if (!item->DeSerialize(swap, fp)) { - delete item; - return false; - } - this->push_back(item); - } else { - // Null elements should keep their place in the vector. - this->push_back(NULL); + if (!DeSerializeElement(swap, fp)) return false; + } + return true; + } + // Enables deserialization of a selection of elements. Note that in order to + // retain the integrity of the stream, the caller must call some combination + // of DeSerializeElement and DeSerializeSkip of the exact number returned in + // *size, assuming a true return. + static bool DeSerializeSize(bool swap, TFile* fp, inT32* size) { + if (fp->FRead(size, sizeof(*size), 1) != 1) return false; + if (swap) Reverse32(size); + return true; + } + // Reads and appends to the vector the next element of the serialization. + bool DeSerializeElement(bool swap, TFile* fp) { + inT8 non_null; + if (fp->FRead(&non_null, sizeof(non_null), 1) != 1) return false; + T* item = NULL; + if (non_null) { + item = new T; + if (!item->DeSerialize(swap, fp)) { + delete item; + return false; } + this->push_back(item); + } else { + // Null elements should keep their place in the vector. + this->push_back(NULL); + } + return true; + } + // Skips the next element of the serialization. + static bool DeSerializeSkip(bool swap, TFile* fp) { + inT8 non_null; + if (fp->FRead(&non_null, sizeof(non_null), 1) != 1) return false; + if (non_null) { + if (!T::SkipDeSerialize(swap, fp)) return false; } return true; } // Sorts the items pointed to by the members of this vector using // t::operator<(). - void sort() { - sort(&sort_ptr_cmp); - } + void sort() { this->GenericVector::sort(&sort_ptr_cmp); } }; } // namespace tesseract @@ -926,6 +949,13 @@ bool GenericVector::DeSerialize(bool swap, tesseract::TFile* fp) { } return true; } +template +bool GenericVector::SkipDeSerialize(bool swap, tesseract::TFile* fp) { + inT32 reserved; + if (fp->FRead(&reserved, sizeof(reserved), 1) != 1) return false; + if (swap) Reverse32(&reserved); + return fp->FRead(NULL, sizeof(T), reserved) == reserved; +} // Writes a vector of classes to the given file. Assumes the existence of // bool T::Serialize(FILE* fp) const that returns false in case of error. @@ -976,6 +1006,16 @@ bool GenericVector::DeSerializeClasses(bool swap, tesseract::TFile* fp) { } return true; } +template +bool GenericVector::SkipDeSerializeClasses(bool swap, tesseract::TFile* fp) { + uinT32 reserved; + if (fp->FRead(&reserved, sizeof(reserved), 1) != 1) return false; + if (swap) Reverse32(&reserved); + for (int i = 0; i < reserved; ++i) { + if (!T::SkipDeSerialize(swap, fp)) return false; + } + return true; +} // This method clear the current object, then, does a shallow copy of // its argument, and finally invalidates its argument. diff --git a/ccutil/globaloc.h b/ccutil/globaloc.h index 41438194c8..60d6b73ae2 100644 --- a/ccutil/globaloc.h +++ b/ccutil/globaloc.h @@ -29,7 +29,7 @@ void SavePixForCrash(int resolution, Pix* pix); void signal_exit(int signal_code); -void err_exit(); +void TESS_API err_exit(); void set_global_loc_code(int loc_code); diff --git a/ccutil/hashfn.h b/ccutil/hashfn.h index be211b0731..73e15be9a8 100644 --- a/ccutil/hashfn.h +++ b/ccutil/hashfn.h @@ -20,16 +20,15 @@ #ifndef HASHFN_H #define HASHFN_H -#ifdef USE_STD_NAMESPACE #if (__cplusplus >= 201103L) || defined(_MSC_VER) // Visual Studio #include #include -#define hash_map std::unordered_map -#if (_MSC_VER >= 1500 && _MSC_VER < 1600) // Visual Studio 2008 -using namespace std::tr1; +#if defined(_MSC_VER) && (_MSC_VER >= 1500 && _MSC_VER < 1600) // VS 2008 +#define TessHashMap std::tr1::unordered_map +#define TessHashSet std::tr1::unordered_set #else // _MSC_VER -using std::unordered_map; -using std::unordered_set; +#define TessHashMap std::unordered_map +#define TessHashSet std::unordered_set #include #define SmartPtr std::unique_ptr #define HAVE_UNIQUE_PTR @@ -41,23 +40,14 @@ using std::unordered_set; #include using __gnu_cxx::hash_map; using __gnu_cxx::hash_set; -#define unordered_map hash_map -#define unordered_set hash_set +#define TessHashMap __gnu_cxx::hash_map +#define TessHashSet __gnu_cxx::hash_set #else #include #include +#define TessHashMap hash_map +#define TessHashSet :hash_set #endif // gcc -#elif (__clang__) -#include -#include -#define hash_map std::unordered_map -#define unordered_set std::unordered_set -#else // USE_STD_NAMESPACE -#include -#include -#define unordered_map hash_map -#define unordered_set hash_set -#endif // USE_STD_NAMESPACE #ifndef HAVE_UNIQUE_PTR // Trivial smart ptr. Expand to add features of std::unique_ptr as required. @@ -73,7 +63,7 @@ template class SmartPtr { return ptr_; } void reset(T* ptr) { - if (ptr_ != NULL) delete ptr_; + delete ptr_; ptr_ = ptr; } bool operator==(const T* ptr) const { diff --git a/ccutil/helpers.h b/ccutil/helpers.h index 51dd3b0b44..a2276bc451 100644 --- a/ccutil/helpers.h +++ b/ccutil/helpers.h @@ -73,7 +73,7 @@ class TRand { // Remove newline (if any) at the end of the string. inline void chomp_string(char *str) { - int last_index = (int)strlen(str) - 1; + int last_index = static_cast(strlen(str)) - 1; while (last_index >= 0 && (str[last_index] == '\n' || str[last_index] == '\r')) { str[last_index--] = '\0'; diff --git a/ccutil/lsterr.h b/ccutil/lsterr.h index 42ed07e326..82497ae80d 100644 --- a/ccutil/lsterr.h +++ b/ccutil/lsterr.h @@ -38,6 +38,6 @@ const ERRCODE NULL_PREV = "Previous element on the list is NULL"; const ERRCODE EMPTY_LIST = "List is empty"; const ERRCODE BAD_PARAMETER = "List parameter error"; const ERRCODE STILL_LINKED = -"Attempting to add an element with non NULL links, to a list"; + "Attempting to add an element with non NULL links, to a list"; #endif #endif diff --git a/ccutil/mainblk.cpp b/ccutil/mainblk.cpp index aa73c55002..36becce9ec 100644 --- a/ccutil/mainblk.cpp +++ b/ccutil/mainblk.cpp @@ -74,7 +74,7 @@ void CCUtil::main_setup(const char *argv0, const char *basename) { #endif /* _WIN32 */ #if defined(TESSDATA_PREFIX) } else { - /* Use tessdata prefix which was compiled in. */ +/* Use tessdata prefix which was compiled in. */ #define _STR(a) #a #define _XSTR(a) _STR(a) datadir = _XSTR(TESSDATA_PREFIX); diff --git a/ccutil/ocrclass.h b/ccutil/ocrclass.h index 3175a6d29a..cb83c6d6b6 100644 --- a/ccutil/ocrclass.h +++ b/ccutil/ocrclass.h @@ -1,7 +1,7 @@ /********************************************************************** * File: ocrclass.h * Description: Class definitions and constants for the OCR API. - * Author: Hewlett-Packard Co + * Author: Hewlett-Packard Co * * (C) Copyright 1996, Hewlett-Packard Co. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -29,7 +29,6 @@ #ifndef __GNUC__ #ifdef _WIN32 -#include #include "gettimeofday.h" #endif #else @@ -110,28 +109,35 @@ typedef struct { /*single character */ * user words found. If it returns true then operation is cancelled. **********************************************************************/ typedef bool (*CANCEL_FUNC)(void* cancel_this, int words); -typedef bool (*PROGRESS_FUNC)(int progress, - int left, int right, int top, int bottom); +typedef bool (*PROGRESS_FUNC)(int progress, int left, int right, int top, + int bottom); class ETEXT_DESC { // output header public: - inT16 count; /// chars in this buffer(0) - inT16 progress; /// percent complete increasing (0-100) + inT16 count; /// chars in this buffer(0) + inT16 progress; /// percent complete increasing (0-100) /** Progress monitor covers word recognition and it does not cover layout * analysis. * See Ray comment in https://github.com/tesseract-ocr/tesseract/pull/27 */ - inT8 more_to_come; /// true if not last - volatile inT8 ocr_alive; /// ocr sets to 1, HP 0 - inT8 err_code; /// for errcode use - CANCEL_FUNC cancel; /// returns true to cancel - PROGRESS_FUNC progress_callback; /// called whenever progress increases - void* cancel_this; /// this or other data for cancel - struct timeval end_time; /** time to stop. expected to be set only by call - * to set_deadline_msecs() */ - EANYCODE_CHAR text[1]; /// character data + inT8 more_to_come; /// true if not last + volatile inT8 ocr_alive; /// ocr sets to 1, HP 0 + inT8 err_code; /// for errcode use + CANCEL_FUNC cancel; /// returns true to cancel + PROGRESS_FUNC progress_callback; /// called whenever progress increases + void* cancel_this; /// this or other data for cancel + struct timeval end_time; /// Time to stop. Expected to be set only + /// by call to set_deadline_msecs(). + EANYCODE_CHAR text[1]; /// character data - ETEXT_DESC() : count(0), progress(0), more_to_come(0), ocr_alive(0), - err_code(0), cancel(NULL), cancel_this(NULL) { + ETEXT_DESC() + : count(0), + progress(0), + more_to_come(0), + ocr_alive(0), + err_code(0), + cancel(NULL), + progress_callback(NULL), + cancel_this(NULL) { end_time.tv_sec = 0; end_time.tv_usec = 0; } diff --git a/ccutil/params.cpp b/ccutil/params.cpp index 9b604fbef4..30604f1129 100644 --- a/ccutil/params.cpp +++ b/ccutil/params.cpp @@ -31,8 +31,7 @@ #define EQUAL '=' tesseract::ParamsVectors *GlobalParams() { - static tesseract::ParamsVectors global_params = - tesseract::ParamsVectors(); + static tesseract::ParamsVectors global_params = tesseract::ParamsVectors(); return &global_params; } @@ -73,7 +72,7 @@ bool ParamUtils::ReadParamsFromFp(FILE *fp, inT64 end_offset, while ((end_offset < 0 || ftell(fp) < end_offset) && fgets(line, MAX_PATH, fp)) { - if (line[0] != '\n' && line[0] != '#') { + if (line[0] != '\r' && line[0] != '\n' && line[0] != '#') { chomp_string(line); // remove newline for (valptr = line; *valptr && *valptr != ' ' && *valptr != '\t'; valptr++); diff --git a/ccutil/params.h b/ccutil/params.h index d49ce3ff8a..c168dbe4c6 100644 --- a/ccutil/params.h +++ b/ccutil/params.h @@ -55,7 +55,7 @@ class ParamUtils { // ORed or ANDed with any current values. // Blank lines and lines beginning # are ignored. // Values may have any whitespace after the name and are the rest of line. - static bool ReadParamsFile( + static bool TESS_API ReadParamsFile( const char *file, // filename to read SetParamConstraint constraint, ParamsVectors *member_params); @@ -252,7 +252,7 @@ class DoubleParam : public Param { // // TODO(daria): remove GlobalParams() when all global Tesseract // parameters are converted to members. -tesseract::ParamsVectors *GlobalParams(); +tesseract::ParamsVectors TESS_API *GlobalParams(); /************************************************************************* * Note on defining parameters. diff --git a/ccutil/platform.h b/ccutil/platform.h index 219f9e31e0..dd9be87ac6 100644 --- a/ccutil/platform.h +++ b/ccutil/platform.h @@ -24,6 +24,10 @@ #define DLLSYM #ifdef _WIN32 +#ifndef NOMINMAX +#define NOMINMAX +#endif /* NOMINMAX */ +#define WIN32_LEAN_AND_MEAN #ifdef __GNUC__ #define ultoa _ultoa #endif /* __GNUC__ */ diff --git a/ccutil/serialis.cpp b/ccutil/serialis.cpp index ff3b278a7e..d1eed58465 100644 --- a/ccutil/serialis.cpp +++ b/ccutil/serialis.cpp @@ -95,7 +95,7 @@ int TFile::FRead(void* buffer, int size, int count) { char* char_buffer = reinterpret_cast(buffer); if (data_->size() - offset_ < required_size) required_size = data_->size() - offset_; - if (required_size > 0) + if (required_size > 0 && char_buffer != NULL) memcpy(char_buffer, &(*data_)[offset_], required_size); offset_ += required_size; return required_size / size; diff --git a/ccutil/strngs.cpp b/ccutil/strngs.cpp index ff3bbac287..5a9cfd0d48 100644 --- a/ccutil/strngs.cpp +++ b/ccutil/strngs.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: strngs.c (Formerly strings.c) * Description: STRING class functions. - * Author: Ray Smith - * Created: Fri Feb 15 09:13:30 GMT 1991 + * Author: Ray Smith + * Created: Fri Feb 15 09:13:30 GMT 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -181,6 +181,14 @@ bool STRING::DeSerialize(bool swap, TFile* fp) { return true; } +// As DeSerialize, but only seeks past the data - hence a static method. +bool STRING::SkipDeSerialize(bool swap, tesseract::TFile* fp) { + inT32 len; + if (fp->FRead(&len, sizeof(len), 1) != 1) return false; + if (swap) ReverseN(&len, sizeof(len)); + return fp->FRead(NULL, 1, len) == len; +} + BOOL8 STRING::contains(const char c) const { return (c != '\0') && (strchr (GetCStr(), c) != NULL); } diff --git a/ccutil/strngs.h b/ccutil/strngs.h index 9308cc67c8..2e65463efd 100644 --- a/ccutil/strngs.h +++ b/ccutil/strngs.h @@ -60,6 +60,8 @@ class TESS_API STRING // Reads from the given file. Returns false in case of error. // If swap is true, assumes a big/little-endian swap is needed. bool DeSerialize(bool swap, tesseract::TFile* fp); + // As DeSerialize, but only seeks past the data - hence a static method. + static bool SkipDeSerialize(bool swap, tesseract::TFile* fp); BOOL8 contains(const char c) const; inT32 length() const; @@ -145,13 +147,11 @@ class TESS_API STRING } // returns the string data part of storage - inline char* GetCStr() { - return ((char *)data_) + sizeof(STRING_HEADER); - }; + inline char* GetCStr() { return ((char*)data_) + sizeof(STRING_HEADER); } inline const char* GetCStr() const { return ((const char *)data_) + sizeof(STRING_HEADER); - }; + } inline bool InvariantOk() const { #if STRING_IS_PROTECTED return (GetHeader()->used_ == 0) ? diff --git a/ccutil/tessdatamanager.h b/ccutil/tessdatamanager.h index fd2685a1d8..e583b70049 100644 --- a/ccutil/tessdatamanager.h +++ b/ccutil/tessdatamanager.h @@ -76,24 +76,24 @@ enum TessdataType { * kTessdataFileSuffixes[i] indicates the file suffix for * tessdata of type i (from TessdataType enum). */ -static const char * const kTessdataFileSuffixes[] = { - kLangConfigFileSuffix, // 0 - kUnicharsetFileSuffix, // 1 - kAmbigsFileSuffix, // 2 - kBuiltInTemplatesFileSuffix, // 3 - kBuiltInCutoffsFileSuffix, // 4 - kNormProtoFileSuffix, // 5 - kPuncDawgFileSuffix, // 6 - kSystemDawgFileSuffix, // 7 - kNumberDawgFileSuffix, // 8 - kFreqDawgFileSuffix, // 9 - kFixedLengthDawgsFileSuffix, // 10 // deprecated - kCubeUnicharsetFileSuffix, // 11 - kCubeSystemDawgFileSuffix, // 12 - kShapeTableFileSuffix, // 13 - kBigramDawgFileSuffix, // 14 - kUnambigDawgFileSuffix, // 15 - kParamsModelFileSuffix, // 16 +static const char *const kTessdataFileSuffixes[] = { + kLangConfigFileSuffix, // 0 + kUnicharsetFileSuffix, // 1 + kAmbigsFileSuffix, // 2 + kBuiltInTemplatesFileSuffix, // 3 + kBuiltInCutoffsFileSuffix, // 4 + kNormProtoFileSuffix, // 5 + kPuncDawgFileSuffix, // 6 + kSystemDawgFileSuffix, // 7 + kNumberDawgFileSuffix, // 8 + kFreqDawgFileSuffix, // 9 + kFixedLengthDawgsFileSuffix, // 10 // deprecated + kCubeUnicharsetFileSuffix, // 11 + kCubeSystemDawgFileSuffix, // 12 + kShapeTableFileSuffix, // 13 + kBigramDawgFileSuffix, // 14 + kUnambigDawgFileSuffix, // 15 + kParamsModelFileSuffix, // 16 }; /** @@ -101,23 +101,23 @@ static const char * const kTessdataFileSuffixes[] = { * of type i (from TessdataType enum) is text, and is binary otherwise. */ static const bool kTessdataFileIsText[] = { - true, // 0 - true, // 1 - true, // 2 - false, // 3 - true, // 4 - true, // 5 - false, // 6 - false, // 7 - false, // 8 - false, // 9 - false, // 10 // deprecated - true, // 11 - false, // 12 - false, // 13 - false, // 14 - false, // 15 - true, // 16 + true, // 0 + true, // 1 + true, // 2 + false, // 3 + true, // 4 + true, // 5 + false, // 6 + false, // 7 + false, // 8 + false, // 9 + false, // 10 // deprecated + true, // 11 + false, // 12 + false, // 13 + false, // 14 + false, // 15 + true, // 16 }; /** diff --git a/ccutil/unicharset.h b/ccutil/unicharset.h index 684655affb..eb9f463068 100644 --- a/ccutil/unicharset.h +++ b/ccutil/unicharset.h @@ -141,7 +141,7 @@ class UNICHARSET { // Custom list of characters and their ligature forms (UTF8) // These map to unicode values in the private use area (PUC) and are supported // by only few font families (eg. Wyld, Adobe Caslon Pro). - static const char* kCustomLigatures[][2]; + static TESS_API const char* kCustomLigatures[][2]; // List of strings for the SpecialUnicharCodes. Keep in sync with the enum. static const char* kSpecialUnicharCodes[SPECIAL_UNICHAR_CODES_COUNT]; @@ -177,12 +177,11 @@ class UNICHARSET { // Return the UNICHAR_ID of a given unichar representation within the // UNICHARSET. - UNICHAR_ID unichar_to_id(const char* const unichar_repr) const; + UNICHAR_ID TESS_API unichar_to_id(const char* const unichar_repr) const; // Return the UNICHAR_ID of a given unichar representation within the // UNICHARSET. Only the first length characters from unichar_repr are used. - UNICHAR_ID unichar_to_id(const char* const unichar_repr, - int length) const; + UNICHAR_ID unichar_to_id(const char* const unichar_repr, int length) const; // Return the minimum number of bytes that matches a legal UNICHAR_ID, // while leaving the rest of the string encodable. Returns 0 if the @@ -235,7 +234,7 @@ class UNICHARSET { } // Add a unichar representation to the set. - void unichar_insert(const char* const unichar_repr); + void TESS_API unichar_insert(const char* const unichar_repr); // Return true if the given unichar id exists within the set. // Relies on the fact that unichar ids are contiguous in the unicharset. @@ -245,7 +244,7 @@ class UNICHARSET { } // Return true if the given unichar representation exists within the set. - bool contains_unichar(const char* const unichar_repr) const; + bool TESS_API contains_unichar(const char* const unichar_repr) const; bool contains_unichar(const char* const unichar_repr, int length) const; // Return true if the given unichar representation corresponds to the given @@ -328,7 +327,7 @@ class UNICHARSET { // Saves the content of the UNICHARSET to the given STRING. // Returns true if the operation is successful. - bool save_to_string(STRING *str) const; + bool TESS_API save_to_string(STRING *str) const; // Load a unicharset from a unicharset file that has been loaded into // the given memory buffer. diff --git a/classify/adaptive.cpp b/classify/adaptive.cpp index 019befb4f6..a713fedc27 100644 --- a/classify/adaptive.cpp +++ b/classify/adaptive.cpp @@ -311,8 +311,8 @@ void Classify::PrintAdaptedTemplates(FILE *File, ADAPT_TEMPLATES Templates) { * @note History: Tue Mar 19 14:11:01 1991, DSJ, Created. */ ADAPT_CLASS ReadAdaptedClass(FILE *File) { - int NumTempProtos; - int NumConfigs; + int32_t NumTempProtos; + int32_t NumConfigs; int i; ADAPT_CLASS Class; TEMP_PROTO TempProto; @@ -330,7 +330,7 @@ ADAPT_CLASS ReadAdaptedClass(FILE *File) { WordsInVectorOfSize (MAX_NUM_CONFIGS), File); /* then read in the list of temporary protos */ - fread ((char *) &NumTempProtos, sizeof (int), 1, File); + fread (&NumTempProtos, sizeof(NumTempProtos), 1, File); Class->TempProtos = NIL_LIST; for (i = 0; i < NumTempProtos; i++) { TempProto = @@ -341,7 +341,7 @@ ADAPT_CLASS ReadAdaptedClass(FILE *File) { } /* then read in the adapted configs */ - fread ((char *) &NumConfigs, sizeof (int), 1, File); + fread (&NumConfigs, sizeof(NumConfigs), 1, File); for (i = 0; i < NumConfigs; i++) if (test_bit (Class->PermConfigs, i)) Class->Config[i].Perm = ReadPermConfig (File); diff --git a/classify/adaptmatch.cpp b/classify/adaptmatch.cpp index 6e995c2159..e4713a04f5 100644 --- a/classify/adaptmatch.cpp +++ b/classify/adaptmatch.cpp @@ -819,7 +819,7 @@ int Classify::GetAdaptiveFeatures(TBLOB *Blob, Features = ExtractPicoFeatures(Blob); NumFeatures = Features->NumFeatures; - if (NumFeatures > UNLIKELY_NUM_FEAT) { + if (NumFeatures == 0 || NumFeatures > UNLIKELY_NUM_FEAT) { FreeFeatureSet(Features); return 0; } @@ -908,8 +908,7 @@ void Classify::AdaptToChar(TBLOB* Blob, CLASS_ID ClassId, int FontinfoId, NumFeatures = GetAdaptiveFeatures(Blob, IntFeatures, &FloatFeatures); if (NumFeatures <= 0) { - FreeFeatureSet(FloatFeatures); - return; + return; // Features already freed by GetAdaptiveFeatures. } // Only match configs with the matching font. @@ -1008,8 +1007,6 @@ void Classify::DisplayAdaptedChar(TBLOB* blob, INT_CLASS_STRUCT* int_class) { #endif } - - /** * This routine adds the result of a classification into * Results. If the new rating is much worse than the current diff --git a/classify/classify.cpp b/classify/classify.cpp index 436efd1f2d..7c11c51f6e 100644 --- a/classify/classify.cpp +++ b/classify/classify.cpp @@ -151,8 +151,8 @@ Classify::Classify() INT_MEMBER(classify_integer_matcher_multiplier, 10, "Integer Matcher Multiplier 0-255: ", this->params()), EnableLearning(true), - INT_MEMBER(il1_adaption_test, 0, "Don't adapt to i/I at beginning of word", - this->params()), + INT_MEMBER(il1_adaption_test, 0, + "Don't adapt to i/I at beginning of word", this->params()), BOOL_MEMBER(classify_bln_numeric_mode, 0, "Assume the input is numbers [0-9].", this->params()), double_MEMBER(speckle_large_max_size, 0.30, "Max large speckle size", diff --git a/classify/cluster.cpp b/classify/cluster.cpp index b723bfa82e..1f82349552 100644 --- a/classify/cluster.cpp +++ b/classify/cluster.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: cluster.c - ** Purpose: Routines for clustering points in N-D space - ** Author: Dan Johnson - ** History: 5/29/89, DSJ, Created. + ** Filename: cluster.c + ** Purpose: Routines for clustering points in N-D space + ** Author: Dan Johnson + ** History: 5/29/89, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -390,11 +390,11 @@ double InvertMatrix(const float* input, int size, float* inv); * This routine creates a new clusterer data structure, * initializes it, and returns a pointer to it. * - * @param SampleSize number of dimensions in feature space - * @param ParamDesc description of each dimension - * @return pointer to the new clusterer data structure - * @note Exceptions: None - * @note History: 5/29/89, DSJ, Created. + * @param SampleSize number of dimensions in feature space + * @param ParamDesc description of each dimension + * @return pointer to the new clusterer data structure + * @note Exceptions: None + * @note History: 5/29/89, DSJ, Created. */ CLUSTERER * MakeClusterer (inT16 SampleSize, const PARAM_DESC ParamDesc[]) { @@ -437,7 +437,6 @@ MakeClusterer (inT16 SampleSize, const PARAM_DESC ParamDesc[]) { return Clusterer; } // MakeClusterer - /** * This routine creates a new sample data structure to hold * the specified feature. This sample is added to the clusterer @@ -445,14 +444,14 @@ MakeClusterer (inT16 SampleSize, const PARAM_DESC ParamDesc[]) { * clustered later), and a pointer to the sample is returned to * the caller. * - * @param Clusterer clusterer data structure to add sample to - * @param Feature feature to be added to clusterer - * @param CharID unique ident. of char that sample came from + * @param Clusterer clusterer data structure to add sample to + * @param Feature feature to be added to clusterer + * @param CharID unique ident. of char that sample came from * - * @return Pointer to the new sample data structure - * @note Exceptions: ALREADYCLUSTERED MakeSample can't be called after + * @return Pointer to the new sample data structure + * @note Exceptions: ALREADYCLUSTERED MakeSample can't be called after * ClusterSamples has been called - * @note History: 5/29/89, DSJ, Created. + * @note History: 5/29/89, DSJ, Created. */ SAMPLE* MakeSample(CLUSTERER * Clusterer, const FLOAT32* Feature, inT32 CharID) { @@ -490,7 +489,6 @@ SAMPLE* MakeSample(CLUSTERER * Clusterer, const FLOAT32* Feature, return (Sample); } // MakeSample - /** * This routine first checks to see if the samples in this * clusterer have already been clustered before; if so, it does @@ -505,12 +503,12 @@ SAMPLE* MakeSample(CLUSTERER * Clusterer, const FLOAT32* Feature, * list of prototypes that best represent the samples given * the constraints specified in Config. * - * @param Clusterer data struct containing samples to be clustered - * @param Config parameters which control clustering process + * @param Clusterer data struct containing samples to be clustered + * @param Config parameters which control clustering process * * @return Pointer to a list of prototypes - * @note Exceptions: None - * @note History: 5/29/89, DSJ, Created. + * @note Exceptions: None + * @note History: 5/29/89, DSJ, Created. */ LIST ClusterSamples(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { //only create cluster tree if samples have never been clustered before @@ -523,10 +521,16 @@ LIST ClusterSamples(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { //compute prototypes starting at the root node in the tree ComputePrototypes(Clusterer, Config); - return (Clusterer->ProtoList); + // We don't need the cluster pointers in the protos any more, so null them + // out, which makes it safe to delete the clusterer. + LIST proto_list = Clusterer->ProtoList; + iterate(proto_list) { + PROTOTYPE *proto = reinterpret_cast(first_node(proto_list)); + proto->Cluster = NULL; + } + return Clusterer->ProtoList; } // ClusterSamples - /** * This routine frees all of the memory allocated to the * specified data structure. It will not, however, free @@ -535,10 +539,10 @@ LIST ClusterSamples(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { * to NULL to indicate that the cluster data structures no * longer exist. Any sample lists that have been obtained * via calls to GetSamples are no longer valid. - * @param Clusterer pointer to data structure to be freed + * @param Clusterer pointer to data structure to be freed * @return None - * @note Exceptions: None - * @note History: 6/6/89, DSJ, Created. + * @note Exceptions: None + * @note History: 6/6/89, DSJ, Created. */ void FreeClusterer(CLUSTERER *Clusterer) { if (Clusterer != NULL) { @@ -558,21 +562,19 @@ void FreeClusterer(CLUSTERER *Clusterer) { } } // FreeClusterer - /** * This routine frees all of the memory allocated to the * specified list of prototypes. The clusters which are * pointed to by the prototypes are not freed. - * @param ProtoList pointer to list of prototypes to be freed + * @param ProtoList pointer to list of prototypes to be freed * @return None - * @note Exceptions: None - * @note History: 6/6/89, DSJ, Created. + * @note Exceptions: None + * @note History: 6/6/89, DSJ, Created. */ void FreeProtoList(LIST *ProtoList) { destroy_nodes(*ProtoList, FreePrototype); } // FreeProtoList - /** * This routine deallocates the memory consumed by the specified * prototype and modifies the corresponding cluster so that it @@ -606,7 +608,6 @@ void FreePrototype(void *arg) { //PROTOTYPE *Prototype) memfree(Prototype); } // FreePrototype - /** * This routine is used to find all of the samples which * belong to a cluster. It starts by removing the top @@ -617,10 +618,10 @@ void FreePrototype(void *arg) { //PROTOTYPE *Prototype) * If all samples have been found, NULL is returned. * InitSampleSearch() must be called * before NextSample() to initialize the search. - * @param SearchState ptr to list containing clusters to be searched - * @return Pointer to the next leaf cluster (sample) or NULL. - * @note Exceptions: None - * @note History: 6/16/89, DSJ, Created. + * @param SearchState ptr to list containing clusters to be searched + * @return Pointer to the next leaf cluster (sample) or NULL. + * @note Exceptions: None + * @note History: 6/16/89, DSJ, Created. */ CLUSTER *NextSample(LIST *SearchState) { CLUSTER *Cluster; @@ -637,29 +638,27 @@ CLUSTER *NextSample(LIST *SearchState) { } } // NextSample - /** * This routine returns the mean of the specified * prototype in the indicated dimension. - * @param Proto prototype to return mean of - * @param Dimension dimension whose mean is to be returned - * @return Mean of Prototype in Dimension + * @param Proto prototype to return mean of + * @param Dimension dimension whose mean is to be returned + * @return Mean of Prototype in Dimension * @note Exceptions: none - * @note History: 7/6/89, DSJ, Created. + * @note History: 7/6/89, DSJ, Created. */ FLOAT32 Mean(PROTOTYPE *Proto, uinT16 Dimension) { return (Proto->Mean[Dimension]); } // Mean - /** * This routine returns the standard deviation of the * prototype in the indicated dimension. - * @param Proto prototype to return standard deviation of - * @param Dimension dimension whose stddev is to be returned - * @return Standard deviation of Prototype in Dimension + * @param Proto prototype to return standard deviation of + * @param Dimension dimension whose stddev is to be returned + * @return Standard deviation of Prototype in Dimension * @note Exceptions: none - * @note History: 7/6/89, DSJ, Created. + * @note History: 7/6/89, DSJ, Created. */ FLOAT32 StandardDeviation(PROTOTYPE *Proto, uinT16 Dimension) { switch (Proto->Style) { @@ -697,10 +696,10 @@ FLOAT32 StandardDeviation(PROTOTYPE *Proto, uinT16 Dimension) { * tree are the individual samples themselves; they have no * sub-clusters. The root node of the tree conceptually contains * all of the samples. - * @param Clusterer data structure holdings samples to be clustered - * @return None (the Clusterer data structure is changed) - * @note Exceptions: None - * @note History: 5/29/89, DSJ, Created. + * @param Clusterer data structure holdings samples to be clustered + * @return None (the Clusterer data structure is changed) + * @note Exceptions: None + * @note History: 5/29/89, DSJ, Created. */ void CreateClusterTree(CLUSTERER *Clusterer) { ClusteringContext context; @@ -760,7 +759,6 @@ void CreateClusterTree(CLUSTERER *Clusterer) { memfree(context.candidates); } // CreateClusterTree - /** * This routine is designed to be used in concert with the * KDWalk routine. It will create a potential cluster for @@ -786,7 +784,6 @@ void MakePotentialClusters(ClusteringContext *context, } } // MakePotentialClusters - /** * This routine searches the specified kd-tree for the nearest * neighbor of the specified cluster. It actually uses the @@ -795,12 +792,12 @@ void MakePotentialClusters(ClusteringContext *context, * neighbor is returned, if it can be found, otherwise NULL is * returned. The distance between the 2 nodes is placed * in the specified variable. - * @param Tree kd-tree to search in for nearest neighbor - * @param Cluster cluster whose nearest neighbor is to be found - * @param Distance ptr to variable to report distance found - * @return Pointer to the nearest neighbor of Cluster, or NULL + * @param Tree kd-tree to search in for nearest neighbor + * @param Cluster cluster whose nearest neighbor is to be found + * @param Distance ptr to variable to report distance found + * @return Pointer to the nearest neighbor of Cluster, or NULL * @note Exceptions: none - * @note History: 5/29/89, DSJ, Created. + * @note History: 5/29/89, DSJ, Created. * 7/13/89, DSJ, Removed visibility of kd-tree node data struct */ CLUSTER * @@ -830,17 +827,16 @@ FindNearestNeighbor(KDTREE * Tree, CLUSTER * Cluster, FLOAT32 * Distance) return BestNeighbor; } // FindNearestNeighbor - /** * This routine creates a new permanent cluster from the * clusters specified in TempCluster. The 2 clusters in * TempCluster are marked as "clustered" and deleted from * the kd-tree. The new cluster is then added to the kd-tree. - * @param Clusterer current clustering environment - * @param TempCluster potential cluster to make permanent + * @param Clusterer current clustering environment + * @param TempCluster potential cluster to make permanent * @return Pointer to the new permanent cluster - * @note Exceptions: none - * @note History: 5/29/89, DSJ, Created. + * @note Exceptions: none + * @note History: 5/29/89, DSJ, Created. * 7/13/89, DSJ, Removed visibility of kd-tree node data struct */ CLUSTER *MakeNewCluster(CLUSTERER *Clusterer, TEMPCLUSTER *TempCluster) { @@ -872,21 +868,20 @@ CLUSTER *MakeNewCluster(CLUSTERER *Clusterer, TEMPCLUSTER *TempCluster) { return Cluster; } // MakeNewCluster - /** * This routine merges two clusters into one larger cluster. * To do this it computes the number of samples in the new * cluster and the mean of the new cluster. The ParamDesc * information is used to ensure that circular dimensions * are handled correctly. - * @param N # of dimensions (size of arrays) - * @param ParamDesc array of dimension descriptions - * @param n1, n2 number of samples in each old cluster - * @param m array to hold mean of new cluster - * @param m1, m2 arrays containing means of old clusters - * @return The number of samples in the new cluster. - * @note Exceptions: None - * @note History: 5/31/89, DSJ, Created. + * @param N # of dimensions (size of arrays) + * @param ParamDesc array of dimension descriptions + * @param n1, n2 number of samples in each old cluster + * @param m array to hold mean of new cluster + * @param m1, m2 arrays containing means of old clusters + * @return The number of samples in the new cluster. + * @note Exceptions: None + * @note History: 5/31/89, DSJ, Created. */ inT32 MergeClusters(inT16 N, PARAM_DESC ParamDesc[], @@ -921,17 +916,16 @@ inT32 MergeClusters(inT16 N, return n; } // MergeClusters - /** * This routine decides which clusters in the cluster tree * should be represented by prototypes, forms a list of these * prototypes, and places the list in the Clusterer data * structure. - * @param Clusterer data structure holding cluster tree - * @param Config parameters used to control prototype generation - * @return None - * @note Exceptions: None - * @note History: 5/30/89, DSJ, Created. + * @param Clusterer data structure holding cluster tree + * @param Config parameters used to control prototype generation + * @return None + * @note Exceptions: None + * @note History: 5/30/89, DSJ, Created. */ void ComputePrototypes(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { LIST ClusterStack = NIL_LIST; @@ -961,8 +955,7 @@ void ComputePrototypes(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { } } // ComputePrototypes - -/** +/** * This routine attempts to create a prototype from the * specified cluster that conforms to the distribution * specified in Config. If there are too few samples in the @@ -972,12 +965,12 @@ void ComputePrototypes(CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { * is generated and NULL is returned. If a prototype can be * found that matches the desired distribution then a pointer * to it is returned, otherwise NULL is returned. - * @param Clusterer data structure holding cluster tree - * @param Config parameters used to control prototype generation - * @param Cluster cluster to be made into a prototype - * @return Pointer to new prototype or NULL - * @note Exceptions: None - * @note History: 6/19/89, DSJ, Created. + * @param Clusterer data structure holding cluster tree + * @param Config parameters used to control prototype generation + * @param Cluster cluster to be made into a prototype + * @return Pointer to new prototype or NULL + * @note Exceptions: None + * @note History: 6/19/89, DSJ, Created. */ PROTOTYPE *MakePrototype(CLUSTERER *Clusterer, CLUSTERCONFIG *Config, @@ -1050,7 +1043,6 @@ PROTOTYPE *MakePrototype(CLUSTERER *Clusterer, return Proto; } // MakePrototype - /** * This routine checks for clusters which are degenerate and * therefore cannot be analyzed in a statistically valid way. @@ -1063,14 +1055,14 @@ PROTOTYPE *MakePrototype(CLUSTERER *Clusterer, * * If the cluster is not degenerate, NULL is returned. * - * @param N number of dimensions - * @param Cluster cluster being analyzed - * @param Statistics statistical info about cluster - * @param Style type of prototype to be generated - * @param MinSamples minimum number of samples in a cluster - * @return Pointer to degenerate prototype or NULL. - * @note Exceptions: None - * @note History: 6/20/89, DSJ, Created. + * @param N number of dimensions + * @param Cluster cluster being analyzed + * @param Statistics statistical info about cluster + * @param Style type of prototype to be generated + * @param MinSamples minimum number of samples in a cluster + * @return Pointer to degenerate prototype or NULL. + * @note Exceptions: None + * @note History: 6/20/89, DSJ, Created. * 7/12/89, DSJ, Changed name and added check for 0 stddev. * 8/8/89, DSJ, Removed check for 0 stddev (handled elsewhere). */ @@ -1110,10 +1102,10 @@ PROTOTYPE *MakeDegenerateProto( //this was MinSample * be split. If not, then a new prototype is formed and * returned to the caller. If there is, then NULL is returned * to the caller. - * @param Clusterer data struct containing samples being clustered + * @param Clusterer data struct containing samples being clustered * @param Config provides the magic number of samples that make a good cluster - * @param Cluster cluster to be made into an elliptical prototype - * @param Statistics statistical info about cluster + * @param Cluster cluster to be made into an elliptical prototype + * @param Statistics statistical info about cluster * @return Pointer to new elliptical prototype or NULL. */ PROTOTYPE *TestEllipticalProto(CLUSTERER *Clusterer, @@ -1215,13 +1207,13 @@ PROTOTYPE *TestEllipticalProto(CLUSTERER *Clusterer, * be approximated by a spherical normal distribution. If it * can be, then a new prototype is formed and returned to the * caller. If it can't be, then NULL is returned to the caller. - * @param Clusterer data struct containing samples being clustered - * @param Cluster cluster to be made into a spherical prototype - * @param Statistics statistical info about cluster - * @param Buckets histogram struct used to analyze distribution - * @return Pointer to new spherical prototype or NULL. - * @note Exceptions: None - * @note History: 6/1/89, DSJ, Created. + * @param Clusterer data struct containing samples being clustered + * @param Cluster cluster to be made into a spherical prototype + * @param Statistics statistical info about cluster + * @param Buckets histogram struct used to analyze distribution + * @return Pointer to new spherical prototype or NULL. + * @note Exceptions: None + * @note History: 6/1/89, DSJ, Created. */ PROTOTYPE *MakeSphericalProto(CLUSTERER *Clusterer, CLUSTER *Cluster, @@ -1247,19 +1239,18 @@ PROTOTYPE *MakeSphericalProto(CLUSTERER *Clusterer, return (Proto); } // MakeSphericalProto - /** * This routine tests the specified cluster to see if it can * be approximated by an elliptical normal distribution. If it * can be, then a new prototype is formed and returned to the * caller. If it can't be, then NULL is returned to the caller. - * @param Clusterer data struct containing samples being clustered - * @param Cluster cluster to be made into an elliptical prototype - * @param Statistics statistical info about cluster - * @param Buckets histogram struct used to analyze distribution - * @return Pointer to new elliptical prototype or NULL. - * @note Exceptions: None - * @note History: 6/12/89, DSJ, Created. + * @param Clusterer data struct containing samples being clustered + * @param Cluster cluster to be made into an elliptical prototype + * @param Statistics statistical info about cluster + * @param Buckets histogram struct used to analyze distribution + * @return Pointer to new elliptical prototype or NULL. + * @note Exceptions: None + * @note History: 6/12/89, DSJ, Created. */ PROTOTYPE *MakeEllipticalProto(CLUSTERER *Clusterer, CLUSTER *Cluster, @@ -1286,7 +1277,6 @@ PROTOTYPE *MakeEllipticalProto(CLUSTERER *Clusterer, return (Proto); } // MakeEllipticalProto - /** * This routine tests each dimension of the specified cluster to * see what distribution would best approximate that dimension. @@ -1295,14 +1285,14 @@ PROTOTYPE *MakeEllipticalProto(CLUSTERER *Clusterer, * be represented by one of these distributions, * then a new prototype is formed and returned to the * caller. If it can't be, then NULL is returned to the caller. - * @param Clusterer data struct containing samples being clustered - * @param Cluster cluster to be made into a prototype - * @param Statistics statistical info about cluster - * @param NormalBuckets histogram struct used to analyze distribution - * @param Confidence confidence level for alternate distributions - * @return Pointer to new mixed prototype or NULL. - * @note Exceptions: None - * @note History: 6/12/89, DSJ, Created. + * @param Clusterer data struct containing samples being clustered + * @param Cluster cluster to be made into a prototype + * @param Statistics statistical info about cluster + * @param NormalBuckets histogram struct used to analyze distribution + * @param Confidence confidence level for alternate distributions + * @return Pointer to new mixed prototype or NULL. + * @note Exceptions: None + * @note History: 6/12/89, DSJ, Created. */ PROTOTYPE *MakeMixedProto(CLUSTERER *Clusterer, CLUSTER *Cluster, @@ -1355,16 +1345,15 @@ PROTOTYPE *MakeMixedProto(CLUSTERER *Clusterer, return (Proto); } // MakeMixedProto - /** * This routine alters the ith dimension of the specified * mixed prototype to be D_random. - * @param i index of dimension to be changed - * @param Proto prototype whose dimension is to be altered - * @param ParamDesc description of specified dimension - * @return None - * @note Exceptions: None - * @note History: 6/20/89, DSJ, Created. + * @param i index of dimension to be changed + * @param Proto prototype whose dimension is to be altered + * @param ParamDesc description of specified dimension + * @return None + * @note Exceptions: None + * @note History: 6/20/89, DSJ, Created. */ void MakeDimRandom(uinT16 i, PROTOTYPE *Proto, PARAM_DESC *ParamDesc) { Proto->Distrib[i] = D_random; @@ -1380,16 +1369,15 @@ void MakeDimRandom(uinT16 i, PROTOTYPE *Proto, PARAM_DESC *ParamDesc) { // note that the proto Weight is irrelevant for D_random protos } // MakeDimRandom - /** * This routine alters the ith dimension of the specified * mixed prototype to be uniform. - * @param i index of dimension to be changed - * @param Proto prototype whose dimension is to be altered - * @param Statistics statistical info about prototype - * @return None - * @note Exceptions: None - * @note History: 6/20/89, DSJ, Created. + * @param i index of dimension to be changed + * @param Proto prototype whose dimension is to be altered + * @param Statistics statistical info about prototype + * @return None + * @note Exceptions: None + * @note History: 6/20/89, DSJ, Created. */ void MakeDimUniform(uinT16 i, PROTOTYPE *Proto, STATISTICS *Statistics) { Proto->Distrib[i] = uniform; @@ -1410,7 +1398,6 @@ void MakeDimUniform(uinT16 i, PROTOTYPE *Proto, STATISTICS *Statistics) { // note that the proto Weight is irrelevant for uniform protos } // MakeDimUniform - /** * This routine searches the cluster tree for all leaf nodes * which are samples in the specified cluster. It computes @@ -1420,12 +1407,12 @@ void MakeDimUniform(uinT16 i, PROTOTYPE *Proto, STATISTICS *Statistics) { * return this information to the caller. An incremental * algorithm for computing statistics is not used because * it will not work with circular dimensions. - * @param N number of dimensions - * @param ParamDesc array of dimension descriptions - * @param Cluster cluster whose stats are to be computed - * @return Pointer to new data structure containing statistics - * @note Exceptions: None - * @note History: 6/2/89, DSJ, Created. + * @param N number of dimensions + * @param ParamDesc array of dimension descriptions + * @param Cluster cluster whose stats are to be computed + * @return Pointer to new data structure containing statistics + * @note Exceptions: None + * @note History: 6/2/89, DSJ, Created. */ STATISTICS * ComputeStatistics (inT16 N, PARAM_DESC ParamDesc[], CLUSTER * Cluster) { @@ -1502,19 +1489,18 @@ ComputeStatistics (inT16 N, PARAM_DESC ParamDesc[], CLUSTER * Cluster) { return (Statistics); } // ComputeStatistics - /** * This routine creates a spherical prototype data structure to * approximate the samples in the specified cluster. * Spherical prototypes have a single variance which is * common across all dimensions. All dimensions are normally * distributed and independent. - * @param N number of dimensions - * @param Cluster cluster to be made into a spherical prototype - * @param Statistics statistical info about samples in cluster - * @return Pointer to a new spherical prototype data structure - * @note Exceptions: None - * @note History: 6/19/89, DSJ, Created. + * @param N number of dimensions + * @param Cluster cluster to be made into a spherical prototype + * @param Statistics statistical info about samples in cluster + * @return Pointer to a new spherical prototype data structure + * @note Exceptions: None + * @note History: 6/19/89, DSJ, Created. */ PROTOTYPE *NewSphericalProto(uinT16 N, CLUSTER *Cluster, @@ -1537,18 +1523,17 @@ PROTOTYPE *NewSphericalProto(uinT16 N, return (Proto); } // NewSphericalProto - /** * This routine creates an elliptical prototype data structure to * approximate the samples in the specified cluster. * Elliptical prototypes have a variance for each dimension. * All dimensions are normally distributed and independent. - * @param N number of dimensions - * @param Cluster cluster to be made into an elliptical prototype - * @param Statistics statistical info about samples in cluster - * @return Pointer to a new elliptical prototype data structure - * @note Exceptions: None - * @note History: 6/19/89, DSJ, Created. + * @param N number of dimensions + * @param Cluster cluster to be made into an elliptical prototype + * @param Statistics statistical info about samples in cluster + * @return Pointer to a new elliptical prototype data structure + * @note Exceptions: None + * @note History: 6/19/89, DSJ, Created. */ PROTOTYPE *NewEllipticalProto(inT16 N, CLUSTER *Cluster, @@ -1579,7 +1564,6 @@ PROTOTYPE *NewEllipticalProto(inT16 N, return (Proto); } // NewEllipticalProto - /** * This routine creates a mixed prototype data structure to * approximate the samples in the specified cluster. @@ -1588,12 +1572,12 @@ PROTOTYPE *NewEllipticalProto(inT16 N, * structure is initially filled in as though it were an * elliptical prototype. The actual distributions of the * dimensions can be altered by other routines. - * @param N number of dimensions - * @param Cluster cluster to be made into a mixed prototype - * @param Statistics statistical info about samples in cluster - * @return Pointer to a new mixed prototype data structure - * @note Exceptions: None - * @note History: 6/19/89, DSJ, Created. + * @param N number of dimensions + * @param Cluster cluster to be made into a mixed prototype + * @param Statistics statistical info about samples in cluster + * @return Pointer to a new mixed prototype data structure + * @note Exceptions: None + * @note History: 6/19/89, DSJ, Created. */ PROTOTYPE *NewMixedProto(inT16 N, CLUSTER *Cluster, STATISTICS *Statistics) { PROTOTYPE *Proto; @@ -1609,16 +1593,15 @@ PROTOTYPE *NewMixedProto(inT16 N, CLUSTER *Cluster, STATISTICS *Statistics) { return (Proto); } // NewMixedProto - /** * This routine allocates memory to hold a simple prototype * data structure, i.e. one without independent distributions * and variances for each dimension. - * @param N number of dimensions - * @param Cluster cluster to be made into a prototype - * @return Pointer to new simple prototype - * @note Exceptions: None - * @note History: 6/19/89, DSJ, Created. + * @param N number of dimensions + * @param Cluster cluster to be made into a prototype + * @return Pointer to new simple prototype + * @note Exceptions: None + * @note History: 6/19/89, DSJ, Created. */ PROTOTYPE *NewSimpleProto(inT16 N, CLUSTER *Cluster) { PROTOTYPE *Proto; @@ -1640,7 +1623,6 @@ PROTOTYPE *NewSimpleProto(inT16 N, CLUSTER *Cluster) { return (Proto); } // NewSimpleProto - /** * This routine returns TRUE if the specified covariance * matrix indicates that all N dimensions are independent of @@ -1653,13 +1635,13 @@ PROTOTYPE *NewSimpleProto(inT16 N, CLUSTER *Cluster) { * coeff[ij] = stddev[ij] / sqrt (stddev[ii] * stddev[jj]) * The covariance matrix is assumed to be symmetric (which * should always be true). - * @param ParamDesc descriptions of each feature space dimension - * @param N number of dimensions - * @param CoVariance ptr to a covariance matrix - * @param Independence max off-diagonal correlation coefficient - * @return TRUE if dimensions are independent, FALSE otherwise - * @note Exceptions: None - * @note History: 6/4/89, DSJ, Created. + * @param ParamDesc descriptions of each feature space dimension + * @param N number of dimensions + * @param CoVariance ptr to a covariance matrix + * @param Independence max off-diagonal correlation coefficient + * @return TRUE if dimensions are independent, FALSE otherwise + * @note Exceptions: None + * @note History: 6/4/89, DSJ, Created. */ BOOL8 Independent (PARAM_DESC ParamDesc[], @@ -1692,7 +1674,6 @@ inT16 N, FLOAT32 * CoVariance, FLOAT32 Independence) { return (TRUE); } // Independent - /** * This routine returns a histogram data structure which can * be used by other routines to place samples into histogram @@ -1703,12 +1684,12 @@ inT16 N, FLOAT32 * CoVariance, FLOAT32 Independence) { * created so that it minimizes the computation time needed * to create a new bucket. * @param clusterer which keeps a bucket_cache for us. - * @param Distribution type of probability distribution to test for - * @param SampleCount number of samples that are available - * @param Confidence probability of a Type I error - * @return Bucket data structure + * @param Distribution type of probability distribution to test for + * @param SampleCount number of samples that are available + * @param Confidence probability of a Type I error + * @return Bucket data structure * @note Exceptions: none - * @note History: Thu Aug 3 12:58:10 1989, DSJ, Created. + * @note History: Thu Aug 3 12:58:10 1989, DSJ, Created. */ BUCKETS *GetBuckets(CLUSTERER* clusterer, DISTRIBUTION Distribution, @@ -1739,7 +1720,6 @@ BUCKETS *GetBuckets(CLUSTERER* clusterer, return Buckets; } // GetBuckets - /** * This routine creates a histogram data structure which can * be used by other routines to place samples into histogram @@ -1751,12 +1731,12 @@ BUCKETS *GetBuckets(CLUSTERER* clusterer, * order to make this possible, a mapping table is * computed which maps "normalized" samples into the * appropriate bucket. - * @param Distribution type of probability distribution to test for - * @param SampleCount number of samples that are available - * @param Confidence probability of a Type I error + * @param Distribution type of probability distribution to test for + * @param SampleCount number of samples that are available + * @param Confidence probability of a Type I error * @return Pointer to new histogram data structure - * @note Exceptions: None - * @note History: 6/4/89, DSJ, Created. + * @note Exceptions: None + * @note History: 6/4/89, DSJ, Created. */ BUCKETS *MakeBuckets(DISTRIBUTION Distribution, uinT32 SampleCount, @@ -1840,7 +1820,6 @@ BUCKETS *MakeBuckets(DISTRIBUTION Distribution, return Buckets; } // MakeBuckets - /** * This routine computes the optimum number of histogram * buckets that should be used in a chi-squared goodness of @@ -1851,7 +1830,7 @@ BUCKETS *MakeBuckets(DISTRIBUTION Distribution, * values. The table is intended for a 0.05 level of * significance (alpha). This routine assumes that it is * equally valid for other alpha's, which may not be true. - * @param SampleCount number of samples to be tested + * @param SampleCount number of samples to be tested * @return Optimum number of histogram buckets * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -1874,7 +1853,6 @@ uinT16 OptimumNumberOfBuckets(uinT32 SampleCount) { return kBucketsTable[Last]; } // OptimumNumberOfBuckets - /** * This routine computes the chi-squared value which will * leave a cumulative probability of Alpha in the right tail @@ -1887,8 +1865,8 @@ uinT16 OptimumNumberOfBuckets(uinT32 SampleCount) { * chi-squared value. Therefore, once a particular chi-squared * value is computed, it is stored in the list and never * needs to be computed again. - * @param DegreesOfFreedom determines shape of distribution - * @param Alpha probability of right tail + * @param DegreesOfFreedom determines shape of distribution + * @param Alpha probability of right tail * @return Desired chi-squared value * @note Exceptions: none * @note History: 6/5/89, DSJ, Created. @@ -1932,19 +1910,19 @@ ComputeChiSquared (uinT16 DegreesOfFreedom, FLOAT64 Alpha) } // ComputeChiSquared - /** * This routine computes the probability density function * of a discrete normal distribution defined by the global * variables kNormalMean, kNormalVariance, and kNormalMagnitude. * Normal magnitude could, of course, be computed in terms of * the normal variance but it is precomputed for efficiency. - * @param x number to compute the normal probability density for + * @param x number to compute the normal probability density for * @note Globals: - * kNormalMean mean of a discrete normal distribution - * kNormalVariance variance of a discrete normal distribution - * kNormalMagnitude magnitude of a discrete normal distribution - * @return The value of the normal distribution at x. + * kNormalMean mean of a discrete normal distribution + * kNormalVariance variance of a discrete normal distribution + * kNormalMagnitude magnitude of a discrete normal + *distribution + * @return The value of the normal distribution at x. * @note Exceptions: None * @note History: 6/4/89, DSJ, Created. */ @@ -1955,12 +1933,11 @@ FLOAT64 NormalDensity(inT32 x) { return kNormalMagnitude * exp(-0.5 * Distance * Distance / kNormalVariance); } // NormalDensity - /** * This routine computes the probability density function * of a uniform distribution at the specified point. The * range of the distribution is from 0 to BUCKETTABLESIZE. - * @param x number to compute the uniform probability density for + * @param x number to compute the uniform probability density for * @return The value of the uniform distribution at x. * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -1974,13 +1951,12 @@ FLOAT64 UniformDensity(inT32 x) { return (FLOAT64) 0.0; } // UniformDensity - /** * This routine computes a trapezoidal approximation to the * integral of a function over a small delta in x. - * @param f1 value of function at x1 - * @param f2 value of function at x2 - * @param Dx x2 - x1 (should always be positive) + * @param f1 value of function at x1 + * @param f2 value of function at x2 + * @param Dx x2 - x1 (should always be positive) * @return Approximation of the integral of the function from x1 to x2. * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -1989,7 +1965,6 @@ FLOAT64 Integral(FLOAT64 f1, FLOAT64 f2, FLOAT64 Dx) { return (f1 + f2) * Dx / 2.0; } // Integral - /** * This routine counts the number of cluster samples which * fall within the various histogram buckets in Buckets. Only @@ -2002,12 +1977,12 @@ FLOAT64 Integral(FLOAT64 f1, FLOAT64 f2, FLOAT64 Dx) { * range and the StdDev is 1/2 the range. A dimension with * zero standard deviation cannot be statistically analyzed. * In this case, a pseudo-analysis is used. - * @param Buckets histogram buckets to count samples - * @param Cluster cluster whose samples are being analyzed - * @param Dim dimension of samples which is being analyzed - * @param ParamDesc description of the dimension - * @param Mean "mean" of the distribution - * @param StdDev "standard deviation" of the distribution + * @param Buckets histogram buckets to count samples + * @param Cluster cluster whose samples are being analyzed + * @param Dim dimension of samples which is being analyzed + * @param ParamDesc description of the dimension + * @param Mean "mean" of the distribution + * @param StdDev "standard deviation" of the distribution * @return None (the Buckets data structure is filled in) * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -2071,16 +2046,15 @@ void FillBuckets(BUCKETS *Buckets, } } // FillBuckets - /** * This routine determines which bucket x falls into in the * discrete normal distribution defined by kNormalMean * and kNormalStdDev. x values which exceed the range of * the discrete distribution are clipped. - * @param ParamDesc used to identify circular dimensions - * @param x value to be normalized - * @param Mean mean of normal distribution - * @param StdDev standard deviation of normal distribution + * @param ParamDesc used to identify circular dimensions + * @param x value to be normalized + * @param Mean mean of normal distribution + * @param StdDev standard deviation of normal distribution * @return Bucket number into which x falls * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -2107,16 +2081,15 @@ uinT16 NormalBucket(PARAM_DESC *ParamDesc, return (uinT16) floor((FLOAT64) X); } // NormalBucket - /** * This routine determines which bucket x falls into in the * discrete uniform distribution defined by * BUCKETTABLESIZE. x values which exceed the range of * the discrete distribution are clipped. - * @param ParamDesc used to identify circular dimensions - * @param x value to be normalized - * @param Mean center of range of uniform distribution - * @param StdDev 1/2 the range of the uniform distribution + * @param ParamDesc used to identify circular dimensions + * @param x value to be normalized + * @param Mean center of range of uniform distribution + * @param StdDev 1/2 the range of the uniform distribution * @return Bucket number into which x falls * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -2143,7 +2116,6 @@ uinT16 UniformBucket(PARAM_DESC *ParamDesc, return (uinT16) floor((FLOAT64) X); } // UniformBucket - /** * This routine performs a chi-square goodness of fit test * on the histogram data in the Buckets data structure. TRUE @@ -2151,7 +2123,7 @@ uinT16 UniformBucket(PARAM_DESC *ParamDesc, * distribution which was specified when the Buckets * structure was originally created. Otherwise FALSE is * returned. - * @param Buckets histogram data to perform chi-square test on + * @param Buckets histogram data to perform chi-square test on * @return TRUE if samples match distribution, FALSE otherwise * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -2176,11 +2148,10 @@ BOOL8 DistributionOK(BUCKETS *Buckets) { return TRUE; } // DistributionOK - /** * This routine frees the memory used by the statistics * data structure. - * @param Statistics pointer to data structure to be freed + * @param Statistics pointer to data structure to be freed * @return None * @note Exceptions: None * @note History: 6/5/89, DSJ, Created. @@ -2192,7 +2163,6 @@ void FreeStatistics(STATISTICS *Statistics) { memfree(Statistics); } // FreeStatistics - /** * This routine properly frees the memory used by a BUCKETS. * @@ -2204,13 +2174,12 @@ void FreeBuckets(BUCKETS *buckets) { Efree(buckets); } // FreeBuckets - /** * This routine frees the memory consumed by the specified * cluster and all of its subclusters. This is done by * recursive calls to FreeCluster(). * - * @param Cluster pointer to cluster to be freed + * @param Cluster pointer to cluster to be freed * * @return None * @@ -2225,7 +2194,6 @@ void FreeCluster(CLUSTER *Cluster) { } } // FreeCluster - /** * This routine computes the degrees of freedom that should * be used in a chi-squared test with the specified number of @@ -2234,8 +2202,8 @@ void FreeCluster(CLUSTER *Cluster) { * computed more easily. This will cause the value of * chi-squared to be higher than the optimum value, resulting * in the chi-square test being more lenient than optimum. - * @param Distribution distribution being tested for - * @param HistogramBuckets number of buckets in chi-square test + * @param Distribution distribution being tested for + * @param HistogramBuckets number of buckets in chi-square test * @return The number of degrees of freedom for a chi-square test * @note Exceptions: none * @note History: Thu Aug 3 14:04:18 1989, DSJ, Created. @@ -2252,7 +2220,6 @@ uinT16 DegreesOfFreedom(DISTRIBUTION Distribution, uinT16 HistogramBuckets) { } // DegreesOfFreedom - /** * This routine is used to search a list of histogram data * structures to find one with the specified number of @@ -2272,7 +2239,6 @@ int NumBucketsMatch(void *arg1, // BUCKETS *Histogram, } // NumBucketsMatch - /** * This routine is used to search a list for a list node * whose contents match Key. It is called by the list @@ -2287,13 +2253,12 @@ int ListEntryMatch(void *arg1, //ListNode } // ListEntryMatch - /** * This routine multiplies each ExpectedCount histogram entry * by NewSampleCount/OldSampleCount so that the histogram * is now adjusted to the new sample count. - * @param Buckets histogram data structure to adjust - * @param NewSampleCount new sample count to adjust to + * @param Buckets histogram data structure to adjust + * @param NewSampleCount new sample count to adjust to * @return none * @note Exceptions: none * @note History: Thu Aug 3 14:31:14 1989, DSJ, Created. @@ -2313,11 +2278,10 @@ void AdjustBuckets(BUCKETS *Buckets, uinT32 NewSampleCount) { } // AdjustBuckets - /** * This routine sets the bucket counts in the specified histogram * to zero. - * @param Buckets histogram data structure to init + * @param Buckets histogram data structure to init * @return none * @note Exceptions: none * @note History: Thu Aug 3 14:31:14 1989, DSJ, Created. @@ -2331,7 +2295,6 @@ void InitBuckets(BUCKETS *Buckets) { } // InitBuckets - /** * This routine is used to search a list of structures which * hold pre-computed chi-squared values for a chi-squared @@ -2355,14 +2318,13 @@ int AlphaMatch(void *arg1, //CHISTRUCT *ChiStruct } // AlphaMatch - /** * This routine allocates a new data structure which is used * to hold a chi-squared value along with its associated * number of degrees of freedom and alpha value. * - * @param DegreesOfFreedom degrees of freedom for new chi value - * @param Alpha confidence level for new chi value + * @param DegreesOfFreedom degrees of freedom for new chi value + * @param Alpha confidence level for new chi value * @return none * @note Exceptions: none * @note History: Fri Aug 4 11:04:59 1989, DSJ, Created. @@ -2377,7 +2339,6 @@ CHISTRUCT *NewChiStruct(uinT16 DegreesOfFreedom, FLOAT64 Alpha) { } // NewChiStruct - /** * This routine attempts to find an x value at which Function * goes to zero (i.e. a root of the function ). It will only @@ -2385,10 +2346,10 @@ CHISTRUCT *NewChiStruct(uinT16 DegreesOfFreedom, FLOAT64 Alpha) { * are no extrema between the solution and the InitialGuess. * The algorithms used are extremely primitive. * - * @param Function function whose zero is to be found - * @param FunctionParams arbitrary data to pass to function - * @param InitialGuess point to start solution search at - * @param Accuracy maximum allowed error + * @param Function function whose zero is to be found + * @param FunctionParams arbitrary data to pass to function + * @param InitialGuess point to start solution search at + * @param Accuracy maximum allowed error * @return Solution of function ( x for which f(x) = 0 ). * @note Exceptions: none * @note History: Fri Aug 4 11:08:59 1989, DSJ, Created. @@ -2440,7 +2401,6 @@ void *FunctionParams, FLOAT64 InitialGuess, FLOAT64 Accuracy) } // Solve - /** * This routine computes the area under a chi density curve * from 0 to x, minus the desired area under the curve. The @@ -2455,8 +2415,8 @@ void *FunctionParams, FLOAT64 InitialGuess, FLOAT64 Accuracy) * integrating the chi density curve in parts to obtain * a series that can be used to compute the area under the * curve. - * @param ChiParams contains degrees of freedom and alpha - * @param x value of chi-squared to evaluate + * @param ChiParams contains degrees of freedom and alpha + * @param x value of chi-squared to evaluate * @return Error between actual and desired area under the chi curve. * @note Exceptions: none * @note History: Fri Aug 4 12:48:41 1989, DSJ, Created. @@ -2480,7 +2440,6 @@ FLOAT64 ChiArea(CHISTRUCT *ChiParams, FLOAT64 x) { } // ChiArea - /** * This routine looks at all samples in the specified cluster. * It computes a running estimate of the percentage of the @@ -2498,10 +2457,10 @@ FLOAT64 ChiArea(CHISTRUCT *ChiParams, FLOAT64 x) { * contained in the same cluster, then the cluster should be * split. * - * @param Clusterer data structure holding cluster tree - * @param Cluster cluster containing samples to be tested - * @param MaxIllegal max percentage of samples allowed to have - * more than 1 feature in the cluster + * @param Clusterer data structure holding cluster tree + * @param Cluster cluster containing samples to be tested + * @param MaxIllegal max percentage of samples allowed to have + * more than 1 feature in the cluster * @return TRUE if the cluster should be split, FALSE otherwise. * @note Exceptions: none * @note History: Wed Aug 30 11:13:05 1989, DSJ, Created. @@ -2562,7 +2521,7 @@ CLUSTER * Cluster, FLOAT32 MaxIllegal) } // MultipleCharSamples /** - * Compute the inverse of a matrix using LU decomposition with partial pivoting. + * Compute the inverse of a matrix using LU decomposition with partial pivoting. * The return value is the sum of norms of the off-diagonal terms of the * product of a and inv. (A measure of the error.) */ diff --git a/classify/cluster.h b/classify/cluster.h index 53ddf87dad..a068a5d58c 100644 --- a/classify/cluster.h +++ b/classify/cluster.h @@ -107,15 +107,15 @@ typedef struct { /*-------------------------------------------------------------------------- Public Function Prototypes --------------------------------------------------------------------------*/ -CLUSTERER *MakeClusterer (inT16 SampleSize, const PARAM_DESC ParamDesc[]); +CLUSTERER TESS_API *MakeClusterer (inT16 SampleSize, const PARAM_DESC ParamDesc[]); -SAMPLE *MakeSample(CLUSTERER * Clusterer, const FLOAT32* Feature, inT32 CharID); +SAMPLE TESS_API *MakeSample(CLUSTERER * Clusterer, const FLOAT32* Feature, inT32 CharID); LIST ClusterSamples(CLUSTERER *Clusterer, CLUSTERCONFIG *Config); -void FreeClusterer(CLUSTERER *Clusterer); +void TESS_API FreeClusterer(CLUSTERER *Clusterer); -void FreeProtoList(LIST *ProtoList); +void TESS_API FreeProtoList(LIST *ProtoList); void FreePrototype(void *arg); // PROTOTYPE *Prototype); @@ -125,7 +125,7 @@ FLOAT32 Mean(PROTOTYPE *Proto, uinT16 Dimension); FLOAT32 StandardDeviation(PROTOTYPE *Proto, uinT16 Dimension); -inT32 MergeClusters(inT16 N, PARAM_DESC ParamDesc[], inT32 n1, inT32 n2, +inT32 TESS_API MergeClusters(inT16 N, PARAM_DESC ParamDesc[], inT32 n1, inT32 n2, FLOAT32 m[], FLOAT32 m1[], FLOAT32 m2[]); //--------------Global Data Definitions and Declarations--------------------------- diff --git a/classify/clusttool.cpp b/classify/clusttool.cpp index d86c3a2407..02e619d273 100644 --- a/classify/clusttool.cpp +++ b/classify/clusttool.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: clustertool.c - ** Purpose: Misc. tools for use with the clustering routines - ** Author: Dan Johnson - ** History: 6/6/89, DSJ, Created. + ** Filename: clustertool.c + ** Purpose: Misc. tools for use with the clustering routines + ** Author: Dan Johnson + ** History: 6/6/89, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -26,9 +26,10 @@ #include //---------------Global Data Definitions and Declarations-------------------- -#define TOKENSIZE 80 //< max size of tokens read from an input file -#define MAXSAMPLESIZE 65535 //< max num of dimensions in feature space -//#define MAXBLOCKSIZE 65535 //< max num of samples in a character (block size) +#define TOKENSIZE 80 //< max size of tokens read from an input file +#define MAXSAMPLESIZE 65535 //< max num of dimensions in feature space +//#define MAXBLOCKSIZE 65535 //< max num of samples in a character (block +// size) /** * This routine reads a single integer from the specified @@ -37,7 +38,7 @@ * @param File open text file to read sample size from * @return Sample size * @note Globals: None - * @note Exceptions: ILLEGALSAMPLESIZE illegal format or range + * @note Exceptions: ILLEGALSAMPLESIZE illegal format or range * @note History: 6/6/89, DSJ, Created. */ uinT16 ReadSampleSize(FILE *File) { @@ -293,7 +294,7 @@ FLOAT32* ReadNFloats(FILE * File, uinT16 N, FLOAT32 Buffer[]) { if (NumFloatsRead != 1) { if ((NumFloatsRead == EOF) && (i == 0)) { if (needs_free) { - Efree(Buffer); + Efree(Buffer); } return NULL; } else { @@ -315,8 +316,7 @@ FLOAT32* ReadNFloats(FILE * File, uinT16 N, FLOAT32 Buffer[]) { * @note Exceptions: None * @note History: 6/6/89, DSJ, Created. */ -void -WriteParamDesc (FILE * File, uinT16 N, PARAM_DESC ParamDesc[]) { +void WriteParamDesc(FILE *File, uinT16 N, const PARAM_DESC ParamDesc[]) { int i; for (i = 0; i < N; i++) { @@ -446,15 +446,10 @@ void WriteProtoStyle(FILE *File, PROTOSTYLE ProtoStyle) { * @note History: 6/12/89, DSJ, Created. */ -void WriteProtoList( - FILE *File, - uinT16 N, - PARAM_DESC ParamDesc[], - LIST ProtoList, - BOOL8 WriteSigProtos, - BOOL8 WriteInsigProtos) -{ - PROTOTYPE *Proto; +void WriteProtoList(FILE *File, uinT16 N, PARAM_DESC ParamDesc[], + LIST ProtoList, BOOL8 WriteSigProtos, + BOOL8 WriteInsigProtos) { + PROTOTYPE *Proto; /* write file header */ fprintf(File,"%0d\n",N); @@ -464,8 +459,8 @@ void WriteProtoList( iterate(ProtoList) { Proto = (PROTOTYPE *) first_node ( ProtoList ); - if (( Proto->Significant && WriteSigProtos ) || - ( ! Proto->Significant && WriteInsigProtos ) ) - WritePrototype( File, N, Proto ); + if ((Proto->Significant && WriteSigProtos) || + (!Proto->Significant && WriteInsigProtos)) + WritePrototype(File, N, Proto); } } diff --git a/classify/clusttool.h b/classify/clusttool.h index e82fa1ef48..a6fe38ea48 100644 --- a/classify/clusttool.h +++ b/classify/clusttool.h @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: clusttool.h - ** Purpose: Definition of clustering utility tools - ** Author: Dan Johnson - ** History: 6/6/89, DSJ, Created. + ** Filename: clusttool.h + ** Purpose: Definition of clustering utility tools + ** Author: Dan Johnson + ** History: 6/6/89, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -36,7 +36,7 @@ PROTOSTYLE ReadProtoStyle(FILE *File); FLOAT32 *ReadNFloats (FILE * File, uinT16 N, FLOAT32 Buffer[]); -void WriteParamDesc (FILE * File, uinT16 N, PARAM_DESC ParamDesc[]); +void WriteParamDesc(FILE *File, uinT16 N, const PARAM_DESC ParamDesc[]); void WritePrototype(FILE *File, uinT16 N, PROTOTYPE *Proto); @@ -44,13 +44,9 @@ void WriteNFloats (FILE * File, uinT16 N, FLOAT32 Array[]); void WriteProtoStyle(FILE *File, PROTOSTYLE ProtoStyle); -void WriteProtoList( - FILE *File, - uinT16 N, - PARAM_DESC ParamDesc[], - LIST ProtoList, - BOOL8 WriteSigProtos, - BOOL8 WriteInsigProtos); +void WriteProtoList(FILE *File, uinT16 N, PARAM_DESC ParamDesc[], + LIST ProtoList, BOOL8 WriteSigProtos, + BOOL8 WriteInsigProtos); //--------------Global Data Definitions and Declarations--------------------- // define errors that can be trapped diff --git a/classify/cutoffs.cpp b/classify/cutoffs.cpp index 4f6417149a..ffb8692ef1 100644 --- a/classify/cutoffs.cpp +++ b/classify/cutoffs.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: cutoffs.c - ** Purpose: Routines to manipulate an array of class cutoffs. - ** Author: Dan Johnson - ** History: Wed Feb 20 09:28:51 1991, DSJ, Created. + ** Filename: cutoffs.c + ** Purpose: Routines to manipulate an array of class cutoffs. + ** Author: Dan Johnson + ** History: Wed Feb 20 09:28:51 1991, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at diff --git a/classify/featdefs.cpp b/classify/featdefs.cpp index ad7b799675..dd31f91d86 100644 --- a/classify/featdefs.cpp +++ b/classify/featdefs.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: featdefs.c - ** Purpose: Definitions of currently defined feature types. - ** Author: Dan Johnson - ** History: Mon May 21 10:26:21 1990, DSJ, Created. + ** Filename: featdefs.c + ** Purpose: Definitions of currently defined feature types. + ** Author: Dan Johnson + ** History: Mon May 21 10:26:21 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -289,13 +289,13 @@ CHAR_DESC ReadCharDescription(const FEATURE_DEFS_STRUCT &FeatureDefs, * the feature type for the feature with the specified short * name. Trap an error if the specified name is not found. * - * Globals: + * Globals: * - none * * @param FeatureDefs definitions of feature types/extractors * @param ShortName short name of a feature type * @return Feature type which corresponds to ShortName. - * @note Exceptions: + * @note Exceptions: * - ILLEGAL_SHORT_NAME * @note History: Wed May 23 15:36:05 1990, DSJ, Created. */ diff --git a/classify/featdefs.h b/classify/featdefs.h index 704bbdfde2..eb0c71ce45 100644 --- a/classify/featdefs.h +++ b/classify/featdefs.h @@ -55,7 +55,7 @@ typedef FEATURE_DEFS_STRUCT *FEATURE_DEFS; /*---------------------------------------------------------------------- Generic functions for manipulating character descriptions ----------------------------------------------------------------------*/ -void InitFeatureDefs(FEATURE_DEFS_STRUCT *featuredefs); +void TESS_API InitFeatureDefs(FEATURE_DEFS_STRUCT *featuredefs); void FreeCharDescription(CHAR_DESC CharDesc); @@ -67,17 +67,17 @@ bool ValidCharDescription(const FEATURE_DEFS_STRUCT &FeatureDefs, void WriteCharDescription(const FEATURE_DEFS_STRUCT& FeatureDefs, CHAR_DESC CharDesc, STRING* str); -CHAR_DESC ReadCharDescription(const FEATURE_DEFS_STRUCT &FeatureDefs, +CHAR_DESC TESS_API ReadCharDescription(const FEATURE_DEFS_STRUCT &FeatureDefs, FILE *File); -int ShortNameToFeatureType(const FEATURE_DEFS_STRUCT &FeatureDefs, +int TESS_API ShortNameToFeatureType(const FEATURE_DEFS_STRUCT &FeatureDefs, const char *ShortName); /**---------------------------------------------------------------------------- Global Data Definitions and Declarations ----------------------------------------------------------------------------**/ extern const FEATURE_DESC_STRUCT MicroFeatureDesc; -extern const FEATURE_DESC_STRUCT PicoFeatDesc; +extern TESS_API const FEATURE_DESC_STRUCT PicoFeatDesc; extern const FEATURE_DESC_STRUCT CharNormDesc; extern const FEATURE_DESC_STRUCT OutlineFeatDesc; extern const FEATURE_DESC_STRUCT IntFeatDesc; diff --git a/classify/fpoint.cpp b/classify/fpoint.cpp index 854bea7b7e..ff5b7b7cf7 100644 --- a/classify/fpoint.cpp +++ b/classify/fpoint.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: fpoint.c - ** Purpose: Abstract data type for a 2D point (floating point coords) - ** Author: Dan Johnson - ** History: Thu Apr 12 10:44:15 1990, DSJ, Created. + ** Filename: fpoint.c + ** Purpose: Abstract data type for a 2D point (floating point coords) + ** Author: Dan Johnson + ** History: Thu Apr 12 10:44:15 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -58,5 +58,4 @@ FLOAT32 NormalizedAngleFrom(FPOINT *Point1, if (Angle < 0.0 || Angle >= FullScale) Angle = 0.0; return (Angle); - } diff --git a/classify/intfeaturemap.h b/classify/intfeaturemap.h index 55c5b5cf5e..93bc896c0e 100644 --- a/classify/intfeaturemap.h +++ b/classify/intfeaturemap.h @@ -75,7 +75,7 @@ class IntFeatureMap { // Copies the given feature_space and uses it as the index feature map // from INT_FEATURE_STRUCT. - void Init(const IntFeatureSpace& feature_space); + void TESS_API Init(const IntFeatureSpace& feature_space); // Helper to return an offset index feature. In this context an offset // feature with a dir of +/-1 is a feature of a similar direction, diff --git a/classify/intfeaturespace.h b/classify/intfeaturespace.h index e1e8e6ec9b..b72c42bdd1 100644 --- a/classify/intfeaturespace.h +++ b/classify/intfeaturespace.h @@ -35,7 +35,7 @@ class IndexMap; // Down-sampling quantization of the INT_FEATURE_STRUCT feature space and // conversion to a single scalar index value, used as a binary feature space. -class IntFeatureSpace { +class TESS_API IntFeatureSpace { public: IntFeatureSpace(); // Default copy constructors and assignment OK! diff --git a/classify/intfx.cpp b/classify/intfx.cpp index 78aa59bbc9..9c9870a2e9 100644 --- a/classify/intfx.cpp +++ b/classify/intfx.cpp @@ -520,7 +520,7 @@ bool ExtractIntFeat(const TBLOB& blob, tesseract::Classify::ExtractFeatures(blob, nonlinear_norm, &bl_features, &cn_features, results, NULL); - if (bl_features.size() == 0 || cn_features.size() == 0 || + if (bl_features.empty() || cn_features.empty() || bl_features.size() > MAX_NUM_INT_FEATURES || cn_features.size() > MAX_NUM_INT_FEATURES) { return false; // Feature extraction failed. diff --git a/classify/intfx.h b/classify/intfx.h index 26c435374c..1cd51a3022 100644 --- a/classify/intfx.h +++ b/classify/intfx.h @@ -48,7 +48,7 @@ const double kStandardFeatureLength = 64.0 / 5; /**---------------------------------------------------------------------------- Public Function Prototypes ----------------------------------------------------------------------------**/ -void InitIntegerFX(); +void TESS_API InitIntegerFX(); // Returns a vector representing the direction of a feature with the given // theta direction in an INT_FEATURE_STRUCT. diff --git a/classify/intmatcher.cpp b/classify/intmatcher.cpp index 8fc135ea45..ff999608f2 100644 --- a/classify/intmatcher.cpp +++ b/classify/intmatcher.cpp @@ -295,7 +295,8 @@ class ClassPruner { HeapSort(num_classes_, sort_key_, sort_index_); } - /** Prints debug info on the class pruner matches for the pruned classes only. */ + /** Prints debug info on the class pruner matches for the pruned classes only. + */ void DebugMatch(const Classify& classify, const INT_TEMPLATES_STRUCT* int_templates, const INT_FEATURE_STRUCT* features) const { @@ -370,8 +371,9 @@ class ClassPruner { private: /** Array[rounded_classes_] of initial counts for each class. */ int *class_count_; - /// Array[rounded_classes_] of modified counts for each class after normalizing - /// for expected number of features, disabled classes, fragments, and xheights. + /// Array[rounded_classes_] of modified counts for each class after + /// normalizing for expected number of features, disabled classes, fragments, + /// and xheights. int *norm_count_; /** Array[rounded_classes_ +1] of pruned counts that gets sorted */ int *sort_key_; @@ -402,8 +404,9 @@ class ClassPruner { * normalization process (by CLASS_INDEX) * @param expected_num_features Array of expected number of features * for each class (by CLASS_INDEX) - * @param results Sorted Array of pruned classes. Must be an array - * of size at least int_templates->NumClasses. + * @param results Sorted Array of pruned classes. Must be an + * array of size at least + * int_templates->NumClasses. * @param keep_this */ int Classify::PruneClasses(const INT_TEMPLATES_STRUCT* int_templates, @@ -606,7 +609,6 @@ int IntegerMatcher::FindGoodProtos( return NumGoodProtos; } - /** * FindBadFeatures finds all features with maximum feature-evidence < * AdaptFeatureThresh. The list is ordered by increasing feature number. @@ -701,7 +703,6 @@ void IntegerMatcher::Init(tesseract::IntParam *classify_debug_level) { evidence_mult_mask_ = ((1 << kIntEvidenceTruncBits) - 1); } - /*---------------------------------------------------------------------------- Private Code ----------------------------------------------------------------------------*/ @@ -717,8 +718,6 @@ void ScratchEvidence::ClearFeatureEvidence(const INT_CLASS class_template) { class_template->NumConfigs * sizeof(feature_evidence_[0])); } - - /** * Print debugging information for Configuations * @return none @@ -742,7 +741,6 @@ void IMDebugConfiguration(int FeatureNum, cprintf ("\n"); } - /** * Print debugging information for Configuations * @return none @@ -795,10 +793,10 @@ int IntegerMatcher::UpdateTablesForFeature( uinT32 XFeatureAddress; uinT32 YFeatureAddress; uinT32 ThetaFeatureAddress; - uinT8 *UINT8Pointer; + uinT8* UINT8Pointer; int ProtoIndex; uinT8 Temp; - int *IntPointer; + int* IntPointer; int ConfigNum; inT32 M3; inT32 A3; @@ -916,7 +914,6 @@ int IntegerMatcher::UpdateTablesForFeature( return SumOverConfigs; } - /** * Print debugging information for Configuations * @return none @@ -1165,8 +1162,6 @@ void ScratchEvidence::UpdateSumOfProtoEvidences( } } - - /** * Normalize Sum of Proto and Feature Evidence by dividing by the sum of * the Feature Lengths and the Proto Lengths for each configuration. @@ -1180,7 +1175,6 @@ void ScratchEvidence::NormalizeSums( } } - /** * Find the best match for the current class and update the Result * with the configuration and match rating. diff --git a/classify/intmatcher.h b/classify/intmatcher.h index 46dbfc5a8d..df678d75ed 100644 --- a/classify/intmatcher.h +++ b/classify/intmatcher.h @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: intmatcher.h - ** Purpose: Interface to high level generic classifier routines. - ** Author: Robert Moss - ** History: Wed Feb 13 15:24:15 MST 1991, RWM, Created. + ** Filename: intmatcher.h + ** Purpose: Interface to high level generic classifier routines. + ** Author: Robert Moss + ** History: Wed Feb 13 15:24:15 MST 1991, RWM, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at diff --git a/classify/intproto.cpp b/classify/intproto.cpp index 4c2f0d9536..473ffd96db 100644 --- a/classify/intproto.cpp +++ b/classify/intproto.cpp @@ -46,9 +46,7 @@ #include "config_auto.h" #endif -using tesseract::FontInfo; using tesseract::FontSet; -using tesseract::FontSpacingInfo; /* match debug display constants*/ #define PROTO_PRUNER_SCALE (4.0) @@ -326,10 +324,8 @@ int AddIntProto(INT_CLASS Class) { Word < Proto->Configs + WERDS_PER_CONFIG_VEC; *Word++ = 0); return (Index); - } - /** * This routine adds Proto to the class pruning tables * for the specified class in Templates. @@ -372,7 +368,6 @@ void AddProtoToClassPruner (PROTO Proto, CLASS_ID ClassId, } } /* AddProtoToClassPruner */ - /** * This routine updates the proto pruner lookup tables * for Class to include a new proto identified by ProtoId @@ -432,7 +427,6 @@ void AddProtoToProtoPruner(PROTO Proto, int ProtoId, FillPPLinearBits(ProtoSet->ProtoPruner[PRUNER_Y], Index, Y, Pad, debug); } /* AddProtoToProtoPruner */ - /** * Returns a quantized bucket for the given param shifted by offset, * notionally (param + offset) * num_buckets, but clipped and casted to the @@ -550,7 +544,6 @@ void Classify::ConvertProto(PROTO Proto, int ProtoId, INT_CLASS Class) { P->A, P->B, P->C, Class->ProtoLengths[ProtoId]); } /* ConvertProto */ - /** * This routine converts from the old floating point format * to the new integer format. @@ -627,7 +620,7 @@ INT_TEMPLATES Classify::CreateIntTemplates(CLASSES FloatProtos, * @note Exceptions: none * @note History: Thu Mar 21 14:45:04 1991, DSJ, Created. */ -void DisplayIntFeature(const INT_FEATURE_STRUCT* Feature, FLOAT32 Evidence) { +void DisplayIntFeature(const INT_FEATURE_STRUCT *Feature, FLOAT32 Evidence) { ScrollView::Color color = GetMatchColorFor(Evidence); RenderIntFeature(IntMatchWindow, Feature, color); if (FeatureDisplayWindow) { @@ -635,7 +628,6 @@ void DisplayIntFeature(const INT_FEATURE_STRUCT* Feature, FLOAT32 Evidence) { } } /* DisplayIntFeature */ - /** * This routine renders the specified proto into a * global display list. @@ -720,7 +712,6 @@ void free_int_class(INT_CLASS int_class) { Efree(int_class); } - /** * This routine allocates a new set of integer templates * initialized to hold 0 classes. @@ -1218,7 +1209,6 @@ FLOAT32 BucketStart(int Bucket, FLOAT32 Offset, int NumBuckets) { } /* BucketStart */ - /** * This routine returns the parameter value which * corresponds to the end of the specified bucket. @@ -1236,7 +1226,6 @@ FLOAT32 BucketEnd(int Bucket, FLOAT32 Offset, int NumBuckets) { return (((FLOAT32) (Bucket + 1) / NumBuckets) - Offset); } /* BucketEnd */ - /** * This routine fills in the section of a class pruner * corresponding to a single x value for a single proto of @@ -1284,7 +1273,6 @@ void DoFill(FILL_SPEC *FillSpec, } } /* DoFill */ - /** * Return TRUE if the specified table filler is done, i.e. * if it has no more lines to fill. @@ -1306,7 +1294,6 @@ BOOL8 FillerDone(TABLE_FILLER *Filler) { } /* FillerDone */ - /** * This routine sets Bit in each bit vector whose * bucket lies within the range Center +- Spread. The fill @@ -1349,7 +1336,6 @@ void FillPPCircularBits(uinT32 ParamTable[NUM_PP_BUCKETS][WERDS_PER_PP_VECTOR], } /* FillPPCircularBits */ - /** * This routine sets Bit in each bit vector whose * bucket lies within the range Center +- Spread. The fill @@ -1516,7 +1502,6 @@ void GetCPPadsForLevel(int Level, } /* GetCPPadsForLevel */ - /** * @param Evidence evidence value to return color for * @return Color which corresponds to specified Evidence value. @@ -1538,7 +1523,6 @@ ScrollView::Color GetMatchColorFor(FLOAT32 Evidence) { return ScrollView::BLUE; } /* GetMatchColorFor */ - /** * This routine returns (in Fill) the specification of * the next line to be filled from Filler. FillerDone() should @@ -1589,7 +1573,6 @@ void GetNextFill(TABLE_FILLER *Filler, FILL_SPEC *Fill) { } /* GetNextFill */ - /** * This routine computes a data structure (Filler) * which can be used to fill in a rectangle surrounding @@ -1723,8 +1706,10 @@ void InitTableFiller (FLOAT32 EndPad, FLOAT32 SidePad, /* translate into bucket positions and deltas */ Filler->X = Bucket8For(Start.x, XS, NB); - Filler->StartDelta = -(inT16) ((Sin / Cos) * 256); - Filler->EndDelta = (inT16) ((Cos / Sin) * 256); + Filler->StartDelta = static_cast(ClipToRange( + -IntCastRounded((Sin / Cos) * 256), MIN_INT16, MAX_INT16)); + Filler->EndDelta = static_cast(ClipToRange( + IntCastRounded((Cos / Sin) * 256), MIN_INT16, MAX_INT16)); XAdjust = BucketEnd(Filler->X, XS, NB) - Start.x; YAdjust = XAdjust * Sin / Cos; @@ -1787,7 +1772,6 @@ void RenderIntFeature(ScrollView *window, const INT_FEATURE_STRUCT* Feature, window->DrawTo(X + Dx, Y + Dy); } /* RenderIntFeature */ - /** * This routine extracts the parameters of the specified * proto from the class description and adds a rendering of diff --git a/classify/intproto.h b/classify/intproto.h index d2c07147a9..262974b83e 100644 --- a/classify/intproto.h +++ b/classify/intproto.h @@ -52,7 +52,7 @@ class FCOORD; #define NUM_CP_BUCKETS 24 #define CLASSES_PER_CP 32 #define NUM_BITS_PER_CLASS 2 -#define CLASS_PRUNER_CLASS_MASK (~(~0 << NUM_BITS_PER_CLASS)) +#define CLASS_PRUNER_CLASS_MASK (~(~0u << NUM_BITS_PER_CLASS)) #define CLASSES_PER_CP_WERD (CLASSES_PER_CP / NUM_BITS_PER_CLASS) #define PROTOS_PER_PP_WERD BITS_PER_WERD #define BITS_PER_CP_VECTOR (CLASSES_PER_CP * NUM_BITS_PER_CLASS) diff --git a/classify/kdtree.cpp b/classify/kdtree.cpp index 61a94f66cc..6ba7086d23 100644 --- a/classify/kdtree.cpp +++ b/classify/kdtree.cpp @@ -70,11 +70,11 @@ class MinK { const Element* elements() { return elements_; } private: - const Key max_key_; //< the maximum possible Key - Element* elements_; //< unsorted array of elements + const Key max_key_; //< the maximum possible Key + Element *elements_; //< unsorted array of elements int elements_count_; //< the number of results collected so far - int k_; //< the number of results we want from the search - int max_index_; //< the index of the result with the largest key + int k_; //< the number of results we want from the search + int max_index_; //< the index of the result with the largest key }; template @@ -117,7 +117,8 @@ bool MinK::insert(Key key, Value value) { //----------------------------------------------------------------------------- -/** Helper class for searching for the k closest points to query_point in tree. */ +/** Helper class for searching for the k closest points to query_point in tree. + */ class KDTreeSearch { public: KDTreeSearch(KDTREE* tree, FLOAT32 *query_point, int k_closest); @@ -241,14 +242,13 @@ void KDStore(KDTREE *Tree, FLOAT32 *Key, void *Data) { *PtrToNode = MakeKDNode(Tree, Key, (void *) Data, Level); } /* KDStore */ - /** - * This routine deletes a node from Tree. The node to be - * deleted is specified by the Key for the node and the Data - * contents of the node. These two pointers must be identical - * to the pointers that were used for the node when it was - * originally stored in the tree. A node will be deleted from - * the tree only if its key and data pointers are identical + * This routine deletes a node from Tree. The node to be + * deleted is specified by the Key for the node and the Data + * contents of the node. These two pointers must be identical + * to the pointers that were used for the node when it was + * originally stored in the tree. A node will be deleted from + * the tree only if its key and data pointers are identical * to Key and Data respectively. The tree is re-formed by removing * the affected subtree and inserting all elements but the root. * @@ -298,7 +298,6 @@ KDDelete (KDTREE * Tree, FLOAT32 Key[], void *Data) { } } /* KDDelete */ - /** * This routine searches the K-D tree specified by Tree and * finds the QuerySize nearest neighbors of Query. All neighbors @@ -442,7 +441,7 @@ void KDTreeSearch::SearchRec(int level, KDNODE *sub_tree) { /*---------------------------------------------------------------------------*/ -/** +/** *Returns the Euclidean distance squared between p1 and p2 for all essential * dimensions. * @param k keys are in k-space @@ -541,7 +540,6 @@ void Walk(KDTREE *tree, void_proc action, void *context, Walk(tree, action, context, sub_tree->Right, NextLevel(tree, level)); } - /** Given a subtree nodes, insert all of its elements into tree. */ void InsertNodes(KDTREE *tree, KDNODE *nodes) { if (nodes == NULL) diff --git a/classify/kdtree.h b/classify/kdtree.h index 1294ea3c6a..259d7078d0 100644 --- a/classify/kdtree.h +++ b/classify/kdtree.h @@ -83,7 +83,7 @@ void FreeKDNode(KDNODE *Node); FLOAT32 DistanceSquared(int k, PARAM_DESC *dim, FLOAT32 p1[], FLOAT32 p2[]); -FLOAT32 ComputeDistance(int k, PARAM_DESC *dim, FLOAT32 p1[], FLOAT32 p2[]); +FLOAT32 TESS_API ComputeDistance(int k, PARAM_DESC *dim, FLOAT32 p1[], FLOAT32 p2[]); int QueryInSearch(KDTREE *tree); diff --git a/classify/mastertrainer.cpp b/classify/mastertrainer.cpp index 866a617a22..849fb06010 100644 --- a/classify/mastertrainer.cpp +++ b/classify/mastertrainer.cpp @@ -214,10 +214,14 @@ void MasterTrainer::AddSample(bool verification, const char* unichar, // Must be called after ReadTrainingSamples, as the current number of images // is used as an offset for page numbers in the samples. void MasterTrainer::LoadPageImages(const char* filename) { + size_t offset = 0; int page; Pix* pix; - for (page = 0; (pix = pixReadTiff(filename, page)) != NULL; ++page) { + for (page = 0; ; page++) { + pix = pixReadFromMultipageTiff(filename, &offset); + if (!pix) break; page_images_.push_back(pix); + if (!offset) break; } tprintf("Loaded %d page images from %s\n", page, filename); } @@ -362,9 +366,11 @@ bool MasterTrainer::LoadFontInfo(const char* filename) { fontinfo.name = font_name; fontinfo.properties = 0; fontinfo.universal_id = 0; - if (tfscanf(fp, "%1024s %i %i %i %i %i\n", font_name, - &italic, &bold, &fixed, &serif, &fraktur) != 6) + if (tfscanf(fp, "%1024s %i %i %i %i %i\n", font_name, &italic, &bold, + &fixed, &serif, &fraktur) != 6) { + delete[] font_name; continue; + } fontinfo.properties = (italic << 0) + (bold << 1) + @@ -373,6 +379,8 @@ bool MasterTrainer::LoadFontInfo(const char* filename) { (fraktur << 4); if (!fontinfo_table_.contains(fontinfo)) { fontinfo_table_.push_back(fontinfo); + } else { + delete[] font_name; } } fclose(fp); diff --git a/classify/mastertrainer.h b/classify/mastertrainer.h index 8cc7158acf..25119d2382 100644 --- a/classify/mastertrainer.h +++ b/classify/mastertrainer.h @@ -66,7 +66,7 @@ struct ShapeDist { // Initially supports shape clustering and mftrainining. // Other important features of the MasterTrainer are conditioning the data // by outlier elimination, replication with perturbation, and serialization. -class MasterTrainer { +class TESS_API MasterTrainer { public: MasterTrainer(NormalizationMode norm_mode, bool shape_analysis, bool replicate_samples, int debug_level); diff --git a/classify/mf.cpp b/classify/mf.cpp index d0c59487e6..37cd2eca0a 100644 --- a/classify/mf.cpp +++ b/classify/mf.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: mf.c - ** Purpose: Micro-feature interface to flexible feature extractor. - ** Author: Dan Johnson - ** History: Thu May 24 09:08:38 1990, DSJ, Created. + ** Filename: mf.c + ** Purpose: Micro-feature interface to flexible feature extractor. + ** Author: Dan Johnson + ** History: Thu May 24 09:08:38 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -36,7 +36,7 @@ * Call the old micro-feature extractor and then copy * the features into the new format. Then deallocate the * old micro-features. - * @param Blob blob to extract micro-features from + * @param Blob blob to extract micro-features from * @param cn_denorm control parameter to feature extractor. * @return Micro-features for Blob. * @note Exceptions: none diff --git a/classify/mfdefs.cpp b/classify/mfdefs.cpp index abe8d0c71a..0f225e8b08 100644 --- a/classify/mfdefs.cpp +++ b/classify/mfdefs.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: mfdefs.c - ** Purpose: Basic routines for manipulating micro-features - ** Author: Dan Johnson - ** History: Mon Jan 22 08:48:58 1990, DSJ, Created. + ** Filename: mfdefs.c + ** Purpose: Basic routines for manipulating micro-features + ** Author: Dan Johnson + ** History: Mon Jan 22 08:48:58 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -32,7 +32,7 @@ * @return New MICROFEATURE * @note History: 7/27/89, DSJ, Created. */ -MICROFEATURE NewMicroFeature() { +MICROFEATURE NewMicroFeature() { return ((MICROFEATURE) Emalloc (sizeof (MFBLOCK))); } /* NewMicroFeature */ @@ -41,10 +41,10 @@ MICROFEATURE NewMicroFeature() { /** * This routine deallocates all of the memory consumed by * a list of micro-features. - * @param MicroFeatures list of micro-features to be freed + * @param MicroFeatures list of micro-features to be freed * @return none * @note History: 7/27/89, DSJ, Created. */ -void FreeMicroFeatures(MICROFEATURES MicroFeatures) { +void FreeMicroFeatures(MICROFEATURES MicroFeatures) { destroy_nodes(MicroFeatures, Efree); } /* FreeMicroFeatures */ diff --git a/classify/mfoutline.cpp b/classify/mfoutline.cpp index 511c34d41f..59593a8523 100644 --- a/classify/mfoutline.cpp +++ b/classify/mfoutline.cpp @@ -35,7 +35,8 @@ ----------------------------------------------------------------------------*/ /*---------------------------------------------------------------------------*/ -/** Convert a blob into a list of MFOUTLINEs (float-based microfeature format). */ +/** Convert a blob into a list of MFOUTLINEs (float-based microfeature format). + */ LIST ConvertBlob(TBLOB *blob) { LIST outlines = NIL_LIST; return (blob == NULL) @@ -344,7 +345,6 @@ void ChangeDirection(MFOUTLINE Start, MFOUTLINE End, DIRECTION Direction) { } /* ChangeDirection */ - /** * This routine normalizes each point in Outline by * translating it to the specified center and scaling it @@ -378,7 +378,6 @@ void CharNormalizeOutline(MFOUTLINE Outline, const DENORM& cn_denorm) { } /* CharNormalizeOutline */ - /** * This routine computes the slope from Start to Finish and * and then computes the approximate direction of the line diff --git a/classify/mfx.cpp b/classify/mfx.cpp index 3da4fb3d0f..6fd8ed5da5 100644 --- a/classify/mfx.cpp +++ b/classify/mfx.cpp @@ -128,7 +128,6 @@ FLOAT32 ComputeOrientation(MFEDGEPT *Start, MFEDGEPT *End) { return (Orientation); } /* ComputeOrientation */ - /** * Convert Outline to MicroFeatures * @param Outline outline to extract micro-features from @@ -164,7 +163,6 @@ MICROFEATURES ConvertToMicroFeatures(MFOUTLINE Outline, return (MicroFeatures); } /* ConvertToMicroFeatures */ - /** * This routine computes the feature parameters which describe * the micro-feature that starts and Start and ends at End. @@ -178,7 +176,7 @@ MICROFEATURES ConvertToMicroFeatures(MFOUTLINE Outline, * @return New micro-feature or NULL if the feature was rejected. * @note Globals: none * @note Exceptions: none - * @note History: + * @note History: * - 7/26/89, DSJ, Created. * - 11/17/89, DSJ, Added handling for Start and End same point. */ diff --git a/classify/mfx.h b/classify/mfx.h index 05ce29cee5..5ed006dcc7 100644 --- a/classify/mfx.h +++ b/classify/mfx.h @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: mfx.h - ** Purpose: Definition of micro-feature extraction routines - ** Author: Dan Johnson - ** History: 5/29/89, DSJ, Created. + ** Filename: mfx.h + ** Purpose: Definition of micro-feature extraction routines + ** Author: Dan Johnson + ** History: 5/29/89, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at diff --git a/classify/normfeat.cpp b/classify/normfeat.cpp index a4ac672a11..f297b3b05d 100644 --- a/classify/normfeat.cpp +++ b/classify/normfeat.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: normfeat.c - ** Purpose: Definition of char normalization features. - ** Author: Dan Johnson - ** History: 12/14/90, DSJ, Created. + ** Filename: normfeat.c + ** Purpose: Definition of char normalization features. + ** Author: Dan Johnson + ** History: 12/14/90, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -33,7 +33,6 @@ FLOAT32 ActualOutlineLength(FEATURE Feature) { return (Feature->Params[CharNormLength] * LENGTH_COMPRESSION); } - /** * Return the character normalization feature for a blob. * diff --git a/classify/normmatch.cpp b/classify/normmatch.cpp index 488cd1652f..b0e066a89d 100644 --- a/classify/normmatch.cpp +++ b/classify/normmatch.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: normmatch.c - ** Purpose: Simple matcher based on character normalization features. - ** Author: Dan Johnson - ** History: Wed Dec 19 16:18:06 1990, DSJ, Created. + ** Filename: normmatch.c + ** Purpose: Simple matcher based on character normalization features. + ** Author: Dan Johnson + ** History: Wed Dec 19 16:18:06 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -197,10 +197,10 @@ double NormEvidenceOf(register double NormAdj) { /*---------------------------------------------------------------------------*/ /** * This routine dumps out detailed normalization match info. - * @param File open text file to dump match debug info to - * @param NumParams # of parameters in proto and feature - * @param Proto[] array of prototype parameters - * @param Feature[] array of feature parameters + * @param File open text file to dump match debug info to + * @param NumParams # of parameters in proto and feature + * @param Proto[] array of prototype parameters + * @param Feature[] array of feature parameters * Globals: none * @return none * @note Exceptions: none diff --git a/classify/ocrfeatures.cpp b/classify/ocrfeatures.cpp index 0895ed0886..7df8135048 100644 --- a/classify/ocrfeatures.cpp +++ b/classify/ocrfeatures.