Merge pull request #1657 from KindDragon/tesseract

Tesseract-OCR library

6 files changed, 138 insertions, 0 deletions

diff --git a/ports/leptonica/CONTROL b/ports/leptonica/CONTROL
new file mode 100644
index 000000000..d0e305c7b
--- /dev/null
+++ b/ports/leptonica/CONTROL
@@ -0,0 +1,4 @@
+Source: leptonica
+Version: 1.74.4
+Description: An open source library containing software that is broadly useful for image processing and image analysis applications
+Build-Depends: libjpeg-turbo, zlib, libpng, tiff, giflib
diff --git a/ports/leptonica/fix-cmakelists.patch b/ports/leptonica/fix-cmakelists.patch
new file mode 100644
index 000000000..18eb8cde5
--- /dev/null
+++ b/ports/leptonica/fix-cmakelists.patch
@@ -0,0 +1,34 @@
+--- a/src/CMakeLists.txt	Sun Jun 11 23:04:50 2017
++++ b/src/CMakeLists.txt	Wed Aug 16 17:29:56 2017
+@@ -27,24 +27,31 @@
+ endif()
+ 
+ if (GIF_LIBRARY)
++    target_include_directories  (leptonica PUBLIC ${GIF_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${GIF_LIBRARY})
+ endif()
+ if (JPEG_LIBRARY)
++    target_include_directories  (leptonica PUBLIC ${JPEG_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${JPEG_LIBRARY})
+ endif()
+ if (JP2K_FOUND)
++    target_include_directories  (leptonica PUBLIC ${JP2K_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${JP2K_LIBRARIES})
+ endif()
+ if (PNG_LIBRARY)
++    target_include_directories  (leptonica PUBLIC ${PNG_INCLUDE_DIRS})
+     target_link_libraries       (leptonica ${PNG_LIBRARY})
+ endif()
+ if (TIFF_LIBRARY)
++    target_include_directories  (leptonica PUBLIC ${TIFF_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${TIFF_LIBRARY})
+ endif()
+ if (WEBP_FOUND)
++    target_include_directories  (leptonica PUBLIC ${WEBP_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${WEBP_LIBRARIES})
+ endif()
+ if (ZLIB_LIBRARY)
++    target_include_directories  (leptonica PUBLIC ${ZLIB_INCLUDE_DIR})
+     target_link_libraries       (leptonica ${ZLIB_LIBRARY})
+ endif()
+ 
diff --git a/ports/leptonica/portfile.cmake b/ports/leptonica/portfile.cmake
new file mode 100644
index 000000000..a5d4330fd
--- /dev/null
+++ b/ports/leptonica/portfile.cmake
@@ -0,0 +1,36 @@
+include(vcpkg_common_functions)
+
+vcpkg_from_github(
+    OUT_SOURCE_PATH SOURCE_PATH
+    REPO DanBloomberg/leptonica
+    REF 1.74.4
+    SHA512 3b9d0be937883f733f72cbdf0b624ec245d9256a8b4622997f437d309efd7ad9695ad1cbe2224d543eb3ef8c44833567b3cc9a95e9a774ef9046b7acaf0ae744
+    HEAD_REF master
+)
+
+vcpkg_apply_patches(
+    SOURCE_PATH ${SOURCE_PATH}
+    PATCHES
+        ${CMAKE_CURRENT_LIST_DIR}/fix-cmakelists.patch
+)
+
+vcpkg_configure_cmake(
+    SOURCE_PATH ${SOURCE_PATH}
+    PREFER_NINJA
+    OPTIONS
+        -DCMAKE_REQUIRED_INCLUDES=${CURRENT_INSTALLED_DIR}/include # for check_include_file()
+)
+
+vcpkg_install_cmake()
+
+vcpkg_fixup_cmake_targets(CONFIG_PATH "cmake")
+
+vcpkg_copy_pdbs()
+
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/include)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/lib/pkgconfig)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/lib/pkgconfig)
