Merge remote-tracking branch 'refs/remotes/origin/master'

Updating from master.

api/baseapi.cpp

@@ -139,7 +139,11 @@ TessBaseAPI::~TessBaseAPI() {
  * Returns the version identifier as a static string. Do not delete.
  */
 const char* TessBaseAPI::Version() {
+#if defined(DEBUG) && defined(GIT_REV)
+  return GIT_REV;
+#else
   return TESSERACT_VERSION_STR;
+#endif
 }
 
 /**

ccutil/platform.h

@@ -28,10 +28,12 @@
 #define ultoa _ultoa
 #endif  /* __GNUC__ */
 #define SIGNED
+#if defined(_MSC_VER)
 #define snprintf _snprintf
 #if (_MSC_VER <= 1400)
 #define vsnprintf _vsnprintf
-#endif /* _WIN32 */
+#endif /* (_MSC_VER <= 1400) */
+#endif /* defined(_MSC_VER) */
 #else
 #define __UNIX__
 #include <limits.h>

configure.ac

@@ -5,12 +5,10 @@
 # ----------------------------------------
 # Initialization
 # ----------------------------------------
-
 AC_PREREQ(2.50)
 AC_INIT([tesseract], [3.04], [http://code.google.com/p/tesseract-ocr/issues/list])
 CXXFLAGS=${CXXFLAGS:-""}
 AC_CONFIG_MACRO_DIR([m4])
-AC_REVISION([$Revision$])
 AC_CONFIG_AUX_DIR(config)
 AC_CONFIG_SRCDIR(api/tesseractmain.cpp)
 AC_PREFIX_DEFAULT(/usr/local)
@@ -21,6 +19,14 @@ AC_PREFIX_DEFAULT(/usr/local)
 PACKAGE_YEAR=2014
 PACKAGE_DATE="08/13"
 
+abs_top_srcdir=`AS_DIRNAME([$0])`
+gitrev="`git --git-dir=${abs_top_srcdir}/.git --work-tree=${abs_top_srcdir} describe --always --tags`"
+if test -n "${gitrev}" ; then
+	AC_REVISION("${gitrev}")
+	AC_DEFINE_UNQUOTED(GIT_REV,"${gitrev}", [Define to be the git revision])
+	echo "Using git revision: ${gitrev}"
+fi
+
 AC_DEFINE_UNQUOTED(PACKAGE_NAME,["${PACKAGE_NAME}"],[Name of package])
 AC_DEFINE_UNQUOTED(PACKAGE_VERSION,["${PACKAGE_VERSION}"],[Version number])
 AC_DEFINE_UNQUOTED(PACKAGE_YEAR,"$PACKAGE_YEAR",[Official year for this release])
@@ -85,7 +91,7 @@ case "${host_os}" in
         AM_CONDITIONAL(ADD_RT, false)
         AM_CONDITIONAL(T_WIN, true)
         AC_SUBST([AM_LDFLAGS], ['-Wl,-no-undefined -Wl,--as-needed'])
-        ;;			
+        ;;
     solaris*)
         LIBS="-lsocket -lnsl -lrt -lxnet"
         AM_CONDITIONAL(ADD_RT, true)

cube/tess_lang_mod_edge.h

@@ -177,7 +177,7 @@ class TessLangModEdge : public LangModEdge {
   // to quickly lookup exisiting edges to converge during search
   inline unsigned int Hash() const {
     return static_cast<unsigned int>(((start_edge_ | end_edge_) ^
-      ((reinterpret_cast<unsigned long int>(dawg_)))) ^
+      ((reinterpret_cast<uintptr_t>(dawg_)))) ^
       ((unsigned int)edge_mask_) ^
       class_id_);
   }

tesseract.pc.in

@@ -10,7 +10,7 @@ Name: @PACKAGE_NAME@
 Description: An OCR Engine that was developed at HP Labs between 1985 and 1995... and now at Google.
 URL: https://code.google.com/p/tesseract-ocr
 Version: @VERSION@
-# Requires.private: lept
+Requires.private: lept
 Libs: -L${libdir} -ltesseract
 Libs.private: -lpthread -llept @OPENCL_LIB@
 Cflags: -I${includedir}

training/fileio.cpp

@@ -76,11 +76,7 @@ bool File::ReadFileToString(const string& filename, string* out) {
     return false;
   InputBuffer in(stream);
   *out = "";
-  string temp;
-  while (in.ReadLine(&temp)) {
-    *out += temp;
-    *out += '\n';
-  }
+  in.Read(out);
   return in.CloseFile();
 }
 
@@ -156,32 +152,17 @@ InputBuffer::~InputBuffer() {
   }
 }
 
-bool InputBuffer::ReadLine(string* out) {
-  ASSERT_HOST(stream_ != NULL);
-  char* line = NULL;
-  int len = -1;
-#ifndef HAVE_GETLINE
-  char line_buf[BUFSIZ];
-  if ((line = fgets(line_buf, BUFSIZ, stream_)) != NULL) {
-    len = strlen(line);
-    if (line_buf[0] != '\0' && line_buf[len - 1] == '\n')
-        line_buf[len - 1] = '\0';
-  } else {
-    return false;
-  }
-  *out = string(line);
-#else
-  size_t line_size;
-  len = getline(&line, &line_size, stream_);
-  if (len < 0) {
-    return false;
+bool InputBuffer::Read(string *out) {
+  char buf[BUFSIZ+1];
+  int l;
+  while((l = fread(buf, 1, BUFSIZ, stream_)) > 0) {
+    if(ferror(stream_)) {
+      clearerr(stream_);
+      return false;
+    }
+    buf[l] = 0;
+    out->append(buf);
   }
-
-  if (len >= 1 && line[len - 1] == '\n')
-    line[len - 1] = '\0';
-  *out = string(line);
-  free(line);
-#endif  // HAVE_GETLINE
   return true;
 }
 

training/fileio.h

@@ -68,6 +68,11 @@ class InputBuffer {
   // Return false if an error occurs or at end-of-file, true otherwise.
   bool ReadLine(string* out);
 
+  // Read data until end-of-file.
+  // The data is stored in '*out'.
+  // Return false if an error occurs, true otherwise.
+  bool Read(string* out);
+
   // Close the FILE* used by InputBuffer.
   // Return false if an error occurs, true otherwise.
   bool CloseFile();