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: features.c - ** Purpose: Generic definition of a feature. - ** Author: Dan Johnson - ** History: Mon May 21 10:49:04 1990, DSJ, Created. + ** Filename: features.c + ** Purpose: Generic definition of a feature. + ** Author: Dan Johnson + ** History: Mon May 21 10:49:04 1990, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -66,12 +66,11 @@ void FreeFeature(FEATURE Feature) { } /* FreeFeature */ - /** * Release the memory consumed by the specified feature * set. This routine also frees the memory consumed by the * features contained in the set. - * @param FeatureSet set of features to be freed + * @param FeatureSet set of features to be freed * @return none * @note History: Mon May 21 13:59:46 1990, DSJ, Created. */ @@ -85,11 +84,10 @@ void FreeFeatureSet(FEATURE_SET FeatureSet) { } } /* FreeFeatureSet */ - /** * Allocate and return a new feature of the specified * type. - * @param FeatureDesc description of feature to be created. + * @param FeatureDesc description of feature to be created. * @return New #FEATURE. * @note History: Mon May 21 14:06:42 1990, DSJ, Created. */ @@ -105,11 +103,10 @@ FEATURE NewFeature(const FEATURE_DESC_STRUCT* FeatureDesc) { } /* NewFeature */ - /** * Allocate and return a new feature set large enough to * hold the specified number of features. - * @param NumFeatures maximum # of features to be put in feature set + * @param NumFeatures maximum # of features to be put in feature set * @return New #FEATURE_SET. * @note History: Mon May 21 14:22:40 1990, DSJ, Created. */ @@ -124,7 +121,6 @@ FEATURE_SET NewFeatureSet(int NumFeatures) { } /* NewFeatureSet */ - /** * Create a new feature of the specified type and read in * the value of its parameters from File. The extra penalty @@ -135,10 +131,11 @@ FEATURE_SET NewFeatureSet(int NumFeatures) { * @param File open text file to read feature from * @param FeatureDesc specifies type of feature to read from File * @return New #FEATURE read from File. - * @note Exceptions: #ILLEGAL_FEATURE_PARAM if text file doesn't match expected format + * @note Exceptions: #ILLEGAL_FEATURE_PARAM if text file doesn't match expected + * format * @note History: Wed May 23 08:53:16 1990, DSJ, Created. */ -FEATURE ReadFeature(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { +FEATURE ReadFeature(FILE* File, const FEATURE_DESC_STRUCT* FeatureDesc) { FEATURE Feature; int i; @@ -153,7 +150,6 @@ FEATURE ReadFeature(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { return (Feature); } /* ReadFeature */ - /** * Create a new feature set of the specified type and read in * the features from File. The correct text representation @@ -165,7 +161,7 @@ FEATURE ReadFeature(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { * @return New feature set read from File. * @note History: Wed May 23 09:17:31 1990, DSJ, Created. */ -FEATURE_SET ReadFeatureSet(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { +FEATURE_SET ReadFeatureSet(FILE* File, const FEATURE_DESC_STRUCT* FeatureDesc) { FEATURE_SET FeatureSet; int NumFeatures; int i; @@ -180,7 +176,6 @@ FEATURE_SET ReadFeatureSet(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { return (FeatureSet); } /* ReadFeatureSet */ - /** * Appends a textual representation of Feature to str. * This representation is simply a list of the N parameters @@ -203,7 +198,6 @@ void WriteFeature(FEATURE Feature, STRING* str) { *str += "\n"; } /* WriteFeature */ - /** * Write a textual representation of FeatureSet to File. * This representation is an integer specifying the number of @@ -224,7 +218,6 @@ void WriteFeatureSet(FEATURE_SET FeatureSet, STRING* str) { } } /* WriteFeatureSet */ - /** * Write a textual representation of FeatureDesc to File * in the old format (i.e. the format used by the clusterer). @@ -240,7 +233,7 @@ void WriteFeatureSet(FEATURE_SET FeatureSet, STRING* str) { * @return none * @note History: Fri May 25 15:27:18 1990, DSJ, Created. */ -void WriteOldParamDesc(FILE *File, const FEATURE_DESC_STRUCT* FeatureDesc) { +void WriteOldParamDesc(FILE* File, const FEATURE_DESC_STRUCT* FeatureDesc) { int i; fprintf (File, "%d\n", FeatureDesc->NumParams); diff --git a/classify/ocrfeatures.h b/classify/ocrfeatures.h index 31a4794ca6..ae00e5a834 100644 --- a/classify/ocrfeatures.h +++ b/classify/ocrfeatures.h @@ -108,7 +108,7 @@ BOOL8 AddFeature(FEATURE_SET FeatureSet, FEATURE Feature); void FreeFeature(FEATURE Feature); -void FreeFeatureSet(FEATURE_SET FeatureSet); +void TESS_API FreeFeatureSet(FEATURE_SET FeatureSet); FEATURE NewFeature(const FEATURE_DESC_STRUCT *FeatureDesc); diff --git a/classify/outfeat.cpp b/classify/outfeat.cpp index b1a4a9be90..76597f7c15 100644 --- a/classify/outfeat.cpp +++ b/classify/outfeat.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: outfeat.c - ** Purpose: Definition of outline-features. - ** Author: Dan Johnson - ** History: 11/13/90, DSJ, Created. + ** Filename: outfeat.c + ** Purpose: Definition of outline-features. + ** Author: Dan Johnson + ** History: 11/13/90, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -40,7 +40,7 @@ namespace tesseract { * @return Outline-features for Blob. * @note Globals: none * @note Exceptions: none - * @note History: + * @note History: * - 11/13/90, DSJ, Created. * - 05/24/91, DSJ, Updated for either char or baseline normalize. */ @@ -115,7 +115,7 @@ void AddOutlineFeatureToSet(FPOINT *Start, * @return none (results are returned in FeatureSet) * @note Globals: none * @note Exceptions: none - * @note History: + * @note History: * - 11/13/90, DSJ, Created. * - 5/24/91, DSJ, Added hidden edge capability. */ diff --git a/classify/picofeat.cpp b/classify/picofeat.cpp index 74beb18f35..a4a39263cf 100644 --- a/classify/picofeat.cpp +++ b/classify/picofeat.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: picofeat.c - ** Purpose: Definition of pico-features. - ** Author: Dan Johnson - ** History: 9/4/90, DSJ, Created. + ** Filename: picofeat.c + ** Purpose: Definition of pico-features. + ** Author: Dan Johnson + ** History: 9/4/90, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -98,7 +98,7 @@ FEATURE_SET Classify::ExtractPicoFeatures(TBLOB *Blob) { * nearest whole number of pico-features. The pico-features * are spaced evenly over the entire segment. * Globals: - * - classify_pico_feature_length length of a single pico-feature + * - classify_pico_feature_length length of a single pico-feature * @param Start starting point of pico-feature * @param End ending point of pico-feature * @param FeatureSet set to add pico-feature to diff --git a/classify/picofeat.h b/classify/picofeat.h index 208b7e7708..966ffc32e7 100644 --- a/classify/picofeat.h +++ b/classify/picofeat.h @@ -61,5 +61,5 @@ extern double_VAR_H(classify_pico_feature_length, 0.05, "Pico Feature Length"); /**---------------------------------------------------------------------------- Global Data Definitions and Declarations ----------------------------------------------------------------------------**/ -extern FLOAT32 PicoFeatureLength; +extern TESS_API FLOAT32 PicoFeatureLength; #endif diff --git a/classify/protos.h b/classify/protos.h index bb2f56b59f..7d36447915 100644 --- a/classify/protos.h +++ b/classify/protos.h @@ -166,13 +166,13 @@ void CopyProto(PROTO Src, PROTO Dest); void FillABC(PROTO Proto); -void FreeClass(CLASS_TYPE Class); +void TESS_API FreeClass(CLASS_TYPE Class); void FreeClassFields(CLASS_TYPE Class); void InitPrototypes(); -CLASS_TYPE NewClass(int NumProtos, int NumConfigs); +CLASS_TYPE TESS_API NewClass(int NumProtos, int NumConfigs); void PrintProtos(CLASS_TYPE Class); diff --git a/classify/shapeclassifier.cpp b/classify/shapeclassifier.cpp index a39c8a24bb..e0ee3373d8 100644 --- a/classify/shapeclassifier.cpp +++ b/classify/shapeclassifier.cpp @@ -176,7 +176,7 @@ void ShapeClassifier::UnicharPrintResults( for (int i = 0; i < results.size(); ++i) { tprintf("%g: c_id=%d=%s", results[i].rating, results[i].unichar_id, GetUnicharset().id_to_unichar(results[i].unichar_id)); - if (results[i].fonts.size() != 0) { + if (!results[i].fonts.empty()) { tprintf(" Font Vector:"); for (int f = 0; f < results[i].fonts.size(); ++f) { tprintf(" %d", results[i].fonts[f].fontinfo_id); diff --git a/classify/shapetable.h b/classify/shapetable.h index d8faae8817..59c3fbc6df 100644 --- a/classify/shapetable.h +++ b/classify/shapetable.h @@ -261,7 +261,7 @@ class Shape { // that the shape represents. // Each UnicharAndFonts also lists the fonts of the unichar_id that were // mapped to the shape during training. -class ShapeTable { +class TESS_API ShapeTable { public: ShapeTable(); // The UNICHARSET reference supplied here, or in set_unicharset below must diff --git a/classify/tessclassifier.h b/classify/tessclassifier.h index 57a04861e2..fcf1870492 100644 --- a/classify/tessclassifier.h +++ b/classify/tessclassifier.h @@ -23,6 +23,7 @@ #define THIRD_PARTY_TESSERACT_CLASSIFY_TESSCLASSIFIER_H_ #include "shapeclassifier.h" +#include "platform.h" namespace tesseract { @@ -33,7 +34,7 @@ class TrainingSample; // Due to limitations in the content of TrainingSample, this currently // only works for the static classifier and only works if the ShapeTable // in classify is not NULL. -class TessClassifier : public ShapeClassifier { +class TESS_API TessClassifier : public ShapeClassifier { public: TessClassifier(bool pruner_only, tesseract::Classify* classify) : pruner_only_(pruner_only), classify_(classify) {} diff --git a/classify/trainingsample.cpp b/classify/trainingsample.cpp index 7fe83b7718..ee6c9d7f83 100644 --- a/classify/trainingsample.cpp +++ b/classify/trainingsample.cpp @@ -209,7 +209,7 @@ void TrainingSample::ExtractCharDesc(int int_feature_type, int geo_type, CHAR_DESC_STRUCT* char_desc) { // Extract the INT features. - if (features_ != NULL) delete [] features_; + delete[] features_; FEATURE_SET_STRUCT* char_features = char_desc->FeatureSets[int_feature_type]; if (char_features == NULL) { tprintf("Error: no features to train on of type %s\n", @@ -230,7 +230,7 @@ void TrainingSample::ExtractCharDesc(int int_feature_type, } } // Extract the Micro features. - if (micro_features_ != NULL) delete [] micro_features_; + delete[] micro_features_; char_features = char_desc->FeatureSets[micro_type]; if (char_features == NULL) { tprintf("Error: no features to train on of type %s\n", diff --git a/classify/trainingsampleset.cpp b/classify/trainingsampleset.cpp index afbf3f420e..93936fcae6 100644 --- a/classify/trainingsampleset.cpp +++ b/classify/trainingsampleset.cpp @@ -96,10 +96,8 @@ bool TrainingSampleSet::DeSerialize(bool swap, FILE* fp) { num_raw_samples_ = samples_.size(); if (!unicharset_.load_from_file(fp)) return false; if (!font_id_map_.DeSerialize(swap, fp)) return false; - if (font_class_array_ != NULL) { - delete font_class_array_; - font_class_array_ = NULL; - } + delete font_class_array_; + font_class_array_ = NULL; inT8 not_null; if (fread(¬_null, sizeof(not_null), 1, fp) != 1) return false; if (not_null) { diff --git a/cmake/BuildFunctions.cmake b/cmake/BuildFunctions.cmake index eea5a396cb..39fd6d7071 100644 --- a/cmake/BuildFunctions.cmake +++ b/cmake/BuildFunctions.cmake @@ -1,3 +1,12 @@ +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. ################################################################################ # # macros and functions diff --git a/cmake/Configure.cmake b/cmake/Configure.cmake index d7f4ac6ad5..bd5b80c5a7 100644 --- a/cmake/Configure.cmake +++ b/cmake/Configure.cmake @@ -1,3 +1,12 @@ +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. ################################################################################ # # configure diff --git a/cmake/FindICU.cmake b/cmake/FindICU.cmake index cd6bf9265c..8381c2eb84 100644 --- a/cmake/FindICU.cmake +++ b/cmake/FindICU.cmake @@ -1,3 +1,12 @@ +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. # This module can find the International Components for Unicode (ICU) Library # # Requirements: diff --git a/cmake/SourceGroups.cmake b/cmake/SourceGroups.cmake index ca87e808d3..7e79927e6c 100644 --- a/cmake/SourceGroups.cmake +++ b/cmake/SourceGroups.cmake @@ -1,3 +1,12 @@ +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. #include(SourceGroups) set(SSRC ${CMAKE_SOURCE_DIR}) @@ -14,6 +23,7 @@ set(H_CPP "(${H}|${CPP})") source_group("Resource files" ".*\\.(rc|ico)") source_group("api" "${SSRC}/api/${H_CPP}") +source_group("arch" "${SSRC}/arch/${H_CPP}") source_group("ccmain" "${SSRC}/ccmain/${H_CPP}") source_group("ccstruct" "${SSRC}/ccstruct/${H_CPP}") source_group("ccutil" "${SSRC}/ccutil/${H_CPP}") @@ -21,6 +31,7 @@ source_group("classify" "${SSRC}/classify/${H_CPP}") source_group("cube" "${SSRC}/cube/${H_CPP}") source_group("cutil" "${SSRC}/cutil/${H_CPP}") source_group("dict" "${SSRC}/dict/${H_CPP}") +source_group("lstm" "${SSRC}/lstm/${H_CPP}") source_group("neural" "${SSRC}/neural_networks/runtime/${H_CPP}") source_group("opencl" "${SSRC}/opencl/${H_CPP}") source_group("textord" "${SSRC}/textord/${H_CPP}") diff --git a/configure.ac b/configure.ac index a775e4fc28..f02cea108e 100644 --- a/configure.ac +++ b/configure.ac @@ -5,8 +5,8 @@ # ---------------------------------------- # Initialization # ---------------------------------------- -AC_PREREQ([2.50]) -AC_INIT([tesseract], [3.05.00dev], [https://github.com/tesseract-ocr/tesseract/issues]) +AC_PREREQ([2.59]) +AC_INIT([tesseract], [3.05.00], [https://github.com/tesseract-ocr/tesseract/issues]) AC_PROG_CXX([g++ clang++]) AC_LANG([C++]) AC_LANG_COMPILER_REQUIRE @@ -18,8 +18,8 @@ AC_PREFIX_DEFAULT([/usr/local]) # Define date of package, etc. Could be useful in auto-generated # documentation. -PACKAGE_YEAR=2015 -PACKAGE_DATE="07/11" +PACKAGE_YEAR=2017 +PACKAGE_DATE="02/16" abs_top_srcdir=`AS_DIRNAME([$0])` gitrev="`git --git-dir=${abs_top_srcdir}/.git --work-tree=${abs_top_srcdir} describe --always --tags`" @@ -43,7 +43,7 @@ GENERIC_LIBRARY_NAME=tesseract # Release versioning GENERIC_MAJOR_VERSION=3 -GENERIC_MINOR_VERSION=4 +GENERIC_MINOR_VERSION=5 GENERIC_MICRO_VERSION=0 # API version (often = GENERIC_MAJOR_VERSION.GENERIC_MINOR_VERSION) @@ -75,6 +75,7 @@ AM_CONDITIONAL([T_WIN], false) AM_CONDITIONAL([MINGW], false) AM_CONDITIONAL([OSX], false) AM_CONDITIONAL([GRAPHICS_DISABLED], false) +AC_SUBST([AM_CPPFLAGS]) OPENCL_INC="/opt/AMDAPP/include" OPENCL_LIBS="-lOpenCL -ltiff" @@ -117,7 +118,7 @@ esac includedir="${includedir}/tesseract" AC_ARG_WITH([extra-includes], - [AC_HELP_STRING([--with-extra-includes=DIR], + [AS_HELP_STRING([--with-extra-includes=DIR], [Define an additional directory for include files])], [if test -d "$withval" ; then CFLAGS="$CFLAGS -I$withval" @@ -126,7 +127,7 @@ AC_ARG_WITH([extra-includes], fi]) AC_ARG_WITH([extra-libraries], - [AC_HELP_STRING([--with-extra-libraries=DIR], + [AS_HELP_STRING([--with-extra-libraries=DIR], [Define an additional directory for library files])], [if test -d "$withval" ; then LDFLAGS="$LDFLAGS -L$withval" @@ -136,8 +137,8 @@ AC_ARG_WITH([extra-libraries], AC_MSG_CHECKING([--enable-graphics argument]) AC_ARG_ENABLE([graphics], - [AC_HELP_STRING([--enable-graphics],[enable graphics (ScrollView) (default)]) -AC_HELP_STRING([--disable-graphics],[disable graphics (ScrollView)])], + [AS_HELP_STRING([--enable-graphics],[enable graphics (ScrollView) (default)]) +AS_HELP_STRING([--disable-graphics],[disable graphics (ScrollView)])], [enable_graphics=$enableval], [enable_graphics="yes"]) AC_MSG_RESULT([$enable_graphics]) @@ -149,12 +150,12 @@ fi # Check if cube should be disabled AC_MSG_CHECKING([whether to disable cube]) AC_ARG_ENABLE([cube], - [AC_HELP_STRING([--disable-cube], [don't build cube support (experimental)])], + [AS_HELP_STRING([--disable-cube], [don't build cube support (experimental)])], [disable_cube="yes"], [disable_cube="no"]) AC_MSG_RESULT([$disable_cube]) AM_CONDITIONAL([NO_CUBE_BUILD], [test "$disable_cube" = "yes"]) if test "$disable_cube" = "yes"; then - AC_SUBST([AM_CPPFLAGS], [-DNO_CUBE_BUILD]) + AM_CPPFLAGS="-DNO_CUBE_BUILD $AM_CPPFLAGS" fi # check whether to build embedded version @@ -166,19 +167,9 @@ AC_ARG_ENABLE([embedded], AC_MSG_RESULT([$enable_embedded]) AM_CONDITIONAL([EMBEDDED], [test "$enable_embedded" = "yes"]) if test "$enable_embedded" = "yes"; then - AC_SUBST([AM_CPPFLAGS], [-DEMBEDDED]) + AM_CPPFLAGS="-DEMBEDDED $AM_CPPFLAGS" fi -# check whether to build OpenMP support -AM_CONDITIONAL([OPENMP], false) -AC_OPENMP -AS_IF([test "x$OPENMP_CFLAGS" != "x"], - [AM_CONDITIONAL([OPENMP], true) - AC_SUBST([AM_CPPFLAGS], ["$OPENMP_CXXFLAGS"]) - AC_DEFINE([OPENMP], [], [Defined when compiled with OpenMP support])] -) - - # check whether to build opencl version AC_MSG_CHECKING([--enable-opencl argument]) AC_ARG_ENABLE([opencl], @@ -227,7 +218,7 @@ case "${host_os}" in if !($have_opencl_lib); then AC_MSG_ERROR([Required OpenCL library not found!]) fi - AC_SUBST([AM_CPPFLAGS], [-DUSE_OPENCL]) + AM_CPPFLAGS="-DUSE_OPENCL $AM_CPPFLAGS" OPENCL_CPPFLAGS="" OPENCL_LDFLAGS="-framework OpenCL" fi @@ -246,7 +237,7 @@ case "${host_os}" in if !($have_tiff); then AC_MSG_ERROR([Required TIFF headers not found! Try to install libtiff-dev?? package.]) fi - AC_SUBST([AM_CPPFLAGS], [-DUSE_OPENCL]) + AM_CPPFLAGS="-DUSE_OPENCL $AM_CPPFLAGS" OPENCL_CPPFLAGS="-I${OPENCL_INC}" OPENCL_LDFLAGS="${OPENCL_LIBS}" fi @@ -261,7 +252,7 @@ AC_SUBST([OPENCL_LDFLAGS]) # http://groups.google.com/group/tesseract-dev/browse_thread/thread/976645ae98189127 AC_MSG_CHECKING([--enable-visibility argument]) AC_ARG_ENABLE([visibility], - [AC_HELP_STRING([--enable-visibility],[enable experimental build with fvisibility (default=no)])], + [AS_HELP_STRING([--enable-visibility],[enable experimental build with fvisibility (default=no)])], [enable_visibility=$enableval], [enable_visibility="no"]) AC_MSG_RESULT([$enable_visibility]) @@ -270,7 +261,7 @@ AM_CONDITIONAL([VISIBILITY], [test "$enable_visibility" = "yes"]) # check whether to build multiple libraries AC_MSG_CHECKING([--enable-multiple-libraries argument]) AC_ARG_ENABLE([multiple-libraries], - [AC_HELP_STRING([--enable-multiple-libraries],[enable multiple libraries (default=no)])], + [AS_HELP_STRING([--enable-multiple-libraries],[enable multiple libraries (default=no)])], [enable_mlibs=$enableval], [enable_mlibs="no"]) AC_MSG_RESULT([$enable_mlibs]) @@ -279,7 +270,7 @@ AM_CONDITIONAL([USING_MULTIPLELIBS], [test "$enable_mlibs" = "yes"]) # Check if tessdata-prefix is disabled AC_MSG_CHECKING([whether to use tessdata-prefix]) AC_ARG_ENABLE([tessdata-prefix], - [AC_HELP_STRING([--disable-tessdata-prefix], + [AS_HELP_STRING([--disable-tessdata-prefix], [don't set TESSDATA-PREFIX during compile])], [tessdata_prefix="no"], [tessdata_prefix="yes"]) AC_MSG_RESULT([$tessdata_prefix]) @@ -288,7 +279,7 @@ AM_CONDITIONAL([NO_TESSDATA_PREFIX], [test "$tessdata_prefix" = "no"]) # Check whether enable debuging AC_MSG_CHECKING([whether to enable debugging]) AC_ARG_ENABLE([debug], - [AC_HELP_STRING([--enable-debug], + [AS_HELP_STRING([--enable-debug], [turn on debugging (default=no)])], [debug=$enableval], [debug="no"]) @@ -447,16 +438,16 @@ else AC_MSG_ERROR([leptonica not found]) fi -AC_MSG_CHECKING([leptonica headers version >= 1.71]) +AC_MSG_CHECKING([leptonica headers version >= 1.74]) AC_PREPROC_IFELSE( [AC_LANG_PROGRAM([#include "allheaders.h"], -[#if (LIBLEPT_MAJOR_VERSION >= 1) && (LIBLEPT_MINOR_VERSION >= 71) +[#if (LIBLEPT_MAJOR_VERSION >= 1) && (LIBLEPT_MINOR_VERSION >= 74) int i = 0; #else #error You need to upgrade your leptonica library! #endif])], [AC_MSG_RESULT(yes)], - [AC_MSG_FAILURE([leptonica 1.71 or higher is required])]) + [AC_MSG_FAILURE([leptonica 1.74 or higher is required])]) AM_CONDITIONAL([ENABLE_TRAINING], true) diff --git a/cppan.yml b/cppan.yml index 8065da79e4..ef38d520c1 100644 --- a/cppan.yml +++ b/cppan.yml @@ -1,124 +1,241 @@ local_settings: - cppan_dir: cppan - -files: - - api/.*\.cpp - - ccmain/.*\.cpp - - ccstruct/.*\.cpp - - ccutil/.*\.cpp - - classify/.*\.cpp - - cube/.*\.cpp - - cutil/.*\.cpp - - dict/.*\.cpp - - neural_networks/runtime/.*\.cpp - - opencl/.*\.cpp - - textord/.*\.cpp - - viewer/.*\.cpp - - wordrec/.*\.cpp - - - api/.*\.h - - ccmain/.*\.h - - ccstruct/.*\.h - - ccutil/.*\.h - - classify/.*\.h - - cube/.*\.h - - cutil/.*\.h - - dict/.*\.h - - neural_networks/runtime/.*\.h - - opencl/.*\.h - - textord/.*\.h - - viewer/.*\.h - - wordrec/.*\.h - - - vs2010/port/.* - -include_directories: - private: - - classify - - cube - - cutil - - dict - - neural_networks/runtime - - opencl - - textord - - vs2010/port - - viewer - - wordrec - public: - - api - - ccmain - - ccstruct - - ccutil - -check_function_exists: - - getline - -check_symbol_exists: - snprintf: stdio.h - -check_include_exists: - - dlfcn.h - - inttypes.h - - limits.h - - malloc.h - - memory.h - - stdbool.h - - stdint.h - - stdlib.h - - strings.h - - string.h - - sys/ipc.h - - sys/shm.h - - sys/stat.h - - sys/types.h - - sys/wait.h - - tiffio.h - - unistd.h - - cairo/cairo-version.h - - CL/cl.h - - OpenCL/cl.h - - pango-1.0/pango/pango-features.h - - unicode/uchar.h - -check_type_size: - - long long int - - off_t - - mbstate_t - - wchar_t - - _Bool - -pre_sources: | - # dummy config file - if (NOT EXISTS ${CMAKE_CURRENT_BINARY_DIR}/config_auto.h) - file(WRITE ${CMAKE_CURRENT_BINARY_DIR}/config_auto.h) - endif() - -post_sources: | - if (NOT WIN32) - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/gettimeofday.cpp") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/gettimeofday.h") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/mathfix.h") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/strcasestr.cpp") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/strcasestr.h") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/strtok_r.cpp") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/strtok_r.h") - list(REMOVE_ITEM src "${CMAKE_CURRENT_SOURCE_DIR}/vs2010/port/vcsversion.h") - endif() - -options: - any: - definitions: - public: - - HAVE_CONFIG_H - - _SILENCE_STDEXT_HASH_DEPRECATION_WARNINGS=1 - - USE_STD_NAMESPACE=1 - - WINDLLNAME="tesseract" - shared: - definitions: - public: TESS_EXPORTS - -dependencies: - private: - # tesseract uses leptonica only internally - # and does not expose its interface to users - pvt.cppan.demo.leptonica: master + #use_shared_libs: true + #generator: Visual Studio 14 2015 Win64 + silent: false + #copy_import_libs: true + build: + c_flags: /W0 + cxx_flags: /W0 + + dependencies: + pvt.cppan.demo.danbloomberg.leptonica: 1 + pvt.cppan.demo.unicode.icu.i18n: "*" + +root_project: pvt.cppan.demo.google.tesseract + +common_settings: + c++: 11 + +projects: + libtesseract: + type: lib + export_all_symbols: true + files: + - api/.*\.cpp + - ccmain/.*\.cpp + - ccstruct/.*\.cpp + - ccutil/.*\.cpp + - classify/.*\.cpp + - cube/.*\.cpp + - cutil/.*\.cpp + - dict/.*\.cpp + - neural_networks/runtime/.*\.cpp + - opencl/.*\.cpp + - textord/.*\.cpp + - viewer/.*\.cpp + - wordrec/.*\.cpp + + - api/.*\.h + - ccmain/.*\.h + - ccstruct/.*\.h + - ccutil/.*\.h + - classify/.*\.h + - cube/.*\.h + - cutil/.*\.h + - dict/.*\.h + - neural_networks/runtime/.*\.h + - opencl/.*\.h + - textord/.*\.h + - viewer/.*\.h + - wordrec/.*\.h + + - vs2010/port/.* + + exclude_from_build: + - api/tesseractmain.cpp + - viewer/svpaint.cpp + + include_directories: + public: + #private: + - classify + - cube + - cutil + - dict + - neural_networks/runtime + - opencl + - textord + - vs2010/port + - viewer + - wordrec + #public: + - api + - ccmain + - ccstruct + - ccutil + + check_function_exists: + - getline + + check_symbol_exists: + snprintf: stdio.h + + check_include_exists: + - dlfcn.h + - inttypes.h + - limits.h + - malloc.h + - memory.h + - stdbool.h + - stdint.h + - stdlib.h + - strings.h + - string.h + - sys/ipc.h + - sys/shm.h + - sys/stat.h + - sys/types.h + - sys/wait.h + - tiffio.h + - unistd.h + - cairo/cairo-version.h + - CL/cl.h + - OpenCL/cl.h + - pango-1.0/pango/pango-features.h + - unicode/uchar.h + + check_type_size: + - long long int + - off_t + - mbstate_t + - wchar_t + - _Bool + + + post_sources: | + file_write_once(${BDIR}/config_auto.h "") + if (NOT WIN32) + remove_src_dir(vs2010/port/*) + endif() + + options: + any: + definitions: + public: + - HAVE_CONFIG_H + - _SILENCE_STDEXT_HASH_DEPRECATION_WARNINGS=1 + - USE_STD_NAMESPACE=1 + - WINDLLNAME="tesseract" + shared: + definitions: + private: + - TESS_EXPORTS + interface: + - TESS_IMPORTS + + dependencies: + pvt.cppan.demo.danbloomberg.leptonica: 1.74 + + tesseract: + files: api/tesseractmain.cpp + dependencies: + - libtesseract + + tessopt: + type: lib + static_only: true + files: training/tessopt.* + include_directories: training + dependencies: libtesseract + + common_training: + type: lib + static_only: true + files: + - training/commandlineflags.cpp + - training/commandlineflags.h + - training/commontraining.cpp + - training/commontraining.h + include_directories: training + dependencies: + - tessopt + + ambiguous_words: + files: training/ambiguous_words.cpp + dependencies: libtesseract + + classifier_tester: + files: training/classifier_tester.cpp + dependencies: common_training + + combine_tessdata: + files: training/combine_tessdata.cpp + dependencies: libtesseract + + cntraining: + files: training/cntraining.cpp + dependencies: common_training + + dawg2wordlist: + files: training/dawg2wordlist.cpp + dependencies: libtesseract + + mftraining: + files: + - training/mftraining.cpp + - training/mergenf.* + dependencies: common_training + + shapeclustering: + files: training/shapeclustering.cpp + dependencies: common_training + + unicharset_extractor: + files: training/unicharset_extractor.cpp + dependencies: tessopt + + wordlist2dawg: + files: training/wordlist2dawg.cpp + dependencies: libtesseract + + unicharset_training: + type: lib + static_only: true + files: + - training/fileio.* + - training/icuerrorcode.h + - training/lstmtester.* + - training/normstrngs.* + - training/unicharset_training_utils.* + include_directories: training + dependencies: + - common_training + - pvt.cppan.demo.unicode.icu.i18n + + set_unicharset_properties: + files: training/set_unicharset_properties.cpp + dependencies: unicharset_training + + text2image: + files: + - training/text2image.cpp + - training/boxchar.cpp + - training/boxchar.h + - training/degradeimage.cpp + - training/degradeimage.h + - training/ligature_table.cpp + - training/ligature_table.h + - training/normstrngs.cpp + - training/normstrngs.h + - training/pango_font_info.cpp + - training/pango_font_info.h + - training/stringrenderer.cpp + - training/stringrenderer.h + - training/tlog.cpp + - training/tlog.h + - training/util.h + - training/icuerrorcode.h + + dependencies: + - unicharset_training + - pvt.cppan.demo.gnome.pango.pangocairo: 1 diff --git a/cube/beam_search.cpp b/cube/beam_search.cpp index fd17a1d59f..37fc35c566 100644 --- a/cube/beam_search.cpp +++ b/cube/beam_search.cpp @@ -36,8 +36,7 @@ BeamSearch::BeamSearch(CubeRecoContext *cntxt, bool word_mode) { void BeamSearch::Cleanup() { if (col_ != NULL) { for (int col = 0; col < col_cnt_; col++) { - if (col_[col]) - delete col_[col]; + delete col_[col]; } delete []col_; } @@ -124,11 +123,6 @@ WordAltList * BeamSearch::Search(SearchObject *srch_obj, LangModel *lang_mod) { // alloc memory for columns col_ = new SearchColumn *[col_cnt_]; - if (!col_) { - fprintf(stderr, "Cube ERROR (BeamSearch::Search): could not construct " - "SearchColumn array\n"); - return NULL; - } memset(col_, 0, col_cnt_ * sizeof(*col_)); // for all possible segments @@ -136,11 +130,6 @@ WordAltList * BeamSearch::Search(SearchObject *srch_obj, LangModel *lang_mod) { // create a search column col_[end_seg - 1] = new SearchColumn(end_seg - 1, cntxt_->Params()->BeamWidth()); - if (!col_[end_seg - 1]) { - fprintf(stderr, "Cube ERROR (BeamSearch::Search): could not construct " - "SearchColumn for column %d\n", end_seg - 1); - return NULL; - } // for all possible start segments int init_seg = MAX(0, end_seg - cntxt_->Params()->MaxSegPerChar()); @@ -356,8 +345,7 @@ CharSamp **BeamSearch::BackTrack(SearchObject *srch_obj, SearchNode *srch_node, return NULL; if (str32) { - if (*str32) - delete [](*str32); // clear existing value + delete [](*str32); // clear existing value *str32 = srch_node->PathString(); if (!*str32) return NULL; @@ -404,11 +392,6 @@ CharSamp **BeamSearch::SplitByNode(SearchObject *srch_obj, // Allocate memory for CharSamp array. CharSamp **chars = new CharSamp *[*char_cnt]; - if (!chars) { - if (char_boxes) - boxaDestroy(char_boxes); - return NULL; - } int ch_idx = *char_cnt - 1; int seg_pt_cnt = srch_obj->SegPtCnt(); diff --git a/cube/beam_search.h b/cube/beam_search.h index cd8fc0110d..31659a7fef 100644 --- a/cube/beam_search.h +++ b/cube/beam_search.h @@ -80,7 +80,7 @@ class BeamSearch { // best-cost path before the alternates list is sorted. inline int BestPresortedNodeIndex() const { return best_presorted_node_idx_; - }; + } private: // Maximum reasonable segmentation point count diff --git a/cube/bmp_8.cpp b/cube/bmp_8.cpp index f7b6e0a1b8..936d344e40 100644 --- a/cube/bmp_8.cpp +++ b/cube/bmp_8.cpp @@ -48,18 +48,14 @@ Bmp8::~Bmp8() { // free buffer void Bmp8::FreeBmpBuffer(unsigned char **buff) { if (buff != NULL) { - if (buff[0] != NULL) { - delete []buff[0]; - } + delete []buff[0]; delete []buff; } } void Bmp8::FreeBmpBuffer(unsigned int **buff) { if (buff != NULL) { - if (buff[0] != NULL) { - delete []buff[0]; - } + delete []buff[0]; delete []buff; } } @@ -76,17 +72,10 @@ unsigned char **Bmp8::CreateBmpBuffer(unsigned char init_val) { stride_ = ((wid_ % 4) == 0) ? wid_ : (4 * (1 + (wid_ / 4))); buff = (unsigned char **) new unsigned char *[hgt_ * sizeof(*buff)]; - if (!buff) { - delete []buff; - return NULL; - } // alloc and init memory for buffer and line buffer buff[0] = (unsigned char *) new unsigned char[stride_ * hgt_ * sizeof(*buff[0])]; - if (!buff[0]) { - return NULL; - } memset(buff[0], init_val, stride_ * hgt_ * sizeof(*buff[0])); @@ -104,16 +93,9 @@ unsigned int ** Bmp8::CreateBmpBuffer(int wid, int hgt, // compute stride (align on 4 byte boundries) buff = (unsigned int **) new unsigned int *[hgt * sizeof(*buff)]; - if (!buff) { - delete []buff; - return NULL; - } // alloc and init memory for buffer and line buffer buff[0] = (unsigned int *) new unsigned int[wid * hgt * sizeof(*buff[0])]; - if (!buff[0]) { - return NULL; - } memset(buff[0], init_val, wid * hgt * sizeof(*buff[0])); @@ -176,9 +158,6 @@ bool Bmp8::LoadFromCharDumpFile(CachedFile *fp) { // alloc memory & read the 3 channel buffer buff = new unsigned char[buf_size]; - if (buff == NULL) { - return false; - } if (fp->Read(buff, buf_size) != buf_size) { delete []buff; @@ -217,9 +196,6 @@ bool Bmp8::LoadFromCharDumpFile(CachedFile *fp) { Bmp8 * Bmp8::FromCharDumpFile(CachedFile *fp) { // create a Bmp8 object Bmp8 *bmp_obj = new Bmp8(0, 0); - if (bmp_obj == NULL) { - return NULL; - } if (bmp_obj->LoadFromCharDumpFile(fp) == false) { delete bmp_obj; @@ -271,9 +247,6 @@ bool Bmp8::LoadFromCharDumpFile(FILE *fp) { // alloc memory & read the 3 channel buffer buff = new unsigned char[buf_size]; - if (buff == NULL) { - return false; - } if (fread(buff, 1, buf_size, fp) != buf_size) { delete []buff; @@ -312,9 +285,6 @@ bool Bmp8::LoadFromCharDumpFile(FILE *fp) { Bmp8 * Bmp8::FromCharDumpFile(FILE *fp) { // create a Bmp8 object Bmp8 *bmp_obj = new Bmp8(0, 0); - if (bmp_obj == NULL) { - return NULL; - } if (bmp_obj->LoadFromCharDumpFile(fp) == false) { delete bmp_obj; @@ -549,9 +519,6 @@ bool Bmp8::SaveBmp2CharDumpFile(FILE *fp) const { // alloc memory & write the 3 channel buffer buff = new unsigned char[buf_size]; - if (buff == NULL) { - return false; - } // copy the data for (y = 0, pix = 0; y < hgt_; y++) { @@ -703,7 +670,7 @@ ConComp ** Bmp8::FindConComps(int *concomp_cnt, int min_size) const { // if there was no foreground pix, then create a new concomp if (master_concomp == NULL) { master_concomp = new ConComp(); - if (master_concomp == NULL || master_concomp->Add(x, y) == false) { + if (master_concomp->Add(x, y) == false) { fprintf(stderr, "Cube ERROR (Bmp8::FindConComps): could not " "allocate or add a connected component\n"); FreeBmpBuffer(out_bmp_array); @@ -715,13 +682,6 @@ ConComp ** Bmp8::FindConComps(int *concomp_cnt, int min_size) const { if ((alloc_concomp_cnt % kConCompAllocChunk) == 0) { ConComp **temp_con_comp = new ConComp *[alloc_concomp_cnt + kConCompAllocChunk]; - if (temp_con_comp == NULL) { - fprintf(stderr, "Cube ERROR (Bmp8::FindConComps): could not " - "extend array of connected components\n"); - FreeBmpBuffer(out_bmp_array); - delete []concomp_array; - return NULL; - } if (alloc_concomp_cnt > 0) { memcpy(temp_con_comp, concomp_array, @@ -778,9 +738,6 @@ bool Bmp8::ComputeTanTable() { // alloc memory for tan table delete []tan_table_; tan_table_ = new float[kDeslantAngleCount]; - if (tan_table_ == NULL) { - return false; - } for (ang_idx = 0, ang_val = kMinDeslantAngle; ang_idx < kDeslantAngleCount; ang_idx++) { @@ -825,10 +782,6 @@ bool Bmp8::Deslant() { int **angle_hist = new int*[kDeslantAngleCount]; for (ang_idx = 0; ang_idx < kDeslantAngleCount; ang_idx++) { angle_hist[ang_idx] = new int[des_wid]; - if (angle_hist[ang_idx] == NULL) { - delete[] angle_hist; - return false; - } memset(angle_hist[ang_idx], 0, des_wid * sizeof(*angle_hist[ang_idx])); } @@ -1010,10 +963,6 @@ bool Bmp8::HorizontalDeslant(double *deslant_angle) { int **angle_hist = new int*[kDeslantAngleCount]; for (ang_idx = 0; ang_idx < kDeslantAngleCount; ang_idx++) { angle_hist[ang_idx] = new int[des_hgt]; - if (angle_hist[ang_idx] == NULL) { - delete[] angle_hist; - return false; - } memset(angle_hist[ang_idx], 0, des_hgt * sizeof(*angle_hist[ang_idx])); } @@ -1122,9 +1071,6 @@ float Bmp8::MeanHorizontalHistogramEntropy() const { int *Bmp8::HorizontalHistogram() const { int *hist = new int[hgt_]; - if (hist == NULL) { - return NULL; - } // compute histograms for (int y = 0; y < hgt_; y++) { diff --git a/cube/cached_file.cpp b/cube/cached_file.cpp index a9a5b2e0b1..84f6f31d66 100644 --- a/cube/cached_file.cpp +++ b/cube/cached_file.cpp @@ -68,9 +68,6 @@ bool CachedFile::Open() { rewind(fp_); // alloc memory for buffer buff_ = new unsigned char[kCacheSize]; - if (buff_ == NULL) { - return false; - } // init counters buff_size_ = 0; buff_pos_ = 0; diff --git a/cube/char_altlist.cpp b/cube/char_altlist.cpp index c0e7776ef2..d4d4b9c543 100644 --- a/cube/char_altlist.cpp +++ b/cube/char_altlist.cpp @@ -56,10 +56,6 @@ bool CharAltList::Insert(int class_id, int cost, void *tag) { alt_cost_ = new int[max_alt_]; alt_tag_ = new void *[max_alt_]; - if (class_id_alt_ == NULL || alt_cost_ == NULL || alt_tag_ == NULL) { - return false; - } - memset(alt_tag_, 0, max_alt_ * sizeof(*alt_tag_)); } @@ -67,9 +63,6 @@ bool CharAltList::Insert(int class_id, int cost, void *tag) { int class_cnt = char_set_->ClassCount(); class_id_cost_ = new int[class_cnt]; - if (class_id_cost_ == NULL) { - return false; - } for (int ich = 0; ich < class_cnt; ich++) { class_id_cost_[ich] = WORST_COST; diff --git a/cube/char_bigrams.cpp b/cube/char_bigrams.cpp index b36b1f6cda..b005c1d2d4 100644 --- a/cube/char_bigrams.cpp +++ b/cube/char_bigrams.cpp @@ -61,11 +61,6 @@ CharBigrams *CharBigrams::Create(const string &data_file_path, // construct a new object CharBigrams *char_bigrams_obj = new CharBigrams(); - if (char_bigrams_obj == NULL) { - fprintf(stderr, "Cube ERROR (CharBigrams::Create): could not create " - "character bigrams object.\n"); - return NULL; - } CharBigramTable *table = &char_bigrams_obj->bigram_table_; table->total_cnt = 0; @@ -90,11 +85,6 @@ CharBigrams *CharBigrams::Create(const string &data_file_path, // expand the bigram table if (ch1 > table->max_char) { CharBigram *char_bigram = new CharBigram[ch1 + 1]; - if (char_bigram == NULL) { - fprintf(stderr, "Cube ERROR (CharBigrams::Create): error allocating " - "additional memory for character bigram table.\n"); - return NULL; - } if (table->char_bigram != NULL && table->max_char >= 0) { memcpy(char_bigram, table->char_bigram, @@ -115,12 +105,6 @@ CharBigrams *CharBigrams::Create(const string &data_file_path, if (ch2 > table->char_bigram[ch1].max_char) { Bigram *bigram = new Bigram[ch2 + 1]; - if (bigram == NULL) { - fprintf(stderr, "Cube ERROR (CharBigrams::Create): error allocating " - "memory for bigram.\n"); - delete char_bigrams_obj; - return NULL; - } if (table->char_bigram[ch1].bigram != NULL && table->char_bigram[ch1].max_char >= 0) { @@ -179,14 +163,14 @@ int CharBigrams::Cost(const char_32 *char_32_ptr, CharSet *char_set) const { if (lower_32 && lower_32[0] != 0) { int cost_lower = MeanCostWithSpaces(lower_32); cost = MIN(cost, cost_lower); - delete [] lower_32; } + delete [] lower_32; char_32 *upper_32 = CubeUtils::ToUpper(char_32_ptr, char_set); if (upper_32 && upper_32[0] != 0) { int cost_upper = MeanCostWithSpaces(upper_32); cost = MIN(cost, cost_upper); - delete [] upper_32; } + delete [] upper_32; } return cost; } diff --git a/cube/char_samp.cpp b/cube/char_samp.cpp index c3493fa13c..f55735be0e 100644 --- a/cube/char_samp.cpp +++ b/cube/char_samp.cpp @@ -106,12 +106,10 @@ CharSamp *CharSamp::FromCharDumpFile(CachedFile *fp) { // the label is not null terminated in the file if (val32 > 0 && val32 < MAX_UINT32) { label32 = new char_32[val32 + 1]; - if (label32 == NULL) { - return NULL; - } // read label if (fp->Read(label32, val32 * sizeof(*label32)) != (val32 * sizeof(*label32))) { + delete [] label32; return NULL; } // null terminate @@ -121,35 +119,40 @@ CharSamp *CharSamp::FromCharDumpFile(CachedFile *fp) { } // read coordinates if (fp->Read(&page, sizeof(page)) != sizeof(page)) { + delete [] label32; return NULL; } if (fp->Read(&left, sizeof(left)) != sizeof(left)) { + delete [] label32; return NULL; } if (fp->Read(&top, sizeof(top)) != sizeof(top)) { + delete [] label32; return NULL; } if (fp->Read(&first_char, sizeof(first_char)) != sizeof(first_char)) { + delete [] label32; return NULL; } if (fp->Read(&last_char, sizeof(last_char)) != sizeof(last_char)) { + delete [] label32; return NULL; } if (fp->Read(&norm_top, sizeof(norm_top)) != sizeof(norm_top)) { + delete [] label32; return NULL; } if (fp->Read(&norm_bottom, sizeof(norm_bottom)) != sizeof(norm_bottom)) { + delete [] label32; return NULL; } if (fp->Read(&norm_aspect_ratio, sizeof(norm_aspect_ratio)) != sizeof(norm_aspect_ratio)) { + delete [] label32; return NULL; } // create the object CharSamp *char_samp = new CharSamp(); - if (char_samp == NULL) { - return NULL; - } // init char_samp->label32_ = label32; char_samp->page_ = page; @@ -195,9 +198,6 @@ CharSamp *CharSamp::FromCharDumpFile(FILE *fp) { // the label is not null terminated in the file if (val32 > 0 && val32 < MAX_UINT32) { label32 = new char_32[val32 + 1]; - if (label32 == NULL) { - return NULL; - } // read label if (fread(label32, 1, val32 * sizeof(*label32), fp) != (val32 * sizeof(*label32))) { @@ -224,10 +224,6 @@ CharSamp *CharSamp::FromCharDumpFile(FILE *fp) { } // create the object CharSamp *char_samp = new CharSamp(); - if (char_samp == NULL) { - delete [] label32; - return NULL; - } // init char_samp->label32_ = label32; char_samp->page_ = page; @@ -250,9 +246,6 @@ CharSamp *CharSamp::FromCharDumpFile(FILE *fp) { // specified width and height CharSamp *CharSamp::Scale(int wid, int hgt, bool isotropic) { CharSamp *scaled_samp = new CharSamp(wid, hgt); - if (scaled_samp == NULL) { - return NULL; - } if (scaled_samp->ScaleFrom(this, isotropic) == false) { delete scaled_samp; return NULL; @@ -274,9 +267,6 @@ CharSamp *CharSamp::FromRawData(int left, int top, int wid, int hgt, unsigned char *data) { // create the object CharSamp *char_samp = new CharSamp(left, top, wid, hgt); - if (char_samp == NULL) { - return NULL; - } if (char_samp->LoadFromRawData(data) == false) { delete char_samp; return NULL; @@ -421,14 +411,6 @@ ConComp **CharSamp::Segment(int *segment_cnt, bool right_2_left, if ((seg_cnt % kConCompAllocChunk) == 0) { ConComp **temp_segm_array = new ConComp *[seg_cnt + kConCompAllocChunk]; - if (temp_segm_array == NULL) { - fprintf(stderr, "Cube ERROR (CharSamp::Segment): could not " - "allocate additional connected components\n"); - delete []concomp_seg_array; - delete []concomp_array; - delete []seg_array; - return NULL; - } if (seg_cnt > 0) { memcpy(temp_segm_array, seg_array, seg_cnt * sizeof(*seg_array)); delete []seg_array; @@ -486,8 +468,6 @@ CharSamp *CharSamp::FromConComps(ConComp **concomp_array, int strt_concomp, bool *id_exist = new bool[id_cnt]; bool *left_most_exist = new bool[id_cnt]; bool *right_most_exist = new bool[id_cnt]; - if (!id_exist || !left_most_exist || !right_most_exist) - return NULL; memset(id_exist, 0, id_cnt * sizeof(*id_exist)); memset(left_most_exist, 0, id_cnt * sizeof(*left_most_exist)); memset(right_most_exist, 0, id_cnt * sizeof(*right_most_exist)); @@ -544,9 +524,6 @@ CharSamp *CharSamp::FromConComps(ConComp **concomp_array, int strt_concomp, (*right_most) = (unq_right_most >= unq_ids); // create the char sample object CharSamp *samp = new CharSamp(left, top, right - left + 1, bottom - top + 1); - if (!samp) { - return NULL; - } // set the foreground pixels for (concomp = strt_concomp; concomp < end_concomp; concomp++) { @@ -594,9 +571,6 @@ CharSamp *CharSamp::FromCharDumpFile(unsigned char **raw_data_ptr) { // the label is not null terminated in the file if (val32 > 0 && val32 < MAX_UINT32) { label32 = new char_32[val32 + 1]; - if (label32 == NULL) { - return NULL; - } // read label memcpy(label32, raw_data, val32 * sizeof(*label32)); raw_data += (val32 * sizeof(*label32)); @@ -608,9 +582,6 @@ CharSamp *CharSamp::FromCharDumpFile(unsigned char **raw_data_ptr) { // create the object CharSamp *char_samp = new CharSamp(); - if (char_samp == NULL) { - return NULL; - } // read coordinates char_samp->label32_ = label32; diff --git a/cube/char_samp.h b/cube/char_samp.h index a3c3063bd6..827e1c37c8 100644 --- a/cube/char_samp.h +++ b/cube/char_samp.h @@ -66,20 +66,14 @@ class CharSamp : public Bmp8 { void SetTop(unsigned short top) { top_ = top; } void SetPage(unsigned short page) { page_ = page; } void SetLabel(char_32 label) { - if (label32_ != NULL) { - delete []label32_; - } + delete []label32_; label32_ = new char_32[2]; - if (label32_ != NULL) { - label32_[0] = label; - label32_[1] = 0; - } + label32_[0] = label; + label32_[1] = 0; } void SetLabel(const char_32 *label32) { - if (label32_ != NULL) { - delete []label32_; - label32_ = NULL; - } + delete []label32_; + label32_ = NULL; if (label32 != NULL) { // remove any byte order marks if any if (label32[0] == 0xfeff) { @@ -87,10 +81,8 @@ class CharSamp : public Bmp8 { } int len = LabelLen(label32); label32_ = new char_32[len + 1]; - if (label32_ != NULL) { - memcpy(label32_, label32, len * sizeof(*label32)); - label32_[len] = 0; - } + memcpy(label32_, label32, len * sizeof(*label32)); + label32_[len] = 0; } } void SetLabel(string str); diff --git a/cube/char_samp_set.cpp b/cube/char_samp_set.cpp index 2a495095ef..9af7581187 100644 --- a/cube/char_samp_set.cpp +++ b/cube/char_samp_set.cpp @@ -40,9 +40,7 @@ void CharSampSet::Cleanup() { // only free samples if owned by class if (own_samples_ == true) { for (int samp_idx = 0; samp_idx < cnt_; samp_idx++) { - if (samp_buff_[samp_idx] != NULL) { - delete samp_buff_[samp_idx]; - } + delete samp_buff_[samp_idx]; } } delete []samp_buff_; @@ -57,9 +55,6 @@ bool CharSampSet::Add(CharSamp *char_samp) { // create an extended buffer CharSamp **new_samp_buff = reinterpret_cast(new CharSamp *[cnt_ + SAMP_ALLOC_BLOCK]); - if (new_samp_buff == NULL) { - return false; - } // copy old contents if (cnt_ > 0) { memcpy(new_samp_buff, samp_buff_, cnt_ * sizeof(*samp_buff_)); @@ -109,10 +104,6 @@ CharSampSet * CharSampSet::FromCharDumpFile(string file_name) { } // create an object CharSampSet *samp_set = new CharSampSet(); - if (samp_set == NULL) { - fclose(fp); - return NULL; - } if (samp_set->LoadCharSamples(fp) == false) { delete samp_set; samp_set = NULL; @@ -148,9 +139,6 @@ bool CharSampSet::EnumSamples(string file_name, CharSampEnum *enum_obj) { i64_pos; // open the file fp_in = new CachedFile(file_name); - if (fp_in == NULL) { - return false; - } i64_size = fp_in->Size(); if (i64_size < 1) { return false; diff --git a/cube/char_set.cpp b/cube/char_set.cpp index 1414d640f4..b2ec5f544a 100644 --- a/cube/char_set.cpp +++ b/cube/char_set.cpp @@ -54,9 +54,6 @@ CharSet::~CharSet() { CharSet *CharSet::Create(TessdataManager *tessdata_manager, UNICHARSET *tess_unicharset) { CharSet *char_set = new CharSet(); - if (char_set == NULL) { - return NULL; - } // First look for Cube's unicharset; if not there, use tesseract's bool cube_unicharset_exists; @@ -119,19 +116,9 @@ bool CharSet::LoadSupportedCharList(FILE *fp, UNICHARSET *tess_unicharset) { } // memory for class strings class_strings_ = new string_32*[class_cnt_]; - if (class_strings_ == NULL) { - fprintf(stderr, "Cube ERROR (CharSet::InitMemory): could not " - "allocate memory for class strings.\n"); - return false; - } // memory for unicharset map if (tess_unicharset) { unicharset_map_ = new int[class_cnt_]; - if (unicharset_map_ == NULL) { - fprintf(stderr, "Cube ERROR (CharSet::InitMemory): could not " - "allocate memory for unicharset map.\n"); - return false; - } } // Read in character strings and add to hash table @@ -154,11 +141,6 @@ bool CharSet::LoadSupportedCharList(FILE *fp, UNICHARSET *tess_unicharset) { } CubeUtils::UTF8ToUTF32(str_line, &str32); class_strings_[class_id] = new string_32(str32); - if (class_strings_[class_id] == NULL) { - fprintf(stderr, "Cube ERROR (CharSet::ReadAndHashStrings): could not " - "allocate memory for class string with class_id=%d.\n", class_id); - return false; - } // Add to hash-table int hash_val = Hash(reinterpret_cast(str32.c_str())); diff --git a/cube/classifier_base.h b/cube/classifier_base.h index 8c2b1bbf9a..98f2f79af4 100644 --- a/cube/classifier_base.h +++ b/cube/classifier_base.h @@ -49,21 +49,15 @@ class CharClassifier { virtual ~CharClassifier() { if (fold_sets_ != NULL) { for (int fold_set = 0; fold_set < fold_set_cnt_; fold_set++) { - if (fold_sets_[fold_set] != NULL) { - delete []fold_sets_[fold_set]; - } + delete []fold_sets_[fold_set]; } delete []fold_sets_; fold_sets_ = NULL; } - if (fold_set_len_ != NULL) { - delete []fold_set_len_; - fold_set_len_ = NULL; - } - if (feat_extract_ != NULL) { - delete feat_extract_; - feat_extract_ = NULL; - } + delete []fold_set_len_; + fold_set_len_ = NULL; + delete feat_extract_; + feat_extract_ = NULL; } // pure virtual functions that need to be implemented by any inheriting class diff --git a/cube/classifier_factory.cpp b/cube/classifier_factory.cpp index a22f0d4ea8..04df263911 100644 --- a/cube/classifier_factory.cpp +++ b/cube/classifier_factory.cpp @@ -56,12 +56,6 @@ CharClassifier *CharClassifierFactory::Create(const string &data_file_path, return NULL; } - if (feat_extract == NULL) { - fprintf(stderr, "Cube ERROR (CharClassifierFactory::Create): unable " - "to instantiate feature extraction object.\n"); - return NULL; - } - // create the classifier object CharClassifier *classifier_obj; switch (params->TypeClassifier()) { @@ -79,12 +73,6 @@ CharClassifier *CharClassifierFactory::Create(const string &data_file_path, return NULL; } - if (classifier_obj == NULL) { - fprintf(stderr, "Cube ERROR (CharClassifierFactory::Create): error " - "allocating memory for character classifier object.\n"); - return NULL; - } - // Init the classifier if (!classifier_obj->Init(data_file_path, lang, lang_mod)) { delete classifier_obj; diff --git a/cube/con_comp.cpp b/cube/con_comp.cpp index 53b1a73b43..a0a926dbd8 100644 --- a/cube/con_comp.cpp +++ b/cube/con_comp.cpp @@ -52,9 +52,6 @@ ConComp::~ConComp() { // adds a pt to the conn comp and updates its boundaries bool ConComp::Add(int x, int y) { ConCompPt *pt_ptr = new ConCompPt(x, y); - if (pt_ptr == NULL) { - return false; - } if (head_ == NULL) { left_ = x; @@ -114,9 +111,6 @@ int *ConComp::CreateHistogram(int max_hist_wnd) { // alloc memo for histogram int *hist_array = new int[wid]; - if (hist_array == NULL) { - return NULL; - } memset(hist_array, 0, wid * sizeof(*hist_array)); @@ -148,9 +142,6 @@ int *ConComp::SegmentHistogram(int *hist_array, int *seg_pt_cnt) { hgt = bottom_ - top_ + 1; int *x_seg_pt = new int[wid]; - if (x_seg_pt == NULL) { - return NULL; - } int seg_pt_wnd = static_cast(hgt * SEG_PT_WND_RATIO); @@ -216,18 +207,9 @@ ConComp **ConComp::Segment(int max_hist_wnd, int *concomp_cnt) { // create concomp array ConComp **concomp_array = new ConComp *[seg_pt_cnt + 1]; - if (concomp_array == NULL) { - delete []x_seg_pt; - return NULL; - } for (int concomp = 0; concomp <= seg_pt_cnt; concomp++) { concomp_array[concomp] = new ConComp(); - if (concomp_array[concomp] == NULL) { - delete []x_seg_pt; - delete []concomp_array; - return NULL; - } // split concomps inherit the ID this concomp concomp_array[concomp]->SetID(id_); diff --git a/cube/conv_net_classifier.cpp b/cube/conv_net_classifier.cpp index ac33cd33b1..e4846ac381 100644 --- a/cube/conv_net_classifier.cpp +++ b/cube/conv_net_classifier.cpp @@ -147,18 +147,7 @@ bool ConvNetCharClassifier::RunNets(CharSamp *char_samp) { // allocate i/p and o/p buffers if needed if (net_input_ == NULL) { net_input_ = new float[feat_cnt]; - if (net_input_ == NULL) { - fprintf(stderr, "Cube ERROR (ConvNetCharClassifier::RunNets): " - "unable to allocate memory for input nodes\n"); - return false; - } - net_output_ = new float[class_cnt]; - if (net_output_ == NULL) { - fprintf(stderr, "Cube ERROR (ConvNetCharClassifier::RunNets): " - "unable to allocate memory for output nodes\n"); - return false; - } } // compute input features @@ -205,11 +194,6 @@ CharAltList *ConvNetCharClassifier::Classify(CharSamp *char_samp) { // create an altlist CharAltList *alt_list = new CharAltList(char_set_, class_cnt); - if (alt_list == NULL) { - fprintf(stderr, "Cube WARNING (ConvNetCharClassifier::Classify): " - "returning emtpy CharAltList\n"); - return NULL; - } for (int out = 1; out < class_cnt; out++) { int cost = CubeUtils::Prob2Cost(net_output_[out]); @@ -261,14 +245,7 @@ bool ConvNetCharClassifier::LoadFoldingSets(const string &data_file_path, fold_set_cnt_ = str_vec.size(); fold_sets_ = new int *[fold_set_cnt_]; - if (fold_sets_ == NULL) { - return false; - } fold_set_len_ = new int[fold_set_cnt_]; - if (fold_set_len_ == NULL) { - fold_set_cnt_ = 0; - return false; - } for (int fold_set = 0; fold_set < fold_set_cnt_; fold_set++) { reinterpret_cast(lang_mod)->RemoveInvalidCharacters( @@ -287,12 +264,6 @@ bool ConvNetCharClassifier::LoadFoldingSets(const string &data_file_path, CubeUtils::UTF8ToUTF32(str_vec[fold_set].c_str(), &str32); fold_set_len_[fold_set] = str32.length(); fold_sets_[fold_set] = new int[fold_set_len_[fold_set]]; - if (fold_sets_[fold_set] == NULL) { - fprintf(stderr, "Cube ERROR (ConvNetCharClassifier::LoadFoldingSets): " - "could not allocate folding set\n"); - fold_set_cnt_ = fold_set; - return false; - } for (int ch = 0; ch < fold_set_len_[fold_set]; ch++) { fold_sets_[fold_set][ch] = char_set_->ClassID(str32[ch]); } @@ -375,14 +346,7 @@ bool ConvNetCharClassifier::LoadNets(const string &data_file_path, // allocate i/p and o/p buffers if needed if (net_input_ == NULL) { net_input_ = new float[feat_cnt]; - if (net_input_ == NULL) { - return false; - } - net_output_ = new float[class_cnt]; - if (net_output_ == NULL) { - return false; - } } return true; diff --git a/cube/cube_line_object.cpp b/cube/cube_line_object.cpp index 0325453740..72fd87ff2b 100644 --- a/cube/cube_line_object.cpp +++ b/cube/cube_line_object.cpp @@ -91,68 +91,62 @@ bool CubeLineObject::Process() { if (word_break_threshold > 0) { // over-allocate phrases object buffer phrases_ = new CubeObject *[con_comp_cnt]; - if (phrases_ != NULL) { - // create a phrase if the horizontal distance between two consecutive - // concomps is higher than threshold - int start_con_idx = 0; - int current_phrase_limit = rtl ? con_comps[0]->Left() : - con_comps[0]->Right(); - - for (int con_idx = 1; con_idx <= con_comp_cnt; con_idx++) { - bool create_new_phrase = true; - // if not at the end, compute the distance between two consecutive - // concomps - if (con_idx < con_comp_cnt) { - int dist = 0; - if (cntxt_->ReadingOrder() == tesseract::CubeRecoContext::R2L) { - dist = current_phrase_limit - con_comps[con_idx]->Right(); - } else { - dist = con_comps[con_idx]->Left() - current_phrase_limit; - } - create_new_phrase = (dist > word_break_threshold); + // create a phrase if the horizontal distance between two consecutive + // concomps is higher than threshold + int start_con_idx = 0; + int current_phrase_limit = rtl ? con_comps[0]->Left() : + con_comps[0]->Right(); + + for (int con_idx = 1; con_idx <= con_comp_cnt; con_idx++) { + bool create_new_phrase = true; + // if not at the end, compute the distance between two consecutive + // concomps + if (con_idx < con_comp_cnt) { + int dist = 0; + if (cntxt_->ReadingOrder() == tesseract::CubeRecoContext::R2L) { + dist = current_phrase_limit - con_comps[con_idx]->Right(); + } else { + dist = con_comps[con_idx]->Left() - current_phrase_limit; } + create_new_phrase = (dist > word_break_threshold); + } - // create a new phrase - if (create_new_phrase) { - // create a phrase corresponding to a range on components - bool left_most; - bool right_most; - CharSamp *phrase_char_samp = - CharSamp::FromConComps(con_comps, start_con_idx, - con_idx - start_con_idx, NULL, - &left_most, &right_most, - line_pix_->h); - if (phrase_char_samp == NULL) { - break; - } - phrases_[phrase_cnt_] = new CubeObject(cntxt_, phrase_char_samp); - if (phrases_[phrase_cnt_] == NULL) { - delete phrase_char_samp; - break; - } - // set the ownership of the charsamp to the cube object - phrases_[phrase_cnt_]->SetCharSampOwnership(true); - phrase_cnt_++; - // advance the starting index to the current index - start_con_idx = con_idx; - // set the limit of the newly starting phrase (if any) - if (con_idx < con_comp_cnt) { - current_phrase_limit = rtl ? con_comps[con_idx]->Left() : - con_comps[con_idx]->Right(); - } + // create a new phrase + if (create_new_phrase) { + // create a phrase corresponding to a range on components + bool left_most; + bool right_most; + CharSamp *phrase_char_samp = + CharSamp::FromConComps(con_comps, start_con_idx, + con_idx - start_con_idx, NULL, + &left_most, &right_most, + line_pix_->h); + if (phrase_char_samp == NULL) { + break; + } + phrases_[phrase_cnt_] = new CubeObject(cntxt_, phrase_char_samp); + // set the ownership of the charsamp to the cube object + phrases_[phrase_cnt_]->SetCharSampOwnership(true); + phrase_cnt_++; + // advance the starting index to the current index + start_con_idx = con_idx; + // set the limit of the newly starting phrase (if any) + if (con_idx < con_comp_cnt) { + current_phrase_limit = rtl ? con_comps[con_idx]->Left() : + con_comps[con_idx]->Right(); + } + } else { + // update the limit of the current phrase + if (cntxt_->ReadingOrder() == tesseract::CubeRecoContext::R2L) { + current_phrase_limit = MIN(current_phrase_limit, + con_comps[con_idx]->Left()); } else { - // update the limit of the current phrase - if (cntxt_->ReadingOrder() == tesseract::CubeRecoContext::R2L) { - current_phrase_limit = MIN(current_phrase_limit, - con_comps[con_idx]->Left()); - } else { - current_phrase_limit = MAX(current_phrase_limit, - con_comps[con_idx]->Right()); - } + current_phrase_limit = MAX(current_phrase_limit, + con_comps[con_idx]->Right()); } } - ret_val = true; } + ret_val = true; } // clean-up connected comps diff --git a/cube/cube_line_segmenter.cpp b/cube/cube_line_segmenter.cpp index 278011f090..4b75dca2cd 100644 --- a/cube/cube_line_segmenter.cpp +++ b/cube/cube_line_segmenter.cpp @@ -126,9 +126,6 @@ Pixa *CubeLineSegmenter::CrackLine(Pix *cracked_line_pix, Box *cracked_line_box, int line_cnt) { // create lines pixa array Pixa **lines_pixa = new Pixa*[line_cnt]; - if (lines_pixa == NULL) { - return NULL; - } memset(lines_pixa, 0, line_cnt * sizeof(*lines_pixa)); @@ -620,9 +617,6 @@ bool CubeLineSegmenter::AddLines(Pixa *lines) { // Index the specific pixa using RTL reading order int *CubeLineSegmenter::IndexRTL(Pixa *pixa) { int *pix_index = new int[pixa->n]; - if (pix_index == NULL) { - return NULL; - } for (int pix = 0; pix < pixa->n; pix++) { pix_index[pix] = pix; diff --git a/cube/cube_object.cpp b/cube/cube_object.cpp index c7dec4d5b8..4d95f71ba2 100644 --- a/cube/cube_object.cpp +++ b/cube/cube_object.cpp @@ -54,47 +54,33 @@ void CubeObject::Init() { // Cleanup function void CubeObject::Cleanup() { - if (alt_list_ != NULL) { - delete alt_list_; - alt_list_ = NULL; - } + delete alt_list_; + alt_list_ = NULL; - if (deslanted_alt_list_ != NULL) { - delete deslanted_alt_list_; - deslanted_alt_list_ = NULL; - } + delete deslanted_alt_list_; + deslanted_alt_list_ = NULL; } CubeObject::~CubeObject() { - if (char_samp_ != NULL && own_char_samp_ == true) { + if (own_char_samp_ == true) { delete char_samp_; char_samp_ = NULL; } - if (srch_obj_ != NULL) { - delete srch_obj_; - srch_obj_ = NULL; - } + delete srch_obj_; + srch_obj_ = NULL; - if (deslanted_srch_obj_ != NULL) { - delete deslanted_srch_obj_; - deslanted_srch_obj_ = NULL; - } + delete deslanted_srch_obj_; + deslanted_srch_obj_ = NULL; - if (beam_obj_ != NULL) { - delete beam_obj_; - beam_obj_ = NULL; - } + delete beam_obj_; + beam_obj_ = NULL; - if (deslanted_beam_obj_ != NULL) { - delete deslanted_beam_obj_; - deslanted_beam_obj_ = NULL; - } + delete deslanted_beam_obj_; + deslanted_beam_obj_ = NULL; - if (deslanted_char_samp_ != NULL) { - delete deslanted_char_samp_; - deslanted_char_samp_ = NULL; - } + delete deslanted_char_samp_; + deslanted_char_samp_ = NULL; Cleanup(); } @@ -129,21 +115,11 @@ WordAltList *CubeObject::Recognize(LangModel *lang_mod, bool word_mode) { // create a beam search object if (beam_obj_ == NULL) { beam_obj_ = new BeamSearch(cntxt_, word_mode); - if (beam_obj_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeObject::Recognize): could not construct " - "BeamSearch\n"); - return NULL; - } } // create a cube search object if (srch_obj_ == NULL) { srch_obj_ = new CubeSearchObject(cntxt_, char_samp_); - if (srch_obj_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeObject::Recognize): could not construct " - "CubeSearchObject\n"); - return NULL; - } } // run a beam search against the tesslang model @@ -156,11 +132,6 @@ WordAltList *CubeObject::Recognize(LangModel *lang_mod, bool word_mode) { if (deslanted_beam_obj_ == NULL) { deslanted_beam_obj_ = new BeamSearch(cntxt_); - if (deslanted_beam_obj_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeObject::Recognize): could not " - "construct deslanted BeamSearch\n"); - return NULL; - } } if (deslanted_srch_obj_ == NULL) { @@ -176,11 +147,6 @@ WordAltList *CubeObject::Recognize(LangModel *lang_mod, bool word_mode) { } deslanted_srch_obj_ = new CubeSearchObject(cntxt_, deslanted_char_samp_); - if (deslanted_srch_obj_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeObject::Recognize): could not " - "construct deslanted CubeSearchObject\n"); - return NULL; - } } // run a beam search against the tesslang model @@ -219,9 +185,6 @@ WordAltList *CubeObject::RecognizePhrase(LangModel *lang_mod) { */ int CubeObject::WordCost(const char *str) { WordListLangModel *lang_mod = new WordListLangModel(cntxt_); - if (lang_mod == NULL) { - return WORST_COST; - } if (lang_mod->AddString(str) == false) { delete lang_mod; @@ -256,9 +219,6 @@ CharAltList *CubeObject::RecognizeChar() { bool CubeObject::Normalize() { // create a cube search object CubeSearchObject *srch_obj = new CubeSearchObject(cntxt_, char_samp_); - if (srch_obj == NULL) { - return false; - } // Perform over-segmentation int seg_cnt = srch_obj->SegPtCnt(); // Only perform normalization if segment count is large enough diff --git a/cube/cube_search_object.cpp b/cube/cube_search_object.cpp index 61294f26b6..ad807193d5 100644 --- a/cube/cube_search_object.cpp +++ b/cube/cube_search_object.cpp @@ -127,36 +127,14 @@ bool CubeSearchObject::Init() { // init cache reco_cache_ = new CharAltList **[segment_cnt_]; - if (reco_cache_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeSearchObject::Init): could not " - "allocate CharAltList array\n"); - return false; - } samp_cache_ = new CharSamp **[segment_cnt_]; - if (samp_cache_ == NULL) { - fprintf(stderr, "Cube ERROR (CubeSearchObject::Init): could not " - "allocate CharSamp array\n"); - return false; - } for (int seg = 0; seg < segment_cnt_; seg++) { reco_cache_[seg] = new CharAltList *[segment_cnt_]; - if (reco_cache_[seg] == NULL) { - fprintf(stderr, "Cube ERROR (CubeSearchObject::Init): could not " - "allocate a single segment's CharAltList array\n"); - return false; - } - memset(reco_cache_[seg], 0, segment_cnt_ * sizeof(*reco_cache_[seg])); samp_cache_[seg] = new CharSamp *[segment_cnt_]; - if (samp_cache_[seg] == NULL) { - fprintf(stderr, "Cube ERROR (CubeSearchObject::Init): could not " - "allocate a single segment's CharSamp array\n"); - return false; - } - memset(samp_cache_[seg], 0, segment_cnt_ * sizeof(*samp_cache_[seg])); } @@ -305,12 +283,10 @@ CharAltList * CubeSearchObject::RecognizeSegment(int start_pt, int end_pt) { exp(-fabs(seg_cnt - 2.0)) * exp(-samp->Width() / static_cast(samp->Height())); - if (alt_list) { - for (int class_idx = 0; class_idx < class_cnt; class_idx++) { - alt_list->Insert(class_idx, CubeUtils::Prob2Cost(prob_val)); - } - reco_cache_[start_pt + 1][end_pt] = alt_list; + for (int class_idx = 0; class_idx < class_cnt; class_idx++) { + alt_list->Insert(class_idx, CubeUtils::Prob2Cost(prob_val)); } + reco_cache_[start_pt + 1][end_pt] = alt_list; } return reco_cache_[start_pt + 1][end_pt]; @@ -353,11 +329,6 @@ bool CubeSearchObject::ComputeSpaceCosts() { // segmentation point int *max_left_x = new int[segment_cnt_ - 1]; int *min_right_x = new int[segment_cnt_ - 1]; - if (!max_left_x || !min_right_x) { - delete []min_right_x; - delete []max_left_x; - return false; - } if (rtl_) { min_right_x[0] = segments_[0]->Left(); max_left_x[segment_cnt_ - 2] = segments_[segment_cnt_ - 1]->Right(); @@ -384,11 +355,6 @@ bool CubeSearchObject::ComputeSpaceCosts() { // trivial cases space_cost_ = new int[segment_cnt_ - 1]; no_space_cost_ = new int[segment_cnt_ - 1]; - if (!space_cost_ || !no_space_cost_) { - delete []min_right_x; - delete []max_left_x; - return false; - } // go through all segmentation points determining the horizontal gap between // the images on both sides of each break points. Use the gap to estimate @@ -400,7 +366,7 @@ bool CubeSearchObject::ComputeSpaceCosts() { float prob = 0.0; // gap is too small => no space - if (gap < min_spc_gap_) { + if (gap < min_spc_gap_ || max_spc_gap_ == min_spc_gap_) { prob = 0.0; } else if (gap > max_spc_gap_) { // gap is too big => definite space diff --git a/cube/cube_tuning_params.cpp b/cube/cube_tuning_params.cpp index ac16c9f5cb..e4a9b0cf02 100644 --- a/cube/cube_tuning_params.cpp +++ b/cube/cube_tuning_params.cpp @@ -54,11 +54,6 @@ CubeTuningParams::~CubeTuningParams() { CubeTuningParams *CubeTuningParams::Create(const string &data_file_path, const string &lang) { CubeTuningParams *obj = new CubeTuningParams(); - if (!obj) { - fprintf(stderr, "Cube ERROR (CubeTuningParams::Create): unable to " - "allocate new tuning params object\n"); - return NULL; - } string tuning_params_file; tuning_params_file = data_file_path + lang; diff --git a/cube/cube_utils.cpp b/cube/cube_utils.cpp index 13c9c236da..4741659d2a 100644 --- a/cube/cube_utils.cpp +++ b/cube/cube_utils.cpp @@ -90,9 +90,6 @@ int CubeUtils::StrCmp(const char_32 *str1, const char_32 *str2) { char_32 *CubeUtils::StrDup(const char_32 *str32) { int len = StrLen(str32); char_32 *new_str = new char_32[len + 1]; - if (new_str == NULL) { - return NULL; - } memcpy(new_str, str32, len * sizeof(*str32)); new_str[len] = 0; return new_str; @@ -165,9 +162,6 @@ unsigned char *CubeUtils::GetImageData(Pix *pix, int left, int top, // copy the char img to a temp buffer unsigned char *temp_buff = new unsigned char[wid * hgt]; - if (temp_buff == NULL) { - return NULL; - } l_int32 w; l_int32 h; l_int32 d; @@ -211,10 +205,6 @@ bool CubeUtils::ReadFileToString(const string &file_name, string *str) { // read the contents rewind(fp); char *buff = new char[file_size]; - if (buff == NULL) { - fclose(fp); - return false; - } int read_bytes = fread(buff, 1, static_cast(file_size), fp); if (read_bytes == file_size) { str->append(buff, file_size); @@ -352,8 +342,6 @@ char_32 *CubeUtils::ToLower(const char_32 *str32, CharSet *char_set) { UNICHARSET *unicharset = char_set->InternalUnicharset(); int len = StrLen(str32); char_32 *lower = new char_32[len + 1]; - if (!lower) - return NULL; for (int i = 0; i < len; ++i) { char_32 ch = str32[i]; if (ch == INVALID_UNICHAR_ID) { @@ -385,8 +373,6 @@ char_32 *CubeUtils::ToUpper(const char_32 *str32, CharSet *char_set) { UNICHARSET *unicharset = char_set->InternalUnicharset(); int len = StrLen(str32); char_32 *upper = new char_32[len + 1]; - if (!upper) - return NULL; for (int i = 0; i < len; ++i) { char_32 ch = str32[i]; if (ch == INVALID_UNICHAR_ID) { diff --git a/cube/hybrid_neural_net_classifier.cpp b/cube/hybrid_neural_net_classifier.cpp index 671a74acdf..29b50d0cbc 100644 --- a/cube/hybrid_neural_net_classifier.cpp +++ b/cube/hybrid_neural_net_classifier.cpp @@ -136,14 +136,7 @@ bool HybridNeuralNetCharClassifier::RunNets(CharSamp *char_samp) { // allocate i/p and o/p buffers if needed if (net_input_ == NULL) { net_input_ = new float[feat_cnt]; - if (net_input_ == NULL) { - return false; - } - net_output_ = new float[class_cnt]; - if (net_output_ == NULL) { - return false; - } } // compute input features @@ -196,9 +189,6 @@ CharAltList *HybridNeuralNetCharClassifier::Classify(CharSamp *char_samp) { // create an altlist CharAltList *alt_list = new CharAltList(char_set_, class_cnt); - if (alt_list == NULL) { - return NULL; - } for (int out = 1; out < class_cnt; out++) { int cost = CubeUtils::Prob2Cost(net_output_[out]); @@ -240,14 +230,7 @@ bool HybridNeuralNetCharClassifier::LoadFoldingSets( CubeUtils::SplitStringUsing(fold_sets_str, "\r\n", &str_vec); fold_set_cnt_ = str_vec.size(); fold_sets_ = new int *[fold_set_cnt_]; - if (fold_sets_ == NULL) { - return false; - } fold_set_len_ = new int[fold_set_cnt_]; - if (fold_set_len_ == NULL) { - fold_set_cnt_ = 0; - return false; - } for (int fold_set = 0; fold_set < fold_set_cnt_; fold_set++) { reinterpret_cast(lang_mod)->RemoveInvalidCharacters( @@ -266,12 +249,6 @@ bool HybridNeuralNetCharClassifier::LoadFoldingSets( CubeUtils::UTF8ToUTF32(str_vec[fold_set].c_str(), &str32); fold_set_len_[fold_set] = str32.length(); fold_sets_[fold_set] = new int[fold_set_len_[fold_set]]; - if (fold_sets_[fold_set] == NULL) { - fprintf(stderr, "Cube ERROR (ConvNetCharClassifier::LoadFoldingSets): " - "could not allocate folding set\n"); - fold_set_cnt_ = fold_set; - return false; - } for (int ch = 0; ch < fold_set_len_[fold_set]; ch++) { fold_sets_[fold_set][ch] = char_set_->ClassID(str32[ch]); } @@ -330,7 +307,7 @@ bool HybridNeuralNetCharClassifier::LoadNets(const string &data_file_path, // split into lines vector str_vec; CubeUtils::SplitStringUsing(str, "\r\n", &str_vec); - if (str_vec.size() <= 0) { + if (str_vec.empty()) { return false; } diff --git a/cube/search_column.cpp b/cube/search_column.cpp index 9a042d016a..71f2222337 100644 --- a/cube/search_column.cpp +++ b/cube/search_column.cpp @@ -62,9 +62,6 @@ bool SearchColumn::Init() { // create hash table if (node_hash_table_ == NULL) { node_hash_table_ = new SearchNodeHashTable(); - if (node_hash_table_ == NULL) { - return false; - } } init_ = true; @@ -144,9 +141,6 @@ SearchNode *SearchColumn::AddNode(LangModEdge *edge, int reco_cost, // node does not exist if (new_node == NULL) { new_node = new SearchNode(cntxt, parent_node, reco_cost, edge, col_idx_); - if (new_node == NULL) { - return NULL; - } // if the max node count has already been reached, check if the cost of // the new node exceeds the max cost. This indicates that it will be pruned @@ -161,10 +155,6 @@ SearchNode *SearchColumn::AddNode(LangModEdge *edge, int reco_cost, // alloc a new buff SearchNode **new_node_buff = new SearchNode *[node_cnt_ + kNodeAllocChunk]; - if (new_node_buff == NULL) { - delete new_node; - return NULL; - } // free existing after copying contents if (node_array_ != NULL) { @@ -195,9 +185,7 @@ SearchNode *SearchColumn::AddNode(LangModEdge *edge, int reco_cost, } // free the edge - if (edge != NULL) { - delete edge; - } + delete edge; } // update Min and Max Costs diff --git a/cube/search_node.cpp b/cube/search_node.cpp index ff5bfbd844..cd46625023 100644 --- a/cube/search_node.cpp +++ b/cube/search_node.cpp @@ -147,9 +147,6 @@ char_32 *SearchNode::PathString() { } char_32 *char_ptr = new char_32[len + 1]; - if (char_ptr == NULL) { - return NULL; - } int ch_idx = len; diff --git a/cube/tess_lang_mod_edge.cpp b/cube/tess_lang_mod_edge.cpp index 4d16f3ac28..911070d3e6 100644 --- a/cube/tess_lang_mod_edge.cpp +++ b/cube/tess_lang_mod_edge.cpp @@ -72,9 +72,6 @@ TessLangModEdge::TessLangModEdge(CubeRecoContext *cntxt, const Dawg *dawg, char *TessLangModEdge::Description() const { char *char_ptr = new char[256]; - if (!char_ptr) { - return NULL; - } char dawg_str[256]; char edge_str[32]; @@ -115,9 +112,8 @@ int TessLangModEdge::CreateChildren(CubeRecoContext *cntxt, for (int i = 0; i < vec.size(); ++i) { const NodeChild &child = vec[i]; if (child.unichar_id == INVALID_UNICHAR_ID) continue; - edge_array[edge_cnt] = + edge_array[edge_cnt++] = new TessLangModEdge(cntxt, dawg, child.edge_ref, child.unichar_id); - if (edge_array[edge_cnt] != NULL) edge_cnt++; } return edge_cnt; } diff --git a/cube/tess_lang_model.cpp b/cube/tess_lang_model.cpp index 5113207260..3a4c7500d7 100644 --- a/cube/tess_lang_model.cpp +++ b/cube/tess_lang_model.cpp @@ -182,9 +182,6 @@ LangModEdge ** TessLangModel::GetEdges(CharAltList *alt_list, // preallocate the edge buffer (*edge_cnt) = dawg_cnt * max_edge_; edge_array = new LangModEdge *[(*edge_cnt)]; - if (edge_array == NULL) { - return NULL; - } for (int dawg_idx = (*edge_cnt) = 0; dawg_idx < dawg_cnt; dawg_idx++) { const Dawg *curr_dawg = GetDawg(dawg_idx); @@ -213,9 +210,6 @@ LangModEdge ** TessLangModel::GetEdges(CharAltList *alt_list, (*edge_cnt) = max_edge_; // allocate memory for edges edge_array = new LangModEdge *[(*edge_cnt)]; - if (edge_array == NULL) { - return NULL; - } // get the FanOut edges from the root of each dawg (*edge_cnt) = FanOut(alt_list, @@ -240,9 +234,6 @@ int TessLangModel::Edges(const char *strng, const Dawg *dawg, // create an edge object edge_array[edge_cnt] = new TessLangModEdge(cntxt_, dawg, edge_ref, class_id); - if (edge_array[edge_cnt] == NULL) { - return 0; - } reinterpret_cast(edge_array[edge_cnt])-> SetEdgeMask(edge_mask); @@ -264,10 +255,6 @@ int TessLangModel::OODEdges(CharAltList *alt_list, EDGE_REF edge_ref, alt_list->ClassCost(class_id) <= max_ood_shape_cost_)) { // create an edge object edge_array[edge_cnt] = new TessLangModEdge(cntxt_, class_id); - if (edge_array[edge_cnt] == NULL) { - return 0; - } - edge_cnt++; } } @@ -368,11 +355,9 @@ int TessLangModel::FanOut(CharAltList *alt_list, const Dawg *dawg, edge_array[edge_cnt] = new TessLangModEdge(cntxt_, dawg, child_edge->StartEdge(), child_edge->EndEdge(), class_id); - if (edge_array[edge_cnt] != NULL) { - reinterpret_cast(edge_array[edge_cnt])-> + reinterpret_cast(edge_array[edge_cnt])-> SetEdgeMask(edge_mask); - edge_cnt++; - } + edge_cnt++; } } } @@ -486,8 +471,6 @@ void TessLangModel::RemoveInvalidCharacters(string *lm_str) { int len = CubeUtils::StrLen(lm_str32.c_str()); char_32 *clean_str32 = new char_32[len + 1]; - if (!clean_str32) - return; int clean_len = 0; for (int i = 0; i < len; ++i) { int class_id = char_set->ClassID((char_32)lm_str32[i]); diff --git a/cube/word_altlist.cpp b/cube/word_altlist.cpp index d6775360ad..f91d56c996 100644 --- a/cube/word_altlist.cpp +++ b/cube/word_altlist.cpp @@ -45,11 +45,6 @@ bool WordAltList::Insert(char_32 *word_str, int cost, void *tag) { word_alt_ = new char_32*[max_alt_]; alt_cost_ = new int[max_alt_]; alt_tag_ = new void *[max_alt_]; - - if (word_alt_ == NULL || alt_cost_ == NULL || alt_tag_ == NULL) { - return false; - } - memset(alt_tag_, 0, max_alt_ * sizeof(*alt_tag_)); } else { // check if alt already exists @@ -69,9 +64,6 @@ bool WordAltList::Insert(char_32 *word_str, int cost, void *tag) { int len = CubeUtils::StrLen(word_str); word_alt_[alt_cnt_] = new char_32[len + 1]; - if (word_alt_[alt_cnt_] == NULL) { - return false; - } if (len > 0) { memcpy(word_alt_[alt_cnt_], word_str, len * sizeof(*word_str)); diff --git a/cube/word_list_lang_model.cpp b/cube/word_list_lang_model.cpp index 67a6a5a985..bb07951d25 100644 --- a/cube/word_list_lang_model.cpp +++ b/cube/word_list_lang_model.cpp @@ -54,9 +54,6 @@ bool WordListLangModel::Init() { // false for now, until Cube has a way to express its preferred debug level. dawg_ = new Trie(DAWG_TYPE_WORD, "", NO_PERM, cntxt_->CharacterSet()->ClassCount(), false); - if (dawg_ == NULL) { - return false; - } init_ = true; return true; } @@ -97,9 +94,6 @@ LangModEdge **WordListLangModel::GetEdges(CharAltList *alt_list, // allocate memory for edges LangModEdge **edge_array = new LangModEdge *[kMaxEdge]; - if (edge_array == NULL) { - return NULL; - } // now get all the emerging edges (*edge_cnt) += TessLangModEdge::CreateChildren(cntxt_, dawg_, edge_ref, diff --git a/cube/word_size_model.cpp b/cube/word_size_model.cpp index 6b9a4530fc..be3ccf734d 100644 --- a/cube/word_size_model.cpp +++ b/cube/word_size_model.cpp @@ -43,11 +43,6 @@ WordSizeModel *WordSizeModel::Create(const string &data_file_path, CharSet *char_set, bool contextual) { WordSizeModel *obj = new WordSizeModel(char_set, contextual); - if (!obj) { - fprintf(stderr, "Cube ERROR (WordSizeModel::Create): unable to allocate " - "new word size model object\n"); - return NULL; - } if (!obj->Init(data_file_path, lang)) { delete obj; @@ -96,19 +91,9 @@ bool WordSizeModel::Init(const string &data_file_path, const string &lang) { FontPairSizeInfo fnt_info; fnt_info.pair_size_info = new PairSizeInfo *[size_class_cnt]; - if (!fnt_info.pair_size_info) { - fprintf(stderr, "Cube ERROR (WordSizeModel::Init): error allcoating " - "memory for font pair size info\n"); - return false; - } fnt_info.pair_size_info[0] = new PairSizeInfo[size_class_cnt * size_class_cnt]; - if (!fnt_info.pair_size_info[0]) { - fprintf(stderr, "Cube ERROR (WordSizeModel::Init): error allocating " - "memory for font pair size info\n"); - return false; - } memset(fnt_info.pair_size_info[0], 0, size_class_cnt * size_class_cnt * sizeof(PairSizeInfo)); diff --git a/cube/word_unigrams.cpp b/cube/word_unigrams.cpp index 70cc9ee1e2..052a025c90 100644 --- a/cube/word_unigrams.cpp +++ b/cube/word_unigrams.cpp @@ -76,32 +76,13 @@ WordUnigrams *WordUnigrams::Create(const string &data_file_path, // allocate memory WordUnigrams *word_unigrams_obj = new WordUnigrams(); - if (word_unigrams_obj == NULL) { - fprintf(stderr, "Cube ERROR (WordUnigrams::Create): could not create " - "word unigrams object.\n"); - return NULL; - } int full_len = str.length(); int word_cnt = str_vec.size() / 2; word_unigrams_obj->words_ = new char*[word_cnt]; word_unigrams_obj->costs_ = new int[word_cnt]; - if (word_unigrams_obj->words_ == NULL || - word_unigrams_obj->costs_ == NULL) { - fprintf(stderr, "Cube ERROR (WordUnigrams::Create): error allocating " - "word unigram fields.\n"); - delete word_unigrams_obj; - return NULL; - } - word_unigrams_obj->words_[0] = new char[full_len]; - if (word_unigrams_obj->words_[0] == NULL) { - fprintf(stderr, "Cube ERROR (WordUnigrams::Create): error allocating " - "word unigram fields.\n"); - delete word_unigrams_obj; - return NULL; - } // construct sorted list of words and costs word_unigrams_obj->word_cnt_ = 0; @@ -163,7 +144,7 @@ int WordUnigrams::Cost(const char_32 *key_str32, CubeUtils::SplitStringUsing(key_str, " \t", &words); // no words => no cost - if (words.size() <= 0) { + if (words.empty()) { return 0; } diff --git a/cutil/Makefile.am b/cutil/Makefile.am index 15b339c8f3..5b0ffc6ebb 100644 --- a/cutil/Makefile.am +++ b/cutil/Makefile.am @@ -7,7 +7,7 @@ endif noinst_HEADERS = \ bitvec.h callcpp.h const.h cutil.h cutil_class.h danerror.h efio.h \ - emalloc.h freelist.h globals.h listio.h \ + emalloc.h freelist.h globals.h \ oldlist.h structures.h if !USING_MULTIPLELIBS @@ -22,7 +22,7 @@ endif libtesseract_cutil_la_SOURCES = \ bitvec.cpp callcpp.cpp cutil.cpp cutil_class.cpp danerror.cpp efio.cpp \ - emalloc.cpp freelist.cpp listio.cpp \ + emalloc.cpp freelist.cpp \ oldlist.cpp structures.cpp diff --git a/cutil/bitvec.h b/cutil/bitvec.h index f70d748b91..d2a364d261 100644 --- a/cutil/bitvec.h +++ b/cutil/bitvec.h @@ -30,29 +30,29 @@ typedef uinT32 *BIT_VECTOR; /*----------------------------------------------------------------------------- Public Function Prototypes -----------------------------------------------------------------------------*/ -#define zero_all_bits(array,length) \ -{\ - int index; /*temporary index*/\ -\ -for (index=0;index -#include -#include -#include "listio.h" - -/*--------------------------------------------------------------------------- - Public Function Code ----------------------------------------------------------------------------*/ -/************************************************************************* - * R E A D L I S T - * - * Read a list of strings from a file. Return the string list to the - * caller. - *************************************************************************/ -LIST read_list(const char *filename) { - FILE *infile; - char s[CHARS_PER_LINE]; - LIST list; - - if ((infile = open_file (filename, "r")) == NULL) - return (NIL_LIST); - - list = NIL_LIST; - while (fgets (s, CHARS_PER_LINE, infile) != NULL) { - s[CHARS_PER_LINE - 1] = '\0'; - if (strlen (s) > 0) { - if (s[strlen (s) - 1] == '\n') - s[strlen (s) - 1] = '\0'; - if (strlen (s) > 0) { - list = push (list, (LIST) strsave (s)); - } - } - } - - fclose(infile); - return (reverse_d (list)); -} diff --git a/cutil/listio.h b/cutil/listio.h deleted file mode 100644 index 7d9c19f777..0000000000 --- a/cutil/listio.h +++ /dev/null @@ -1,43 +0,0 @@ -/* -*-C-*- -################################################################################ -# -# File: listio.h -# Description: List I/O processing procedures. -# Author: Mark Seaman, Software Productivity -# Created: Thu Jul 23 13:24:09 1987 -# Modified: Mon Oct 16 11:38:52 1989 (Mark Seaman) marks@hpgrlt -# Language: C -# Package: N/A -# Status: Reusable Software Component -# -# (c) Copyright 1987, Hewlett-Packard Company. -** Licensed under the Apache License, Version 2.0 (the "License"); -** you may not use this file except in compliance with the License. -** You may obtain a copy of the License at -** http://www.apache.org/licenses/LICENSE-2.0 -** Unless required by applicable law or agreed to in writing, software -** distributed under the License is distributed on an "AS IS" BASIS, -** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -** See the License for the specific language governing permissions and -** limitations under the License. -# -################################################################################ - * Revision 1.5 89/06/27 11:56:00 11:56:00 marks (Mark Seaman) - * Fixed MAC_OR_DOS bug - * - - This file contains the interface definitions to a set of general purpose - list I/O routines. - -***********************************************************************/ -#ifndef LISTIO_H -#define LISTIO_H - -#include -#include "oldlist.h" - -/*---------------------------------------------------------------------------- - Public Function Prototypes ---------------------------------------------------------------------------*/ -LIST read_list(const char *filename); -#endif diff --git a/cutil/oldlist.cpp b/cutil/oldlist.cpp index 52c0d8680a..9e3f6f4c06 100644 --- a/cutil/oldlist.cpp +++ b/cutil/oldlist.cpp @@ -206,8 +206,8 @@ void destroy_nodes(LIST list, void_dest destructor) { destructor = memfree; while (list != NIL_LIST) { - (*destructor) (first_node (list)); - list = pop (list); + if (first_node(list) != NULL) (*destructor)(first_node(list)); + list = pop(list); } } @@ -401,7 +401,6 @@ LIST s_adjoin(LIST var_list, void *variable, int_compare compare) { return (push_last (var_list, variable)); } - /********************************************************************** * s e a r c h * diff --git a/dict/context.cpp b/dict/context.cpp index a9acb137c3..368ad462b4 100644 --- a/dict/context.cpp +++ b/dict/context.cpp @@ -32,28 +32,22 @@ namespace tesseract { static const int kMinAbsoluteGarbageWordLength = 10; static const float kMinAbsoluteGarbageAlphanumFrac = 0.5f; -const int case_state_table[6][4] = { { - /* 0. Beginning of word */ - /* P U L D */ - /* -1. Error on case */ - 0, 1, 5, 4 - }, - { /* 1. After initial capital */ - 0, 3, 2, 4 - }, - { /* 2. After lower case */ - 0, -1, 2, -1 - }, - { /* 3. After upper case */ - 0, 3, -1, 4 - }, - { /* 4. After a digit */ - 0, -1, -1, 4 - }, - { /* 5. After initial lower case */ - 5, -1, 2, -1 - }, - }; +const int case_state_table[6][4] = { + {/* 0. Beginning of word */ + /* P U L D */ + /* -1. Error on case */ + 0, 1, 5, 4}, + {/* 1. After initial capital */ + 0, 3, 2, 4}, + {/* 2. After lower case */ + 0, -1, 2, -1}, + {/* 3. After upper case */ + 0, 3, -1, 4}, + {/* 4. After a digit */ + 0, -1, -1, 4}, + {/* 5. After initial lower case */ + 5, -1, 2, -1}, +}; int Dict::case_ok(const WERD_CHOICE &word, const UNICHARSET &unicharset) { int state = 0; diff --git a/dict/dawg.h b/dict/dawg.h index b37e771503..bd789f3722 100644 --- a/dict/dawg.h +++ b/dict/dawg.h @@ -128,7 +128,7 @@ class Dawg { inline const STRING &lang() const { return lang_; } inline PermuterType permuter() const { return perm_; } - virtual ~Dawg() {}; + virtual ~Dawg() {} /// Returns true if the given word is in the Dawg. bool word_in_dawg(const WERD_CHOICE &word) const; @@ -183,13 +183,20 @@ class Dawg { /// of the given unichar_id. virtual void unichar_id_to_patterns(UNICHAR_ID unichar_id, const UNICHARSET &unicharset, - GenericVector *vec) const {}; + GenericVector *vec) const { + (void)unichar_id; + (void)unicharset; + (void)vec; + } /// Returns the given EDGE_REF if the EDGE_RECORD that it points to has /// a self loop and the given unichar_id matches the unichar_id stored in the /// EDGE_RECORD, returns NO_EDGE otherwise. virtual EDGE_REF pattern_loop_edge( EDGE_REF edge_ref, UNICHAR_ID unichar_id, bool word_end) const { + (void)edge_ref; + (void)unichar_id; + (void)word_end; return false; } @@ -368,14 +375,6 @@ struct DawgPosition { class DawgPositionVector : public GenericVector { public: - /// Overload destructor, since clear() does not delete data_[] any more. - ~DawgPositionVector() { - if (size_reserved_ > 0) { - delete[] data_; - size_used_ = 0; - size_reserved_ = 0; - } - } /// Overload clear() in order to avoid allocating/deallocating memory /// when clearing the vector and re-inserting entries into it later. void clear() { size_used_ = 0; } diff --git a/dict/dict.cpp b/dict/dict.cpp index 6f8e0249a1..0dcc7d6c6e 100644 --- a/dict/dict.cpp +++ b/dict/dict.cpp @@ -30,13 +30,12 @@ namespace tesseract { class Image; -Dict::Dict(CCUtil* ccutil) +Dict::Dict(CCUtil *ccutil) : letter_is_okay_(&tesseract::Dict::def_letter_is_okay), probability_in_context_(&tesseract::Dict::def_probability_in_context), params_model_classify_(NULL), ccutil_(ccutil), - STRING_MEMBER(user_words_file, "", - "A filename of user-provided words.", + STRING_MEMBER(user_words_file, "", "A filename of user-provided words.", getCCUtil()->params()), STRING_INIT_MEMBER(user_words_suffix, "", "A suffix of user-provided words located in tessdata.", @@ -54,33 +53,41 @@ Dict::Dict(CCUtil* ccutil) getCCUtil()->params()), BOOL_INIT_MEMBER(load_unambig_dawg, true, "Load unambiguous word dawg.", getCCUtil()->params()), - BOOL_INIT_MEMBER(load_punc_dawg, true, "Load dawg with punctuation" - " patterns.", getCCUtil()->params()), - BOOL_INIT_MEMBER(load_number_dawg, true, "Load dawg with number" - " patterns.", getCCUtil()->params()), - BOOL_INIT_MEMBER(load_bigram_dawg, true, "Load dawg with special word " - "bigrams.", getCCUtil()->params()), + BOOL_INIT_MEMBER(load_punc_dawg, true, + "Load dawg with punctuation" + " patterns.", + getCCUtil()->params()), + BOOL_INIT_MEMBER(load_number_dawg, true, + "Load dawg with number" + " patterns.", + getCCUtil()->params()), + BOOL_INIT_MEMBER(load_bigram_dawg, true, + "Load dawg with special word " + "bigrams.", + getCCUtil()->params()), double_MEMBER(xheight_penalty_subscripts, 0.125, "Score penalty (0.1 = 10%) added if there are subscripts " "or superscripts in a word, but it is otherwise OK.", getCCUtil()->params()), double_MEMBER(xheight_penalty_inconsistent, 0.25, "Score penalty (0.1 = 10%) added if an xheight is " - "inconsistent.", getCCUtil()->params()), + "inconsistent.", + getCCUtil()->params()), double_MEMBER(segment_penalty_dict_frequent_word, 1.0, "Score multiplier for word matches which have good case and" "are frequent in the given language (lower is better).", getCCUtil()->params()), double_MEMBER(segment_penalty_dict_case_ok, 1.1, "Score multiplier for word matches that have good case " - "(lower is better).", getCCUtil()->params()), + "(lower is better).", + getCCUtil()->params()), double_MEMBER(segment_penalty_dict_case_bad, 1.3125, "Default score multiplier for word matches, which may have " "case issues (lower is better).", getCCUtil()->params()), double_MEMBER(segment_penalty_ngram_best_choice, 1.24, - "Multipler to for the best choice from the ngram model.", - getCCUtil()->params()), + "Multipler to for the best choice from the ngram model.", + getCCUtil()->params()), double_MEMBER(segment_penalty_dict_nonword, 1.25, "Score multiplier for glyph fragment segmentations which " "do not match a dictionary word (lower is better).", @@ -88,11 +95,13 @@ Dict::Dict(CCUtil* ccutil) double_MEMBER(segment_penalty_garbage, 1.50, "Score multiplier for poorly cased strings that are not in" " the dictionary and generally look like garbage (lower is" - " better).", getCCUtil()->params()), + " better).", + getCCUtil()->params()), STRING_MEMBER(output_ambig_words_file, "", "Output file for ambiguities found in the dictionary", getCCUtil()->params()), - INT_MEMBER(dawg_debug_level, 0, "Set to 1 for general debug info" + INT_MEMBER(dawg_debug_level, 0, + "Set to 1 for general debug info" ", to 2 for more details, to 3 to see all the debug messages", getCCUtil()->params()), INT_MEMBER(hyphen_debug_level, 0, "Debug level for hyphenated words.", @@ -109,12 +118,12 @@ Dict::Dict(CCUtil* ccutil) "Certainty threshold for non-dict words", getCCUtil()->params()), double_MEMBER(stopper_phase2_certainty_rejection_offset, 1.0, - "Reject certainty offset", - getCCUtil()->params()), + "Reject certainty offset", getCCUtil()->params()), INT_MEMBER(stopper_smallword_size, 2, "Size of dict word to be treated as non-dict word", getCCUtil()->params()), - double_MEMBER(stopper_certainty_per_char, -0.50, "Certainty to add" + double_MEMBER(stopper_certainty_per_char, -0.50, + "Certainty to add" " for each dict char above small word size.", getCCUtil()->params()), double_MEMBER(stopper_allowable_character_badness, 3.0, @@ -130,9 +139,9 @@ Dict::Dict(CCUtil* ccutil) "Deprecated- backward compatibility only", getCCUtil()->params()), INT_MEMBER(tessedit_truncate_wordchoice_log, 10, - "Max words to keep in list", - getCCUtil()->params()), - STRING_MEMBER(word_to_debug, "", "Word for which stopper debug" + "Max words to keep in list", getCCUtil()->params()), + STRING_MEMBER(word_to_debug, "", + "Word for which stopper debug" " information should be printed to stdout", getCCUtil()->params()), STRING_MEMBER(word_to_debug_lengths, "", @@ -141,10 +150,10 @@ Dict::Dict(CCUtil* ccutil) INT_MEMBER(fragments_debug, 0, "Debug character fragments", getCCUtil()->params()), BOOL_MEMBER(segment_nonalphabetic_script, false, - "Don't use any alphabetic-specific tricks." - "Set to true in the traineddata config file for" - " scripts that are cursive or inherently fixed-pitch", - getCCUtil()->params()), + "Don't use any alphabetic-specific tricks." + "Set to true in the traineddata config file for" + " scripts that are cursive or inherently fixed-pitch", + getCCUtil()->params()), BOOL_MEMBER(save_doc_words, 0, "Save Document Words", getCCUtil()->params()), double_MEMBER(doc_dict_pending_threshold, 0.0, @@ -152,8 +161,10 @@ Dict::Dict(CCUtil* ccutil) getCCUtil()->params()), double_MEMBER(doc_dict_certainty_threshold, -2.25, "Worst certainty for words that can be inserted into the" - "document dictionary", getCCUtil()->params()), - INT_MEMBER(max_permuter_attempts, 10000, "Maximum number of different" + "document dictionary", + getCCUtil()->params()), + INT_MEMBER(max_permuter_attempts, 10000, + "Maximum number of different" " character choices to consider during permutation." " This limit is especially useful when user patterns" " are specified, since overly generic patterns can result in" @@ -179,7 +190,8 @@ Dict::Dict(CCUtil* ccutil) } Dict::~Dict() { - if (hyphen_word_ != NULL) delete hyphen_word_; + End(); + delete hyphen_word_; if (output_ambig_words_file_ != NULL) fclose(output_ambig_words_file_); } @@ -190,10 +202,8 @@ DawgCache *Dict::GlobalDawgCache() { return &cache; } -void Dict::Load(DawgCache *dawg_cache) { - STRING name; - STRING &lang = getCCUtil()->lang; - +// Sets up ready for a Load. +void Dict::SetupForLoad(DawgCache *dawg_cache) { if (dawgs_.length() != 0) this->End(); apostrophe_unichar_id_ = getUnicharset().unichar_to_id(kApostropheSymbol); @@ -208,10 +218,10 @@ void Dict::Load(DawgCache *dawg_cache) { dawg_cache_ = new DawgCache(); dawg_cache_is_ours_ = true; } +} - TessdataManager &tessdata_manager = getCCUtil()->tessdata_manager; - const char *data_file_name = tessdata_manager.GetDataFileName().string(); - +// Loads the dawgs needed by Tesseract. Call FinishLoad() after. +void Dict::Load(const char *data_file_name, const STRING &lang) { // Load dawgs_. if (load_punc_dawg) { punc_dawg_ = dawg_cache_->GetSquishedDawg( @@ -243,6 +253,7 @@ void Dict::Load(DawgCache *dawg_cache) { if (unambig_dawg_) dawgs_ += unambig_dawg_; } + STRING name; if (((STRING &)user_words_suffix).length() > 0 || ((STRING &)user_words_file).length() > 0) { Trie *trie_ptr = new Trie(DAWG_TYPE_WORD, lang, USER_DAWG_PERM, @@ -288,8 +299,13 @@ void Dict::Load(DawgCache *dawg_cache) { // This dawg is temporary and should not be searched by letter_is_ok. pending_words_ = new Trie(DAWG_TYPE_WORD, lang, NO_PERM, getUnicharset().size(), dawg_debug_level); +} - // Construct a list of corresponding successors for each dawg. Each entry i +// Completes the loading process after Load(). +// Returns false if no dictionaries were loaded. +bool Dict::FinishLoad() { + if (dawgs_.empty()) return false; + // Construct a list of corresponding successors for each dawg. Each entry, i, // in the successors_ vector is a vector of integers that represent the // indices into the dawgs_ vector of the successors for dawg i. successors_.reserve(dawgs_.length()); @@ -304,6 +320,7 @@ void Dict::Load(DawgCache *dawg_cache) { } successors_ += lst; } + return true; } void Dict::End() { @@ -323,10 +340,8 @@ void Dict::End() { dawgs_.clear(); successors_.clear(); document_words_ = NULL; - if (pending_words_ != NULL) { - delete pending_words_; - pending_words_ = NULL; - } + delete pending_words_; + pending_words_ = NULL; } // Returns true if in light of the current state unichar_id is allowed @@ -356,6 +371,7 @@ int Dict::def_letter_is_okay(void* void_dawg_args, // Initialization. PermuterType curr_perm = NO_PERM; dawg_args->updated_dawgs->clear(); + dawg_args->valid_end = false; // Go over the active_dawgs vector and insert DawgPosition records // with the updated ref (an edge with the corresponding unichar id) into @@ -393,6 +409,9 @@ int Dict::def_letter_is_okay(void* void_dawg_args, dawg_debug_level > 0, "Append transition from punc dawg to current dawgs: "); if (sdawg->permuter() > curr_perm) curr_perm = sdawg->permuter(); + if (sdawg->end_of_word(dawg_edge) && + punc_dawg->end_of_word(punc_transition_edge)) + dawg_args->valid_end = true; } } } @@ -407,6 +426,7 @@ int Dict::def_letter_is_okay(void* void_dawg_args, dawg_debug_level > 0, "Extend punctuation dawg: "); if (PUNC_PERM > curr_perm) curr_perm = PUNC_PERM; + if (punc_dawg->end_of_word(punc_edge)) dawg_args->valid_end = true; } continue; } @@ -424,6 +444,7 @@ int Dict::def_letter_is_okay(void* void_dawg_args, dawg_debug_level > 0, "Return to punctuation dawg: "); if (dawg->permuter() > curr_perm) curr_perm = dawg->permuter(); + if (punc_dawg->end_of_word(punc_edge)) dawg_args->valid_end = true; } } @@ -433,8 +454,8 @@ int Dict::def_letter_is_okay(void* void_dawg_args, // possible edges, not only for the exact unichar_id, but also // for all its character classes (alpha, digit, etc). if (dawg->type() == DAWG_TYPE_PATTERN) { - ProcessPatternEdges(dawg, pos, unichar_id, word_end, - dawg_args->updated_dawgs, &curr_perm); + ProcessPatternEdges(dawg, pos, unichar_id, word_end, dawg_args, + &curr_perm); // There can't be any successors to dawg that is of type // DAWG_TYPE_PATTERN, so we are done examining this DawgPosition. continue; @@ -461,6 +482,9 @@ int Dict::def_letter_is_okay(void* void_dawg_args, continue; } if (dawg->permuter() > curr_perm) curr_perm = dawg->permuter(); + if (dawg->end_of_word(edge) && + (punc_dawg == NULL || punc_dawg->end_of_word(pos.punc_ref))) + dawg_args->valid_end = true; dawg_args->updated_dawgs->add_unique( DawgPosition(pos.dawg_index, edge, pos.punc_index, pos.punc_ref, false), @@ -485,7 +509,7 @@ int Dict::def_letter_is_okay(void* void_dawg_args, void Dict::ProcessPatternEdges(const Dawg *dawg, const DawgPosition &pos, UNICHAR_ID unichar_id, bool word_end, - DawgPositionVector *updated_dawgs, + DawgArgs *dawg_args, PermuterType *curr_perm) const { NODE_REF node = GetStartingNode(dawg, pos.dawg_ref); // Try to find the edge corresponding to the exact unichar_id and to all the @@ -508,7 +532,8 @@ void Dict::ProcessPatternEdges(const Dawg *dawg, const DawgPosition &pos, tprintf("Letter found in pattern dawg %d\n", pos.dawg_index); } if (dawg->permuter() > *curr_perm) *curr_perm = dawg->permuter(); - updated_dawgs->add_unique( + if (dawg->end_of_word(edge)) dawg_args->valid_end = true; + dawg_args->updated_dawgs->add_unique( DawgPosition(pos.dawg_index, edge, pos.punc_index, pos.punc_ref, pos.back_to_punc), dawg_debug_level > 0, @@ -804,5 +829,12 @@ bool Dict::valid_punctuation(const WERD_CHOICE &word) { return false; } +/// Returns true if the language is space-delimited (not CJ, or T). +bool Dict::IsSpaceDelimitedLang() const { + const UNICHARSET &u_set = getUnicharset(); + if (u_set.han_sid() > 0) return false; + if (u_set.katakana_sid() > 0) return false; + return true; +} } // namespace tesseract diff --git a/dict/dict.h b/dict/dict.h index 326f1235d5..a3883b5124 100644 --- a/dict/dict.h +++ b/dict/dict.h @@ -23,7 +23,6 @@ #include "dawg.h" #include "dawg_cache.h" #include "host.h" -#include "oldlist.h" #include "ratngs.h" #include "stopper.h" #include "trie.h" @@ -76,11 +75,13 @@ enum XHeightConsistencyEnum {XH_GOOD, XH_SUBNORMAL, XH_INCONSISTENT}; struct DawgArgs { DawgArgs(DawgPositionVector *d, DawgPositionVector *up, PermuterType p) - : active_dawgs(d), updated_dawgs(up), permuter(p) {} + : active_dawgs(d), updated_dawgs(up), permuter(p), valid_end(false) {} DawgPositionVector *active_dawgs; DawgPositionVector *updated_dawgs; PermuterType permuter; + // True if the current position is a valid word end. + bool valid_end; }; class Dict { @@ -243,7 +244,7 @@ class Dict { CHAR_FRAGMENT_INFO *char_frag_info); /* stopper.cpp *************************************************************/ - bool NoDangerousAmbig(WERD_CHOICE *BestChoice, + bool TESS_API NoDangerousAmbig(WERD_CHOICE *BestChoice, DANGERR *fixpt, bool fix_replaceable, MATRIX* ratings); @@ -294,7 +295,13 @@ class Dict { /// Initialize Dict class - load dawgs from [lang].traineddata and /// user-specified wordlist and parttern list. static DawgCache *GlobalDawgCache(); - void Load(DawgCache *dawg_cache); + // Sets up ready for a Load. + void SetupForLoad(DawgCache *dawg_cache); + // Loads the dawgs needed by Tesseract. Call FinishLoad() after. + void Load(const char *data_file_name, const STRING &lang); + // Completes the loading process after Load(). + // Returns false if no dictionaries were loaded. + bool FinishLoad(); void End(); // Resets the document dictionary analogous to ResetAdaptiveClassifier. @@ -374,10 +381,11 @@ class Dict { double def_probability_in_context( const char* lang, const char* context, int context_bytes, const char* character, int character_bytes) { - (void) context; - (void) context_bytes; - (void) character; - (void) character_bytes; + (void)lang; + (void)context; + (void)context_bytes; + (void)character; + (void)character_bytes; return 0.0; } double ngram_probability_in_context(const char* lang, @@ -397,9 +405,7 @@ class Dict { } inline void SetWildcardID(UNICHAR_ID id) { wildcard_unichar_id_ = id; } - inline UNICHAR_ID WildcardID() const { - return wildcard_unichar_id_; - } + inline UNICHAR_ID WildcardID() const { return wildcard_unichar_id_; } /// Return the number of dawgs in the dawgs_ vector. inline int NumDawgs() const { return dawgs_.size(); } /// Return i-th dawg pointer recorded in the dawgs_ vector. @@ -436,7 +442,7 @@ class Dict { /// edges were found. void ProcessPatternEdges(const Dawg *dawg, const DawgPosition &info, UNICHAR_ID unichar_id, bool word_end, - DawgPositionVector *updated_dawgs, + DawgArgs *dawg_args, PermuterType *current_permuter) const; /// Read/Write/Access special purpose dawgs which contain words @@ -483,6 +489,8 @@ class Dict { inline void SetWordsegRatingAdjustFactor(float f) { wordseg_rating_adjust_factor_ = f; } + /// Returns true if the language is space-delimited (not CJ, or T). + bool IsSpaceDelimitedLang() const; private: /** Private member variables. */ diff --git a/dict/stopper.cpp b/dict/stopper.cpp index 660b4c8cf3..a0e3c02b2e 100644 --- a/dict/stopper.cpp +++ b/dict/stopper.cpp @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: stopper.c - ** Purpose: Stopping criteria for word classifier. - ** Author: Dan Johnson - ** History: Mon Apr 29 14:56:49 1991, DSJ, Created. + ** Filename: stopper.c + ** Purpose: Stopping criteria for word classifier. + ** Author: Dan Johnson + ** History: Mon Apr 29 14:56:49 1991, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at @@ -41,7 +41,6 @@ #pragma warning(disable:4800) // int/bool warnings #endif -using tesseract::ScriptPos; /*---------------------------------------------------------------------------- Private Code ----------------------------------------------------------------------------*/ diff --git a/dict/stopper.h b/dict/stopper.h index b028b0ee29..58d23734dd 100644 --- a/dict/stopper.h +++ b/dict/stopper.h @@ -1,10 +1,10 @@ /****************************************************************************** - ** Filename: stopper.h - ** Purpose: Stopping criteria for word classifier. - ** Author: Dan Johnson - ** History: Wed May 1 09:42:57 1991, DSJ, Created. + ** Filename: stopper.h + ** Purpose: Stopping criteria for word classifier. + ** Author: Dan Johnson + ** History: Wed May 1 09:42:57 1991, DSJ, Created. ** - ** (c) Copyright Hewlett-Packard Company, 1988. + ** (c) Copyright Hewlett-Packard Company, 1988. ** Licensed under the Apache License, Version 2.0 (the "License"); ** you may not use this file except in compliance with the License. ** You may obtain a copy of the License at diff --git a/dict/trie.cpp b/dict/trie.cpp index 6a7a8d1e35..ac7dd33cdc 100644 --- a/dict/trie.cpp +++ b/dict/trie.cpp @@ -276,7 +276,6 @@ bool Trie::add_word_to_dawg(const WERD_CHOICE &word, NODE_REF Trie::new_dawg_node() { TRIE_NODE_RECORD *node = new TRIE_NODE_RECORD(); - if (node == NULL) return 0; // failed to create new node nodes_.push_back(node); return nodes_.length() - 1; } diff --git a/doc/ambiguous_words.1.html b/doc/ambiguous_words.1.html index 3fd5f7f1f6..be74b62d0d 100644 --- a/doc/ambiguous_words.1.html +++ b/doc/ambiguous_words.1.html @@ -1,790 +1,790 @@ - - - - - -AMBIGUOUS_WORDS(1) - - - - - -
-
-

