Mercurial > hgrepos > Python2 > PyMuPDF
diff mupdf-source/thirdparty/tesseract/CMakeLists.txt @ 2:b50eed0cc0ef upstream
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
The directory name has changed: no version number in the expanded directory now.
| author | Franz Glasner <fzglas.hg@dom66.de> |
|---|---|
| date | Mon, 15 Sep 2025 11:43:07 +0200 |
| parents | |
| children |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/mupdf-source/thirdparty/tesseract/CMakeLists.txt Mon Sep 15 11:43:07 2025 +0200 @@ -0,0 +1,986 @@ +# +# tesseract +# + +# ############################################################################## +# +# cmake settings +# +# ############################################################################## + +cmake_minimum_required(VERSION 3.10 FATAL_ERROR) + +# In-source builds are disabled. +if("${CMAKE_CURRENT_SOURCE_DIR}" STREQUAL "${CMAKE_CURRENT_BINARY_DIR}") + message( + FATAL_ERROR + "CMake generation is not possible within the source directory!" + "\n Remove the CMakeCache.txt file and try again from another folder, " + "e.g.:\n " + "\n rm CMakeCache.txt" + "\n mkdir build" + "\n cd build" + "\n cmake ..") +endif() + +set(CMAKE_MODULE_PATH "${CMAKE_MODULE_PATH};${CMAKE_CURRENT_SOURCE_DIR}/cmake") + +set(EXECUTABLE_OUTPUT_PATH "${CMAKE_BINARY_DIR}/bin") +set(CMAKE_RUNTIME_OUTPUT_DIRECTORY "${EXECUTABLE_OUTPUT_PATH}") + +# Use solution folders. +set_property(GLOBAL PROPERTY USE_FOLDERS ON) +set_property(GLOBAL PROPERTY PREDEFINED_TARGETS_FOLDER "CMake Targets") + +if(NOT ${CMAKE_VERSION} VERSION_LESS "3.15.0") + if(WIN32) + cmake_policy(SET CMP0091 NEW) + message(STATUS "Setting policy CMP0091 to NEW") + endif() +endif() + +# ############################################################################## +# +# project settings +# +# ############################################################################## + +project(tesseract C CXX) + +# Get version with components from VERSION file. +file(STRINGS "VERSION" VERSION_PLAIN) +string(REGEX REPLACE "^([^.]*)\\..*" "\\1" VERSION_MAJOR ${VERSION_PLAIN}) +string(REGEX REPLACE "^[^.]*\\.([^.]*)\\..*" "\\1" VERSION_MINOR + ${VERSION_PLAIN}) +string(REGEX REPLACE "^[^.]*\\.[^.]*\\.([0-9]*).*" "\\1" VERSION_PATCH + ${VERSION_PLAIN}) +if(EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/.git) + execute_process(COMMAND git --git-dir ${CMAKE_CURRENT_SOURCE_DIR}/.git + describe --abbrev=4 OUTPUT_VARIABLE GIT_REV) + string(REGEX REPLACE "\n$" "" PACKAGE_VERSION "${GIT_REV}") +endif() +if(NOT PACKAGE_VERSION) + set(PACKAGE_VERSION ${VERSION_PLAIN}) +endif() + +# Provide also same macro names as autoconf (see configure.ac). +set(GENERIC_MAJOR_VERSION ${VERSION_MAJOR}) +set(GENERIC_MINOR_VERSION ${VERSION_MINOR}) +set(GENERIC_MICRO_VERSION ${VERSION_PATCH}) + +set(MINIMUM_LEPTONICA_VERSION 1.74) + +# ############################################################################## +# +# options +# +# ############################################################################## + +message(STATUS "Configuring tesseract version ${PACKAGE_VERSION}...") + +if(WIN32) + option(SW_BUILD "Build with sw" ON) +else() + option(SW_BUILD "Build with sw" OFF) +endif() +option(OPENMP_BUILD "Build with openmp support" OFF) # see issue #1662 +option(GRAPHICS_DISABLED "Disable disable graphics (ScrollView)" OFF) +option(DISABLED_LEGACY_ENGINE "Disable the legacy OCR engine" OFF) +option(ENABLE_LTO "Enable link-time optimization" OFF) +option(FAST_FLOAT "Enable float for LSTM" ON) +option(ENABLE_NATIVE + "Enable optimization for host CPU (could break HW compatibility)" OFF) +# see +# https://stackoverflow.com/questions/52653025/why-is-march-native-used-so-rarely +option(BUILD_TRAINING_TOOLS "Build training tools" ON) +option(BUILD_TESTS "Build tests" OFF) +option(USE_SYSTEM_ICU "Use system ICU" OFF) +option(DISABLE_TIFF "Disable build with libtiff (if available)" OFF) +option(DISABLE_ARCHIVE "Disable build with libarchive (if available)" OFF) +option(DISABLE_CURL "Disable build with libcurl (if