SharpMuPDF/mupdf/thirdparty/tesseract/ChangeLog

2024-11-10 - V5.5.0
* Set hOCR capabilities ocrp_dir and ocrp_lang unconditionally.
* Calculate row bounding box in single-word mode per (issue #4304).
* Reduce clock syscalls (#4303).
* Several small performance and other code fixes.
* Modernized code.
* Print time for tessedit_timing_debug in milliseconds.
* Print time for ErrorCounter::ComputeErrorRate in milliseconds.
* cmake: Correctly set the soversion based on SemVer properties.
* Do not export PDBs for static libraries (issue #4279).
* Several other small fixes and improvements for builds and CI.
* Modernize code for renderers and remove filename conversion for Windows (#4330).
* Add build rule for Windows installer.
* Support symbolic values for --oem and --psm options.
* Remove Tensorflow support.
* Add RISC-V V support (#4346).
* Remove broken GitHub action msys2-4.1.1.

2024-06-11 - V5.4.1
* Avoid FP overflow in NormEvidenceOf (fixes issue #4257) (#4259)
* Small build fixes and code improvements (#4262, #4263, #4266, #4267)

2024-06-06 - V5.4.0
* Small build fixes and code improvements
  (#4241, #4243, #4244, #4245, #4246, #4248, #4249, #4250, #4253)

2024-05-19 - V5.4.0-rc2
* Fix setup of datadir on installations with Conda (issue #4230) (#4240)
* Fix FP exception in Wordrec::angle_change (issue #4242) (#4243)

2024-05-12 - V5.4.0-rc1
* Build fixes, code refactoring and other smaller changes.
* Fix grey result of indexed PNG in pdfrenderer.
* Rename frk -> deu_latf (ISO 639-3, ISO 15924).
* Remove broken Dockerfile.
* Fixes for several issues reported by Coverity Scan.
* Remove unsupported OpenCL code and related API functions (#4220).
* Facilitate vectorization for generic build (#4223).
* Add PAGE XML renderer / export (#4214).
* Support training without lstmf files.
* Improve CCUtil::main_setup (fixes issue #4230 related to Coda).
* Allow for text angle/gradient to be retrieved (#4070).

2024-01-18 - V5.3.4
* Fixes for scrollview
* Fixes for autoconf, clang and sw builds
* Improve OCR for an image URL
  * Fail on curl download errors
  * New parameter curl_cookiefile
  * Set User-Agent: header field in HTTP request for curl downloads
* Output directory list from "combine_tessdata -d" to stdout
* Other small improvements for code and documentation.

2023-10-05 - V5.3.3
* Small code fixes and improvements to fix Coverity Scan issues.
* Disable -mfpu=neon for aarch64.
* Fix build without git clone in cloned directory (required for FreeBSD).
* Other build fixes for autotools, cmake and sw.
* Fix regression in layout detection which was introduced in release 5.0.0.
* Fix regression which prevented loading of submodels, introduced in release 5.0.0-rc2.
* Other small improvements for code and documentation.

2023-07-11 - V5.3.2
* Updates for snap package building.
* Support for Sgaw and W Pwo Karen languages in the Myanmar validator (#4065).
* Improve format of logging from lstmtraining.
* Use less digits in filenames of checkpoints written by lstmtraining.
* Replace deprecated sprintf.
* Remove unused code in function fix_rep_char.
* Avoid 32 bit overflow in multiplication (fixes 3 CodeQL CI alerts).
* Avoid conversions from std::string to char* to std::string.
* Abort with error message if OSD is requested with LSTM-only model.
* cmake: allow to disable tiff (-DDISABLE_TIFF=ON).
* cmake: provide info about disabled LibArchive and CURL.
* cmake: check if leptonica was build with tiff support.
* Remove old broken GitHub action vcpkg-4.1.1 (fixes issue #4078).
* Create config.yml.
* Fix typos.

2023-04-01 - V5.3.1
 * Bug fixes for some special scenarios:
   * Fix issue #4010.
   * textord: Catch empty rows in block iterator (fixes #4039).
   * Fix FP division by zero (issue #3995).
 * Improve documentation and log messages.
 * Build fixes and improvements (mainly for cmake).

2022-12-22 - V5.3.0
 * Minor updates for documentation and cmake builds.

