path: root/tesseract.patch

diff '--color=auto' --unified --recursive --text doccreator-git/software/DocCreator/CMakeLists.txt doccreator-git2/software/DocCreator/CMakeLists.txt
--- doccreator-git/software/DocCreator/CMakeLists.txt	2023-06-30 14:18:05.876354271 -0500
+++ doccreator-git2/software/DocCreator/CMakeLists.txt	2023-06-30 14:23:40.120257279 -0500
@@ -27,272 +27,7 @@
 
 FIND_PACKAGE(OpenGL REQUIRED)  #required ???
 
-
-OPTION(BUILD_TESSERACT "Build tesseract from sources" OFF)
-
-IF(NOT BUILD_TESSERACT)
-  FIND_PACKAGE(Tesseract QUIET)
-  #if Tesseract is found, Tesseract_TESSDATA_PARENT_DIR should be set
-  IF(NOT Tesseract_FOUND)
-    SET(BUILD_TESSERACT ON CACHE STRING "Build Leptonica/Tesseract from sources." FORCE)
-  else()
-    message(STATUS "Found Tesseract.")
-  ENDIF(NOT Tesseract_FOUND)
-ENDIF(NOT BUILD_TESSERACT)
-
-
-######################################## TESSERACT
-
-IF(BUILD_TESSERACT)
-  
-  message(STATUS "Will download and build leptonica and tesseract from sources")
-
-  SET(LEPTONICA_VERSION 1.80.0)
-  SET(TESSERACT_VERSION 41) #library suffixe
-  SET(TESSERACT_ARCHIVE_VERSION 4.1.1)
-  SET(LEPTONICA_SRC_DIR "${PROJECT_SOURCE_DIR}/thirdparty/leptonica-${LEPTONICA_VERSION}")
-  SET(TESSERACT_SRC_DIR "${PROJECT_SOURCE_DIR}/thirdparty/tesseract-${TESSERACT_ARCHIVE_VERSION}")
-  #TODO: check that directories exist
-
-  IF (NOT EXISTS ${LEPTONICA_SRC_DIR} OR NOT EXISTS ${TESSERACT_SRC_DIR} OR NOT EXISTS ${PROJECT_SOURCE_DIR}/thirdparty/tessdata/fra.traineddata) 
-
-    SET(LEPTONICA_URL "https://github.com/DanBloomberg/leptonica/archive/${LEPTONICA_VERSION}.tar.gz")
-    #SET(LEPTONICA_URL "http://dept-info.labri.fr/~mansenca/leptonica-${LEPTONICA_VERSION}.tar.gz") 
-    SET(TESSERACT_URL "https://github.com/tesseract-ocr/tesseract/archive/${TESSERACT_ARCHIVE_VERSION}.tar.gz")
-    SET(LEPTONICA_ARCHIVE "leptonica.tar.gz")
-    SET(TESSERACT_ARCHIVE "tesseract.tar.gz")
-    IF(NOT EXISTS ${PROJECT_SOURCE_DIR}/thirdparty/${TESSERACT_ARCHIVE})
-      MESSAGE(STATUS "  downloading leptonica & tesseract")
-      FILE(DOWNLOAD ${LEPTONICA_URL} ${PROJECT_SOURCE_DIR}/thirdparty/${LEPTONICA_ARCHIVE} STATUS status LOG log)
-      LIST(GET status 0 status_code)
-      LIST(GET status 1 status_string)
-      IF(NOT status_code EQUAL 0)
-        MESSAGE(FATAL_ERROR "error: downloading ${LEPTONICA_URL} failed: status code=${status_code} string=${status_string}, log=${log}")
-      ENDIF()
-      
-      FILE(DOWNLOAD ${TESSERACT_URL} ${PROJECT_SOURCE_DIR}/thirdparty/${TESSERACT_ARCHIVE} STATUS status LOG log)
-      LIST(GET status 0 status_code)
-      LIST(GET status 1 status_string)
-      IF(NOT status_code EQUAL 0)
-        MESSAGE(FATAL_ERROR "error: downloading ${TESSERACT_URL} failed: status code=${status_code} string=${status_string}, log=${log}")
-      ENDIF()
-      MESSAGE(STATUS "  downloading done.")
-    ENDIF()
-    #download tessdata
-    SET(TESSDATA_URL https://github.com/tesseract-ocr/tessdata_best/raw/master)