available)" OFF) +option(INSTALL_CONFIGS "Install tesseract configs" ON) + +if(NOT ${CMAKE_VERSION} VERSION_LESS "3.15.0") + if(WIN32 AND MSVC) + option(WIN32_MT_BUILD "Build with MT flag for MSVC" OFF) + endif() +endif() + +# ############################################################################## +# +# compiler and linker +# +# ############################################################################## + +if(CMAKE_CXX_COMPILER_ID MATCHES "Clang") + set(CLANG 1) +endif() + +if(NOT CMAKE_BUILD_TYPE) + message(STATUS "Setting build type to 'Release' as none was specified.") + set(CMAKE_BUILD_TYPE + Release + CACHE STRING "Choose the type of build." FORCE) + set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release") +endif() + +include(CheckCXXCompilerFlag) + +set(CMAKE_CXX_STANDARD 17) +if("cxx_std_20" IN_LIST CMAKE_CXX_COMPILE_FEATURES) + set(CMAKE_CXX_STANDARD 20) +endif() +set(CMAKE_CXX_STANDARD_REQUIRED ON) +if(NOT CMAKE_CXX_COMPILER_ID STREQUAL "GNU") + # cygwin gnu c++ needs to use -std=gnu++17 instead of -std=c++17 + set(CMAKE_CXX_EXTENSIONS OFF) +endif() + +if(BUILD_SHARED_LIBS) + set(CMAKE_CXX_VISIBILITY_PRESET hidden) +endif() + +# LTO +cmake_policy(SET CMP0069 NEW) +include(CheckIPOSupported) +check_ipo_supported(RESULT LTO_SUPPORTED OUTPUT error) +if(LTO_SUPPORTED) + message(STATUS "IPO / LTO supported") +else() + message(STATUS "IPO / LTO not supported: <${error}>") +endif() + +set(MARCH_NATIVE_OPT OFF) +if(ENABLE_NATIVE) + check_cxx_compiler_flag("-march=native" COMPILER_SUPPORTS_MARCH_NATIVE) + if(COMPILER_SUPPORTS_MARCH_NATIVE) + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} -march=native") + if(NOT CLANG AND MSVC) + # clang-cl does not know this argument + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} -mtune=native") + endif() + set(MARCH_NATIVE_OPT ON) + endif(COMPILER_SUPPORTS_MARCH_NATIVE) +endif(ENABLE_NATIVE) + +message(STATUS "CMAKE_SYSTEM_PROCESSOR=<${CMAKE_SYSTEM_PROCESSOR}>") + +if(CMAKE_SYSTEM_PROCESSOR MATCHES "x86|x86_64|AMD64|amd64|i386|i686") + + set(HAVE_NEON FALSE) + if(MSVC) + set(HAVE_AVX ON) + set(AVX_COMPILE_FLAGS "/arch:AVX") + add_definitions("-DHAVE_AVX") + + set(HAVE_AVX2 ON) + set(AVX2_COMPILE_FLAGS "/arch:AVX2") + add_definitions("-DHAVE_AVX2") + + set(HAVE_AVX512F ON) + set(AVX512F_COMPILE_FLAGS "/arch:AVX512") + add_definitions("-DHAVE_AVX512F") + + set(HAVE_FMA ON) + set(FMA_COMPILE_FLAGS "-D__FMA__") + add_definitions("-DHAVE_FMA") + + set(HAVE_SSE4_1 ON) + set(SSE4_1_COMPILE_FLAGS "-D__SSE4_1__") + add_definitions("-DHAVE_SSE4_1") + + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} -openmp:experimental") + add_definitions("-DOPENMP_SIMD") + + # clang with MSVC compatibility + if(CLANG) + set(CMAKE_CXX_FLAGS + "${CMAKE_CXX_FLAGS} -Wno-microsoft-unqualified-friend") + if(HAVE_FMA) + set(FMA_COMPILE_FLAGS "-mfma ${FMA_COMPILE_FLAGS}") + endif(HAVE_FMA) + if(HAVE_SSE4_1) + set(SSE4_1_COMPILE_FLAGS "-msse4.1 ${SSE4_1_COMPILE_FLAGS}") + endif(HAVE_SSE4_1) + endif(CLANG) + else() # if not MSVC + check_cxx_compiler_flag("-mavx" HAVE_AVX) + if(HAVE_AVX) + set(AVX_COMPILE_FLAGS "-mavx") + add_definitions("-DHAVE_AVX") + endif(HAVE_AVX) + + check_cxx_compiler_flag("-mavx2" HAVE_AVX2) + if(HAVE_AVX2) + set(AVX2_COMPILE_FLAGS "-mavx2") + add_definitions("-DHAVE_AVX2") + endif() + + check_cxx_compiler_flag("-mavx512f" HAVE_AVX512F) + if(HAVE_AVX512F) + set(AVX512F_COMPILE_FLAGS "-mavx512f") + add_definitions("-DHAVE_AVX512F") + endif() + + check_cxx_compiler_flag("-mfma" HAVE_FMA) + if(HAVE_FMA) + set(FMA_COMPILE_FLAGS "-mfma") + add_definitions("-DHAVE_FMA") + endif() + + check_cxx_compiler_flag("-msse4.1" HAVE_SSE4_1) + if(HAVE_SSE4_1) + set(SSE4_1_COMPILE_FLAGS "-msse4.1") + add_definitions("-DHAVE_SSE4_1") + endif() + + check_cxx_compiler_flag("-fopenmp-simd" OPENMP_SIMD) + if(OPENMP_SIMD) + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} -fopenmp-simd") + add_definitions("-DOPENMP_SIMD") + endif(OPENMP_SIMD) + endif(MSVC) + +elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64.*|AARCH64.