Update from Leptonica 1.74.1 to 1.74.2

[stweil]

The newer version contains fixes for the pixUnsharpMaskingGray*
functions which are relevant for Tesseract (used in ImageData::PreScale
which calls pixScale).

Signed-off-by: Stefan Weil [email protected]

[stweil]

See Dan's comment for issue #644.

[stweil]

Should we refuse builds with Leptonica < 1.74.2? Then additional changes for CMakeLists.txt and configure.ac would be needed. The newer version not only fixes problems with undefined memory for lstmtraining, but also for tesseract --oem 1:

With Leptonica 1.74.1:

$ valgrind bin/ndebug/x86_64-linux-gnu/api/tesseract --oem 1 ~/Bilder/ocr/hello_world.png /tmp/hello
==21183== Memcheck, a memory error detector
==21183== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==21183== Using Valgrind-3.12.0.SVN and LibVEX; rerun with -h for copyright info
==21183== Command: bin/ndebug/x86_64-linux-gnu/api/tesseract --oem 1 /home/stweil/Bilder/ocr/hello_world.png /tmp/hello
==21183== 
==21183== Conditional jump or move depends on uninitialised value(s)
==21183==    at 0x4ECF07A: nextOnPixelInRasterLow (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4ECF29F: nextOnPixelInRaster (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4ED03A9: pixConnCompBB (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8B8DC: ??? (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8C021: ??? (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8CABA: bmfCreate (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x1710B8: DebugPixa (debugpixa.h:15)
==21183==    by 0x1710B8: tesseract::Tesseract::Tesseract() (tesseractclass.cpp:623)
==21183==    by 0x11F99B: tesseract::TessBaseAPI::Init(char const*, int, char const*, tesseract::OcrEngineMode, char**, int, GenericVector<STRING> const*, GenericVector<STRING> const*, bool, bool (*)(STRING const&, GenericVector<char>*)) (baseapi.cpp:320)
==21183==    by 0x11FE25: tesseract::TessBaseAPI::Init(char const*, char const*, tesseract::OcrEngineMode, char**, int, GenericVector<STRING> const*, GenericVector<STRING> const*, bool) (baseapi.cpp:284)
==21183==    by 0x116273: main (tesseractmain.cpp:440)
==21183== 
==21183== Conditional jump or move depends on uninitialised value(s)
==21183==    at 0x4ECF12E: nextOnPixelInRasterLow (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4ECF29F: nextOnPixelInRaster (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4ED03A9: pixConnCompBB (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8B8DC: ??? (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8C021: ??? (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x4E8CABA: bmfCreate (in /usr/lib/x86_64-linux-gnu/liblept.so.5.0.1)
==21183==    by 0x1710B8: DebugPixa (debugpixa.h:15)
==21183==    by 0x1710B8: tesseract::Tesseract::Tesseract() (tesseractclass.cpp:623)
==21183==    by 0x11F99B: tesseract::TessBaseAPI::Init(char const*, int, char const*, tesseract::OcrEngineMode, char**, int, GenericVector<STRING> const*, GenericVector<STRING> const*, bool, bool (*)(STRING const&, GenericVector<char>*)) (baseapi.cpp:320)
==21183==    by 0x11FE25: tesseract::TessBaseAPI::Init(char const*, char const*, tesseract::OcrEngineMode, char**, int, GenericVector<STRING> const*, GenericVector<STRING> const*, bool) (baseapi.cpp:284)
==21183==    by 0x116273: main (tesseractmain.cpp:440)
==21183== 
Tesseract Open Source OCR Engine v4.00.00alpha with Leptonica
==21183== 
==21183== HEAP SUMMARY:
==21183==     in use at exit: 3,360 bytes in 7 blocks
==21183==   total heap usage: 653,804 allocs, 653,797 frees, 143,432,299 bytes allocated
==21183== 
==21183== LEAK SUMMARY:
==21183==    definitely lost: 0 bytes in 0 blocks
==21183==    indirectly lost: 0 bytes in 0 blocks
==21183==      possibly lost: 912 bytes in 3 blocks
==21183==    still reachable: 2,448 bytes in 4 blocks
==21183==         suppressed: 0 bytes in 0 blocks
==21183== Rerun with --leak-check=full to see details of leaked memory
==21183== 
==21183== For counts of detected and suppressed errors, rerun with: -v
==21183== Use --track-origins=yes to see where uninitialised values come from
==21183== ERROR SUMMARY: 169 errors from 2 contexts (suppressed: 0 from 0)

With Leptonica 1.74.2:

$ valgrind bin/ndebug/x86_64-linux-gnu/api/tesseract --oem 1 ~/Bilder/ocr/hello_world.png /tmp/hello2
==22040== Memcheck, a memory error detector
==22040== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==22040== Using Valgrind-3.12.0.SVN and LibVEX; rerun with -h for copyright info
==22040== Command: bin/ndebug/x86_64-linux-gnu/api/tesseract --oem 1 /home/stweil/Bilder/ocr/hello_world.png /tmp/hello2
==22040== 
Tesseract Open Source OCR Engine v4.00.00alpha with Leptonica
==22040== 
==22040== HEAP SUMMARY:
==22040==     in use at exit: 3,360 bytes in 7 blocks
==22040==   total heap usage: 653,807 allocs, 653,800 frees, 143,423,100 bytes allocated
==22040== 
==22040== LEAK SUMMARY:
==22040==    definitely lost: 0 bytes in 0 blocks
==22040==    indirectly lost: 0 bytes in 0 blocks
==22040==      possibly lost: 912 bytes in 3 blocks
==22040==    still reachable: 2,448 bytes in 4 blocks
==22040==         suppressed: 0 bytes in 0 blocks
==22040== Rerun with --leak-check=full to see details of leaked memory
==22040== 
==22040== For counts of detected and suppressed errors, rerun with: -v
==22040== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

[stweil]

Tesseract cmake builds are broken with Leptonica 1.74.2 because Leptonica now requires an installation: the Leptonica_INCLUDE_DIRS refer to /usr/local/include. Therefore Travis CI fails. @egorpugin, how should we fix that?

[egorpugin]

Someone broke my leptonica cmake scripts? :)

[egorpugin]

Should we refuse builds with Leptonica < 1.74.2?

We can do anything, because it's dev branch.
+1 from me