2022-12-13 - V5.3.0-rc1
 * Fix the training tools for the legacy OCR engine (fix issue #3925).
 * PDF renderer: Ignore non-text blocks (fix issue #3957).
 * Remove colormap before thresholding (fix issue #3940).
 * Fix a number of performance issues reported by Coverity Scan.
 * Training tools: Replace call of exit function by return statement in main function.
 * Fix double free in function vigorous_noise_removal (fix issue #3876).
 * Create to_win if needed in Textord::make_spline_rows (fix issue #3875).
 * Bug fixes for ScrollView viewer:
   * Fix memory issues in ScrollView::MessageReceiver.
   * Catch potential nullptr in SVNetwork::SVNetwork.
   * Move svpaint.cpp from src/viewer to src/.
   * Add rule for svpaint executable in Autotools.
 * Bug fixes and improvements for build tools:
   * Fix AMD64 detection with autobuild on FreeBSD (fix issue #3964).
   * Fix tesseract.pc generated from CMake to match Autotools.
   * Detect availability of AVX512-VNNI.
   * configure.ac: fix build on aarch64_be.
 * Drop CI for old versions of macOS and Ubuntu.

2022-07-06 - V5.2.0
  * Improvements and fixes for continuous integration,
    autoconf and cmake builds.
  * Set /Os for some 32 bit MS compilers (fixes #3769).
  * Improve comments and other documentation.
  * Add initial support for Intel AVX512F.
  * Fix for very large PDF files on 32 bit hosts (fixes #3805).
  * Fix NEON detection on FreeBSD.
  * Fix regression with UZN files (fixes #3837).
  * Fix calling delete[] for memory allocated by malloc in C API.
  * Add an API function to init tesseract with traineddata from memory
    (fixes #3691).
  * Replace direct access to Leptonica internal data structures by
    function calls and support latest releases of Leptonica.
  * Replace std::regex by std::string functions (fixes issue #3830).
  * Use compiled-in TESSDATA_PREFIX also on Windows (fixes #3767).
  * Add new parameter 'invert_threshold', change the default threshold
    from 0.5 to 0.7 and mark parameter 'tessedit_do_invert' as deprecated.

2022-03-01 - V5.1.0
  * Handle image and line regions in output formats ALTO, hOCR and text.
  * New parameter curl_timeout for curl_easy_setop.
  * Build fixes and improvements.
  * Catch nullptr in PageIterator::Orientation to improve robustness.
  * Remove unused code.

2022-01-06 - V5.0.1
  * Add SPDX-License-Identifier to public include files.
  * Support redirections when running OCR on a URL.
  * Lots of fixes and improvements for cmake builds.
    Distributions should use the autoconf build.
  * Fix broken msys2 build with gcc 11.
  * Fix parameter certainty_scale (was duplicated).
  * Fix some compiler warnings and clean code.
  * Correctly detect amd64 and i386 on FreeBSD.
  * Add libarchive and libcurl in continuous integration actions.
  * Update submodule googletest to release v1.11.0.

2021-11-22 - V5.0.0
  * Faster training and recognition by default (float instead of
    double calculations)
  * More options for binarization
  * Improved support for ARM NEON
  * Modernized code
  * Removed proprietary data types like GenericVector and STRING
    from public API
  * pdf.ttf no longer needed, now integrated into the code
  * Faster flat build with automake
  * New options for combine_tessdata to show details of traineddata files
  * Improved training messages
  * Improved unit tests and fuzzing tests
  * Lots of bug fixes

2021-11-15 - V4.1.3
  * Fix build regression for autoconf build

2021-11-14 - V4.1.2
  * Add RowAttributes getter to PageIterator
  * Allow line images with larger width for training
  * Fix memory leaks
  * Improve build process
  * Don't output empty ALTO sourceImageInformation (issue #2700)
  * Extend URI support for Tesseract with libcurl
  * Abort LSTM training with integer model (fixes issue #1573)
  * Update documentation
  * Make automake builds less noisy by default
  * Don't use -march=native in automake builds

2019-12-26 - V4.1.1
  * Implemented sw build (cppan is depreciated)
  * Improved cmake build
  * Code cleanup and optimization
  * A lot of bug fixes...

2019-07-07 - V4.1.0
  * Added new renders Alto, LSTMBox, WordStrBox.
  * Added character boxes in hOCR output.
  * Added python training scripts (experimental) as alternative shell scripts.
  * Better support AVX / AVX2 / SSE.
  * Disable OpenMP support by default (see e.g. #1171, #1081).
  * Fix for bounding box problem.
  * Implemented support for whitelist/blacklist in LSTM engine.
  * Improved cmake configuration.
  * Code modernization and improvements.
  * A lot of bug fixes...