-    IF(NOT EXISTS ${PROJECT_SOURCE_DIR}/thirdparty/tessdata/eng.traineddata)
-      MESSAGE(STATUS "  downloading tesseract language models")
-      SET(TESSDATA_ENG_URL "${TESSDATA_URL}/eng.traineddata")
-      FILE(DOWNLOAD ${TESSDATA_ENG_URL} ${PROJECT_SOURCE_DIR}/thirdparty/tessdata/eng.traineddata STATUS status LOG log)
-      LIST(GET status 0 status_code)
-      LIST(GET status 1 status_string)
-      IF(NOT status_code EQUAL 0)
-        MESSAGE(FATAL_ERROR "error: downloading ${TESSDATA_ENG_URL} failed: status code=${status_code} string=${status_string}, log=${log}")
-      ENDIF()
-
-      SET(TESSDATA_FRA_URL "${TESSDATA_URL}/fra.traineddata")
-      FILE(DOWNLOAD ${TESSDATA_FRA_URL} ${PROJECT_SOURCE_DIR}/thirdparty/tessdata/fra.traineddata STATUS status LOG log)
-      LIST(GET status 0 status_code)
-      LIST(GET status 1 status_string)
-      IF(NOT status_code EQUAL 0)
-        MESSAGE(FATAL_ERROR "error: downloading ${TESSDATA_FRA_URL} failed: status code=${status_code} string=${status_string}, log=${log}")
-      ENDIF()
-      MESSAGE(STATUS "  downloading done.")
-    ENDIF()
-
-    EXECUTE_PROCESS(COMMAND ${CMAKE_COMMAND} -E tar xzf ${PROJECT_SOURCE_DIR}/thirdparty/${LEPTONICA_ARCHIVE} WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}/thirdparty )
-    EXECUTE_PROCESS(COMMAND ${CMAKE_COMMAND} -E tar xzf ${PROJECT_SOURCE_DIR}/thirdparty/${TESSERACT_ARCHIVE} WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}/thirdparty )
-    MESSAGE(STATUS "  decompression done.")
-  ENDIF()
-
-
-
-  IF(NOT MSVC AND NOT MINGW)
-    SET(LEPTONICA_LIBRARY "${PROJECT_BINARY_DIR}/leptonica/lib/${CMAKE_STATIC_LIBRARY_PREFIX}leptonica${CMAKE_STATIC_LIBRARY_SUFFIX}")
-    SET(TESSERACT_LIBRARY "${PROJECT_BINARY_DIR}/tesseract/lib/${CMAKE_STATIC_LIBRARY_PREFIX}tesseract${CMAKE_STATIC_LIBRARY_SUFFIX}")
-  ELSE(NOT MSVC AND NOT MINGW)
-
-    #set(MY_DEFINITION 
-    #  $<$<CONFIG:Debug>:definition_for_debug>
-    #  $<$<CONFIG:RelWithDebInfo>:definition_for_rel_with_debug>
-    #  $<$<CONFIG:Release>:definition_for_release> 
-    #  $<$<CONFIG:MinSizeRel>:definition_for_tight_release>
-    #  )
-    
-    #set(DEBUG_EXT "$<$<CONFIG:Debug>:d>")
-    set(DEBUG_EXT "")
-    IF(NOT CMAKE_CONFIGURATION_TYPES)
-      #mono-config, e.g., ninja
-      IF("${CMAKE_BUILD_TYPE}" STREQUAL "Debug" OR
-         "${CMAKE_BUILD_TYPE}" STREQUAL "RelWithDebugInfo")
-	       set(DEBUG_EXT "d")
-      ENDIF()
-    ENDIF()
-    
-
-    SET(LEPTONICA_LIBRARY "${PROJECT_BINARY_DIR}/leptonica/lib/${CMAKE_STATIC_LIBRARY_PREFIX}leptonica-${LEPTONICA_VERSION}${DEBUG_EXT}${CMAKE_STATIC_LIBRARY_SUFFIX}")
-    SET(TESSERACT_LIBRARY "${PROJECT_BINARY_DIR}/tesseract/lib/${CMAKE_STATIC_LIBRARY_PREFIX}tesseract${TESSERACT_VERSION}${DEBUG_EXT}${CMAKE_STATIC_LIBRARY_SUFFIX}")
-
-  ENDIF(NOT MSVC AND NOT MINGW)
-
-
-  include(ExternalProject)
-
-  #build static library with -DSTATIC=ON
-  ExternalProject_Add(leptonica
-    URL ${LEPTONICA_SRC_DIR}
-    PREFIX ${PROJECT_BINARY_DIR}/leptonica