SYNOPSIS

-
-

ambiguous_words [-l lang] TESSDATADIR WORDLIST AMBIGUOUSFILE

-
-
-
-

DESCRIPTION

-
-

ambiguous_words(1) runs Tesseract in a special mode, and for each word -in word list, produces a set of words which Tesseract thinks might be -ambiguous with it. TESSDATADIR must be set to the absolute path of -a directory containing tessdata/lang.traineddata.

-
-
-
-

SEE ALSO

-
-

tesseract(1)

-
-
-
-

COPYING

-
-

Copyright (C) 2012 Google, Inc. -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +AMBIGUOUS_WORDS(1) + + + + + +
+
+

SYNOPSIS

+
+

ambiguous_words [-l lang] TESSDATADIR WORDLIST AMBIGUOUSFILE

+
+
+
+

DESCRIPTION

+
+

ambiguous_words(1) runs Tesseract in a special mode, and for each word +in word list, produces a set of words which Tesseract thinks might be +ambiguous with it. TESSDATADIR must be set to the absolute path of +a directory containing tessdata/lang.traineddata.

+
+
+
+

SEE ALSO

+
+

tesseract(1)

+
+
+
+

COPYING

+
+

Copyright (C) 2012 Google, Inc. +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/ambiguous_words.1.xml b/doc/ambiguous_words.1.xml index 6293866ceb..4900c6eb93 100644 --- a/doc/ambiguous_words.1.xml +++ b/doc/ambiguous_words.1.xml @@ -1,43 +1,43 @@ - - - - - - - AMBIGUOUS_WORDS(1) - - -ambiguous_words -1 -  -  - - - ambiguous_words - generate sets of words Tesseract is likely to find ambiguous - - -ambiguous_words [-l lang] TESSDATADIR WORDLIST AMBIGUOUSFILE - - -DESCRIPTION -ambiguous_words(1) runs Tesseract in a special mode, and for each word -in word list, produces a set of words which Tesseract thinks might be -ambiguous with it. TESSDATADIR must be set to the absolute path of -a directory containing tessdata/lang.traineddata. - - -SEE ALSO -tesseract(1) - - -COPYING -Copyright (C) 2012 Google, Inc. -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + AMBIGUOUS_WORDS(1) + + +ambiguous_words +1 +  +  + + + ambiguous_words + generate sets of words Tesseract is likely to find ambiguous + + +ambiguous_words [-l lang] TESSDATADIR WORDLIST AMBIGUOUSFILE + + +DESCRIPTION +ambiguous_words(1) runs Tesseract in a special mode, and for each word +in word list, produces a set of words which Tesseract thinks might be +ambiguous with it. TESSDATADIR must be set to the absolute path of +a directory containing tessdata/lang.traineddata. + + +SEE ALSO +tesseract(1) + + +COPYING +Copyright (C) 2012 Google, Inc. +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/cntraining.1.html b/doc/cntraining.1.html index 706d3bd0f4..7653061e1e 100644 --- a/doc/cntraining.1.html +++ b/doc/cntraining.1.html @@ -1,805 +1,805 @@ - - - - - -CNTRAINING(1) - - - - - -
-
-

SYNOPSIS

-
-

cntraining [-D dir] FILE…

-
-
-
-

DESCRIPTION

-
-

cntraining takes a list of .tr files, from which it generates the -normproto data file (the character normalization sensitivity -prototypes).

-
-
-
-

OPTIONS

-
-
-
--D dir -
-
-

- Directory to write output files to. -

-
-
-
-
-
-

SEE ALSO

-
-

tesseract(1), shapeclustering(1), mftraining(1)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

COPYING

-
-

Copyright (c) Hewlett-Packard Company, 1988 -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +CNTRAINING(1) + + + + + +
+
+

SYNOPSIS

+
+

cntraining [-D dir] FILE…

+
+
+
+

DESCRIPTION

+
+

cntraining takes a list of .tr files, from which it generates the +normproto data file (the character normalization sensitivity +prototypes).

+
+
+
+

OPTIONS

+
+
+
+-D dir +
+
+

+ Directory to write output files to. +

+
+
+
+
+
+

SEE ALSO

+
+

tesseract(1), shapeclustering(1), mftraining(1)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

COPYING

+
+

Copyright (c) Hewlett-Packard Company, 1988 +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/cntraining.1.xml b/doc/cntraining.1.xml index 6795f12f2c..6efc99be1d 100644 --- a/doc/cntraining.1.xml +++ b/doc/cntraining.1.xml @@ -1,58 +1,58 @@ - - - - - - - CNTRAINING(1) - - -cntraining -1 -  -  - - - cntraining - character normalization training for Tesseract - - -cntraining [-D dir] FILE… - - -DESCRIPTION -cntraining takes a list of .tr files, from which it generates the -normproto data file (the character normalization sensitivity -prototypes). - - -OPTIONS - - - --D dir - - - - Directory to write output files to. - - - - - - -SEE ALSO -tesseract(1), shapeclustering(1), mftraining(1) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -COPYING -Copyright (c) Hewlett-Packard Company, 1988 -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + CNTRAINING(1) + + +cntraining +1 +  +  + + + cntraining + character normalization training for Tesseract + + +cntraining [-D dir] FILE… + + +DESCRIPTION +cntraining takes a list of .tr files, from which it generates the +normproto data file (the character normalization sensitivity +prototypes). + + +OPTIONS + + + +-D dir + + + + Directory to write output files to. + + + + + + +SEE ALSO +tesseract(1), shapeclustering(1), mftraining(1) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +COPYING +Copyright (c) Hewlett-Packard Company, 1988 +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/combine_tessdata.1.asc b/doc/combine_tessdata.1.asc index d93de7ea0f..7b5295f227 100644 --- a/doc/combine_tessdata.1.asc +++ b/doc/combine_tessdata.1.asc @@ -11,7 +11,7 @@ SYNOPSIS DESCRIPTION ----------- -combine_tessdata(1) is the main program to combine/extract/overwrite +combine_tessdata(1) is the main program to combine/extract/overwrite tessdata components in [lang].traineddata files. To combine all the individual tessdata components (unicharset, DAWGs, diff --git a/doc/combine_tessdata.1.html b/doc/combine_tessdata.1.html index 8de474b33b..a7f699f939 100644 --- a/doc/combine_tessdata.1.html +++ b/doc/combine_tessdata.1.html @@ -1,1014 +1,1014 @@ - - - - - -COMBINE_TESSDATA(1) - - - - - -
-
-

SYNOPSIS

-
-

combine_tessdata [OPTION] FILE…

-
-
-
-

DESCRIPTION

-
-

combine_tessdata(1) is the main program to combine/extract/overwrite -tessdata components in [lang].traineddata files.

-

To combine all the individual tessdata components (unicharset, DAWGs, -classifier templates, ambiguities, language configs) located at, say, -/home/$USER/temp/eng.* run:

-
-
-
combine_tessdata /home/$USER/temp/eng.
-
-

The result will be a combined tessdata file /home/$USER/temp/eng.traineddata

-

Specify option -e if you would like to extract individual components -from a combined traineddata file. For example, to extract language config -file and the unicharset from tessdata/eng.traineddata run:

-
-
-
combine_tessdata -e tessdata/eng.traineddata \
-  /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset
-
-

The desired config file and unicharset will be written to -/home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset

-

Specify option -o to overwrite individual components of the given -[lang].traineddata file. For example, to overwrite language config -and unichar ambiguities files in tessdata/eng.traineddata use:

-
-
-
combine_tessdata -o tessdata/eng.traineddata \
-  /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharambigs
-
-

As a result, tessdata/eng.traineddata will contain the new language config -and unichar ambigs, plus all the original DAWGs, classifier templates, etc.

-

Note: the file names of the files to extract to and to overwrite from should -have the appropriate file suffixes (extensions) indicating their tessdata -component type (.unicharset for the unicharset, .unicharambigs for unichar -ambigs, etc). See k*FileSuffix variable in ccutil/tessdatamanager.h.

-

Specify option -u to unpack all the components to the specified path:

-
-
-
combine_tessdata -u tessdata/eng.traineddata /home/$USER/temp/eng.
-
-

This will create /home/$USER/temp/eng.* files with individual tessdata -components from tessdata/eng.traineddata.

-
-
-
-

OPTIONS

-
-

-e .traineddata FILE…: - Extracts the specified components from the .traineddata file

-

-o .traineddata FILE…: - Overwrites the specified components of the .traineddata file - with those provided on the comand line.

-

-u .traineddata PATHPREFIX - Unpacks the .traineddata using the provided prefix.

-
-
-
-

CAVEATS

-
-

Prefix refers to the full file prefix, including period (.)

-
-
-
-

COMPONENTS

-
-

The components in a Tesseract lang.traineddata file as of -Tesseract 3.02 are briefly described below; For more information on -many of these files, see -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-lang.config -
-
-

- (Optional) Language-specific overrides to default config variables. -

-
-
-lang.unicharset -
-
-

- (Required) The list of symbols that Tesseract recognizes, with properties. - See unicharset(5). -

-
-
-lang.unicharambigs -
-
-

- (Optional) This file contains information on pairs of recognized symbols - which are often confused. For example, rn and m. -

-
-
-lang.inttemp -
-
-

- (Required) Character shape templates for each unichar. Produced by - mftraining(1). -

-
-
-lang.pffmtable -
-
-

- (Required) The number of features expected for each unichar. - Produced by mftraining(1) from .tr files. -

-
-
-lang.normproto -
-
-

- (Required) Character normalization prototypes generated by cntraining(1) - from .tr files. -

-
-
-lang.punc-dawg -
-
-

- (Optional) A dawg made from punctuation patterns found around words. - The "word" part is replaced by a single space. -

-
-
-lang.word-dawg -
-
-

- (Optional) A dawg made from dictionary words from the language. -

-
-
-lang.number-dawg -
-
-

- (Optional) A dawg made from tokens which originally contained digits. - Each digit is replaced by a space character. -

-
-
-lang.freq-dawg -
-
-

- (Optional) A dawg made from the most frequent words which would have - gone into word-dawg. -

-
-
-lang.fixed-length-dawgs -
-
-

- (Optional) Several dawgs of different fixed lengths — useful for - languages like Chinese. -

-
-
-lang.cube-unicharset -
-
-

- (Optional) A unicharset for cube, if cube was trained on a different set - of symbols. -

-
-
-lang.cube-word-dawg -
-
-

- (Optional) A word dawg for cube’s alternate unicharset. Not needed if Cube - was trained with Tesseract’s unicharset. -

-
-
-lang.shapetable -
-
-

- (Optional) When present, a shapetable is an extra layer between the character - classifier and the word recognizer that allows the character classifier to - return a collection of unichar ids and fonts instead of a single unichar-id - and font. -

-
-
-lang.bigram-dawg -
-
-

- (Optional) A dawg of word bigrams where the words are separated by a space - and each digit is replaced by a ?. -

-
-
-lang.unambig-dawg -
-
-

- (Optional) TODO: Describe. -

-
-
-lang.params-training-model -
-
-

- (Optional) TODO: Describe. -

-
-
-
-
-
-

HISTORY

-
-

combine_tessdata(1) first appeared in version 3.00 of Tesseract

-
-
-
-

SEE ALSO

-
-

tesseract(1), wordlist2dawg(1), cntraining(1), mftraining(1), unicharset(5), -unicharambigs(5)

-
-
-
-

COPYING

-
-

Copyright (C) 2009, Google Inc. -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +COMBINE_TESSDATA(1) + + + + + +
+
+

SYNOPSIS

+
+

combine_tessdata [OPTION] FILE…

+
+
+
+

DESCRIPTION

+
+

combine_tessdata(1) is the main program to combine/extract/overwrite +tessdata components in [lang].traineddata files.

+

To combine all the individual tessdata components (unicharset, DAWGs, +classifier templates, ambiguities, language configs) located at, say, +/home/$USER/temp/eng.* run:

+
+
+
combine_tessdata /home/$USER/temp/eng.
+
+

The result will be a combined tessdata file /home/$USER/temp/eng.traineddata

+

Specify option -e if you would like to extract individual components +from a combined traineddata file. For example, to extract language config +file and the unicharset from tessdata/eng.traineddata run:

+
+
+
combine_tessdata -e tessdata/eng.traineddata \
+  /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset
+
+

The desired config file and unicharset will be written to +/home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset

+

Specify option -o to overwrite individual components of the given +[lang].traineddata file. For example, to overwrite language config +and unichar ambiguities files in tessdata/eng.traineddata use:

+
+
+
combine_tessdata -o tessdata/eng.traineddata \
+  /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharambigs
+
+

As a result, tessdata/eng.traineddata will contain the new language config +and unichar ambigs, plus all the original DAWGs, classifier templates, etc.

+

Note: the file names of the files to extract to and to overwrite from should +have the appropriate file suffixes (extensions) indicating their tessdata +component type (.unicharset for the unicharset, .unicharambigs for unichar +ambigs, etc). See k*FileSuffix variable in ccutil/tessdatamanager.h.

+

Specify option -u to unpack all the components to the specified path:

+
+
+
combine_tessdata -u tessdata/eng.traineddata /home/$USER/temp/eng.
+
+

This will create /home/$USER/temp/eng.* files with individual tessdata +components from tessdata/eng.traineddata.

+
+
+
+

OPTIONS

+
+

-e .traineddata FILE…: + Extracts the specified components from the .traineddata file

+

-o .traineddata FILE…: + Overwrites the specified components of the .traineddata file + with those provided on the comand line.

+

-u .traineddata PATHPREFIX + Unpacks the .traineddata using the provided prefix.

+
+
+
+

CAVEATS

+
+

Prefix refers to the full file prefix, including period (.)

+
+
+
+

COMPONENTS

+
+

The components in a Tesseract lang.traineddata file as of +Tesseract 3.02 are briefly described below; For more information on +many of these files, see +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+lang.config +
+
+

+ (Optional) Language-specific overrides to default config variables. +

+
+
+lang.unicharset +
+
+

+ (Required) The list of symbols that Tesseract recognizes, with properties. + See unicharset(5). +

+
+
+lang.unicharambigs +
+
+

+ (Optional) This file contains information on pairs of recognized symbols + which are often confused. For example, rn and m. +

+
+
+lang.inttemp +
+
+

+ (Required) Character shape templates for each unichar. Produced by + mftraining(1). +

+
+
+lang.pffmtable +
+
+

+ (Required) The number of features expected for each unichar. + Produced by mftraining(1) from .tr files. +

+
+
+lang.normproto +
+
+

+ (Required) Character normalization prototypes generated by cntraining(1) + from .tr files. +

+
+
+lang.punc-dawg +
+
+

+ (Optional) A dawg made from punctuation patterns found around words. + The "word" part is replaced by a single space. +

+
+
+lang.word-dawg +
+
+

+ (Optional) A dawg made from dictionary words from the language. +

+
+
+lang.number-dawg +
+
+

+ (Optional) A dawg made from tokens which originally contained digits. + Each digit is replaced by a space character. +

+
+
+lang.freq-dawg +
+
+

+ (Optional) A dawg made from the most frequent words which would have + gone into word-dawg. +

+
+
+lang.fixed-length-dawgs +
+
+

+ (Optional) Several dawgs of different fixed lengths — useful for + languages like Chinese. +

+
+
+lang.cube-unicharset +
+
+

+ (Optional) A unicharset for cube, if cube was trained on a different set + of symbols. +

+
+
+lang.cube-word-dawg +
+
+

+ (Optional) A word dawg for cube’s alternate unicharset. Not needed if Cube + was trained with Tesseract’s unicharset. +

+
+
+lang.shapetable +
+
+

+ (Optional) When present, a shapetable is an extra layer between the character + classifier and the word recognizer that allows the character classifier to + return a collection of unichar ids and fonts instead of a single unichar-id + and font. +

+
+
+lang.bigram-dawg +
+
+

+ (Optional) A dawg of word bigrams where the words are separated by a space + and each digit is replaced by a ?. +

+
+
+lang.unambig-dawg +
+
+

+ (Optional) TODO: Describe. +

+
+
+lang.params-training-model +
+
+

+ (Optional) TODO: Describe. +

+
+
+
+
+
+

HISTORY

+
+

combine_tessdata(1) first appeared in version 3.00 of Tesseract

+
+
+
+

SEE ALSO

+
+

tesseract(1), wordlist2dawg(1), cntraining(1), mftraining(1), unicharset(5), +unicharambigs(5)

+
+
+
+

COPYING

+
+

Copyright (C) 2009, Google Inc. +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/combine_tessdata.1.xml b/doc/combine_tessdata.1.xml index 1a43995fb5..693e1343b5 100644 --- a/doc/combine_tessdata.1.xml +++ b/doc/combine_tessdata.1.xml @@ -1,281 +1,281 @@ - - - - - - - COMBINE_TESSDATA(1) - - -combine_tessdata -1 -  -  - - - combine_tessdata - combine/extract/overwrite Tesseract data - - -combine_tessdata [OPTION] FILE… - - -DESCRIPTION -combine_tessdata(1) is the main program to combine/extract/overwrite -tessdata components in [lang].traineddata files. -To combine all the individual tessdata components (unicharset, DAWGs, -classifier templates, ambiguities, language configs) located at, say, -/home/$USER/temp/eng.* run: -combine_tessdata /home/$USER/temp/eng. -The result will be a combined tessdata file /home/$USER/temp/eng.traineddata -Specify option -e if you would like to extract individual components -from a combined traineddata file. For example, to extract language config -file and the unicharset from tessdata/eng.traineddata run: -combine_tessdata -e tessdata/eng.traineddata \ - /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset -The desired config file and unicharset will be written to -/home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset -Specify option -o to overwrite individual components of the given -[lang].traineddata file. For example, to overwrite language config -and unichar ambiguities files in tessdata/eng.traineddata use: -combine_tessdata -o tessdata/eng.traineddata \ - /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharambigs -As a result, tessdata/eng.traineddata will contain the new language config -and unichar ambigs, plus all the original DAWGs, classifier templates, etc. -Note: the file names of the files to extract to and to overwrite from should -have the appropriate file suffixes (extensions) indicating their tessdata -component type (.unicharset for the unicharset, .unicharambigs for unichar -ambigs, etc). See k*FileSuffix variable in ccutil/tessdatamanager.h. -Specify option -u to unpack all the components to the specified path: -combine_tessdata -u tessdata/eng.traineddata /home/$USER/temp/eng. -This will create /home/$USER/temp/eng.* files with individual tessdata -components from tessdata/eng.traineddata. - - -OPTIONS --e .traineddata FILE…: - Extracts the specified components from the .traineddata file --o .traineddata FILE…: - Overwrites the specified components of the .traineddata file - with those provided on the comand line. --u .traineddata PATHPREFIX - Unpacks the .traineddata using the provided prefix. - - -CAVEATS -Prefix refers to the full file prefix, including period (.) - - -COMPONENTS -The components in a Tesseract lang.traineddata file as of -Tesseract 3.02 are briefly described below; For more information on -many of these files, see -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - - -lang.config - - - - (Optional) Language-specific overrides to default config variables. - - - - - -lang.unicharset - - - - (Required) The list of symbols that Tesseract recognizes, with properties. - See unicharset(5). - - - - - -lang.unicharambigs - - - - (Optional) This file contains information on pairs of recognized symbols - which are often confused. For example, rn and m. - - - - - -lang.inttemp - - - - (Required) Character shape templates for each unichar. Produced by - mftraining(1). - - - - - -lang.pffmtable - - - - (Required) The number of features expected for each unichar. - Produced by mftraining(1) from .tr files. - - - - - -lang.normproto - - - - (Required) Character normalization prototypes generated by cntraining(1) - from .tr files. - - - - - -lang.punc-dawg - - - - (Optional) A dawg made from punctuation patterns found around words. - The "word" part is replaced by a single space. - - - - - -lang.word-dawg - - - - (Optional) A dawg made from dictionary words from the language. - - - - - -lang.number-dawg - - - - (Optional) A dawg made from tokens which originally contained digits. - Each digit is replaced by a space character. - - - - - -lang.freq-dawg - - - - (Optional) A dawg made from the most frequent words which would have - gone into word-dawg. - - - - - -lang.fixed-length-dawgs - - - - (Optional) Several dawgs of different fixed lengths — useful for - languages like Chinese. - - - - - -lang.cube-unicharset - - - - (Optional) A unicharset for cube, if cube was trained on a different set - of symbols. - - - - - -lang.cube-word-dawg - - - - (Optional) A word dawg for cube’s alternate unicharset. Not needed if Cube - was trained with Tesseract’s unicharset. - - - - - -lang.shapetable - - - - (Optional) When present, a shapetable is an extra layer between the character - classifier and the word recognizer that allows the character classifier to - return a collection of unichar ids and fonts instead of a single unichar-id - and font. - - - - - -lang.bigram-dawg - - - - (Optional) A dawg of word bigrams where the words are separated by a space - and each digit is replaced by a ?. - - - - - -lang.unambig-dawg - - - - (Optional) TODO: Describe. - - - - - -lang.params-training-model - - - - (Optional) TODO: Describe. - - - - - - -HISTORY -combine_tessdata(1) first appeared in version 3.00 of Tesseract - - -SEE ALSO -tesseract(1), wordlist2dawg(1), cntraining(1), mftraining(1), unicharset(5), -unicharambigs(5) - - -COPYING -Copyright (C) 2009, Google Inc. -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + COMBINE_TESSDATA(1) + + +combine_tessdata +1 +  +  + + + combine_tessdata + combine/extract/overwrite Tesseract data + + +combine_tessdata [OPTION] FILE… + + +DESCRIPTION +combine_tessdata(1) is the main program to combine/extract/overwrite +tessdata components in [lang].traineddata files. +To combine all the individual tessdata components (unicharset, DAWGs, +classifier templates, ambiguities, language configs) located at, say, +/home/$USER/temp/eng.* run: +combine_tessdata /home/$USER/temp/eng. +The result will be a combined tessdata file /home/$USER/temp/eng.traineddata +Specify option -e if you would like to extract individual components +from a combined traineddata file. For example, to extract language config +file and the unicharset from tessdata/eng.traineddata run: +combine_tessdata -e tessdata/eng.traineddata \ + /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset +The desired config file and unicharset will be written to +/home/$USER/temp/eng.config /home/$USER/temp/eng.unicharset +Specify option -o to overwrite individual components of the given +[lang].traineddata file. For example, to overwrite language config +and unichar ambiguities files in tessdata/eng.traineddata use: +combine_tessdata -o tessdata/eng.traineddata \ + /home/$USER/temp/eng.config /home/$USER/temp/eng.unicharambigs +As a result, tessdata/eng.traineddata will contain the new language config +and unichar ambigs, plus all the original DAWGs, classifier templates, etc. +Note: the file names of the files to extract to and to overwrite from should +have the appropriate file suffixes (extensions) indicating their tessdata +component type (.unicharset for the unicharset, .unicharambigs for unichar +ambigs, etc). See k*FileSuffix variable in ccutil/tessdatamanager.h. +Specify option -u to unpack all the components to the specified path: +combine_tessdata -u tessdata/eng.traineddata /home/$USER/temp/eng. +This will create /home/$USER/temp/eng.* files with individual tessdata +components from tessdata/eng.traineddata. + + +OPTIONS +-e .traineddata FILE…: + Extracts the specified components from the .traineddata file +-o .traineddata FILE…: + Overwrites the specified components of the .traineddata file + with those provided on the comand line. +-u .traineddata PATHPREFIX + Unpacks the .traineddata using the provided prefix. + + +CAVEATS +Prefix refers to the full file prefix, including period (.) + + +COMPONENTS +The components in a Tesseract lang.traineddata file as of +Tesseract 3.02 are briefly described below; For more information on +many of these files, see +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + + +lang.config + + + + (Optional) Language-specific overrides to default config variables. + + + + + +lang.unicharset + + + + (Required) The list of symbols that Tesseract recognizes, with properties. + See unicharset(5). + + + + + +lang.unicharambigs + + + + (Optional) This file contains information on pairs of recognized symbols + which are often confused. For example, rn and m. + + + + + +lang.inttemp + + + + (Required) Character shape templates for each unichar. Produced by + mftraining(1). + + + + + +lang.pffmtable + + + + (Required) The number of features expected for each unichar. + Produced by mftraining(1) from .tr files. + + + + + +lang.normproto + + + + (Required) Character normalization prototypes generated by cntraining(1) + from .tr files. + + + + + +lang.punc-dawg + + + + (Optional) A dawg made from punctuation patterns found around words. + The "word" part is replaced by a single space. + + + + + +lang.word-dawg + + + + (Optional) A dawg made from dictionary words from the language. + + + + + +lang.number-dawg + + + + (Optional) A dawg made from tokens which originally contained digits. + Each digit is replaced by a space character. + + + + + +lang.freq-dawg + + + + (Optional) A dawg made from the most frequent words which would have + gone into word-dawg. + + + + + +lang.fixed-length-dawgs + + + + (Optional) Several dawgs of different fixed lengths — useful for + languages like Chinese. + + + + + +lang.cube-unicharset + + + + (Optional) A unicharset for cube, if cube was trained on a different set + of symbols. + + + + + +lang.cube-word-dawg + + + + (Optional) A word dawg for cube’s alternate unicharset. Not needed if Cube + was trained with Tesseract’s unicharset. + + + + + +lang.shapetable + + + + (Optional) When present, a shapetable is an extra layer between the character + classifier and the word recognizer that allows the character classifier to + return a collection of unichar ids and fonts instead of a single unichar-id + and font. + + + + + +lang.bigram-dawg + + + + (Optional) A dawg of word bigrams where the words are separated by a space + and each digit is replaced by a ?. + + + + + +lang.unambig-dawg + + + + (Optional) TODO: Describe. + + + + + +lang.params-training-model + + + + (Optional) TODO: Describe. + + + + + + +HISTORY +combine_tessdata(1) first appeared in version 3.00 of Tesseract + + +SEE ALSO +tesseract(1), wordlist2dawg(1), cntraining(1), mftraining(1), unicharset(5), +unicharambigs(5) + + +COPYING +Copyright (C) 2009, Google Inc. +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/dawg2wordlist.1.html b/doc/dawg2wordlist.1.html index b700fe186d..0b2645dfb7 100644 --- a/doc/dawg2wordlist.1.html +++ b/doc/dawg2wordlist.1.html @@ -1,802 +1,802 @@ - - - - - -DAWG2WORDLIST(1) - - - - - -
-
-

SYNOPSIS

-
-

dawg2wordlist UNICHARSET DAWG WORDLIST

-
-
-
-

DESCRIPTION

-
-

dawg2wordlist(1) converts a Tesseract Directed Acyclic Word -Graph (DAWG) to a list of words using a unicharset as key.

-
-
-
-

OPTIONS

-
-

UNICHARSET - The unicharset of the language. This is the unicharset - generated by mftraining(1).

-

DAWG - The input DAWG, created by wordlist2dawg(1)

-

WORDLIST - Plain text (output) file in UTF-8, one word per line

-
-
-
-

SEE ALSO

-
-

tesseract(1), mftraining(1), wordlist2dawg(1), unicharset(5), -combine_tessdata(1)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

COPYING

-
-

Copyright (C) 2012 Google, Inc. -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +DAWG2WORDLIST(1) + + + + + +
+
+

SYNOPSIS

+
+

dawg2wordlist UNICHARSET DAWG WORDLIST

+
+
+
+

DESCRIPTION

+
+

dawg2wordlist(1) converts a Tesseract Directed Acyclic Word +Graph (DAWG) to a list of words using a unicharset as key.

+
+
+
+

OPTIONS

+
+

UNICHARSET + The unicharset of the language. This is the unicharset + generated by mftraining(1).