*") + + set(HAVE_AVX FALSE) + set(HAVE_AVX2 FALSE) + set(HAVE_AVX512F FALSE) + set(HAVE_FMA FALSE) + set(HAVE_SSE4_1 FALSE) + set(HAVE_NEON TRUE) + +elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "arm.*") + + set(HAVE_AVX FALSE) + set(HAVE_AVX2 FALSE) + set(HAVE_AVX512F FALSE) + set(HAVE_FMA FALSE) + set(HAVE_SSE4_1 FALSE) + + check_cxx_compiler_flag("-mfpu=neon" HAVE_NEON) + if(HAVE_NEON) + set(NEON_COMPILE_FLAGS "-mfpu=neon") + endif(HAVE_NEON) + +else() + + set(HAVE_AVX FALSE) + set(HAVE_AVX2 FALSE) + set(HAVE_AVX512F FALSE) + set(HAVE_FMA FALSE) + set(HAVE_NEON FALSE) + set(HAVE_SSE4_1 FALSE) + +endif(CMAKE_SYSTEM_PROCESSOR MATCHES "x86|x86_64|AMD64|amd64|i386|i686") + +if(HAVE_NEON) + message(STATUS "LTO build is not supported on arm/RBPi.") + set(ENABLE_LTO FALSE) # enable LTO cause fatal error on arm/RBPi +endif() + +# Compiler specific environment +if(CMAKE_COMPILER_IS_GNUCXX OR MINGW) + set(CMAKE_CXX_FLAGS_DEBUG + "${CMAKE_CXX_FLAGS_DEBUG} -Wall -DDEBUG -pedantic -Og") +elseif(MSVC) + add_definitions(-D_CRT_SECURE_NO_WARNINGS) + add_definitions(-D_CRT_NONSTDC_NO_DEPRECATE) # strdup + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /utf-8") + if(NOT CLANG) + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /MP") + endif() + # Hide some warnings for release target wd4244 'argument': conversion from + # 'uint64_t' to 'unsigned int', possible loss of data wd4251 needs to have + # dll-interface wd4267 return': conversion from 'size_t' to 'int', possible + # loss of data wd4275 non dll-interface class wd4305 ...truncation from + # 'double' to 'float' + set(CMAKE_CXX_FLAGS_RELEASE + "${CMAKE_CXX_FLAGS_RELEASE} /wd4244 /wd4305 /wd4267 /wd4251 /wd4275 /wd4005" + ) + set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /wd4068") + # Don't use /Wall because it generates too many warnings. + set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /W0 /bigobj") + # MT flag + if(WIN32_MT_BUILD) + set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>") + message(STATUS "Building with static CRT.") + endif() +endif() +if(CLANG) # clang all platforms + set(CMAKE_CXX_FLAGS_RELEASE + "${CMAKE_CXX_FLAGS_RELEASE} -Wno-unused-command-line-argument") + set(CMAKE_CXX_FLAGS_DEBUG + "${CMAKE_CXX_FLAGS_DEBUG} -Wall -DDEBUG -pedantic -O0") +endif() + +if(OPENMP_BUILD + AND MSVC + AND "${MSVC_VERSION}" LESS 1929) + set(OPENMP_BUILD OFF) +endif() +if(OPENMP_BUILD) + find_package(OpenMP QUIET) + # https://stackoverflow.com/questions/12399422 + # how-to-set-linker-flags-for-openmp-in-cmakes-try-compile-function + if(NOT OpenMP_FOUND + AND CLANG + AND WIN32) + # workaround because find_package(OpenMP) does not work for clang-cl + # https://gitlab.kitware.com/cmake/cmake/issues/19404 + check_include_file_cxx(omp.h HAVE_OMP_H_INCLUDE) + find_library(OpenMP_LIBRARY NAMES omp libomp.lib) + message(">> OpenMP_LIBRARY: ${OpenMP_LIBRARY}") + if(MSVC) + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /openmp") + else() + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fopenmp") + endif() + set(OpenMP_FOUND 1) + # OpenMP 3.1 is fully supported from Clang 3.8.0 + add_definitions(-D_OPENMP=201107) + endif() + if(MSVC) + # Note: -openmp:llvm is available for X64 from MSVC 16.9 from MSVC 16.10 + # Preview 2 there is support also for x86 and arm64 + # https://devblogs.microsoft.com/cppblog/openmp-updates-and-fixes-for-cpp-in-visual-studio-2019-16-10/ + if("${OpenMP_CXX_FLAGS}" STREQUAL "-openmp") + set(OpenMP_CXX_FLAGS "-openmp:llvm") + endif() + endif() + if(OpenMP_FOUND) + message(">> OpenMP_FOUND ${OpenMP_FOUND} version: ${OpenMP_CXX_VERSION}") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}") + if(NOT TARGET OpenMP::OpenMP_CXX) + add_library(OpenMP::OpenMP_CXX IMPORTED INTERFACE) + endif() + endif() +endif() + +if(CYGWIN) + add_definitions(-D__CYGWIN__) +elseif(UNIX) + if(NOT ANDROID) + set(LIB_pthread pthread) + endif() +elseif(WIN32) + set(LIB_Ws2_32 Ws2_32) +endif() + +add_definitions("-DCMAKE_BUILD") + +# ############################################################################## +# +# packages +# +# ############################################################################## +include(CheckFunctions) + +if(SW_BUILD) + find_package(SW REQUIRED) + if(BUILD_SHARED_LIBS) + set(SW_BUILD_SHARED_LIBS 1) + else() + set(SW_BUILD_SHARED_LIBS 0) + endif() + sw_add_package(org.sw.demo.danbloomberg.leptonica + org.sw.demo.libarchive.libarchive) + if(BUILD_TRAINING_TOOLS) + sw_add_package(org.sw.demo.gnome.pango.pangocairo + org.sw.demo.unicode.icu.i18n) + endif() + sw_execute() +else() + find_package(PkgConfig) + # Check for required library. option -DLeptonica_DIR=path => cmake hint where + # to find leptonica + find_package(Leptonica ${MINIMUM_LEPTONICA_VERSION} CONFIG) + if(NOT Leptonica_FOUND AND PKG_CONFIG_EXECUTABLE) + pkg_check_modules(Leptonica lept>=${MINIMUM_LEPTONICA_VERSION}) + link_directories(${Leptonica_LIBRARY_DIRS}) + endif() + if(NOT Leptonica_FOUND) + message(FATAL_ERROR "Cannot find required library Leptonica. Quitting!") + else() + message(STATUS "Found leptonica version: ${Leptonica_VERSION}") + endif(NOT Leptonica_FOUND) + include_directories(${Leptonica_INCLUDE_DIRS}) + + check_leptonica_tiff_support() + if ((NOT LEPT_TIFF_RESULT EQUAL 0) AND LEPT_TIFF_COMPILE_SUCCESS) + message(NOTICE "Leptonica was build without TIFF support! Disabling TIFF support...") + set(DISABLE_TIFF ON) + elseif(NOT ${CMAKE_VERSION} VERSION_LESS "3.25") + message(STATUS "Leptonica was build with TIFF support.") + endif() + + # Check for optional libraries. + if(DISABLE_TIFF) + set(HAVE_TIFFIO_H OFF) + message(STATUS "TIFF support disabled.") + else(DISABLE_TIFF) + find_package(TIFF) # for tesseract + if(NOT TIFF_FOUND AND PKG_CONFIG_EXECUTABLE) + # try PKG_CONFIG to find libtiff if cmake failed + pkg_check_modules(TIFF libtiff-4) + endif() + if(TIFF_FOUND) + set(HAVE_TIFFIO_H ON) + include_directories(${TIFF_INCLUDE_DIRS}) + endif(TIFF_FOUND) + endif(DISABLE_TIFF) + if(DISABLE_ARCHIVE) + set(HAVE_LIBARCHIVE OFF) + message(STATUS "LibArchive support disabled.") + else(DISABLE_ARCHIVE) + find_package(LibArchive) + if(NOT LibArchive_FOUND AND PKG_CONFIG_EXECUTABLE) + # try PKG_CONFIG to find libarchive if cmake failed + pkg_check_modules(LibArchive libarchive) + endif() + if(LibArchive_FOUND) + set(HAVE_LIBARCHIVE ON) + include_directories(${LibArchive_INCLUDE_DIRS}) + endif(LibArchive_FOUND) + endif(DISABLE_ARCHIVE) + if(DISABLE_CURL) + set(HAVE_LIBCURL OFF) + message(STATUS "CURL support disabled.") + else(DISABLE_CURL) + find_package(CURL) + if(NOT CURL_FOUND AND PKG_CONFIG_EXECUTABLE) + # try PKG_CONFIG to find libcurl if cmake failed + pkg_check_modules(CURL libcurl) + endif() + if(CURL_FOUND) + set(HAVE_LIBCURL ON) + include_directories(${CURL_INCLUDE_DIRS}) + endif(CURL_FOUND) + endif(DISABLE_CURL) +endif() + +# ############################################################################## +# +# configure +# +# ############################################################################## + +if(MSVC) + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} /fp:fast") +else() + set(DOTPRODUCT_FLAGS "${DOTPRODUCT_FLAGS} -O3 -ffast-math") +endif() + +include (GNUInstallDirs) + +set(AUTOCONFIG_SRC ${CMAKE_CURRENT_BINARY_DIR}/config_auto.h.in) +set(AUTOCONFIG ${CMAKE_CURRENT_BINARY_DIR}/config_auto.h) +add_definitions(-DHAVE_CONFIG_H) + +if(GRAPHICS_DISABLED) + message("ScrollView debugging disabled.") +endif() +set(CMAKE_REQUIRED_INCLUDES + ${CMAKE_REQUIRED_INCLUDES} "${CMAKE_PREFIX_PATH}/include" + ${CMAKE_INSTALL_INCLUDEDIR}) +include(Configure) + +configure_file(${AUTOCONFIG_SRC} ${AUTOCONFIG} @ONLY) + +set(INCLUDE_DIR ${CMAKE_INSTALL_INCLUDEDIR}) +set(LIBRARY_DIRS ${CMAKE_INSTALL_LIBDIR}) + +configure_file(${CMAKE_CURRENT_SOURCE_DIR}/include/tesseract/version.h.in + ${CMAKE_CURRENT_BINARY_DIR}/include/tesseract/version.h @ONLY) + +include(CMakePackageConfigHelpers) +include(GenerateExportHeader) + +# show summary of configuration +if(${CMAKE_BUILD_TYPE} MATCHES Debug) + set(COMPILER_FLAGS "${CMAKE_CXX_FLAGS} ${CMAKE_CXX_FLAGS_DEBUG}") +elseif(${CMAKE_BUILD_TYPE} MATCHES