-    PATCH_COMMAND patch -p0 < ${PROJECT_SOURCE_DIR}/thirdparty/leptonica-1.80.0.patch
-    CMAKE_ARGS -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DCMAKE_INSTALL_PREFIX=${PROJECT_BINARY_DIR}/leptonica -DSTATIC=ON -DSW_BUILD=OFF -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
-    BUILD_BYPRODUCTS ${LEPTONICA_LIBRARY}
-    )
-
-  #-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER} is to ensure that leptonica&tesseract are built with the same compiler than DocCreator
-  # It is for example useful on Linux, when compiled with a non-default compiler.
-
-  ExternalProject_Get_Property(leptonica binary_dir)
-  ExternalProject_Get_Property(leptonica source_dir)
-  ExternalProject_Get_Property(leptonica install_dir)
-  set(leptonica_prefix "${install_dir}")
-
-  set(Lept_INCLUDE_DIRS "${install_dir}/include;${install_dir}/include/leptonica") #(not needed to install tesseract), just later
-  set(lept_lib_dirs "${install_dir}/lib")
-
-  #library is in ${binary_dir}/src
-  #headers are in ${source_dir}/src
-
-  ADD_LIBRARY(leptonicaLib STATIC IMPORTED GLOBAL)
-  SET_TARGET_PROPERTIES(leptonicaLib PROPERTIES IMPORTED_LOCATION ${LEPTONICA_LIBRARY})
-  ADD_DEPENDENCIES(leptonicaLib leptonica)
-
-  ExternalProject_Get_Property(leptonica install_dir)
-  message("leptonica install_dir=${install_dir}")
-  message("Leptonica_INCLUDE_DIRS=${Leptonica_INCLUDE_DIRS}")
-  set(Lept_INCLUDE_DIRS "${install_dir}/include;${install_dir}/include/leptonica")
-  message("moi Lept_INCLUDE_DIRS=${Lept_INCLUDE_DIRS}")
-  message("Leptonica_LIBRARIES=${Leptonica_LIBRARIES}")
-
-  #Tesseract is built with -std=c++11
-  #If clang is used, we ensure that libc++ is also used.
-  SET(ADDITIONAL_FLAG "")
-  IF ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang")
-    SET(ADDITIONAL_FLAG "-DCMAKE_CXX_FLAGS=-stdlib=libc++")
-  ENDIF()
-
-  ExternalProject_Add(tesseract
-    DEPENDS leptonica
-    URL ${TESSERACT_SRC_DIR}
-    PREFIX ${PROJECT_BINARY_DIR}/tesseract
-    CMAKE_ARGS -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DLeptonica_DIR=${install_dir}/cmake -DCMAKE_INSTALL_PREFIX=${PROJECT_BINARY_DIR}/tesseract -DSTATIC=ON -DBUILD_TRAINING_TOOLS=OFF -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER} ${ADDITIONAL_FLAG}
-    BUILD_BYPRODUCTS ${TESSERACT_LIBRARY}
-    )
-
-  ExternalProject_Get_Property(tesseract binary_dir)
-  ExternalProject_Get_Property(tesseract source_dir)
-  ExternalProject_Get_Property(tesseract install_dir)
-  set(tesseract_prefix "${install_dir}")
-  set(Tesseract_INCLUDE_DIRS "${install_dir}/include;${install_dir}/include/tesseract") 
-  set(tess_lib_dirs "${install_dir}/lib") 
-
-  ADD_LIBRARY(tesseractLib STATIC IMPORTED GLOBAL)
-  SET_TARGET_PROPERTIES(tesseractLib PROPERTIES IMPORTED_LOCATION ${TESSERACT_LIBRARY})
-  ADD_DEPENDENCIES(tesseractLib tesseract)
-
-  #TODO: the following libraries are already searched by leptonica.
-  #they are in ${install_dir}/cmake/LeptonicaTargets-release.cmake
-  # but I don't know how to retrieve them.
-  # As we use ExternalProject_Add(),