+

DAWG + The input DAWG, created by wordlist2dawg(1)

+

WORDLIST + Plain text (output) file in UTF-8, one word per line

+
+
+
+

SEE ALSO

+
+

tesseract(1), mftraining(1), wordlist2dawg(1), unicharset(5), +combine_tessdata(1)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

COPYING

+
+

Copyright (C) 2012 Google, Inc. +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/dawg2wordlist.1.xml b/doc/dawg2wordlist.1.xml index c73113191c..ee960ad9fc 100644 --- a/doc/dawg2wordlist.1.xml +++ b/doc/dawg2wordlist.1.xml @@ -1,53 +1,53 @@ - - - - - - - DAWG2WORDLIST(1) - - -dawg2wordlist -1 -  -  - - - dawg2wordlist - convert a Tesseract DAWG to a wordlist - - -dawg2wordlist UNICHARSET DAWG WORDLIST - - -DESCRIPTION -dawg2wordlist(1) converts a Tesseract Directed Acyclic Word -Graph (DAWG) to a list of words using a unicharset as key. - - -OPTIONS -UNICHARSET - The unicharset of the language. This is the unicharset - generated by mftraining(1). -DAWG - The input DAWG, created by wordlist2dawg(1) -WORDLIST - Plain text (output) file in UTF-8, one word per line - - -SEE ALSO -tesseract(1), mftraining(1), wordlist2dawg(1), unicharset(5), -combine_tessdata(1) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -COPYING -Copyright (C) 2012 Google, Inc. -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + DAWG2WORDLIST(1) + + +dawg2wordlist +1 +  +  + + + dawg2wordlist + convert a Tesseract DAWG to a wordlist + + +dawg2wordlist UNICHARSET DAWG WORDLIST + + +DESCRIPTION +dawg2wordlist(1) converts a Tesseract Directed Acyclic Word +Graph (DAWG) to a list of words using a unicharset as key. + + +OPTIONS +UNICHARSET + The unicharset of the language. This is the unicharset + generated by mftraining(1). +DAWG + The input DAWG, created by wordlist2dawg(1) +WORDLIST + Plain text (output) file in UTF-8, one word per line + + +SEE ALSO +tesseract(1), mftraining(1), wordlist2dawg(1), unicharset(5), +combine_tessdata(1) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +COPYING +Copyright (C) 2012 Google, Inc. +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/mftraining.1.asc b/doc/mftraining.1.asc index 85e1263ade..43fe533a16 100644 --- a/doc/mftraining.1.asc +++ b/doc/mftraining.1.asc @@ -24,12 +24,12 @@ OPTIONS -F 'font_properties_file':: (Input) font properties file, each line is of the following form, where each field other than the font name is 0 or 1: - + *font_name* *italic* *bold* *fixed_pitch* *serif* *fraktur* -X 'xheights_file':: (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] - + *font_name* *xheight* -D 'dir':: diff --git a/doc/mftraining.1.html b/doc/mftraining.1.html index 4abdfd6a6c..41a3804457 100644 --- a/doc/mftraining.1.html +++ b/doc/mftraining.1.html @@ -1,847 +1,847 @@ - - - - - -MFTRAINING(1) - - - - - -
-
-

SYNOPSIS

-
-

mftraining -U unicharset -O lang.unicharset FILE…

-
-
-
-

DESCRIPTION

-
-

mftraining takes a list of .tr files, from which it generates the -files inttemp (the shape prototypes), shapetable, and pffmtable -(the number of expected features for each character). (A fourth file -called Microfeat is also written by this program, but it is not used.)

-
-
-
-

OPTIONS

-
-
-
--U FILE -
-
-

- (Input) The unicharset generated by unicharset_extractor(1) -

-
-
--F font_properties_file -
-
-

- (Input) font properties file, each line is of the following form, where each field other than the font name is 0 or 1: -

-
-
-
*font_name* *italic* *bold* *fixed_pitch* *serif* *fraktur*
-
-
-
--X xheights_file -
-
-

- (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] -

-
-
-
*font_name* *xheight*
-
-
-
--D dir -
-
-

- Directory to write output files to. -

-
-
--O FILE -
-
-

- (Output) The output unicharset that will be given to combine_tessdata(1) -

-
-
-
-
-
-

SEE ALSO

-
-

tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), -shapeclustering(1), unicharset(5)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

COPYING

-
-

Copyright (C) Hewlett-Packard Company, 1988 -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +MFTRAINING(1) + + + + + +
+
+

SYNOPSIS

+
+

mftraining -U unicharset -O lang.unicharset FILE…

+
+
+
+

DESCRIPTION

+
+

mftraining takes a list of .tr files, from which it generates the +files inttemp (the shape prototypes), shapetable, and pffmtable +(the number of expected features for each character). (A fourth file +called Microfeat is also written by this program, but it is not used.)

+
+
+
+

OPTIONS

+
+
+
+-U FILE +
+
+

+ (Input) The unicharset generated by unicharset_extractor(1) +

+
+
+-F font_properties_file +
+
+

+ (Input) font properties file, each line is of the following form, where each field other than the font name is 0 or 1: +

+
+
+
*font_name* *italic* *bold* *fixed_pitch* *serif* *fraktur*
+
+
+
+-X xheights_file +
+
+

+ (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] +

+
+
+
*font_name* *xheight*
+
+
+
+-D dir +
+
+

+ Directory to write output files to. +

+
+
+-O FILE +
+
+

+ (Output) The output unicharset that will be given to combine_tessdata(1) +

+
+
+
+
+
+

SEE ALSO

+
+

tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), +shapeclustering(1), unicharset(5)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

COPYING

+
+

Copyright (C) Hewlett-Packard Company, 1988 +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/mftraining.1.xml b/doc/mftraining.1.xml index 239178a5c1..10b3c6d2e5 100644 --- a/doc/mftraining.1.xml +++ b/doc/mftraining.1.xml @@ -1,102 +1,102 @@ - - - - - - - MFTRAINING(1) - - -mftraining -1 -  -  - - - mftraining - feature training for Tesseract - - -mftraining -U unicharset -O lang.unicharset FILE… - - -DESCRIPTION -mftraining takes a list of .tr files, from which it generates the -files inttemp (the shape prototypes), shapetable, and pffmtable -(the number of expected features for each character). (A fourth file -called Microfeat is also written by this program, but it is not used.) - - -OPTIONS - - - --U FILE - - - - (Input) The unicharset generated by unicharset_extractor(1) - - - - - --F font_properties_file - - - - (Input) font properties file, each line is of the following form, where each field other than the font name is 0 or 1: - -*font_name* *italic* *bold* *fixed_pitch* *serif* *fraktur* - - - - --X xheights_file - - - - (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] - -*font_name* *xheight* - - - - --D dir - - - - Directory to write output files to. - - - - - --O FILE - - - - (Output) The output unicharset that will be given to combine_tessdata(1) - - - - - - -SEE ALSO -tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), -shapeclustering(1), unicharset(5) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -COPYING -Copyright (C) Hewlett-Packard Company, 1988 -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + MFTRAINING(1) + + +mftraining +1 +  +  + + + mftraining + feature training for Tesseract + + +mftraining -U unicharset -O lang.unicharset FILE… + + +DESCRIPTION +mftraining takes a list of .tr files, from which it generates the +files inttemp (the shape prototypes), shapetable, and pffmtable +(the number of expected features for each character). (A fourth file +called Microfeat is also written by this program, but it is not used.) + + +OPTIONS + + + +-U FILE + + + + (Input) The unicharset generated by unicharset_extractor(1) + + + + + +-F font_properties_file + + + + (Input) font properties file, each line is of the following form, where each field other than the font name is 0 or 1: + +*font_name* *italic* *bold* *fixed_pitch* *serif* *fraktur* + + + + +-X xheights_file + + + + (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] + +*font_name* *xheight* + + + + +-D dir + + + + Directory to write output files to. + + + + + +-O FILE + + + + (Output) The output unicharset that will be given to combine_tessdata(1) + + + + + + +SEE ALSO +tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), +shapeclustering(1), unicharset(5) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +COPYING +Copyright (C) Hewlett-Packard Company, 1988 +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/shapeclustering.1.asc b/doc/shapeclustering.1.asc index 81ca0dbc09..0a1bfb035b 100644 --- a/doc/shapeclustering.1.asc +++ b/doc/shapeclustering.1.asc @@ -35,7 +35,7 @@ OPTIONS -X 'xheights_file':: (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] - + 'font_name' 'xheight' -O 'FILE':: diff --git a/doc/shapeclustering.1.html b/doc/shapeclustering.1.html index 845d49a815..5fca944fc8 100644 --- a/doc/shapeclustering.1.html +++ b/doc/shapeclustering.1.html @@ -1,850 +1,850 @@ - - - - - -SHAPECLUSTERING(1) - - - - - -
-
-

SYNOPSIS

-
-

shapeclustering -D output_dir - -U unicharset -O mfunicharset - -F font_props -X xheights - FILE…

-
-
-
-

DESCRIPTION

-
-

shapeclustering(1) takes extracted feature .tr files (generated by -tesseract(1) run in a special mode from box files) and produces a -file shapetable and an enhanced unicharset. This program is still -experimental, and is not required (yet) for training Tesseract.

-
-
-
-

OPTIONS

-
-
-
--U FILE -
-
-

- The unicharset generated by unicharset_extractor(1). -

-
-
--D dir -
-
-

- Directory to write output files to. -

-
-
--F font_properties_file -
-
-

- (Input) font properties file, where each line is of the following form, where each field other than the font name is 0 or 1: -

-
-
-
'font_name' 'italic' 'bold' 'fixed_pitch' 'serif' 'fraktur'
-
-
-
--X xheights_file -
-
-

- (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] -

-
-
-
'font_name' 'xheight'
-
-
-
--O FILE -
-
-

- The output unicharset that will be given to combine_tessdata(1). -

-
-
-
-
-
-

SEE ALSO

-
-

tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), -unicharset(5)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

COPYING

-
-

Copyright (C) Google, 2011 -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +SHAPECLUSTERING(1) + + + + + +
+
+

SYNOPSIS

+
+

shapeclustering -D output_dir + -U unicharset -O mfunicharset + -F font_props -X xheights + FILE…

+
+
+
+

DESCRIPTION

+
+

shapeclustering(1) takes extracted feature .tr files (generated by +tesseract(1) run in a special mode from box files) and produces a +file shapetable and an enhanced unicharset. This program is still +experimental, and is not required (yet) for training Tesseract.

+
+
+
+

OPTIONS

+
+
+
+-U FILE +
+
+

+ The unicharset generated by unicharset_extractor(1). +

+
+
+-D dir +
+
+

+ Directory to write output files to. +

+
+
+-F font_properties_file +
+
+

+ (Input) font properties file, where each line is of the following form, where each field other than the font name is 0 or 1: +

+
+
+
'font_name' 'italic' 'bold' 'fixed_pitch' 'serif' 'fraktur'
+
+
+
+-X xheights_file +
+
+

+ (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] +

+
+
+
'font_name' 'xheight'
+
+
+
+-O FILE +
+
+

+ The output unicharset that will be given to combine_tessdata(1). +

+
+
+
+
+
+

SEE ALSO

+
+

tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), +unicharset(5)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

COPYING

+
+

Copyright (C) Google, 2011 +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/shapeclustering.1.xml b/doc/shapeclustering.1.xml index d02bcf8db9..933789ad3c 100644 --- a/doc/shapeclustering.1.xml +++ b/doc/shapeclustering.1.xml @@ -1,105 +1,105 @@ - - - - - - - SHAPECLUSTERING(1) - - -shapeclustering -1 -  -  - - - shapeclustering - shape clustering training for Tesseract - - -shapeclustering -D output_dir - -U unicharset -O mfunicharset - -F font_props -X xheights - FILE… - - -DESCRIPTION -shapeclustering(1) takes extracted feature .tr files (generated by -tesseract(1) run in a special mode from box files) and produces a -file shapetable and an enhanced unicharset. This program is still -experimental, and is not required (yet) for training Tesseract. - - -OPTIONS - - - --U FILE - - - - The unicharset generated by unicharset_extractor(1). - - - - - --D dir - - - - Directory to write output files to. - - - - - --F font_properties_file - - - - (Input) font properties file, where each line is of the following form, where each field other than the font name is 0 or 1: - -'font_name' 'italic' 'bold' 'fixed_pitch' 'serif' 'fraktur' - - - - --X xheights_file - - - - (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] - -'font_name' 'xheight' - - - - --O FILE - - - - The output unicharset that will be given to combine_tessdata(1). - - - - - - -SEE ALSO -tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), -unicharset(5) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -COPYING -Copyright (C) Google, 2011 -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + SHAPECLUSTERING(1) + + +shapeclustering +1 +  +  + + + shapeclustering + shape clustering training for Tesseract + + +shapeclustering -D output_dir + -U unicharset -O mfunicharset + -F font_props -X xheights + FILE… + + +DESCRIPTION +shapeclustering(1) takes extracted feature .tr files (generated by +tesseract(1) run in a special mode from box files) and produces a +file shapetable and an enhanced unicharset. This program is still +experimental, and is not required (yet) for training Tesseract. + + +OPTIONS + + + +-U FILE + + + + The unicharset generated by unicharset_extractor(1). + + + + + +-D dir + + + + Directory to write output files to. + + + + + +-F font_properties_file + + + + (Input) font properties file, where each line is of the following form, where each field other than the font name is 0 or 1: + +'font_name' 'italic' 'bold' 'fixed_pitch' 'serif' 'fraktur' + + + + +-X xheights_file + + + + (Input) x heights file, each line is of the following form, where xheight is calculated as the pixel x height of a character drawn at 32pt on 300 dpi. [ That is, if base x height + ascenders + descenders = 133, how much is x height? ] + +'font_name' 'xheight' + + + + +-O FILE + + + + The output unicharset that will be given to combine_tessdata(1). + + + + + + +SEE ALSO +tesseract(1), cntraining(1), unicharset_extractor(1), combine_tessdata(1), +unicharset(5) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +COPYING +Copyright (C) Google, 2011 +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/tesseract.1 b/doc/tesseract.1 index 95128fec99..89107f0312 100644 --- a/doc/tesseract.1 +++ b/doc/tesseract.1 @@ -84,7 +84,7 @@ Set value for control parameter\&. Multiple \-c arguments are allowed\&. The language to use\&. If none is specified, English is assumed\&. Multiple languages may be specified, separated by plus characters\&. Tesseract uses 3\-character ISO 639\-2 language codes\&. (See LANGUAGES) .RE .PP -\fI\-psm N\fR +\fI\--psm N\fR .RS 4 Set Tesseract to only run a subset of layout analysis and assume a certain form of image\&. The options for \fBN\fR @@ -139,7 +139,7 @@ pdf \- Output in pdf instead of a text file\&. .RE .RE .sp -\fBNota Bene:\fR The options \fI\-l lang\fR and \fI\-psm N\fR must occur before any \fIconfigfile\fR\&. +\fBNota Bene:\fR The options \fI\-l lang\fR and \fI\--psm N\fR must occur before any \fIconfigfile\fR\&. .SH "SINGLE OPTIONS" .PP \fI\-v\fR diff --git a/doc/tesseract.1.asc b/doc/tesseract.1.asc index d6f34d5060..312aae07f6 100644 --- a/doc/tesseract.1.asc +++ b/doc/tesseract.1.asc @@ -54,7 +54,7 @@ OPTIONS Multiple languages may be specified, separated by plus characters. Tesseract uses 3-character ISO 639-2 language codes. (See LANGUAGES) -'-psm N':: +'--psm N':: Set Tesseract to only run a subset of layout analysis and assume a certain form of image. The options for *N* are: @@ -67,7 +67,7 @@ OPTIONS 6 = Assume a single uniform block of text. 7 = Treat the image as a single text line. 8 = Treat the image as a single word. - 9 = Treat the image as a single word in a circle. + 9 = Treat the image as a single word in a circle. 10 = Treat the image as a single character. 'configfile':: @@ -78,7 +78,7 @@ OPTIONS * hocr - Output in hOCR format instead of as a text file. * pdf - Output in pdf instead of a text file. -*Nota Bene:* The options '-l lang' and '-psm N' must occur +*Nota Bene:* The options '-l lang' and '--psm N' must occur before any 'configfile'. @@ -264,11 +264,11 @@ on read_pattern_list(). HISTORY ------- -The engine was developed at Hewlett Packard Laboratories Bristol and at -Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more -changes made in 1996 to port to Windows, and some C\+\+izing in 1998. A -lot of the code was written in C, and then some more was written in C\+\+. -The C\++ code makes heavy use of a list system using macros. This predates +The engine was developed at Hewlett Packard Laboratories Bristol and at +Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more +changes made in 1996 to port to Windows, and some C\+\+izing in 1998. A +lot of the code was written in C, and then some more was written in C\+\+. +The C\+\+ code makes heavy use of a list system using macros. This predates stl, was portable before stl, and is more efficient than stl lists, but has the big negative that if you do get a segmentation violation, it is hard to debug. @@ -276,18 +276,18 @@ debug. Version 2.00 brought Unicode (UTF-8) support, six languages, and the ability to train Tesseract. -Tesseract was included in UNLV's Fourth Annual Test of OCR Accuracy. +Tesseract was included in UNLV's Fourth Annual Test of OCR Accuracy. See . With Tesseract 2.00, -scripts are now included to allow anyone to reproduce some of these tests. -See for more +scripts are now included to allow anyone to reproduce some of these tests. +See for more details. -Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, -and Korean. It also introduces a new, single-file based system of managing +Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, +and Korean. It also introduces a new, single-file based system of managing language data. -Tesseract 3.02 adds BiDirectional text support, the ability to recognize -multiple languages in a single image, and improved layout analysis. +Tesseract 3.02 adds BiDirectional text support, the ability to recognize +multiple languages in a single image, and improved layout analysis. For further details, see the file ReleaseNotes included with the distribution. diff --git a/doc/tesseract.1.html b/doc/tesseract.1.html index 90c5dae78c..d0addae65b 100644 --- a/doc/tesseract.1.html +++ b/doc/tesseract.1.html @@ -1,1163 +1,1163 @@ - - - - - -TESSERACT(1) - - - - - -
-
-

SYNOPSIS

-
-

tesseract imagename|stdin outputbase|stdout [options…] [configfile…]

-
-
-
-

DESCRIPTION

-
-

tesseract(1) is a commercial quality OCR engine originally developed at HP -between 1985 and 1995. In 1995, this engine was among the top 3 evaluated by -UNLV. It was open-sourced by HP and UNLV in 2005, and has been developed -at Google since then.

-
-
-
-

IN/OUT ARGUMENTS

-
-
-
-imagename -
-
-

- The name of the input image. Most image file formats (anything - readable by Leptonica) are supported. -

-
-
-stdin -
-
-

- Instruction to read data from standard input -

-
-
-outputbase -
-
-

- The basename of the output file (to which the appropriate extension - will be appended). By default the output will be named outbase.txt. -

-
-
-stdout -
-
-

- Instruction to sent output data to standard output -

-
-
-
-
-
-

OPTIONS

-
-
-
---tessdata-dir /path -
-
-

- Specify the location of tessdata path -

-
-
---user-words /path/to/file -
-
-

- Specify the location of user words file -

-
-
---user-patterns /path/to/file specify -
-
-

- The location of user patterns file -

-
-
--c configvar=value -
-
-

- Set value for control parameter. Multiple -c arguments are allowed. -

-
-
--l lang -
-
-

- The language to use. If none is specified, English is assumed. - Multiple languages may be specified, separated by plus characters. - Tesseract uses 3-character ISO 639-2 language codes. (See LANGUAGES) -

-
-
--psm N -
-
-

- Set Tesseract to only run a subset of layout analysis and assume - a certain form of image. The options for N are: -

-
-
-
0 = Orientation and script detection (OSD) only.
-1 = Automatic page segmentation with OSD.
-2 = Automatic page segmentation, but no OSD, or OCR.
-3 = Fully automatic page segmentation, but no OSD. (Default)
-4 = Assume a single column of text of variable sizes.
-5 = Assume a single uniform block of vertically aligned text.
-6 = Assume a single uniform block of text.
-7 = Treat the image as a single text line.
-8 = Treat the image as a single word.
-9 = Treat the image as a single word in a circle.
-10 = Treat the image as a single character.
-
-
-
-configfile -
-
-

- The name of a config to use. A config is a plaintext file which - contains a list of variables and their values, one per line, with a - space separating variable from value. Interesting config files - include:
-

-
    -
  • -

    -hocr - Output in hOCR format instead of as a text file. -

    -
    • -
  • -

    -pdf - Output in pdf instead of a text file. -

    -
    • -
-
-
-

Nota Bene: The options -l lang and -psm N must occur -before any configfile.

-
-
-
-

SINGLE OPTIONS

-
-
-
--v -
-
-

- Returns the current version of the tesseract(1) executable. -

-
-
---list-langs -
-
-

- list available languages for tesseract engine. Can be used with --tessdata-dir. -

-
-
---print-parameters -
-
-

- print tesseract parameters to the stdout. -

-
-
-
-
-
-

LANGUAGES

-
-

There are currently language packs available for the following languages -(in https://github.com/tesseract-ocr/tessdata):

-

afr (Afrikaans) -amh (Amharic) -ara (Arabic) -asm (Assamese) -aze (Azerbaijani) -aze_cyrl (Azerbaijani - Cyrilic) -bel (Belarusian) -ben (Bengali) -bod (Tibetan) -bos (Bosnian) -bul (Bulgarian) -cat (Catalan; Valencian) -ceb (Cebuano) -ces (Czech) -chi_sim (Chinese - Simplified) -chi_tra (Chinese - Traditional) -chr (Cherokee) -cym (Welsh) -dan (Danish) -dan_frak (Danish - Fraktur) -deu (German) -deu_frak (German - Fraktur) -dzo (Dzongkha) -ell (Greek, Modern (1453-)) -eng (English) -enm (English, Middle (1100-1500)) -epo (Esperanto) -equ (Math / equation detection module) -est (Estonian) -eus (Basque) -fas (Persian) -fin (Finnish) -fra (French) -frk (Frankish) -frm (French, Middle (ca.1400-1600)) -gle (Irish) -glg (Galician) -grc (Greek, Ancient (to 1453)) -guj (Gujarati) -hat (Haitian; Haitian Creole) -heb (Hebrew) -hin (Hindi) -hrv (Croatian) -hun (Hungarian) -iku (Inuktitut) -ind (Indonesian) -isl (Icelandic) -ita (Italian) -ita_old (Italian - Old) -jav (Javanese) -jpn (Japanese) -kan (Kannada) -kat (Georgian) -kat_old (Georgian - Old) -kaz (Kazakh) -khm (Central Khmer) -kir (Kirghiz; Kyrgyz) -kor (Korean) -kur (Kurdish) -lao (Lao) -lat (Latin) -lav (Latvian) -lit (Lithuanian) -mal (Malayalam) -mar (Marathi) -mkd (Macedonian) -mlt (Maltese) -msa (Malay) -mya (Burmese) -nep (Nepali) -nld (Dutch; Flemish) -nor (Norwegian) -ori (Oriya) -osd (Orientation and script detection module) -pan (Panjabi; Punjabi) -pol (Polish) -por (Portuguese) -pus (Pushto; Pashto) -ron (Romanian; Moldavian; Moldovan) -rus (Russian) -san (Sanskrit) -sin (Sinhala; Sinhalese) -slk (Slovak) -slk_frak (Slovak - Fraktur) -slv (Slovenian) -spa (Spanish; Castilian) -spa_old (Spanish; Castilian - Old) -sqi (Albanian) -srp (Serbian) -srp_latn (Serbian - Latin) -swa (Swahili) -swe (Swedish) -syr (Syriac) -tam (Tamil) -tel (Telugu) -tgk (Tajik) -tgl (Tagalog) -tha (Thai) -tir (Tigrinya) -tur (Turkish) -uig (Uighur; Uyghur) -ukr (Ukrainian) -urd (Urdu) -uzb (Uzbek) -uzb_cyrl (Uzbek - Cyrilic) -vie (Vietnamese) -yid (Yiddish)

-

To use a non-standard language pack named foo.traineddata, set the -TESSDATA_PREFIX environment variable so the file can be found at -TESSDATA_PREFIX/tessdata/foo.traineddata and give Tesseract the -argument -l foo.

-
-
-
-

CONFIG FILES AND AUGMENTING WITH USER DATA

-
-

Tesseract config files consist of lines with variable-value pairs (space -separated). The variables are documented as flags in the source code like -the following one in tesseractclass.h:

-

STRING_VAR_H(tessedit_char_blacklist, "", - "Blacklist of chars not to recognize");

-

These variables may enable or disable various features of the engine, and -may cause it to load (or not load) various data. For instance, let’s suppose -you want to OCR in English, but suppress the normal dictionary and load an -alternative word list and an alternative list of patterns — these two files -are the most commonly used extra data files.

-

If your language pack is in /path/to/eng.traineddata and the hocr config -is in /path/to/configs/hocr then create three new files:

-

/path/to/eng.user-words:

-
-
the
-quick
-brown
-fox
-jumped
-
-
-

/path/to/eng.user-patterns:

-
-
1-\d\d\d-GOOG-411
-www.\n\\\*.com
-
-
-

/path/to/configs/bazaar:

-
-
load_system_dawg     F
-load_freq_dawg       F
-user_words_suffix    user-words
-user_patterns_suffix user-patterns
-
-
-

Now, if you pass the word bazaar as a trailing command line parameter -to Tesseract, Tesseract will not bother loading the system dictionary nor -the dictionary of frequent words and will load and use the eng.user-words -and eng.user-patterns files you provided. The former is a simple word list, -one per line. The format of the latter is documented in dict/trie.h -on read_pattern_list().

-
-
-
-

HISTORY

-
-

The engine was developed at Hewlett Packard Laboratories Bristol and at -Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more -changes made in 1996 to port to Windows, and some C++izing in 1998. A -lot of the code was written in C, and then some more was written in C++. -The C\++ code makes heavy use of a list system using macros. This predates -stl, was portable before stl, and is more efficient than stl lists, but has -the big negative that if you do get a segmentation violation, it is hard to -debug.

-

Version 2.00 brought Unicode (UTF-8) support, six languages, and the ability -to train Tesseract.

-

Tesseract was included in UNLV’s Fourth Annual Test of OCR Accuracy. -See https://github.com/tesseract-ocr/docs/blob/master/AT-1995.pdf. With Tesseract 2.00, -scripts are now included to allow anyone to reproduce some of these tests. -See https://github.com/tesseract-ocr/tesseract/wiki/TestingTesseract for more -details.

-

Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, -and Korean. It also introduces a new, single-file based system of managing -language data.

-

Tesseract 3.02 adds BiDirectional text support, the ability to recognize -multiple languages in a single image, and improved layout analysis.

-

For further details, see the file ReleaseNotes included with the distribution.

-
-
-
-

RESOURCES

-
-

Main web site: https://github.com/tesseract-ocr
-Information on training: https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

SEE ALSO

-
-

ambiguous_words(1), cntraining(1), combine_tessdata(1), dawg2wordlist(1), -shape_training(1), mftraining(1), unicharambigs(5), unicharset(5), -unicharset_extractor(1), wordlist2dawg(1)

-
-
-
-

AUTHOR

-
-

Tesseract development was led at Hewlett-Packard and Google by Ray Smith. -The development team has included:

-

Ahmad Abdulkader, Chris Newton, Dan Johnson, Dar-Shyang Lee, David Eger, -Eric Wiseblatt, Faisal Shafait, Hiroshi Takenaka, Joe Liu, Joern Wanke, -Mark Seaman, Mickey Namiki, Nicholas Beato, Oded Fuhrmann, Phil Cheatle, -Pingping Xiu, Pong Eksombatchai (Chantat), Ranjith Unnikrishnan, Raquel -Romano, Ray Smith, Rika Antonova, Robert Moss, Samuel Charron, Sheelagh -Lloyd, Shobhit Saxena, and Thomas Kielbus.

-
-
-
-

COPYING

-
-

Licensed under the Apache License, Version 2.0

-
-
-
-
- - - + + + + + +TESSERACT(1) + + + + + +
+
+

SYNOPSIS

+
+

tesseract imagename|stdin outputbase|stdout [options…] [configfile…]

+
+
+
+

DESCRIPTION

+
+

tesseract(1) is a commercial quality OCR engine originally developed at HP +between 1985 and 1995. In 1995, this engine was among the top 3 evaluated by +UNLV. It was open-sourced by HP and UNLV in 2005, and has been developed +at Google since then.

+
+
+
+

IN/OUT ARGUMENTS

+
+
+
+imagename +
+
+

+ The name of the input image. Most image file formats (anything + readable by Leptonica) are supported. +

+
+
+stdin +
+
+

+ Instruction to read data from standard input +

+
+
+outputbase +
+
+

+ The basename of the output file (to which the appropriate extension + will be appended). By default the output will be named outbase.txt. +

+
+
+stdout +
+
+

+ Instruction to sent output data to standard output +

+
+
+
+
+
+

OPTIONS

+
+
+
+--tessdata-dir /path +
+
+

+ Specify the location of tessdata path +

+
+
+--user-words /path/to/file +
+
+

+ Specify the location of user words file +

+
+
+--user-patterns /path/to/file specify +
+
+

+ The location of user patterns file +

+
+
+-c configvar=value +
+
+

+ Set value for control parameter. Multiple -c arguments are allowed. +

+
+
+-l lang +
+
+

+ The language to use. If none is specified, English is assumed. + Multiple languages may be specified, separated by plus characters. + Tesseract uses 3-character ISO 639-2 language codes. (See LANGUAGES) +

+
+
+--psm N +
+
+

+ Set Tesseract to only run a subset of layout analysis and assume + a certain form of image. The options for N are: +

+
+
+
0 = Orientation and script detection (OSD) only.
+1 = Automatic page segmentation with OSD.
+2 = Automatic page segmentation, but no OSD, or OCR.
+3 = Fully automatic page segmentation, but no OSD. (Default)
+4 = Assume a single column of text of variable sizes.
+5 = Assume a single uniform block of vertically aligned text.
+6 = Assume a single uniform block of text.
+7 = Treat the image as a single text line.
+8 = Treat the image as a single word.
+9 = Treat the image as a single word in a circle.
+10 = Treat the image as a single character.
+
+
+
+configfile +
+
+

+ The name of a config to use. A config is a plaintext file which + contains a list of variables and their values, one per line, with a + space separating variable from value. Interesting config files + include:
+

+
    +
  • +

    +hocr - Output in hOCR format instead of as a text file. +

    +
    • +
  • +

    +pdf - Output in pdf instead of a text file. +

    +
    • +
+
+
+

Nota Bene: The options -l lang and --psm N must occur +before any configfile.

+
+
+
+

SINGLE OPTIONS

+
+
+
+-v +
+
+

+ Returns the current version of the tesseract(1) executable. +

+
+
+--list-langs +
+
+

+ list available languages for tesseract engine. Can be used with --tessdata-dir. +

+
+
+--print-parameters +
+
+

+ print tesseract parameters to the stdout. +

+
+
+
+
+
+

LANGUAGES

+
+

There are currently language packs available for the following languages +(in https://github.com/tesseract-ocr/tessdata):

+

afr (Afrikaans) +amh (Amharic) +ara (Arabic) +asm (Assamese) +aze (Azerbaijani) +aze_cyrl (Azerbaijani - Cyrilic) +bel (Belarusian) +ben (Bengali) +bod (Tibetan) +bos (Bosnian) +bul (Bulgarian) +cat (Catalan; Valencian) +ceb (Cebuano) +ces (Czech) +chi_sim (Chinese - Simplified) +chi_tra (Chinese - Traditional) +chr (Cherokee) +cym (Welsh) +dan (Danish) +dan_frak (Danish - Fraktur) +deu (German) +deu_frak (German - Fraktur) +dzo (Dzongkha) +ell (Greek, Modern (1453-)) +eng (English) +enm (English, Middle (1100-1500)) +epo (Esperanto) +equ (Math / equation detection module) +est (Estonian) +eus (Basque) +fas (Persian) +fin (Finnish) +fra (French) +frk (Frankish) +frm (French, Middle (ca.1400-1600)) +gle (Irish) +glg (Galician) +grc (Greek, Ancient (to 1453)) +guj (Gujarati) +hat (Haitian; Haitian Creole) +heb (Hebrew) +hin (Hindi) +hrv (Croatian) +hun (Hungarian) +iku (Inuktitut) +ind (Indonesian) +isl (Icelandic) +ita (Italian) +ita_old (Italian - Old) +jav (Javanese) +jpn (Japanese) +kan (Kannada) +kat (Georgian) +kat_old (Georgian - Old) +kaz (Kazakh) +khm (Central Khmer) +kir (Kirghiz; Kyrgyz) +kor (Korean) +kur (Kurdish) +lao (Lao) +lat (Latin) +lav (Latvian) +lit (Lithuanian) +mal (Malayalam) +mar (Marathi) +mkd (Macedonian) +mlt (Maltese) +msa (Malay) +mya (Burmese) +nep (Nepali) +nld (Dutch; Flemish) +nor (Norwegian) +ori (Oriya) +osd (Orientation and script detection module) +pan (Panjabi; Punjabi) +pol (Polish) +por (Portuguese) +pus (Pushto; Pashto) +ron (Romanian; Moldavian; Moldovan) +rus (Russian) +san (Sanskrit) +sin (Sinhala; Sinhalese) +slk (Slovak) +slk_frak (Slovak - Fraktur) +slv (Slovenian) +spa (Spanish; Castilian) +spa_old (Spanish; Castilian - Old) +sqi (Albanian) +srp (Serbian) +srp_latn (Serbian - Latin) +swa (Swahili) +swe (Swedish) +syr (Syriac) +tam (Tamil) +tel (Telugu) +tgk (Tajik) +tgl (Tagalog) +tha (Thai) +tir (Tigrinya) +tur (Turkish) +uig (Uighur; Uyghur) +ukr (Ukrainian) +urd (Urdu) +uzb (Uzbek) +uzb_cyrl (Uzbek - Cyrilic) +vie (Vietnamese) +yid (Yiddish)

+

To use a non-standard language pack named foo.traineddata, set the +TESSDATA_PREFIX environment variable so the file can be found at +TESSDATA_PREFIX/tessdata/foo.traineddata and give Tesseract the +argument -l foo.

+
+
+
+

CONFIG FILES AND AUGMENTING WITH USER DATA

+
+

Tesseract config files consist of lines with variable-value pairs (space +separated). The variables are documented as flags in the source code like +the following one in tesseractclass.h:

+

STRING_VAR_H(tessedit_char_blacklist, "", + "Blacklist of chars not to recognize");

+

These variables may enable or disable various features of the engine, and +may cause it to load (or not load) various data. For instance, let’s suppose +you want to OCR in English, but suppress the normal dictionary and load an +alternative word list and an alternative list of patterns — these two files +are the most commonly used extra data files.

+

If your language pack is in /path/to/eng.traineddata and the hocr config +is in /path/to/configs/hocr then create three new files:

+

/path/to/eng.user-words:

+
+
the
+quick
+brown
+fox
+jumped
+
+
+

/path/to/eng.user-patterns:

+
+
1-\d\d\d-GOOG-411
+www.\n\\\*.com
+
+
+

/path/to/configs/bazaar:

+
+
load_system_dawg     F
+load_freq_dawg       F
+user_words_suffix    user-words
+user_patterns_suffix user-patterns
+
+
+

Now, if you pass the word bazaar as a trailing command line parameter +to Tesseract, Tesseract will not bother loading the system dictionary nor +the dictionary of frequent words and will load and use the eng.user-words +and eng.user-patterns files you provided. The former is a simple word list, +one per line. The format of the latter is documented in dict/trie.h +on read_pattern_list().

+
+
+
+

HISTORY

+
+

The engine was developed at Hewlett Packard Laboratories Bristol and at +Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more +changes made in 1996 to port to Windows, and some C++izing in 1998. A +lot of the code was written in C, and then some more was written in C++. +The C\++ code makes heavy use of a list system using macros. This predates +stl, was portable before stl, and is more efficient than stl lists, but has +the big negative that if you do get a segmentation violation, it is hard to +debug.

+

Version 2.00 brought Unicode (UTF-8) support, six languages, and the ability +to train Tesseract.

+

Tesseract was included in UNLV’s Fourth Annual Test of OCR Accuracy. +See https://github.com/tesseract-ocr/docs/blob/master/AT-1995.pdf. With Tesseract 2.00, +scripts are now included to allow anyone to reproduce some of these tests. +See https://github.com/tesseract-ocr/tesseract/wiki/TestingTesseract for more +details.

+

Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, +and Korean. It also introduces a new, single-file based system of managing +language data.

+

Tesseract 3.02 adds BiDirectional text support, the ability to recognize +multiple languages in a single image, and improved layout analysis.

+

For further details, see the file ReleaseNotes included with the distribution.

+
+
+
+

RESOURCES

+
+

Main web site: https://github.com/tesseract-ocr
+Information on training: https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

SEE ALSO

+
+

ambiguous_words(1), cntraining(1), combine_tessdata(1), dawg2wordlist(1), +shape_training(1), mftraining(1), unicharambigs(5), unicharset(5), +unicharset_extractor(1), wordlist2dawg(1)

+
+
+
+

AUTHOR

+
+

Tesseract development was led at Hewlett-Packard and Google by Ray Smith. +The development team has included:

+

Ahmad Abdulkader, Chris Newton, Dan Johnson, Dar-Shyang Lee, David Eger, +Eric Wiseblatt, Faisal Shafait, Hiroshi Takenaka, Joe Liu, Joern Wanke, +Mark Seaman, Mickey Namiki, Nicholas Beato, Oded Fuhrmann, Phil Cheatle, +Pingping Xiu, Pong Eksombatchai (Chantat), Ranjith Unnikrishnan, Raquel +Romano, Ray Smith, Rika Antonova, Robert Moss, Samuel Charron, Sheelagh +Lloyd, Shobhit Saxena, and Thomas Kielbus.

+
+
+
+

COPYING

+
+

Licensed under the Apache License, Version 2.0

+
+
+
+
+ + + diff --git a/doc/tesseract.1.xml b/doc/tesseract.1.xml index 2f971caa7b..8ddce87cd6 100644 --- a/doc/tesseract.1.xml +++ b/doc/tesseract.1.xml @@ -1,424 +1,424 @@ - - - - - - - TESSERACT(1) - - -tesseract -1 -  -  - - - tesseract - command-line OCR engine - - -tesseract imagename|stdin outputbase|stdout [options…] [configfile…] - - -DESCRIPTION -tesseract(1) is a commercial quality OCR engine originally developed at HP -between 1985 and 1995. In 1995, this engine was among the top 3 evaluated by -UNLV. It was open-sourced by HP and UNLV in 2005, and has been developed -at Google since then. - - -IN/OUT ARGUMENTS - - - -imagename - - - - The name of the input image. Most image file formats (anything - readable by Leptonica) are supported. - - - - - -stdin - - - - Instruction to read data from standard input - - - - - -outputbase - - - - The basename of the output file (to which the appropriate extension - will be appended). By default the output will be named outbase.txt. - - - - - -stdout - - - - Instruction to sent output data to standard output - - - - - - -OPTIONS - - - ---tessdata-dir /path - - - - Specify the location of tessdata path - - - - - ---user-words /path/to/file - - - - Specify the location of user words file - - - - - ---user-patterns /path/to/file specify - - - - The location of user patterns file - - - - - --c configvar=value - - - - Set value for control parameter. Multiple -c arguments are allowed. - - - - - --l lang - - - - The language to use. If none is specified, English is assumed. - Multiple languages may be specified, separated by plus characters. - Tesseract uses 3-character ISO 639-2 language codes. (See LANGUAGES) - - - - - --psm N - - - - Set Tesseract to only run a subset of layout analysis and assume - a certain form of image. The options for N are: - -0 = Orientation and script detection (OSD) only. -1 = Automatic page segmentation with OSD. -2 = Automatic page segmentation, but no OSD, or OCR. -3 = Fully automatic page segmentation, but no OSD. (Default) -4 = Assume a single column of text of variable sizes. -5 = Assume a single uniform block of vertically aligned text. -6 = Assume a single uniform block of text. -7 = Treat the image as a single text line. -8 = Treat the image as a single word. -9 = Treat the image as a single word in a circle. -10 = Treat the image as a single character. - - - - -configfile - - - - The name of a config to use. A config is a plaintext file which - contains a list of variables and their values, one per line, with a - space separating variable from value. Interesting config files - include: - - - - -hocr - Output in hOCR format instead of as a text file. - - - - -pdf - Output in pdf instead of a text file. - - - - - - -Nota Bene: The options -l lang and -psm N must occur -before any configfile. - - -SINGLE OPTIONS - - - --v - - - - Returns the current version of the tesseract(1) executable. - - - - - ---list-langs - - - - list available languages for tesseract engine. Can be used with --tessdata-dir. - - - - - ---print-parameters - - - - print tesseract parameters to the stdout. - - - - - - -LANGUAGES -There are currently language packs available for the following languages -(in https://github.com/tesseract-ocr/tessdata): -afr (Afrikaans) -amh (Amharic) -ara (Arabic) -asm (Assamese) -aze (Azerbaijani) -aze_cyrl (Azerbaijani - Cyrilic) -bel (Belarusian) -ben (Bengali) -bod (Tibetan) -bos (Bosnian) -bul (Bulgarian) -cat (Catalan; Valencian) -ceb (Cebuano) -ces (Czech) -chi_sim (Chinese - Simplified) -chi_tra (Chinese - Traditional) -chr (Cherokee) -cym (Welsh) -dan (Danish) -dan_frak (Danish - Fraktur) -deu (German) -deu_frak (German - Fraktur) -dzo (Dzongkha) -ell (Greek, Modern (1453-)) -eng (English) -enm (English, Middle (1100-1500)) -epo (Esperanto) -equ (Math / equation detection module) -est (Estonian) -eus (Basque) -fas (Persian) -fin (Finnish) -fra (French) -frk (Frankish) -frm (French, Middle (ca.1400-1600)) -gle (Irish) -glg (Galician) -grc (Greek, Ancient (to 1453)) -guj (Gujarati) -hat (Haitian; Haitian Creole) -heb (Hebrew) -hin (Hindi) -hrv (Croatian) -hun (Hungarian) -iku (Inuktitut) -ind (Indonesian) -isl (Icelandic) -ita (Italian) -ita_old (Italian - Old) -jav (Javanese) -jpn (Japanese) -kan (Kannada) -kat (Georgian) -kat_old (Georgian - Old) -kaz (Kazakh) -khm (Central Khmer) -kir (Kirghiz; Kyrgyz) -kor (Korean) -kur (Kurdish) -lao (Lao) -lat (Latin) -lav (Latvian) -lit (Lithuanian) -mal (Malayalam) -mar (Marathi) -mkd (Macedonian) -mlt (Maltese) -msa (Malay) -mya (Burmese) -nep (Nepali) -nld (Dutch; Flemish) -nor (Norwegian) -ori (Oriya) -osd (Orientation and script detection module) -pan (Panjabi; Punjabi) -pol (Polish) -por (Portuguese) -pus (Pushto; Pashto) -ron (Romanian; Moldavian; Moldovan) -rus (Russian) -san (Sanskrit) -sin (Sinhala; Sinhalese) -slk (Slovak) -slk_frak (Slovak - Fraktur) -slv (Slovenian) -spa (Spanish; Castilian) -spa_old (Spanish; Castilian - Old) -sqi (Albanian) -srp (Serbian) -srp_latn (Serbian - Latin) -swa (Swahili) -swe (Swedish) -syr (Syriac) -tam (Tamil) -tel (Telugu) -tgk (Tajik) -tgl (Tagalog) -tha (Thai) -tir (Tigrinya) -tur (Turkish) -uig (Uighur; Uyghur) -ukr (Ukrainian) -urd (Urdu) -uzb (Uzbek) -uzb_cyrl (Uzbek - Cyrilic) -vie (Vietnamese) -yid (Yiddish) -To use a non-standard language pack named foo.traineddata, set the -TESSDATA_PREFIX environment variable so the file can be found at -TESSDATA_PREFIX/tessdata/foo.traineddata and give Tesseract the -argument -l foo. - - -CONFIG FILES AND AUGMENTING WITH USER DATA -Tesseract config files consist of lines with variable-value pairs (space -separated). The variables are documented as flags in the source code like -the following one in tesseractclass.h: -STRING_VAR_H(tessedit_char_blacklist, "", - "Blacklist of chars not to recognize"); -These variables may enable or disable various features of the engine, and -may cause it to load (or not load) various data. For instance, let’s suppose -you want to OCR in English, but suppress the normal dictionary and load an -alternative word list and an alternative list of patterns — these two files -are the most commonly used extra data files. -If your language pack is in /path/to/eng.traineddata and the hocr config -is in /path/to/configs/hocr then create three new files: -/path/to/eng.user-words: -
-the -quick -brown -fox -jumped -
-/path/to/eng.user-patterns: -
-1-\d\d\d-GOOG-411 -www.\n\\\*.com -
-/path/to/configs/bazaar: -
-load_system_dawg F -load_freq_dawg F -user_words_suffix user-words -user_patterns_suffix user-patterns -
-Now, if you pass the word bazaar as a trailing command line parameter -to Tesseract, Tesseract will not bother loading the system dictionary nor -the dictionary of frequent words and will load and use the eng.user-words -and eng.user-patterns files you provided. The former is a simple word list, -one per line. The format of the latter is documented in dict/trie.h -on read_pattern_list(). - - -HISTORY -The engine was developed at Hewlett Packard Laboratories Bristol and at -Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more -changes made in 1996 to port to Windows, and some C++izing in 1998. A -lot of the code was written in C, and then some more was written in C++. -The C\++ code makes heavy use of a list system using macros. This predates -stl, was portable before stl, and is more efficient than stl lists, but has -the big negative that if you do get a segmentation violation, it is hard to -debug. -Version 2.00 brought Unicode (UTF-8) support, six languages, and the ability -to train Tesseract. -Tesseract was included in UNLV’s Fourth Annual Test of OCR Accuracy. -See https://github.com/tesseract-ocr/docs/blob/master/AT-1995.pdf. With Tesseract 2.00, -scripts are now included to allow anyone to reproduce some of these tests. -See https://github.com/tesseract-ocr/tesseract/wiki/TestingTesseract for more -details. -Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, -and Korean. It also introduces a new, single-file based system of managing -language data. -Tesseract 3.02 adds BiDirectional text support, the ability to recognize -multiple languages in a single image, and improved layout analysis. -For further details, see the file ReleaseNotes included with the distribution. - - -RESOURCES -Main web site: https://github.com/tesseract-ocr -Information on training: https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -SEE ALSO -ambiguous_words(1), cntraining(1), combine_tessdata(1), dawg2wordlist(1), -shape_training(1), mftraining(1), unicharambigs(5), unicharset(5), -unicharset_extractor(1), wordlist2dawg(1) - - -AUTHOR -Tesseract development was led at Hewlett-Packard and Google by Ray Smith. -The development team has included: -Ahmad Abdulkader, Chris Newton, Dan Johnson, Dar-Shyang Lee, David Eger, -Eric Wiseblatt, Faisal Shafait, Hiroshi Takenaka, Joe Liu, Joern Wanke, -Mark Seaman, Mickey Namiki, Nicholas Beato, Oded Fuhrmann, Phil Cheatle, -Pingping Xiu, Pong Eksombatchai (Chantat), Ranjith Unnikrishnan, Raquel -Romano, Ray Smith, Rika Antonova, Robert Moss, Samuel Charron, Sheelagh -Lloyd, Shobhit Saxena, and Thomas Kielbus. - - -COPYING -Licensed under the Apache License, Version 2.0 - - + + + + + + + TESSERACT(1) + + +tesseract +1 +  +  + + + tesseract + command-line OCR engine + + +tesseract imagename|stdin outputbase|stdout [options…] [configfile…] + + +DESCRIPTION +tesseract(1) is a commercial quality OCR engine originally developed at HP +between 1985 and 1995. In 1995, this engine was among the top 3 evaluated by +UNLV. It was open-sourced by HP and UNLV in 2005, and has been developed +at Google since then. + + +IN/OUT ARGUMENTS + + + +imagename + + + + The name of the input image. Most image file formats (anything + readable by Leptonica) are supported. + + + + + +stdin + + + + Instruction to read data from standard input + + + + + +outputbase + + + + The basename of the output file (to which the appropriate extension + will be appended). By default the output will be named outbase.txt. + + + + + +stdout + + + + Instruction to sent output data to standard output + + + + + + +OPTIONS + + + +--tessdata-dir /path + + + + Specify the location of tessdata path + + + + + +--user-words /path/to/file + + + + Specify the location of user words file + + + + + +--user-patterns /path/to/file specify + + + + The location of user patterns file + + + + + +-c configvar=value + + + + Set value for control parameter. Multiple -c arguments are allowed. + + + + + +-l lang + + + + The language to use. If none is specified, English is assumed. + Multiple languages may be specified, separated by plus characters. + Tesseract uses 3-character ISO 639-2 language codes. (See LANGUAGES) + + + + + +--psm N + + + + Set Tesseract to only run a subset of layout analysis and assume + a certain form of image. The options for N are: + +0 = Orientation and script detection (OSD) only. +1 = Automatic page segmentation with OSD. +2 = Automatic page segmentation, but no OSD, or OCR. +3 = Fully automatic page segmentation, but no OSD. (Default) +4 = Assume a single column of text of variable sizes. +5 = Assume a single uniform block of vertically aligned text. +6 = Assume a single uniform block of text. +7 = Treat the image as a single text line. +8 = Treat the image as a single word. +9 = Treat the image as a single word in a circle. +10 = Treat the image as a single character. + + + + +configfile + + + + The name of a config to use. A config is a plaintext file which + contains a list of variables and their values, one per line, with a + space separating variable from value. Interesting config files + include: + + + + +hocr - Output in hOCR format instead of as a text file. + + + + +pdf - Output in pdf instead of a text file. + + + + + + +Nota Bene: The options -l lang and --psm N must occur +before any configfile. + + +SINGLE OPTIONS + + + +-v + + + + Returns the current version of the tesseract(1) executable. + + + + + +--list-langs + + + + list available languages for tesseract engine. Can be used with --tessdata-dir. + + + + + +--print-parameters + + + + print tesseract parameters to the stdout. + + + + + + +LANGUAGES +There are currently language packs available for the following languages +(in https://github.com/tesseract-ocr/tessdata): +afr (Afrikaans) +amh (Amharic) +ara (Arabic) +asm (Assamese) +aze (Azerbaijani) +aze_cyrl (Azerbaijani - Cyrilic) +bel (Belarusian) +ben (Bengali) +bod (Tibetan) +bos (Bosnian) +bul (Bulgarian) +cat (Catalan; Valencian) +ceb (Cebuano) +ces (Czech) +chi_sim (Chinese - Simplified) +chi_tra (Chinese - Traditional) +chr (Cherokee) +cym (Welsh) +dan (Danish) +dan_frak (Danish - Fraktur) +deu (German) +deu_frak (German - Fraktur) +dzo (Dzongkha) +ell (Greek, Modern (1453-)) +eng (English) +enm (English, Middle (1100-1500)) +epo (Esperanto) +equ (Math / equation detection module) +est (Estonian) +eus (Basque) +fas (Persian) +fin (Finnish) +fra (French) +frk (Frankish) +frm (French, Middle (ca.1400-1600)) +gle (Irish) +glg (Galician) +grc (Greek, Ancient (to 1453)) +guj (Gujarati) +hat (Haitian; Haitian Creole) +heb (Hebrew) +hin (Hindi) +hrv (Croatian) +hun (Hungarian) +iku (Inuktitut) +ind (Indonesian) +isl (Icelandic) +ita (Italian) +ita_old (Italian - Old) +jav (Javanese) +jpn (Japanese) +kan (Kannada) +kat (Georgian) +kat_old (Georgian - Old) +kaz (Kazakh) +khm (Central Khmer) +kir (Kirghiz; Kyrgyz) +kor (Korean) +kur (Kurdish) +lao (Lao) +lat (Latin) +lav (Latvian) +lit (Lithuanian) +mal (Malayalam) +mar (Marathi) +mkd (Macedonian) +mlt (Maltese) +msa (Malay) +mya (Burmese) +nep (Nepali) +nld (Dutch; Flemish) +nor (Norwegian) +ori (Oriya) +osd (Orientation and script detection module) +pan (Panjabi; Punjabi) +pol (Polish) +por (Portuguese) +pus (Pushto; Pashto) +ron (Romanian; Moldavian; Moldovan) +rus (Russian) +san (Sanskrit) +sin (Sinhala; Sinhalese) +slk (Slovak) +slk_frak (Slovak - Fraktur) +slv (Slovenian) +spa (Spanish; Castilian) +spa_old (Spanish; Castilian - Old) +sqi (Albanian) +srp (Serbian) +srp_latn (Serbian - Latin) +swa (Swahili) +swe (Swedish) +syr (Syriac) +tam (Tamil) +tel (Telugu) +tgk (Tajik) +tgl (Tagalog) +tha (Thai) +tir (Tigrinya) +tur (Turkish) +uig (Uighur; Uyghur) +ukr (Ukrainian) +urd (Urdu) +uzb (Uzbek) +uzb_cyrl (Uzbek - Cyrilic) +vie (Vietnamese) +yid (Yiddish) +To use a non-standard language pack named foo.traineddata, set the +TESSDATA_PREFIX environment variable so the file can be found at +TESSDATA_PREFIX/tessdata/foo.traineddata and give Tesseract the +argument -l foo. + + +CONFIG FILES AND AUGMENTING WITH USER DATA +Tesseract config files consist of lines with variable-value pairs (space +separated). The variables are documented as flags in the source code like +the following one in tesseractclass.h: +STRING_VAR_H(tessedit_char_blacklist, "", + "Blacklist of chars not to recognize"); +These variables may enable or disable various features of the engine, and +may cause it to load (or not load) various data. For instance, let’s suppose +you want to OCR in English, but suppress the normal dictionary and load an +alternative word list and an alternative list of patterns — these two files +are the most commonly used extra data files. +If your language pack is in /path/to/eng.traineddata and the hocr config +is in /path/to/configs/hocr then create three new files: +/path/to/eng.user-words: +
+the +quick +brown +fox +jumped +
+/path/to/eng.user-patterns: +
+1-\d\d\d-GOOG-411 +www.\n\\\*.com +
+/path/to/configs/bazaar: +
+load_system_dawg F +load_freq_dawg F +user_words_suffix user-words +user_patterns_suffix user-patterns +
+Now, if you pass the word bazaar as a trailing command line parameter +to Tesseract, Tesseract will not bother loading the system dictionary nor +the dictionary of frequent words and will load and use the eng.user-words +and eng.user-patterns files you provided. The former is a simple word list, +one per line. The format of the latter is documented in dict/trie.h +on read_pattern_list(). + + +HISTORY +The engine was developed at Hewlett Packard Laboratories Bristol and at +Hewlett Packard Co, Greeley Colorado between 1985 and 1994, with some more +changes made in 1996 to port to Windows, and some C++izing in 1998. A +lot of the code was written in C, and then some more was written in C++. +The C\++ code makes heavy use of a list system using macros. This predates +stl, was portable before stl, and is more efficient than stl lists, but has +the big negative that if you do get a segmentation violation, it is hard to +debug. +Version 2.00 brought Unicode (UTF-8) support, six languages, and the ability +to train Tesseract. +Tesseract was included in UNLV’s Fourth Annual Test of OCR Accuracy. +See https://github.com/tesseract-ocr/docs/blob/master/AT-1995.pdf. With Tesseract 2.00, +scripts are now included to allow anyone to reproduce some of these tests. +See https://github.com/tesseract-ocr/tesseract/wiki/TestingTesseract for more +details. +Tesseract 3.00 adds a number of new languages, including Chinese, Japanese, +and Korean. It also introduces a new, single-file based system of managing +language data. +Tesseract 3.02 adds BiDirectional text support, the ability to recognize +multiple languages in a single image, and improved layout analysis. +For further details, see the file ReleaseNotes included with the distribution. + + +RESOURCES +Main web site: https://github.com/tesseract-ocr +Information on training: https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +SEE ALSO +ambiguous_words(1), cntraining(1), combine_tessdata(1), dawg2wordlist(1), +shape_training(1), mftraining(1), unicharambigs(5), unicharset(5), +unicharset_extractor(1), wordlist2dawg(1) + + +AUTHOR +Tesseract development was led at Hewlett-Packard and Google by Ray Smith. +The development team has included: +Ahmad Abdulkader, Chris Newton, Dan Johnson, Dar-Shyang Lee, David Eger, +Eric Wiseblatt, Faisal Shafait, Hiroshi Takenaka, Joe Liu, Joern Wanke, +Mark Seaman, Mickey Namiki, Nicholas Beato, Oded Fuhrmann, Phil Cheatle, +Pingping Xiu, Pong Eksombatchai (Chantat), Ranjith Unnikrishnan, Raquel +Romano, Ray Smith, Rika Antonova, Robert Moss, Samuel Charron, Sheelagh +Lloyd, Shobhit Saxena, and Thomas Kielbus. + + +COPYING +Licensed under the Apache License, Version 2.0 + + diff --git a/doc/unicharambigs.5.asc b/doc/unicharambigs.5.asc index 7ce25e4478..079f6d53de 100644 --- a/doc/unicharambigs.5.asc +++ b/doc/unicharambigs.5.asc @@ -38,7 +38,7 @@ EXAMPLE 3 i i i 1 m 0 ............................... -In this example, all instances of the '2' character sequence '''' will +In this example, all instances of the '2' character sequence '''' will *always* be replaced by the '1' character sequence '"'; a '1' character sequence 'm' *may* be replaced by the '2' character sequence 'rn', and the '3' character sequence *may* be replaced by the '1' character diff --git a/doc/unicharambigs.5.html b/doc/unicharambigs.5.html index c6a645e69c..bb9fb291a3 100644 --- a/doc/unicharambigs.5.html +++ b/doc/unicharambigs.5.html @@ -1,875 +1,875 @@ - - - - - -UNICHARAMBIGS(5) - - - - - -
-
-

DESCRIPTION

-
-

The unicharambigs file (a component of traineddata, see combine_tessdata(1) ) -is used by Tesseract to represent possible ambiguities between characters, -or groups of characters.

-

The file contains a number of lines, laid out as follow:

-
-
-
[num] <TAB> [char(s)] <TAB> [num] <TAB> [char(s)] <TAB> [num]
-
-
- - - - - - - - - - - - - - - - - - - - -
-Field one -
-		 -

-the number of characters contained in field two -

-
-Field two -
-		 -

-the character sequence to be replaced -

-
-Field three -
-		 -

-the number of characters contained in field four -

-
-Field four -
-		 -

-the character sequence used to replace field two -

-
-Field five -
-		 -

-contains either 1 or 0. 1 denotes a mandatory -replacement, 0 denotes an optional replacement. -

-
-

Characters appearing in fields two and four should appear in -unicharset. The numbers in fields one and three refer to the -number of unichars (not bytes).

-
-
-
-

EXAMPLE

-
-
-
-
2       ' '     1       "     1
-1       m       2       r n   0
-3       i i i   1       m     0
-
-

In this example, all instances of the 2 character sequence '' will -always be replaced by the 1 character sequence "; a 1 character -sequence m may be replaced by the 2 character sequence rn, and -the 3 character sequence may be replaced by the 1 character -sequence m.

-
-
-
-

HISTORY

-
-

The unicharambigs file first appeared in Tesseract 3.00; prior to that, a -similar format, called DangAmbigs (dangerous ambiguities) was used: the -format was almost identical, except only mandatory replacements could be -specified, and field 5 was absent.

-
-
-
-

BUGS

-
-

This is a documentation "bug": it’s not currently clear what should be done -in the case of ligatures (such as fi) which may also appear as regular -letters in the unicharset.

-
-
-
-

SEE ALSO

-
-

tesseract(1), unicharset(5)

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +UNICHARAMBIGS(5) + + + + + +
+
+

DESCRIPTION

+
+

The unicharambigs file (a component of traineddata, see combine_tessdata(1) ) +is used by Tesseract to represent possible ambiguities between characters, +or groups of characters.

+

The file contains a number of lines, laid out as follow:

+
+
+
[num] <TAB> [char(s)] <TAB> [num] <TAB> [char(s)] <TAB> [num]
+
+
+ + + + + + + + + + + + + + + + + + + + +
+Field one +
+		 +

+the number of characters contained in field two +

+
+Field two +
+		 +

+the character sequence to be replaced +

+
+Field three +
+		 +

+the number of characters contained in field four +

+
+Field four +
+		 +

+the character sequence used to replace field two +

+
+Field five +
+		 +

+contains either 1 or 0. 1 denotes a mandatory +replacement, 0 denotes an optional replacement. +

+
+

Characters appearing in fields two and four should appear in +unicharset. The numbers in fields one and three refer to the +number of unichars (not bytes).

+
+
+
+

EXAMPLE

+
+
+
+
2       ' '     1       "     1
+1       m       2       r n   0
+3       i i i   1       m     0
+
+

In this example, all instances of the 2 character sequence '' will +always be replaced by the 1 character sequence "; a 1 character +sequence m may be replaced by the 2 character sequence rn, and +the 3 character sequence may be replaced by the 1 character +sequence m.

+
+
+
+

HISTORY

+
+

The unicharambigs file first appeared in Tesseract 3.00; prior to that, a +similar format, called DangAmbigs (dangerous ambiguities) was used: the +format was almost identical, except only mandatory replacements could be +specified, and field 5 was absent.

+
+
+
+

BUGS

+
+

This is a documentation "bug": it’s not currently clear what should be done +in the case of ligatures (such as fi) which may also appear as regular +letters in the unicharset.

+
+
+
+

SEE ALSO

+
+

tesseract(1), unicharset(5)

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/unicharambigs.5.xml b/doc/unicharambigs.5.xml index 75b3c66431..cbc0f50e50 100644 --- a/doc/unicharambigs.5.xml +++ b/doc/unicharambigs.5.xml @@ -1,126 +1,126 @@ - - - - - - - UNICHARAMBIGS(5) - - -unicharambigs -5 -  -  - - - unicharambigs - Tesseract unicharset ambiguities - - -DESCRIPTION -The unicharambigs file (a component of traineddata, see combine_tessdata(1) ) -is used by Tesseract to represent possible ambiguities between characters, -or groups of characters. -The file contains a number of lines, laid out as follow: -[num] <TAB> [char(s)] <TAB> [num] <TAB> [char(s)] <TAB> [num] - - - - -Field one - - - - -the number of characters contained in field two - - - - - - -Field two - - - - -the character sequence to be replaced - - - - - - -Field three - - - - -the number of characters contained in field four - - - - - - -Field four - - - - -the character sequence used to replace field two - - - - - - -Field five - - - - -contains either 1 or 0. 1 denotes a mandatory -replacement, 0 denotes an optional replacement. - - - - -Characters appearing in fields two and four should appear in -unicharset. The numbers in fields one and three refer to the -number of unichars (not bytes). - - -EXAMPLE -2 ' ' 1 " 1 -1 m 2 r n 0 -3 i i i 1 m 0 -In this example, all instances of the 2 character sequence '' will -always be replaced by the 1 character sequence "; a 1 character -sequence m may be replaced by the 2 character sequence rn, and -the 3 character sequence may be replaced by the 1 character -sequence m. - - -HISTORY -The unicharambigs file first appeared in Tesseract 3.00; prior to that, a -similar format, called DangAmbigs (dangerous ambiguities) was used: the -format was almost identical, except only mandatory replacements could be -specified, and field 5 was absent. - - -BUGS -This is a documentation "bug": it’s not currently clear what should be done -in the case of ligatures (such as fi) which may also appear as regular -letters in the unicharset. - - -SEE ALSO -tesseract(1), unicharset(5) - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + UNICHARAMBIGS(5) + + +unicharambigs +5 +  +  + + + unicharambigs + Tesseract unicharset ambiguities + + +DESCRIPTION +The unicharambigs file (a component of traineddata, see combine_tessdata(1) ) +is used by Tesseract to represent possible ambiguities between characters, +or groups of characters. +The file contains a number of lines, laid out as follow: +[num] <TAB> [char(s)] <TAB> [num] <TAB> [char(s)] <TAB> [num] + + + + +Field one + + + + +the number of characters contained in field two + + + + + + +Field two + + + + +the character sequence to be replaced + + + + + + +Field three + + + + +the number of characters contained in field four + + + + + + +Field four + + + + +the character sequence used to replace field two + + + + + + +Field five + + + + +contains either 1 or 0. 1 denotes a mandatory +replacement, 0 denotes an optional replacement. + + + + +Characters appearing in fields two and four should appear in +unicharset. The numbers in fields one and three refer to the +number of unichars (not bytes). + + +EXAMPLE +2 ' ' 1 " 1 +1 m 2 r n 0 +3 i i i 1 m 0 +In this example, all instances of the 2 character sequence '' will +always be replaced by the 1 character sequence "; a 1 character +sequence m may be replaced by the 2 character sequence rn, and +the 3 character sequence may be replaced by the 1 character +sequence m. + + +HISTORY +The unicharambigs file first appeared in Tesseract 3.00; prior to that, a +similar format, called DangAmbigs (dangerous ambiguities) was used: the +format was almost identical, except only mandatory replacements could be +specified, and field 5 was absent. + + +BUGS +This is a documentation "bug": it’s not currently clear what should be done +in the case of ligatures (such as fi) which may also appear as regular +letters in the unicharset. + + +SEE ALSO +tesseract(1), unicharset(5) + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/unicharset.5.html b/doc/unicharset.5.html index 0f16c9e5e5..f3c3e7a9fc 100644 --- a/doc/unicharset.5.html +++ b/doc/unicharset.5.html @@ -1,965 +1,965 @@ - - - - - -UNICHARSET(5) - - - - - -
-
-

DESCRIPTION

-
-

Tesseract’s unicharset file contains information on each symbol -(unichar) the Tesseract OCR engine is trained to recognize.

-

A unicharset file (i.e. eng.unicharset) is distributed as part of a -Tesseract language pack (i.e. eng.traineddata). For information on -extracting the unicharset file, see combine_tessdata(1).

-

The first line of a unicharset file contains the number of unichars in -the file. After this line, each subsequent line provides information for -a single unichar. The first such line contains a placeholder reserved for -the space character. Each unichar is referred to within Tesseract by its -Unichar ID, which is the line number (minus 1) within the unicharset file. -Therefore, space gets unichar 0.

-

Each unichar line in the unicharset file (v2+) may have four space-separated fields:

-
-
-
'character' 'properties' 'script' 'id'
-
-

Starting with Tesseract v3.02, more information may be given for each unichar:

-
-
-
'character' 'properties' 'glyph_metrics' 'script' 'other_case' 'direction' 'mirror' 'normed_form'
-
-

Entries:

-
-
-character -
-
-

-The UTF-8 encoded string to be produced for this unichar. -

-
-
-properties -
-
-

-An integer mask of character properties, one per bit. - From least to most significant bit, these are: isalpha, islower, isupper, - isdigit, ispunctuation. -

-
-
-glyph_metrics -
-
-

-Ten comma-separated integers representing various standards - for where this glyph is to be found within a baseline-normalized coordinate - system where 128 is normalized to x-height. -

-
    -
  • -

    -min_bottom, max_bottom: the ranges where the bottom of the character can - be found. -

    -
    • -
  • -

    -min_top, max_top: the ranges where the top of the character may be found. -

    -
    • -
  • -

    -min_width, max_width: horizontal width of the character. -

    -
    • -
  • -

    -min_bearing, max_bearing: how far from the usual start position does the - leftmost part of the character begin. -

    -
    • -
  • -

    -min_advance, max_advance: how far from the printer’s cell left do we - advance to begin the next character. -

    -
    • -
-
-
-script -
-
-

-Name of the script (Latin, Common, Greek, Cyrillic, Han, null). -

-
-
-other_case -
-
-

-The Unichar ID of the other case version of this character - (upper or lower). -

-
-
-direction -
-
-

-The Unicode BiDi direction of this character, as defined by - ICU’s enum UCharDirection. (0 = Left to Right, 1 = Right to Left, - 2 = European Number…) -

-
-
-mirror -
-
-

-The Unichar ID of the BiDirectional mirror of this character. - For example the mirror of open paren is close paren, but Latin Capital C - has no mirror, so it remains a Latin Capital C. -

-
-
-normed_form -
-
-

-The UTF-8 representation of a "normalized form" of this unichar - for the purpose of blaming a module for errors given ground truth text. - For instance, a left or right single quote may normalize to an ASCII quote. -

-
-
-
-
-
-

EXAMPLE (v2)

-
-
-
-
; 10 Common 46
-b 3 Latin 59
-W 5 Latin 40
-7 8 Common 66
-= 0 Common 93
-
-

";" is a punctuation character. Its properties are thus represented by the -binary number 10000 (10 in hexadecimal).

-

"b" is an alphabetic character and a lower case character. Its properties are -thus represented by the binary number 00011 (3 in hexadecimal).

-

"W" is an alphabetic character and an upper case character. Its properties are -thus represented by the binary number 00101 (5 in hexadecimal).

-

"7" is just a digit. Its properties are thus represented by the binary number -01000 (8 in hexadecimal).

-

"=" is not punctuation nor a digit nor an alphabetic character. Its properties -are thus represented by the binary number 00000 (0 in hexadecimal).

-

Japanese or Chinese alphabetic character properties are represented by the -binary number 00001 (1 in hexadecimal): they are alphabetic, but neither -upper nor lower case.

-
-
-
-

EXAMPLE (v3.02)

-
-
-
-
110
-NULL 0 NULL 0
-N 5 59,68,216,255,87,236,0,27,104,227 Latin 11 0 1 N
-Y 5 59,68,216,255,91,205,0,47,91,223 Latin 33 0 2 Y
-1 8 59,69,203,255,45,128,0,66,74,173 Common 3 2 3 1
-9 8 18,66,203,255,89,156,0,39,104,173 Common 4 2 4 9
-a 3 58,65,186,198,85,164,0,26,97,185 Latin 56 0 5 a
-. . .
-
-
-
-
-

CAVEATS

-
-

Although the unicharset reader maintains the ability to read unicharsets -of older formats and will assign default values to missing fields, -the accuracy will be degraded.

-

Further, most other data files are indexed by the unicharset file, -so changing it without re-generating the others is likely to have dire -consequences.

-
-
-
-

HISTORY

-
-

The unicharset format first appeared with Tesseract 2.00, which was the -first version to support languages other than English. The unicharset file -contained only the first two fields, and the "ispunctuation" property was -absent (punctuation was regarded as "0", as "=" is in the above example.

-
-
-
-

SEE ALSO

-
-

tesseract(1), combine_tessdata(1), unicharset_extractor(1)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +UNICHARSET(5) + + + + + +
+
+

DESCRIPTION

+
+

Tesseract’s unicharset file contains information on each symbol +(unichar) the Tesseract OCR engine is trained to recognize.

+

A unicharset file (i.e. eng.unicharset) is distributed as part of a +Tesseract language pack (i.e. eng.traineddata). For information on +extracting the unicharset file, see combine_tessdata(1).

+

The first line of a unicharset file contains the number of unichars in +the file. After this line, each subsequent line provides information for +a single unichar. The first such line contains a placeholder reserved for +the space character. Each unichar is referred to within Tesseract by its +Unichar ID, which is the line number (minus 1) within the unicharset file. +Therefore, space gets unichar 0.

+

Each unichar line in the unicharset file (v2+) may have four space-separated fields:

+
+
+
'character' 'properties' 'script' 'id'
+
+

Starting with Tesseract v3.02, more information may be given for each unichar:

+
+
+
'character' 'properties' 'glyph_metrics' 'script' 'other_case' 'direction' 'mirror' 'normed_form'
+
+

Entries:

+
+
+character +
+
+

+The UTF-8 encoded string to be produced for this unichar. +

+
+
+properties +
+
+

+An integer mask of character properties, one per bit. + From least to most significant bit, these are: isalpha, islower, isupper, + isdigit, ispunctuation. +

+
+
+glyph_metrics +
+
+

+Ten comma-separated integers representing various standards + for where this glyph is to be found within a baseline-normalized coordinate + system where 128 is normalized to x-height. +

+
    +
  • +

    +min_bottom, max_bottom: the ranges where the bottom of the character can + be found. +

    +
    • +
  • +

    +min_top, max_top: the ranges where the top of the character may be found. +

    +
    • +
  • +

    +min_width, max_width: horizontal width of the character. +

    +
    • +
  • +

    +min_bearing, max_bearing: how far from the usual start position does the + leftmost part of the character begin. +

    +
    • +
  • +

    +min_advance, max_advance: how far from the printer’s cell left do we + advance to begin the next character. +

    +
    • +
+
+
+script +
+
+

+Name of the script (Latin, Common, Greek, Cyrillic, Han, null). +

+
+
+other_case +
+
+

+The Unichar ID of the other case version of this character + (upper or lower). +

+
+
+direction +
+
+

+The Unicode BiDi direction of this character, as defined by + ICU’s enum UCharDirection. (0 = Left to Right, 1 = Right to Left, + 2 = European Number…) +

+
+
+mirror +
+
+

+The Unichar ID of the BiDirectional mirror of this character. + For example the mirror of open paren is close paren, but Latin Capital C + has no mirror, so it remains a Latin Capital C. +

+
+
+normed_form +
+
+

+The UTF-8 representation of a "normalized form" of this unichar + for the purpose of blaming a module for errors given ground truth text. + For instance, a left or right single quote may normalize to an ASCII quote. +

+
+
+
+
+
+

EXAMPLE (v2)

+
+
+
+
; 10 Common 46
+b 3 Latin 59
+W 5 Latin 40
+7 8 Common 66
+= 0 Common 93
+
+

";" is a punctuation character. Its properties are thus represented by the +binary number 10000 (10 in hexadecimal).

+

"b" is an alphabetic character and a lower case character. Its properties are +thus represented by the binary number 00011 (3 in hexadecimal).

+

"W" is an alphabetic character and an upper case character. Its properties are +thus represented by the binary number 00101 (5 in hexadecimal).

+

"7" is just a digit. Its properties are thus represented by the binary number +01000 (8 in hexadecimal).

+

"=" is not punctuation nor a digit nor an alphabetic character. Its properties +are thus represented by the binary number 00000 (0 in hexadecimal).

+

Japanese or Chinese alphabetic character properties are represented by the +binary number 00001 (1 in hexadecimal): they are alphabetic, but neither +upper nor lower case.

+
+
+
+

EXAMPLE (v3.02)

+
+
+
+
110
+NULL 0 NULL 0
+N 5 59,68,216,255,87,236,0,27,104,227 Latin 11 0 1 N
+Y 5 59,68,216,255,91,205,0,47,91,223 Latin 33 0 2 Y
+1 8 59,69,203,255,45,128,0,66,74,173 Common 3 2 3 1
+9 8 18,66,203,255,89,156,0,39,104,173 Common 4 2 4 9
+a 3 58,65,186,198,85,164,0,26,97,185 Latin 56 0 5 a
+. . .
+
+
+
+
+

CAVEATS

+
+

Although the unicharset reader maintains the ability to read unicharsets +of older formats and will assign default values to missing fields, +the accuracy will be degraded.

+

Further, most other data files are indexed by the unicharset file, +so changing it without re-generating the others is likely to have dire +consequences.

+
+
+
+

HISTORY

+
+

The unicharset format first appeared with Tesseract 2.00, which was the +first version to support languages other than English. The unicharset file +contained only the first two fields, and the "ispunctuation" property was +absent (punctuation was regarded as "0", as "=" is in the above example.

+
+
+
+

SEE ALSO

+
+

tesseract(1), combine_tessdata(1), unicharset_extractor(1)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/unicharset.5.xml b/doc/unicharset.5.xml index 9ae6257e60..40e03c6eea 100644 --- a/doc/unicharset.5.xml +++ b/doc/unicharset.5.xml @@ -1,219 +1,219 @@ - - - - - - - UNICHARSET(5) - - -unicharset -5 -  -  - - - unicharset - character properties file used by tesseract(1) - - -DESCRIPTION -Tesseract’s unicharset file contains information on each symbol -(unichar) the Tesseract OCR engine is trained to recognize. -A unicharset file (i.e. eng.unicharset) is distributed as part of a -Tesseract language pack (i.e. eng.traineddata). For information on -extracting the unicharset file, see combine_tessdata(1). -The first line of a unicharset file contains the number of unichars in -the file. After this line, each subsequent line provides information for -a single unichar. The first such line contains a placeholder reserved for -the space character. Each unichar is referred to within Tesseract by its -Unichar ID, which is the line number (minus 1) within the unicharset file. -Therefore, space gets unichar 0. -Each unichar line in the unicharset file (v2+) may have four space-separated fields: -'character' 'properties' 'script' 'id' -Starting with Tesseract v3.02, more information may be given for each unichar: -'character' 'properties' 'glyph_metrics' 'script' 'other_case' 'direction' 'mirror' 'normed_form' -Entries: - - - -character - - - -The UTF-8 encoded string to be produced for this unichar. - - - - - -properties - - - -An integer mask of character properties, one per bit. - From least to most significant bit, these are: isalpha, islower, isupper, - isdigit, ispunctuation. - - - - - -glyph_metrics - - - -Ten comma-separated integers representing various standards - for where this glyph is to be found within a baseline-normalized coordinate - system where 128 is normalized to x-height. - - - - -min_bottom, max_bottom: the ranges where the bottom of the character can - be found. - - - - -min_top, max_top: the ranges where the top of the character may be found. - - - - -min_width, max_width: horizontal width of the character. - - - - -min_bearing, max_bearing: how far from the usual start position does the - leftmost part of the character begin. - - - - -min_advance, max_advance: how far from the printer’s cell left do we - advance to begin the next character. - - - - - - - -script - - - -Name of the script (Latin, Common, Greek, Cyrillic, Han, null). - - - - - -other_case - - - -The Unichar ID of the other case version of this character - (upper or lower). - - - - - -direction - - - -The Unicode BiDi direction of this character, as defined by - ICU’s enum UCharDirection. (0 = Left to Right, 1 = Right to Left, - 2 = European Number…) - - - - - -mirror - - - -The Unichar ID of the BiDirectional mirror of this character. - For example the mirror of open paren is close paren, but Latin Capital C - has no mirror, so it remains a Latin Capital C. - - - - - -normed_form - - - -The UTF-8 representation of a "normalized form" of this unichar - for the purpose of blaming a module for errors given ground truth text. - For instance, a left or right single quote may normalize to an ASCII quote. - - - - - - -EXAMPLE (v2) -; 10 Common 46 -b 3 Latin 59 -W 5 Latin 40 -7 8 Common 66 -= 0 Common 93 -";" is a punctuation character. Its properties are thus represented by the -binary number 10000 (10 in hexadecimal). -"b" is an alphabetic character and a lower case character. Its properties are -thus represented by the binary number 00011 (3 in hexadecimal). -"W" is an alphabetic character and an upper case character. Its properties are -thus represented by the binary number 00101 (5 in hexadecimal). -"7" is just a digit. Its properties are thus represented by the binary number -01000 (8 in hexadecimal). -"=" is not punctuation nor a digit nor an alphabetic character. Its properties -are thus represented by the binary number 00000 (0 in hexadecimal). -Japanese or Chinese alphabetic character properties are represented by the -binary number 00001 (1 in hexadecimal): they are alphabetic, but neither -upper nor lower case. - - -EXAMPLE (v3.02) -110 -NULL 0 NULL 0 -N 5 59,68,216,255,87,236,0,27,104,227 Latin 11 0 1 N -Y 5 59,68,216,255,91,205,0,47,91,223 Latin 33 0 2 Y -1 8 59,69,203,255,45,128,0,66,74,173 Common 3 2 3 1 -9 8 18,66,203,255,89,156,0,39,104,173 Common 4 2 4 9 -a 3 58,65,186,198,85,164,0,26,97,185 Latin 56 0 5 a -. . . - - -CAVEATS -Although the unicharset reader maintains the ability to read unicharsets -of older formats and will assign default values to missing fields, -the accuracy will be degraded. -Further, most other data files are indexed by the unicharset file, -so changing it without re-generating the others is likely to have dire -consequences. - - -HISTORY -The unicharset format first appeared with Tesseract 2.00, which was the -first version to support languages other than English. The unicharset file -contained only the first two fields, and the "ispunctuation" property was -absent (punctuation was regarded as "0", as "=" is in the above example. - - -SEE ALSO -tesseract(1), combine_tessdata(1), unicharset_extractor(1) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + UNICHARSET(5) + + +unicharset +5 +  +  + + + unicharset + character properties file used by tesseract(1) + + +DESCRIPTION +Tesseract’s unicharset file contains information on each symbol +(unichar) the Tesseract OCR engine is trained to recognize. +A unicharset file (i.e. eng.unicharset) is distributed as part of a +Tesseract language pack (i.e. eng.traineddata). For information on +extracting the unicharset file, see combine_tessdata(1). +The first line of a unicharset file contains the number of unichars in +the file. After this line, each subsequent line provides information for +a single unichar. The first such line contains a placeholder reserved for +the space character. Each unichar is referred to within Tesseract by its +Unichar ID, which is the line number (minus 1) within the unicharset file. +Therefore, space gets unichar 0. +Each unichar line in the unicharset file (v2+) may have four space-separated fields: +'character' 'properties' 'script' 'id' +Starting with Tesseract v3.02, more information may be given for each unichar: +'character' 'properties' 'glyph_metrics' 'script' 'other_case' 'direction' 'mirror' 'normed_form' +Entries: + + + +character + + + +The UTF-8 encoded string to be produced for this unichar. + + + + + +properties + + + +An integer mask of character properties, one per bit. + From least to most significant bit, these are: isalpha, islower, isupper, + isdigit, ispunctuation. + + + + + +glyph_metrics + + + +Ten comma-separated integers representing various standards + for where this glyph is to be found within a baseline-normalized coordinate + system where 128 is normalized to x-height. + + + + +min_bottom, max_bottom: the ranges where the bottom of the character can + be found. + + + + +min_top, max_top: the ranges where the top of the character may be found. + + + + +min_width, max_width: horizontal width of the character. + + + + +min_bearing, max_bearing: how far from the usual start position does the + leftmost part of the character begin. + + + + +min_advance, max_advance: how far from the printer’s cell left do we + advance to begin the next character. + + + + + + + +script + + + +Name of the script (Latin, Common, Greek, Cyrillic, Han, null). + + + + + +other_case + + + +The Unichar ID of the other case version of this character + (upper or lower). + + + + + +direction + + + +The Unicode BiDi direction of this character, as defined by + ICU’s enum UCharDirection. (0 = Left to Right, 1 = Right to Left, + 2 = European Number…) + + + + + +mirror + + + +The Unichar ID of the BiDirectional mirror of this character. + For example the mirror of open paren is close paren, but Latin Capital C + has no mirror, so it remains a Latin Capital C. + + + + + +normed_form + + + +The UTF-8 representation of a "normalized form" of this unichar + for the purpose of blaming a module for errors given ground truth text. + For instance, a left or right single quote may normalize to an ASCII quote. + + + + + + +EXAMPLE (v2) +; 10 Common 46 +b 3 Latin 59 +W 5 Latin 40 +7 8 Common 66 += 0 Common 93 +";" is a punctuation character. Its properties are thus represented by the +binary number 10000 (10 in hexadecimal). +"b" is an alphabetic character and a lower case character. Its properties are +thus represented by the binary number 00011 (3 in hexadecimal). +"W" is an alphabetic character and an upper case character. Its properties are +thus represented by the binary number 00101 (5 in hexadecimal). +"7" is just a digit. Its properties are thus represented by the binary number +01000 (8 in hexadecimal). +"=" is not punctuation nor a digit nor an alphabetic character. Its properties +are thus represented by the binary number 00000 (0 in hexadecimal). +Japanese or Chinese alphabetic character properties are represented by the +binary number 00001 (1 in hexadecimal): they are alphabetic, but neither +upper nor lower case. + + +EXAMPLE (v3.02) +110 +NULL 0 NULL 0 +N 5 59,68,216,255,87,236,0,27,104,227 Latin 11 0 1 N +Y 5 59,68,216,255,91,205,0,47,91,223 Latin 33 0 2 Y +1 8 59,69,203,255,45,128,0,66,74,173 Common 3 2 3 1 +9 8 18,66,203,255,89,156,0,39,104,173 Common 4 2 4 9 +a 3 58,65,186,198,85,164,0,26,97,185 Latin 56 0 5 a +. . . + + +CAVEATS +Although the unicharset reader maintains the ability to read unicharsets +of older formats and will assign default values to missing fields, +the accuracy will be degraded. +Further, most other data files are indexed by the unicharset file, +so changing it without re-generating the others is likely to have dire +consequences. + + +HISTORY +The unicharset format first appeared with Tesseract 2.00, which was the +first version to support languages other than English. The unicharset file +contained only the first two fields, and the "ispunctuation" property was +absent (punctuation was regarded as "0", as "=" is in the above example. + + +SEE ALSO +tesseract(1), combine_tessdata(1), unicharset_extractor(1) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/unicharset_extractor.1.asc b/doc/unicharset_extractor.1.asc index c972783a8e..bde21ab3ba 100644 --- a/doc/unicharset_extractor.1.asc +++ b/doc/unicharset_extractor.1.asc @@ -11,9 +11,9 @@ SYNOPSIS DESCRIPTION ----------- -Tesseract needs to know the set of possible characters it can output. -To generate the unicharset data file, use the unicharset_extractor -program on the same training pages bounding box files as used for +Tesseract needs to know the set of possible characters it can output. +To generate the unicharset data file, use the unicharset_extractor +program on the same training pages bounding box files as used for clustering: unicharset_extractor fontfile_1.box fontfile_2.box ... @@ -21,19 +21,19 @@ clustering: The unicharset will be put into the file 'dir/unicharset', or simply './unicharset' if no output directory is provided. -Tesseract also needs to have access to character properties isalpha, -isdigit, isupper, islower, ispunctuation. all of this auxilury data +Tesseract also needs to have access to character properties isalpha, +isdigit, isupper, islower, ispunctuation. all of this auxilury data and more is encoded in this file. (See unicharset(5)) -If your system supports the wctype functions, these values will be set -automatically by unicharset_extractor and there is no need to edit the -unicharset file. On some older systems (eg Windows 95), the unicharset +If your system supports the wctype functions, these values will be set +automatically by unicharset_extractor and there is no need to edit the +unicharset file. On some older systems (eg Windows 95), the unicharset file must be edited by hand to add these property description codes. -*NOTE* The unicharset file must be regenerated whenever inttemp, normproto -and pffmtable are generated (i.e. they must all be recreated when the box -file is changed) as they have to be in sync. This is made easier than in -previous versions by running unicharset_extractor before mftraining and +*NOTE* The unicharset file must be regenerated whenever inttemp, normproto +and pffmtable are generated (i.e. they must all be recreated when the box +file is changed) as they have to be in sync. This is made easier than in +previous versions by running unicharset_extractor before mftraining and cntraining, and giving the unicharset to mftraining. SEE ALSO diff --git a/doc/unicharset_extractor.1.html b/doc/unicharset_extractor.1.html index a6ac9e898b..6fdeb5e953 100644 --- a/doc/unicharset_extractor.1.html +++ b/doc/unicharset_extractor.1.html @@ -1,815 +1,815 @@ - - - - - -UNICHARSET_EXTRACTOR(1) - - - - - -
-
-

SYNOPSIS

-
-

unicharset_extractor [-D dir] FILE…

-
-
-
-

DESCRIPTION

-
-

Tesseract needs to know the set of possible characters it can output. -To generate the unicharset data file, use the unicharset_extractor -program on the same training pages bounding box files as used for -clustering:

-
-
-
unicharset_extractor fontfile_1.box fontfile_2.box ...
-
-

The unicharset will be put into the file dir/unicharset, or simply -./unicharset if no output directory is provided.