2018-10-29 - V4.0.0
  * Added new neural network system based on LSTMs, with major accuracy gains.
  * Improvements to PDF rendering.
  * Fixes to trainingdata rendering.
  * Added LSTM models+lang models to 101 languages. (tessdata repository)
  * Improved multi-page TIFF handling.
  * Fixed damage to binary images when processing PDFs.
  * Fixes to training process to allow incremental training from a recognition model.
  * Made LSTM the default engine, pushed cube out.
  * Deleted cube code.
  * Changed OEModes --oem 0 for legacy tesseract engine, --oem 1 for LSTM, --oem 2 for both, --oem 3 for default.
  * Avoid use of Leptonica debug parameters or functions.
  * Fixed multi-language mode.
  * Removed support for VS2010.
  * Added Support for VS2015 and VS2017 with CPPAN.
  * Implemented invisible text only for PDF.
  * Added AVX / SSE support for Windows.
  * Enabled OpenMP support.
  * Parameter unlv_tilde_crunching change to false.
  * Miscellaneous Fixes.
  * Detailed Changelog can be found at https://tesseract-ocr.github.io/tessdoc/4.0x-Changelog.html and https://tesseract-ocr.github.io/tessdoc/ReleaseNotes.html#tesseract-release-notes-oct-29-2018---v400

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.
  * Require Leptonica 1.74 or higher.
  * 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.
  * Simplify build and run of ScrollView.
  * Improved PDF output for OS X Preview utility.
  * INCOMPATIBLE fix to hOCR line height information - commit 134ebc3.
  * Added option to build Tesseract without Cube OCR engine (-DNO_CUBE_BUILD).
  * Enable OpenMP support.
  * Many bug fixes.

2015-07-11 - V3.04.00
  * Tesseract development is now done with Git and hosted at github.com (Previously we used Subversion as a VCS and code.google.com for hosting).
  * Tesseract now requires leptonica 1.71 or a higher version.
  * Removed official support for VS 2008.
  * Added support for 39 additional scripts/languages, including: amh, asm, aze_cyrl, bod, bos, ceb, cym, dzo, fas, gle, guj, hat, iku, jav, kat, kat_old, kaz, khm, kir, kur, lao, lat, mar, mya, nep, ori, pan, pus, san, sin, srp_latn, syr, tgk, tir, uig, urd, uzb, uzb_cyrl, yid
  * Major updates to training system as a result of extensive testing on 100 languages.
  * New training data for over 100 languages
  * Improved performance with PIC compilation option.
  * Significant change to invisible font system in pdf output to improve correctness and compatibility with external programs, particularly ghostscript.
  * Improved font identification.
  * Major change to improve layout analysis for heavily diacritic languages: Thai, Vietnamese, Kannada, Telugu etc.
  * Fixed problems with shifted baselines so recognition can recover from layout analysis errors.
  * Major refactor to improve speed on difficult images, especially when running a heap checker.
  * Moved params from global in page layout to tesseractclass.
  * Improved single column layout analysis.
  * Allow ocr output to multiple formats using tesseract command line executable.
  * Fixed issues with mixed eng+ara scripts.
  * Improved script consistency in numbers.
  * Major refactor of control.cpp to enable line recognition.
  * Added tesstrain.sh - a master training script.
  * Added ability to text2image training tool to just list available fonts.
  * Added ability to text2image to underline words.
  * Improved efficiency of image processing for PDF output.
  * Added parameter description for each parameter listed with 'print-parameters' command line option.
  * Added font info to hOCR output.
  * Enabled streaming input and output of multi-page documents.
  * Many bug fixes.

2014-02-04 - V3.03(rc1)
  * Added new training tool text2image to generate box/tif file pairs from
    text and truetype fonts.
  * Added support for PDF output with searchable text.
  * Removed entire IMAGE class and all code in image directory.
  * Tesseract executable: support for output to stdout; limited support for one
    page images from stdin  (especially on Windows)
  * Added Renderer to API to allow document-level processing and output
    of document formats, like hOCR, PDF.
  * Major refactor of word-level recognition, beam search, eliminating dead code.
  * Refactored classifier to make it easier to add new ones.
  * Generalized feature extractor to allow feature extraction from greyscale.
  * Improved sub/superscript treatment.
  * Improved baseline fit.
  * Added set_unicharset_properties to training tools.
  * Many bug fixes.
  * More training source data included.