-  # LeptonicaTargets-release.cmake is only created during the build phase (and not the configuration phase)
-  # We cannot use find_package(leptonica) to get them.
-
-  FIND_PACKAGE(ZLIB QUIET)  #ZLIB brought by PNG on linux but not on OSX 
-  IF(ZLIB_FOUND)
-    SET(OTHER_LIBS ${ZLIB_LIBRARIES})
-  ENDIF()
-
-  FIND_PACKAGE(PNG QUIET)
-  IF(PNG_FOUND)
-    SET(OTHER_LIBS "${PNG_LIBRARIES};${OTHER_LIBS}")
-  ENDIF()
-
-  FIND_PACKAGE(JPEG QUIET)
-  IF(JPEG_FOUND)
-    SET(OTHER_LIBS "${JPEG_LIBRARIES};${OTHER_LIBS}")
-  ENDIF()
-
-  FIND_PACKAGE(TIFF QUIET)
-  IF(TIFF_FOUND)
-    SET(OTHER_LIBS "${TIFF_LIBRARIES};${OTHER_LIBS}")
-  ENDIF()
-
-  FIND_PACKAGE(GIF QUIET)
-  IF(GIF_FOUND)
-    SET(OTHER_LIBS "${GIF_LIBRARIES};${OTHER_LIBS}")
-  ENDIF()
-
-  message("OTHER_LIBS=${OTHER_LIBS}")
-  
-  FIND_PACKAGE(PkgConfig QUIET)
-  IF (PKG_CONFIG_FOUND)
-    message("search with pkg config")
-    pkg_check_modules(WEBP QUIET libwebp)
-    IF(WEBP_FOUND)
-      #SET(OTHER_LIBS "${WEBP_LIBRARIES};${OTHER_LIBS}")
-      message("WEBP_FOUND=${WEBP_FOUND} by pkg: $WEBP_LIBRARY=${WEBP_LIBRARY} $WEBP_LIBRARIES=${WEBP_LIBRARIES} WEBP=${WEBP}")
-    ENDIF()
-    pkg_check_modules(JP2K QUIET libopenjp2)
-    IF(JP2K_FOUND)
-      #SET(OTHER_LIBS "${JP2K_LIBRARIES};${OTHER_LIBS}")
-      message("JP2K_FOUND=${JP2K_FOUND} by pkg: $JP2K_LIBRARY=${WEBP_LIBRARY} $JP2K_LIBRARIES=${WEBP_LIBRARIES} JP2K=${JP2K}")
-    ENDIF()
-  ENDIF()
-  if(NOT WEBP)
-    message("search webp directly")
-    find_path(WEBP_INCLUDE_DIR /webp/decode.h)
-    find_library(WEBP_LIBRARY NAMES webp)
-    if (WEBP_INCLUDE_DIR AND WEBP_LIBRARY)
-      message("WEBP_FOUND2=${WEBP_FOUND}: $WEBP_LIBRARY=${WEBP_LIBRARY} $WEBP_LIBRARIES=${WEBP_LIBRARIES} WEBP=${WEBP}")
-      set(OTHER_LIBS "${WEBP_LIBRARY};${OTHER_LIBS}")
-    endif()
-  endif()
-  if(NOT JP2K)
-    message("search jp2k directly")
-    find_path(JP2K_INCLUDE_DIR /openjpeg-2.3/openjpeg.h)
-    find_library(JP2K_LIBRARY NAMES openjp2)
-    if (JP2K_INCLUDE_DIR AND JP2K_LIBRARY)
-      message("JP2K_FOUND2=${JP2K_FOUND}: $JP2K_LIBRARY=${JP2K_LIBRARY} $JP2K_LIBRARIES=${JP2K_LIBRARIES} JP2K=${JP2K}")
-      set(OTHER_LIBS "${JP2K_LIBRARY};${OTHER_LIBS}")
-    endif()
-  endif()
-
-  message("OTHER_LIBS=${OTHER_LIBS}")
-    
-  SET(Lept_LIBRARIES "${LEPTONICA_LIBRARY};${OTHER_LIBS}")
-  SET(Tesseract_LIBRARIES ${TESSERACT_LIBRARY})
-  IF(MSVC OR MINGW)
-    SET(Tesseract_LIBRARIES "${TESSERACT_LIBRARY};Ws2_32")
-  ENDIF()
-
-  message("Lept_LIBRARIES=${Lept_LIBRARIES}")
-  message("Tesseract_LIBRARIES=${Tesseract_LIBRARIES}")
-
-
-  #install tessdata
-  install(DIRECTORY ${PROJECT_SOURCE_DIR}/thirdparty/tessdata DESTINATION share/DocCreator)
-
-  set(Tesseract_TESSDATA_PARENT_DIR "") #empty to be found dynamically
-
-ENDIF()
-
-SET(Tesseract_LIBRARIES "${Tesseract_LIBRARIES};${Lept_LIBRARIES}")
-message("Tesseract_LIBRARIES=${Tesseract_LIBRARIES}")
-
-
-######################################## TESSERACT
-
-
-
-
+include(FindPkgConfig)
 