-

Tesseract also needs to have access to character properties isalpha, -isdigit, isupper, islower, ispunctuation. all of this auxilury data -and more is encoded in this file. (See unicharset(5))

-

If your system supports the wctype functions, these values will be set -automatically by unicharset_extractor and there is no need to edit the -unicharset file. On some older systems (eg Windows 95), the unicharset -file must be edited by hand to add these property description codes.

-

NOTE The unicharset file must be regenerated whenever inttemp, normproto -and pffmtable are generated (i.e. they must all be recreated when the box -file is changed) as they have to be in sync. This is made easier than in -previous versions by running unicharset_extractor before mftraining and -cntraining, and giving the unicharset to mftraining.

-
-
-
-

SEE ALSO

-
-

tesseract(1), unicharset(5)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

HISTORY

-
-

unicharset_extractor first appeared in Tesseract 2.00.

-
-
-
-

COPYING

-
-

Copyright (C) 2006, Google Inc. -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +UNICHARSET_EXTRACTOR(1) + + + + + +
+
+

SYNOPSIS

+
+

unicharset_extractor [-D dir] FILE…

+
+
+
+

DESCRIPTION

+
+

Tesseract needs to know the set of possible characters it can output. +To generate the unicharset data file, use the unicharset_extractor +program on the same training pages bounding box files as used for +clustering:

+
+
+
unicharset_extractor fontfile_1.box fontfile_2.box ...
+
+

The unicharset will be put into the file dir/unicharset, or simply +./unicharset if no output directory is provided.

+

Tesseract also needs to have access to character properties isalpha, +isdigit, isupper, islower, ispunctuation. all of this auxilury data +and more is encoded in this file. (See unicharset(5))

+

If your system supports the wctype functions, these values will be set +automatically by unicharset_extractor and there is no need to edit the +unicharset file. On some older systems (eg Windows 95), the unicharset +file must be edited by hand to add these property description codes.

+

NOTE The unicharset file must be regenerated whenever inttemp, normproto +and pffmtable are generated (i.e. they must all be recreated when the box +file is changed) as they have to be in sync. This is made easier than in +previous versions by running unicharset_extractor before mftraining and +cntraining, and giving the unicharset to mftraining.

+
+
+
+

SEE ALSO

+
+

tesseract(1), unicharset(5)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

HISTORY

+
+

unicharset_extractor first appeared in Tesseract 2.00.

+
+
+
+

COPYING

+
+

Copyright (C) 2006, Google Inc. +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/unicharset_extractor.1.xml b/doc/unicharset_extractor.1.xml index bea4d1e16e..45087a8c64 100644 --- a/doc/unicharset_extractor.1.xml +++ b/doc/unicharset_extractor.1.xml @@ -1,63 +1,63 @@ - - - - - - - UNICHARSET_EXTRACTOR(1) - - -unicharset_extractor -1 -  -  - - - unicharset_extractor - extract unicharset from Tesseract boxfiles - - -unicharset_extractor [-D dir] FILE… - - -DESCRIPTION -Tesseract needs to know the set of possible characters it can output. -To generate the unicharset data file, use the unicharset_extractor -program on the same training pages bounding box files as used for -clustering: -unicharset_extractor fontfile_1.box fontfile_2.box ... -The unicharset will be put into the file dir/unicharset, or simply -./unicharset if no output directory is provided. -Tesseract also needs to have access to character properties isalpha, -isdigit, isupper, islower, ispunctuation. all of this auxilury data -and more is encoded in this file. (See unicharset(5)) -If your system supports the wctype functions, these values will be set -automatically by unicharset_extractor and there is no need to edit the -unicharset file. On some older systems (eg Windows 95), the unicharset -file must be edited by hand to add these property description codes. -NOTE The unicharset file must be regenerated whenever inttemp, normproto -and pffmtable are generated (i.e. they must all be recreated when the box -file is changed) as they have to be in sync. This is made easier than in -previous versions by running unicharset_extractor before mftraining and -cntraining, and giving the unicharset to mftraining. - - -SEE ALSO -tesseract(1), unicharset(5) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -HISTORY -unicharset_extractor first appeared in Tesseract 2.00. - - -COPYING -Copyright (C) 2006, Google Inc. -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + UNICHARSET_EXTRACTOR(1) + + +unicharset_extractor +1 +  +  + + + unicharset_extractor + extract unicharset from Tesseract boxfiles + + +unicharset_extractor [-D dir] FILE… + + +DESCRIPTION +Tesseract needs to know the set of possible characters it can output. +To generate the unicharset data file, use the unicharset_extractor +program on the same training pages bounding box files as used for +clustering: +unicharset_extractor fontfile_1.box fontfile_2.box ... +The unicharset will be put into the file dir/unicharset, or simply +./unicharset if no output directory is provided. +Tesseract also needs to have access to character properties isalpha, +isdigit, isupper, islower, ispunctuation. all of this auxilury data +and more is encoded in this file. (See unicharset(5)) +If your system supports the wctype functions, these values will be set +automatically by unicharset_extractor and there is no need to edit the +unicharset file. On some older systems (eg Windows 95), the unicharset +file must be edited by hand to add these property description codes. +NOTE The unicharset file must be regenerated whenever inttemp, normproto +and pffmtable are generated (i.e. they must all be recreated when the box +file is changed) as they have to be in sync. This is made easier than in +previous versions by running unicharset_extractor before mftraining and +cntraining, and giving the unicharset to mftraining. + + +SEE ALSO +tesseract(1), unicharset(5) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +HISTORY +unicharset_extractor first appeared in Tesseract 2.00. + + +COPYING +Copyright (C) 2006, Google Inc. +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/doc/wordlist2dawg.1.html b/doc/wordlist2dawg.1.html index 58e5cab4fa..733570511a 100644 --- a/doc/wordlist2dawg.1.html +++ b/doc/wordlist2dawg.1.html @@ -1,820 +1,820 @@ - - - - - -WORDLIST2DAWG(1) - - - - - -
-
-

SYNOPSIS

-
-

wordlist2dawg WORDLIST DAWG lang.unicharset

-

wordlist2dawg -t WORDLIST DAWG lang.unicharset

-

wordlist2dawg -r 1 WORDLIST DAWG lang.unicharset

-

wordlist2dawg -r 2 WORDLIST DAWG lang.unicharset

-

wordlist2dawg -l <short> <long> WORDLIST DAWG lang.unicharset

-
-
-
-

DESCRIPTION

-
-

wordlist2dawg(1) converts a wordlist to a Directed Acyclic Word Graph -(DAWG) for use with Tesseract. A DAWG is a compressed, space and time -efficient representation of a word list.

-
-
-
-

OPTIONS

-
-

-t - Verify that a given dawg file is equivalent to a given wordlist.

-

-r 1 - Reverse a word if it contains an RTL character.

-

-r 2 - Reverse all words.

-

-l <short> <long> - Produce a file with several dawgs in it, one each for words - of length <short>, <short+1>,… <long>

-
-
-
-

ARGUMENTS

-
-

WORDLIST - A plain text file in UTF-8, one word per line.

-

DAWG - The output DAWG to write.

-

lang.unicharset - The unicharset of the language. This is the unicharset - generated by mftraining(1).

-
-
-
-

SEE ALSO

-
-

tesseract(1), combine_tessdata(1), dawg2wordlist(1)

-

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

-
-
-
-

COPYING

-
-

Copyright (C) 2006 Google, Inc. -Licensed under the Apache License, Version 2.0

-
-
-
-

AUTHOR

-
-

The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present).

-
-
-
-
- - - + + + + + +WORDLIST2DAWG(1) + + + + + +
+
+

SYNOPSIS

+
+

wordlist2dawg WORDLIST DAWG lang.unicharset

+

wordlist2dawg -t WORDLIST DAWG lang.unicharset

+

wordlist2dawg -r 1 WORDLIST DAWG lang.unicharset

+

wordlist2dawg -r 2 WORDLIST DAWG lang.unicharset

+

wordlist2dawg -l <short> <long> WORDLIST DAWG lang.unicharset

+
+
+
+

DESCRIPTION

+
+

wordlist2dawg(1) converts a wordlist to a Directed Acyclic Word Graph +(DAWG) for use with Tesseract. A DAWG is a compressed, space and time +efficient representation of a word list.

+
+
+
+

OPTIONS

+
+

-t + Verify that a given dawg file is equivalent to a given wordlist.

+

-r 1 + Reverse a word if it contains an RTL character.

+

-r 2 + Reverse all words.

+

-l <short> <long> + Produce a file with several dawgs in it, one each for words + of length <short>, <short+1>,… <long>

+
+
+
+

ARGUMENTS

+
+

WORDLIST + A plain text file in UTF-8, one word per line.

+

DAWG + The output DAWG to write.

+

lang.unicharset + The unicharset of the language. This is the unicharset + generated by mftraining(1).

+
+
+
+

SEE ALSO

+
+

tesseract(1), combine_tessdata(1), dawg2wordlist(1)

+

https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract

+
+
+
+

COPYING

+
+

Copyright (C) 2006 Google, Inc. +Licensed under the Apache License, Version 2.0

+
+
+
+

AUTHOR

+
+

The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present).