Release) + set(COMPILER_FLAGS "${CMAKE_CXX_FLAGS} ${CMAKE_CXX_FLAGS_RELEASE}") + if(LTO_SUPPORTED AND ENABLE_LTO) + set(CMAKE_INTERPROCEDURAL_OPTIMIZATION TRUE) + else() + set(CMAKE_INTERPROCEDURAL_OPTIMIZATION FALSE) + endif() # LTO_SUPPORTED +endif() + +if(CMAKE_SIZEOF_VOID_P EQUAL 8) + set(BUILD_ARCH "64 bits") +elseif(CMAKE_SIZEOF_VOID_P EQUAL 4) + set(BUILD_ARCH "32 bits") +endif() + +message(STATUS) +message(STATUS "General configuration for Tesseract ${PACKAGE_VERSION}") +message(STATUS "--------------------------------------------------------") +message(STATUS "Build type: ${CMAKE_BUILD_TYPE} ${BUILD_ARCH}") +message(STATUS "Compiler: ${CMAKE_CXX_COMPILER_ID}") +message(STATUS "Used standard: C++${CMAKE_CXX_STANDARD}") +message(STATUS "CXX compiler options: ${COMPILER_FLAGS}") +get_directory_property(DirCompDefs COMPILE_DEFINITIONS) +message(STATUS "Compile definitions = ${DirCompDefs}") +message(STATUS "Linker options: ${CMAKE_EXE_LINKER_FLAGS} " + "${CMAKE_EXE_LINKER_FLAGS_${CMAKE_BUILD_TYPE_UP}}") +message(STATUS "Install directory: ${CMAKE_INSTALL_PREFIX}") +message(STATUS "HAVE_AVX: ${HAVE_AVX}") +message(STATUS "HAVE_AVX2: ${HAVE_AVX2}") +message(STATUS "HAVE_AVX512F: ${HAVE_AVX512F}") +message(STATUS "HAVE_FMA: ${HAVE_FMA}") +message(STATUS "HAVE_SSE4_1: ${HAVE_SSE4_1}") +message(STATUS "MARCH_NATIVE_OPT: ${MARCH_NATIVE_OPT}") +message(STATUS "HAVE_NEON: ${HAVE_NEON}") +message(STATUS "Link-time optimization: ${CMAKE_INTERPROCEDURAL_OPTIMIZATION}") +message(STATUS "--------------------------------------------------------") +message(STATUS "Build with sw [SW_BUILD]: ${SW_BUILD}") +message(STATUS "Build with openmp support [OPENMP_BUILD]: ${OPENMP_BUILD}") +message(STATUS "Build with libarchive support [HAVE_LIBARCHIVE]: " + "${HAVE_LIBARCHIVE}") +message(STATUS "Build with libcurl support [HAVE_LIBCURL]: ${HAVE_LIBCURL}") +message(STATUS "Enable float for LSTM [FAST_FLOAT]: ${FAST_FLOAT}") +message(STATUS "Enable optimization for host CPU (could break HW compatibility)" + " [ENABLE_NATIVE]: ${ENABLE_NATIVE}") +message(STATUS "Disable disable graphics (ScrollView) [GRAPHICS_DISABLED]: " + "${GRAPHICS_DISABLED}") +message(STATUS "Disable the legacy OCR engine [DISABLED_LEGACY_ENGINE]: " + "${DISABLED_LEGACY_ENGINE}") +message(STATUS "Build training tools [BUILD_TRAINING_TOOLS]: " + "${BUILD_TRAINING_TOOLS}") +message(STATUS "Build tests [BUILD_TESTS]: ${BUILD_TESTS}") +message(STATUS "Use system ICU Library [USE_SYSTEM_ICU]: ${USE_SYSTEM_ICU}") +message( + STATUS "Install tesseract configs [INSTALL_CONFIGS]: ${INSTALL_CONFIGS}") +message(STATUS "--------------------------------------------------------") +message(STATUS) + +# ############################################################################## +# +# build +# +# ############################################################################## + +include(BuildFunctions) +include(SourceGroups) + +add_definitions(-D_SILENCE_STDEXT_HASH_DEPRECATION_WARNINGS=1) + +include_directories(${CMAKE_CURRENT_BINARY_DIR}) +include_directories(${CMAKE_CURRENT_BINARY_DIR}/include) +if(ANDROID_TOOLCHAIN) + include_directories(${ANDROID_TOOLCHAIN}/sysroot/usr/include) + add_compile_definitions(__ANDROID_API_FUTURE__) +endif() + +# ############################################################################## +# LIBRARY tesseract +# ############################################################################## + +file( + GLOB + TESSERACT_SRC + src/ccmain/*.cpp + src/ccstruct/*.cpp + src/ccutil/*.cpp + src/classify/*.cpp + src/cutil/*.cpp + src/dict/*.cpp + src/lstm/*.cpp + src/textord/*.cpp + src/viewer/*.cpp + src/wordrec/*.cpp) + +if(DISABLED_LEGACY_ENGINE) + # prepend path to list of source files + function(prepend_path srcs path) + set(tmp, "") + foreach(src IN LISTS ${srcs}) + list(APPEND tmp ${path}/${src}) + endforeach(src ${srcs}) + set(${srcs} + ${tmp} + PARENT_SCOPE) + endfunction() + + set(TESSERACT_SRC_LEGACY + src/ccmain/adaptions.cpp + src/ccmain/docqual.cpp + src/ccmain/equationdetect.cpp + src/ccmain/fixspace.cpp + src/ccmain/fixxht.cpp + src/ccmain/osdetect.cpp + src/ccmain/par_control.cpp + src/ccmain/recogtraining.cpp + src/ccmain/superscript.cpp + src/ccmain/tessbox.cpp + src/ccmain/tfacepp.cpp + src/ccstruct/fontinfo.cpp + src/ccstruct/params_training_featdef.cpp + src/ccutil/ambigs.cpp + src/ccutil/bitvector.cpp + src/ccutil/indexmapbidi.cpp + src/classify/adaptive.cpp + src/classify/adaptmatch.cpp + src/classify/blobclass.cpp + src/classify/cluster.cpp + src/classify/clusttool.cpp + src/classify/cutoffs.cpp + src/classify/featdefs.cpp + src/classify/float2int.cpp + src/classify/fpoint.cpp + src/classify/intfeaturespace.cpp + src/classify/intfx.cpp + src/classify/intmatcher.cpp + src/classify/intproto.cpp + src/classify/kdtree.cpp + src/classify/mf.cpp + src/classify/mfoutline.cpp + src/classify/mfx.cpp + src/classify/normfeat.cpp + src/classify/normmatch.cpp + src/classify/ocrfeatures.cpp + src/classify/outfeat.cpp + src/classify/picofeat.cpp + src/classify/protos.cpp + src/classify/shapeclassifier.cpp + src/classify/shapetable.cpp + src/classify/tessclassifier.cpp + src/classify/trainingsample.cpp + src/dict/permdawg.cpp + src/dict/hyphen.cpp + src/wordrec/associate.cpp + src/wordrec/chop.cpp + src/wordrec/chopper.cpp + src/wordrec/drawfx.cpp + src/wordrec/findseam.cpp + src/wordrec/gradechop.cpp + src/wordrec/language_model.cpp + src/wordrec/lm_consistency.cpp + src/wordrec/lm_pain_points.cpp + src/wordrec/lm_state.cpp + src/wordrec/outlines.cpp + src/wordrec/params_model.cpp + src/wordrec/pieces.cpp + src/wordrec/plotedges.cpp + src/wordrec/render.cpp + src/wordrec/segsearch.cpp + src/wordrec/wordclass.cpp) + prepend_path(TESSERACT_SRC_LEGACY "${CMAKE_CURRENT_SOURCE_DIR}") + list(REMOVE_ITEM TESSERACT_SRC ${TESSERACT_SRC_LEGACY}) +endif(DISABLED_LEGACY_ENGINE) + +list(APPEND arch_files src/arch/dotproduct.cpp src/arch/simddetect.cpp + src/arch/intsimdmatrix.cpp) + +if(DOTPRODUCT_FLAGS) + set_source_files_properties(src/arch/dotproduct.cpp + PROPERTIES COMPILE_FLAGS ${DOTPRODUCT_FLAGS}) +endif(DOTPRODUCT_FLAGS) +if(HAVE_AVX) + list(APPEND arch_files_opt src/arch/dotproductavx.cpp) + set_source_files_properties(src/arch/dotproductavx.cpp + PROPERTIES COMPILE_FLAGS ${AVX_COMPILE_FLAGS}) +endif(HAVE_AVX) +if(HAVE_AVX2) + list(APPEND arch_files_opt src/arch/intsimdmatrixavx2.cpp + src/arch/dotproductavx.cpp) + set_source_files_properties(src/arch/intsimdmatrixavx2.cpp + PROPERTIES COMPILE_FLAGS ${AVX2_COMPILE_FLAGS}) +endif(HAVE_AVX2) +if(HAVE_AVX512F) + list(APPEND arch_files_opt src/arch/dotproductavx512.cpp) + set_source_files_properties(src/arch/dotproductavx512.cpp + PROPERTIES COMPILE_FLAGS ${AVX512F_COMPILE_FLAGS}) +endif(HAVE_AVX512F) +if(HAVE_FMA) + list(APPEND arch_files_opt src/arch/dotproductfma.cpp) + set_source_files_properties(src/arch/dotproductfma.cpp + PROPERTIES COMPILE_FLAGS ${FMA_COMPILE_FLAGS}) +endif(HAVE_FMA) +if(HAVE_SSE4_1) + list(APPEND arch_files_opt src/arch/dotproductsse.cpp + src/arch/intsimdmatrixsse.cpp) + set_source_files_properties( + src/arch/dotproductsse.cpp src/arch/intsimdmatrixsse.cpp + PROPERTIES COMPILE_FLAGS ${SSE4_1_COMPILE_FLAGS}) +endif(HAVE_SSE4_1) +if(HAVE_NEON) + list(APPEND arch_files_opt src/arch/dotproductneon.cpp + src/arch/intsimdmatrixneon.cpp) + if(NEON_COMPILE_FLAGS) + set_source_files_properties( + src/arch/dotproductneon.cpp src/arch/intsimdmatrixneon.cpp + PROPERTIES COMPILE_FLAGS ${NEON_COMPILE_FLAGS}) + endif() +endif(HAVE_NEON) + +file( + GLOB_RECURSE + TESSERACT_HDR + include/* + src/arch/*.h + src/ccmain/*.h + src/ccstruct/*.h + src/ccutil/*.h + src/classify/*.h + src/cutil/*.h + src/dict/*.h + src/lstm/*.h + src/textord/*.h + src/viewer/*.h + src/wordrec/*.h) + +set(TESSERACT_SRC + ${TESSERACT_SRC} + src/api/baseapi.cpp + src/api/capi.cpp + src/api/renderer.cpp + src/api/altorenderer.cpp + src/api/pagerenderer.cpp + src/api/hocrrenderer.cpp + src/api/lstmboxrenderer.cpp + src/api/pdfrenderer.cpp + src/api/wordstrboxrenderer.cpp) + +set(TESSERACT_CONFIGS + tessdata/configs/alto + tessdata/configs/ambigs.train + tessdata/configs/api_config + tessdata/configs/bazaar + tessdata/configs/bigram + tessdata/configs/box.train + tessdata/configs/box.train.stderr + tessdata/configs/digits + tessdata/configs/get.images + tessdata/configs/hocr + tessdata/configs/inter + tessdata/configs/kannada + tessdata/configs/linebox + tessdata/configs/logfile + tessdata/configs/lstm.train + tessdata/configs/lstmbox + tessdata/configs/lstmdebug + tessdata/configs/makebox + tessdata/configs/page + tessdata/configs/pdf + tessdata/configs/quiet + tessdata/configs/rebox + tessdata/configs/strokewidth + tessdata/configs/tsv + tessdata/configs/txt + tessdata/configs/unlv + tessdata/configs/wordstrbox) + +set(TESSERACT_TESSCONFIGS + tessdata/tessconfigs/batch tessdata/tessconfigs/batch.nochop + tessdata/tessconfigs/matdemo tessdata/tessconfigs/msdemo + tessdata/tessconfigs/nobatch tessdata/tessconfigs/segdemo) + +set(LIBTESSFILES ${TESSERACT_SRC} ${arch_files} ${arch_files_opt} + ${TESSERACT_HDR}) + +source_group(TREE ${CMAKE_CURRENT_SOURCE_DIR} FILES ${LIBTESSFILES}) + +add_library(libtesseract ${LIBTESSFILES}) +target_include_directories( + libtesseract BEFORE + PRIVATE src + PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/arch> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/ccmain> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/ccstruct> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/ccutil> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/classify> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/cutil> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/dict> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/lstm> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/textord> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/viewer> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/wordrec> + $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/src/training>) +if(BUILD_SHARED_LIBS) + target_compile_definitions( + libtesseract + PRIVATE -DTESS_EXPORTS + INTERFACE -DTESS_IMPORTS) + # generate_export_header (libtesseract EXPORT_MACRO_NAME TESS_API) +endif() +target_link_libraries(libtesseract PRIVATE ${LIB_Ws2_32} ${LIB_pthread}) +if(OpenMP_CXX_FOUND) + target_link_libraries(libtesseract PUBLIC OpenMP::OpenMP_CXX) +endif() +if(LibArchive_FOUND) + target_link_libraries(libtesseract PUBLIC ${LibArchive_LIBRARIES}) +endif(LibArchive_FOUND) +if(CURL_FOUND) + if(NOT CURL_LIBRARIES) + target_link_libraries(libtesseract PUBLIC CURL::libcurl) + else() + target_link_libraries(libtesseract PUBLIC ${CURL_LIBRARIES}) + endif() +endif(CURL_FOUND) + +set_target_properties( + libtesseract PROPERTIES VERSION + ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_PATCH}) +set_target_properties( + libtesseract PROPERTIES SOVERSION + ${VERSION_MAJOR}.${VERSION_MINOR}) + +set_target_properties( + libtesseract + PROPERTIES + OUTPUT_NAME + tesseract$<$<BOOL:${WIN32}>:${VERSION_MAJOR}${VERSION_MINOR}$<$<CONFIG:DEBUG>:d>> +) + +if(SW_BUILD) + target_link_libraries(libtesseract PUBLIC org.sw.demo.danbloomberg.leptonica + org.sw.demo.libarchive.libarchive) + file(WRITE ${CMAKE_CURRENT_BINARY_DIR}/TesseractTargets.cmake + "include(${CMAKE_CURRENT_BINARY_DIR}/cppan.cmake)\n") + export( + TARGETS libtesseract + APPEND + FILE ${CMAKE_CURRENT_BINARY_DIR}/TesseractTargets.cmake + NAMESPACE Tesseract::) +else() + target_link_libraries(libtesseract PUBLIC ${Leptonica_LIBRARIES}) + export( + TARGETS libtesseract + FILE ${CMAKE_CURRENT_BINARY_DIR}/TesseractTargets.cmake + NAMESPACE Tesseract::) +endif() + +if(WIN32 + AND CLANG + AND OPENMP_BUILD) + # Workaround for "libomp.lib is not automatically added on Windows" see: + # http://lists.llvm.org/pipermail/openmp-dev/2015-August/000857.html + target_link_libraries(libtesseract PRIVATE ${OpenMP_LIBRARY}) +endif() + +if(ANDROID) + add_definitions(-DANDROID) + find_package(CpuFeaturesNdkCompat REQUIRED) + target_include_directories( + libtesseract + PRIVATE "${CpuFeaturesNdkCompat_DIR}/../../../include/ndk_compat") + target_link_libraries(libtesseract PRIVATE CpuFeatures::ndk_compat) +endif() + +# ############################################################################## +# EXECUTABLE tesseract +# ############################################################################## + +add_executable(tesseract src/tesseract.cpp) +target_link_libraries(tesseract libtesseract) +if(HAVE_TIFFIO_H AND WIN32) + target_link_libraries(tesseract ${TIFF_LIBRARIES}) +endif() + +if(OPENMP_BUILD AND UNIX) + target_link_libraries(tesseract pthread) +endif() + +# ############################################################################## + +if(BUILD_TESTS + AND EXISTS + ${CMAKE_CURRENT_SOURCE_DIR}/unittest/third_party/googletest/CMakeLists.txt +) + add_subdirectory(unittest/third_party/googletest) +endif() + +if(BUILD_TRAINING_TOOLS) + add_subdirectory(src/training) +endif() + +get_target_property(tesseract_NAME libtesseract NAME) +get_target_property(tesseract_VERSION libtesseract VERSION) +get_target_property(tesseract_OUTPUT_NAME libtesseract OUTPUT_NAME) + +configure_file(tesseract.pc.cmake ${CMAKE_CURRENT_BINARY_DIR}/tesseract.pc.in + @ONLY) +# to resolve generator expression in OUTPUT_NAME +file( + GENERATE + OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/tesseract_$<CONFIG>.pc + INPUT ${CMAKE_CURRENT_BINARY_DIR}/tesseract.pc.in) + +configure_package_config_file( + cmake/templates/TesseractConfig.cmake.in + ${CMAKE_CURRENT_BINARY_DIR}/cmake/tesseract/TesseractConfig.cmake + INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/tesseract + PATH_VARS INCLUDE_DIR LIBRARY_DIRS) +write_basic_package_version_file( + ${CMAKE_CURRENT_BINARY_DIR}/cmake/tesseract/TesseractConfigVersion.cmake + VERSION ${PACKAGE_VERSION} + COMPATIBILITY SameMajorVersion) + +install( + FILES ${CMAKE_CURRENT_BINARY_DIR}/tesseract_$<CONFIG>.pc + DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig + RENAME tesseract.pc) +install(TARGETS tesseract DESTINATION bin) +if (MSVC) + install(FILES $<TARGET_PDB_FILE:${PROJECT_NAME}> DESTINATION bin OPTIONAL) +endif() +install( + TARGETS libtesseract + EXPORT TesseractTargets + RUNTIME DESTINATION bin + RUNTIME DESTINATION bin + LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR} + ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}) +if (MSVC AND BUILD_SHARED_LIBS) + install(FILES $<TARGET_PDB_FILE:libtesseract> DESTINATION bin OPTIONAL) +endif() +install( + EXPORT TesseractTargets + NAMESPACE Tesseract:: + DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/tesseract) +install(DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/cmake + DESTINATION ${CMAKE_INSTALL_LIBDIR}) + +install( + FILES include/tesseract/baseapi.h + include/tesseract/capi.h + include/tesseract/renderer.h + ${CMAKE_CURRENT_BINARY_DIR}/include/tesseract/version.h + include/tesseract/ltrresultiterator.h + include/tesseract/pageiterator.h + include/tesseract/resultiterator.h + include/tesseract/osdetect.h + include/tesseract/publictypes.h + include/tesseract/ocrclass.h + include/tesseract/export.h + include/tesseract/unichar.h + # ${CMAKE_CURRENT_BINARY_DIR}/src/endianness.h + DESTINATION include/tesseract) + +if(INSTALL_CONFIGS) + install(FILES ${TESSERACT_CONFIGS} + DESTINATION ${CMAKE_INSTALL_DATAROOTDIR}/tessdata/configs) + install(FILES ${TESSERACT_TESSCONFIGS} + DESTINATION ${CMAKE_INSTALL_DATAROOTDIR}/tessdata/tessconfigs) +endif() + +# ############################################################################## +# uninstall target +# ############################################################################## +if(NOT TARGET uninstall) + configure_file( + "${CMAKE_CURRENT_SOURCE_DIR}/cmake/templates/cmake_uninstall.cmake.in" + "${CMAKE_CURRENT_BINARY_DIR}/cmake_uninstall.cmake" IMMEDIATE @ONLY) + + add_custom_target( + uninstall + COMMENT "Uninstall installed files" + COMMAND ${CMAKE_COMMAND} -P + ${CMAKE_CURRENT_BINARY_DIR}/cmake_uninstall.cmake) +endif() + +# ##############################################################################