 SET(
   DocCreatorQ_headers
@@ -555,7 +290,13 @@
 
 ADD_SUBDIRECTORY(${CMAKE_CURRENT_SOURCE_DIR}/src/Degradations/Distortion3DModel)
 
-SET(EXTRA_LIBS ${QT_LIBRARIES} ${Lipsum4Qt_LIBRARIES} ${DocCreator_LIBRARIES} ${OpenCV_LIBS} ${DOC3D_LIBRARIES} ${Tesseract_LIBRARIES})
+# Shared libraries
+pkg_check_modules(libraries
+	REQUIRED IMPORTED_TARGET
+		tesseract
+)
+
+SET(EXTRA_LIBS ${QT_LIBRARIES} ${Lipsum4Qt_LIBRARIES} ${DocCreator_LIBRARIES} ${OpenCV_LIBS} ${DOC3D_LIBRARIES})
 
 #message("EXTRA_LIBS=${EXTRA_LIBS}")
 
@@ -579,7 +320,7 @@
 
 ADD_LIBRARY(${LIBNAME} STATIC ${DocCreatorQ_sources} ${DocCreatorQ_headers} ${MOC_LIB})
 TARGET_INCLUDE_DIRECTORIES(${LIBNAME} PUBLIC ${FRAMEWORK_INCLUDE_DIRS} $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src>)
-TARGET_LINK_LIBRARIES(${LIBNAME} ${EXTRA_LIBS})
+TARGET_LINK_LIBRARIES(${LIBNAME} ${EXTRA_LIBS} PkgConfig::libraries)
 TARGET_COMPILE_OPTIONS(${LIBNAME} PRIVATE ${WARNING_FLAGS})
 SET_TARGET_PROPERTIES(${LIBNAME} PROPERTIES
   CXX_STANDARD 11
@@ -588,7 +329,7 @@
   )
 