+
+
+
+
+ + + diff --git a/doc/wordlist2dawg.1.xml b/doc/wordlist2dawg.1.xml index 907d3a574d..bad256fe70 100644 --- a/doc/wordlist2dawg.1.xml +++ b/doc/wordlist2dawg.1.xml @@ -1,69 +1,69 @@ - - - - - - - WORDLIST2DAWG(1) - - -wordlist2dawg -1 -  -  - - - wordlist2dawg - convert a wordlist to a DAWG for Tesseract - - -wordlist2dawg WORDLIST DAWG lang.unicharset -wordlist2dawg -t WORDLIST DAWG lang.unicharset -wordlist2dawg -r 1 WORDLIST DAWG lang.unicharset -wordlist2dawg -r 2 WORDLIST DAWG lang.unicharset -wordlist2dawg -l <short> <long> WORDLIST DAWG lang.unicharset - - -DESCRIPTION -wordlist2dawg(1) converts a wordlist to a Directed Acyclic Word Graph -(DAWG) for use with Tesseract. A DAWG is a compressed, space and time -efficient representation of a word list. - - -OPTIONS --t - Verify that a given dawg file is equivalent to a given wordlist. --r 1 - Reverse a word if it contains an RTL character. --r 2 - Reverse all words. --l <short> <long> - Produce a file with several dawgs in it, one each for words - of length <short>, <short+1>,… <long> - - -ARGUMENTS -WORDLIST - A plain text file in UTF-8, one word per line. -DAWG - The output DAWG to write. -lang.unicharset - The unicharset of the language. This is the unicharset - generated by mftraining(1). - - -SEE ALSO -tesseract(1), combine_tessdata(1), dawg2wordlist(1) -https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract - - -COPYING -Copyright (C) 2006 Google, Inc. -Licensed under the Apache License, Version 2.0 - - -AUTHOR -The Tesseract OCR engine was written by Ray Smith and his research groups -at Hewlett Packard (1985-1995) and Google (2006-present). - - + + + + + + + WORDLIST2DAWG(1) + + +wordlist2dawg +1 +  +  + + + wordlist2dawg + convert a wordlist to a DAWG for Tesseract + + +wordlist2dawg WORDLIST DAWG lang.unicharset +wordlist2dawg -t WORDLIST DAWG lang.unicharset +wordlist2dawg -r 1 WORDLIST DAWG lang.unicharset +wordlist2dawg -r 2 WORDLIST DAWG lang.unicharset +wordlist2dawg -l <short> <long> WORDLIST DAWG lang.unicharset + + +DESCRIPTION +wordlist2dawg(1) converts a wordlist to a Directed Acyclic Word Graph +(DAWG) for use with Tesseract. A DAWG is a compressed, space and time +efficient representation of a word list. + + +OPTIONS +-t + Verify that a given dawg file is equivalent to a given wordlist. +-r 1 + Reverse a word if it contains an RTL character. +-r 2 + Reverse all words. +-l <short> <long> + Produce a file with several dawgs in it, one each for words + of length <short>, <short+1>,… <long> + + +ARGUMENTS +WORDLIST + A plain text file in UTF-8, one word per line. +DAWG + The output DAWG to write. +lang.unicharset + The unicharset of the language. This is the unicharset + generated by mftraining(1). + + +SEE ALSO +tesseract(1), combine_tessdata(1), dawg2wordlist(1) +https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract + + +COPYING +Copyright (C) 2006 Google, Inc. +Licensed under the Apache License, Version 2.0 + + +AUTHOR +The Tesseract OCR engine was written by Ray Smith and his research groups +at Hewlett Packard (1985-1995) and Google (2006-present). + + diff --git a/java/Makefile.am b/java/Makefile.am index fddbc6f9ec..af3b1885c3 100644 --- a/java/Makefile.am +++ b/java/Makefile.am @@ -36,19 +36,20 @@ SCROLLVIEW_CLASSES = \ com/google/scrollview/ScrollView.class SCROLLVIEW_LIBS = \ - $(srcdir)/piccolo2d-core-3.0.jar \ - $(srcdir)/piccolo2d-extras-3.0.jar + piccolo2d-core-3.0.jar \ + piccolo2d-extras-3.0.jar -CLASSPATH = $(srcdir)/piccolo2d-core-3.0.jar:$(srcdir)/piccolo2d-extras-3.0.jar +CLASSPATH = piccolo2d-core-3.0.jar:piccolo2d-extras-3.0.jar ScrollView.jar : $(SCROLLVIEW_CLASSES) - $(JAR) cfm $@ Manifest.txt com/google/scrollview/*.class \ + $(JAR) cfm $@ $(srcdir)/Manifest.txt com/google/scrollview/*.class \ com/google/scrollview/events/*.class com/google/scrollview/ui/*.class -$(SCROLLVIEW_CLASSES) : $(SCROLLVIEW_FILES) +$(SCROLLVIEW_CLASSES) : $(SCROLLVIEW_FILES) $(SCROLLVIEW_LIBS) $(JAVAC) -encoding UTF8 -sourcepath $(srcdir) -classpath $(CLASSPATH) $(SCROLLVIEW_FILES) -d $(builddir) -fetch-jars : +.PHONY: fetch-jars +fetch-jars $(SCROLLVIEW_LIBS): curl -L http://search.maven.org/remotecontent?filepath=org/piccolo2d/piccolo2d-core/3.0/piccolo2d-core-3.0.jar > piccolo2d-core-3.0.jar curl -L http://search.maven.org/remotecontent?filepath=org/piccolo2d/piccolo2d-extras/3.0/piccolo2d-extras-3.0.jar > piccolo2d-extras-3.0.jar @@ -64,7 +65,7 @@ uninstall: endif clean : - rm -f ScrollView.jar *.class $(srcdir)/*.class + rm -f ScrollView.jar $(SCROLLVIEW_CLASSES) # all-am does nothing, to make the java part optional. all all-am install : diff --git a/neural_networks/runtime/input_file_buffer.cpp b/neural_networks/runtime/input_file_buffer.cpp index c3ca67b604..2ab6d1b341 100644 --- a/neural_networks/runtime/input_file_buffer.cpp +++ b/neural_networks/runtime/input_file_buffer.cpp @@ -4,12 +4,21 @@ // // input_file_buffer.h: Declarations of a class for an object that // represents an input file buffer. +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include #include "input_file_buffer.h" namespace tesseract { -// default and only contsructor +// default and only constructor InputFileBuffer::InputFileBuffer(const string &file_name) : file_name_(file_name) { fp_ = NULL; diff --git a/neural_networks/runtime/input_file_buffer.h b/neural_networks/runtime/input_file_buffer.h index 5aa7465c41..51110c4753 100644 --- a/neural_networks/runtime/input_file_buffer.h +++ b/neural_networks/runtime/input_file_buffer.h @@ -5,6 +5,15 @@ // input_file_buffer.h: Declarations of a class for an object that // represents an input file buffer. // +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef INPUT_FILE_BUFFER_H #define INPUT_FILE_BUFFER_H diff --git a/neural_networks/runtime/neural_net.cpp b/neural_networks/runtime/neural_net.cpp index fd2c65af45..30fa4f5704 100644 --- a/neural_networks/runtime/neural_net.cpp +++ b/neural_networks/runtime/neural_net.cpp @@ -4,6 +4,15 @@ // // neural_net.cpp: Declarations of a class for an object that // represents an arbitrary network of neurons +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. // #include #include @@ -148,9 +157,6 @@ bool NeuralNet::CreateFastNet() { node->fan_in_cnt = neurons_[node_idx].fan_in_cnt(); // allocate memory for fan-in nodes node->inputs = new WeightedNode[node->fan_in_cnt]; - if (node->inputs == NULL) { - return false; - } for (int fan_in = 0; fan_in < node->fan_in_cnt; fan_in++) { // identify fan-in neuron const int id = neurons_[node_idx].fan_in(fan_in)->id(); @@ -213,9 +219,6 @@ NeuralNet *NeuralNet::FromFile(const string file_name) { NeuralNet *NeuralNet::FromInputBuffer(InputFileBuffer *ib) { // create a new net object NeuralNet *net_obj = new NeuralNet(); - if (net_obj == NULL) { - return NULL; - } // load the net if (!net_obj->ReadBinary(ib)) { delete net_obj; diff --git a/neural_networks/runtime/neural_net.h b/neural_networks/runtime/neural_net.h index 91d0d68a24..ff3c35323e 100644 --- a/neural_networks/runtime/neural_net.h +++ b/neural_networks/runtime/neural_net.h @@ -5,6 +5,15 @@ // neural_net.h: Declarations of a class for an object that // represents an arbitrary network of neurons // +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef NEURAL_NET_H #define NEURAL_NET_H @@ -131,9 +140,6 @@ class NeuralNet { } // set the size of the neurons vector neurons_ = new Neuron[neuron_cnt_]; - if (neurons_ == NULL) { - return false; - } // read & validate inputs if (input_buff->Read(&read_val, sizeof(read_val)) != sizeof(read_val)) { return false; diff --git a/neural_networks/runtime/neuron.cpp b/neural_networks/runtime/neuron.cpp index 36309082d9..f1ff7b2316 100644 --- a/neural_networks/runtime/neuron.cpp +++ b/neural_networks/runtime/neuron.cpp @@ -4,6 +4,15 @@ // // neuron.cpp: The implementation of a class for an object // that represents a single neuron in a neural network +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include "neuron.h" #include "input_file_buffer.h" diff --git a/neural_networks/runtime/neuron.h b/neural_networks/runtime/neuron.h index a13d4a2eb7..8021902f3f 100644 --- a/neural_networks/runtime/neuron.h +++ b/neural_networks/runtime/neuron.h @@ -5,6 +5,15 @@ // neuron.h: Declarations of a class for an object that // represents a single neuron in a neural network // +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef NEURON_H #define NEURON_H diff --git a/neural_networks/runtime/sigmoid_table.cpp b/neural_networks/runtime/sigmoid_table.cpp index f170a10844..0be73b6f8f 100644 --- a/neural_networks/runtime/sigmoid_table.cpp +++ b/neural_networks/runtime/sigmoid_table.cpp @@ -3,6 +3,15 @@ // Author: ahmadab@google.com (Ahmad Abdulkader) // // sigmoid_table.cpp: Sigmoid function lookup table +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include "neuron.h" diff --git a/opencl/oclkernels.h b/opencl/oclkernels.h index b3a8316b12..5a34fb7a22 100644 --- a/opencl/oclkernels.h +++ b/opencl/oclkernels.h @@ -1,4 +1,12 @@ - +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef _OCL_KERNEL_H_ #define _OCL_KERNEL_H_ #ifndef USE_EXTERNAL_KERNEL @@ -50,7 +58,7 @@ KERNEL( ) KERNEL( -\n__kernel void pixSubtract(__global int *dword, __global int *sword, +\n__kernel void pixSubtract(__global int *dword, __global int *sword, const int wpl, const int h, __global int *outword) { const unsigned int row = get_global_id(1); @@ -105,15 +113,15 @@ KERNEL( unsigned int prevword, nextword, currword,tempword; unsigned int destword; const int col = pos % wpl; - + //Ignore the execss if (pos >= (wpl * h)) return; - - - currword = *(sword + pos); + + + currword = *(sword + pos); destword = currword; - + //Handle boundary conditions if(col==0) prevword=0; @@ -124,9 +132,9 @@ KERNEL( nextword=0; else nextword = *(sword + pos + 1); - + //Loop unrolled - + //1 bit to left and 1 bit to right //Get the max value on LHS of every pixel tempword = (prevword << (31)) | ((currword >> 1)); @@ -142,10 +150,10 @@ KERNEL( //Get max value on RHS of every pixel tempword = (currword << 2) | (nextword >> (30)); destword |= tempword; - - + + *(dword + pos) = destword; - + }\n ) @@ -200,7 +208,7 @@ KERNEL( unsigned int destword, tempword, lastword, currword; unsigned int lnextword, lprevword, rnextword, rprevword, firstword, secondword; int i, j, siter, eiter; - + //Ignore the execss if (pos >= (wpl*h) || (xn < 1 && xp < 1)) return; @@ -225,7 +233,7 @@ KERNEL( firstword = 0x0; else firstword = *(sword + pos - 1); - + //Get next word if (col == (wpl - 1)) secondword = 0x0; @@ -237,7 +245,7 @@ KERNEL( { //Get the max value on LHS of every pixel tempword = ((i == parbitsxp) && (parbitsxp != parbitsxn)) ? 0x0 : (firstword << (32-i)) | ((currword >> i)); - + destword |= tempword; //Get max value on RHS of every pixel @@ -266,11 +274,11 @@ KERNEL( else firstword = *(sword + row*wpl + siter); - if (eiter >= wpl) + if (eiter >= wpl) lastword = 0x0; else lastword = *(sword + row*wpl + eiter); - + for ( i = 1; i < nwords; i++) { //Gets LHS words @@ -280,14 +288,14 @@ KERNEL( secondword = *(sword + row*wpl + siter + i); lprevword = firstword << (32 - parbitsxn) | secondword >> parbitsxn; - + firstword = secondword; if ((siter + i + 1) < 0) secondword = 0x0; else secondword = *(sword + row*wpl + siter + i + 1); - + lnextword = firstword << (32 - parbitsxn) | secondword >> parbitsxn; //Gets RHS words @@ -295,7 +303,7 @@ KERNEL( firstword = 0x0; else firstword = *(sword + row*wpl + eiter - i); - + rnextword = firstword << parbitsxp | lastword >> (32 - parbitsxp); lastword = firstword; @@ -325,7 +333,7 @@ KERNEL( lastword = firstword; firstword = secondword; } - + *(dword + pos) = destword; }\n ) @@ -342,14 +350,14 @@ KERNEL( unsigned int prevword, nextword, currword,tempword; unsigned int destword; int i; - + //Ignore the execss if (pos >= (wpl * h)) return; - currword = *(sword + pos); + currword = *(sword + pos); destword = currword; - + //Handle boundary conditions if(col==0) prevword=0; @@ -360,7 +368,7 @@ KERNEL( nextword=0; else nextword = *(sword + pos + 1); - + for (i = 1; i <= halfwidth; i++) { //Get the max value on LHS of every pixel @@ -377,7 +385,7 @@ KERNEL( //Get max value on RHS of every pixel tempword = (currword << i) | (nextword >> (32 - i)); - + destword |= tempword; } @@ -397,7 +405,7 @@ KERNEL( unsigned int tempword; unsigned int destword; int i, siter, eiter; - + //Ignore the execss if (row >= h || col >= wpl) return; @@ -427,27 +435,27 @@ KERNEL( unsigned int prevword, nextword, currword,tempword; unsigned int destword; const int col = pos % wpl; - + //Ignore the execss if (pos >= (wpl * h)) return; - - currword = *(sword + pos); + + currword = *(sword + pos); destword = currword; - + //Handle boundary conditions if(col==0) prevword=0xffffffff; else prevword = *(sword + pos - 1); - + if(col==(wpl - 1)) nextword=0xffffffff; else nextword = *(sword + pos + 1); - + //Loop unrolled - + //1 bit to left and 1 bit to right //Get the min value on LHS of every pixel tempword = (prevword << (31)) | ((currword >> 1)); @@ -463,10 +471,10 @@ KERNEL( //Get min value on RHS of every pixel tempword = (currword << 2) | (nextword >> (30)); destword &= tempword; - - + + *(dword + pos) = destword; - + }\n ) @@ -491,7 +499,7 @@ KERNEL( if (row < 2 || row >= (h - 2)) { destword = 0x0; - } + } else { //2 words above @@ -518,7 +526,7 @@ KERNEL( tempword = *(sword + i*wpl + col); destword &= tempword; - if (col == 0) + if (col == 0) { destword &= fwmask; } @@ -534,7 +542,7 @@ KERNEL( ) KERNEL( -\n__kernel void morphoErodeHor(__global int *sword,__global int *dword, const int xp, const int xn, const int wpl, +\n__kernel void morphoErodeHor(__global int *sword,__global int *dword, const int xp, const int xn, const int wpl, const int h, const char isAsymmetric, const int rwmask, const int lwmask) { const int col = get_global_id(0); @@ -569,7 +577,7 @@ KERNEL( firstword = 0xffffffff; else firstword = *(sword + pos - 1); - + //Get next word if (col == (wpl - 1)) secondword = 0xffffffff; @@ -585,7 +593,7 @@ KERNEL( //Get max value on RHS of every pixel tempword = ((i == parbitsxp) && (parbitsxp != parbitsxn)) ? 0xffffffff : (currword << i) | (secondword >> (32 - i)); - + //tempword = (currword << i) | (secondword >> (32 - i)); destword &= tempword; } @@ -614,18 +622,18 @@ KERNEL( *(dword + pos) = destword; return; } - + if (siter < 0) firstword = 0xffffffff; else firstword = *(sword + row*wpl + siter); - if (eiter >= wpl) + if (eiter >= wpl) lastword = 0xffffffff; else lastword = *(sword + row*wpl + eiter); - - + + for ( i = 1; i < nwords; i++) { //Gets LHS words @@ -635,14 +643,14 @@ KERNEL( secondword = *(sword + row*wpl + siter + i); lprevword = firstword << (32 - parbitsxp) | secondword >> (parbitsxp); - + firstword = secondword; if ((siter + i + 1) < 0) secondword = 0xffffffff; else secondword = *(sword + row*wpl + siter + i + 1); - + lnextword = firstword << (32 - parbitsxp) | secondword >> (parbitsxp); //Gets RHS words @@ -650,7 +658,7 @@ KERNEL( firstword = 0xffffffff; else firstword = *(sword + row*wpl + eiter - i); - + rnextword = firstword << parbitsxn | lastword >> (32 - parbitsxn); lastword = firstword; @@ -680,7 +688,7 @@ KERNEL( lastword = firstword; firstword = secondword; } - + if (isAsymmetric) { //Clear boundary pixels @@ -700,8 +708,8 @@ KERNEL( KERNEL( \n__kernel void morphoErodeHor_32word(__global int *sword,__global int *dword, - const int halfwidth, const int wpl, - const int h, const char clearBoundPixH, + const int halfwidth, const int wpl, + const int h, const char clearBoundPixH, const int rwmask, const int lwmask, const char isEven) { @@ -715,25 +723,25 @@ KERNEL( if (pos >= (wpl * h)) return; - currword = *(sword + pos); + currword = *(sword + pos); destword = currword; - + //Handle boundary conditions if(col==0) prevword=0xffffffff; else prevword = *(sword + pos - 1); - + if(col==(wpl - 1)) nextword=0xffffffff; else nextword = *(sword + pos + 1); - + for (i = 1; i <= halfwidth; i++) { //Get the min value on LHS of every pixel tempword = (prevword << (32-i)) | ((currword >> i)); - + destword &= tempword; //Get min value on RHS of every pixel @@ -751,7 +759,7 @@ KERNEL( if (clearBoundPixH) { - if (col == 0) + if (col == 0) { destword &= rwmask; } @@ -767,7 +775,7 @@ KERNEL( KERNEL( \n__kernel void morphoErodeVer(__global int *sword,__global int *dword, - const int yp, + const int yp, const int wpl, const int h, const char clearBoundPixV, const int yn) { @@ -776,7 +784,7 @@ KERNEL( const unsigned int pos = row * wpl + col; unsigned int tempword, destword; int i, siter, eiter; - + //Ignore the execss if (row >= h || col >= wpl) return; @@ -796,7 +804,7 @@ KERNEL( //Clear boundary pixels if (clearBoundPixV && ((row < yp) || ((h - row) <= yn))) - { + { destword = 0x0; } @@ -884,23 +892,23 @@ KERNEL( \n __global const uchar* data, \n uint numPixels, \n __global uint *histBuffer) { // each wg will write HIST_SIZE*NUM_CHANNELS into this result; cpu will accumulate across wg's -\n +\n \n /* declare variables */ -\n +\n \n // work indices \n size_t groupId = get_group_id(0); \n size_t localId = get_local_id(0); // 0 -> 256-1 \n size_t globalId = get_global_id(0); // 0 -> 8*10*256-1=20480-1 \n uint numThreads = get_global_size(0); -\n +\n \n /* accumulate in global memory */ \n for ( uint pc = get_global_id(0); pc < numPixels; pc += get_global_size(0) ) { \n uchar value = data[ pc ]; \n int idx = value * get_global_size(0) + get_global_id(0); \n histBuffer[ idx ]++; -\n +\n \n } -\n +\n \n } // kernel_HistogramRectAllChannels_Grey ) @@ -993,35 +1001,35 @@ void kernel_HistogramRectOneChannelReduction( KERNEL( // unused - // each work group (x256) handles a histogram bin + // each work group (x256) handles a histogram bin \n __attribute__((reqd_work_group_size(256, 1, 1))) \n __kernel \n void kernel_HistogramRectAllChannelsReduction_Grey( \n int n, // pixel redundancy that needs to be accumulated \n __global uint *histBuffer, \n __global uint* histResult) { // each wg accumulates 1 bin -\n +\n \n /* declare variables */ -\n +\n \n // work indices \n size_t groupId = get_group_id(0); \n size_t localId = get_local_id(0); // 0 -> 256-1 \n size_t globalId = get_global_id(0); // 0 -> 8*10*256-1=20480-1 \n uint numThreads = get_global_size(0); \n unsigned int hist = 0; -\n +\n \n /* accumulate in global memory */ \n for ( uint p = 0; p < n; p+=GROUP_SIZE) { \n hist += histBuffer[ (get_group_id(0)*n + p)]; \n } -\n +\n \n /* reduction in local memory */ \n // populate local memory \n __local unsigned int localHist[GROUP_SIZE]; \n localHist[localId] = hist; \n barrier(CLK_LOCAL_MEM_FENCE); -\n +\n \n for (int stride = GROUP_SIZE/2; stride >= 1; stride /= 2) { \n if (localId < stride) { \n hist = localHist[ (localId+stride)]; @@ -1032,12 +1040,11 @@ KERNEL( \n } \n barrier(CLK_LOCAL_MEM_FENCE); \n } -\n +\n \n if (localId == 0) \n histResult[get_group_id(0)] = localHist[0]; -\n +\n \n } // kernel_HistogramRectAllChannelsReduction_Grey - ) // ThresholdRectToPix Kernel @@ -1092,7 +1099,8 @@ void kernel_ThresholdRectToPix( for ( int c = 0; c < NUM_CHANNELS; c++) { unsigned char pixChan = pixels.s[p*NUM_CHANNELS + c]; if (pHi_Values[c] >= 0 && (pixChan > pThresholds[c]) == (pHi_Values[c] == 0)) { - word |= (((uint)0x80000000) >> ((b*PIXELS_PER_BURST+p)&31)); + const uint kTopBit = 0x80000000; + word |= (kTopBit >> ((b*PIXELS_PER_BURST+p)&31)); } } } @@ -1146,10 +1154,10 @@ void kernel_ThresholdRectToPix_OneChan( // for each pixel in burst for ( int p = 0; p < PIXELS_PER_BURST; p++) { - + //int littleEndianIdx = p ^ 3; //int bigEndianIdx = p; - int idx = + int idx = \n#ifdef __ENDIAN_LITTLE__\n p ^ 3; \n#else\n @@ -1157,52 +1165,50 @@ void kernel_ThresholdRectToPix_OneChan( \n#endif\n unsigned char pixChan = pixels.s[idx]; if (pHi_Values[0] >= 0 && (pixChan > pThresholds[0]) == (pHi_Values[0] == 0)) { - word |= (0x80000000 >> ((b*PIXELS_PER_BURST+p)&31)); + const uint kTopBit = 0x80000000; + word |= (kTopBit >> ((b*PIXELS_PER_BURST+p)&31)); } } } pix[w] = word; } } - ) - KERNEL( -\n#define RED_SHIFT 24\n -\n#define GREEN_SHIFT 16\n -\n#define BLUE_SHIFT 8\n +\n#define RED_SHIFT 24\n +\n#define GREEN_SHIFT 16\n +\n#define BLUE_SHIFT 8\n \n#define SET_DATA_BYTE( pdata, n, val ) (*(l_uint8 *)((l_uintptr_t)((l_uint8 *)(pdata) + (n)) ^ 3) = (val))\n \n \n__attribute__((reqd_work_group_size(256, 1, 1)))\n \n__kernel\n \nvoid kernel_RGBToGray( __global const unsigned int *srcData, - __global unsigned char *dstData, + __global unsigned char *dstData, int srcWPL, int dstWPL, int height, int width, - float rwt, - float gwt, - float bwt ) { - + float rwt, + float gwt, + float bwt ) { + // pixel index int pixelIdx = get_global_id(0); if (pixelIdx >= height*width) return; - unsigned int word = srcData[pixelIdx]; - int output = (rwt * ((word >> RED_SHIFT) & 0xff) + + unsigned int word = srcData[pixelIdx]; + int output = (rwt * ((word >> RED_SHIFT) & 0xff) + gwt * ((word >> GREEN_SHIFT) & 0xff) + - bwt * ((word >> BLUE_SHIFT) & 0xff) + 0.5); + bwt * ((word >> BLUE_SHIFT) & 0xff) + 0.5f); // SET_DATA_BYTE dstData[pixelIdx] = output; } ) -#endif ; // close char* -#endif // USE_EXTERNAL_KERNEL -//#endif //_OCL_KERNEL_H_ +#endif // USE_EXTERNAL_KERNEL +#endif //_OCL_KERNEL_H_ /* vim:set shiftwidth=4 softtabstop=4 expandtab: */ diff --git a/opencl/opencl_device_selection.h b/opencl/opencl_device_selection.h index 74272b35c0..b595be62d4 100644 --- a/opencl/opencl_device_selection.h +++ b/opencl/opencl_device_selection.h @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifdef USE_OPENCL #ifndef DEVICE_SELECTION_H #define DEVICE_SELECTION_H @@ -59,12 +68,12 @@ typedef struct { typedef ds_status (*ds_score_release)(void* score); static ds_status releaseDSProfile(ds_profile* profile, ds_score_release sr) { ds_status status = DS_SUCCESS; - if (profile!=NULL) { - if (profile->devices!=NULL && sr!=NULL) { + if (profile != NULL) { + if (profile->devices != NULL && sr != NULL) { unsigned int i; for (i = 0; i < profile->numDevices; i++) { - if (profile->devices[i].oclDeviceName) free(profile->devices[i].oclDeviceName); - if (profile->devices[i].oclDriverVersion) free(profile->devices[i].oclDriverVersion); + free(profile->devices[i].oclDeviceName); + free(profile->devices[i].oclDriverVersion); status = sr(profile->devices[i].score); if (status != DS_SUCCESS) break; @@ -81,19 +90,15 @@ static ds_status initDSProfile(ds_profile** p, const char* version) { int numDevices; cl_uint numPlatforms; cl_platform_id* platforms = NULL; - cl_device_id* devices = NULL; + cl_device_id* devices = NULL; ds_status status = DS_SUCCESS; - ds_profile* profile = NULL; unsigned int next; unsigned int i; - if (p == NULL) - return DS_INVALID_PROFILE; + if (p == NULL) return DS_INVALID_PROFILE; + ds_profile* profile = (ds_profile*)malloc(sizeof(ds_profile)); + if (profile == NULL) return DS_MEMORY_ERROR; - profile = (ds_profile*)malloc(sizeof(ds_profile)); - if (profile == NULL) - return DS_MEMORY_ERROR; - memset(profile, 0, sizeof(ds_profile)); clGetPlatformIDs(0, NULL, &numPlatforms); @@ -123,7 +128,8 @@ static ds_status initDSProfile(ds_profile** p, const char* version) { } profile->numDevices = numDevices+1; // +1 to numDevices to include the native CPU - profile->devices = (ds_device*)malloc(profile->numDevices*sizeof(ds_device)); + profile->devices = + (ds_device*)malloc(profile->numDevices * sizeof(ds_device)); if (profile->devices == NULL) { profile->numDevices = 0; status = DS_MEMORY_ERROR; @@ -143,14 +149,14 @@ static ds_status initDSProfile(ds_profile** p, const char* version) { profile->devices[next].type = DS_DEVICE_OPENCL_DEVICE; profile->devices[next].oclDeviceID = devices[j]; - clGetDeviceInfo(profile->devices[next].oclDeviceID, CL_DEVICE_NAME - , DS_DEVICE_NAME_LENGTH, &buffer, NULL); + clGetDeviceInfo(profile->devices[next].oclDeviceID, CL_DEVICE_NAME, + DS_DEVICE_NAME_LENGTH, &buffer, NULL); length = strlen(buffer); profile->devices[next].oclDeviceName = (char*)malloc(length+1); memcpy(profile->devices[next].oclDeviceName, buffer, length+1); - clGetDeviceInfo(profile->devices[next].oclDeviceID, CL_DRIVER_VERSION - , DS_DEVICE_NAME_LENGTH, &buffer, NULL); + clGetDeviceInfo(profile->devices[next].oclDeviceID, CL_DRIVER_VERSION, + DS_DEVICE_NAME_LENGTH, &buffer, NULL); length = strlen(buffer); profile->devices[next].oclDriverVersion = (char*)malloc(length+1); memcpy(profile->devices[next].oclDriverVersion, buffer, length+1); @@ -160,15 +166,14 @@ static ds_status initDSProfile(ds_profile** p, const char* version) { profile->version = version; cleanup: - if (platforms) free(platforms); - if (devices) free(devices); + free(platforms); + free(devices); if (status == DS_SUCCESS) { *p = profile; } else { if (profile) { - if (profile->devices) - free(profile->devices); + free(profile->devices); free(profile); } } @@ -206,8 +211,7 @@ static ds_status profileDevices(ds_profile* profile, switch (type) { case DS_EVALUATE_NEW_ONLY: - if (profile->devices[i].score != NULL) - break; + if (profile->devices[i].score != NULL) break; // else fall through case DS_EVALUATE_ALL: evaluatorStatus = evaluator(profile->devices+i, evaluatorData); @@ -252,14 +256,11 @@ static ds_status writeProfileToFile(ds_profile* profile, ds_score_serializer serializer, const char* file) { ds_status status = DS_SUCCESS; - FILE* profileFile = NULL; + if (profile == NULL) return DS_INVALID_PROFILE; - if (profile == NULL) - return DS_INVALID_PROFILE; - - profileFile = fopen(file, "wb"); - if (profileFile==NULL) { + FILE* profileFile = fopen(file, "wb"); + if (profileFile == NULL) { status = DS_FILE_ERROR; } else { @@ -322,7 +323,8 @@ static ds_status writeProfileToFile(ds_profile* profile, fwrite(DS_TAG_SCORE, sizeof(char), strlen(DS_TAG_SCORE), profileFile); status = serializer(profile->devices+i, &serializedScore, &serializedScoreSize); - if (status == DS_SUCCESS && serializedScore!=NULL && serializedScoreSize > 0) { + if (status == DS_SUCCESS && serializedScore != NULL && + serializedScoreSize > 0) { fwrite(serializedScore, sizeof(char), serializedScoreSize, profileFile); free(serializedScore); } @@ -338,23 +340,21 @@ static ds_status writeProfileToFile(ds_profile* profile, static ds_status readProFile(const char* fileName, char** content, size_t* contentSize) { - FILE * input = NULL; size_t size = 0; - char* binary = NULL; *contentSize = 0; *content = NULL; - input = fopen(fileName, "rb"); - if(input == NULL) { + FILE* input = fopen(fileName, "rb"); + if (input == NULL) { return DS_FILE_ERROR; } fseek(input, 0L, SEEK_END); size = ftell(input); rewind(input); - binary = (char*)malloc(size); - if(binary == NULL) { + char* binary = (char*)malloc(size); + if (binary == NULL) { fclose(input); return DS_FILE_ERROR; } @@ -401,8 +401,7 @@ static ds_status readProfileFromFile(ds_profile* profile, const char* contentEnd = NULL; size_t contentSize; - if (profile==NULL) - return DS_INVALID_PROFILE; + if (profile == NULL) return DS_INVALID_PROFILE; status = readProFile(file, &contentStart, &contentSize); if (status == DS_SUCCESS) { @@ -424,7 +423,7 @@ static ds_status readProfileFromFile(ds_profile* profile, dataStart += strlen(DS_TAG_VERSION); dataEnd = findString(dataStart, contentEnd, DS_TAG_VERSION_END); - if (dataEnd==NULL) { + if (dataEnd == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } @@ -456,27 +455,27 @@ static ds_status readProfileFromFile(ds_profile* profile, const char* deviceDriverEnd; dataStart = findString(currentPosition, contentEnd, DS_TAG_DEVICE); - if (dataStart==NULL) { + if (dataStart == NULL) { // nothing useful remain, quit... break; } dataStart+=strlen(DS_TAG_DEVICE); dataEnd = findString(dataStart, contentEnd, DS_TAG_DEVICE_END); - if (dataEnd==NULL) { + if (dataEnd == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } // parse the device type deviceTypeStart = findString(dataStart, contentEnd, DS_TAG_DEVICE_TYPE); - if (deviceTypeStart==NULL) { + if (deviceTypeStart == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } deviceTypeStart+=strlen(DS_TAG_DEVICE_TYPE); deviceTypeEnd = findString(deviceTypeStart, contentEnd, DS_TAG_DEVICE_TYPE_END); - if (deviceTypeEnd==NULL) { + if (deviceTypeEnd == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } @@ -487,14 +486,14 @@ static ds_status readProfileFromFile(ds_profile* profile, if (deviceType == DS_DEVICE_OPENCL_DEVICE) { deviceNameStart = findString(dataStart, contentEnd, DS_TAG_DEVICE_NAME); - if (deviceNameStart==NULL) { + if (deviceNameStart == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } deviceNameStart+=strlen(DS_TAG_DEVICE_NAME); deviceNameEnd = findString(deviceNameStart, contentEnd, DS_TAG_DEVICE_NAME_END); - if (deviceNameEnd==NULL) { + if (deviceNameEnd == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } @@ -502,19 +501,18 @@ static ds_status readProfileFromFile(ds_profile* profile, deviceDriverStart = findString(dataStart, contentEnd, DS_TAG_DEVICE_DRIVER_VERSION); - if (deviceDriverStart==NULL) { + if (deviceDriverStart == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } deviceDriverStart+=strlen(DS_TAG_DEVICE_DRIVER_VERSION); deviceDriverEnd = findString(deviceDriverStart, contentEnd, DS_TAG_DEVICE_DRIVER_VERSION_END); - if (deviceDriverEnd ==NULL) { + if (deviceDriverEnd == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } - // check if this device is on the system for (i = 0; i < profile->numDevices; i++) { if (profile->devices[i].type == DS_DEVICE_OPENCL_DEVICE) { @@ -530,7 +528,7 @@ static ds_status readProfileFromFile(ds_profile* profile, && strncmp(profile->devices[i].oclDriverVersion, deviceDriverStart, driverVersionLength)==0) { deviceScoreStart = findString(dataStart, contentEnd, DS_TAG_SCORE); - if (deviceNameStart==NULL) { + if (deviceNameStart == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } @@ -552,7 +550,7 @@ static ds_status readProfileFromFile(ds_profile* profile, for (i = 0; i < profile->numDevices; i++) { if (profile->devices[i].type == DS_DEVICE_NATIVE_CPU) { deviceScoreStart = findString(dataStart, contentEnd, DS_TAG_SCORE); - if (deviceScoreStart==NULL) { + if (deviceScoreStart == NULL) { status = DS_PROFILE_FILE_ERROR; goto cleanup; } @@ -574,23 +572,9 @@ static ds_status readProfileFromFile(ds_profile* profile, } } cleanup: - if (contentStart!=NULL) free(contentStart); + free(contentStart); return status; } -static ds_status getNumDeviceWithEmptyScore(ds_profile* profile, - unsigned int* num) { - unsigned int i; - if (profile == NULL || num==NULL) - return DS_MEMORY_ERROR; - *num=0; - for (i = 0; i < profile->numDevices; i++) { - if (profile->devices[i].score == NULL) { - *num++; - } - } - return DS_SUCCESS; -} - #endif #endif diff --git a/opencl/openclwrapper.cpp b/opencl/openclwrapper.cpp index f35fcd439d..c3bb745127 100644 --- a/opencl/openclwrapper.cpp +++ b/opencl/openclwrapper.cpp @@ -1,7 +1,14 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifdef _WIN32 -#include #include - #else #include #include @@ -16,23 +23,24 @@ #include "thresholder.h" #if ON_APPLE -#include #include +#include #endif /* Convenience macro to test the version of Leptonica. */ #if defined(LIBLEPT_MAJOR_VERSION) && defined(LIBLEPT_MINOR_VERSION) -# define TESSERACT_LIBLEPT_PREREQ(maj, min) \ - ((LIBLEPT_MAJOR_VERSION) > (maj) || ((LIBLEPT_MAJOR_VERSION) == (maj) && (LIBLEPT_MINOR_VERSION) >= (min))) +#define TESSERACT_LIBLEPT_PREREQ(maj, min) \ + ((LIBLEPT_MAJOR_VERSION) > (maj) || \ + ((LIBLEPT_MAJOR_VERSION) == (maj) && (LIBLEPT_MINOR_VERSION) >= (min))) #else -# define TESSERACT_LIBLEPT_PREREQ(maj, min) 0 +#define TESSERACT_LIBLEPT_PREREQ(maj, min) 0 #endif -#if TESSERACT_LIBLEPT_PREREQ(1,73) -# define CALLOC LEPT_CALLOC -# define FREE LEPT_FREE +#if TESSERACT_LIBLEPT_PREREQ(1, 73) +#define CALLOC LEPT_CALLOC +#define FREE LEPT_FREE #endif #ifdef USE_OPENCL @@ -40,36 +48,28 @@ #include "opencl_device_selection.h" GPUEnv OpenclDevice::gpuEnv; - bool OpenclDevice::deviceIsSelected = false; ds_device OpenclDevice::selectedDevice; - int OpenclDevice::isInited = 0; static l_int32 MORPH_BC = ASYMMETRIC_MORPH_BC; static const l_uint32 lmask32[] = { - 0x80000000, 0xc0000000, 0xe0000000, 0xf0000000, - 0xf8000000, 0xfc000000, 0xfe000000, 0xff000000, - 0xff800000, 0xffc00000, 0xffe00000, 0xfff00000, - 0xfff80000, 0xfffc0000, 0xfffe0000, 0xffff0000, - 0xffff8000, 0xffffc000, 0xffffe000, 0xfffff000, - 0xfffff800, 0xfffffc00, 0xfffffe00, 0xffffff00, - 0xffffff80, 0xffffffc0, 0xffffffe0, 0xfffffff0, - 0xfffffff8, 0xfffffffc, 0xfffffffe, 0xffffffff -}; + 0x80000000, 0xc0000000, 0xe0000000, 0xf0000000, 0xf8000000, 0xfc000000, + 0xfe000000, 0xff000000, 0xff800000, 0xffc00000, 0xffe00000, 0xfff00000, + 0xfff80000, 0xfffc0000, 0xfffe0000, 0xffff0000, 0xffff8000, 0xffffc000, + 0xffffe000, 0xfffff000, 0xfffff800, 0xfffffc00, 0xfffffe00, 0xffffff00, + 0xffffff80, 0xffffffc0, 0xffffffe0, 0xfffffff0, 0xfffffff8, 0xfffffffc, + 0xfffffffe, 0xffffffff}; static const l_uint32 rmask32[] = { - 0x00000001, 0x00000003, 0x00000007, 0x0000000f, - 0x0000001f, 0x0000003f, 0x0000007f, 0x000000ff, - 0x000001ff, 0x000003ff, 0x000007ff, 0x00000fff, - 0x00001fff, 0x00003fff, 0x00007fff, 0x0000ffff, - 0x0001ffff, 0x0003ffff, 0x0007ffff, 0x000fffff, - 0x001fffff, 0x003fffff, 0x007fffff, 0x00ffffff, - 0x01ffffff, 0x03ffffff, 0x07ffffff, 0x0fffffff, - 0x1fffffff, 0x3fffffff, 0x7fffffff, 0xffffffff -}; + 0x00000001, 0x00000003, 0x00000007, 0x0000000f, 0x0000001f, 0x0000003f, + 0x0000007f, 0x000000ff, 0x000001ff, 0x000003ff, 0x000007ff, 0x00000fff, + 0x00001fff, 0x00003fff, 0x00007fff, 0x0000ffff, 0x0001ffff, 0x0003ffff, + 0x0007ffff, 0x000fffff, 0x001fffff, 0x003fffff, 0x007fffff, 0x00ffffff, + 0x01ffffff, 0x03ffffff, 0x07ffffff, 0x0fffffff, 0x1fffffff, 0x3fffffff, + 0x7fffffff, 0xffffffff}; struct tiff_transform { int vflip; /* if non-zero, image needs a vertical fip */ @@ -90,7 +90,7 @@ static struct tiff_transform tiff_orientation_transforms[] = { {0, 0, -1} }; -static const l_int32 MAX_PAGES_IN_TIFF_FILE = 3000; +static const l_int32 MAX_PAGES_IN_TIFF_FILE = 3000; cl_mem pixsCLBuffer, pixdCLBuffer, pixdCLIntermediate; //Morph operations buffers cl_mem pixThBuffer; //output from thresholdtopix calculation @@ -100,7 +100,8 @@ KernelEnv rEnv; // substitute invalid characters in device name with _ void legalizeFileName( char *fileName) { //printf("fileName: %s\n", fileName); - const char* invalidChars = "/\?:*\"><| "; // space is valid but can cause headaches + const char *invalidChars = + "/\?:*\"><| "; // space is valid but can cause headaches // for each invalid char for (int i = 0; i < strlen(invalidChars); i++) { char invalidStr[4]; @@ -111,10 +112,11 @@ void legalizeFileName( char *fileName) { // initial ./ is valid for present directory //if (*pos == '.') pos++; //if (*pos == '/') pos++; - for ( char *pos = strstr(fileName, invalidStr); pos != NULL; pos = strstr(pos+1, invalidStr)) { - //printf("\tfound: %s, ", pos); - pos[0] = '_'; - //printf("fileName: %s\n", fileName); + for (char *pos = strstr(fileName, invalidStr); pos != NULL; + pos = strstr(pos + 1, invalidStr)) { + // printf("\tfound: %s, ", pos); + pos[0] = '_'; + // printf("fileName: %s\n", fileName); } } } @@ -127,39 +129,41 @@ void populateGPUEnvFromDevice( GPUEnv *gpuInfo, cl_device_id device ) { gpuInfo->mpDevID = device; gpuInfo->mpArryDevsID = new cl_device_id[1]; gpuInfo->mpArryDevsID[0] = gpuInfo->mpDevID; - clStatus = clGetDeviceInfo(gpuInfo->mpDevID, CL_DEVICE_TYPE , sizeof(cl_device_type), (void *) &gpuInfo->mDevType , &size); + clStatus = + clGetDeviceInfo(gpuInfo->mpDevID, CL_DEVICE_TYPE, + sizeof(cl_device_type), &gpuInfo->mDevType, &size); CHECK_OPENCL( clStatus, "populateGPUEnv::getDeviceInfo(TYPE)"); // platform - clStatus = clGetDeviceInfo(gpuInfo->mpDevID, CL_DEVICE_PLATFORM , sizeof(cl_platform_id), (void *) &gpuInfo->mpPlatformID , &size); + clStatus = + clGetDeviceInfo(gpuInfo->mpDevID, CL_DEVICE_PLATFORM, + sizeof(cl_platform_id), &gpuInfo->mpPlatformID, &size); CHECK_OPENCL( clStatus, "populateGPUEnv::getDeviceInfo(PLATFORM)"); // context cl_context_properties props[3]; props[0] = CL_CONTEXT_PLATFORM; props[1] = (cl_context_properties) gpuInfo->mpPlatformID; props[2] = 0; - gpuInfo->mpContext = clCreateContext(props, 1, &gpuInfo->mpDevID, NULL, NULL, &clStatus); + gpuInfo->mpContext = clCreateContext(props, 1, &gpuInfo->mpDevID, NULL, + NULL, &clStatus); CHECK_OPENCL( clStatus, "populateGPUEnv::createContext"); // queue cl_command_queue_properties queueProperties = 0; gpuInfo->mpCmdQueue = clCreateCommandQueue( gpuInfo->mpContext, gpuInfo->mpDevID, queueProperties, &clStatus ); CHECK_OPENCL( clStatus, "populateGPUEnv::createCommandQueue"); - } int OpenclDevice::LoadOpencl() { #ifdef WIN32 - HINSTANCE HOpenclDll = NULL; - void * OpenclDll = NULL; - //fprintf(stderr, " LoadOpenclDllxx... \n"); - OpenclDll = static_cast( HOpenclDll ); - OpenclDll = LoadLibrary( "openCL.dll" ); - if ( !static_cast( OpenclDll ) ) - { - fprintf(stderr, "[OD] Load opencl.dll failed!\n"); - FreeLibrary( static_cast( OpenclDll ) ); - return 0; - + HINSTANCE HOpenclDll = NULL; + void *OpenclDll = NULL; + // fprintf(stderr, " LoadOpenclDllxx... \n"); + OpenclDll = static_cast(HOpenclDll); + OpenclDll = LoadLibrary("openCL.dll"); + if (!static_cast(OpenclDll)) { + fprintf(stderr, "[OD] Load opencl.dll failed!\n"); + FreeLibrary(static_cast(OpenclDll)); + return 0; } fprintf(stderr, "[OD] Load opencl.dll successful!\n"); #endif @@ -182,60 +186,57 @@ cl_mem allocateZeroCopyBuffer(KernelEnv rEnv, l_uint32 *hostbuffer, size_t nElem return membuffer; } -PIX* mapOutputCLBuffer(KernelEnv rEnv, cl_mem clbuffer, PIX* pixd, PIX* pixs, int elements, cl_mem_flags flags, bool memcopy = false, bool sync = true) -{ - PROCNAME("mapOutputCLBuffer"); - if (!pixd) - { - if (memcopy) - { - if ((pixd = pixCreateTemplate(pixs)) == NULL) - (PIX *)ERROR_PTR("pixd not made", procName, NULL); - } - else - { - if ((pixd = pixCreateHeader(pixGetWidth(pixs), pixGetHeight(pixs), pixGetDepth(pixs))) == NULL) - (PIX *)ERROR_PTR("pixd not made", procName, NULL); - } - } - l_uint32 *pValues = (l_uint32 *)clEnqueueMapBuffer(rEnv.mpkCmdQueue, clbuffer, CL_TRUE, flags, 0, - elements * sizeof(l_uint32), 0, NULL, NULL, NULL ); - - if (memcopy) - { - memcpy(pixGetData(pixd), pValues, elements * sizeof(l_uint32)); - } - else - { - pixSetData(pixd, pValues); +PIX *mapOutputCLBuffer(KernelEnv rEnv, cl_mem clbuffer, PIX *pixd, PIX *pixs, + int elements, cl_mem_flags flags, bool memcopy = false, + bool sync = true) { + PROCNAME("mapOutputCLBuffer"); + if (!pixd) { + if (memcopy) { + if ((pixd = pixCreateTemplate(pixs)) == NULL) + (PIX *)ERROR_PTR("pixd not made", procName, NULL); + } else { + if ((pixd = pixCreateHeader(pixGetWidth(pixs), pixGetHeight(pixs), + pixGetDepth(pixs))) == NULL) + (PIX *)ERROR_PTR("pixd not made", procName, NULL); } + } + l_uint32 *pValues = (l_uint32 *)clEnqueueMapBuffer( + rEnv.mpkCmdQueue, clbuffer, CL_TRUE, flags, 0, + elements * sizeof(l_uint32), 0, NULL, NULL, NULL); + + if (memcopy) { + memcpy(pixGetData(pixd), pValues, elements * sizeof(l_uint32)); + } else { + pixSetData(pixd, pValues); + } - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue,clbuffer,pValues,0,NULL,NULL); + clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, clbuffer, pValues, 0, NULL, + NULL); - if (sync) - { - clFinish( rEnv.mpkCmdQueue ); - } + if (sync) { + clFinish(rEnv.mpkCmdQueue); + } - return pixd; + return pixd; } cl_mem allocateIntBuffer( KernelEnv rEnv, const l_uint32 *_pValues, size_t nElements, cl_int *pStatus , bool sync = false) { - cl_mem xValues = clCreateBuffer( rEnv.mpkContext, (cl_mem_flags) (CL_MEM_READ_WRITE), - nElements * sizeof(l_int32), NULL, pStatus); + cl_mem xValues = + clCreateBuffer(rEnv.mpkContext, (cl_mem_flags)(CL_MEM_READ_WRITE), + nElements * sizeof(l_int32), NULL, pStatus); - if (_pValues != NULL) - { - l_int32 *pValues = (l_int32 *)clEnqueueMapBuffer( rEnv.mpkCmdQueue, xValues, CL_TRUE, CL_MAP_WRITE, 0, - nElements * sizeof(l_int32), 0, NULL, NULL, NULL ); + if (_pValues != NULL) { + l_int32 *pValues = (l_int32 *)clEnqueueMapBuffer( + rEnv.mpkCmdQueue, xValues, CL_TRUE, CL_MAP_WRITE, 0, + nElements * sizeof(l_int32), 0, NULL, NULL, NULL); - memcpy(pValues, _pValues, nElements * sizeof(l_int32)); + memcpy(pValues, _pValues, nElements * sizeof(l_int32)); - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue,xValues,pValues,0,NULL,NULL); + clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, xValues, pValues, 0, NULL, + NULL); - if (sync) - clFinish( rEnv.mpkCmdQueue ); + if (sync) clFinish(rEnv.mpkCmdQueue); } return xValues; @@ -244,27 +245,25 @@ PIX* mapOutputCLBuffer(KernelEnv rEnv, cl_mem clbuffer, PIX* pixd, PIX* pixs, in void OpenclDevice::releaseMorphCLBuffers() { - if (pixdCLIntermediate != NULL) - clReleaseMemObject(pixdCLIntermediate); - if (pixsCLBuffer != NULL) - clReleaseMemObject(pixsCLBuffer); - if (pixdCLBuffer != NULL) - clReleaseMemObject(pixdCLBuffer); - if (pixThBuffer != NULL) - clReleaseMemObject(pixThBuffer); - pixdCLIntermediate = pixsCLBuffer = pixdCLBuffer = pixThBuffer = NULL; + if (pixdCLIntermediate != NULL) clReleaseMemObject(pixdCLIntermediate); + if (pixsCLBuffer != NULL) clReleaseMemObject(pixsCLBuffer); + if (pixdCLBuffer != NULL) clReleaseMemObject(pixdCLBuffer); + if (pixThBuffer != NULL) clReleaseMemObject(pixThBuffer); + pixdCLIntermediate = pixsCLBuffer = pixdCLBuffer = pixThBuffer = NULL; } int OpenclDevice::initMorphCLAllocations(l_int32 wpl, l_int32 h, PIX* pixs) { SetKernelEnv( &rEnv ); - if (pixThBuffer != NULL) - { - pixsCLBuffer = allocateZeroCopyBuffer(rEnv, NULL, wpl*h, CL_MEM_ALLOC_HOST_PTR, &clStatus); + if (pixThBuffer != NULL) { + pixsCLBuffer = allocateZeroCopyBuffer(rEnv, NULL, wpl * h, + CL_MEM_ALLOC_HOST_PTR, &clStatus); - //Get the output from ThresholdToPix operation - clStatus = clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixThBuffer, pixsCLBuffer, 0, 0, sizeof(l_uint32) * wpl*h, 0, NULL, NULL); + // Get the output from ThresholdToPix operation + clStatus = + clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixThBuffer, pixsCLBuffer, 0, 0, + sizeof(l_uint32) * wpl * h, 0, NULL, NULL); } else { @@ -275,9 +274,11 @@ int OpenclDevice::initMorphCLAllocations(l_int32 wpl, l_int32 h, PIX* pixs) pixsCLBuffer = allocateZeroCopyBuffer(rEnv, srcdata, wpl*h, CL_MEM_USE_HOST_PTR, &clStatus); } - pixdCLBuffer = allocateZeroCopyBuffer(rEnv, NULL, wpl*h, CL_MEM_ALLOC_HOST_PTR, &clStatus); + pixdCLBuffer = allocateZeroCopyBuffer(rEnv, NULL, wpl * h, + CL_MEM_ALLOC_HOST_PTR, &clStatus); - pixdCLIntermediate = allocateZeroCopyBuffer(rEnv, NULL, wpl*h, CL_MEM_ALLOC_HOST_PTR, &clStatus); + pixdCLIntermediate = allocateZeroCopyBuffer( + rEnv, NULL, wpl * h, CL_MEM_ALLOC_HOST_PTR, &clStatus); return (int)clStatus; } @@ -296,7 +297,6 @@ PERF_COUNT_SUB("LoadOpencl") #endif // sets up environment, compiles programs - InitOpenclRunEnv_DeviceSelection( 0 ); //PERF_COUNT_SUB("called InitOpenclRunEnv_DS") //PERF_COUNT_END @@ -409,20 +409,20 @@ int OpenclDevice::BinaryGenerated( const char * clFileName, FILE ** fhandle ) int status = 0; char *str = NULL; FILE *fd = NULL; - char fileName[256] = { 0 }, cl_name[128] = { 0 }; + char fileName[256] = {0}, cl_name[128] = {0}; char deviceName[1024]; - clStatus = clGetDeviceInfo( gpuEnv.mpArryDevsID[i], CL_DEVICE_NAME, sizeof(deviceName), deviceName, NULL ); - CHECK_OPENCL( clStatus, "clGetDeviceInfo" ); - str = (char*) strstr( clFileName, (char*) ".cl" ); - memcpy( cl_name, clFileName, str - clFileName ); + clStatus = clGetDeviceInfo(gpuEnv.mpArryDevsID[i], CL_DEVICE_NAME, + sizeof(deviceName), deviceName, NULL); + CHECK_OPENCL(clStatus, "clGetDeviceInfo"); + str = (char *)strstr(clFileName, (char *)".cl"); + memcpy(cl_name, clFileName, str - clFileName); cl_name[str - clFileName] = '\0'; - sprintf( fileName, "%s-%s.bin", cl_name, deviceName ); + sprintf(fileName, "%s-%s.bin", cl_name, deviceName); legalizeFileName(fileName); - fd = fopen( fileName, "rb" ); - status = ( fd != NULL ) ? 1 : 0; - if ( fd != NULL ) - { - *fhandle = fd; + fd = fopen(fileName, "rb"); + status = (fd != NULL) ? 1 : 0; + if (fd != NULL) { + *fhandle = fd; } return status; @@ -434,9 +434,8 @@ int OpenclDevice::CachedOfKernerPrg( const GPUEnv *gpuEnvCached, const char * cl { if ( strcasecmp( gpuEnvCached->mArryKnelSrcFile[i], clFileName ) == 0 ) { - if ( gpuEnvCached->mpArryPrograms[i] != NULL ) - { - return 1; + if (gpuEnvCached->mpArryPrograms[i] != NULL) { + return 1; } } } @@ -445,11 +444,10 @@ int OpenclDevice::CachedOfKernerPrg( const GPUEnv *gpuEnvCached, const char * cl } int OpenclDevice::WriteBinaryToFile( const char* fileName, const char* birary, size_t numBytes ) { - FILE *output = NULL; - output = fopen( fileName, "wb" ); - if ( output == NULL ) - { - return 0; + FILE *output = NULL; + output = fopen(fileName, "wb"); + if (output == NULL) { + return 0; } fwrite( birary, sizeof(char), numBytes, output ); @@ -462,36 +460,37 @@ int OpenclDevice::GeneratBinFromKernelSource( cl_program program, const char * c { unsigned int i = 0; cl_int clStatus; - size_t *binarySizes, numDevices=0; + size_t *binarySizes; + cl_uint numDevices; cl_device_id *mpArryDevsID; char **binaries, *str = NULL; - clStatus = clGetProgramInfo( program, CL_PROGRAM_NUM_DEVICES, - sizeof(numDevices), &numDevices, NULL ); + clStatus = clGetProgramInfo(program, CL_PROGRAM_NUM_DEVICES, + sizeof(numDevices), &numDevices, NULL); CHECK_OPENCL( clStatus, "clGetProgramInfo" ); mpArryDevsID = (cl_device_id*) malloc( sizeof(cl_device_id) * numDevices ); - if ( mpArryDevsID == NULL ) - { - return 0; + if (mpArryDevsID == NULL) { + return 0; } /* grab the handles to all of the devices in the program. */ - clStatus = clGetProgramInfo( program, CL_PROGRAM_DEVICES, - sizeof(cl_device_id) * numDevices, mpArryDevsID, NULL ); + clStatus = clGetProgramInfo(program, CL_PROGRAM_DEVICES, + sizeof(cl_device_id) * numDevices, mpArryDevsID, + NULL); CHECK_OPENCL( clStatus, "clGetProgramInfo" ); /* figure out the sizes of each of the binaries. */ binarySizes = (size_t*) malloc( sizeof(size_t) * numDevices ); - clStatus = clGetProgramInfo( program, CL_PROGRAM_BINARY_SIZES, - sizeof(size_t) * numDevices, binarySizes, NULL ); + clStatus = + clGetProgramInfo(program, CL_PROGRAM_BINARY_SIZES, + sizeof(size_t) * numDevices, binarySizes, NULL); CHECK_OPENCL( clStatus, "clGetProgramInfo" ); /* copy over all of the generated binaries. */ binaries = (char**) malloc( sizeof(char *) * numDevices ); - if ( binaries == NULL ) - { - return 0; + if (binaries == NULL) { + return 0; } for ( i = 0; i < numDevices; i++ ) @@ -499,19 +498,18 @@ int OpenclDevice::GeneratBinFromKernelSource( cl_program program, const char * c if ( binarySizes[i] != 0 ) { binaries[i] = (char*) malloc( sizeof(char) * binarySizes[i] ); - if ( binaries[i] == NULL ) - { - return 0; + if (binaries[i] == NULL) { + return 0; } } else { - binaries[i] = NULL; + binaries[i] = NULL; } } - clStatus = clGetProgramInfo( program, CL_PROGRAM_BINARIES, - sizeof(char *) * numDevices, binaries, NULL ); + clStatus = clGetProgramInfo(program, CL_PROGRAM_BINARIES, + sizeof(char *) * numDevices, binaries, NULL); CHECK_OPENCL(clStatus,"clGetProgramInfo"); /* dump out each binary into its own separate file. */ @@ -523,7 +521,7 @@ int OpenclDevice::GeneratBinFromKernelSource( cl_program program, const char * c { char deviceName[1024]; clStatus = clGetDeviceInfo(mpArryDevsID[i], CL_DEVICE_NAME, - sizeof(deviceName), deviceName, NULL); + sizeof(deviceName), deviceName, NULL); CHECK_OPENCL( clStatus, "clGetDeviceInfo" ); str = (char*) strstr( clFileName, (char*) ".cl" ); @@ -543,45 +541,34 @@ int OpenclDevice::GeneratBinFromKernelSource( cl_program program, const char * c // Release all resouces and memory for ( i = 0; i < numDevices; i++ ) { - if ( binaries[i] != NULL ) - { - free( binaries[i] ); - binaries[i] = NULL; - } + free(binaries[i]); + binaries[i] = NULL; } - if ( binaries != NULL ) - { - free( binaries ); - binaries = NULL; - } + free(binaries); + binaries = NULL; - if ( binarySizes != NULL ) - { - free( binarySizes ); - binarySizes = NULL; - } + free(binarySizes); + binarySizes = NULL; + + free(mpArryDevsID); + mpArryDevsID = NULL; - if ( mpArryDevsID != NULL ) - { - free( mpArryDevsID ); - mpArryDevsID = NULL; - } return 1; } void copyIntBuffer( KernelEnv rEnv, cl_mem xValues, const l_uint32 *_pValues, size_t nElements, cl_int *pStatus ) { - l_int32 *pValues = (l_int32 *)clEnqueueMapBuffer( rEnv.mpkCmdQueue, xValues, CL_TRUE, CL_MAP_WRITE, 0, - nElements * sizeof(l_int32), 0, NULL, NULL, NULL ); - clFinish( rEnv.mpkCmdQueue ); - if (_pValues != NULL) - { - for ( int i = 0; i < (int)nElements; i++ ) - pValues[i] = (l_int32)_pValues[i]; + l_int32 *pValues = (l_int32 *)clEnqueueMapBuffer( + rEnv.mpkCmdQueue, xValues, CL_TRUE, CL_MAP_WRITE, 0, + nElements * sizeof(l_int32), 0, NULL, NULL, NULL); + clFinish(rEnv.mpkCmdQueue); + if (_pValues != NULL) { + for (int i = 0; i < (int)nElements; i++) pValues[i] = (l_int32)_pValues[i]; } - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue,xValues,pValues,0,NULL,NULL); + clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, xValues, pValues, 0, NULL, + NULL); //clFinish( rEnv.mpkCmdQueue ); return; } @@ -595,7 +582,7 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) const char *source; size_t source_size[1]; int b_error, binary_status, binaryExisted, idx; - size_t numDevices; + cl_uint numDevices; cl_device_id *mpArryDevsID; FILE *fd, *fd1; const char* filename = "kernel.cl"; @@ -615,14 +602,13 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) //PERF_COUNT_SUB("BinaryGenerated") if ( binaryExisted == 1 ) { - clStatus = clGetContextInfo( gpuInfo->mpContext, CL_CONTEXT_NUM_DEVICES, - sizeof(numDevices), &numDevices, NULL ); - CHECK_OPENCL( clStatus, "clGetContextInfo" ); + clStatus = clGetContextInfo(gpuInfo->mpContext, CL_CONTEXT_NUM_DEVICES, + sizeof(numDevices), &numDevices, NULL); + CHECK_OPENCL(clStatus, "clGetContextInfo"); - mpArryDevsID = (cl_device_id*) malloc( sizeof(cl_device_id) * numDevices ); - if ( mpArryDevsID == NULL ) - { - return 0; + mpArryDevsID = (cl_device_id *)malloc(sizeof(cl_device_id) * numDevices); + if (mpArryDevsID == NULL) { + return 0; } //PERF_COUNT_SUB("get numDevices") b_error = 0; @@ -649,8 +635,9 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) //PERF_COUNT_SUB("read file") fd = NULL; // grab the handles to all of the devices in the context. - clStatus = clGetContextInfo( gpuInfo->mpContext, CL_CONTEXT_DEVICES, - sizeof( cl_device_id ) * numDevices, mpArryDevsID, NULL ); + clStatus = clGetContextInfo(gpuInfo->mpContext, CL_CONTEXT_DEVICES, + sizeof(cl_device_id) * numDevices, + mpArryDevsID, NULL); CHECK_OPENCL( clStatus, "clGetContextInfo" ); //PERF_COUNT_SUB("get devices") //fprintf(stderr, "[OD] Create kernel from binary\n"); @@ -662,7 +649,7 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) free( binary ); free( mpArryDevsID ); mpArryDevsID = NULL; -//PERF_COUNT_SUB("binaryExisted") + // PERF_COUNT_SUB("binaryExisted") } else { @@ -674,9 +661,8 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) //PERF_COUNT_SUB("!binaryExisted") } - if ( gpuInfo->mpArryPrograms[idx] == (cl_program) NULL ) - { - return 0; + if (gpuInfo->mpArryPrograms[idx] == (cl_program) NULL) { + return 0; } //char options[512]; @@ -685,15 +671,17 @@ int OpenclDevice::CompileKernelFile( GPUEnv *gpuInfo, const char *buildOption ) PERF_COUNT_START("OD::CompileKernel::clBuildProgram") if (!gpuInfo->mnIsUserCreated) { - clStatus = clBuildProgram(gpuInfo->mpArryPrograms[idx], 1, gpuInfo->mpArryDevsID, - buildOption, NULL, NULL); -//PERF_COUNT_SUB("clBuildProgram notUserCreated") + clStatus = + clBuildProgram(gpuInfo->mpArryPrograms[idx], 1, gpuInfo->mpArryDevsID, + buildOption, NULL, NULL); + // PERF_COUNT_SUB("clBuildProgram notUserCreated") } else { - clStatus = clBuildProgram(gpuInfo->mpArryPrograms[idx], 1, &(gpuInfo->mpDevID), - buildOption, NULL, NULL); -//PERF_COUNT_SUB("clBuildProgram isUserCreated") + clStatus = + clBuildProgram(gpuInfo->mpArryPrograms[idx], 1, &(gpuInfo->mpDevID), + buildOption, NULL, NULL); + // PERF_COUNT_SUB("clBuildProgram isUserCreated") } PERF_COUNT_END if ( clStatus != CL_SUCCESS ) @@ -701,13 +689,15 @@ PERF_COUNT_END printf ("BuildProgram error!\n"); if ( !gpuInfo->mnIsUserCreated ) { - clStatus = clGetProgramBuildInfo( gpuInfo->mpArryPrograms[idx], gpuInfo->mpArryDevsID[0], - CL_PROGRAM_BUILD_LOG, 0, NULL, &length ); + clStatus = clGetProgramBuildInfo( + gpuInfo->mpArryPrograms[idx], gpuInfo->mpArryDevsID[0], + CL_PROGRAM_BUILD_LOG, 0, NULL, &length); } else { - clStatus = clGetProgramBuildInfo( gpuInfo->mpArryPrograms[idx], gpuInfo->mpDevID, - CL_PROGRAM_BUILD_LOG, 0, NULL, &length); + clStatus = clGetProgramBuildInfo( + gpuInfo->mpArryPrograms[idx], gpuInfo->mpDevID, + CL_PROGRAM_BUILD_LOG, 0, NULL, &length); } if ( clStatus != CL_SUCCESS ) { @@ -715,9 +705,8 @@ PERF_COUNT_END return 0; } buildLog = (char*) malloc( length ); - if ( buildLog == (char*) NULL ) - { - return 0; + if (buildLog == (char *)NULL) { + return 0; } if ( !gpuInfo->mnIsUserCreated ) { @@ -736,10 +725,9 @@ PERF_COUNT_END } fd1 = fopen( "kernel-build.log", "w+" ); - if ( fd1 != NULL ) - { - fwrite( buildLog, sizeof(char), length, fd1 ); - fclose( fd1 ); + if (fd1 != NULL) { + fwrite(buildLog, sizeof(char), length, fd1); + fclose(fd1); } free( buildLog ); @@ -782,36 +770,40 @@ PERF_COUNT_START("pixReadFromTiffKernel") l_uint32 *pResult = (l_uint32 *)malloc(w*h * sizeof(l_uint32)); rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "composeRGBPixel", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel"); + CHECK_OPENCL(clStatus, "clCreateKernel composeRGBPixel"); //Allocate input and output OCL buffers valuesCl = allocateZeroCopyBuffer(rEnv, tiffdata, w*h, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, &clStatus); outputCl = allocateZeroCopyBuffer(rEnv, pResult, w*h, CL_MEM_WRITE_ONLY | CL_MEM_USE_HOST_PTR, &clStatus); //Kernel arguments - clStatus = clSetKernelArg( rEnv.mpkKernel, 0, sizeof(cl_mem), (void *)&valuesCl ); + clStatus = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &valuesCl); CHECK_OPENCL( clStatus, "clSetKernelArg"); - clStatus = clSetKernelArg( rEnv.mpkKernel, 1, sizeof(w), (void *)&w ); + clStatus = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(w), &w); CHECK_OPENCL( clStatus, "clSetKernelArg" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 2, sizeof(h), (void *)&h ); + clStatus = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(h), &h); CHECK_OPENCL( clStatus, "clSetKernelArg" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 3, sizeof(wpl), (void *)&wpl ); + clStatus = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); CHECK_OPENCL( clStatus, "clSetKernelArg" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 4, sizeof(cl_mem), (void *)&outputCl ); + clStatus = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(cl_mem), &outputCl); CHECK_OPENCL( clStatus, "clSetKernelArg"); //Kernel enqueue PERF_COUNT_SUB("before") - clStatus = clEnqueueNDRangeKernel( rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, NULL, globalThreads, localThreads, 0, NULL, NULL ); - CHECK_OPENCL( clStatus, "clEnqueueNDRangeKernel" ); - - /* map results back from gpu */ - void *ptr = clEnqueueMapBuffer(rEnv.mpkCmdQueue, outputCl, CL_TRUE, CL_MAP_READ, 0, w*h * sizeof(l_uint32), 0, NULL, NULL, &clStatus); - CHECK_OPENCL( clStatus, "clEnqueueMapBuffer outputCl"); - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, outputCl, ptr, 0, NULL, NULL); - - //Sync - clFinish( rEnv.mpkCmdQueue ); +clStatus = + clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, NULL, + globalThreads, localThreads, 0, NULL, NULL); +CHECK_OPENCL(clStatus, "clEnqueueNDRangeKernel"); + +/* map results back from gpu */ +void *ptr = clEnqueueMapBuffer(rEnv.mpkCmdQueue, outputCl, CL_TRUE, CL_MAP_READ, + 0, w * h * sizeof(l_uint32), 0, NULL, NULL, + &clStatus); +CHECK_OPENCL(clStatus, "clEnqueueMapBuffer outputCl"); +clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, outputCl, ptr, 0, NULL, NULL); + +// Sync +clFinish(rEnv.mpkCmdQueue); PERF_COUNT_SUB("kernel & map") PERF_COUNT_END return pResult; @@ -827,18 +819,17 @@ PIX *pix; PROCNAME("pixReadTiff"); if (!filename) - return (PIX *)ERROR_PTR("filename not defined", procName, NULL); + return (PIX *)ERROR_PTR("filename not defined", procName, NULL); if ((fp = fopenReadStream(filename)) == NULL) - return (PIX *)ERROR_PTR("image file not found", procName, NULL); + return (PIX *)ERROR_PTR("image file not found", procName, NULL); if ((pix = pixReadStreamTiffCl(fp, n)) == NULL) { - fclose(fp); - return (PIX *)ERROR_PTR("pix not read", procName, NULL); + fclose(fp); + return (PIX *)ERROR_PTR("pix not read", procName, NULL); } fclose(fp); PERF_COUNT_END return pix; - } TIFF * OpenclDevice::fopenTiffCl(FILE *fp, @@ -848,13 +839,12 @@ l_int32 fd; PROCNAME("fopenTiff"); - if (!fp) - return (TIFF *)ERROR_PTR("stream not opened", procName, NULL); + if (!fp) return (TIFF *)ERROR_PTR("stream not opened", procName, NULL); if (!modestring) - return (TIFF *)ERROR_PTR("modestring not defined", procName, NULL); + return (TIFF *)ERROR_PTR("modestring not defined", procName, NULL); if ((fd = fileno(fp)) < 0) - return (TIFF *)ERROR_PTR("invalid file descriptor", procName, NULL); + return (TIFF *)ERROR_PTR("invalid file descriptor", procName, NULL); lseek(fd, 0, SEEK_SET); return TIFFFdOpen(fd, "TIFFstream", modestring); @@ -1039,8 +1029,8 @@ tiffCloseCallback(thandle_t handle) *mstream->poutdata = mstream->buffer; *mstream->poutsize = mstream->hw; } - FREE(mstream); /* never free the buffer! */ - return 0; + FREE(mstream); /* never free the buffer! */ + return 0; } @@ -1104,26 +1094,26 @@ size_t *pdatasize) PROCNAME("fopenTiffMemstream"); if (!filename) - return (TIFF *)ERROR_PTR("filename not defined", procName, NULL); - if (!operation) - return (TIFF *)ERROR_PTR("operation not defined", procName, NULL); - if (!pdata) - return (TIFF *)ERROR_PTR("&data not defined", procName, NULL); - if (!pdatasize) - return (TIFF *)ERROR_PTR("&datasize not defined", procName, NULL); - if (!strcmp(operation, "r") && !strcmp(operation, "w")) - return (TIFF *)ERROR_PTR("operation not 'r' or 'w'}", procName, NULL); - - if (!strcmp(operation, "r")) - mstream = memstreamCreateForRead(*pdata, *pdatasize); - else - mstream = memstreamCreateForWrite(pdata, pdatasize); + return (TIFF *)ERROR_PTR("filename not defined", procName, NULL); + if (!operation) + return (TIFF *)ERROR_PTR("operation not defined", procName, NULL); + if (!pdata) + return (TIFF *)ERROR_PTR("&data not defined", procName, NULL); + if (!pdatasize) + return (TIFF *)ERROR_PTR("&datasize not defined", procName, NULL); + if (!strcmp(operation, "r") && !strcmp(operation, "w")) + return (TIFF *)ERROR_PTR("operation not 'r' or 'w'}", procName, + NULL); + + if (!strcmp(operation, "r")) + mstream = memstreamCreateForRead(*pdata, *pdatasize); + else + mstream = memstreamCreateForWrite(pdata, pdatasize); - return TIFFClientOpen(filename, operation, mstream, - tiffReadCallback, tiffWriteCallback, - tiffSeekCallback, tiffCloseCallback, - tiffSizeCallback, tiffMapCallback, - tiffUnmapCallback); + return TIFFClientOpen(filename, operation, mstream, tiffReadCallback, + tiffWriteCallback, tiffSeekCallback, + tiffCloseCallback, tiffSizeCallback, + tiffMapCallback, tiffUnmapCallback); } @@ -1134,38 +1124,38 @@ OpenclDevice::pixReadMemTiffCl(const l_uint8 *data,size_t size,l_int32 n) l_int32 i, pagefound; PIX *pix; TIFF *tif; - //L_MEMSTREAM *memStream; - PROCNAME("pixReadMemTiffCl"); - - if (!data) - return (PIX *)ERROR_PTR("data pointer is NULL", procName, NULL); - - if ((tif = fopenTiffMemstream("", "r", (l_uint8 **)&data, &size)) == NULL) - return (PIX *)ERROR_PTR("tif not opened", procName, NULL); - - pagefound = FALSE; - pix = NULL; - for (i = 0; i < MAX_PAGES_IN_TIFF_FILE; i++) { - if (i == n) { - pagefound = TRUE; - if ((pix = pixReadFromTiffStreamCl(tif)) == NULL) { - TIFFCleanup(tif); - return (PIX *)ERROR_PTR("pix not read", procName, NULL); - } - break; - } - if (TIFFReadDirectory(tif) == 0) - break; - } + // L_MEMSTREAM *memStream; + PROCNAME("pixReadMemTiffCl"); - if (pagefound == FALSE) { - L_WARNING("tiff page %d not found", procName); - TIFFCleanup(tif); - return NULL; - } + if (!data) + return (PIX *)ERROR_PTR("data pointer is NULL", procName, NULL); + + if ((tif = fopenTiffMemstream("", "r", (l_uint8 **)&data, &size)) == + NULL) + return (PIX *)ERROR_PTR("tif not opened", procName, NULL); - TIFFCleanup(tif); - return pix; + pagefound = FALSE; + pix = NULL; + for (i = 0; i < MAX_PAGES_IN_TIFF_FILE; i++) { + if (i == n) { + pagefound = TRUE; + if ((pix = pixReadFromTiffStreamCl(tif)) == NULL) { + TIFFCleanup(tif); + return (PIX *)ERROR_PTR("pix not read", procName, NULL); + } + break; + } + if (TIFFReadDirectory(tif) == 0) break; + } + + if (pagefound == FALSE) { + L_WARNING("tiff page %d not found\n", procName, i); + TIFFCleanup(tif); + return NULL; + } + + TIFFCleanup(tif); + return pix; } PIX * @@ -1178,11 +1168,10 @@ TIFF *tif; PROCNAME("pixReadStreamTiff"); - if (!fp) - return (PIX *)ERROR_PTR("stream not defined", procName, NULL); + if (!fp) return (PIX *)ERROR_PTR("stream not defined", procName, NULL); if ((tif = fopenTiffCl(fp, "rb")) == NULL) - return (PIX *)ERROR_PTR("tif not opened", procName, NULL); + return (PIX *)ERROR_PTR("tif not opened", procName, NULL); pagefound = FALSE; pix = NULL; @@ -1190,8 +1179,8 @@ TIFF *tif; if (i == n) { pagefound = TRUE; if ((pix = pixReadFromTiffStreamCl(tif)) == NULL) { - TIFFCleanup(tif); - return (PIX *)ERROR_PTR("pix not read", procName, NULL); + TIFFCleanup(tif); + return (PIX *)ERROR_PTR("pix not read", procName, NULL); } break; } @@ -1252,7 +1241,6 @@ void compare(l_uint32 *cpu, l_uint32 *gpu,int size) } } printf("\nit matches\n"); - } //OpenCL implementation of pixReadFromTiffStream. @@ -1272,73 +1260,70 @@ PIXCMAP *cmap; PROCNAME("pixReadFromTiffStream"); - if (!tif) - return (PIX *)ERROR_PTR("tif not defined", procName, NULL); - + if (!tif) return (PIX *)ERROR_PTR("tif not defined", procName, NULL); TIFFGetFieldDefaulted(tif, TIFFTAG_BITSPERSAMPLE, &bps); TIFFGetFieldDefaulted(tif, TIFFTAG_SAMPLESPERPIXEL, &spp); bpp = bps * spp; if (bpp > 32) - return (PIX *)ERROR_PTR("can't handle bpp > 32", procName, NULL); + return (PIX *)ERROR_PTR("can't handle bpp > 32", procName, NULL); if (spp == 1) d = bps; else if (spp == 3 || spp == 4) d = 32; else - return (PIX *)ERROR_PTR("spp not in set {1,3,4}", procName, NULL); + return (PIX *)ERROR_PTR("spp not in set {1,3,4}", procName, NULL); TIFFGetField(tif, TIFFTAG_IMAGEWIDTH, &w); TIFFGetField(tif, TIFFTAG_IMAGELENGTH, &h); tiffbpl = TIFFScanlineSize(tif); if ((pix = pixCreate(w, h, d)) == NULL) - return (PIX *)ERROR_PTR("pix not made", procName, NULL); + return (PIX *)ERROR_PTR("pix not made", procName, NULL); data = (l_uint8 *)pixGetData(pix); wpl = pixGetWpl(pix); bpl = 4 * wpl; - if (spp == 1) { - if ((linebuf = (l_uint8 *)CALLOC(tiffbpl + 1, sizeof(l_uint8))) == NULL) - return (PIX *)ERROR_PTR("calloc fail for linebuf", procName, NULL); - - for (i = 0 ; i < h ; i++) { - if (TIFFReadScanline(tif, linebuf, i, 0) < 0) { - FREE(linebuf); - pixDestroy(&pix); - return (PIX *)ERROR_PTR("line read fail", procName, NULL); - } - memcpy((char *)data, (char *)linebuf, tiffbpl); - data += bpl; + if ((linebuf = (l_uint8 *)CALLOC(tiffbpl + 1, sizeof(l_uint8))) == + NULL) + return (PIX *)ERROR_PTR("calloc fail for linebuf", procName, NULL); + + for (i = 0; i < h; i++) { + if (TIFFReadScanline(tif, linebuf, i, 0) < 0) { + FREE(linebuf); + pixDestroy(&pix); + return (PIX *)ERROR_PTR("line read fail", procName, NULL); + } + memcpy((char *)data, (char *)linebuf, tiffbpl); + data += bpl; } if (bps <= 8) pixEndianByteSwap(pix); else - pixEndianTwoByteSwap(pix); + pixEndianTwoByteSwap(pix); FREE(linebuf); - } - else { - if ((tiffdata = (l_uint32 *)CALLOC(w * h, sizeof(l_uint32))) == NULL) { - pixDestroy(&pix); - return (PIX *)ERROR_PTR("calloc fail for tiffdata", procName, NULL); - } - if (!TIFFReadRGBAImageOriented(tif, w, h, (uint32 *)tiffdata, - ORIENTATION_TOPLEFT, 0)) { - FREE(tiffdata); - pixDestroy(&pix); - return (PIX *)ERROR_PTR("failed to read tiffdata", procName, NULL); - } - line = pixGetData(pix); + } else { + if ((tiffdata = (l_uint32 *)CALLOC(w * h, sizeof(l_uint32))) == NULL) { + pixDestroy(&pix); + return (PIX *)ERROR_PTR("calloc fail for tiffdata", procName, NULL); + } + if (!TIFFReadRGBAImageOriented(tif, w, h, (uint32 *)tiffdata, + ORIENTATION_TOPLEFT, 0)) { + FREE(tiffdata); + pixDestroy(&pix); + return (PIX *)ERROR_PTR("failed to read tiffdata", procName, NULL); + } + line = pixGetData(pix); - //Invoke the OpenCL kernel for pixReadFromTiff - l_uint32* output_gpu=pixReadFromTiffKernel(tiffdata,w,h,wpl,line); + // Invoke the OpenCL kernel for pixReadFromTiff + l_uint32 *output_gpu = pixReadFromTiffKernel(tiffdata, w, h, wpl, line); - pixSetData(pix, output_gpu); - // pix already has data allocated, it now points to output_gpu? - FREE(tiffdata); - FREE(line); - //FREE(output_gpu); + pixSetData(pix, output_gpu); + // pix already has data allocated, it now points to output_gpu? + FREE(tiffdata); + FREE(line); + // FREE(output_gpu); } if (getTiffStreamResolutionCl(tif, &xres, &yres) == 0) { @@ -1352,32 +1337,28 @@ PIXCMAP *cmap; pixSetInputFormat(pix, comptype); if (TIFFGetField(tif, TIFFTAG_COLORMAP, &redmap, &greenmap, &bluemap)) { - - if ((cmap = pixcmapCreate(bps)) == NULL) { - pixDestroy(&pix); - return (PIX *)ERROR_PTR("cmap not made", procName, NULL); + if ((cmap = pixcmapCreate(bps)) == NULL) { + pixDestroy(&pix); + return (PIX *)ERROR_PTR("cmap not made", procName, NULL); } ncolors = 1 << bps; for (i = 0; i < ncolors; i++) pixcmapAddColor(cmap, redmap[i] >> 8, greenmap[i] >> 8, bluemap[i] >> 8); pixSetColormap(pix, cmap); - } - else { - if (!TIFFGetField(tif, TIFFTAG_PHOTOMETRIC, &photometry)) { - - if (tiffcomp == COMPRESSION_CCITTFAX3 || - tiffcomp == COMPRESSION_CCITTFAX4 || - tiffcomp == COMPRESSION_CCITTRLE || - tiffcomp == COMPRESSION_CCITTRLEW) { - photometry = PHOTOMETRIC_MINISWHITE; - } - else - photometry = PHOTOMETRIC_MINISBLACK; - } - if ((d == 1 && photometry == PHOTOMETRIC_MINISBLACK) || - (d == 8 && photometry == PHOTOMETRIC_MINISWHITE)) - pixInvert(pix, pix); + } else { + if (!TIFFGetField(tif, TIFFTAG_PHOTOMETRIC, &photometry)) { + if (tiffcomp == COMPRESSION_CCITTFAX3 || + tiffcomp == COMPRESSION_CCITTFAX4 || + tiffcomp == COMPRESSION_CCITTRLE || + tiffcomp == COMPRESSION_CCITTRLEW) { + photometry = PHOTOMETRIC_MINISWHITE; + } else + photometry = PHOTOMETRIC_MINISBLACK; + } + if ((d == 1 && photometry == PHOTOMETRIC_MINISBLACK) || + (d == 8 && photometry == PHOTOMETRIC_MINISWHITE)) + pixInvert(pix, pix); } if (TIFFGetField(tif, TIFFTAG_ORIENTATION, &orientation)) { @@ -1415,6 +1396,7 @@ pixDilateCL_55(l_int32 wpl, l_int32 h) localThreads[1] = GROUPSIZE_HMORY; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateHor_5x5", &status ); + CHECK_OPENCL(status, "clCreateKernel morphoDilateHor_5x5"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -1424,24 +1406,12 @@ pixDilateCL_55(l_int32 wpl, l_int32 h) 1, sizeof(cl_mem), &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(h), - (const void *)&h); - - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(h), &h); + + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); //Swap source and dest buffers pixtemp = pixsCLBuffer; @@ -1457,6 +1427,7 @@ pixDilateCL_55(l_int32 wpl, l_int32 h) localThreads[1] = GROUPSIZE_Y; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateVer_5x5", &status ); + CHECK_OPENCL(status, "clCreateKernel morphoDilateVer_5x5"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -1466,23 +1437,11 @@ pixDilateCL_55(l_int32 wpl, l_int32 h) 1, sizeof(cl_mem), &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(h), - (const void *)&h); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(h), &h); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); return status; } @@ -1509,6 +1468,7 @@ pixErodeCL_55(l_int32 wpl, l_int32 h) localThreads[1] = GROUPSIZE_HMORY; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoErodeHor_5x5", &status ); + CHECK_OPENCL(status, "clCreateKernel morphoErodeHor_5x5"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -1518,24 +1478,12 @@ pixErodeCL_55(l_int32 wpl, l_int32 h) 1, sizeof(cl_mem), &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(h), - (const void *)&h); - - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(h), &h); + + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); //Swap source and dest buffers pixtemp = pixsCLBuffer; @@ -1551,6 +1499,7 @@ pixErodeCL_55(l_int32 wpl, l_int32 h) localThreads[1] = GROUPSIZE_Y; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoErodeVer_5x5", &status ); + CHECK_OPENCL(status, "clCreateKernel morphoErodeVer_5x5"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -1560,31 +1509,13 @@ pixErodeCL_55(l_int32 wpl, l_int32 h) 1, sizeof(cl_mem), &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(fwmask), - (const void *)&fwmask); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(lwmask), - (const void *)&lwmask); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(h), &h); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(fwmask), &fwmask); + status = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(lwmask), &lwmask); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); return status; } @@ -1625,101 +1556,56 @@ pixDilateCL(l_int32 hsize, l_int32 vsize, l_int32 wpl, l_int32 h) if (xp > 31 || xn > 31) { - //Generic case. - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateHor", &status ); - - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &pixsCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(xp), - (const void *)&xp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(xn), - (const void *)&xn); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(h), - (const void *)&h); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); - - if (yp > 0 || yn > 0) - { - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; + // Generic case. + rEnv.mpkKernel = + clCreateKernel(rEnv.mpkProgram, "morphoDilateHor", &status); + CHECK_OPENCL(status, "clCreateKernel morphoDilateHor"); + + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &pixsCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &pixdCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(xp), &xp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(xn), &xn); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(h), &h); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); + + if (yp > 0 || yn > 0) { + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; } } else if (xp > 0 || xn > 0 ) { - //Specific Horizontal pass kernel for half width < 32 - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateHor_32word", &status ); - isEven = (xp != xn); - - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &pixsCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(xp), - (const void *)&xp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(isEven), - (const void *)&isEven); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); - - if (yp > 0 || yn > 0) - { - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; - } + // Specific Horizontal pass kernel for half width < 32 + rEnv.mpkKernel = + clCreateKernel(rEnv.mpkProgram, "morphoDilateHor_32word", &status); + CHECK_OPENCL(status, "clCreateKernel morphoDilateHor_32word"); + isEven = (xp != xn); + + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &pixsCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &pixdCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(xp), &xp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(isEven), &isEven); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); + + if (yp > 0 || yn > 0) { + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; + } } if (yp > 0 || yn > 0) { rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoDilateVer", &status ); + CHECK_OPENCL(status, "clCreateKernel morphoDilateVer"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -1729,280 +1615,161 @@ pixDilateCL(l_int32 hsize, l_int32 vsize, l_int32 wpl, l_int32 h) 1, sizeof(cl_mem), &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(yp), - (const void *)&yp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(yn), - (const void *)&yn); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(yp), &yp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(yn), &yn); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); } - return status; } //Morphology Erode operation. Invokes the relevant OpenCL kernels -cl_int -pixErodeCL(l_int32 hsize, l_int32 vsize, l_uint32 wpl, l_uint32 h) -{ - - l_int32 xp, yp, xn, yn; - SEL* sel; - size_t globalThreads[2]; - size_t localThreads[2]; - cl_mem pixtemp; - cl_int status; - int gsize; - char isAsymmetric = (MORPH_BC == ASYMMETRIC_MORPH_BC); - l_uint32 rwmask, lwmask; - char isEven; - - sel = selCreateBrick(vsize, hsize, vsize / 2, hsize / 2, SEL_HIT); - - selFindMaxTranslations(sel, &xp, &yp, &xn, &yn); - selDestroy(&sel); - OpenclDevice::SetKernelEnv( &rEnv ); - - if (hsize == 5 && vsize == 5 && isAsymmetric) - { - //Specific kernel for 5x5 - status = pixErodeCL_55(wpl, h); - return status; - } - - lwmask = lmask32[31 - (xn & 31)]; - rwmask = rmask32[31 - (xp & 31)]; - - //global and local work dimensions for Horizontal pass - gsize = (wpl + GROUPSIZE_X - 1)/ GROUPSIZE_X * GROUPSIZE_X; - globalThreads[0] = gsize; - gsize = (h + GROUPSIZE_Y - 1)/ GROUPSIZE_Y * GROUPSIZE_Y; - globalThreads[1] = gsize; - localThreads[0] = GROUPSIZE_X; - localThreads[1] = GROUPSIZE_Y; - - //Horizontal Pass - if (xp > 31 || xn > 31 ) - { - //Generic case. - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoErodeHor", &status ); - - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &pixsCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(xp), - (const void *)&xp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(xn), - (const void *)&xn); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 6, - sizeof(isAsymmetric), - (const void *)&isAsymmetric); - status = clSetKernelArg(rEnv.mpkKernel, - 7, - sizeof(rwmask), - (const void *)&rwmask); - status = clSetKernelArg(rEnv.mpkKernel, - 8, - sizeof(lwmask), - (const void *)&lwmask); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); - - if (yp > 0 || yn > 0) - { - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; - } - } - else if (xp > 0 || xn > 0) - { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoErodeHor_32word", &status ); - isEven = (xp != xn); - - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &pixsCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(xp), - (const void *)&xp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(isAsymmetric), - (const void *)&isAsymmetric); - status = clSetKernelArg(rEnv.mpkKernel, - 6, - sizeof(rwmask), - (const void *)&rwmask); - status = clSetKernelArg(rEnv.mpkKernel, - 7, - sizeof(lwmask), - (const void *)&lwmask); - status = clSetKernelArg(rEnv.mpkKernel, - 8, - sizeof(isEven), - (const void *)&isEven); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); +cl_int pixErodeCL(l_int32 hsize, l_int32 vsize, l_uint32 wpl, l_uint32 h) { + l_int32 xp, yp, xn, yn; + SEL *sel; + size_t globalThreads[2]; + size_t localThreads[2]; + cl_mem pixtemp; + cl_int status; + int gsize; + char isAsymmetric = (MORPH_BC == ASYMMETRIC_MORPH_BC); + l_uint32 rwmask, lwmask; + char isEven; + + sel = selCreateBrick(vsize, hsize, vsize / 2, hsize / 2, SEL_HIT); + + selFindMaxTranslations(sel, &xp, &yp, &xn, &yn); + selDestroy(&sel); + OpenclDevice::SetKernelEnv(&rEnv); + + if (hsize == 5 && vsize == 5 && isAsymmetric) { + // Specific kernel for 5x5 + status = pixErodeCL_55(wpl, h); + return status; + } - if (yp > 0 || yn > 0) - { - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; - } + lwmask = lmask32[31 - (xn & 31)]; + rwmask = rmask32[31 - (xp & 31)]; + + // global and local work dimensions for Horizontal pass + gsize = (wpl + GROUPSIZE_X - 1) / GROUPSIZE_X * GROUPSIZE_X; + globalThreads[0] = gsize; + gsize = (h + GROUPSIZE_Y - 1) / GROUPSIZE_Y * GROUPSIZE_Y; + globalThreads[1] = gsize; + localThreads[0] = GROUPSIZE_X; + localThreads[1] = GROUPSIZE_Y; + + // Horizontal Pass + if (xp > 31 || xn > 31) { + // Generic case. + rEnv.mpkKernel = clCreateKernel(rEnv.mpkProgram, "morphoErodeHor", &status); + + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &pixsCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &pixdCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(xp), &xp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(xn), &xn); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(h), &h); + status = + clSetKernelArg(rEnv.mpkKernel, 6, sizeof(isAsymmetric), &isAsymmetric); + status = clSetKernelArg(rEnv.mpkKernel, 7, sizeof(rwmask), &rwmask); + status = clSetKernelArg(rEnv.mpkKernel, 8, sizeof(lwmask), &lwmask); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); + + if (yp > 0 || yn > 0) { + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; + } + } else if (xp > 0 || xn > 0) { + rEnv.mpkKernel = + clCreateKernel(rEnv.mpkProgram, "morphoErodeHor_32word", &status); + isEven = (xp != xn); + + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &pixsCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &pixdCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(xp), &xp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = + clSetKernelArg(rEnv.mpkKernel, 5, sizeof(isAsymmetric), &isAsymmetric); + status = clSetKernelArg(rEnv.mpkKernel, 6, sizeof(rwmask), &rwmask); + status = clSetKernelArg(rEnv.mpkKernel, 7, sizeof(lwmask), &lwmask); + status = clSetKernelArg(rEnv.mpkKernel, 8, sizeof(isEven), &isEven); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); + + if (yp > 0 || yn > 0) { + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; } + } - //Vertical Pass - if (yp > 0 || yn > 0) - { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "morphoErodeVer", &status ); - - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &pixsCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &pixdCLBuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(yp), - (const void *)&yp); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clSetKernelArg(rEnv.mpkKernel, - 5, - sizeof(isAsymmetric), - (const void *)&isAsymmetric); - status = clSetKernelArg(rEnv.mpkKernel, - 6, - sizeof(yn), - (const void *)&yn); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); - } + // Vertical Pass + if (yp > 0 || yn > 0) { + rEnv.mpkKernel = clCreateKernel(rEnv.mpkProgram, "morphoErodeVer", &status); + CHECK_OPENCL(status, "clCreateKernel morphoErodeVer"); + + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &pixsCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &pixdCLBuffer); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(yp), &yp); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = + clSetKernelArg(rEnv.mpkKernel, 5, sizeof(isAsymmetric), &isAsymmetric); + status = clSetKernelArg(rEnv.mpkKernel, 6, sizeof(yn), &yn); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); + } - return status; + return status; } // OpenCL implementation of Morphology Dilate //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixDilateBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, l_int32 vsize, bool reqDataCopy = false) -{ - l_uint32 wpl, h; +PIX *OpenclDevice::pixDilateBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, + l_int32 vsize, bool reqDataCopy = false) { + l_uint32 wpl, h; - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); - clStatus = pixDilateCL(hsize, vsize, wpl, h); + clStatus = pixDilateCL(hsize, vsize, wpl, h); - if (reqDataCopy) - { - pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl*h, CL_MAP_READ, false); - } + if (reqDataCopy) { + pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl * h, + CL_MAP_READ, false); + } - return pixd; + return pixd; } // OpenCL implementation of Morphology Erode //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixErodeBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, l_int32 vsize, bool reqDataCopy = false) -{ - l_uint32 wpl, h; +PIX *OpenclDevice::pixErodeBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, + l_int32 vsize, bool reqDataCopy = false) { + l_uint32 wpl, h; - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); - clStatus = pixErodeCL(hsize, vsize, wpl, h); + clStatus = pixErodeCL(hsize, vsize, wpl, h); - if (reqDataCopy) - { - pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl*h, CL_MAP_READ); - } + if (reqDataCopy) { + pixd = + mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl * h, CL_MAP_READ); + } - return pixd; + return pixd; } //Morphology Open operation. Invokes the relevant OpenCL kernels @@ -2045,50 +1812,40 @@ pixCloseCL(l_int32 hsize, l_int32 vsize, l_int32 wpl, l_int32 h) // OpenCL implementation of Morphology Close //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixCloseBrickCL(PIX *pixd, - PIX *pixs, - l_int32 hsize, - l_int32 vsize, - bool reqDataCopy = false) -{ - l_uint32 wpl, h; +PIX *OpenclDevice::pixCloseBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, + l_int32 vsize, bool reqDataCopy = false) { + l_uint32 wpl, h; - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); - clStatus = pixCloseCL(hsize, vsize, wpl, h); + clStatus = pixCloseCL(hsize, vsize, wpl, h); - if (reqDataCopy) - { - pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl*h, CL_MAP_READ); - } + if (reqDataCopy) { + pixd = + mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl * h, CL_MAP_READ); + } - return pixd; + return pixd; } // OpenCL implementation of Morphology Open //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixOpenBrickCL(PIX *pixd, - PIX *pixs, - l_int32 hsize, - l_int32 vsize, - bool reqDataCopy = false) -{ - l_uint32 wpl, h; +PIX *OpenclDevice::pixOpenBrickCL(PIX *pixd, PIX *pixs, l_int32 hsize, + l_int32 vsize, bool reqDataCopy = false) { + l_uint32 wpl, h; - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); - clStatus = pixOpenCL(hsize, vsize, wpl, h); + clStatus = pixOpenCL(hsize, vsize, wpl, h); - if (reqDataCopy) - { - pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl*h, CL_MAP_READ); - } + if (reqDataCopy) { + pixd = + mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl * h, CL_MAP_READ); + } - return pixd; + return pixd; } //pix OR operation: outbuffer = buffer1 | buffer2 @@ -2106,6 +1863,7 @@ pixORCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, cl_mem buffer2, cl_mem ou globalThreads[1] = gsize; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "pixOR", &status ); + CHECK_OPENCL(status, "clCreateKernel pixOR"); status = clSetKernelArg(rEnv.mpkKernel, 0, @@ -2119,23 +1877,11 @@ pixORCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, cl_mem buffer2, cl_mem ou 2, sizeof(cl_mem), &outbuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); return status; } @@ -2155,6 +1901,7 @@ pixANDCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, cl_mem buffer2, cl_mem o globalThreads[1] = gsize; rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "pixAND", &status ); + CHECK_OPENCL(status, "clCreateKernel pixAND"); // Enqueue a kernel run call. status = clSetKernelArg(rEnv.mpkKernel, @@ -2169,102 +1916,64 @@ pixANDCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, cl_mem buffer2, cl_mem o 2, sizeof(cl_mem), &outbuffer); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(h), - (const void *)&h); - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(h), &h); + status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, + NULL, globalThreads, localThreads, 0, + NULL, NULL); return status; } //output = buffer1 & ~(buffer2) -cl_int -pixSubtractCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, cl_mem buffer2, cl_mem outBuffer = NULL) -{ - cl_int status; - size_t globalThreads[2]; - int gsize; - size_t localThreads[] = {GROUPSIZE_X, GROUPSIZE_Y}; - - gsize = (wpl + GROUPSIZE_X - 1)/ GROUPSIZE_X * GROUPSIZE_X; - globalThreads[0] = gsize; - gsize = (h + GROUPSIZE_Y - 1)/ GROUPSIZE_Y * GROUPSIZE_Y; - globalThreads[1] = gsize; - - if (outBuffer != NULL) - { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "pixSubtract", &status ); - } - else - { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "pixSubtract_inplace", &status ); - } +cl_int pixSubtractCL_work(l_uint32 wpl, l_uint32 h, cl_mem buffer1, + cl_mem buffer2, cl_mem outBuffer = NULL) { + cl_int status; + size_t globalThreads[2]; + int gsize; + size_t localThreads[] = {GROUPSIZE_X, GROUPSIZE_Y}; + + gsize = (wpl + GROUPSIZE_X - 1) / GROUPSIZE_X * GROUPSIZE_X; + globalThreads[0] = gsize; + gsize = (h + GROUPSIZE_Y - 1) / GROUPSIZE_Y * GROUPSIZE_Y; + globalThreads[1] = gsize; + + if (outBuffer != NULL) { + rEnv.mpkKernel = clCreateKernel(rEnv.mpkProgram, "pixSubtract", &status); + CHECK_OPENCL(status, "clCreateKernel pixSubtract"); + } else { + rEnv.mpkKernel = + clCreateKernel(rEnv.mpkProgram, "pixSubtract_inplace", &status); + CHECK_OPENCL(status, "clCreateKernel pixSubtract_inplace"); + } - // Enqueue a kernel run call. - status = clSetKernelArg(rEnv.mpkKernel, - 0, - sizeof(cl_mem), - &buffer1); - status = clSetKernelArg(rEnv.mpkKernel, - 1, - sizeof(cl_mem), - &buffer2); - status = clSetKernelArg(rEnv.mpkKernel, - 2, - sizeof(wpl), - (const void *)&wpl); - status = clSetKernelArg(rEnv.mpkKernel, - 3, - sizeof(h), - (const void *)&h); - if (outBuffer != NULL) - { - status = clSetKernelArg(rEnv.mpkKernel, - 4, - sizeof(cl_mem), - (const void *)&outBuffer); - } - status = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 2, - NULL, - globalThreads, - localThreads, - 0, - NULL, - NULL); + // Enqueue a kernel run call. + status = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &buffer1); + status = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(cl_mem), &buffer2); + status = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(wpl), &wpl); + status = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(h), &h); + if (outBuffer != NULL) { + status = clSetKernelArg(rEnv.mpkKernel, 4, sizeof(cl_mem), &outBuffer); + } + status = + clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 2, NULL, + globalThreads, localThreads, 0, NULL, NULL); - return status; + return status; } // OpenCL implementation of Subtract pix //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixSubtractCL(PIX *pixd, PIX *pixs1, PIX *pixs2, bool reqDataCopy = false) -{ - l_uint32 wpl, h; +PIX *OpenclDevice::pixSubtractCL(PIX *pixd, PIX *pixs1, PIX *pixs2, + bool reqDataCopy = false) { + l_uint32 wpl, h; - PROCNAME("pixSubtractCL"); + PROCNAME("pixSubtractCL"); - if (!pixs1) - return (PIX *)ERROR_PTR("pixs1 not defined", procName, pixd); - if (!pixs2) - return (PIX *)ERROR_PTR("pixs2 not defined", procName, pixd); - if (pixGetDepth(pixs1) != pixGetDepth(pixs2)) - return (PIX *)ERROR_PTR("depths of pixs* unequal", procName, pixd); + if (!pixs1) return (PIX *)ERROR_PTR("pixs1 not defined", procName, pixd); + if (!pixs2) return (PIX *)ERROR_PTR("pixs2 not defined", procName, pixd); + if (pixGetDepth(pixs1) != pixGetDepth(pixs2)) + return (PIX *)ERROR_PTR("depths of pixs* unequal", procName, pixd); #if EQUAL_SIZE_WARNING if (!pixSizesEqual(pixs1, pixs2)) @@ -2287,200 +1996,206 @@ OpenclDevice::pixSubtractCL(PIX *pixd, PIX *pixs1, PIX *pixs2, bool reqDataCo // OpenCL implementation of Hollow pix //Note: Assumes the source and dest opencl buffer are initialized. No check done -PIX* -OpenclDevice::pixHollowCL(PIX *pixd, - PIX *pixs, - l_int32 close_hsize, - l_int32 close_vsize, - l_int32 open_hsize, - l_int32 open_vsize, - bool reqDataCopy = false) -{ - l_uint32 wpl, h; - cl_mem pixtemp; - - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); - - //First step : Close Morph operation: Dilate followed by Erode - clStatus = pixCloseCL(close_hsize, close_vsize, wpl, h); - - //Store the output of close operation in an intermediate buffer - //this will be later used for pixsubtract - clStatus = clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, 0, sizeof(int) * wpl*h, 0, NULL, NULL); - - //Second step: Open Operation - Erode followed by Dilate - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; - - clStatus = pixOpenCL(open_hsize, open_vsize, wpl, h); - - //Third step: Subtract : (Close - Open) - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixdCLIntermediate; - pixdCLIntermediate = pixtemp; - - clStatus = pixSubtractCL_work(wpl, h, pixdCLBuffer, pixsCLBuffer); - - if (reqDataCopy) - { - //Read back output data from OCL buffer to cpu - pixd = mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl*h, CL_MAP_READ); - } - return pixd; +PIX *OpenclDevice::pixHollowCL(PIX *pixd, PIX *pixs, l_int32 close_hsize, + l_int32 close_vsize, l_int32 open_hsize, + l_int32 open_vsize, bool reqDataCopy = false) { + l_uint32 wpl, h; + cl_mem pixtemp; + + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); + + // First step : Close Morph operation: Dilate followed by Erode + clStatus = pixCloseCL(close_hsize, close_vsize, wpl, h); + + // Store the output of close operation in an intermediate buffer + // this will be later used for pixsubtract + clStatus = + clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, + 0, sizeof(int) * wpl * h, 0, NULL, NULL); + + // Second step: Open Operation - Erode followed by Dilate + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; + + clStatus = pixOpenCL(open_hsize, open_vsize, wpl, h); + + // Third step: Subtract : (Close - Open) + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixdCLIntermediate; + pixdCLIntermediate = pixtemp; + + clStatus = pixSubtractCL_work(wpl, h, pixdCLBuffer, pixsCLBuffer); + + if (reqDataCopy) { + // Read back output data from OCL buffer to cpu + pixd = + mapOutputCLBuffer(rEnv, pixdCLBuffer, pixd, pixs, wpl * h, CL_MAP_READ); + } + return pixd; } // OpenCL implementation of Get Lines from pix function //Note: Assumes the source and dest opencl buffer are initialized. No check done -void -OpenclDevice::pixGetLinesCL(PIX *pixd, - PIX *pixs, - PIX** pix_vline, - PIX** pix_hline, - PIX** pixClosed, - bool getpixClosed, - l_int32 close_hsize, l_int32 close_vsize, - l_int32 open_hsize, l_int32 open_vsize, - l_int32 line_hsize, l_int32 line_vsize) -{ - l_uint32 wpl, h; - cl_mem pixtemp; - - wpl = pixGetWpl(pixs); - h = pixGetHeight(pixs); - - //First step : Close Morph operation: Dilate followed by Erode - clStatus = pixCloseCL(close_hsize, close_vsize, wpl, h); - - //Copy the Close output to CPU buffer - if (getpixClosed) - { - *pixClosed = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pixClosed, pixs, wpl*h, CL_MAP_READ, true, false); - } +void OpenclDevice::pixGetLinesCL(PIX *pixd, PIX *pixs, PIX **pix_vline, + PIX **pix_hline, PIX **pixClosed, + bool getpixClosed, l_int32 close_hsize, + l_int32 close_vsize, l_int32 open_hsize, + l_int32 open_vsize, l_int32 line_hsize, + l_int32 line_vsize) { + l_uint32 wpl, h; + cl_mem pixtemp; + + wpl = pixGetWpl(pixs); + h = pixGetHeight(pixs); + + // First step : Close Morph operation: Dilate followed by Erode + clStatus = pixCloseCL(close_hsize, close_vsize, wpl, h); + + // Copy the Close output to CPU buffer + if (getpixClosed) { + *pixClosed = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pixClosed, pixs, + wpl * h, CL_MAP_READ, true, false); + } - //Store the output of close operation in an intermediate buffer - //this will be later used for pixsubtract - clStatus = clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, 0, sizeof(int) * wpl*h, 0, NULL, NULL); + // Store the output of close operation in an intermediate buffer + // this will be later used for pixsubtract + clStatus = + clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, + 0, sizeof(int) * wpl * h, 0, NULL, NULL); - //Second step: Open Operation - Erode followed by Dilate - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; + // Second step: Open Operation - Erode followed by Dilate + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; - clStatus = pixOpenCL(open_hsize, open_vsize, wpl, h); + clStatus = pixOpenCL(open_hsize, open_vsize, wpl, h); - //Third step: Subtract : (Close - Open) - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixdCLIntermediate; - pixdCLIntermediate = pixtemp; + // Third step: Subtract : (Close - Open) + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixdCLIntermediate; + pixdCLIntermediate = pixtemp; - clStatus = pixSubtractCL_work(wpl, h, pixdCLBuffer, pixsCLBuffer); + clStatus = pixSubtractCL_work(wpl, h, pixdCLBuffer, pixsCLBuffer); - //Store the output of Hollow operation in an intermediate buffer - //this will be later used - clStatus = clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, 0, sizeof(int) * wpl*h, 0, NULL, NULL); + // Store the output of Hollow operation in an intermediate buffer + // this will be later used + clStatus = + clEnqueueCopyBuffer(rEnv.mpkCmdQueue, pixdCLBuffer, pixdCLIntermediate, 0, + 0, sizeof(int) * wpl * h, 0, NULL, NULL); - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLBuffer; - pixdCLBuffer = pixtemp; + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLBuffer; + pixdCLBuffer = pixtemp; - //Fourth step: Get vertical line - //pixOpenBrick(NULL, pix_hollow, 1, min_line_length); - clStatus = pixOpenCL(1, line_vsize, wpl, h); + // Fourth step: Get vertical line + // pixOpenBrick(NULL, pix_hollow, 1, min_line_length); + clStatus = pixOpenCL(1, line_vsize, wpl, h); - //Copy the vertical line output to CPU buffer - *pix_vline = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pix_vline, pixs, wpl*h, CL_MAP_READ, true, false); + // Copy the vertical line output to CPU buffer + *pix_vline = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pix_vline, pixs, wpl * h, + CL_MAP_READ, true, false); - pixtemp = pixsCLBuffer; - pixsCLBuffer = pixdCLIntermediate; - pixdCLIntermediate = pixtemp; + pixtemp = pixsCLBuffer; + pixsCLBuffer = pixdCLIntermediate; + pixdCLIntermediate = pixtemp; - //Fifth step: Get horizontal line - //pixOpenBrick(NULL, pix_hollow, min_line_length, 1); - clStatus = pixOpenCL(line_hsize, 1, wpl, h); + // Fifth step: Get horizontal line + // pixOpenBrick(NULL, pix_hollow, min_line_length, 1); + clStatus = pixOpenCL(line_hsize, 1, wpl, h); - //Copy the horizontal line output to CPU buffer - *pix_hline = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pix_hline, pixs, wpl*h, CL_MAP_READ, true, true); + // Copy the horizontal line output to CPU buffer + *pix_hline = mapOutputCLBuffer(rEnv, pixdCLBuffer, *pix_hline, pixs, wpl * h, + CL_MAP_READ, true, true); - return; + return; } - /************************************************************************* * HistogramRect * Otsu Thresholding Operations * histogramAllChannels is laid out as all channel 0, then all channel 1... * only supports 1 or 4 channels (bytes_per_pixel) ************************************************************************/ -int OpenclDevice::HistogramRectOCL( - const unsigned char* imageData, - int bytes_per_pixel, - int bytes_per_line, - int left, // always 0 - int top, // always 0 - int width, - int height, - int kHistogramSize, - int* histogramAllChannels) -{ -PERF_COUNT_START("HistogramRectOCL") - cl_int clStatus; - int retVal= 0; - KernelEnv histKern; - SetKernelEnv( &histKern ); - KernelEnv histRedKern; - SetKernelEnv( &histRedKern ); - /* map imagedata to device as read only */ - // USE_HOST_PTR uses onion+ bus which is slowest option; also happens to be coherent which we don't need. - // faster option would be to allocate initial image buffer - // using a garlic bus memory type - cl_mem imageBuffer = clCreateBuffer( histKern.