+
+# Handle copyright
+file(COPY ${SOURCE_PATH}/leptonica-license.txt DESTINATION ${CURRENT_PACKAGES_DIR}/share/leptonica)
+file(RENAME ${CURRENT_PACKAGES_DIR}/share/leptonica/leptonica-license.txt ${CURRENT_PACKAGES_DIR}/share/leptonica/copyright)
diff --git a/ports/tesseract/CONTROL b/ports/tesseract/CONTROL
new file mode 100644
index 000000000..5bbc7b7e5
--- /dev/null
+++ b/ports/tesseract/CONTROL
@@ -0,0 +1,4 @@
+Source: tesseract
+Version: 3.05.01
+Description: An OCR Engine that was developed at HP Labs between 1985 and 1995... and now at Google.
+Build-Depends: leptonica, icu
diff --git a/ports/tesseract/portfile.cmake b/ports/tesseract/portfile.cmake
new file mode 100644
index 000000000..31ed2cd1f
--- /dev/null
+++ b/ports/tesseract/portfile.cmake
@@ -0,0 +1,49 @@
+include(vcpkg_common_functions)
+
+if (VCPKG_LIBRARY_LINKAGE STREQUAL dynamic)
+    message(STATUS "Warning: Dynamic building not supported yet. Building static.")
+    set(VCPKG_LIBRARY_LINKAGE static)
+endif()
+
+vcpkg_from_github(
+    OUT_SOURCE_PATH SOURCE_PATH
+    REPO tesseract-ocr/tesseract
+    REF 3.05.01
+    SHA512 a49c20c98386684cd89582e57b772811204fad8e5ff18214fb0da109f73629c70845054985e31e8deeb49107fbcf56e546aff661f08eb5dd60fbf83dbe976e81
+    HEAD_REF master
+)
+
+vcpkg_apply_patches(
+    SOURCE_PATH ${SOURCE_PATH}
+    PATCHES
+        ${CMAKE_CURRENT_LIST_DIR}/use-vcpkg-icu.patch
+)
+
+vcpkg_configure_cmake(
+    SOURCE_PATH ${SOURCE_PATH}
+    PREFER_NINJA
+    OPTIONS
+        -DSTATIC=ON
+        -DUSE_SYSTEM_ICU=True
+)
+
+vcpkg_install_cmake()
+
+vcpkg_fixup_cmake_targets(CONFIG_PATH "cmake")
+
+# Install tool
+file(MAKE_DIRECTORY ${CURRENT_PACKAGES_DIR}/tools/tesseract)
+file(COPY ${CURRENT_PACKAGES_DIR}/bin/tesseract.exe DESTINATION ${CURRENT_PACKAGES_DIR}/tools/tesseract)
+vcpkg_copy_tool_dependencies(${CURRENT_PACKAGES_DIR}/tools/tesseract)
+
+vcpkg_copy_pdbs()
+
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/include)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/bin)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/bin)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/lib/pkgconfig)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/lib/pkgconfig)
+
+# Handle copyright
+file(COPY ${SOURCE_PATH}/LICENSE DESTINATION ${CURRENT_PACKAGES_DIR}/share/tesseract)
+file(RENAME ${CURRENT_PACKAGES_DIR}/share/tesseract/LICENSE ${CURRENT_PACKAGES_DIR}/share/tesseract/copyright)
diff --git a/ports/tesseract/use-vcpkg-icu.patch b/ports/tesseract/use-vcpkg-icu.patch
new file mode 100644
index 000000000..9c75726e1
--- /dev/null
+++ b/ports/tesseract/use-vcpkg-icu.patch
@@ -0,0 +1,11 @@
+--- a/training/CMakeLists.txt	Thu Aug 17 02:46:15 2017
++++ b/training/CMakeLists.txt	Thu Aug 17 02:46:15 2017
+@@ -11,7 +11,7 @@
+ endif()
+ 
+ # experimental
+-if (MSVC AND NOT CPPAN_BUILD)
++if (MSVC AND NOT CPPAN_BUILD AND NOT USE_SYSTEM_ICU)
+ 
+ include(CheckTypeSize)
+ check_type_size("void *" SIZEOF_VOID_P)