 ADD_EXECUTABLE(${PROGNAME} ${DocCreator_sources} ${DocCreator_headers} ${DocCreator_rcs} ${WIN_RESOURCE_FILE})
-TARGET_INCLUDE_DIRECTORIES(${PROGNAME} PRIVATE ${FRAMEWORK_INCLUDE_DIRS} ${DOC3D_INCLUDE_DIRS} $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src> $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/GenerateDocument> ${CMAKE_BINARY_DIR} ${CMAKE_CURRENT_BINARY_DIR} ${Lept_INCLUDE_DIRS} ${Tesseract_INCLUDE_DIRS})
+TARGET_INCLUDE_DIRECTORIES(${PROGNAME} PRIVATE ${FRAMEWORK_INCLUDE_DIRS} ${DOC3D_INCLUDE_DIRS} $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src> $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/GenerateDocument> ${CMAKE_BINARY_DIR} ${CMAKE_CURRENT_BINARY_DIR} ${Lept_INCLUDE_DIRS})
 SET_TARGET_PROPERTIES(${PROGNAME} PROPERTIES INSTALL_RPATH "${PROG_RPATH}")
 TARGET_COMPILE_OPTIONS(${PROGNAME} PRIVATE ${WARNING_FLAGS})
 SET_TARGET_PROPERTIES(${PROGNAME} PROPERTIES
@@ -599,7 +340,7 @@
 # Cf http://www.kitware.com/blog/home/post/510
 #As CMAKE_MACOSX_RPATH is defined to 1,
 # Cmake  will automatically add an RPATH to this executable which is the absolute location of libDocCreator in the build tree.
-TARGET_LINK_LIBRARIES(${PROGNAME} ${EXTRA_LIBS} ${LIBNAME})
+TARGET_LINK_LIBRARIES(${PROGNAME} ${EXTRA_LIBS} ${LIBNAME} PkgConfig::libraries)
 #TARGET_LINK_LIBRARIES(${PROGNAME} ${ALL_LIBS})
 
 
@@ -617,7 +358,7 @@
 
 set(PROG_RPATH "@loader_path/../lib") #B: necessary as we install libs (LipsumQt, framework, doc3D) in ../lib directory [Non-bundle case]
 set(PROG_RPATH "${PROG_RPATH};${QT_INSTALL_PATH}") #B: necessary as Qt libs are marked @rpath in exe
-set(PROG_RPATH "${PROG_RPATH};${LEPTONICA_INSTALL_PATH};${TESSERACT_INSTALL_PATH}") #B: added to be able to launch app without install
+set(PROG_RPATH "${PROG_RPATH};${LEPTONICA_INSTALL_PATH}") #B: added to be able to launch app without install
 #message("PROG_RPATH=${PROG_RPATH}")
 SET_TARGET_PROPERTIES(${PROGNAME} PROPERTIES INSTALL_RPATH "${PROG_RPATH}")
 
@@ -628,12 +369,6 @@
 #GET_TARGET_PROPERTY(CURRENT_INSTALL_RPATHS2 ${PROGNAME} INSTALL_RPATH)
 #message("DEBUG ======= CURRENT_INSTALL_RPATHS2=${CURRENT_INSTALL_RPATHS2}")
 
-IF(BUILD_TESSERACT)
-  add_dependencies(${PROGNAME} tesseractLib)  #B: to be able to compile with "make -j 9"
-ENDIF(BUILD_TESSERACT)
-
-
-
 IF(UNIX)
   install(TARGETS ${PROGNAME} RUNTIME DESTINATION bin)
 ENDIF(UNIX)