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, width*height*bytes_per_pixel*sizeof(char), (void *)imageData, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer imageBuffer"); - - /* setup work group size parameters */ - int block_size = 256; - cl_uint numCUs; - clStatus = clGetDeviceInfo( gpuEnv.mpDevID, CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(numCUs), &numCUs, NULL); - CHECK_OPENCL( clStatus, "clCreateBuffer imageBuffer"); - - int requestedOccupancy = 10; - int numWorkGroups = numCUs * requestedOccupancy; - int numThreads = block_size*numWorkGroups; - size_t local_work_size[] = {static_cast(block_size)}; - size_t global_work_size[] = {static_cast(numThreads)}; - size_t red_global_work_size[] = {static_cast(block_size*kHistogramSize*bytes_per_pixel)}; - - /* map histogramAllChannels as write only */ - int numBins = kHistogramSize*bytes_per_pixel*numWorkGroups; - - cl_mem histogramBuffer = clCreateBuffer( histKern.mpkContext, CL_MEM_READ_WRITE | CL_MEM_USE_HOST_PTR, kHistogramSize*bytes_per_pixel*sizeof(int), (void *)histogramAllChannels, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer histogramBuffer"); - - /* intermediate histogram buffer */ - int histRed = 256; - int tmpHistogramBins = kHistogramSize*bytes_per_pixel*histRed; - - cl_mem tmpHistogramBuffer = clCreateBuffer( histKern.mpkContext, CL_MEM_READ_WRITE, tmpHistogramBins*sizeof(cl_uint), NULL, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer tmpHistogramBuffer"); - - /* atomic sync buffer */ - int *zeroBuffer = new int[1]; - zeroBuffer[0] = 0; - cl_mem atomicSyncBuffer = clCreateBuffer( histKern.mpkContext, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, sizeof(cl_int), (void *)zeroBuffer, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer atomicSyncBuffer"); - delete[] zeroBuffer; - //Create kernel objects based on bytes_per_pixel - if (bytes_per_pixel == 1) - { - histKern.mpkKernel = clCreateKernel( histKern.mpkProgram, "kernel_HistogramRectOneChannel", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel kernel_HistogramRectOneChannel"); - - histRedKern.mpkKernel = clCreateKernel( histRedKern.mpkProgram, "kernel_HistogramRectOneChannelReduction", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel kernel_HistogramRectOneChannelReduction"); - } else { +int OpenclDevice::HistogramRectOCL(unsigned char *imageData, + int bytes_per_pixel, int bytes_per_line, + int left, // always 0 + int top, // always 0 + int width, int height, int kHistogramSize, + int *histogramAllChannels) { + PERF_COUNT_START("HistogramRectOCL") + cl_int clStatus; + int retVal = 0; + KernelEnv histKern; + SetKernelEnv(&histKern); + KernelEnv histRedKern; + SetKernelEnv(&histRedKern); + /* map imagedata to device as read only */ + // USE_HOST_PTR uses onion+ bus which is slowest option; also happens to be + // coherent which we don't need. + // faster option would be to allocate initial image buffer + // using a garlic bus memory type + cl_mem imageBuffer = clCreateBuffer( + histKern.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, + width * height * bytes_per_pixel * sizeof(char), imageData, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer imageBuffer"); + + /* setup work group size parameters */ + int block_size = 256; + cl_uint numCUs; + clStatus = clGetDeviceInfo(gpuEnv.mpDevID, CL_DEVICE_MAX_COMPUTE_UNITS, + sizeof(numCUs), &numCUs, NULL); + CHECK_OPENCL(clStatus, "clCreateBuffer imageBuffer"); + + int requestedOccupancy = 10; + int numWorkGroups = numCUs * requestedOccupancy; + int numThreads = block_size * numWorkGroups; + size_t local_work_size[] = {static_cast(block_size)}; + size_t global_work_size[] = {static_cast(numThreads)}; + size_t red_global_work_size[] = { + static_cast(block_size * kHistogramSize * bytes_per_pixel)}; + + /* map histogramAllChannels as write only */ + int numBins = kHistogramSize * bytes_per_pixel * numWorkGroups; + + cl_mem histogramBuffer = clCreateBuffer( + histKern.mpkContext, CL_MEM_READ_WRITE | CL_MEM_USE_HOST_PTR, + kHistogramSize * bytes_per_pixel * sizeof(int), histogramAllChannels, + &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer histogramBuffer"); + + /* intermediate histogram buffer */ + int histRed = 256; + int tmpHistogramBins = kHistogramSize * bytes_per_pixel * histRed; + + cl_mem tmpHistogramBuffer = + clCreateBuffer(histKern.mpkContext, CL_MEM_READ_WRITE, + tmpHistogramBins * sizeof(cl_uint), NULL, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer tmpHistogramBuffer"); + + /* atomic sync buffer */ + int *zeroBuffer = new int[1]; + zeroBuffer[0] = 0; + cl_mem atomicSyncBuffer = clCreateBuffer( + histKern.mpkContext, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, + sizeof(cl_int), zeroBuffer, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer atomicSyncBuffer"); + delete[] zeroBuffer; + // Create kernel objects based on bytes_per_pixel + if (bytes_per_pixel == 1) { + histKern.mpkKernel = clCreateKernel( + histKern.mpkProgram, "kernel_HistogramRectOneChannel", &clStatus); + CHECK_OPENCL(clStatus, "clCreateKernel kernel_HistogramRectOneChannel"); + + histRedKern.mpkKernel = + clCreateKernel(histRedKern.mpkProgram, + "kernel_HistogramRectOneChannelReduction", &clStatus); + CHECK_OPENCL(clStatus, + "clCreateKernel kernel_HistogramRectOneChannelReduction"); + } else { histKern.mpkKernel = clCreateKernel( histKern.mpkProgram, "kernel_HistogramRectAllChannels", &clStatus ); CHECK_OPENCL( clStatus, "clCreateKernel kernel_HistogramRectAllChannels"); @@ -2491,72 +2206,77 @@ PERF_COUNT_START("HistogramRectOCL") void *ptr; //Initialize tmpHistogramBuffer buffer - ptr = clEnqueueMapBuffer(histKern.mpkCmdQueue, tmpHistogramBuffer, CL_TRUE, CL_MAP_WRITE, 0, tmpHistogramBins*sizeof(cl_uint), 0, NULL, NULL, &clStatus); + ptr = clEnqueueMapBuffer( + histKern.mpkCmdQueue, tmpHistogramBuffer, CL_TRUE, CL_MAP_WRITE, 0, + tmpHistogramBins * sizeof(cl_uint), 0, NULL, NULL, &clStatus); CHECK_OPENCL( clStatus, "clEnqueueMapBuffer tmpHistogramBuffer"); memset(ptr, 0, tmpHistogramBins*sizeof(cl_uint)); - clEnqueueUnmapMemObject(histKern.mpkCmdQueue, tmpHistogramBuffer, ptr, 0, NULL, NULL); + clEnqueueUnmapMemObject(histKern.mpkCmdQueue, tmpHistogramBuffer, ptr, 0, + NULL, NULL); /* set kernel 1 arguments */ - clStatus = clSetKernelArg( histKern.mpkKernel, 0, sizeof(cl_mem), (void *)&imageBuffer ); + clStatus = + clSetKernelArg(histKern.mpkKernel, 0, sizeof(cl_mem), &imageBuffer); CHECK_OPENCL( clStatus, "clSetKernelArg imageBuffer"); cl_uint numPixels = width*height; - clStatus = clSetKernelArg( histKern.mpkKernel, 1, sizeof(cl_uint), (void *)&numPixels ); + clStatus = + clSetKernelArg(histKern.mpkKernel, 1, sizeof(cl_uint), &numPixels); CHECK_OPENCL( clStatus, "clSetKernelArg numPixels" ); - clStatus = clSetKernelArg( histKern.mpkKernel, 2, sizeof(cl_mem), (void *)&tmpHistogramBuffer ); + clStatus = clSetKernelArg(histKern.mpkKernel, 2, sizeof(cl_mem), + &tmpHistogramBuffer); CHECK_OPENCL( clStatus, "clSetKernelArg tmpHistogramBuffer"); /* set kernel 2 arguments */ int n = numThreads/bytes_per_pixel; - clStatus = clSetKernelArg( histRedKern.mpkKernel, 0, sizeof(cl_int), (void *)&n ); + clStatus = clSetKernelArg(histRedKern.mpkKernel, 0, sizeof(cl_int), &n); CHECK_OPENCL( clStatus, "clSetKernelArg imageBuffer"); - clStatus = clSetKernelArg( histRedKern.mpkKernel, 1, sizeof(cl_mem), (void *)&tmpHistogramBuffer ); + clStatus = clSetKernelArg(histRedKern.mpkKernel, 1, sizeof(cl_mem), + &tmpHistogramBuffer); CHECK_OPENCL( clStatus, "clSetKernelArg tmpHistogramBuffer"); - clStatus = clSetKernelArg( histRedKern.mpkKernel, 2, sizeof(cl_mem), (void *)&histogramBuffer ); + clStatus = clSetKernelArg(histRedKern.mpkKernel, 2, sizeof(cl_mem), + &histogramBuffer); CHECK_OPENCL( clStatus, "clSetKernelArg histogramBuffer"); /* launch histogram */ PERF_COUNT_SUB("before") - clStatus = clEnqueueNDRangeKernel( - histKern.mpkCmdQueue, - histKern.mpkKernel, - 1, NULL, global_work_size, local_work_size, - 0, NULL, NULL ); - CHECK_OPENCL( clStatus, "clEnqueueNDRangeKernel kernel_HistogramRectAllChannels" ); - clFinish( histKern.mpkCmdQueue ); - if(clStatus !=0) - { - retVal = -1; - } +clStatus = clEnqueueNDRangeKernel(histKern.mpkCmdQueue, histKern.mpkKernel, 1, + NULL, global_work_size, local_work_size, 0, + NULL, NULL); +CHECK_OPENCL(clStatus, + "clEnqueueNDRangeKernel kernel_HistogramRectAllChannels"); +clFinish(histKern.mpkCmdQueue); +if (clStatus != 0) { + retVal = -1; + } /* launch histogram */ clStatus = clEnqueueNDRangeKernel( - histRedKern.mpkCmdQueue, - histRedKern.mpkKernel, - 1, NULL, red_global_work_size, local_work_size, - 0, NULL, NULL ); + histRedKern.mpkCmdQueue, histRedKern.mpkKernel, 1, NULL, + red_global_work_size, local_work_size, 0, NULL, NULL); CHECK_OPENCL( clStatus, "clEnqueueNDRangeKernel kernel_HistogramRectAllChannelsReduction" ); clFinish( histRedKern.mpkCmdQueue ); - if(clStatus !=0) - { - retVal = -1; - } -PERF_COUNT_SUB("redKernel") + if (clStatus != 0) { + retVal = -1; + } + PERF_COUNT_SUB("redKernel") /* map results back from gpu */ - ptr = clEnqueueMapBuffer(histRedKern.mpkCmdQueue, histogramBuffer, CL_TRUE, CL_MAP_READ, 0, kHistogramSize*bytes_per_pixel*sizeof(int), 0, NULL, NULL, &clStatus); + ptr = clEnqueueMapBuffer(histRedKern.mpkCmdQueue, histogramBuffer, CL_TRUE, + CL_MAP_READ, 0, + kHistogramSize * bytes_per_pixel * sizeof(int), 0, + NULL, NULL, &clStatus); CHECK_OPENCL( clStatus, "clEnqueueMapBuffer histogramBuffer"); - if(clStatus !=0) - { - retVal = -1; - } - clEnqueueUnmapMemObject(histRedKern.mpkCmdQueue, histogramBuffer, ptr, 0, NULL, NULL); + if (clStatus != 0) { + retVal = -1; + } + clEnqueueUnmapMemObject(histRedKern.mpkCmdQueue, histogramBuffer, ptr, 0, + NULL, NULL); clReleaseMemObject(histogramBuffer); clReleaseMemObject(imageBuffer); PERF_COUNT_SUB("after") PERF_COUNT_END - return retVal; - +return retVal; } /************************************************************************* @@ -2564,111 +2284,118 @@ PERF_COUNT_END * from the class, using thresholds/hi_values to the output IMAGE. * only supports 1 or 4 channels ************************************************************************/ -int OpenclDevice::ThresholdRectToPixOCL( - const unsigned char* imageData, - int bytes_per_pixel, - int bytes_per_line, - const int* thresholds, - const int* hi_values, - Pix** pix, - int height, - int width, - int top, - int left) { -PERF_COUNT_START("ThresholdRectToPixOCL") - int retVal =0; - /* create pix result buffer */ - *pix = pixCreate(width, height, 1); - uinT32* pixData = pixGetData(*pix); - int wpl = pixGetWpl(*pix); - int pixSize = wpl*height*sizeof(uinT32); // number of pixels - - cl_int clStatus; - KernelEnv rEnv; - SetKernelEnv( &rEnv ); - - /* setup work group size parameters */ - int block_size = 256; - cl_uint numCUs = 6; - clStatus = clGetDeviceInfo( gpuEnv.mpDevID, CL_DEVICE_MAX_COMPUTE_UNITS, sizeof(numCUs), &numCUs, NULL); - CHECK_OPENCL( clStatus, "clCreateBuffer imageBuffer"); - - int requestedOccupancy = 10; - int numWorkGroups = numCUs * requestedOccupancy; - int numThreads = block_size*numWorkGroups; - size_t local_work_size[] = {(size_t) block_size}; - size_t global_work_size[] = {(size_t) numThreads}; - - /* map imagedata to device as read only */ - // USE_HOST_PTR uses onion+ bus which is slowest option; also happens to be coherent which we don't need. - // faster option would be to allocate initial image buffer - // using a garlic bus memory type - cl_mem imageBuffer = clCreateBuffer( rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, width*height*bytes_per_pixel*sizeof(char), (void *)imageData, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer imageBuffer"); - - /* map pix as write only */ - pixThBuffer = clCreateBuffer( rEnv.mpkContext, CL_MEM_READ_WRITE | CL_MEM_USE_HOST_PTR, pixSize, (void *)pixData, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer pix"); - - /* map thresholds and hi_values */ - cl_mem thresholdsBuffer = clCreateBuffer( rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, bytes_per_pixel*sizeof(int), (void *)thresholds, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer thresholdBuffer"); - cl_mem hiValuesBuffer = clCreateBuffer( rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, bytes_per_pixel*sizeof(int), (void *)hi_values, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer hiValuesBuffer"); - - /* compile kernel */ - if (bytes_per_pixel == 4) { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "kernel_ThresholdRectToPix", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel kernel_ThresholdRectToPix"); - } else { - rEnv.mpkKernel = clCreateKernel( rEnv.mpkProgram, "kernel_ThresholdRectToPix_OneChan", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel kernel_ThresholdRectToPix_OneChan"); - } +int OpenclDevice::ThresholdRectToPixOCL(unsigned char *imageData, + int bytes_per_pixel, int bytes_per_line, + int *thresholds, int *hi_values, + Pix **pix, int height, int width, + int top, int left) { + PERF_COUNT_START("ThresholdRectToPixOCL") + int retVal = 0; + /* create pix result buffer */ + *pix = pixCreate(width, height, 1); + uinT32 *pixData = pixGetData(*pix); + int wpl = pixGetWpl(*pix); + int pixSize = wpl * height * sizeof(uinT32); // number of pixels + + cl_int clStatus; + KernelEnv rEnv; + SetKernelEnv(&rEnv); + + /* setup work group size parameters */ + int block_size = 256; + cl_uint numCUs = 6; + clStatus = clGetDeviceInfo(gpuEnv.mpDevID, CL_DEVICE_MAX_COMPUTE_UNITS, + sizeof(numCUs), &numCUs, NULL); + CHECK_OPENCL(clStatus, "clCreateBuffer imageBuffer"); + + int requestedOccupancy = 10; + int numWorkGroups = numCUs * requestedOccupancy; + int numThreads = block_size * numWorkGroups; + size_t local_work_size[] = {(size_t)block_size}; + size_t global_work_size[] = {(size_t)numThreads}; + + /* map imagedata to device as read only */ + // USE_HOST_PTR uses onion+ bus which is slowest option; also happens to be + // coherent which we don't need. + // faster option would be to allocate initial image buffer + // using a garlic bus memory type + cl_mem imageBuffer = clCreateBuffer( + rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, + width * height * bytes_per_pixel * sizeof(char), imageData, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer imageBuffer"); + + /* map pix as write only */ + pixThBuffer = + clCreateBuffer(rEnv.mpkContext, CL_MEM_READ_WRITE | CL_MEM_USE_HOST_PTR, + pixSize, pixData, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer pix"); + + /* map thresholds and hi_values */ + cl_mem thresholdsBuffer = + clCreateBuffer(rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, + bytes_per_pixel * sizeof(int), thresholds, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer thresholdBuffer"); + cl_mem hiValuesBuffer = + clCreateBuffer(rEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, + bytes_per_pixel * sizeof(int), hi_values, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer hiValuesBuffer"); + + /* compile kernel */ + if (bytes_per_pixel == 4) { + rEnv.mpkKernel = + clCreateKernel(rEnv.mpkProgram, "kernel_ThresholdRectToPix", &clStatus); + CHECK_OPENCL(clStatus, "clCreateKernel kernel_ThresholdRectToPix"); + } else { + rEnv.mpkKernel = clCreateKernel( + rEnv.mpkProgram, "kernel_ThresholdRectToPix_OneChan", &clStatus); + CHECK_OPENCL(clStatus, "clCreateKernel kernel_ThresholdRectToPix_OneChan"); + } - /* set kernel arguments */ - clStatus = clSetKernelArg( rEnv.mpkKernel, 0, sizeof(cl_mem), (void *)&imageBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg imageBuffer"); - cl_uint numPixels = width*height; - clStatus = clSetKernelArg( rEnv.mpkKernel, 1, sizeof(int), (void *)&height ); - CHECK_OPENCL( clStatus, "clSetKernelArg height" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 2, sizeof(int), (void *)&width ); - CHECK_OPENCL( clStatus, "clSetKernelArg width" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 3, sizeof(int), (void *)&wpl ); - CHECK_OPENCL( clStatus, "clSetKernelArg wpl" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 4, sizeof(cl_mem), (void *)&thresholdsBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg thresholdsBuffer" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 5, sizeof(cl_mem), (void *)&hiValuesBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg hiValuesBuffer" ); - clStatus = clSetKernelArg( rEnv.mpkKernel, 6, sizeof(cl_mem), (void *)&pixThBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg pixThBuffer"); - - /* launch kernel & wait */ -PERF_COUNT_SUB("before") - clStatus = clEnqueueNDRangeKernel( - rEnv.mpkCmdQueue, - rEnv.mpkKernel, - 1, NULL, global_work_size, local_work_size, - 0, NULL, NULL ); - CHECK_OPENCL( clStatus, "clEnqueueNDRangeKernel kernel_ThresholdRectToPix" ); - clFinish( rEnv.mpkCmdQueue ); -PERF_COUNT_SUB("kernel") - if(clStatus !=0) - { - printf("Setting return value to -1\n"); - retVal = -1; - } - /* map results back from gpu */ - void *ptr = clEnqueueMapBuffer(rEnv.mpkCmdQueue, pixThBuffer, CL_TRUE, CL_MAP_READ, 0, pixSize, 0, NULL, NULL, &clStatus); - CHECK_OPENCL( clStatus, "clEnqueueMapBuffer histogramBuffer"); - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, pixThBuffer, ptr, 0, NULL, NULL); + /* set kernel arguments */ + clStatus = clSetKernelArg(rEnv.mpkKernel, 0, sizeof(cl_mem), &imageBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg imageBuffer"); + cl_uint numPixels = width * height; + clStatus = clSetKernelArg(rEnv.mpkKernel, 1, sizeof(int), &height); + CHECK_OPENCL(clStatus, "clSetKernelArg height"); + clStatus = clSetKernelArg(rEnv.mpkKernel, 2, sizeof(int), &width); + CHECK_OPENCL(clStatus, "clSetKernelArg width"); + clStatus = clSetKernelArg(rEnv.mpkKernel, 3, sizeof(int), &wpl); + CHECK_OPENCL(clStatus, "clSetKernelArg wpl"); + clStatus = + clSetKernelArg(rEnv.mpkKernel, 4, sizeof(cl_mem), &thresholdsBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg thresholdsBuffer"); + clStatus = clSetKernelArg(rEnv.mpkKernel, 5, sizeof(cl_mem), &hiValuesBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg hiValuesBuffer"); + clStatus = clSetKernelArg(rEnv.mpkKernel, 6, sizeof(cl_mem), &pixThBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg pixThBuffer"); + + /* launch kernel & wait */ + PERF_COUNT_SUB("before") + clStatus = clEnqueueNDRangeKernel(rEnv.mpkCmdQueue, rEnv.mpkKernel, 1, + NULL, global_work_size, local_work_size, + 0, NULL, NULL); + CHECK_OPENCL(clStatus, "clEnqueueNDRangeKernel kernel_ThresholdRectToPix"); + clFinish(rEnv.mpkCmdQueue); + PERF_COUNT_SUB("kernel") + if (clStatus != 0) { + printf("Setting return value to -1\n"); + retVal = -1; + } + /* map results back from gpu */ + void *ptr = + clEnqueueMapBuffer(rEnv.mpkCmdQueue, pixThBuffer, CL_TRUE, CL_MAP_READ, 0, + pixSize, 0, NULL, NULL, &clStatus); + CHECK_OPENCL(clStatus, "clEnqueueMapBuffer histogramBuffer"); + clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, pixThBuffer, ptr, 0, NULL, + NULL); - clReleaseMemObject(imageBuffer); - clReleaseMemObject(thresholdsBuffer); - clReleaseMemObject(hiValuesBuffer); + clReleaseMemObject(imageBuffer); + clReleaseMemObject(thresholdsBuffer); + clReleaseMemObject(hiValuesBuffer); -PERF_COUNT_SUB("after") -PERF_COUNT_END -return retVal; + PERF_COUNT_SUB("after") + PERF_COUNT_END + return retVal; } @@ -2745,7 +2472,6 @@ void populateTessScoreEvaluationInputData( TessScoreEvaluationInputData *input ) float fractionBlack = 0.1; // how much of the image should be blackened int numSpots = (height*width)*fractionBlack/(maxLineWidth*maxLineWidth/2/2); for (int i = 0; i < numSpots; i++) { - int lineWidth = rand()%maxLineWidth; int col = lineWidth + rand()%(width-2*lineWidth); int row = lineWidth + rand()%(height-2*lineWidth); @@ -2776,15 +2502,14 @@ typedef struct _TessDeviceScore { *****************************************************************************/ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData input, ds_device_type type ) { - double time = 0; #if ON_WINDOWS LARGE_INTEGER freq, time_funct_start, time_funct_end; QueryPerformanceFrequency(&freq); #elif ON_APPLE - mach_timebase_info_data_t info = { 0, 0 }; + mach_timebase_info_data_t info = {0, 0}; mach_timebase_info(&info); - long long start,stop; + long long start, stop; #else timespec time_funct_start, time_funct_end; #endif @@ -2795,21 +2520,22 @@ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData inpu if (type == DS_DEVICE_OPENCL_DEVICE) { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif OpenclDevice::gpuEnv = *env; int wpl = pixGetWpl(input.pix); - OpenclDevice::pixReadFromTiffKernel(tiffdata, input.width, input.height, wpl, NULL); + OpenclDevice::pixReadFromTiffKernel(tiffdata, input.width, input.height, + wpl, NULL); #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -2818,8 +2544,8 @@ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData inpu } else { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif @@ -2832,7 +2558,6 @@ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData inpu int idx = 0; for (i = 0; i < input.height ; i++) { for (j = 0; j < input.width; j++) { - l_uint32 tiffword = tiffdata[i * input.width + j]; l_int32 rval = ((tiffword) & 0xff); l_int32 gval = (((tiffword) >> 8) & 0xff); @@ -2845,9 +2570,9 @@ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData inpu #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -2862,15 +2587,14 @@ double composeRGBPixelMicroBench( GPUEnv *env, TessScoreEvaluationInputData inpu } double histogramRectMicroBench( GPUEnv *env, TessScoreEvaluationInputData input, ds_device_type type ) { - double time; #if ON_WINDOWS LARGE_INTEGER freq, time_funct_start, time_funct_end; QueryPerformanceFrequency(&freq); -#elif ON_APPLE - mach_timebase_info_data_t info = { 0, 0 }; +#elif ON_APPLE + mach_timebase_info_data_t info = {0, 0}; mach_timebase_info(&info); - long long start,stop; + long long start, stop; #else timespec time_funct_start, time_funct_end; #endif @@ -2882,58 +2606,56 @@ double histogramRectMicroBench( GPUEnv *env, TessScoreEvaluationInputData input, int kHistogramSize = 256; int bytes_per_line = input.width*input.numChannels; int *histogramAllChannels = new int[kHistogramSize*input.numChannels]; - int retVal= 0; + int retVal = 0; // function call if (type == DS_DEVICE_OPENCL_DEVICE) { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif OpenclDevice::gpuEnv = *env; int wpl = pixGetWpl(input.pix); - retVal= OpenclDevice::HistogramRectOCL(input.imageData, input.numChannels, bytes_per_line, top, left, input.width, input.height, kHistogramSize, histogramAllChannels); + retVal = OpenclDevice::HistogramRectOCL( + input.imageData, input.numChannels, bytes_per_line, top, left, + input.width, input.height, kHistogramSize, histogramAllChannels); #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - if(retVal ==0) - { - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; - } - else - { - time= FLT_MAX; - } +#elif ON_APPLE + stop = mach_absolute_time(); + if (retVal == 0) { + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; + } else { + time = FLT_MAX; + } #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; #endif } else { - int *histogram = new int[kHistogramSize]; #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif for (int ch = 0; ch < input.numChannels; ++ch) { - tesseract::HistogramRect(input.pix, input.numChannels, - left, top, input.width, input.height, histogram); + tesseract::HistogramRect(input.pix, input.numChannels, left, top, + input.width, input.height, histogram); } #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -2959,13 +2681,13 @@ void ThresholdRectToPix_Native(const unsigned char* imagedata, int height = pixGetHeight(*pix); *pix = pixCreate(width, height, 1); - uinT32* pixdata = pixGetData(*pix); + uinT32 *pixdata = pixGetData(*pix); int wpl = pixGetWpl(*pix); const unsigned char* srcdata = imagedata + top * bytes_per_line + left * bytes_per_pixel; for (int y = 0; y < height; ++y) { - const uinT8* linedata = srcdata; - uinT32* pixline = pixdata + y * wpl; + const uinT8 *linedata = srcdata; + uinT32 *pixline = pixdata + y * wpl; for (int x = 0; x < width; ++x, linedata += bytes_per_pixel) { bool white_result = true; for (int ch = 0; ch < bytes_per_pixel; ++ch) { @@ -2985,16 +2707,15 @@ void ThresholdRectToPix_Native(const unsigned char* imagedata, } double thresholdRectToPixMicroBench( GPUEnv *env, TessScoreEvaluationInputData input, ds_device_type type ) { - double time; - int retVal =0; + int retVal = 0; #if ON_WINDOWS LARGE_INTEGER freq, time_funct_start, time_funct_end; QueryPerformanceFrequency(&freq); -#elif ON_APPLE - mach_timebase_info_data_t info = { 0, 0 }; +#elif ON_APPLE + mach_timebase_info_data_t info = {0, 0}; mach_timebase_info(&info); - long long start,stop; + long long start, stop; #else timespec time_funct_start, time_funct_end; #endif @@ -3020,29 +2741,29 @@ double thresholdRectToPixMicroBench( GPUEnv *env, TessScoreEvaluationInputData i if (type == DS_DEVICE_OPENCL_DEVICE) { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif OpenclDevice::gpuEnv = *env; int wpl = pixGetWpl(input.pix); - retVal= OpenclDevice::ThresholdRectToPixOCL(input.imageData, input.numChannels, bytes_per_line, thresholds, hi_values, &input.pix, input.height, input.width, top, left); + retVal = OpenclDevice::ThresholdRectToPixOCL( + input.imageData, input.numChannels, bytes_per_line, thresholds, + hi_values, &input.pix, input.height, input.width, top, left); #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - if(retVal ==0) - { - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9;; - } - else - { - time= FLT_MAX; - } +#elif ON_APPLE + stop = mach_absolute_time(); + if (retVal == 0) { + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; + ; + } else { + time = FLT_MAX; + } #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); @@ -3055,8 +2776,8 @@ double thresholdRectToPixMicroBench( GPUEnv *env, TessScoreEvaluationInputData i thresholder.SetImage( input.pix ); #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif @@ -3066,9 +2787,9 @@ double thresholdRectToPixMicroBench( GPUEnv *env, TessScoreEvaluationInputData i #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -3087,10 +2808,10 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in #if ON_WINDOWS LARGE_INTEGER freq, time_funct_start, time_funct_end; QueryPerformanceFrequency(&freq); -#elif ON_APPLE - mach_timebase_info_data_t info = { 0, 0 }; +#elif ON_APPLE + mach_timebase_info_data_t info = {0, 0}; mach_timebase_info(&info); - long long start,stop; + long long start, stop; #else timespec time_funct_start, time_funct_end; #endif @@ -3108,8 +2829,8 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in if (type == DS_DEVICE_OPENCL_DEVICE) { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif @@ -3117,16 +2838,19 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in OpenclDevice::gpuEnv = *env; OpenclDevice::initMorphCLAllocations(wpl, input.height, input.pix); Pix *pix_vline = NULL, *pix_hline = NULL, *pix_closed = NULL; - OpenclDevice::pixGetLinesCL(NULL, input.pix, &pix_vline, &pix_hline, &pix_closed, true, closing_brick, closing_brick, max_line_width, max_line_width, min_line_length, min_line_length); + OpenclDevice::pixGetLinesCL( + NULL, input.pix, &pix_vline, &pix_hline, &pix_closed, true, + closing_brick, closing_brick, max_line_width, max_line_width, + min_line_length, min_line_length); OpenclDevice::releaseMorphCLBuffers(); #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -3134,16 +2858,18 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in } else { #if ON_WINDOWS QueryPerformanceCounter(&time_funct_start); -#elif ON_APPLE - start = mach_absolute_time(); +#elif ON_APPLE + start = mach_absolute_time(); #else clock_gettime( CLOCK_MONOTONIC, &time_funct_start ); #endif // native serial code Pix *src_pix = input.pix; - Pix *pix_closed = pixCloseBrick(NULL, src_pix, closing_brick, closing_brick); - Pix *pix_solid = pixOpenBrick(NULL, pix_closed, max_line_width, max_line_width); + Pix *pix_closed = + pixCloseBrick(NULL, src_pix, closing_brick, closing_brick); + Pix *pix_solid = + pixOpenBrick(NULL, pix_closed, max_line_width, max_line_width); Pix *pix_hollow = pixSubtract(NULL, pix_closed, pix_solid); pixDestroy(&pix_solid); Pix *pix_vline = pixOpenBrick(NULL, pix_hollow, 1, min_line_length); @@ -3153,9 +2879,9 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in #if ON_WINDOWS QueryPerformanceCounter(&time_funct_end); time = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); -#elif ON_APPLE - stop = mach_absolute_time(); - time = ((stop - start) * (double) info.numer / info.denom) / 1.0E9; +#elif ON_APPLE + stop = mach_absolute_time(); + time = ((stop - start) * (double)info.numer / info.denom) / 1.0E9; #else clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); time = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; @@ -3173,11 +2899,10 @@ double getLineMasksMorphMicroBench( GPUEnv *env, TessScoreEvaluationInputData in #include "stdlib.h" - // encode score object as byte string ds_status serializeScore( ds_device* device, void **serializedScore, unsigned int* serializedScoreSize ) { *serializedScoreSize = sizeof(TessDeviceScore); - *serializedScore = (void *) new unsigned char[*serializedScoreSize]; + *serializedScore = new unsigned char[*serializedScoreSize]; memcpy(*serializedScore, device->score, *serializedScoreSize); return DS_SUCCESS; } @@ -3190,14 +2915,13 @@ ds_status deserializeScore( ds_device* device, const unsigned char* serializedSc return DS_SUCCESS; } -ds_status releaseScore( void* score ) { +ds_status releaseScore(void *score) { delete (TessDeviceScore *)score; return DS_SUCCESS; } // evaluate devices ds_status evaluateScoreForDevice( ds_device *device, void *inputData) { - // overwrite statuc gpuEnv w/ current device // so native opencl calls can be used; they use static gpuEnv printf("\n[DS] Device: \"%s\" (%s) evaluation...\n", device->oclDeviceName, device->type==DS_DEVICE_OPENCL_DEVICE ? "OpenCL" : "Native" ); @@ -3233,15 +2957,13 @@ ds_status evaluateScoreForDevice( ds_device *device, void *inputData) { float composeRGBPixelWeight = 1.2f; float histogramRectWeight = 2.4f; float thresholdRectToPixWeight = 4.5f; - float getLineMasksMorphWeight = 5.0f; - - float weightedTime = - composeRGBPixelWeight * composeRGBPixelTime + - histogramRectWeight * histogramRectTime + - thresholdRectToPixWeight * thresholdRectToPixTime + - getLineMasksMorphWeight * getLineMasksMorphTime - ; - device->score = (void *)new TessDeviceScore; + float getLineMasksMorphWeight = 5.0f; + + float weightedTime = composeRGBPixelWeight * composeRGBPixelTime + + histogramRectWeight * histogramRectTime + + thresholdRectToPixWeight * thresholdRectToPixTime + + getLineMasksMorphWeight * getLineMasksMorphTime; + device->score = new TessDeviceScore; ((TessDeviceScore *)device->score)->time = weightedTime; printf("[DS] Device: \"%s\" (%s) evaluated\n", device->oclDeviceName, device->type==DS_DEVICE_OPENCL_DEVICE ? "OpenCL" : "Native" ); @@ -3256,83 +2978,104 @@ ds_status evaluateScoreForDevice( ds_device *device, void *inputData) { // initial call to select device ds_device OpenclDevice::getDeviceSelection( ) { if (!deviceIsSelected) { -PERF_COUNT_START("getDeviceSelection") - // check if opencl is available at runtime - if( 1 == LoadOpencl() ) { - // opencl is available -//PERF_COUNT_SUB("LoadOpencl") - // setup devices - ds_status status; - ds_profile *profile; - status = initDSProfile( &profile, "v0.1" ); -PERF_COUNT_SUB("initDSProfile") - // try reading scores from file - const char *fileName = "tesseract_opencl_profile_devices.dat"; - status = readProfileFromFile( profile, deserializeScore, fileName); - if (status != DS_SUCCESS) { - // need to run evaluation - printf("[DS] Profile file not available (%s); performing profiling.\n", fileName); - - // create input data - TessScoreEvaluationInputData input; - populateTessScoreEvaluationInputData( &input ); -//PERF_COUNT_SUB("populateTessScoreEvaluationInputData") - // perform evaluations - unsigned int numUpdates; - status = profileDevices( profile, DS_EVALUATE_ALL, evaluateScoreForDevice, (void *)&input, &numUpdates ); -PERF_COUNT_SUB("profileDevices") - // write scores to file - if ( status == DS_SUCCESS ) { - status = writeProfileToFile( profile, serializeScore, fileName); -PERF_COUNT_SUB("writeProfileToFile") - if ( status == DS_SUCCESS ) { - printf("[DS] Scores written to file (%s).\n", fileName); + PERF_COUNT_START("getDeviceSelection") + // check if opencl is available at runtime + if (1 == LoadOpencl()) { + // opencl is available + // PERF_COUNT_SUB("LoadOpencl") + // setup devices + ds_status status; + ds_profile *profile; + status = initDSProfile(&profile, "v0.1"); + PERF_COUNT_SUB("initDSProfile") + // try reading scores from file + const char *fileName = "tesseract_opencl_profile_devices.dat"; + status = readProfileFromFile(profile, deserializeScore, fileName); + if (status != DS_SUCCESS) { + // need to run evaluation + printf("[DS] Profile file not available (%s); performing profiling.\n", + fileName); + + // create input data + TessScoreEvaluationInputData input; + populateTessScoreEvaluationInputData(&input); + // PERF_COUNT_SUB("populateTessScoreEvaluationInputData") + // perform evaluations + unsigned int numUpdates; + status = profileDevices(profile, DS_EVALUATE_ALL, + evaluateScoreForDevice, &input, &numUpdates); + PERF_COUNT_SUB("profileDevices") + // write scores to file + if (status == DS_SUCCESS) { + status = writeProfileToFile(profile, serializeScore, fileName); + PERF_COUNT_SUB("writeProfileToFile") + if (status == DS_SUCCESS) { + printf("[DS] Scores written to file (%s).\n", fileName); + } else { + printf( + "[DS] Error saving scores to file (%s); scores not written to " + "file.\n", + fileName); + } } else { - printf("[DS] Error saving scores to file (%s); scores not written to file.\n", fileName); + printf( + "[DS] Unable to evaluate performance; scores not written to " + "file.\n"); } } else { - printf("[DS] Unable to evaluate performance; scores not written to file.\n"); + PERF_COUNT_SUB("readProfileFromFile") + printf("[DS] Profile read from file (%s).\n", fileName); } - } else { -PERF_COUNT_SUB("readProfileFromFile") - printf("[DS] Profile read from file (%s).\n", fileName); - } - - // we now have device scores either from file or evaluation - // select fastest using custom Tesseract selection algorithm - float bestTime = FLT_MAX; // begin search with worst possible time - int bestDeviceIdx = -1; - for (int d = 0; d < profile->numDevices; d++) { - ds_device device = profile->devices[d]; - TessDeviceScore score = *(TessDeviceScore *)device.score; - - float time = score.time; - printf("[DS] Device[%i] %i:%s score is %f\n", d+1, device.type, device.oclDeviceName, time); - if (time < bestTime) { - bestTime = time; + // we now have device scores either from file or evaluation + // select fastest using custom Tesseract selection algorithm + float bestTime = FLT_MAX; // begin search with worst possible time + int bestDeviceIdx = -1; + for (int d = 0; d < profile->numDevices; d++) { + ds_device device = profile->devices[d]; + TessDeviceScore score = *(TessDeviceScore *)device.score; + + float time = score.time; + printf("[DS] Device[%i] %i:%s score is %f\n", d + 1, device.type, + device.oclDeviceName, time); + if (time < bestTime) { + bestTime = time; bestDeviceIdx = d; + } } - } - printf("[DS] Selected Device[%i]: \"%s\" (%s)\n", bestDeviceIdx+1, profile->devices[bestDeviceIdx].oclDeviceName, profile->devices[bestDeviceIdx].type==DS_DEVICE_OPENCL_DEVICE ? "OpenCL" : "Native"); - // cleanup - // TODO: call destructor for profile object? + printf("[DS] Selected Device[%i]: \"%s\" (%s)\n", bestDeviceIdx + 1, + profile->devices[bestDeviceIdx].oclDeviceName, + profile->devices[bestDeviceIdx].type == DS_DEVICE_OPENCL_DEVICE + ? "OpenCL" + : "Native"); + // cleanup + // TODO: call destructor for profile object? - bool overrided = false; + bool overridden = false; char *overrideDeviceStr = getenv("TESSERACT_OPENCL_DEVICE"); if (overrideDeviceStr != NULL) { int overrideDeviceIdx = atoi(overrideDeviceStr); - if (overrideDeviceIdx > 0 && overrideDeviceIdx <= profile->numDevices ) { - printf("[DS] Overriding Device Selection (TESSERACT_OPENCL_DEVICE=%s, %i)\n", overrideDeviceStr, overrideDeviceIdx); + if (overrideDeviceIdx > 0 && overrideDeviceIdx <= profile->numDevices) { + printf( + "[DS] Overriding Device Selection (TESSERACT_OPENCL_DEVICE=%s, " + "%i)\n", + overrideDeviceStr, overrideDeviceIdx); bestDeviceIdx = overrideDeviceIdx - 1; - overrided = true; + overridden = true; } else { - printf("[DS] Ignoring invalid TESSERACT_OPENCL_DEVICE=%s ([1,%i] are valid devices).\n", overrideDeviceStr, profile->numDevices); + printf( + "[DS] Ignoring invalid TESSERACT_OPENCL_DEVICE=%s ([1,%i] are " + "valid devices).\n", + overrideDeviceStr, profile->numDevices); } } - if (overrided) { - printf("[DS] Overridden Device[%i]: \"%s\" (%s)\n", bestDeviceIdx+1, profile->devices[bestDeviceIdx].oclDeviceName, profile->devices[bestDeviceIdx].type==DS_DEVICE_OPENCL_DEVICE ? "OpenCL" : "Native"); + if (overridden) { + printf("[DS] Overridden Device[%i]: \"%s\" (%s)\n", bestDeviceIdx + 1, + profile->devices[bestDeviceIdx].oclDeviceName, + profile->devices[bestDeviceIdx].type == DS_DEVICE_OPENCL_DEVICE + ? "OpenCL" + : "Native"); } selectedDevice = profile->devices[bestDeviceIdx]; // cleanup @@ -3347,10 +3090,10 @@ PERF_COUNT_SUB("readProfileFromFile") selectedDevice.oclDriverVersion = NULL; } deviceIsSelected = true; -PERF_COUNT_SUB("select from Profile") -PERF_COUNT_END + PERF_COUNT_SUB("select from Profile") + PERF_COUNT_END } -//PERF_COUNT_END + // PERF_COUNT_END return selectedDevice; } @@ -3365,8 +3108,6 @@ bool OpenclDevice::selectedDeviceIsNativeCPU() { return (device.type == DS_DEVICE_NATIVE_CPU); } - - /*! * pixConvertRGBToGray() from leptonica, converted to opencl kernel * @@ -3378,111 +3119,112 @@ bool OpenclDevice::selectedDeviceIsNativeCPU() { * Notes: * (1) Use a weighted average of the RGB values. */ -#define SET_DATA_BYTE( pdata, n, val ) (*(l_uint8 *)((l_uintptr_t)((l_uint8 *)(pdata) + (n)) ^ 3) = (val)) +#define SET_DATA_BYTE(pdata, n, val) \ + (*(l_uint8 *)((l_uintptr_t)((l_uint8 *)(pdata) + (n)) ^ 3) = (val)) -Pix * OpenclDevice::pixConvertRGBToGrayOCL( - Pix *srcPix, // 32-bit source - float rwt, - float gwt, - float bwt ) -{ -PERF_COUNT_START("pixConvertRGBToGrayOCL") - Pix *dstPix; // 8-bit destination +Pix *OpenclDevice::pixConvertRGBToGrayOCL(Pix *srcPix, // 32-bit source + float rwt, float gwt, float bwt) { + PERF_COUNT_START("pixConvertRGBToGrayOCL") + Pix *dstPix; // 8-bit destination - if (rwt < 0.0 || gwt < 0.0 || bwt < 0.0) return NULL; - - if (rwt == 0.0 && gwt == 0.0 && bwt == 0.0) { - // magic numbers from leptonica - rwt = 0.3; - gwt = 0.5; - bwt = 0.2; - } - // normalize - float sum = rwt + gwt + bwt; - rwt /= sum; - gwt /= sum; - bwt /= sum; - - // source pix - int w, h; - pixGetDimensions(srcPix, &w, &h, NULL); - //printf("Image is %i x %i\n", w, h); - unsigned int *srcData = pixGetData(srcPix); - int srcWPL = pixGetWpl(srcPix); - int srcSize = srcWPL * h * sizeof(unsigned int); - - // destination pix - if ((dstPix = pixCreate(w, h, 8)) == NULL) - return NULL; - pixCopyResolution(dstPix, srcPix); - unsigned int *dstData = pixGetData(dstPix); - int dstWPL = pixGetWpl(dstPix); - int dstWords = dstWPL * h; - int dstSize = dstWords * sizeof(unsigned int); - //printf("dstSize = %i\n", dstSize); -PERF_COUNT_SUB("pix setup") - - // opencl objects - cl_int clStatus; - KernelEnv kEnv; - SetKernelEnv( &kEnv ); - - // source buffer - cl_mem srcBuffer = clCreateBuffer( kEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, srcSize, (void *)srcData, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer srcBuffer"); - - // destination buffer - cl_mem dstBuffer = clCreateBuffer( kEnv.mpkContext, CL_MEM_WRITE_ONLY | CL_MEM_USE_HOST_PTR, dstSize, (void *)dstData, &clStatus ); - CHECK_OPENCL( clStatus, "clCreateBuffer dstBuffer"); - - // setup work group size parameters - int block_size = 256; - int numWorkGroups = ((h*w+block_size-1) / block_size ); - int numThreads = block_size*numWorkGroups; - size_t local_work_size[] = {static_cast(block_size)}; - size_t global_work_size[] = {static_cast(numThreads)}; - //printf("Enqueueing %i threads for %i output pixels\n", numThreads, w*h); - - /* compile kernel */ - kEnv.mpkKernel = clCreateKernel( kEnv.mpkProgram, "kernel_RGBToGray", &clStatus ); - CHECK_OPENCL( clStatus, "clCreateKernel kernel_RGBToGray"); - - - /* set kernel arguments */ - clStatus = clSetKernelArg( kEnv.mpkKernel, 0, sizeof(cl_mem), (void *)&srcBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg srcBuffer"); - clStatus = clSetKernelArg( kEnv.mpkKernel, 1, sizeof(cl_mem), (void *)&dstBuffer ); - CHECK_OPENCL( clStatus, "clSetKernelArg dstBuffer"); - clStatus = clSetKernelArg( kEnv.mpkKernel, 2, sizeof(int), (void *)&srcWPL ); - CHECK_OPENCL( clStatus, "clSetKernelArg srcWPL" ); - clStatus = clSetKernelArg( kEnv.mpkKernel, 3, sizeof(int), (void *)&dstWPL ); - CHECK_OPENCL( clStatus, "clSetKernelArg dstWPL" ); - clStatus = clSetKernelArg( kEnv.mpkKernel, 4, sizeof(int), (void *)&h ); - CHECK_OPENCL( clStatus, "clSetKernelArg height" ); - clStatus = clSetKernelArg( kEnv.mpkKernel, 5, sizeof(int), (void *)&w ); - CHECK_OPENCL( clStatus, "clSetKernelArg width" ); - clStatus = clSetKernelArg( kEnv.mpkKernel, 6, sizeof(float), (void *)&rwt ); - CHECK_OPENCL( clStatus, "clSetKernelArg rwt" ); - clStatus = clSetKernelArg( kEnv.mpkKernel, 7, sizeof(float), (void *)&gwt ); - CHECK_OPENCL( clStatus, "clSetKernelArg gwt"); - clStatus = clSetKernelArg( kEnv.mpkKernel, 8, sizeof(float), (void *)&bwt ); - CHECK_OPENCL( clStatus, "clSetKernelArg bwt"); - - /* launch kernel & wait */ -PERF_COUNT_SUB("before") - clStatus = clEnqueueNDRangeKernel( - kEnv.mpkCmdQueue, - kEnv.mpkKernel, - 1, NULL, global_work_size, local_work_size, - 0, NULL, NULL ); - CHECK_OPENCL( clStatus, "clEnqueueNDRangeKernel kernel_RGBToGray" ); - clFinish( kEnv.mpkCmdQueue ); -PERF_COUNT_SUB("kernel") + if (rwt < 0.0 || gwt < 0.0 || bwt < 0.0) return NULL; - /* map results back from gpu */ - void *ptr = clEnqueueMapBuffer(kEnv.mpkCmdQueue, dstBuffer, CL_TRUE, CL_MAP_READ, 0, dstSize, 0, NULL, NULL, &clStatus); - CHECK_OPENCL( clStatus, "clEnqueueMapBuffer dstBuffer"); - clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, dstBuffer, ptr, 0, NULL, NULL); + if (rwt == 0.0 && gwt == 0.0 && bwt == 0.0) { + // magic numbers from leptonica + rwt = 0.3; + gwt = 0.5; + bwt = 0.2; + } + // normalize + float sum = rwt + gwt + bwt; + rwt /= sum; + gwt /= sum; + bwt /= sum; + + // source pix + int w, h; + pixGetDimensions(srcPix, &w, &h, NULL); + // printf("Image is %i x %i\n", w, h); + unsigned int *srcData = pixGetData(srcPix); + int srcWPL = pixGetWpl(srcPix); + int srcSize = srcWPL * h * sizeof(unsigned int); + + // destination pix + if ((dstPix = pixCreate(w, h, 8)) == NULL) return NULL; + pixCopyResolution(dstPix, srcPix); + unsigned int *dstData = pixGetData(dstPix); + int dstWPL = pixGetWpl(dstPix); + int dstWords = dstWPL * h; + int dstSize = dstWords * sizeof(unsigned int); + // printf("dstSize = %i\n", dstSize); + PERF_COUNT_SUB("pix setup") + + // opencl objects + cl_int clStatus; + KernelEnv kEnv; + SetKernelEnv(&kEnv); + + // source buffer + cl_mem srcBuffer = + clCreateBuffer(kEnv.mpkContext, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, + srcSize, srcData, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer srcBuffer"); + + // destination buffer + cl_mem dstBuffer = + clCreateBuffer(kEnv.mpkContext, CL_MEM_WRITE_ONLY | CL_MEM_USE_HOST_PTR, + dstSize, dstData, &clStatus); + CHECK_OPENCL(clStatus, "clCreateBuffer dstBuffer"); + + // setup work group size parameters + int block_size = 256; + int numWorkGroups = ((h * w + block_size - 1) / block_size); + int numThreads = block_size * numWorkGroups; + size_t local_work_size[] = {static_cast(block_size)}; + size_t global_work_size[] = {static_cast(numThreads)}; + // printf("Enqueueing %i threads for %i output pixels\n", numThreads, w*h); + + /* compile kernel */ + kEnv.mpkKernel = + clCreateKernel(kEnv.mpkProgram, "kernel_RGBToGray", &clStatus); + CHECK_OPENCL(clStatus, "clCreateKernel kernel_RGBToGray"); + + /* set kernel arguments */ + clStatus = clSetKernelArg(kEnv.mpkKernel, 0, sizeof(cl_mem), &srcBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg srcBuffer"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 1, sizeof(cl_mem), &dstBuffer); + CHECK_OPENCL(clStatus, "clSetKernelArg dstBuffer"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 2, sizeof(int), &srcWPL); + CHECK_OPENCL(clStatus, "clSetKernelArg srcWPL"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 3, sizeof(int), &dstWPL); + CHECK_OPENCL(clStatus, "clSetKernelArg dstWPL"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 4, sizeof(int), &h); + CHECK_OPENCL(clStatus, "clSetKernelArg height"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 5, sizeof(int), &w); + CHECK_OPENCL(clStatus, "clSetKernelArg width"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 6, sizeof(float), &rwt); + CHECK_OPENCL(clStatus, "clSetKernelArg rwt"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 7, sizeof(float), &gwt); + CHECK_OPENCL(clStatus, "clSetKernelArg gwt"); + clStatus = clSetKernelArg(kEnv.mpkKernel, 8, sizeof(float), &bwt); + CHECK_OPENCL(clStatus, "clSetKernelArg bwt"); + + /* launch kernel & wait */ + PERF_COUNT_SUB("before") + clStatus = clEnqueueNDRangeKernel(kEnv.mpkCmdQueue, kEnv.mpkKernel, 1, + NULL, global_work_size, local_work_size, + 0, NULL, NULL); + CHECK_OPENCL(clStatus, "clEnqueueNDRangeKernel kernel_RGBToGray"); + clFinish(kEnv.mpkCmdQueue); + PERF_COUNT_SUB("kernel") + + /* map results back from gpu */ + void *ptr = + clEnqueueMapBuffer(kEnv.mpkCmdQueue, dstBuffer, CL_TRUE, CL_MAP_READ, 0, + dstSize, 0, NULL, NULL, &clStatus); + CHECK_OPENCL(clStatus, "clEnqueueMapBuffer dstBuffer"); + clEnqueueUnmapMemObject(rEnv.mpkCmdQueue, dstBuffer, ptr, 0, NULL, + NULL); #if 0 // validate: compute on cpu @@ -3519,13 +3261,12 @@ PERF_COUNT_SUB("kernel") //printf("\n"); } #endif - // release opencl objects - clReleaseMemObject(srcBuffer); - clReleaseMemObject(dstBuffer); + // release opencl objects + clReleaseMemObject(srcBuffer); + clReleaseMemObject(dstBuffer); - -PERF_COUNT_END - // success - return dstPix; + PERF_COUNT_END + // success + return dstPix; } -#endif +#endif \ No newline at end of file diff --git a/opencl/openclwrapper.h b/opencl/openclwrapper.h index ae52a80155..e09e371dbb 100644 --- a/opencl/openclwrapper.h +++ b/opencl/openclwrapper.h @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include #include "allheaders.h" #include "pix.h" @@ -10,7 +19,8 @@ // including CL/cl.h doesn't occur until USE_OPENCL defined below // platform preprocessor commands -#if defined( WIN32 ) || defined( __WIN32__ ) || defined( _WIN32 ) || defined( __CYGWIN__ ) || defined( __MINGW32__ ) +#if defined(WIN32) || defined(__WIN32__) || defined(_WIN32) || \ + defined(__CYGWIN__) || defined(__MINGW32__) #define ON_WINDOWS 1 #define ON_LINUX 0 #define ON_APPLE 0 @@ -80,21 +90,23 @@ time_sub_start = time_funct_start; \ time_sub_end = time_funct_start; -#define PERF_COUNT_END \ - QueryPerformanceCounter(&time_funct_end); \ - elapsed_time_sec = (time_funct_end.QuadPart-time_funct_start.QuadPart)/(double)(freq.QuadPart); \ - printf(PERF_COUNT_REPORT_STR, funct_name, "total", elapsed_time_sec); +#define PERF_COUNT_END \ + QueryPerformanceCounter(&time_funct_end); \ + elapsed_time_sec = (time_funct_end.QuadPart - time_funct_start.QuadPart) / \ + (double)(freq.QuadPart); \ + printf(PERF_COUNT_REPORT_STR, funct_name, "total", elapsed_time_sec); #else #define PERF_COUNT_START(FUNCT_NAME) #define PERF_COUNT_END #endif #if PERF_COUNT_VERBOSE >= 3 -#define PERF_COUNT_SUB(SUB) \ - QueryPerformanceCounter(&time_sub_end); \ - elapsed_time_sec = (time_sub_end.QuadPart-time_sub_start.QuadPart)/(double)(freq.QuadPart); \ - printf(PERF_COUNT_REPORT_STR, funct_name, SUB, elapsed_time_sec); \ - time_sub_start = time_sub_end; +#define PERF_COUNT_SUB(SUB) \ + QueryPerformanceCounter(&time_sub_end); \ + elapsed_time_sec = (time_sub_end.QuadPart - time_sub_start.QuadPart) / \ + (double)(freq.QuadPart); \ + printf(PERF_COUNT_REPORT_STR, funct_name, SUB, elapsed_time_sec); \ + time_sub_start = time_sub_end; #else #define PERF_COUNT_SUB(SUB) #endif @@ -112,21 +124,25 @@ time_sub_start = time_funct_start; \ time_sub_end = time_funct_start; -#define PERF_COUNT_END \ - clock_gettime( CLOCK_MONOTONIC, &time_funct_end ); \ - elapsed_time_sec = (time_funct_end.tv_sec - time_funct_start.tv_sec)*1.0 + (time_funct_end.tv_nsec - time_funct_start.tv_nsec)/1000000000.0; \ - printf(PERF_COUNT_REPORT_STR, funct_name, "total", elapsed_time_sec); +#define PERF_COUNT_END \ + clock_gettime(CLOCK_MONOTONIC, &time_funct_end); \ + elapsed_time_sec = \ + (time_funct_end.tv_sec - time_funct_start.tv_sec) * 1.0 + \ + (time_funct_end.tv_nsec - time_funct_start.tv_nsec) / 1000000000.0; \ + printf(PERF_COUNT_REPORT_STR, funct_name, "total", elapsed_time_sec); #else #define PERF_COUNT_START(FUNCT_NAME) #define PERF_COUNT_END #endif #if PERF_COUNT_VERBOSE >= 3 -#define PERF_COUNT_SUB(SUB) \ - clock_gettime( CLOCK_MONOTONIC, &time_sub_end ); \ - elapsed_time_sec = (time_sub_end.tv_sec - time_sub_start.tv_sec)*1.0 + (time_sub_end.tv_nsec - time_sub_start.tv_nsec)/1000000000.0; \ - printf(PERF_COUNT_REPORT_STR, funct_name, SUB, elapsed_time_sec); \ - time_sub_start = time_sub_end; +#define PERF_COUNT_SUB(SUB) \ + clock_gettime(CLOCK_MONOTONIC, &time_sub_end); \ + elapsed_time_sec = \ + (time_sub_end.tv_sec - time_sub_start.tv_sec) * 1.0 + \ + (time_sub_end.tv_nsec - time_sub_start.tv_nsec) / 1000000000.0; \ + printf(PERF_COUNT_REPORT_STR, funct_name, SUB, elapsed_time_sec); \ + time_sub_start = time_sub_end; #else #define PERF_COUNT_SUB(SUB) #endif @@ -253,12 +269,12 @@ class OpenclDevice // OpenCL implementation of Morphology (Hollow = Closed - Open) static PIX* pixHollowCL(PIX *pixd, PIX *pixs, l_int32 close_hsize, l_int32 close_vsize, l_int32 open_hsize, l_int32 open_vsize, bool reqDataCopy); - static void pixGetLinesCL(PIX *pixd, PIX *pixs, - PIX** pix_vline, PIX** pix_hline, - PIX** pixClosed, bool getpixClosed, - l_int32 close_hsize, l_int32 close_vsize, - l_int32 open_hsize, l_int32 open_vsize, - l_int32 line_hsize, l_int32 line_vsize); + static void pixGetLinesCL(PIX *pixd, PIX *pixs, PIX **pix_vline, + PIX **pix_hline, PIX **pixClosed, + bool getpixClosed, l_int32 close_hsize, + l_int32 close_vsize, l_int32 open_hsize, + l_int32 open_vsize, l_int32 line_hsize, + l_int32 line_vsize); //int InitOpenclAttr( OpenCLEnv * env ); //int ReleaseKernel( KernelEnv * env ); @@ -279,34 +295,23 @@ class OpenclDevice static void FreeOpenclDll(); #endif - inline static int AddKernelConfig( int kCount, const char *kName ); /* for binarization */ - static int HistogramRectOCL( - const unsigned char *imagedata, - int bytes_per_pixel, - int bytes_per_line, - int left, - int top, - int width, - int height, - int kHistogramSize, - int *histogramAllChannels); - - static int ThresholdRectToPixOCL( - const unsigned char* imagedata, - int bytes_per_pixel, - int bytes_per_line, - const int* thresholds, - const int* hi_values, - Pix** pix, - int rect_height, - int rect_width, - int rect_top, - int rect_left); - - static Pix * pixConvertRGBToGrayOCL( Pix *pix, float weightRed = 0.3, float weightGreen = 0.5, float weightBlue = 0.2 ); + static int HistogramRectOCL(unsigned char *imagedata, int bytes_per_pixel, + int bytes_per_line, int left, int top, + int width, int height, int kHistogramSize, + int *histogramAllChannels); + + static int ThresholdRectToPixOCL(unsigned char *imagedata, + int bytes_per_pixel, int bytes_per_line, + int *thresholds, int *hi_values, Pix **pix, + int rect_height, int rect_width, + int rect_top, int rect_left); + + static Pix *pixConvertRGBToGrayOCL(Pix *pix, float weightRed = 0.3, + float weightGreen = 0.5, + float weightBlue = 0.2); static ds_device getDeviceSelection(); static ds_device selectedDevice; diff --git a/snap/snapcraft.yaml b/snap/snapcraft.yaml new file mode 100644 index 0000000000..4dd68d4a44 --- /dev/null +++ b/snap/snapcraft.yaml @@ -0,0 +1,34 @@ +name: tesseract +version: master +summary: open source optical character recognition engine +description: | + Tesseract has unicode (UTF-8) support, and can recognize more than 100 + languages "out of the box". It can be trained to recognize other languages. + Tesseract supports various output formats: plain-text, hocr(html), pdf. + +grade: devel # must be 'stable' to release into candidate/stable channels +confinement: strict + +apps: + tesseract: + command: env TESSDATA_PREFIX=$SNAP_USER_COMMON tesseract + plugs: [home] + +parts: + tesseract: + source: . + plugin: autotools + build-packages: + - autoconf-archive + - pkg-config + - libpng12-dev + - libjpeg8-dev + - libtiff5-dev + - zlib1g-dev + - libicu-dev + - libpango1.0-dev + - libcairo2-dev + after: [leptonica] + leptonica: + source: http://www.leptonica.org/source/leptonica-1.74.1.tar.gz + plugin: autotools diff --git a/tessdata/configs/box.train.stderr b/tessdata/configs/box.train.stderr index 6fc51fdd5e..d44ff2b2c7 100644 --- a/tessdata/configs/box.train.stderr +++ b/tessdata/configs/box.train.stderr @@ -1,7 +1,7 @@ -file_type .bl -#tessedit_use_nn F -textord_fast_pitch_test T -tessedit_single_match 0 +file_type .bl +#tessedit_use_nn F +textord_fast_pitch_test T +tessedit_single_match 0 tessedit_zero_rejection T tessedit_minimal_rejection F tessedit_write_rep_codes F diff --git a/tessdata/pdf.ttf b/tessdata/pdf.ttf index 578974a9e8..d1472b20ef 100644 Binary files a/tessdata/pdf.ttf and b/tessdata/pdf.ttf differ diff --git a/testing/reorgdata.sh b/testing/reorgdata.sh index 141de4a6f4..8cee64ff80 100755 --- a/testing/reorgdata.sh +++ b/testing/reorgdata.sh @@ -1,4 +1,13 @@ #!/bin/bash +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. if [ $# -ne 1 ] then diff --git a/testing/runtestset.sh b/testing/runtestset.sh index 0c9595f9aa..5c2a7e7aa2 100755 --- a/testing/runtestset.sh +++ b/testing/runtestset.sh @@ -64,7 +64,7 @@ do srcdir="$imdir" fi # echo "$srcdir/$page.tif" - $tess $srcdir/$page.tif $resdir/$page -psm 6 $config 2>&1 |grep -v "OCR Engine" + $tess $srcdir/$page.tif $resdir/$page --psm 6 $config 2>&1 |grep -v "OCR Engine" if [ -r times.txt ] then read t bounding_box(); + const TBOX& box = bbox->bounding_box(); int left_x = box.left(); int right_x = box.right(); int top_y = box.top(); diff --git a/textord/baselinedetect.cpp b/textord/baselinedetect.cpp index a2b0173949..9bbd999e15 100644 --- a/textord/baselinedetect.cpp +++ b/textord/baselinedetect.cpp @@ -850,7 +850,8 @@ void BaselineDetect::ComputeBaselineSplinesAndXheights(const ICOORD& page_tr, Pix* pix_spline = pix_debug_ ? pixConvertTo32(pix_debug_) : NULL; for (int i = 0; i < blocks_.size(); ++i) { BaselineBlock* bl_block = blocks_[i]; - bl_block->PrepareForSplineFitting(page_tr, remove_noise); + if (enable_splines) + bl_block->PrepareForSplineFitting(page_tr, remove_noise); bl_block->FitBaselineSplines(enable_splines, show_final_rows, textord); if (pix_spline) { bl_block->DrawPixSpline(pix_spline); diff --git a/textord/bbgrid.cpp b/textord/bbgrid.cpp index 06114748f5..4cadcdcf2e 100644 --- a/textord/bbgrid.cpp +++ b/textord/bbgrid.cpp @@ -231,7 +231,7 @@ Pix* GridReducedPix(const TBOX& box, int gridsize, // Note that the Pix is used upside-down, with (0, 0) being the bottom-left. Pix* TraceOutlineOnReducedPix(C_OUTLINE* outline, int gridsize, ICOORD bleft, int* left, int* bottom) { - TBOX box = outline->bounding_box(); + const TBOX& box = outline->bounding_box(); Pix* pix = GridReducedPix(box, gridsize, bleft, left, bottom); int wpl = pixGetWpl(pix); l_uint32* data = pixGetData(pix); @@ -257,7 +257,7 @@ Pix* TraceOutlineOnReducedPix(C_OUTLINE* outline, int gridsize, // As TraceOutlineOnReducedPix above, but on a BLOCK instead of a C_OUTLINE. Pix* TraceBlockOnReducedPix(BLOCK* block, int gridsize, ICOORD bleft, int* left, int* bottom) { - TBOX box = block->bounding_box(); + const TBOX& box = block->bounding_box(); Pix* pix = GridReducedPix(box, gridsize, bleft, left, bottom); int wpl = pixGetWpl(pix); l_uint32* data = pixGetData(pix); diff --git a/textord/bbgrid.h b/textord/bbgrid.h index d16b902ecf..066b5bae1e 100644 --- a/textord/bbgrid.h +++ b/textord/bbgrid.h @@ -364,7 +364,7 @@ template class GridSearch { // An iterator over the list at (x_, y_) in the grid_. BBC_C_IT it_; // Set of unique returned elements used when unique_mode_ is true. - unordered_set > returns_; + TessHashSet > returns_; }; // Sort function to sort a BBC by bounding_box().left(). @@ -623,7 +623,7 @@ void BBGrid::DisplayBoxes(ScrollView* tab_win) { gsearch.StartFullSearch(); BBC* bbox; while ((bbox = gsearch.NextFullSearch()) != NULL) { - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); int left_x = box.left(); int right_x = box.right(); int top_y = box.top(); diff --git a/textord/blkocc.h b/textord/blkocc.h index 89462dc86f..f27bb9a5a5 100644 --- a/textord/blkocc.h +++ b/textord/blkocc.h @@ -52,9 +52,8 @@ class REGION_OCC:public ELIST_LINK float max_x; //Highest x in region inT16 region_type; //Type of crossing - REGION_OCC() { - }; //constructor used - //only in COPIER etc + REGION_OCC() {} // constructor used + // only in COPIER etc REGION_OCC( //constructor float min, float max, @@ -72,12 +71,12 @@ ELISTIZEH (REGION_OCC) Adapted from the following procedure so that it can be used in the bands class in an include file... -BOOL8 range_in_band[ +BOOL8 range_in_band[ range within band? -inT16 band_max, -inT16 band_min, -inT16 range_max, -inT16 range_min] +inT16 band_max, +inT16 band_min, +inT16 range_max, +inT16 range_min] { if ( (range_min >= band_min) && (range_max < band_max) ) return TRUE; @@ -91,12 +90,12 @@ inT16 range_min] Adapted from the following procedure so that it can be used in the bands class in an include file... -BOOL8 range_overlaps_band[ +BOOL8 range_overlaps_band[ range crosses band? -inT16 band_max, -inT16 band_min, -inT16 range_max, -inT16 range_min] +inT16 band_max, +inT16 band_min, +inT16 range_max, +inT16 range_min] { if ( (range_max >= band_min) && (range_min < band_max) ) return TRUE; diff --git a/textord/ccnontextdetect.cpp b/textord/ccnontextdetect.cpp index 1cb0e4c6c7..f6a7d8f41a 100644 --- a/textord/ccnontextdetect.cpp +++ b/textord/ccnontextdetect.cpp @@ -305,7 +305,7 @@ bool CCNonTextDetect::BlobOverlapsTooMuch(BLOBNBOX* blob, int max_overlaps) { // Search the grid to see what intersects it. // Setup a Rectangle search for overlapping this blob. BlobGridSearch rsearch(this); - TBOX box = blob->bounding_box(); + const TBOX& box = blob->bounding_box(); rsearch.StartRectSearch(box); rsearch.SetUniqueMode(true); BLOBNBOX* neighbour; diff --git a/textord/colpartition.cpp b/textord/colpartition.cpp index effb5a9112..0d0b4ca39e 100644 --- a/textord/colpartition.cpp +++ b/textord/colpartition.cpp @@ -918,7 +918,7 @@ void ColPartition::ComputeLimits() { for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) { bbox = it.data(); if (non_leader_count == 0 || bbox->flow() != BTFT_LEADER) { - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); int area = box.area(); top_stats.add(box.top(), area); bottom_stats.add(box.bottom(), area); @@ -1181,8 +1181,8 @@ bool ColPartition::MarkAsLeaderIfMonospaced() { if (best_end == NULL) { tprintf("No path\n"); } else { - tprintf("Total cost = %d vs allowed %d\n", - best_end->total_cost(), blob_count); + tprintf("Total cost = %d vs allowed %d\n", best_end->total_cost(), + blob_count); } } delete [] projection; @@ -1632,6 +1632,10 @@ TO_BLOCK* ColPartition::MakeBlock(const ICOORD& bleft, const ICOORD& tright, ColPartition_LIST* used_parts) { if (block_parts->empty()) return NULL; // Nothing to do. + // If the block_parts are not in reading order, then it will make an invalid + // block polygon and bounding_box, so sort by bounding box now just to make + // sure. + block_parts->sort(&ColPartition::SortByBBox); ColPartition_IT it(block_parts); ColPartition* part = it.data(); PolyBlockType type = part->type(); @@ -2121,7 +2125,7 @@ void ColPartition::RefinePartnersByOverlap(bool upper, // Return true if bbox belongs better in this than other. bool ColPartition::ThisPartitionBetter(BLOBNBOX* bbox, const ColPartition& other) { - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); // Margins take priority. int left = box.left(); int right = box.right(); diff --git a/textord/colpartition.h b/textord/colpartition.h index 5c941cce15..7fcbc0004e 100644 --- a/textord/colpartition.h +++ b/textord/colpartition.h @@ -704,6 +704,25 @@ class ColPartition : public ELIST2_LINK { // doing a SideSearch when you want things in the same page column. bool IsInSameColumnAs(const ColPartition& part) const; + // Sort function to sort by bounding box. + static int SortByBBox(const void* p1, const void* p2) { + const ColPartition* part1 = + *reinterpret_cast(p1); + const ColPartition* part2 = + *reinterpret_cast(p2); + int mid_y1 = part1->bounding_box_.y_middle(); + int mid_y2 = part2->bounding_box_.y_middle(); + if ((part2->bounding_box_.bottom() <= mid_y1 && + mid_y1 <= part2->bounding_box_.top()) || + (part1->bounding_box_.bottom() <= mid_y2 && + mid_y2 <= part1->bounding_box_.top())) { + // Sort by increasing x. + return part1->bounding_box_.x_middle() - part2->bounding_box_.x_middle(); + } + // Sort by decreasing y. + return mid_y2 - mid_y1; + } + // Sets the column bounds. Primarily used in testing. void set_first_column(int column) { first_column_ = column; diff --git a/textord/colpartitiongrid.cpp b/textord/colpartitiongrid.cpp index 4d703fbe7d..86c883280e 100644 --- a/textord/colpartitiongrid.cpp +++ b/textord/colpartitiongrid.cpp @@ -86,7 +86,7 @@ void ColPartitionGrid::HandleClick(int x, int y) { ColPartition* neighbour; FCOORD click(x, y); while ((neighbour = radsearch.NextRadSearch()) != NULL) { - TBOX nbox = neighbour->bounding_box(); + const TBOX& nbox = neighbour->bounding_box(); if (nbox.contains(click)) { tprintf("Block box:"); neighbour->bounding_box().print(); @@ -1037,7 +1037,7 @@ void ColPartitionGrid::ListFindMargins(ColPartitionSet** best_columns, ColPartition* part = part_it.data(); ColPartitionSet* columns = NULL; if (best_columns != NULL) { - TBOX part_box = part->bounding_box(); + const TBOX& part_box = part->bounding_box(); // Get the columns from the y grid coord. int grid_x, grid_y; GridCoords(part_box.left(), part_box.bottom(), &grid_x, &grid_y); @@ -1376,7 +1376,7 @@ void ColPartitionGrid::FindMergeCandidates(const ColPartition* part, // combined box to see if anything else is inappropriately overlapped. if (!part_box.contains(c_box) && !c_box.contains(part_box)) { // Search the combined rectangle to see if anything new is overlapped. - // This is a preliminary test designed to quickly weed-out stupid + // This is a preliminary test designed to quickly weed-out poor // merge candidates that would create a big list of overlapped objects // for the squared-order overlap analysis. Eg. vertical and horizontal // line-like objects that overlap real text when merged: @@ -1569,7 +1569,7 @@ BlobRegionType ColPartitionGrid::SmoothInOneDirection( const TBOX& im_box, const FCOORD& rerotation, bool debug, const ColPartition& part, int* best_distance) { // Set up a rectangle search bounded by the part. - TBOX part_box = part.bounding_box(); + const TBOX& part_box = part.bounding_box(); TBOX search_box; ICOORD dist_scaling; ComputeSearchBoxAndScaling(direction, part_box, gridsize(), @@ -1619,10 +1619,10 @@ BlobRegionType ColPartitionGrid::SmoothInOneDirection( image_bias - htext_score >= kSmoothDecisionMargin && image_bias - vtext_score >= kSmoothDecisionMargin) { *best_distance = dists[NPT_IMAGE][0]; - if (dists[NPT_WEAK_VTEXT].size() > 0 && + if (!dists[NPT_WEAK_VTEXT].empty() && *best_distance > dists[NPT_WEAK_VTEXT][0]) *best_distance = dists[NPT_WEAK_VTEXT][0]; - if (dists[NPT_WEAK_HTEXT].size() > 0 && + if (!dists[NPT_WEAK_HTEXT].empty() && *best_distance > dists[NPT_WEAK_HTEXT][0]) *best_distance = dists[NPT_WEAK_HTEXT][0]; return BRT_POLYIMAGE; diff --git a/textord/devanagari_processing.h b/textord/devanagari_processing.h index 990a5dfe39..0d070decd0 100644 --- a/textord/devanagari_processing.h +++ b/textord/devanagari_processing.h @@ -1,5 +1,14 @@ // Copyright 2008 Google Inc. All Rights Reserved. // Author: shobhitsaxena@google.com (Shobhit Saxena) +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef TESSERACT_TEXTORD_DEVNAGARI_PROCESSING_H_ #define TESSERACT_TEXTORD_DEVNAGARI_PROCESSING_H_ @@ -41,9 +50,7 @@ class PixelHistogram { length_ = 0; } - int* hist() const { - return hist_; - } + int* hist() const { return hist_; } int length() const { return length_; diff --git a/textord/drawedg.h b/textord/drawedg.h index 6bf062d4ee..ef5ed5e202 100644 --- a/textord/drawedg.h +++ b/textord/drawedg.h @@ -1,8 +1,9 @@ /********************************************************************** * File: drawedg.h (Formerly drawedge.h) - * Description: Collection of functions to draw things to do with edge detection. - * Author: Ray Smith - * Created: Thu Jun 06 13:29:20 BST 1991 + * Description: Collection of functions to draw things to do with edge + *detection. + * Author: Ray Smith + * Created: Thu Jun 06 13:29:20 BST 1991 * * (C) Copyright 1991, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); diff --git a/textord/fpchop.cpp b/textord/fpchop.cpp index 4c18338b8f..699d419620 100644 --- a/textord/fpchop.cpp +++ b/textord/fpchop.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: fpchop.cpp (Formerly fp_chop.c) * Description: Code to chop fixed pitch text into character cells. - * Author: Ray Smith - * Created: Thu Sep 16 11:14:15 BST 1993 + * Author: Ray Smith + * Created: Thu Sep 16 11:14:15 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -259,8 +259,8 @@ void split_to_blob( //split the blob pitch_error, left_coutlines, right_coutlines); - if (blob != NULL) - delete blob; //free it + + delete blob; } /********************************************************************** @@ -730,7 +730,6 @@ C_OUTLINE *join_chopped_fragments( //join pieces return NULL; } - /********************************************************************** * join_segments * diff --git a/textord/gap_map.cpp b/textord/gap_map.cpp index 2f8440e601..421208784d 100644 --- a/textord/gap_map.cpp +++ b/textord/gap_map.cpp @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include "statistc.h" #include "gap_map.h" diff --git a/textord/gap_map.h b/textord/gap_map.h index 914e8dbdea..227db3646b 100644 --- a/textord/gap_map.h +++ b/textord/gap_map.h @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #ifndef GAP_MAP_H #define GAP_MAP_H diff --git a/textord/imagefind.cpp b/textord/imagefind.cpp index c119e69f95..14442d52ed 100644 --- a/textord/imagefind.cpp +++ b/textord/imagefind.cpp @@ -77,7 +77,7 @@ Pix* ImageFind::FindImages(Pix* pix) { // Leptonica will print an error message and return NULL if we call // pixGenHalftoneMask(pixr, NULL, ...) with too small image, so we // want to bypass that. - if (pixGetWidth(pixr) < kMinImageFindSize || + if (pixGetWidth(pixr) < kMinImageFindSize || pixGetHeight(pixr) < kMinImageFindSize) { pixDestroy(&pixr); return pixCreate(pixGetWidth(pix), pixGetHeight(pix), 1); @@ -1115,7 +1115,7 @@ static bool TestWeakIntersectedPart(const TBOX& im_box, ColPartition* part) { if (part->flow() < BTFT_STRONG_CHAIN) { // A weak partition intersects the box. - TBOX part_box = part->bounding_box(); + const TBOX& part_box = part->bounding_box(); if (im_box.contains(part_box)) { int area = part_box.area(); int intersect_area = IntersectArea(part_box, part_list); @@ -1180,7 +1180,7 @@ static bool ScanForOverlappingText(ColPartitionGrid* part_grid, TBOX* box) { part->flow() == BTFT_STRONG_CHAIN) { // Text intersects the box. any_text_in_padded_rect = true; - TBOX part_box = part->bounding_box(); + const TBOX& part_box = part->bounding_box(); if (box->overlap(part_box)) { return true; } diff --git a/textord/makerow.cpp b/textord/makerow.cpp index 1df4855b40..a5749ad680 100644 --- a/textord/makerow.cpp +++ b/textord/makerow.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: makerow.cpp (Formerly makerows.c) * Description: Code to arrange blobs into rows of text. - * Author: Ray Smith - * Created: Mon Sep 21 14:34:48 BST 1992 + * Author: Ray Smith + * Created: Mon Sep 21 14:34:48 BST 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -507,8 +507,7 @@ void vigorous_noise_removal(TO_BLOCK* block) { continue; // Looks OK. } // It might be noise so get rid of it. - if (blob->cblob() != NULL) - delete blob->cblob(); + delete blob->cblob(); delete b_it.extract(); } else { prev = blob; @@ -671,7 +670,7 @@ BOOL8 find_best_dropout_row( //find neighbours TO_ROW_IT *row_it, //current position BOOL8 testing_on //correct orientation ) { - inT32 next_index; //of neighbouring row + inT32 next_index; // of neighbouring row inT32 row_offset; //from current row inT32 abs_dist; //absolute distance inT8 row_inc; //increment to row_index @@ -1786,7 +1785,7 @@ static int CountOverlaps(const TBOX& box, int min_height, BLOBNBOX_IT blob_it(blobs); for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) { BLOBNBOX* blob = blob_it.data(); - TBOX blob_box = blob->bounding_box(); + const TBOX &blob_box = blob->bounding_box(); if (blob_box.height() >= min_height && box.major_overlap(blob_box)) { ++overlaps; } diff --git a/textord/oldbasel.cpp b/textord/oldbasel.cpp index c73fe8d57b..99e55fdbb6 100644 --- a/textord/oldbasel.cpp +++ b/textord/oldbasel.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: oldbasel.cpp (Formerly oldbl.c) * Description: A re-implementation of the old baseline algorithm. - * Author: Ray Smith - * Created: Wed Oct 6 09:41:48 BST 1993 + * Author: Ray Smith + * Created: Wed Oct 6 09:41:48 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -122,7 +122,7 @@ void Textord::correlate_lines(TO_BLOCK *block, float gradient) { TO_ROW **rows; //array of ptrs int rowcount; /*no of rows to do */ int rowindex; /*no of row */ - //iterator + // iterator TO_ROW_IT row_it = block->get_rows (); rowcount = row_it.length (); @@ -1018,61 +1018,6 @@ int *partcount /*no of partitions */ return bestpart; } - -///*merge_partitions(partids,partcount,blobcount,bestpart) discards funny looking -//partitions and gives all the rest partid 0*/ -// -//merge_partitions(partids,partcount,blobcount,bestpart) -//register char *partids; /*partition numbers*/ -//int partcount; /*no of partitions*/ -//int blobcount; /*no of blobs*/ -//int bestpart; /*best partition*/ -//{ -// int blobindex; /*no along text line*/ -// int runlength; /*run of same partition*/ -// int bestrun; /*biggest runlength*/ -// -// bestrun=0; /*no runs yet*/ -// runlength=1; -// for (blobindex=1;blobindexbestrun) -// bestrun=runlength; /*find biggest run*/ -// runlength=1; /*new run*/ -// } -// else -// { runlength++; -// } -// } -// if (runlength>bestrun) -// bestrun=runlength; -// -// for (blobindex=0;blobindex=blobcount -// || partids[blobindex]!=partids[blobindex+1]) -// /*loner*/ -// && (bestrun>2 || partids[blobindex]!=bestpart)) -// { partids[blobindex]=partcount; /*discard loner*/ -// } -// else if (blobindex+1=blobcount -// || partids[blobindex]!=partids[blobindex+2]) -// && (bestrun>3 || partids[blobindex]!=bestpart)) -// { partids[blobindex]=partcount; /*discard both*/ -// partids[blobindex+1]=partcount; -// } -// } -// } -// for (blobindex=0;blobindex= MINASCRISE && partsizes[partition] > poscount) { @@ -1459,8 +1402,8 @@ int blobcount, /*blobs in blobcoords */ QSPLINE * baseline, /*established */ float jumplimit /*min ascender height */ ) { - int blobindex; /*current blob */ - /*height statistics */ + int blobindex; /*current blob */ + /*height statistics */ STATS heightstat (0, MAXHEIGHT); int height; /*height of blob */ int xcentre; /*centre of blob */ diff --git a/textord/pithsync.h b/textord/pithsync.h index f9ba479e2f..386426be72 100644 --- a/textord/pithsync.h +++ b/textord/pithsync.h @@ -1,8 +1,8 @@ /********************************************************************** * File: pithsync.h (Formerly pitsync2.h) * Description: Code to find the optimum fixed pitch segmentation of some blobs. - * Author: Ray Smith - * Created: Thu Nov 19 11:48:05 GMT 1992 + * Author: Ray Smith + * Created: Thu Nov 19 11:48:05 GMT 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -66,7 +66,7 @@ class FPCUTPT inT16 pitch, //proposed pitch inT16 pitch_error); //allowed tolerance - inT32 position() { //access func + inT32 position() { // access func return xpos; } double cost_function() { diff --git a/textord/pitsync1.h b/textord/pitsync1.h index c2fb9bec65..5374b003dd 100644 --- a/textord/pitsync1.h +++ b/textord/pitsync1.h @@ -1,8 +1,8 @@ /********************************************************************** * File: pitsync1.h (Formerly pitsync.h) * Description: Code to find the optimum fixed pitch segmentation of some blobs. - * Author: Ray Smith - * Created: Thu Nov 19 11:48:05 GMT 1992 + * Author: Ray Smith + * Created: Thu Nov 19 11:48:05 GMT 1992 * * (C) Copyright 1992, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -46,7 +46,7 @@ class FPSEGPT:public ELIST_LINK FPSEGPT_LIST *prev_list); //previous segment FPSEGPT(FPCUTPT *cutpt); //build from new type - inT32 position() { //access func + inT32 position() { // access func return xpos; } double cost_function() { diff --git a/textord/scanedg.cpp b/textord/scanedg.cpp index dbb3b66298..0430843619 100644 --- a/textord/scanedg.cpp +++ b/textord/scanedg.cpp @@ -335,7 +335,7 @@ void join_edges(CRACKEDGE *edge1, // edges to join if (edge1->pos.x() + edge1->stepx != edge2->pos.x() || edge1->pos.y() + edge1->stepy != edge2->pos.y()) { CRACKEDGE *tempedge = edge1; - edge1 = edge2; // swap araound + edge1 = edge2; // swap around edge2 = tempedge; } diff --git a/textord/strokewidth.cpp b/textord/strokewidth.cpp index 5d0fdc5133..059aa9b85e 100644 --- a/textord/strokewidth.cpp +++ b/textord/strokewidth.cpp @@ -393,7 +393,7 @@ void StrokeWidth::GradeBlobsIntoPartitions( } static void PrintBoxWidths(BLOBNBOX* neighbour) { - TBOX nbox = neighbour->bounding_box(); + const TBOX& nbox = neighbour->bounding_box(); tprintf("Box (%d,%d)->(%d,%d): h-width=%.1f, v-width=%.1f p-width=%1.f\n", nbox.left(), nbox.bottom(), nbox.right(), nbox.top(), neighbour->horz_stroke_width(), neighbour->vert_stroke_width(), @@ -1939,7 +1939,7 @@ ScrollView* StrokeWidth::DisplayGoodBlobs(const char* window_name, gsearch.StartFullSearch(); BLOBNBOX* bbox; while ((bbox = gsearch.NextFullSearch()) != NULL) { - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); int left_x = box.left(); int right_x = box.right(); int top_y = box.top(); diff --git a/textord/tabfind.cpp b/textord/tabfind.cpp index dc7a072b7d..30bad8bb3c 100644 --- a/textord/tabfind.cpp +++ b/textord/tabfind.cpp @@ -229,7 +229,7 @@ void TabFind::GutterWidthAndNeighbourGap(int tab_x, int mean_height, bbox->flow() == BTFT_TEXT_ON_IMAGE, 0.0, *gutter_width, box.top(), box.bottom()); if (gutter_bbox != NULL) { - TBOX gutter_box = gutter_bbox->bounding_box(); + const TBOX& gutter_box = gutter_bbox->bounding_box(); *gutter_width = left ? tab_x - gutter_box.right() : gutter_box.left() - tab_x; } @@ -261,7 +261,7 @@ void TabFind::GutterWidthAndNeighbourGap(int tab_x, int mean_height, int neighbour_edge = left ? RightEdgeForBox(box, true, false) : LeftEdgeForBox(box, true, false); if (neighbour != NULL) { - TBOX n_box = neighbour->bounding_box(); + const TBOX& n_box = neighbour->bounding_box(); if (debug) { tprintf("Found neighbour:"); n_box.print(); diff --git a/textord/tablefind.cpp b/textord/tablefind.cpp index 425bdbc218..b68655a57c 100644 --- a/textord/tablefind.cpp +++ b/textord/tablefind.cpp @@ -550,7 +550,7 @@ void TableFinder::GroupColumnBlocks(ColSegment_LIST* new_blocks, // iterate through the source list for (src_it.mark_cycle_pt(); !src_it.cycled_list(); src_it.forward()) { ColSegment* src_seg = src_it.data(); - TBOX src_box = src_seg->bounding_box(); + const TBOX& src_box = src_seg->bounding_box(); bool match_found = false; // iterate through the destination list to find a matching column block for (dest_it.mark_cycle_pt(); !dest_it.cycled_list(); dest_it.forward()) { @@ -1342,7 +1342,7 @@ void TableFinder::GetTableRegions(ColSegment_LIST* table_columns, // create a bool array to hold projection on y-axis bool* table_region = new bool[page_height]; while ((part = gsearch.NextFullSearch()) != NULL) { - TBOX part_box = part->bounding_box(); + const TBOX& part_box = part->bounding_box(); // reset the projection array for (int i = 0; i < page_height; i++) { table_region[i] = false; @@ -1974,7 +1974,7 @@ void TableFinder::DisplayColPartitionConnections( ColPartition* upper_part = part->nearest_neighbor_above(); if (upper_part) { - TBOX upper_box = upper_part->bounding_box(); + const TBOX& upper_box = upper_part->bounding_box(); int mid_x = (left_x + right_x) / 2; int mid_y = (top_y + bottom_y) / 2; int other_x = (upper_box.left() + upper_box.right()) / 2; @@ -1985,7 +1985,7 @@ void TableFinder::DisplayColPartitionConnections( } ColPartition* lower_part = part->nearest_neighbor_below(); if (lower_part) { - TBOX lower_box = lower_part->bounding_box(); + const TBOX& lower_box = lower_part->bounding_box(); int mid_x = (left_x + right_x) / 2; int mid_y = (top_y + bottom_y) / 2; int other_x = (lower_box.left() + lower_box.right()) / 2; @@ -2098,7 +2098,7 @@ void TableFinder::MakeTableBlocks(ColPartitionGrid* grid, table_search.StartFullSearch(); ColSegment* table; while ((table = table_search.NextFullSearch()) != NULL) { - TBOX table_box = table->bounding_box(); + const TBOX& table_box = table->bounding_box(); // Start a rect search on table_box GridSearch rectsearch(grid); diff --git a/textord/tabvector.cpp b/textord/tabvector.cpp index c8d508f0df..f3e99fa38f 100644 --- a/textord/tabvector.cpp +++ b/textord/tabvector.cpp @@ -435,7 +435,7 @@ bool TabVector::SimilarTo(const ICOORD& vertical, vsearch.StartVerticalSearch(left, right, top_y); BLOBNBOX* blob; while ((blob = vsearch.NextVerticalSearch(true)) != NULL) { - TBOX box = blob->bounding_box(); + const TBOX& box = blob->bounding_box(); if (box.top() > bottom_y) return true; // Nothing found. if (box.bottom() < top_y) @@ -523,12 +523,12 @@ const char* kAlignmentNames[] = { // Print basic information about this tab vector. void TabVector::Print(const char* prefix) { - tprintf("%s %s (%d,%d)->(%d,%d) w=%d s=%d, sort key=%d, boxes=%d," - " partners=%d\n", - prefix, kAlignmentNames[alignment_], - startpt_.x(), startpt_.y(), endpt_.x(), endpt_.y(), - mean_width_, percent_score_, sort_key_, - boxes_.length(), partners_.length()); + tprintf( + "%s %s (%d,%d)->(%d,%d) w=%d s=%d, sort key=%d, boxes=%d," + " partners=%d\n", + prefix, kAlignmentNames[alignment_], startpt_.x(), startpt_.y(), + endpt_.x(), endpt_.y(), mean_width_, percent_score_, sort_key_, + boxes_.length(), partners_.length()); } // Print basic information about this tab vector and every box in it. @@ -806,7 +806,7 @@ bool TabVector::Fit(ICOORD vertical, bool force_parallel) { // Fit a line to all the boxes in the list. for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) { BLOBNBOX* bbox = it.data(); - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); int x1 = IsRightTab() ? box.right() : box.left(); ICOORD boxpt(x1, box.bottom()); linepoints.Add(boxpt); @@ -831,7 +831,7 @@ bool TabVector::Fit(ICOORD vertical, bool force_parallel) { int width_count = 0; for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) { BLOBNBOX* bbox = it.data(); - TBOX box = bbox->bounding_box(); + const TBOX& box = bbox->bounding_box(); mean_width_ += box.width(); ++width_count; int x1 = IsRightTab() ? box.right() : box.left(); diff --git a/textord/textlineprojection.cpp b/textord/textlineprojection.cpp index 6018e5fdaa..2651a19b33 100644 --- a/textord/textlineprojection.cpp +++ b/textord/textlineprojection.cpp @@ -760,7 +760,7 @@ void TextlineProjection::TruncateToImageBounds(TPOINT* pt) const { pt->y = ClipToRange(pt->y, 0, pixGetHeight(pix_) - 1); } #ifdef _MSC_VER -#pragma optimize( "", on ) +#pragma optimize("", on) #endif // _MSC_VER // Transform tesseract image coordinates to coordinates used in the projection. diff --git a/textord/textord.cpp b/textord/textord.cpp index 1f7e8a8869..94ef49c783 100644 --- a/textord/textord.cpp +++ b/textord/textord.cpp @@ -33,7 +33,8 @@ namespace tesseract { Textord::Textord(CCStruct* ccstruct) - : ccstruct_(ccstruct), use_cjk_fp_model_(false), + : ccstruct_(ccstruct), + use_cjk_fp_model_(false), // makerow.cpp /////////////////////////////////////////// BOOL_MEMBER(textord_single_height_mode, false, "Script has no xheight, so use a single mode", @@ -46,24 +47,20 @@ Textord::Textord(CCStruct* ccstruct) "old_to_method.", ccstruct_->params()), BOOL_MEMBER(tosp_only_use_prop_rows, true, - "Block stats to use fixed pitch rows?", - ccstruct_->params()), + "Block stats to use fixed pitch rows?", ccstruct_->params()), BOOL_MEMBER(tosp_force_wordbreak_on_punct, false, "Force word breaks on punct to break long lines in non-space " "delimited langs", ccstruct_->params()), - BOOL_MEMBER(tosp_use_pre_chopping, false, - "Space stats use prechopping?", + BOOL_MEMBER(tosp_use_pre_chopping, false, "Space stats use prechopping?", ccstruct_->params()), BOOL_MEMBER(tosp_old_to_bug_fix, false, "Fix suspected bug in old code", ccstruct_->params()), - BOOL_MEMBER(tosp_block_use_cert_spaces, true, - "Only stat OBVIOUS spaces", + BOOL_MEMBER(tosp_block_use_cert_spaces, true, "Only stat OBVIOUS spaces", ccstruct_->params()), BOOL_MEMBER(tosp_row_use_cert_spaces, true, "Only stat OBVIOUS spaces", ccstruct_->params()), - BOOL_MEMBER(tosp_narrow_blobs_not_cert, true, - "Only stat OBVIOUS spaces", + BOOL_MEMBER(tosp_narrow_blobs_not_cert, true, "Only stat OBVIOUS spaces", ccstruct_->params()), BOOL_MEMBER(tosp_row_use_cert_spaces1, true, "Only stat OBVIOUS spaces", ccstruct_->params()), @@ -78,30 +75,24 @@ Textord::Textord(CCStruct* ccstruct) "Don't restrict kn->sp fuzzy limit to tables", ccstruct_->params()), BOOL_MEMBER(tosp_stats_use_xht_gaps, true, - "Use within xht gap for wd breaks", - ccstruct_->params()), + "Use within xht gap for wd breaks", ccstruct_->params()), BOOL_MEMBER(tosp_use_xht_gaps, true, "Use within xht gap for wd breaks", ccstruct_->params()), BOOL_MEMBER(tosp_only_use_xht_gaps, false, - "Only use within xht gap for wd breaks", - ccstruct_->params()), + "Only use within xht gap for wd breaks", ccstruct_->params()), BOOL_MEMBER(tosp_rule_9_test_punct, false, - "Don't chng kn to space next to punct", - ccstruct_->params()), + "Don't chng kn to space next to punct", ccstruct_->params()), BOOL_MEMBER(tosp_flip_fuzz_kn_to_sp, true, "Default flip", ccstruct_->params()), BOOL_MEMBER(tosp_flip_fuzz_sp_to_kn, true, "Default flip", ccstruct_->params()), BOOL_MEMBER(tosp_improve_thresh, false, "Enable improvement heuristic", ccstruct_->params()), - INT_MEMBER(tosp_debug_level, 0, "Debug data", - ccstruct_->params()), + INT_MEMBER(tosp_debug_level, 0, "Debug data", ccstruct_->params()), INT_MEMBER(tosp_enough_space_samples_for_median, 3, - "or should we use mean", - ccstruct_->params()), + "or should we use mean", ccstruct_->params()), INT_MEMBER(tosp_redo_kern_limit, 10, - "No.samples reqd to reestimate for row", - ccstruct_->params()), + "No.samples reqd to reestimate for row", ccstruct_->params()), INT_MEMBER(tosp_few_samples, 40, "No.gaps reqd with 1 large gap to treat as a table", ccstruct_->params()), @@ -114,30 +105,24 @@ Textord::Textord(CCStruct* ccstruct) "Factor for defining space threshold in terms of space and " "kern sizes", ccstruct_->params()), - double_MEMBER(tosp_threshold_bias1, 0, - "how far between kern and space?", + double_MEMBER(tosp_threshold_bias1, 0, "how far between kern and space?", ccstruct_->params()), - double_MEMBER(tosp_threshold_bias2, 0, - "how far between kern and space?", + double_MEMBER(tosp_threshold_bias2, 0, "how far between kern and space?", ccstruct_->params()), double_MEMBER(tosp_narrow_fraction, 0.3, "Fract of xheight for narrow", ccstruct_->params()), double_MEMBER(tosp_narrow_aspect_ratio, 0.48, - "narrow if w/h less than this", - ccstruct_->params()), + "narrow if w/h less than this", ccstruct_->params()), double_MEMBER(tosp_wide_fraction, 0.52, "Fract of xheight for wide", ccstruct_->params()), double_MEMBER(tosp_wide_aspect_ratio, 0.0, "wide if w/h less than this", ccstruct_->params()), double_MEMBER(tosp_fuzzy_space_factor, 0.6, - "Fract of xheight for fuzz sp", - ccstruct_->params()), + "Fract of xheight for fuzz sp", ccstruct_->params()), double_MEMBER(tosp_fuzzy_space_factor1, 0.5, - "Fract of xheight for fuzz sp", - ccstruct_->params()), + "Fract of xheight for fuzz sp", ccstruct_->params()), double_MEMBER(tosp_fuzzy_space_factor2, 0.72, - "Fract of xheight for fuzz sp", - ccstruct_->params()), + "Fract of xheight for fuzz sp", ccstruct_->params()), double_MEMBER(tosp_gap_factor, 0.83, "gap ratio to flip sp->kern", ccstruct_->params()), double_MEMBER(tosp_kern_gap_factor1, 2.0, "gap ratio to flip kern->sp", @@ -156,14 +141,11 @@ Textord::Textord(CCStruct* ccstruct) "Fract of kerns reqd for isolated row stats", ccstruct_->params()), double_MEMBER(tosp_table_kn_sp_ratio, 2.25, - "Min difference of kn & sp in table", - ccstruct_->params()), + "Min difference of kn & sp in table", ccstruct_->params()), double_MEMBER(tosp_table_xht_sp_ratio, 0.33, - "Expect spaces bigger than this", - ccstruct_->params()), + "Expect spaces bigger than this", ccstruct_->params()), double_MEMBER(tosp_table_fuzzy_kn_sp_ratio, 3.0, - "Fuzzy if less than this", - ccstruct_->params()), + "Fuzzy if less than this", ccstruct_->params()), double_MEMBER(tosp_fuzzy_kn_fraction, 0.5, "New fuzzy kn alg", ccstruct_->params()), double_MEMBER(tosp_fuzzy_sp_fraction, 0.5, "New fuzzy sp alg", @@ -172,20 +154,16 @@ Textord::Textord(CCStruct* ccstruct) "Don't trust spaces less than this time kn", ccstruct_->params()), double_MEMBER(tosp_init_guess_kn_mult, 2.2, - "Thresh guess - mult kn by this", - ccstruct_->params()), + "Thresh guess - mult kn by this", ccstruct_->params()), double_MEMBER(tosp_init_guess_xht_mult, 0.28, - "Thresh guess - mult xht by this", - ccstruct_->params()), + "Thresh guess - mult xht by this", ccstruct_->params()), double_MEMBER(tosp_max_sane_kn_thresh, 5.0, - "Multiplier on kn to limit thresh", - ccstruct_->params()), + "Multiplier on kn to limit thresh", ccstruct_->params()), double_MEMBER(tosp_flip_caution, 0.0, "Don't autoflip kn to sp when large separation", ccstruct_->params()), double_MEMBER(tosp_large_kerning, 0.19, - "Limit use of xht gap with large kns", - ccstruct_->params()), + "Limit use of xht gap with large kns", ccstruct_->params()), double_MEMBER(tosp_dont_fool_with_small_kerns, -1, "Limit use of xht gap with odd small kns", ccstruct_->params()), @@ -193,11 +171,9 @@ Textord::Textord(CCStruct* ccstruct) "Don't reduce box if the top left is non blank", ccstruct_->params()), double_MEMBER(tosp_silly_kn_sp_gap, 0.2, - "Don't let sp minus kn get too small", - ccstruct_->params()), + "Don't let sp minus kn get too small", ccstruct_->params()), double_MEMBER(tosp_pass_wide_fuzz_sp_to_context, 0.75, - "How wide fuzzies need context", - ccstruct_->params()), + "How wide fuzzies need context", ccstruct_->params()), // tordmain.cpp /////////////////////////////////////////// BOOL_MEMBER(textord_no_rejects, false, "Don't remove noise blobs", ccstruct_->params()), @@ -206,34 +182,27 @@ Textord::Textord(CCStruct* ccstruct) BOOL_MEMBER(textord_show_boxes, false, "Display unsorted blobs", ccstruct_->params()), INT_MEMBER(textord_max_noise_size, 7, "Pixel size of noise", - ccstruct_->params()), + ccstruct_->params()), INT_MEMBER(textord_baseline_debug, 0, "Baseline debug level", - ccstruct_->params()), + ccstruct_->params()), double_MEMBER(textord_blob_size_bigile, 95, "Percentile for large blobs", ccstruct_->params()), double_MEMBER(textord_noise_area_ratio, 0.7, - "Fraction of bounding box for noise", - ccstruct_->params()), + "Fraction of bounding box for noise", ccstruct_->params()), double_MEMBER(textord_blob_size_smallile, 20, - "Percentile for small blobs", - ccstruct_->params()), + "Percentile for small blobs", ccstruct_->params()), double_MEMBER(textord_initialx_ile, 0.75, - "Ile of sizes for xheight guess", - ccstruct_->params()), + "Ile of sizes for xheight guess", ccstruct_->params()), double_MEMBER(textord_initialasc_ile, 0.90, - "Ile of sizes for xheight guess", - ccstruct_->params()), - INT_MEMBER(textord_noise_sizefraction, 10, - "Fraction of size for maxima", + "Ile of sizes for xheight guess", ccstruct_->params()), + INT_MEMBER(textord_noise_sizefraction, 10, "Fraction of size for maxima", ccstruct_->params()), double_MEMBER(textord_noise_sizelimit, 0.5, - "Fraction of x for big t count", - ccstruct_->params()), + "Fraction of x for big t count", ccstruct_->params()), INT_MEMBER(textord_noise_translimit, 16, "Transitions for normal blob", ccstruct_->params()), double_MEMBER(textord_noise_normratio, 2.0, - "Dot to norm ratio for deletion", - ccstruct_->params()), + "Dot to norm ratio for deletion", ccstruct_->params()), BOOL_MEMBER(textord_noise_rejwords, true, "Reject noise-like words", ccstruct_->params()), BOOL_MEMBER(textord_noise_rejrows, true, "Reject noise-like rows", @@ -242,24 +211,20 @@ Textord::Textord(CCStruct* ccstruct) "xh fract height error for norm blobs", ccstruct_->params()), double_MEMBER(textord_noise_sxfract, 0.4, - "xh fract width error for norm blobs", - ccstruct_->params()), - double_MEMBER(textord_noise_hfract, 1.0/64, + "xh fract width error for norm blobs", ccstruct_->params()), + double_MEMBER(textord_noise_hfract, 1.0 / 64, "Height fraction to discard outlines as speckle noise", ccstruct_->params()), INT_MEMBER(textord_noise_sncount, 1, "super norm blobs to save row", ccstruct_->params()), double_MEMBER(textord_noise_rowratio, 6.0, - "Dot to norm ratio for deletion", - ccstruct_->params()), + "Dot to norm ratio for deletion", ccstruct_->params()), BOOL_MEMBER(textord_noise_debug, false, "Debug row garbage detector", ccstruct_->params()), double_MEMBER(textord_blshift_maxshift, 0.00, "Max baseline shift", ccstruct_->params()), double_MEMBER(textord_blshift_xfraction, 9.99, - "Min size of baseline shift", - ccstruct_->params()) { -} + "Min size of baseline shift", ccstruct_->params()) {} Textord::~Textord() { } @@ -324,10 +289,9 @@ void Textord::TextordPage(PageSegMode pageseg_mode, const FCOORD& reskew, BaselineDetect baseline_detector(textord_baseline_debug, reskew, to_blocks); baseline_detector.ComputeStraightBaselines(use_box_bottoms); - baseline_detector.ComputeBaselineSplinesAndXheights(page_tr_, true, - textord_heavy_nr, - textord_show_final_rows, - this); + baseline_detector.ComputeBaselineSplinesAndXheights( + page_tr_, pageseg_mode != PSM_RAW_LINE, textord_heavy_nr, + textord_show_final_rows, this); // Now make the words in the lines. if (PSM_WORD_FIND_ENABLED(pageseg_mode)) { // SINGLE_LINE uses the old word maker on the single line. diff --git a/textord/topitch.cpp b/textord/topitch.cpp index ae9999f7db..cfde683b0b 100644 --- a/textord/topitch.cpp +++ b/textord/topitch.