2012-02-01 - V3.02
  * Moved ResultIterator/PageIterator to ccmain.
  * Added Right-to-left/Bidi capability in the output iterators for Hebrew/Arabic.
  * Added paragraph detection in layout analysis/post OCR.
  * Fixed inconsistent xheight during training and over-chopping.
  * Added simultaneous multi-language capability.
  * Refactored top-level word recognition module.
  * Added experimental equation detector.
  * Improved handling of resolution from input images.
  * Blamer module added for error analysis.
  * Cleaned up externally used namespace by removing includes from baseapi.h.
  * Removed dead memory mangagement code.
  * Tidied up constraints on control parameters.
  * Added support for ShapeTable in classifier and training.
  * Refactored class pruner.
  * Fixed training leaks and randomness.
  * Major improvements to layout analysis for better image detection, diacritic detection, better textline finding, better tabstop finding.
  * Improved line detection and removal.
  * Added fixed pitch chopper for CJK.
  * Added UNICHARSET to WERD_CHOICE to make mult-language handling easier.
  * Fixed problems with internally scaled images.
  * Added page and bbox to string in tr files to identify source of training data better.
  * Fixes to Hindi Shiroreka splitter.
  * Added word bigram correction.
  * Reduced stack memory consumption and eliminated some ugly typedefs.
  * Added new uniform classifier API.
  * Added new training error counter.
  * Fixed endian bug in dawg reader.
  * Many other fixes, including the way in which the chopper finds chops and messes with the outline while it does so.

2010-11-29 - V3.01
  * Removed old/dead serialise/deserialize methods on *LISTIZED classes.
  * Total rewrite of DENORM to better encapsulate operation and make
    for potential to extract features from images.
  * Thread-safety! Moved all critical global and static variables to members of the appropriate class. Tesseract is now thread-safe (multiple instances can be used in parallel in multiple threads.) with the minor exception that some control parameters are still global and affect all threads.
  * Added Cube, a new recognizer for Arabic. Cube can also be used in combination with normal Tesseract for other languages with an improvement in accuracy at the cost of (much) lower speed. *There is no training module for Cube yet.*
  * `OcrEngineMode` in `Init` replaces `AccuracyVSpeed` to control cube.
  * Greatly improved segmentation search with consequent accuracy and speed improvements, especially for Chinese.
  * Added `PageIterator` and `ResultIterator` as cleaner ways to get the full results out of Tesseract, that are not currently provided by any of the `TessBaseAPI::Get*` methods. All other methods, such as the `ETEXT_STRUCT` in particular are deprecated and will be deleted in the future.
  * ApplyBoxes totally rewritten to make training easier. It can now cope with touching/overlapping training characters, and a new boxfile format allows word boxes instead of character boxes, BUT to use that you have to have already bootstrapped the language with character boxes. "Cyclic dependency" on traineddata.
  * Auto orientation and script detection added to page layout analysis.
  * Deleted *lots* of dead code.
  * Fixxht module replaced with scalable data-driven module.
  * Output font characteristics accuracy improved.
  * Removed the double conversion at each classification.
  * Upgraded oldest structs to be classes and deprecated PBLOB.
  * Removed non-deterministic baseline fit.
  * Added fixed length dawgs for Chinese.
  * Handling of vertical text improved.
  * Handling of leader dots improved.
  * Table detection greatly improved.
  * Fixed a couple of memory leaks.
  * Fixed font labels on output text. (Not perfect, but a lot better than before.)
  * Cleanup and more bug fixes
  * Special treatments for Hindi.
  * Support for build in VS2010 with Microsoft Windows SDK for Windows 7 (thanks to Michael Lutz)