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: topitch.cpp (Formerly to_pitch.c) * Description: Code to determine fixed pitchness and the pitch if fixed. - * Author: Ray Smith - * Created: Tue Aug 24 16:57:29 BST 1993 + * Author: Ray Smith + * Created: Tue Aug 24 16:57:29 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -1084,7 +1084,7 @@ BOOL8 count_pitch_stats( //find lines return FALSE; prev_valid = FALSE; prev_centre = 0; - prev_right = 0; //stop compiler warning + prev_right = 0; // stop compiler warning joined_box = blob_it.data ()->bounding_box (); do { blob_it.forward (); @@ -1285,8 +1285,6 @@ float tune_row_pitch2( //find fp cells return initial_pitch; } sum_proj = new STATS[textord_pitch_range * 2 + 1]; - if (sum_proj == NULL) - return initial_pitch; for (pitch_delta = -textord_pitch_range; pitch_delta <= textord_pitch_range; pitch_delta++) diff --git a/textord/tordmain.cpp b/textord/tordmain.cpp index f09a186d4f..0eaf843ec3 100644 --- a/textord/tordmain.cpp +++ b/textord/tordmain.cpp @@ -251,6 +251,7 @@ void Textord::filter_blobs(ICOORD page_tr, // top right &block->noise_blobs, &block->small_blobs, &block->large_blobs); + if (block->line_size == 0) block->line_size = 1; block->line_spacing = block->line_size * (tesseract::CCStruct::kDescenderFraction + tesseract::CCStruct::kXHeightFraction + @@ -360,7 +361,7 @@ void Textord::cleanup_nontext_block(BLOCK* block) { // Non-text blocks must contain at least one row. ROW_IT row_it(block->row_list()); if (row_it.empty()) { - TBOX box = block->bounding_box(); + const TBOX& box = block->bounding_box(); float height = box.height(); inT32 xstarts[2] = {box.left(), box.right()}; double coeffs[3] = {0.0, 0.0, static_cast(box.bottom())}; @@ -769,6 +770,7 @@ void Textord::TransferDiacriticsToBlockGroups(BLOBNBOX_LIST* diacritic_blobs, PointerVector word_ptrs; for (int g = 0; g < groups.size(); ++g) { const BlockGroup* group = groups[g]; + if (group->bounding_box.null_box()) continue; WordGrid word_grid(group->min_xheight, group->bounding_box.botleft(), group->bounding_box.topright()); for (int b = 0; b < group->blocks.size(); ++b) { diff --git a/textord/tospace.cpp b/textord/tospace.cpp index bec346ef52..4358436917 100644 --- a/textord/tospace.cpp +++ b/textord/tospace.cpp @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. /********************************************************************** * tospace.cpp * @@ -419,9 +428,8 @@ void Textord::row_spacing_stats( if (suspected_table && (row->space_size < tosp_table_kn_sp_ratio * row->kern_size)) { if (tosp_debug_level > 5) - tprintf ("B:%d R:%d -- DON'T BELIEVE SPACE %3.2f %d %3.2f.\n", - block_idx, row_idx, - row->kern_size, row->space_threshold, row->space_size); + tprintf("B:%d R:%d -- DON'T BELIEVE SPACE %3.2f %d %3.2f.\n", block_idx, + row_idx, row->kern_size, row->space_threshold, row->space_size); row->space_threshold = (inT32) (tosp_table_kn_sp_ratio * row->kern_size); row->space_size = MAX (row->space_threshold + 1, row->xheight); @@ -441,10 +449,9 @@ void Textord::row_spacing_stats( MAX (tosp_min_sane_kn_sp * MAX (row->kern_size, 2.5), row->xheight / 2); if (tosp_debug_level > 5) - tprintf - ("B:%d R:%d -- DON'T BELIEVE SPACE %3.2f %d %3.2f -> %3.2f.\n", - block_idx, row_idx, row->kern_size, row->space_threshold, - row->space_size, sane_space); + tprintf("B:%d R:%d -- DON'T BELIEVE SPACE %3.2f %d %3.2f -> %3.2f.\n", + block_idx, row_idx, row->kern_size, row->space_threshold, + row->space_size, sane_space); row->space_size = sane_space; row->space_threshold = inT32 (floor ((row->space_size + row->kern_size) / @@ -455,10 +462,9 @@ void Textord::row_spacing_stats( MAX (row->kern_size, 2.5))); if (row->space_threshold > sane_threshold) { if (tosp_debug_level > 5) - tprintf ("B:%d R:%d -- DON'T BELIEVE THRESH %3.2f %d %3.2f->%d.\n", - block_idx, row_idx, - row->kern_size, - row->space_threshold, row->space_size, sane_threshold); + tprintf("B:%d R:%d -- DON'T BELIEVE THRESH %3.2f %d %3.2f->%d.\n", + block_idx, row_idx, row->kern_size, row->space_threshold, + row->space_size, sane_threshold); row->space_threshold = sane_threshold; if (row->space_size <= sane_threshold) row->space_size = row->space_threshold + 1.0f; @@ -498,7 +504,7 @@ void Textord::row_spacing_stats( MIN (inT32 (ceil (tosp_fuzzy_space_factor * row->xheight)), inT32 (row->space_size)); if (row->min_space <= row->space_threshold) - //Don't be silly + // Don't be silly row->min_space = row->space_threshold + 1; /* Lets try to guess the max certain kern gap by looking at the cluster of @@ -559,7 +565,7 @@ void Textord::row_spacing_stats( row->kern_size)); } if (row->max_nonspace > row->space_threshold) { - //Don't be silly + // Don't be silly row->max_nonspace = row->space_threshold; } @@ -700,8 +706,8 @@ BOOL8 Textord::isolated_row_stats(TO_ROW *row, ((small_gaps_count / (float) total) < tosp_enough_small_gaps) || (total - small_gaps_count < 1)) { if (tosp_debug_level > 5) - tprintf ("B:%d R:%d -- Can't do isolated row stats.\n", - block_idx, row_idx); + tprintf("B:%d R:%d -- Can't do isolated row stats.\n", block_idx, + row_idx); return FALSE; } blob_it.set_to_list (row->blob_list ()); @@ -1130,10 +1136,10 @@ ROW *Textord::make_prop_words( else blanks = 0; if (tosp_debug_level > 5) - tprintf - ("Repch wd at EOL (%d,%d). rep spacing %5.2f; Lgap:%d (%d blanks)\n", - word->bounding_box ().left (), word->bounding_box ().bottom (), - repetition_spacing, current_gap, blanks); + tprintf( + "Repch wd at EOL (%d,%d). rep spacing %5.2f; Lgap:%d (%d blanks)\n", + word->bounding_box().left(), word->bounding_box().bottom(), + repetition_spacing, current_gap, blanks); word->set_blanks (blanks); //NO uncertainty word->set_flag (W_FUZZY_SP, FALSE); @@ -1326,9 +1332,10 @@ BOOL8 Textord::make_a_word_break( we may need to set PARTICULAR spaces to fuzzy or not. The values will ONLY be used if the function returns TRUE - ie the word is to be broken. */ - blanks = (uinT8) (current_gap / row->space_size); - if (blanks < 1) - blanks = 1; + int num_blanks = current_gap; + if (row->space_size > 1.0f) + num_blanks = IntCastRounded(current_gap / row->space_size); + blanks = static_cast(ClipToRange(num_blanks, 1, MAX_UINT8)); fuzzy_sp = FALSE; fuzzy_non = FALSE; /* @@ -1686,10 +1693,9 @@ void Textord::mark_gap( blob.bottom () + blob.height () / 2.0f); } if (tosp_debug_level > 5) - tprintf (" (%d,%d) Sp<->Kn Rule %d %d %d %d %d %d\n", - blob.left () - current_gap / 2, blob.bottom (), rule, - prev_gap, prev_blob_width, current_gap, - next_blob_width, next_gap); + tprintf(" (%d,%d) Sp<->Kn Rule %d %d %d %d %d %d\n", + blob.left() - current_gap / 2, blob.bottom(), rule, prev_gap, + prev_blob_width, current_gap, next_blob_width, next_gap); } #endif @@ -1727,8 +1733,7 @@ BOOL8 Textord::ignore_big_gap(TO_ROW *row, inT16 right) { inT16 gap = right - left + 1; - if (tosp_ignore_big_gaps > 999) - return FALSE; //Don't ignore + if (tosp_ignore_big_gaps > 999) return FALSE; // Don't ignore if (tosp_ignore_big_gaps > 0) return (gap > tosp_ignore_big_gaps * row->xheight); if (gap > tosp_ignore_very_big_gaps * row->xheight) @@ -1750,7 +1755,6 @@ BOOL8 Textord::ignore_big_gap(TO_ROW *row, return FALSE; } - /********************************************************************** * reduced_box_next * diff --git a/textord/tovars.cpp b/textord/tovars.cpp index 71114358ac..6b1b833248 100644 --- a/textord/tovars.cpp +++ b/textord/tovars.cpp @@ -1,8 +1,8 @@ /********************************************************************** * File: tovars.cpp (Formerly to_vars.c) * Description: Variables used by textord. - * Author: Ray Smith - * Created: Tue Aug 24 16:55:02 BST 1993 + * Author: Ray Smith + * Created: Tue Aug 24 16:55:02 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -49,8 +49,8 @@ EXTERN double_VAR (textord_words_default_minspace, 0.6, EXTERN double_VAR (textord_words_min_minspace, 0.3, "Fraction of xheight"); EXTERN double_VAR (textord_words_default_nonspace, 0.2, "Fraction of xheight"); -EXTERN double_VAR (textord_words_initial_lower, 0.25, -"Max initial cluster size"); +EXTERN double_VAR(textord_words_initial_lower, 0.25, + "Max initial cluster size"); EXTERN double_VAR (textord_words_initial_upper, 0.15, "Min initial cluster spacing"); EXTERN double_VAR (textord_words_minlarge, 0.75, @@ -67,7 +67,7 @@ EXTERN double_VAR (textord_pitch_rowsimilarity, 0.08, "Fraction of xheight for sameness"); EXTERN BOOL_VAR (textord_pitch_scalebigwords, FALSE, "Scale scores on big words"); -EXTERN double_VAR (words_initial_lower, 0.5, "Max initial cluster size"); +EXTERN double_VAR(words_initial_lower, 0.5, "Max initial cluster size"); EXTERN double_VAR (words_initial_upper, 0.15, "Min initial cluster spacing"); EXTERN double_VAR (words_default_prop_nonspace, 0.25, "Fraction of xheight"); EXTERN double_VAR (words_default_fixed_space, 0.75, "Fraction of xheight"); diff --git a/textord/tovars.h b/textord/tovars.h index 99edae1d2c..46315bb96b 100644 --- a/textord/tovars.h +++ b/textord/tovars.h @@ -1,8 +1,8 @@ /********************************************************************** * File: tovars.h (Formerly to_vars.h) * Description: Variables used by textord. - * Author: Ray Smith - * Created: Tue Aug 24 16:55:02 BST 1993 + * Author: Ray Smith + * Created: Tue Aug 24 16:55:02 BST 1993 * * (C) Copyright 1993, Hewlett-Packard Ltd. ** Licensed under the Apache License, Version 2.0 (the "License"); @@ -51,8 +51,8 @@ extern double_VAR_H (textord_words_default_minspace, 0.6, extern double_VAR_H (textord_words_min_minspace, 0.3, "Fraction of xheight"); extern double_VAR_H (textord_words_default_nonspace, 0.2, "Fraction of xheight"); -extern double_VAR_H (textord_words_initial_lower, 0.25, -"Max initial cluster size"); +extern double_VAR_H(textord_words_initial_lower, 0.25, + "Max initial cluster size"); extern double_VAR_H (textord_words_initial_upper, 0.15, "Min initial cluster spacing"); extern double_VAR_H (textord_words_minlarge, 0.75, @@ -69,7 +69,7 @@ extern double_VAR_H (textord_pitch_rowsimilarity, 0.08, "Fraction of xheight for sameness"); extern BOOL_VAR_H (textord_pitch_scalebigwords, FALSE, "Scale scores on big words"); -extern double_VAR_H (words_initial_lower, 0.5, "Max initial cluster size"); +extern double_VAR_H(words_initial_lower, 0.5, "Max initial cluster size"); extern double_VAR_H (words_initial_upper, 0.15, "Min initial cluster spacing"); extern double_VAR_H (words_default_prop_nonspace, 0.25, diff --git a/training/CMakeLists.txt b/training/CMakeLists.txt index a733e73ea0..c13ed77479 100644 --- a/training/CMakeLists.txt +++ b/training/CMakeLists.txt @@ -2,10 +2,16 @@ # tesseract # -if (STATIC OR NOT (WIN32 OR CYGWIN)) +if (NOT CPPAN_BUILD AND NOT (WIN32 OR CYGWIN)) + return() +endif() + +if (CPPAN_BUILD) + set(ICU_FOUND 1) +endif() # experimental -if (MSVC) +if (MSVC AND NOT CPPAN_BUILD) include(CheckTypeSize) check_type_size("void *" SIZEOF_VOID_P) @@ -48,10 +54,12 @@ endif() set(ICU_ROOT ${icu_dir}/icu) -endif(MSVC) +endif() # experimental +if (NOT CPPAN_BUILD) find_package(ICU COMPONENTS uc i18n) +endif() ######################################## # LIBRARY tessopt @@ -74,7 +82,7 @@ set(common_training_hdr commontraining.h ) add_library (common_training ${common_training_src} ${common_training_hdr}) -target_link_libraries (common_training tesseract tessopt) +target_link_libraries (common_training libtesseract tessopt) project_group (common_training "Training Tools") @@ -83,7 +91,7 @@ project_group (common_training "Training Tools") ######################################## add_executable (ambiguous_words ambiguous_words.cpp) -target_link_libraries (ambiguous_words tesseract) +target_link_libraries (ambiguous_words libtesseract) project_group (ambiguous_words "Training Tools") @@ -101,7 +109,7 @@ project_group (classifier_tester "Training Tools") ######################################## add_executable (combine_tessdata combine_tessdata.cpp) -target_link_libraries (combine_tessdata tesseract) +target_link_libraries (combine_tessdata libtesseract) project_group (combine_tessdata "Training Tools") @@ -119,7 +127,7 @@ project_group (cntraining "Training Tools") ######################################## add_executable (dawg2wordlist dawg2wordlist.cpp) -target_link_libraries (dawg2wordlist tesseract) +target_link_libraries (dawg2wordlist libtesseract) project_group (dawg2wordlist "Training Tools") @@ -146,7 +154,7 @@ project_group (shapeclustering "Training Tools") ######################################## add_executable (unicharset_extractor unicharset_extractor.cpp) -target_link_libraries (unicharset_extractor tesseract tessopt) +target_link_libraries (unicharset_extractor libtesseract tessopt) project_group (unicharset_extractor "Training Tools") @@ -155,7 +163,7 @@ project_group (unicharset_extractor "Training Tools") ######################################## add_executable (wordlist2dawg wordlist2dawg.cpp) -target_link_libraries (wordlist2dawg tesseract) +target_link_libraries (wordlist2dawg libtesseract) project_group (wordlist2dawg "Training Tools") @@ -165,7 +173,9 @@ project_group (wordlist2dawg "Training Tools") if (ICU_FOUND) +if (NOT CPPAN_BUILD) include_directories(${ICU_INCLUDE_DIRS}) +endif() add_executable (set_unicharset_properties set_unicharset_properties.cpp @@ -177,7 +187,11 @@ add_executable (set_unicharset_properties normstrngs.h icuerrorcode.h ) +if (NOT CPPAN_BUILD) target_link_libraries (set_unicharset_properties common_training ${ICU_LIBRARIES}) +else() +target_link_libraries (set_unicharset_properties common_training pvt.cppan.demo.unicode.icu.i18n) +endif() project_group (set_unicharset_properties "Training Tools") @@ -221,7 +235,7 @@ endif() add_executable (text2image ${text2image_src}) target_include_directories (text2image BEFORE PRIVATE ${Cairo_INCLUDE_DIRS} ${Pango_INCLUDE_DIRS}) target_compile_definitions (text2image PRIVATE -DPANGO_ENABLE_ENGINE) -target_link_libraries (text2image tesseract common_training +target_link_libraries (text2image libtesseract common_training ${ICU_LIBRARIES} ${Pango_LIBRARIES} ${Cairo_LIBRARIES} @@ -233,6 +247,5 @@ project_group (text2image "Training Tools") endif(PKG_CONFIG_FOUND) endif(ICU_FOUND) -endif(STATIC OR NOT (WIN32 OR CYGWIN)) ############################################################################### diff --git a/training/Makefile.am b/training/Makefile.am index fe3d85bcdc..2506226654 100644 --- a/training/Makefile.am +++ b/training/Makefile.am @@ -10,14 +10,7 @@ AM_CPPFLAGS += \ EXTRA_DIST = language-specific.sh tesstrain.sh tesstrain_utils.sh -if MINGW -# try static build -#AM_LDFLAGS += -all-static -#libic=-lsicuin -licudt -lsicuuc -libicu=-licuin -licuuc -else libicu=-licui18n -licuuc -endif # TODO: training programs can not be linked to shared library created # with -fvisibility if VISIBILITY diff --git a/training/boxchar.cpp b/training/boxchar.cpp index b99c12a600..d3d5160895 100644 --- a/training/boxchar.cpp +++ b/training/boxchar.cpp @@ -49,7 +49,8 @@ void BoxChar::AddBox(int x, int y, int width, int height) { } /* static */ -void BoxChar::TranslateBoxes(int xshift, int yshift, vector* boxes) { +void BoxChar::TranslateBoxes(int xshift, int yshift, + std::vector* boxes) { for (int i = 0; i < boxes->size(); ++i) { BOX* box = (*boxes)[i]->box_; if (box != NULL) { @@ -62,7 +63,7 @@ void BoxChar::TranslateBoxes(int xshift, int yshift, vector* boxes) { // Prepares for writing the boxes to a file by inserting newlines, spaces, // and re-ordering so the boxes are strictly left-to-right. /* static */ -void BoxChar::PrepareToWrite(vector* boxes) { +void BoxChar::PrepareToWrite(std::vector* boxes) { bool rtl_rules = ContainsMostlyRTL(*boxes); bool vertical_rules = MostlyVertical(*boxes); InsertNewlines(rtl_rules, vertical_rules, boxes); @@ -73,13 +74,12 @@ void BoxChar::PrepareToWrite(vector* boxes) { if (rtl_rules) { ReorderRTLText(boxes); } - tprintf("Rtl = %d ,vertical=%d\n", rtl_rules, vertical_rules); } // Inserts newline (tab) characters into the vector at newline positions. /* static */ void BoxChar::InsertNewlines(bool rtl_rules, bool vertical_rules, - vector* boxes) { + std::vector* boxes) { int prev_i = -1; int max_shift = 0; for (int i = 0; i < boxes->size(); ++i) { @@ -142,7 +142,7 @@ void BoxChar::InsertNewlines(bool rtl_rules, bool vertical_rules, // Converts NULL boxes to space characters, with appropriate bounding boxes. /* static */ void BoxChar::InsertSpaces(bool rtl_rules, bool vertical_rules, - vector* boxes) { + std::vector* boxes) { // After InsertNewlines, any remaining null boxes are not newlines, and are // singletons, so add a box to each remaining null box. for (int i = 1; i + 1 < boxes->size(); ++i) { @@ -198,7 +198,7 @@ void BoxChar::InsertSpaces(bool rtl_rules, bool vertical_rules, // Reorders text in a right-to-left script in left-to-right order. /* static */ -void BoxChar::ReorderRTLText(vector* boxes) { +void BoxChar::ReorderRTLText(std::vector* boxes) { // After adding newlines and spaces, this task is simply a matter of sorting // by left each group of boxes between newlines. BoxCharPtrSort sorter; @@ -212,7 +212,7 @@ void BoxChar::ReorderRTLText(vector* boxes) { // Returns true if the vector contains mostly RTL characters. /* static */ -bool BoxChar::ContainsMostlyRTL(const vector& boxes) { +bool BoxChar::ContainsMostlyRTL(const std::vector& boxes) { int num_rtl = 0, num_ltr = 0; for (int i = 0; i < boxes.size(); ++i) { // Convert the unichar to UTF32 representation @@ -241,7 +241,7 @@ bool BoxChar::ContainsMostlyRTL(const vector& boxes) { // Returns true if the text is mostly laid out vertically. /* static */ -bool BoxChar::MostlyVertical(const vector& boxes) { +bool BoxChar::MostlyVertical(const std::vector& boxes) { inT64 total_dx = 0, total_dy = 0; for (int i = 1; i < boxes.size(); ++i) { if (boxes[i - 1]->box_ != NULL && boxes[i]->box_ != NULL && @@ -260,7 +260,7 @@ bool BoxChar::MostlyVertical(const vector& boxes) { // Returns the total length of all the strings in the boxes. /* static */ -int BoxChar::TotalByteLength(const vector& boxes) { +int BoxChar::TotalByteLength(const std::vector& boxes) { int total_length = 0; for (int i = 0; i < boxes.size(); ++i) total_length += boxes[i]->ch_.size(); return total_length; @@ -270,7 +270,8 @@ int BoxChar::TotalByteLength(const vector& boxes) { // The rotation is in radians clockwise about the given center. /* static */ void BoxChar::RotateBoxes(float rotation, int xcenter, int ycenter, - int start_box, int end_box, vector* boxes) { + int start_box, int end_box, + std::vector* boxes) { Boxa* orig = boxaCreate(0); for (int i = start_box; i < end_box; ++i) { BOX* box = (*boxes)[i]->box_; @@ -290,14 +291,21 @@ void BoxChar::RotateBoxes(float rotation, int xcenter, int ycenter, const int kMaxLineLength = 1024; /* static */ void BoxChar::WriteTesseractBoxFile(const string& filename, int height, - const vector& boxes) { + const std::vector& boxes) { + string output = GetTesseractBoxStr(height, boxes); + File::WriteStringToFileOrDie(output, filename); +} + +/* static */ +string BoxChar::GetTesseractBoxStr(int height, + const std::vector& boxes) { string output; char buffer[kMaxLineLength]; for (int i = 0; i < boxes.size(); ++i) { const Box* box = boxes[i]->box_; if (box == NULL) { tprintf("Error: Call PrepareToWrite before WriteTesseractBoxFile!!\n"); - return; + return ""; } int nbytes = snprintf(buffer, kMaxLineLength, "%s %d %d %d %d %d\n", @@ -305,6 +313,7 @@ void BoxChar::WriteTesseractBoxFile(const string& filename, int height, box->x + box->w, height - box->y, boxes[i]->page_); output.append(buffer, nbytes); } - File::WriteStringToFileOrDie(output, filename); + return output; } + } // namespace tesseract diff --git a/training/boxchar.h b/training/boxchar.h index 27b568a143..3748c4abad 100644 --- a/training/boxchar.h +++ b/training/boxchar.h @@ -100,6 +100,9 @@ class BoxChar { // is needed to convert to tesseract coordinates. static void WriteTesseractBoxFile(const string& name, int height, const vector& boxes); + // Gets the tesseract box file as a string from the vector of boxes. + // The image height is needed to convert to tesseract coordinates. + static string GetTesseractBoxStr(int height, const vector& boxes); private: string ch_; diff --git a/training/classifier_tester.cpp b/training/classifier_tester.cpp index 48f3781ebb..ed7e50cd2f 100644 --- a/training/classifier_tester.cpp +++ b/training/classifier_tester.cpp @@ -48,9 +48,9 @@ enum ClassifierName { const char* names[] = {"pruner", "full", #ifndef NO_CUBE_BUILD - "cube", "cubetess", + "cube", "cubetess", #endif // NO_CUBE_BUILD - NULL }; + NULL}; static tesseract::ShapeClassifier* InitializeClassifier( const char* classifer_name, const UNICHARSET& unicharset, @@ -80,7 +80,7 @@ static tesseract::ShapeClassifier* InitializeClassifier( tesseract::Classify* classify = NULL; if ( #ifndef NO_CUBE_BUILD - classifier == CN_CUBE || classifier == CN_CUBETESS || + classifier == CN_CUBE || classifier == CN_CUBETESS || #endif // NO_CUBE_BUILD classifier == CN_PRUNER || classifier == CN_FULL) { #ifndef NO_CUBE_BUILD diff --git a/training/cntraining.cpp b/training/cntraining.cpp index ab19ddb93d..6f4f42aebe 100644 --- a/training/cntraining.cpp +++ b/training/cntraining.cpp @@ -20,7 +20,6 @@ ** limitations under the License. ******************************************************************************/ - /*---------------------------------------------------------------------------- Include Files and Type Defines ----------------------------------------------------------------------------*/ @@ -53,10 +52,8 @@ int main ( Private Function Prototypes ----------------------------------------------------------------------------*/ -void WriteNormProtos ( - const char *Directory, - LIST LabeledProtoList, - CLUSTERER *Clusterer); +void WriteNormProtos(const char *Directory, LIST LabeledProtoList, + const FEATURE_DESC_STRUCT *feature_desc); /* PARAMDESC *ConvertToPARAMDESC( @@ -81,7 +78,6 @@ CLUSTERCONFIG CNConfig = elliptical, 0.025, 0.05, 0.8, 1e-3, 0 }; - /*---------------------------------------------------------------------------- Public Code ----------------------------------------------------------------------------*/ @@ -134,8 +130,7 @@ CLUSTERCONFIG CNConfig = * @note Exceptions: none * @note History: Fri Aug 18 08:56:17 1989, DSJ, Created. */ -int main(int argc, char* argv[]) -{ +int main(int argc, char *argv[]) { // Set the global Config parameters before parsing the command line. Config = CNConfig; @@ -165,13 +160,18 @@ int main(int argc, char* argv[]) // reduce the min samples: // Config.MinSamples = 0.5 / num_fonts; pCharList = CharList; + // The norm protos will count the source protos, so we keep them here in + // freeable_protos, so they can be freed later. + GenericVector freeable_protos; iterate(pCharList) { //Cluster - if (Clusterer) - FreeClusterer(Clusterer); CharSample = (LABELEDLIST)first_node(pCharList); Clusterer = SetUpForClustering(FeatureDefs, CharSample, PROGRAM_FEATURE_TYPE); + if (Clusterer == NULL) { // To avoid a SIGSEGV + fprintf(stderr, "Error: NULL clusterer!\n"); + return 1; + } float SavedMinSamples = Config.MinSamples; // To disable the tendency to produce a single cluster for all fonts, // make MagicSamples an impossible to achieve number: @@ -190,21 +190,21 @@ int main(int argc, char* argv[]) } Config.MinSamples = SavedMinSamples; AddToNormProtosList(&NormProtoList, ProtoList, CharSample->Label); + freeable_protos.push_back(ProtoList); + FreeClusterer(Clusterer); } FreeTrainingSamples(CharList); - if (Clusterer == NULL) { // To avoid a SIGSEGV - fprintf(stderr, "Error: NULL clusterer!\n"); - return 1; - } - WriteNormProtos(FLAGS_D.c_str(), NormProtoList, Clusterer); + int desc_index = ShortNameToFeatureType(FeatureDefs, PROGRAM_FEATURE_TYPE); + WriteNormProtos(FLAGS_D.c_str(), NormProtoList, + FeatureDefs.FeatureDesc[desc_index]); FreeNormProtoList(NormProtoList); - FreeProtoList(&ProtoList); - FreeClusterer(Clusterer); + for (int i = 0; i < freeable_protos.size(); ++i) { + FreeProtoList(&freeable_protos[i]); + } printf ("\n"); return 0; } // main - /*---------------------------------------------------------------------------- Private Code ----------------------------------------------------------------------------*/ @@ -216,16 +216,13 @@ int main(int argc, char* argv[]) * of the samples. * @param Directory directory to place sample files into * @param LabeledProtoList List of labeled protos -* @param Clusterer The CLUSTERER to use +* @param feature_desc Description of the features * @return none * @note Exceptions: none * @note History: Fri Aug 18 16:17:06 1989, DSJ, Created. */ -void WriteNormProtos ( - const char *Directory, - LIST LabeledProtoList, - CLUSTERER *Clusterer) -{ +void WriteNormProtos(const char *Directory, LIST LabeledProtoList, + const FEATURE_DESC_STRUCT *feature_desc) { FILE *File; STRING Filename; LABELEDLIST LabeledProto; @@ -240,8 +237,8 @@ void WriteNormProtos ( Filename += "normproto"; printf ("\nWriting %s ...", Filename.string()); File = Efopen (Filename.string(), "wb"); - fprintf(File,"%0d\n",Clusterer->SampleSize); - WriteParamDesc(File,Clusterer->SampleSize,Clusterer->ParamDesc); + fprintf(File, "%0d\n", feature_desc->NumParams); + WriteParamDesc(File, feature_desc->NumParams, feature_desc->ParamDesc); iterate(LabeledProtoList) { LabeledProto = (LABELEDLIST) first_node (LabeledProtoList); @@ -256,7 +253,7 @@ void WriteNormProtos ( exit(1); } fprintf(File, "\n%s %d\n", LabeledProto->Label, N); - WriteProtos(File, Clusterer->SampleSize, LabeledProto->List, true, false); + WriteProtos(File, feature_desc->NumParams, LabeledProto->List, true, false); } fclose (File); diff --git a/training/commandlineflags.cpp b/training/commandlineflags.cpp index 06bfbe6589..56dcb84221 100644 --- a/training/commandlineflags.cpp +++ b/training/commandlineflags.cpp @@ -1,3 +1,12 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. #include "commandlineflags.h" #ifdef USE_STD_NAMESPACE @@ -106,36 +115,36 @@ void PrintCommandLineFlags() { if (!strncmp(GlobalParams()->int_params[i]->name_str(), kFlagNamePrefix, kFlagNamePrefixLen)) { printf(" --%s %s (type:int default:%d)\n", - GlobalParams()->int_params[i]->name_str() + kFlagNamePrefixLen, - GlobalParams()->int_params[i]->info_str(), - inT32(*(GlobalParams()->int_params[i]))); + GlobalParams()->int_params[i]->name_str() + kFlagNamePrefixLen, + GlobalParams()->int_params[i]->info_str(), + inT32(*(GlobalParams()->int_params[i]))); } } for (int i = 0; i < GlobalParams()->double_params.size(); ++i) { if (!strncmp(GlobalParams()->double_params[i]->name_str(), kFlagNamePrefix, kFlagNamePrefixLen)) { printf(" --%s %s (type:double default:%g)\n", - GlobalParams()->double_params[i]->name_str() + kFlagNamePrefixLen, - GlobalParams()->double_params[i]->info_str(), - static_cast(*(GlobalParams()->double_params[i]))); + GlobalParams()->double_params[i]->name_str() + kFlagNamePrefixLen, + GlobalParams()->double_params[i]->info_str(), + static_cast(*(GlobalParams()->double_params[i]))); } } for (int i = 0; i < GlobalParams()->bool_params.size(); ++i) { if (!strncmp(GlobalParams()->bool_params[i]->name_str(), kFlagNamePrefix, kFlagNamePrefixLen)) { printf(" --%s %s (type:bool default:%s)\n", - GlobalParams()->bool_params[i]->name_str() + kFlagNamePrefixLen, - GlobalParams()->bool_params[i]->info_str(), - (BOOL8(*(GlobalParams()->bool_params[i])) ? "true" : "false")); + GlobalParams()->bool_params[i]->name_str() + kFlagNamePrefixLen, + GlobalParams()->bool_params[i]->info_str(), + (BOOL8(*(GlobalParams()->bool_params[i])) ? "true" : "false")); } } for (int i = 0; i < GlobalParams()->string_params.size(); ++i) { if (!strncmp(GlobalParams()->string_params[i]->name_str(), kFlagNamePrefix, kFlagNamePrefixLen)) { printf(" --%s %s (type:string default:%s)\n", - GlobalParams()->string_params[i]->name_str() + kFlagNamePrefixLen, - GlobalParams()->string_params[i]->info_str(), - GlobalParams()->string_params[i]->string()); + GlobalParams()->string_params[i]->name_str() + kFlagNamePrefixLen, + GlobalParams()->string_params[i]->info_str(), + GlobalParams()->string_params[i]->string()); } } } diff --git a/training/commontraining.cpp b/training/commontraining.cpp index 1c4cc832bc..57ecdbecef 100644 --- a/training/commontraining.cpp +++ b/training/commontraining.cpp @@ -39,7 +39,6 @@ #include using tesseract::CCUtil; -using tesseract::FontInfo; using tesseract::IntFeatureSpace; using tesseract::ParamUtils; using tesseract::ShapeTable; @@ -312,9 +311,7 @@ const char *GetNextFilename(int argc, const char* const * argv) { return argv[tessoptind++]; else return NULL; -} /* GetNextFilename */ - - +} /* GetNextFilename */ /*---------------------------------------------------------------------------*/ /** @@ -328,11 +325,8 @@ const char *GetNextFilename(int argc, const char* const * argv) { * @note Exceptions: none * @note History: Fri Aug 18 15:57:41 1989, DSJ, Created. */ -LABELEDLIST FindList ( - LIST List, - char *Label) -{ - LABELEDLIST LabeledList; +LABELEDLIST FindList(LIST List, char* Label) { + LABELEDLIST LabeledList; iterate (List) { @@ -342,7 +336,7 @@ LABELEDLIST FindList ( } return (NULL); -} /* FindList */ +} /* FindList */ /*---------------------------------------------------------------------------*/ /** @@ -354,10 +348,8 @@ LABELEDLIST FindList ( * @note Exceptions: none * @note History: Fri Aug 18 16:08:46 1989, DSJ, Created. */ -LABELEDLIST NewLabeledList ( - const char *Label) -{ - LABELEDLIST LabeledList; +LABELEDLIST NewLabeledList(const char* Label) { + LABELEDLIST LabeledList; LabeledList = (LABELEDLIST) Emalloc (sizeof (LABELEDLISTNODE)); LabeledList->Label = (char*)Emalloc (strlen (Label)+1); @@ -367,7 +359,7 @@ LABELEDLIST NewLabeledList ( LabeledList->font_sample_count = 0; return (LabeledList); -} /* NewLabeledList */ +} /* NewLabeledList */ /*---------------------------------------------------------------------------*/ // TODO(rays) This is now used only by cntraining. Convert cntraining to use @@ -386,7 +378,7 @@ LABELEDLIST NewLabeledList ( * @return none * @note Globals: none * @note Exceptions: none - * @note History: + * @note History: * - Fri Aug 18 13:11:39 1989, DSJ, Created. * - Tue May 17 1998 simplifications to structure, illiminated * font, and feature specification levels of structure. @@ -460,17 +452,17 @@ void FreeTrainingSamples(LIST CharList) { FEATURE_SET FeatureSet; LIST FeatureList; - - iterate(CharList) { /* iterate through all of the fonts */ + LIST nodes = CharList; + iterate(CharList) { /* iterate through all of the fonts */ char_sample = (LABELEDLIST) first_node(CharList); FeatureList = char_sample->List; - iterate(FeatureList) { /* iterate through all of the classes */ + iterate(FeatureList) { /* iterate through all of the classes */ FeatureSet = (FEATURE_SET) first_node(FeatureList); FreeFeatureSet(FeatureSet); } FreeLabeledList(char_sample); } - destroy(CharList); + destroy(nodes); } /* FreeTrainingSamples */ /*---------------------------------------------------------------------------*/ @@ -532,15 +524,15 @@ CLUSTERER *SetUpForClustering(const FEATURE_DEFS_STRUCT &FeatureDefs, } CharID++; } - if ( Sample != NULL ) free( Sample ); - return( Clusterer ); + free(Sample); + return Clusterer; -} /* SetUpForClustering */ +} /* SetUpForClustering */ /*------------------------------------------------------------------------*/ void MergeInsignificantProtos(LIST ProtoList, const char* label, - CLUSTERER *Clusterer, CLUSTERCONFIG *Config) { - PROTOTYPE *Prototype; + CLUSTERER* Clusterer, CLUSTERCONFIG* Config) { + PROTOTYPE* Prototype; bool debug = strcmp(FLAGS_test_ch.c_str(), label) == 0; LIST pProtoList = ProtoList; @@ -600,7 +592,7 @@ void MergeInsignificantProtos(LIST ProtoList, const char* label, Prototype->Significant = true; } } -} /* MergeInsignificantProtos */ +} /* MergeInsignificantProtos */ /*-----------------------------------------------------------------------------*/ void CleanUpUnusedData( @@ -695,14 +687,11 @@ LIST RemoveInsignificantProtos( } FreeProtoList(&ProtoList); return (NewProtoList); -} /* RemoveInsignificantProtos */ +} /* RemoveInsignificantProtos */ /*----------------------------------------------------------------------------*/ -MERGE_CLASS FindClass ( - LIST List, - const char *Label) -{ - MERGE_CLASS MergeClass; +MERGE_CLASS FindClass(LIST List, const char* Label) { + MERGE_CLASS MergeClass; iterate (List) { @@ -712,13 +701,11 @@ MERGE_CLASS FindClass ( } return (NULL); -} /* FindClass */ +} /* FindClass */ /*---------------------------------------------------------------------------*/ -MERGE_CLASS NewLabeledClass ( - const char *Label) -{ - MERGE_CLASS MergeClass; +MERGE_CLASS NewLabeledClass(const char* Label) { + MERGE_CLASS MergeClass; MergeClass = new MERGE_CLASS_NODE; MergeClass->Label = (char*)Emalloc (strlen (Label)+1); @@ -726,7 +713,7 @@ MERGE_CLASS NewLabeledClass ( MergeClass->Class = NewClass (MAX_NUM_PROTOS, MAX_NUM_CONFIGS); return (MergeClass); -} /* NewLabeledClass */ +} /* NewLabeledClass */ /*-----------------------------------------------------------------------------*/ /** @@ -738,38 +725,37 @@ MERGE_CLASS NewLabeledClass ( * @note Exceptions: none * @note History: Fri Aug 18 17:44:27 1989, DSJ, Created. */ -void FreeLabeledClassList ( - LIST ClassList) -{ - MERGE_CLASS MergeClass; +void FreeLabeledClassList(LIST ClassList) { + MERGE_CLASS MergeClass; - iterate (ClassList) /* iterate through all of the fonts */ + LIST nodes = ClassList; + iterate(ClassList) /* iterate through all of the fonts */ { MergeClass = (MERGE_CLASS) first_node (ClassList); free (MergeClass->Label); FreeClass(MergeClass->Class); delete MergeClass; } - destroy (ClassList); + destroy(nodes); -} /* FreeLabeledClassList */ +} /* FreeLabeledClassList */ /* SetUpForFloat2Int */ CLASS_STRUCT* SetUpForFloat2Int(const UNICHARSET& unicharset, LIST LabeledClassList) { - MERGE_CLASS MergeClass; - CLASS_TYPE Class; - int NumProtos; - int NumConfigs; - int NumWords; - int i, j; - float Values[3]; - PROTO NewProto; - PROTO OldProto; - BIT_VECTOR NewConfig; - BIT_VECTOR OldConfig; - - // printf("Float2Int ...\n"); + MERGE_CLASS MergeClass; + CLASS_TYPE Class; + int NumProtos; + int NumConfigs; + int NumWords; + int i, j; + float Values[3]; + PROTO NewProto; + PROTO OldProto; + BIT_VECTOR NewConfig; + BIT_VECTOR OldConfig; + + // printf("Float2Int ...\n"); CLASS_STRUCT* float_classes = new CLASS_STRUCT[unicharset.size()]; iterate(LabeledClassList) @@ -821,9 +807,9 @@ CLASS_STRUCT* SetUpForFloat2Int(const UNICHARSET& unicharset, void Normalize ( float *Values) { - register float Slope; - register float Intercept; - register float Normalizer; + float Slope; + float Intercept; + float Normalizer; Slope = tan (Values [2] * 2 * PI); Intercept = Values [1] - Slope * Values [0]; @@ -835,20 +821,20 @@ void Normalize ( } // Normalize /*-------------------------------------------------------------------------*/ -void FreeNormProtoList ( - LIST CharList) +void FreeNormProtoList(LIST CharList) { - LABELEDLIST char_sample; + LABELEDLIST char_sample; - iterate (CharList) /* iterate through all of the fonts */ + LIST nodes = CharList; + iterate(CharList) /* iterate through all of the fonts */ { char_sample = (LABELEDLIST) first_node (CharList); FreeLabeledList (char_sample); } - destroy (CharList); + destroy(nodes); -} // FreeNormProtoList +} // FreeNormProtoList /*---------------------------------------------------------------------------*/ void AddToNormProtosList( @@ -869,19 +855,16 @@ void AddToNormProtosList( } /*---------------------------------------------------------------------------*/ -int NumberOfProtos( - LIST ProtoList, - BOOL8 CountSigProtos, - BOOL8 CountInsigProtos) -{ +int NumberOfProtos(LIST ProtoList, BOOL8 CountSigProtos, + BOOL8 CountInsigProtos) { int N = 0; - PROTOTYPE *Proto; + PROTOTYPE* Proto; iterate(ProtoList) { Proto = (PROTOTYPE *) first_node ( ProtoList ); - if (( Proto->Significant && CountSigProtos ) || - ( ! Proto->Significant && CountInsigProtos ) ) + if ((Proto->Significant && CountSigProtos) || + (!Proto->Significant && CountInsigProtos)) N++; } return(N); diff --git a/training/degradeimage.cpp b/training/degradeimage.cpp index f9c3cfb048..333f3703dc 100644 --- a/training/degradeimage.cpp +++ b/training/degradeimage.cpp @@ -22,10 +22,36 @@ #include #include "allheaders.h" // from leptonica +#include "genericvector.h" #include "helpers.h" // For TRand. +#include "rect.h" namespace tesseract { +// A randomized perspective distortion can be applied to synthetic input. +// The perspective distortion comes from leptonica, which uses 2 sets of 4 +// corners to determine the distortion. There are random values for each of +// the x numbers x0..x3 and y0..y3, except for x2 and x3 which are instead +// defined in terms of a single shear value. This reduces the degrees of +// freedom enough to make the distortion more realistic than it would otherwise +// be if all 8 coordinates could move independently. +// One additional factor is used for the color of the pixels that don't exist +// in the source image. +// Name for each of the randomizing factors. +enum FactorNames { + FN_INCOLOR, + FN_Y0, + FN_Y1, + FN_Y2, + FN_Y3, + FN_X0, + FN_X1, + FN_SHEAR, + // x2 = x1 - shear + // x3 = x0 + shear + FN_NUM_FACTORS +}; + // Rotation is +/- kRotationRange radians. const float kRotationRange = 0.02f; // Number of grey levels to shift by for each exposure step. @@ -144,4 +170,141 @@ Pix* DegradeImage(Pix* input, int exposure, TRand* randomizer, return input; } +// Creates and returns a Pix distorted by various means according to the bool +// flags. If boxes is not NULL, the boxes are resized/positioned according to +// any spatial distortion and also by the integer reduction factor box_scale +// so they will match what the network will output. +// Returns NULL on error. The returned Pix must be pixDestroyed. +Pix* PrepareDistortedPix(const Pix* pix, bool perspective, bool invert, + bool white_noise, bool smooth_noise, bool blur, + int box_reduction, TRand* randomizer, + GenericVector* boxes) { + Pix* distorted = pixCopy(NULL, const_cast(pix)); + // Things to do to synthetic training data. + if (invert && randomizer->SignedRand(1.0) < 0) + pixInvert(distorted, distorted); + if ((white_noise || smooth_noise) && randomizer->SignedRand(1.0) > 0.0) { + // TODO(rays) Cook noise in a more thread-safe manner than rand(). + // Attempt to make the sequences reproducible. + srand(randomizer->IntRand()); + Pix* pixn = pixAddGaussianNoise(distorted, 8.0); + pixDestroy(&distorted); + if (smooth_noise) { + distorted = pixBlockconv(pixn, 1, 1); + pixDestroy(&pixn); + } else { + distorted = pixn; + } + } + if (blur && randomizer->SignedRand(1.0) > 0.0) { + Pix* blurred = pixBlockconv(distorted, 1, 1); + pixDestroy(&distorted); + distorted = blurred; + } + if (perspective) + GeneratePerspectiveDistortion(0, 0, randomizer, &distorted, boxes); + if (boxes != NULL) { + for (int b = 0; b < boxes->size(); ++b) { + (*boxes)[b].scale(1.0f / box_reduction); + if ((*boxes)[b].width() <= 0) + (*boxes)[b].set_right((*boxes)[b].left() + 1); + } + } + return distorted; +} + +// Distorts anything that has a non-null pointer with the same pseudo-random +// perspective distortion. Width and height only need to be set if there +// is no pix. If there is a pix, then they will be taken from there. +void GeneratePerspectiveDistortion(int width, int height, TRand* randomizer, + Pix** pix, GenericVector* boxes) { + if (pix != NULL && *pix != NULL) { + width = pixGetWidth(*pix); + height = pixGetHeight(*pix); + } + float* im_coeffs = NULL; + float* box_coeffs = NULL; + l_int32 incolor = + ProjectiveCoeffs(width, height, randomizer, &im_coeffs, &box_coeffs); + if (pix != NULL && *pix != NULL) { + // Transform the image. + Pix* transformed = pixProjective(*pix, im_coeffs, incolor); + if (transformed == NULL) { + tprintf("Projective transformation failed!!\n"); + return; + } + pixDestroy(pix); + *pix = transformed; + } + if (boxes != NULL) { + // Transform the boxes. + for (int b = 0; b < boxes->size(); ++b) { + int x1, y1, x2, y2; + const TBOX& box = (*boxes)[b]; + projectiveXformSampledPt(box_coeffs, box.left(), height - box.top(), &x1, + &y1); + projectiveXformSampledPt(box_coeffs, box.right(), height - box.bottom(), + &x2, &y2); + TBOX new_box1(x1, height - y2, x2, height - y1); + projectiveXformSampledPt(box_coeffs, box.left(), height - box.bottom(), + &x1, &y1); + projectiveXformSampledPt(box_coeffs, box.right(), height - box.top(), &x2, + &y2); + TBOX new_box2(x1, height - y1, x2, height - y2); + (*boxes)[b] = new_box1.bounding_union(new_box2); + } + } + free(im_coeffs); + free(box_coeffs); +} + +// Computes the coefficients of a randomized projective transformation. +// The image transform requires backward transformation coefficient, and the +// box transform the forward coefficients. +// Returns the incolor arg to pixProjective. +int ProjectiveCoeffs(int width, int height, TRand* randomizer, + float** im_coeffs, float** box_coeffs) { + // Setup "from" points. + Pta* src_pts = ptaCreate(4); + ptaAddPt(src_pts, 0.0f, 0.0f); + ptaAddPt(src_pts, width, 0.0f); + ptaAddPt(src_pts, width, height); + ptaAddPt(src_pts, 0.0f, height); + // Extract factors from pseudo-random sequence. + float factors[FN_NUM_FACTORS]; + float shear = 0.0f; // Shear is signed. + for (int i = 0; i < FN_NUM_FACTORS; ++i) { + // Everything is squared to make wild values rarer. + if (i == FN_SHEAR) { + // Shear is signed. + shear = randomizer->SignedRand(0.5 / 3.0); + shear = shear >= 0.0 ? shear * shear : -shear * shear; + // Keep the sheared points within the original rectangle. + if (shear < -factors[FN_X0]) shear = -factors[FN_X0]; + if (shear > factors[FN_X1]) shear = factors[FN_X1]; + factors[i] = shear; + } else if (i != FN_INCOLOR) { + factors[i] = fabs(randomizer->SignedRand(1.0)); + if (i <= FN_Y3) + factors[i] *= 5.0 / 8.0; + else + factors[i] *= 0.5; + factors[i] *= factors[i]; + } + } + // Setup "to" points. + Pta* dest_pts = ptaCreate(4); + ptaAddPt(dest_pts, factors[FN_X0] * width, factors[FN_Y0] * height); + ptaAddPt(dest_pts, (1.0f - factors[FN_X1]) * width, factors[FN_Y1] * height); + ptaAddPt(dest_pts, (1.0f - factors[FN_X1] + shear) * width, + (1 - factors[FN_Y2]) * height); + ptaAddPt(dest_pts, (factors[FN_X0] + shear) * width, + (1 - factors[FN_Y3]) * height); + getProjectiveXformCoeffs(dest_pts, src_pts, im_coeffs); + getProjectiveXformCoeffs(src_pts, dest_pts, box_coeffs); + ptaDestroy(&src_pts); + ptaDestroy(&dest_pts); + return factors[FN_INCOLOR] > 0.5f ? L_BRING_IN_WHITE : L_BRING_IN_BLACK; +} + } // namespace tesseract diff --git a/training/degradeimage.h b/training/degradeimage.h index 2add6282f8..a7af9565ff 100644 --- a/training/degradeimage.h +++ b/training/degradeimage.h @@ -20,12 +20,13 @@ #ifndef TESSERACT_TRAINING_DEGRADEIMAGE_H_ #define TESSERACT_TRAINING_DEGRADEIMAGE_H_ -struct Pix; +#include "allheaders.h" +#include "genericvector.h" +#include "helpers.h" // For TRand. +#include "rect.h" namespace tesseract { -class TRand; - // Degrade the pix as if by a print/copy/scan cycle with exposure > 0 // corresponding to darkening on the copier and <0 lighter and 0 not copied. // If rotation is not NULL, the clockwise rotation in radians is saved there. @@ -34,6 +35,27 @@ class TRand; struct Pix* DegradeImage(struct Pix* input, int exposure, TRand* randomizer, float* rotation); +// Creates and returns a Pix distorted by various means according to the bool +// flags. If boxes is not NULL, the boxes are resized/positioned according to +// any spatial distortion and also by the integer reduction factor box_scale +// so they will match what the network will output. +// Returns NULL on error. The returned Pix must be pixDestroyed. +Pix* PrepareDistortedPix(const Pix* pix, bool perspective, bool invert, + bool white_noise, bool smooth_noise, bool blur, + int box_reduction, TRand* randomizer, + GenericVector* boxes); +// Distorts anything that has a non-null pointer with the same pseudo-random +// perspective distortion. Width and height only need to be set if there +// is no pix. If there is a pix, then they will be taken from there. +void GeneratePerspectiveDistortion(int width, int height, TRand* randomizer, + Pix** pix, GenericVector* boxes); +// Computes the coefficients of a randomized projective transformation. +// The image transform requires backward transformation coefficient, and the +// box transform the forward coefficients. +// Returns the incolor arg to pixProjective. +int ProjectiveCoeffs(int width, int height, TRand* randomizer, + float** im_coeffs, float** box_coeffs); + } // namespace tesseract #endif // TESSERACT_TRAINING_DEGRADEIMAGE_H_ diff --git a/training/fileio.cpp b/training/fileio.cpp index e3e43bd023..bb1f4afcef 100644 --- a/training/fileio.cpp +++ b/training/fileio.cpp @@ -81,8 +81,9 @@ bool File::ReadFileToString(const string& filename, string* out) { } string File::JoinPath(const string& prefix, const string& suffix) { - return (!prefix.size() || prefix[prefix.size() - 1] == '/') ? - prefix + suffix : prefix + "/" + suffix; + return (prefix.empty() || prefix[prefix.size() - 1] == '/') + ? prefix + suffix + : prefix + "/" + suffix; } bool File::Delete(const char* pathname) { diff --git a/training/language-specific.sh b/training/language-specific.sh index a62f1e3cf3..b3c38cf876 100755 --- a/training/language-specific.sh +++ b/training/language-specific.sh @@ -1,4 +1,14 @@ # +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# http://www.apache.org/licenses/LICENSE-2.0 +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# # Set some language specific variables. Works in conjunction with # tesstrain.sh # @@ -868,6 +878,9 @@ set_lang_specific_parameters() { AMBIGS_FILTER_DENOMINATOR="100000" LEADING="32" MEAN_COUNT="40" # Default for latin script. + # Language to mix with the language for maximum accuracy. Defaults to eng. + # If no language is good, set to the base language. + MIX_LANG="eng" case ${lang} in # Latin languages. @@ -959,11 +972,13 @@ set_lang_specific_parameters() { WORD_DAWG_SIZE=1000000 test -z "$FONTS" && FONTS=( "${EARLY_LATIN_FONTS[@]}" );; - # Cyrillic script-based languages. + # Cyrillic script-based languages. It is bad to mix Latin with Cyrillic. rus ) test -z "$FONTS" && FONTS=( "${RUSSIAN_FONTS[@]}" ) + MIX_LANG="rus" NUMBER_DAWG_FACTOR=0.05 WORD_DAWG_SIZE=1000000 ;; aze_cyrl | bel | bul | kaz | mkd | srp | tgk | ukr | uzb_cyrl ) + MIX_LANG="${lang}" test -z "$FONTS" && FONTS=( "${RUSSIAN_FONTS[@]}" ) ;; # Special code for performing Cyrillic language-id that is trained on diff --git a/training/ligature_table.h b/training/ligature_table.h index ecae7943dd..83e7dc3c4d 100644 --- a/training/ligature_table.h +++ b/training/ligature_table.h @@ -32,7 +32,7 @@ namespace tesseract { class PangoFontInfo; // defined in pango_font_info.h // Map to substitute strings for ligatures. -typedef hash_map LigHash; +typedef TessHashMap LigHash; class LigatureTable { public: diff --git a/training/mftraining.cpp b/training/mftraining.cpp index 60314a1cdf..9e2e250927 100644 --- a/training/mftraining.cpp +++ b/training/mftraining.cpp @@ -64,9 +64,6 @@ #include "tprintf.h" #include "unicity_table.h" -using tesseract::Classify; -using tesseract::FontInfo; -using tesseract::FontSpacingInfo; using tesseract::IndexMapBiDi; using tesseract::MasterTrainer; using tesseract::Shape; @@ -305,6 +302,9 @@ int main (int argc, char **argv) { *shape_table, float_classes, inttemp_file.string(), pffmtable_file.string()); + for (int c = 0; c < unicharset->size(); ++c) { + FreeClassFields(&float_classes[c]); + } delete [] float_classes; FreeLabeledClassList(mf_classes); delete trainer; diff --git a/training/normstrngs.cpp b/training/normstrngs.cpp index acffeee13d..e7cac21f4b 100644 --- a/training/normstrngs.cpp +++ b/training/normstrngs.cpp @@ -113,12 +113,12 @@ bool is_double_quote(const char32 ch) { return false; } -STRING NormalizeUTF8String(const char* str8) { +STRING NormalizeUTF8String(bool decompose, const char* str8) { GenericVector str32, out_str32, norm_str; UTF8ToUTF32(str8, &str32); for (int i = 0; i < str32.length(); ++i) { norm_str.clear(); - NormalizeChar32(str32[i], &norm_str); + NormalizeChar32(str32[i], decompose, &norm_str); for (int j = 0; j < norm_str.length(); ++j) { out_str32.push_back(norm_str[j]); } @@ -128,10 +128,10 @@ STRING NormalizeUTF8String(const char* str8) { return out_str8; } -void NormalizeChar32(char32 ch, GenericVector* str) { +void NormalizeChar32(char32 ch, bool decompose, GenericVector* str) { IcuErrorCode error_code; const icu::Normalizer2* nfkc = icu::Normalizer2::getInstance( - NULL, "nfkc", UNORM2_COMPOSE, error_code); + NULL, "nfkc", decompose ? UNORM2_DECOMPOSE : UNORM2_COMPOSE, error_code); error_code.assertSuccess(); error_code.reset(); diff --git a/training/normstrngs.h b/training/normstrngs.h index 71e7b8da08..6fca3193ab 100644 --- a/training/normstrngs.h +++ b/training/normstrngs.h @@ -39,11 +39,16 @@ void UTF32ToUTF8(const GenericVector& str32, STRING* utf8_str); // assumption of this function is that the input is already as fully composed // as it can be, but may require some compatibility normalizations or just // OCR evaluation related normalizations. -void NormalizeChar32(char32 ch, GenericVector* str); +void NormalizeChar32(char32 ch, bool decompose, GenericVector* str); // Normalize a UTF8 string. Same as above, but for UTF8-encoded strings, that // can contain multiple UTF32 code points. -STRING NormalizeUTF8String(const char* str8); +STRING NormalizeUTF8String(bool decompose, const char* str8); +// Default behavior is to compose, until it is proven that decomposed benefits +// at least one language. +inline STRING NormalizeUTF8String(const char* str8) { + return NormalizeUTF8String(false, str8); +} // Apply just the OCR-specific normalizations and return the normalized char. char32 OCRNormalize(char32 ch); diff --git a/training/pango_font_info.cpp b/training/pango_font_info.cpp index c2b508c02d..41e352eae4 100644 --- a/training/pango_font_info.cpp +++ b/training/pango_font_info.cpp @@ -60,15 +60,6 @@ STRING_PARAM_FLAG(fontconfig_tmpdir, "/tmp", "Overrides fontconfig default temporary dir"); -BOOL_PARAM_FLAG(fontconfig_refresh_cache, false, - "Does a one-time deletion of cache files from the " - "fontconfig_tmpdir before initializing fontconfig."); -BOOL_PARAM_FLAG(fontconfig_refresh_config_file, true, - "Does a one-time reset of the fontconfig config file to point" - " to fonts_dir before initializing fontconfig. Set to true" - " if fontconfig_refresh_cache is true. Set it to false to use" - " multiple instances in separate processes without having to" - " rescan the fonts_dir, using a previously setup font cache"); #ifndef USE_STD_NAMESPACE #include "ocr/trainingdata/typesetting/legacy_fonts.h" @@ -91,7 +82,8 @@ namespace tesseract { // in pixels. const int kDefaultResolution = 300; -bool PangoFontInfo::fontconfig_initialized_ = false; +string PangoFontInfo::fonts_dir_; +string PangoFontInfo::cache_dir_; PangoFontInfo::PangoFontInfo() : desc_(NULL), resolution_(kDefaultResolution) { Clear(); @@ -119,6 +111,8 @@ void PangoFontInfo::Clear() { } } +PangoFontInfo::~PangoFontInfo() { pango_font_description_free(desc_); } + string PangoFontInfo::DescriptionName() const { if (!desc_) return ""; char* desc_str = pango_font_description_to_string(desc_); @@ -127,59 +121,63 @@ string PangoFontInfo::DescriptionName() const { return desc_name; } -// Initializes Fontconfig for use by writing a fake fonts.conf file into the -// FLAGS_fontconfigs_tmpdir directory, that points to the supplied -// fonts_dir, and then overrides the FONTCONFIG_PATH environment variable -// to point to this fonts.conf file. If force_clear, the cache is refreshed -// even if it has already been initialized. -void PangoFontInfo::InitFontConfig(bool force_clear, const string& fonts_dir) { - if ((fontconfig_initialized_ && !force_clear) || fonts_dir.empty()) { - fontconfig_initialized_ = true; - return; - } - if (FLAGS_fontconfig_refresh_cache || force_clear) { - File::DeleteMatchingFiles(File::JoinPath( - FLAGS_fontconfig_tmpdir.c_str(), "*cache-?").c_str()); - } - if (FLAGS_fontconfig_refresh_config_file || FLAGS_fontconfig_refresh_cache || - force_clear) { - const int MAX_FONTCONF_FILESIZE = 1024; - char fonts_conf_template[MAX_FONTCONF_FILESIZE]; - snprintf(fonts_conf_template, MAX_FONTCONF_FILESIZE, - "\n" - "\n" - "\n" - "%s\n" - "%s\n" - "\n" - "", fonts_dir.c_str(), - FLAGS_fontconfig_tmpdir.c_str()); - string fonts_conf_file = File::JoinPath(FLAGS_fontconfig_tmpdir.c_str(), - "fonts.conf"); - File::WriteStringToFileOrDie(fonts_conf_template, fonts_conf_file); +// If not already initialized, initializes FontConfig by setting its +// environment variable and creating a fonts.conf file that points to the +// FLAGS_fonts_dir and the cache to FLAGS_fontconfig_tmpdir. +/* static */ +void PangoFontInfo::SoftInitFontConfig() { + if (fonts_dir_.empty()) { + HardInitFontConfig(FLAGS_fonts_dir.c_str(), + FLAGS_fontconfig_tmpdir.c_str()); } +} + +// Re-initializes font config, whether or not already initialized. +// If already initialized, any existing cache is deleted, just to be sure. +/* static */ +void PangoFontInfo::HardInitFontConfig(const string& fonts_dir, + const string& cache_dir) { + if (!cache_dir_.empty()) { + File::DeleteMatchingFiles( + File::JoinPath(cache_dir_.c_str(), "*cache-?").c_str()); + } + const int MAX_FONTCONF_FILESIZE = 1024; + char fonts_conf_template[MAX_FONTCONF_FILESIZE]; + cache_dir_ = cache_dir; + fonts_dir_ = fonts_dir; + snprintf(fonts_conf_template, MAX_FONTCONF_FILESIZE, + "\n" + "\n" + "\n" + "%s\n" + "%s\n" + "\n" + "", + fonts_dir.c_str(), cache_dir_.c_str()); + string fonts_conf_file = File::JoinPath(cache_dir_.c_str(), "fonts.conf"); + File::WriteStringToFileOrDie(fonts_conf_template, fonts_conf_file); #ifdef _WIN32 std::string env("FONTCONFIG_PATH="); - env.append(FLAGS_fontconfig_tmpdir.c_str()); + env.append(cache_dir_.c_str()); putenv(env.c_str()); putenv("LANG=en_US.utf8"); #else - setenv("FONTCONFIG_PATH", FLAGS_fontconfig_tmpdir.c_str(), true); + setenv("FONTCONFIG_PATH", cache_dir_.c_str(), true); // Fix the locale so that the reported font names are consistent. setenv("LANG", "en_US.utf8", true); #endif // _WIN32 - if (!fontconfig_initialized_ || force_clear) { - if (FcInitReinitialize() != FcTrue) { - tprintf("FcInitiReinitialize failed!!\n"); - } + + if (FcInitReinitialize() != FcTrue) { + tprintf("FcInitiReinitialize failed!!\n"); } - fontconfig_initialized_ = true; FontUtils::ReInit(); + // Clear Pango's font cache too. + pango_cairo_font_map_set_default(NULL); } static void ListFontFamilies(PangoFontFamily*** families, int* n_families) { - PangoFontInfo::InitFontConfig(false, FLAGS_fonts_dir.c_str()); + PangoFontInfo::SoftInitFontConfig(); PangoFontMap* font_map = pango_cairo_font_map_get_default(); DISABLE_HEAP_LEAK_CHECK; pango_font_map_list_families(font_map, families, n_families); @@ -253,7 +251,7 @@ bool PangoFontInfo::ParseFontDescriptionName(const string& name) { // in the font map. Note that if the font is wholly missing, this could // correspond to a completely different font family and face. PangoFont* PangoFontInfo::ToPangoFont() const { - InitFontConfig(false, FLAGS_fonts_dir.c_str()); + SoftInitFontConfig(); PangoFontMap* font_map = pango_cairo_font_map_get_default(); PangoContext* context = pango_context_new(); pango_cairo_context_set_resolution(context, resolution_); @@ -437,10 +435,15 @@ bool PangoFontInfo::CanRenderString(const char* utf8_word, int len, PangoGlyph dotted_circle_glyph; PangoFont* font = run->item->analysis.font; - PangoGlyphString * glyphs = pango_glyph_string_new(); +#ifdef _WIN32 // Fixme! Leaks memory and breaks unittests. + PangoGlyphString* glyphs = pango_glyph_string_new(); char s[] = "\xc2\xa7"; pango_shape(s, sizeof(s), &(run->item->analysis), glyphs); dotted_circle_glyph = glyphs->glyphs[0].glyph; +#else + dotted_circle_glyph = pango_fc_font_get_glyph( + reinterpret_cast(font), kDottedCircleGlyph); +#endif if (TLOG_IS_ON(2)) { PangoFontDescription* desc = pango_font_describe(font); @@ -519,22 +522,21 @@ vector FontUtils::available_fonts_; // cache list bool FontUtils::IsAvailableFont(const char* input_query_desc, string* best_match) { string query_desc(input_query_desc); - if (PANGO_VERSION <= 12005) { - // Strip commas and any ' Medium' substring in the name. - query_desc.erase(std::remove(query_desc.begin(), query_desc.end(), ','), - query_desc.end()); - const string kMediumStr = " Medium"; - std::size_t found = query_desc.find(kMediumStr); - if (found != std::string::npos) { - query_desc.erase(found, kMediumStr.length()); - } +#if (PANGO_VERSION <= 12005) + // Strip commas and any ' Medium' substring in the name. + query_desc.erase(std::remove(query_desc.begin(), query_desc.end(), ','), + query_desc.end()); + const string kMediumStr = " Medium"; + std::size_t found = query_desc.find(kMediumStr); + if (found != std::string::npos) { + query_desc.erase(found, kMediumStr.length()); } - +#endif PangoFontDescription *desc = pango_font_description_from_string( query_desc.c_str()); PangoFont* selected_font = NULL; { - PangoFontInfo::InitFontConfig(false, FLAGS_fonts_dir.c_str()); + PangoFontInfo::SoftInitFontConfig(); PangoFontMap* font_map = pango_cairo_font_map_get_default(); PangoContext* context = pango_context_new(); pango_context_set_font_map(context, font_map); @@ -589,7 +591,7 @@ static bool ShouldIgnoreFontFamilyName(const char* query) { // Outputs description names of available fonts. /* static */ const vector& FontUtils::ListAvailableFonts() { - if (available_fonts_.size()) { + if (!available_fonts_.empty()) { return available_fonts_; } #ifndef USE_STD_NAMESPACE @@ -686,9 +688,8 @@ void FontUtils::GetAllRenderableCharacters(const vector& fonts, // Utilities written to be backward compatible with StringRender /* static */ -int FontUtils::FontScore(const unordered_map& ch_map, - const string& fontname, - int* raw_score, +int FontUtils::FontScore(const TessHashMap& ch_map, + const string& fontname, int* raw_score, vector* ch_flags) { PangoFontInfo font_info; if (!font_info.ParseFontDescriptionName(fontname)) { @@ -703,7 +704,7 @@ int FontUtils::FontScore(const unordered_map& ch_map, } *raw_score = 0; int ok_chars = 0; - for (unordered_map::const_iterator it = ch_map.begin(); + for (TessHashMap::const_iterator it = ch_map.begin(); it != ch_map.end(); ++it) { bool covered = (IsWhitespace(it->first) || (pango_coverage_get(coverage, it->first) @@ -721,7 +722,7 @@ int FontUtils::FontScore(const unordered_map& ch_map, /* static */ -string FontUtils::BestFonts(const unordered_map& ch_map, +string FontUtils::BestFonts(const TessHashMap& ch_map, vector > >* fonts) { const double kMinOKFraction = 0.99; // Weighted fraction of characters that must be renderable in a font to make diff --git a/training/pango_font_info.h b/training/pango_font_info.h index 421139a163..5c04bf0a9c 100644 --- a/training/pango_font_info.h +++ b/training/pango_font_info.h @@ -24,10 +24,16 @@ #include #include +#include "commandlineflags.h" #include "hashfn.h" #include "host.h" -#include "util.h" #include "pango/pango-font.h" +#include "pango/pango.h" +#include "pango/pangocairo.h" +#include "util.h" + +DECLARE_STRING_PARAM_FLAG(fonts_dir); +DECLARE_STRING_PARAM_FLAG(fontconfig_tmpdir); typedef signed int char32; @@ -44,6 +50,7 @@ class PangoFontInfo { DECORATIVE, }; PangoFontInfo(); + ~PangoFontInfo(); // Initialize from parsing a font description name, defined as a string of the // format: // "FamilyName [FaceName] [PointSize]" @@ -73,7 +80,7 @@ class PangoFontInfo { // If true, returns individual graphemes. Any whitespace characters in the // original string are also included in the list. bool CanRenderString(const char* utf8_word, int len, - vector* graphemes) const; + std::vector* graphemes) const; bool CanRenderString(const char* utf8_word, int len) const; // Retrieves the x_bearing and x_advance for the given utf8 character in the @@ -83,25 +90,29 @@ class PangoFontInfo { bool GetSpacingProperties(const string& utf8_char, int* x_bearing, int* x_advance) const; - // Initializes FontConfig by setting its environment variable and creating - // a fonts.conf file that points to the given fonts_dir. Once initialized, - // it is not re-initialized unless force_clear is true. - static void InitFontConfig(bool force_clear, const string& fonts_dir); + // If not already initialized, initializes FontConfig by setting its + // environment variable and creating a fonts.conf file that points to the + // FLAGS_fonts_dir and the cache to FLAGS_fontconfig_tmpdir. + static void SoftInitFontConfig(); + // Re-initializes font config, whether or not already initialized. + // If already initialized, any existing cache is deleted, just to be sure. + static void HardInitFontConfig(const string& fonts_dir, + const string& cache_dir); // Accessors string DescriptionName() const; // Font Family name eg. "Arial" const string& family_name() const { return family_name_; } // Size in points (1/72"), rounded to the nearest integer. - int font_size() const { return font_size_; } - bool is_bold() const { return is_bold_; } - bool is_italic() const { return is_italic_; } - bool is_smallcaps() const { return is_smallcaps_; } - bool is_monospace() const { return is_monospace_; } - bool is_fraktur() const { return is_fraktur_; } + int font_size() const { return font_size_; } + bool is_bold() const { return is_bold_; } + bool is_italic() const { return is_italic_; } + bool is_smallcaps() const { return is_smallcaps_; } + bool is_monospace() const { return is_monospace_; } + bool is_fraktur() const { return is_fraktur_; } FontTypeEnum font_type() const { return font_type_; } - int resolution() const { return resolution_; } + int resolution() const { return resolution_; } void set_resolution(const int resolution) { resolution_ = resolution; } @@ -130,8 +141,14 @@ class PangoFontInfo { int resolution_; // Fontconfig operates through an environment variable, so it intrinsically // cannot be thread-friendly, but you can serialize multiple independent - // font configurations by calling InitFontConfig(true, path). - static bool fontconfig_initialized_; + // font configurations by calling HardInitFontConfig(fonts_dir, cache_dir). + // These hold the last initialized values set by HardInitFontConfig or + // the first call to SoftInitFontConfig. + // Directory to be scanned for font files. + static string fonts_dir_; + // Directory to store the cache of font information. (Can be the same as + // fonts_dir_) + static string cache_dir_; private: PangoFontInfo(const PangoFontInfo&); @@ -152,29 +169,29 @@ class FontUtils { // best_match is not NULL, the closest matching font is returned there. static bool IsAvailableFont(const char* font_desc, string* best_match); // Outputs description names of available fonts. - static const vector& ListAvailableFonts(); + static const std::vector& ListAvailableFonts(); // Picks font among available fonts that covers and can render the given word, // and returns the font description name and the decomposition of the word to // graphemes. Returns false if no suitable font was found. static bool SelectFont(const char* utf8_word, const int utf8_len, - string* font_name, vector* graphemes); + string* font_name, std::vector* graphemes); // Picks font among all_fonts that covers and can render the given word, // and returns the font description name and the decomposition of the word to // graphemes. Returns false if no suitable font was found. static bool SelectFont(const char* utf8_word, const int utf8_len, - const vector& all_fonts, - string* font_name, vector* graphemes); + const std::vector& all_fonts, + string* font_name, std::vector* graphemes); // Returns a bitmask where the value of true at index 'n' implies that unicode // value 'n' is renderable by at least one available font. - static void GetAllRenderableCharacters(vector* unichar_bitmap); + static void GetAllRenderableCharacters(std::vector* unichar_bitmap); // Variant of the above function that inspects only the provided font names. - static void GetAllRenderableCharacters(const vector& font_names, - vector* unichar_bitmap); + static void GetAllRenderableCharacters(const std::vector& font_names, + std::vector* unichar_bitmap); static void GetAllRenderableCharacters(const string& font_name, - vector* unichar_bitmap); + std::vector* unichar_bitmap); // NOTE: The following utilities were written to be backward compatible with // StringRender. @@ -185,23 +202,24 @@ class FontUtils { // In the flags vector, each flag is set according to whether the // corresponding character (in order of iterating ch_map) can be rendered. // The return string is a list of the acceptable fonts that were used. - static string BestFonts(const unordered_map& ch_map, - vector > >* font_flag); + static string BestFonts( + const TessHashMap& ch_map, + std::vector > >* font_flag); // FontScore returns the weighted renderability score of the given // hash map character table in the given font. The unweighted score // is also returned in raw_score. // The values in the bool vector ch_flags correspond to whether the // corresponding character (in order of iterating ch_map) can be rendered. - static int FontScore(const unordered_map& ch_map, + static int FontScore(const TessHashMap& ch_map, const string& fontname, int* raw_score, - vector* ch_flags); + std::vector* ch_flags); // PangoFontInfo is reinitialized, so clear the static list of fonts. static void ReInit(); private: - static vector available_fonts_; // cache list + static std::vector available_fonts_; // cache list }; } // namespace tesseract diff --git a/training/set_unicharset_properties.cpp b/training/set_unicharset_properties.cpp index 00844ecb56..691c6dcfa6 100644 --- a/training/set_unicharset_properties.cpp +++ b/training/set_unicharset_properties.cpp @@ -1,3 +1,13 @@ +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + // This program reads a unicharset file, puts the result in a UNICHARSET // object, fills it with properties about the unichars it contains and writes // the result back to a file. diff --git a/training/stringrenderer.cpp b/training/stringrenderer.cpp index fbff55fb11..07a0e2003d 100644 --- a/training/stringrenderer.cpp +++ b/training/stringrenderer.cpp @@ -52,7 +52,7 @@ static const int kDefaultOutputResolution = 300; // Word joiner (U+2060) inserted after letters in ngram mode, as per // recommendation in http://unicode.org/reports/tr14/ to avoid line-breaks at // hyphens and other non-alpha characters. -static const char* kWordJoinerUTF8 = "\xE2\x81\xA0"; //u8"\u2060"; +static const char* kWordJoinerUTF8 = "\xE2\x81\xA0"; // u8"\u2060"; static const char32 kWordJoiner = 0x2060; static bool IsCombiner(int ch) { @@ -108,6 +108,7 @@ StringRenderer::StringRenderer(const string& font_desc, int page_width, underline_start_prob_(0), underline_continuation_prob_(0), underline_style_(PANGO_UNDERLINE_SINGLE), + features_(NULL), drop_uncovered_chars_(true), strip_unrenderable_words_(false), add_ligatures_(false), @@ -120,7 +121,6 @@ StringRenderer::StringRenderer(const string& font_desc, int page_width, box_padding_(0), total_chars_(0), font_index_(0), - features_(NULL), last_offset_(0) { pen_color_[0] = 0.0; pen_color_[1] = 0.0; @@ -209,8 +209,7 @@ void StringRenderer::SetLayoutProperties() { #if (PANGO_VERSION_MAJOR == 1 && PANGO_VERSION_MINOR >= 38) if (add_ligatures_) { set_features("liga, clig, dlig, hlig"); - PangoAttribute* feature_attr = - pango_attr_font_features_new(features_); + PangoAttribute* feature_attr = pango_attr_font_features_new(features_); pango_attr_list_change(attr_list, feature_attr); } #endif @@ -245,7 +244,7 @@ void StringRenderer::SetWordUnderlineAttributes(const string& page_text) { int offset = 0; TRand rand; bool started_underline = false; - PangoAttribute* und_attr = nullptr; + PangoAttribute* und_attr = NULL; while (offset < page_text.length()) { offset += SpanUTF8Whitespace(text + offset); @@ -264,7 +263,7 @@ void StringRenderer::SetWordUnderlineAttributes(const string& page_text) { // previous word. pango_attr_list_insert(attr_list, und_attr); started_underline = false; - und_attr = nullptr; + und_attr = NULL; } } if (!started_underline && RandBool(underline_start_prob_, &rand)) { @@ -348,6 +347,11 @@ void StringRenderer::ClearBoxes() { boxaDestroy(&page_boxes_); } +string StringRenderer::GetBoxesStr() { + BoxChar::PrepareToWrite(&boxchars_); + return BoxChar::GetTesseractBoxStr(page_height_, boxchars_); +} + void StringRenderer::WriteAllBoxes(const string& filename) { BoxChar::PrepareToWrite(&boxchars_); BoxChar::WriteTesseractBoxFile(filename, page_height_, boxchars_); @@ -396,7 +400,7 @@ bool StringRenderer::GetClusterStrings(vector* cluster_text) { it != start_byte_to_text.end(); ++it) { cluster_text->push_back(it->second); } - return cluster_text->size(); + return !cluster_text->empty(); } // Merges an array of BoxChars into words based on the identification of @@ -496,7 +500,7 @@ void StringRenderer::ComputeClusterBoxes() { const int end_byte_index = cluster_start_to_end_index[start_byte_index]; string cluster_text = string(text + start_byte_index, end_byte_index - start_byte_index); - if (cluster_text.size() && cluster_text[0] == '\n') { + if (!cluster_text.empty() && cluster_text[0] == '\n') { tlog(2, "Skipping newlines at start of text.\n"); continue; } @@ -596,11 +600,12 @@ void StringRenderer::ComputeClusterBoxes() { all_boxes = boxaCreate(0); boxaAddBox(all_boxes, page_boxchars[i]->mutable_box(), L_CLONE); } - boxaGetExtent(all_boxes, NULL, NULL, &page_box); - boxaDestroy(&all_boxes); - if (page_boxes_ == NULL) - page_boxes_ = boxaCreate(0); - boxaAddBox(page_boxes_, page_box, L_INSERT); + if (all_boxes != NULL) { + boxaGetExtent(all_boxes, NULL, NULL, &page_box); + boxaDestroy(&all_boxes); + if (page_boxes_ == NULL) page_boxes_ = boxaCreate(0); + boxaAddBox(page_boxes_, page_box, L_INSERT); + } } diff --git a/training/stringrenderer.h b/training/stringrenderer.h index 942b7fddce..b4646f71b1 100644 --- a/training/stringrenderer.h +++ b/training/stringrenderer.h @@ -90,7 +90,7 @@ class StringRenderer { void set_underline_style(const PangoUnderline style) { underline_style_ = style; } - void set_features(const char *features) { + void set_features(const char* features) { free(features_); features_ = strdup(features); } @@ -130,16 +130,12 @@ class StringRenderer { const PangoFontInfo& font() const { return font_; } - int h_margin() const { - return h_margin_; - } - int v_margin() const { - return v_margin_; - } + int h_margin() const { return h_margin_; } + int v_margin() const { return v_margin_; } // Get the boxchars of all clusters rendered thus far (or since the last call // to ClearBoxes()). - const vector& GetBoxes() const; + const std::vector& GetBoxes() const; // Get the rendered page bounding boxes of all pages created thus far (or // since last call to ClearBoxes()). Boxa* GetPageBoxes() const; @@ -148,6 +144,9 @@ class StringRenderer { void RotatePageBoxes(float rotation); // Delete all boxes. void ClearBoxes(); + // Returns the boxes in a boxfile string. + string GetBoxesStr(); + // Writes the boxes to a boxfile. void WriteAllBoxes(const string& filename); // Removes space-delimited words from the string that are not renderable by // the current font and returns the count of such words. @@ -172,8 +171,8 @@ class StringRenderer { void SetWordUnderlineAttributes(const string& page_text); // Compute bounding boxes around grapheme clusters. void ComputeClusterBoxes(); - void CorrectBoxPositionsToLayout(vector* boxchars); - bool GetClusterStrings(vector* cluster_text); + void CorrectBoxPositionsToLayout(std::vector* boxchars); + bool GetClusterStrings(std::vector* cluster_text); int FindFirstPageBreakOffset(const char* text, int text_length); PangoFontInfo font_; @@ -189,7 +188,7 @@ class StringRenderer { double underline_start_prob_; double underline_continuation_prob_; PangoUnderline underline_style_; - char *features_; + char* features_; // Text filtering options bool drop_uncovered_chars_; bool strip_unrenderable_words_; @@ -205,13 +204,13 @@ class StringRenderer { int page_; // Boxes and associated text for all pages rendered with RenderToImage() since // the last call to ClearBoxes(). - vector boxchars_; + std::vector boxchars_; int box_padding_; // Bounding boxes for pages since the last call to ClearBoxes(). Boxa* page_boxes_; // Objects cached for subsequent calls to RenderAllFontsToImage() - hash_map char_map_; // Time-saving char histogram. + TessHashMap char_map_; // Time-saving char histogram. int total_chars_; // Number in the string to be rendered. int font_index_; // Index of next font to use in font list. int last_offset_; // Offset returned from last successful rendering diff --git a/training/tesstrain_utils.sh b/training/tesstrain_utils.sh index c45d00378d..906a20ac4f 100755 --- a/training/tesstrain_utils.sh +++ b/training/tesstrain_utils.sh @@ -90,8 +90,8 @@ parse_flags() { --) break;; --fontlist) - fn=0 - FONTS="" + fn=0 + FONTS="" while test $j -lt ${#ARGV[@]}; do test -z "${ARGV[$j]}" && break test `echo ${ARGV[$j]} | cut -c -2` = "--" && break @@ -199,7 +199,7 @@ generate_font_image() { local common_args="--fontconfig_tmpdir=${FONT_CONFIG_CACHE}" common_args+=" --fonts_dir=${FONTS_DIR} --strip_unrenderable_words" - common_args+=" --fontconfig_refresh_config_file=false --leading=${LEADING}" + common_args+=" --leading=${LEADING}" common_args+=" --char_spacing=${CHAR_SPACING} --exposure=${EXPOSURE}" common_args+=" --outputbase=${outbase}" diff --git a/training/text2image.cpp b/training/text2image.cpp index 406669dc10..946f6facda 100644 --- a/training/text2image.cpp +++ b/training/text2image.cpp @@ -251,6 +251,8 @@ void ExtractFontProperties(const string &utf8_text, // the input consists of the separated characters. NOTE(ranjith): As per // behdad@ this is not currently controllable at the level of the Pango // API. + // The most frequent of all is a single character "word" made by the CJK + // segmenter. // Safeguard against these cases here by just skipping the bigram. if (IsWhitespaceBox(boxes[b+1])) { continue; @@ -445,7 +447,7 @@ int main(int argc, char** argv) { string pango_name; if (!FontUtils::IsAvailableFont(FLAGS_font.c_str(), &pango_name)) { tprintf("Could not find font named %s.\n", FLAGS_font.c_str()); - if (!pango_name.empty()) { + if (!pango_name.empty()) { tprintf("Pango suggested font %s.\n", pango_name.c_str()); } tprintf("Please correct --font arg.\n"); @@ -523,7 +525,7 @@ int main(int argc, char** argv) { if (FLAGS_render_ngrams && !FLAGS_unicharset_file.empty() && !unicharset.load_from_file(FLAGS_unicharset_file.c_str())) { tprintf("Failed to load unicharset from file %s\n", - FLAGS_unicharset_file.c_str()); + FLAGS_unicharset_file.c_str()); exit(1); } @@ -604,7 +606,8 @@ int main(int argc, char** argv) { rotation = -1 * page_rotation[page_num]; } if (FLAGS_degrade_image) { - pix = DegradeImage(pix, FLAGS_exposure, &randomizer, FLAGS_rotate_image ? &rotation : NULL); + pix = DegradeImage(pix, FLAGS_exposure, &randomizer, + FLAGS_rotate_image ? &rotation : NULL); } render.RotatePageBoxes(rotation); diff --git a/training/unicharset_training_utils.cpp b/training/unicharset_training_utils.cpp index 10aaf0e6c3..efa3a22cd5 100644 --- a/training/unicharset_training_utils.cpp +++ b/training/unicharset_training_utils.cpp @@ -37,7 +37,8 @@ namespace tesseract { // Helper sets the character attribute properties and sets up the script table. // Does not set tops and bottoms. -void SetupBasicProperties(bool report_errors, UNICHARSET* unicharset) { +void SetupBasicProperties(bool report_errors, bool decompose, + UNICHARSET* unicharset) { for (int unichar_id = 0; unichar_id < unicharset->size(); ++unichar_id) { // Convert any custom ligatures. const char* unichar_str = unicharset->id_to_unichar(unichar_id); @@ -129,7 +130,7 @@ void SetupBasicProperties(bool report_errors, UNICHARSET* unicharset) { } // Record normalized version of this unichar. - STRING normed_str = tesseract::NormalizeUTF8String(unichar_str); + STRING normed_str = tesseract::NormalizeUTF8String(decompose, unichar_str); if (unichar_id != 0 && normed_str.length() > 0) { unicharset->set_normed(unichar_id, normed_str.c_str()); } else { @@ -158,7 +159,7 @@ void SetPropertiesForInputFile(const string& script_dir, // Set unichar properties tprintf("Setting unichar properties\n"); - SetupBasicProperties(true, &unicharset); + SetupBasicProperties(true, false, &unicharset); string xheights_str; for (int s = 0; s < unicharset.get_script_table_size(); ++s) { // Load the unicharset for the script if available. diff --git a/training/unicharset_training_utils.h b/training/unicharset_training_utils.h index ff2262875d..f03e12ace4 100644 --- a/training/unicharset_training_utils.h +++ b/training/unicharset_training_utils.h @@ -33,7 +33,13 @@ namespace tesseract { // Helper sets the character attribute properties and sets up the script table. // Does not set tops and bottoms. -void SetupBasicProperties(bool report_errors, UNICHARSET* unicharset); +void SetupBasicProperties(bool report_errors, bool decompose, + UNICHARSET* unicharset); +// Default behavior is to compose, until it is proven that decomposed benefits +// at least one language. +inline void SetupBasicProperties(bool report_errors, UNICHARSET* unicharset) { + SetupBasicProperties(report_errors, false, unicharset); +} // Helper to set the properties for an input unicharset file, writes to the // output file. If an appropriate script unicharset can be found in the diff --git a/viewer/scrollview.cpp b/viewer/scrollview.cpp index ac059d5469..d4eb6d46a5 100644 --- a/viewer/scrollview.cpp +++ b/viewer/scrollview.cpp @@ -37,9 +37,9 @@ #include "scrollview.h" #ifdef _MSC_VER -#pragma warning(disable:4786) // Don't give stupid warnings for stl -#pragma warning(disable:4018) // signed/unsigned warnings -#pragma warning(disable:4530) // exception warnings +#pragma warning(disable : 4786) // Don't give irrelevant warnings for stl +#pragma warning(disable : 4018) // signed/unsigned warnings +#pragma warning(disable : 4530) // exception warnings #endif const int kSvPort = 8461; diff --git a/viewer/scrollview.h b/viewer/scrollview.h index 12fac4e5db..cf1e182efb 100644 --- a/viewer/scrollview.h +++ b/viewer/scrollview.h @@ -89,7 +89,7 @@ class SVEventHandler { // Gets called by the SV Window. Does nothing on default, overwrite this // to implement the desired behaviour - virtual void Notify(const SVEvent* sve) { } + virtual void Notify(const SVEvent* sve) { (void)sve; } }; // The ScrollView class provides the expernal API to the scrollviewer process. @@ -327,7 +327,7 @@ class ScrollView { // be unique among menubar eventIDs. void MenuItem(const char* parent, const char* name, int cmdEvent); -// This adds a new checkbox entry, which might initially be flagged. + // This adds a new checkbox entry, which might initially be flagged. void MenuItem(const char* parent, const char* name, int cmdEvent, bool flagged); diff --git a/viewer/svpaint.cpp b/viewer/svpaint.cpp index 4d2f49d951..c267257bac 100644 --- a/viewer/svpaint.cpp +++ b/viewer/svpaint.cpp @@ -1,6 +1,15 @@ // Copyright 2007 Google Inc. All Rights Reserved. // // Author: Joern Wanke +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// http://www.apache.org/licenses/LICENSE-2.0 +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. // // Simple drawing program to illustrate ScrollView capabilities. // diff --git a/viewer/svutil.cpp b/viewer/svutil.cpp index 80e11019e5..34a22861eb 100644 --- a/viewer/svutil.cpp +++ b/viewer/svutil.cpp @@ -22,6 +22,7 @@ #include #ifdef _WIN32 +#include struct addrinfo { struct sockaddr* ai_addr; int ai_addrlen; @@ -31,13 +32,13 @@ struct addrinfo { }; #else #include +#include #include #include #include #include #include #include -#include #include #include #ifdef __linux__ @@ -56,10 +57,34 @@ struct addrinfo { #include "config_auto.h" #endif -#ifndef GRAPHICS_DISABLED - #include "svutil.h" +SVMutex::SVMutex() { +#ifdef _WIN32 + mutex_ = CreateMutex(0, FALSE, 0); +#else + pthread_mutex_init(&mutex_, NULL); +#endif +} + +void SVMutex::Lock() { +#ifdef _WIN32 + WaitForSingleObject(mutex_, INFINITE); +#else + pthread_mutex_lock(&mutex_); +#endif +} + +void SVMutex::Unlock() { +#ifdef _WIN32 + ReleaseMutex(mutex_); +#else + pthread_mutex_unlock(&mutex_); +#endif +} + +#ifndef GRAPHICS_DISABLED + const int kMaxMsgSize = 4096; // Signals a thread to exit. @@ -161,29 +186,6 @@ void SVSemaphore::Wait() { #endif } -SVMutex::SVMutex() { -#ifdef _WIN32 - mutex_ = CreateMutex(0, FALSE, 0); -#else - pthread_mutex_init(&mutex_, NULL); -#endif -} - -void SVMutex::Lock() { -#ifdef _WIN32 - WaitForSingleObject(mutex_, INFINITE); -#else - pthread_mutex_lock(&mutex_); -#endif -} - -void SVMutex::Unlock() { -#ifdef _WIN32 - ReleaseMutex(mutex_); -#else - pthread_mutex_unlock(&mutex_); -#endif -} // Create new thread. @@ -200,7 +202,10 @@ void SVSync::StartThread(void *(*func)(void*), void* arg) { &threadid); // returns the thread identifier #else pthread_t helper; - pthread_create(&helper, NULL, func, arg); + pthread_attr_t attr; + pthread_attr_init(&attr); + pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); + pthread_create(&helper, &attr, func, arg); #endif } @@ -214,7 +219,7 @@ void SVNetwork::Send(const char* msg) { // Send the whole buffer. void SVNetwork::Flush() { mutex_send_->Lock(); - while (msg_buffer_out_.size() > 0) { + while (!msg_buffer_out_.empty()) { int i = send(stream_, msg_buffer_out_.c_str(), msg_buffer_out_.length(), 0); msg_buffer_out_.erase(0, i); } @@ -302,7 +307,8 @@ static std::string ScrollViewCommand(std::string scrollview_path) { const char* cmd_template = "-Djava.library.path=%s -jar %s/ScrollView.jar"; #else - const char* cmd_template = "-c \"trap 'kill %%1' 0 1 2 ; java " + const char* cmd_template = + "-c \"trap 'kill %%1' 0 1 2 ; java " "-Xms1024m -Xmx2048m -jar %s/ScrollView.jar" " & wait\""; #endif diff --git a/viewer/svutil.h b/viewer/svutil.h index ccfce917fe..667c052083 100644 --- a/viewer/svutil.h +++ b/viewer/svutil.h @@ -26,6 +26,7 @@ #ifdef _WIN32 #ifndef __GNUC__ +#include "platform.h" #include #if defined(_MSC_VER) && _MSC_VER < 1900 #define snprintf _snprintf @@ -102,6 +103,17 @@ class SVMutex { #endif }; +// Auto-unlocking object that locks a mutex on construction and unlocks it +// on destruction. +class SVAutoLock { + public: + explicit SVAutoLock(SVMutex* mutex) : mutex_(mutex) { mutex->Lock(); } + ~SVAutoLock() { mutex_->Unlock(); } + + private: + SVMutex* mutex_; +}; + /// The SVNetwork class takes care of the remote connection for ScrollView /// This means setting up and maintaining a remote connection, sending and /// receiving messages and closing the connection. diff --git a/vs2010/include/tesseract_versionnumbers.props b/vs2010/include/tesseract_versionnumbers.props index e16989d0e7..77bb3e03ab 100644 --- a/vs2010/include/tesseract_versionnumbers.props +++ b/vs2010/include/tesseract_versionnumbers.props @@ -4,9 +4,9 @@ - 303 - 3,3,0,0 - 3.03 + 305 + 3,5,0,0 + 3.05 <_ProjectFileVersion>10.0.40219.1 diff --git a/vs2010/libtesseract/libtesseract.rc b/vs2010/libtesseract/libtesseract.rc index f72d17f5a2..a4c5fc4568 100644 --- a/vs2010/libtesseract/libtesseract.rc +++ b/vs2010/libtesseract/libtesseract.rc @@ -72,7 +72,7 @@ BEGIN VALUE "FileDescription", "Tesseract OCR library" VALUE "FileVersion", "3, 5, 0, 0" VALUE "InternalName", "libtesseract" - VALUE "LegalCopyright", "Copyright (C) 2015 Google, Inc. Licensed under the Apache License, Version 2.0" + VALUE "LegalCopyright", "Copyright (C) 2016 Google, Inc. Licensed under the Apache License, Version 2.0" VALUE "OriginalFilename", "libtesseract" VALUE "ProductName", "Tesseract OCR Library" VALUE "ProductVersion", "3, 5, 0, 0" diff --git a/vs2010/libtesseract/libtesseract.vcxproj b/vs2010/libtesseract/libtesseract.vcxproj index 9b734d5d52..10762fe92e 100644 --- a/vs2010/libtesseract/libtesseract.vcxproj +++ b/vs2010/libtesseract/libtesseract.vcxproj @@ -27,7 +27,7 @@ - libtesseract304 + libtesseract305 {D14F28C7-0CAE-4C37-B174-40FDEFBD4FE0} libtesseract @@ -448,7 +448,6 @@ copy "$(TargetDir)$(TargetName).lib" ..\..\..\lib - @@ -716,7 +715,6 @@ copy "$(TargetDir)$(TargetName).lib" ..\..\..\lib - @@ -867,4 +865,4 @@ copy "$(TargetDir)$(TargetName).lib" ..\..\..\lib - \ No newline at end of file + diff --git a/vs2010/libtesseract/libtesseract.vcxproj.filters b/vs2010/libtesseract/libtesseract.vcxproj.filters index 911f3f7b9f..aa18a673e2 100644 --- a/vs2010/libtesseract/libtesseract.vcxproj.filters +++ b/vs2010/libtesseract/libtesseract.vcxproj.filters @@ -361,9 +361,6 @@ Source Files - - Source Files - Source Files @@ -1170,9 +1167,6 @@ Header Files - - Header Files - Header Files @@ -1610,4 +1604,4 @@ Resource Files - \ No newline at end of file + diff --git a/vs2010/port/vcsversion.h b/vs2010/port/vcsversion.h index 6d5bed8074..d620733040 100644 --- a/vs2010/port/vcsversion.h +++ b/vs2010/port/vcsversion.h @@ -1,2 +1,2 @@ -#define GIT_REV "3.04.00" +#define GIT_REV "3.05.00dev" diff --git a/vs2010/tesseract/tesseract.rc b/vs2010/tesseract/tesseract.rc index f2fc5248dc..48ae7fc88b 100644 --- a/vs2010/tesseract/tesseract.rc +++ b/vs2010/tesseract/tesseract.rc @@ -72,7 +72,7 @@ BEGIN VALUE "FileDescription", "Tesseract command-line OCR engine" VALUE "FileVersion", "3,5,0,0" VALUE "InternalName", "tesseract" - VALUE "LegalCopyright", "Copyright (C) 2015 Google, Inc. Licensed under the Apache License, Version 2.0" + VALUE "LegalCopyright", "Copyright (C) 2016 Google, Inc. Licensed under the Apache License, Version 2.0" VALUE "OriginalFilename", "tesseract.exe" VALUE "ProductName", "Tesseract-OCR" VALUE "ProductVersion", "3.05.00dev" diff --git a/wordrec/associate.h b/wordrec/associate.h index 3d6fc44708..10b1e0b7d9 100644 --- a/wordrec/associate.h +++ b/wordrec/associate.h @@ -47,9 +47,7 @@ struct AssociateStats { gap_sum = 0; } - void Print() { - tprintf("AssociateStats: w(%g %d)\n", shape_cost, bad_shape); - } + void Print() { tprintf("AssociateStats: s(%g %d)\n", shape_cost, bad_shape); } float shape_cost; // cost of blob shape bool bad_shape; // true if the shape of the blob is unacceptable diff --git a/wordrec/chopper.cpp b/wordrec/chopper.cpp index 69a458bc2c..dfda3e9183 100644 --- a/wordrec/chopper.cpp +++ b/wordrec/chopper.cpp @@ -426,7 +426,7 @@ void Wordrec::chop_word_main(WERD_RES *word) { if (word->best_choice == NULL) { // SegSearch found no valid paths, so just use the leading diagonal. - word->FakeWordFromRatings(); + word->FakeWordFromRatings(TOP_CHOICE_PERM); } word->RebuildBestState(); // If we finished without a hyphen at the end of the word, let the next word @@ -568,9 +568,7 @@ int Wordrec::select_blob_to_split( for (x = 0; x < blob_choices.size(); ++x) { if (blob_choices[x] == NULL) { - if (fragments != NULL) { - delete[] fragments; - } + delete[] fragments; return x; } else { blob_choice = blob_choices[x]; @@ -614,9 +612,7 @@ int Wordrec::select_blob_to_split( } } } - if (fragments != NULL) { - delete[] fragments; - } + delete[] fragments; // TODO(daria): maybe a threshold of badness for // worst_near_fragment would be useful. return worst_index_near_fragment != -1 ? diff --git a/wordrec/language_model.cpp b/wordrec/language_model.cpp index f0e3be66f2..99710478ed 100644 --- a/wordrec/language_model.cpp +++ b/wordrec/language_model.cpp @@ -32,7 +32,7 @@ #include "params.h" #include "params_training_featdef.h" -#if defined(_MSC_VER) || defined(ANDROID) +#if (defined(_MSC_VER) && _MSC_VER < 1900) || defined(ANDROID) double log2(double n) { return log(n) / log(2.0); } @@ -988,7 +988,7 @@ float LanguageModel::ComputeNgramCost(const char *unichar, unichar, context_ptr, CertaintyScore(certainty)/denom, prob, ngram_and_classifier_cost); } - if (modified_context != NULL) delete[] modified_context; + delete[] modified_context; return ngram_and_classifier_cost; } diff --git a/wordrec/lm_state.h b/wordrec/lm_state.h index 623bbb5e7f..6229e9b350 100644 --- a/wordrec/lm_state.h +++ b/wordrec/lm_state.h @@ -48,8 +48,8 @@ typedef unsigned char LanguageModelFlagsType; /// Each ViterbiStateEntry contains information from various components of the /// language model: dawgs in which the path is found, character ngram model /// probability of the path, script/chartype/font consistency info, state for -/// language-specific heuristics (e.g. hyphenated and compound words, lower/upper -/// case preferences, etc). +/// language-specific heuristics (e.g. hyphenated and compound words, +/// lower/upper case preferences, etc). /// /// Each ViterbiStateEntry also contains the parent pointer, so that the path /// that it represents (WERD_CHOICE) can be constructed by following these @@ -165,13 +165,13 @@ struct ViterbiStateEntry : public ELIST_LINK { /// Various information about the characters on the path represented /// by this ViterbiStateEntry. - float ratings_sum; //< sum of ratings of character on the path - float min_certainty; //< minimum certainty on the path - int adapted; //< number of BLOB_CHOICES from adapted templates - int length; //< number of characters on the path + float ratings_sum; //< sum of ratings of character on the path + float min_certainty; //< minimum certainty on the path + int adapted; //< number of BLOB_CHOICES from adapted templates + int length; //< number of characters on the path float outline_length; //< length of the outline so far LMConsistencyInfo consistency_info; //< path consistency info - AssociateStats associate_stats; //< character widths/gaps/seams + AssociateStats associate_stats; //< character widths/gaps/seams /// Flags for marking the entry as a top choice path with /// the smallest rating or lower/upper case letters). diff --git a/wordrec/measure.h b/wordrec/measure.h index 9c73906853..894938e55a 100644 --- a/wordrec/measure.h +++ b/wordrec/measure.h @@ -60,10 +60,9 @@ typedef struct * Add one more sample to a measurement. **********************************************************************/ -#define ADD_SAMPLE(m,s) \ -(m.sum_of_samples += (float) (s), \ - m.sum_of_squares += (float) (s) * (float) (s), \ - ++m.num_samples) +#define ADD_SAMPLE(m, s) \ + (m.sum_of_samples += (float)(s), \ + m.sum_of_squares += (float)(s) * (float)(s), ++m.num_samples) /********************************************************************** * mean @@ -71,10 +70,8 @@ typedef struct * Return the mean value of the measurement. **********************************************************************/ -#define MEAN(m) \ -((m).num_samples ? \ - ((float) ((m).sum_of_samples / (m).num_samples)) : \ - 0) +#define MEAN(m) \ + ((m).num_samples ? ((float)((m).sum_of_samples / (m).num_samples)) : 0) /********************************************************************** * new_measurement @@ -83,10 +80,8 @@ typedef struct * samples. **********************************************************************/ -#define new_measurement(m) \ -((m).num_samples = 0, \ - (m).sum_of_samples = 0, \ - (m).sum_of_squares = 0) +#define new_measurement(m) \ + ((m).num_samples = 0, (m).sum_of_samples = 0, (m).sum_of_squares = 0) /********************************************************************** * number_of_samples @@ -112,13 +107,12 @@ typedef struct * Return the variance of the measurement. **********************************************************************/ -#define VARIANCE(m) \ -(((m).num_samples > 1) ? \ - ((float) \ - (((m).num_samples * (m).sum_of_squares - \ - (m).sum_of_samples * (m).sum_of_samples) / \ - (((m).num_samples - 1) * (m).num_samples))) : \ - 0) +#define VARIANCE(m) \ + (((m).num_samples > 1) \ + ? ((float)(((m).num_samples * (m).sum_of_squares - \ + (m).sum_of_samples * (m).sum_of_samples) / \ + (((m).num_samples - 1) * (m).num_samples))) \ + : 0) /********************************************************************** * print_summary @@ -126,10 +120,8 @@ typedef struct * Summarize a MEASUREMENT record. **********************************************************************/ -#define print_summary(string,measure) \ -cprintf ("\t%-20s \tn = %d, \tm = %4.2f, \ts = %4.2f\n ", \ - string, \ - number_of_samples (measure), \ - MEAN (measure), \ - standard_deviation (measure)) +#define print_summary(string, measure) \ + cprintf("\t%-20s \tn = %d, \tm = %4.2f, \ts = %4.2f\n ", string, \ + number_of_samples(measure), MEAN(measure), \ + standard_deviation(measure)) #endif diff --git a/wordrec/pieces.cpp b/wordrec/pieces.cpp index 04e340396e..f7b406d5bc 100644 --- a/wordrec/pieces.cpp +++ b/wordrec/pieces.cpp @@ -267,7 +267,6 @@ void Wordrec::merge_and_put_fragment_lists(inT16 row, inT16 column, delete [] choice_lists_it; } - /********************************************************************** * get_fragment_lists * diff --git a/wordrec/tface.cpp b/wordrec/tface.cpp index e21fcb8829..b1fc1779fb 100644 --- a/wordrec/tface.cpp +++ b/wordrec/tface.cpp @@ -49,7 +49,11 @@ void Wordrec::program_editup(const char *textbase, if (textbase != NULL) imagefile = textbase; InitFeatureDefs(&feature_defs_); InitAdaptiveClassifier(init_classifier); - if (init_dict) getDict().Load(Dict::GlobalDawgCache()); + if (init_dict) { + getDict().SetupForLoad(Dict::GlobalDawgCache()); + getDict().Load(tessdata_manager.GetDataFileName().string(), lang); + getDict().FinishLoad(); + } pass2_ok_split = chop_ok_split; }