2010-09-21 - V3.00
  * Preparations for thread safety:
     * Changed TessBaseAPI methods to be non-static
     * Created a class hierarchy for the directories to hold instance data,
       and began moving code into the classes.
     * Moved thresholding code to a separate class.
  * Added major new page layout analysis module.
  * Added HOCR output (issues 221, 263: thanks to amkryukov).
  * Added Leptonica as main image I/O and handling. Currently optional,
    but in future releases linking with Leptonica will be mandatory.
  * Ambiguity table rewritten to allow definite replacements in place
    of fix_quotes.
  * Added TessdataManager to combine data files into a single file.
  * Some dead code deleted.
  * VC++6 no longer supported. It can't cope with the use of templates.
  * Many more languages added.
  * Doxygenation of most of the function header comments.
  * Added man pages.
  * Added bash completion script (issue 247: thanks to neskiem)
  * Fix integer overview in thresholding (issue 366: thanks to Cyanide.Drake)
  * Add Danish Fraktur support (issues 300, 360: thanks to
    dsl602230@vip.cybercity.dk)
  * Fix file pointer leak (issue 359, thanks to yukihiro.nakadaira)
  * Fix an error using user-words (Issue 345: thanks to max.markin)
  * Fix a memory leak in tablefind.cpp (Issue 342, thanks to zdravco)
  * Fix a segfault due to double fclose (Issue 320, thanks to souther)
  * Fix an automake error (Issue 318, thanks to ichanjz)
  * Fix a Win32 crash on fileFormatIsTiff() (Issues 304, 316, 317, 330, 347,
    349, 352: thanks to nguyenq87, max.markin, zdenop)
  * Fixed a number of errors in newer (stricter) versions of VC++ (Issues
    301, among others)

2009-06-30 - V2.04
  * Integrated bug fixes and patches and misc changes for portability.
  * Integrated a patch to remove some of the "access" macros.
  * Removed dependence on lua from the viewer, speeding it up
    dramatically.
  * Fixed the viewer so it compiles and runs properly!
  * Specifically fixing issues: 1, 63, 67, 71, 76, 81, 82, 106, 111,
   112, 128, 129, 130, 133, 135, 142, 143, 145, 147, 153, 154, 160,
   165, 170, 175, 177, 187, 192, 195, 199, 201, 205, 209, 108, 169

2008-04-22 - V2.03
  * Fixed crash introduced in 2.02.
  * Fixed lack of tessembedded.cpp in distribution.
  * Added test for leptonica header files and conditional test for lib.

2008-04-21 - V2.02 (again)
  * Fixed namespace collisions with jpeg library (INT32).
  * Portability fixes for Windows for new code.
  * Updates to autoconf system for new code.

2008-01-23 - V2.02
  * Improvements to clustering, training and classifier.
  * Major internationalization improvements for large-character-set
  * languages, eg Kannada.
  * Removed some compiler warnings.
  * Added multipage tiff support for training and running.
  * Updated graphics output to talk to new java-based viewer.
  * Added ability to save n-best lists.
  * Added leptonica support for more file types.
  * Improved Init/End to make them safe.
  * Reduced memory use of dictionaries.
  * Added some new APIs to TessBaseAPI.

2007-08-27 - V2.01
  * Fixed UTF8 input problems with box file reader.
  * Fixed various infinite loops and crashes in dawg code.
  * Removed include of config_auto.h from host.h.
  * Added automatic wctype encoding to unicharset_extractor.
  * Fixed dawg table too full error.
  * Removed svn files from tarball.
  * Added new functions to tessdll.
  * Increased maximum utf8 string in a classification result to 8.

2007-07-02 - V2.00
  * Converted internal character handling to UTF8.
  * Trained with 6 languages.
  * Added unicharset_extractor, wordlist2dawg.
  * Added boxfile creation mode.
  * Added UNLV regression test capability.
  * Fixed problems with copyright and registered symbols.
  * Fixed extern "C" declarations problem.

2007-05-15 - V1.04
  * Added dll exports for Windows.
  * Fixed name collisions with stl etc.
  * Made some preliminary changes ready for unicodeization.
  * Several bug fixes discovered during unicodeization.

2007-02-02 - V1.03
  * Added mftraining and cntraining.
  * Added baseapi with adaptive thresholding for grey and color.
  * Fixed many memory leaks.
  * Fixed several bugs including lack of use of adaptive classifier.
  * Added ifdefs to eliminate graphics code and add embedded platform support.
  * Incorporated several patches, including 64-bit builds, Mac builds.
  * Minor accuracy improvements.

2006-10-04 - V1.02
  * Removed dependency on Aspirin.
  * Fixed a few missing Apache license headers.
  * Removed $log.

2006-09-07 - V1.01.
  * Added mfcpch.cpp and getopt.cpp for VC++.
  * Fixed problem with greyscale images and no libtiff.
  * Stopped debug window from being used for the usage output.
  * Fixed load of inttemp for big-endian architectures.
  * Fixed some Mac compilation issues.

2006-06-16 - V1.0 of open source Tesseract checked-in.