[libmupdf] add ocr feature (#32084)

* add libmupdf ocr feature WIP * Fix compile defs * Fix ocr feature * Add version * increment libmupdf port version * add libmupdf version * revert version change
microsoft · Jul 5, 2023 · d06a922 · d06a922
1 parent dad2e6f
commit d06a922
Show file tree

Hide file tree

Showing 5 changed files with 53 additions and 4 deletions.
diff --git a/ports/libmupdf/CMakeLists.txt b/ports/libmupdf/CMakeLists.txt
@@ -1,5 +1,7 @@
 cmake_minimum_required(VERSION 3.9)
-project(libmupdf C)
+project(libmupdf)
+
+option(ENABLE_OCR "Build with OCR" OFF)
 
 set(CMAKE_DEBUG_POSTFIX d)
 
@@ -25,18 +27,35 @@ find_package(ZLIB REQUIRED)
 find_package(OpenJPEG CONFIG REQUIRED)
 find_library(JBIG2DEC_LIB NAMES jbig2decd jbig2dec)
 
-file(GLOB_RECURSE SOURCES "source/*.c" "generated/*.c")
+if (ENABLE_OCR)
+  find_package(Leptonica CONFIG REQUIRED)
+  find_package(Tesseract CONFIG REQUIRED)
+endif()
+
+file(GLOB_RECURSE SOURCES "source/*.c" "generated/*.c" "source/*.h")
 list(FILTER SOURCES EXCLUDE REGEX "source/tools/[a-z]*\\.c$")
 list(FILTER SOURCES EXCLUDE REGEX "source/tests/.*.c$")
 list(FILTER SOURCES EXCLUDE REGEX "source/fitz/output-docx.c")
 
+if (ENABLE_OCR)
+  list(APPEND SOURCES "source/fitz/tessocr.cpp")
+endif()
+
 add_library(libmupdf ${SOURCES})
 
 if(WIN32)
   target_compile_definitions(libmupdf PRIVATE -DSHARE_JPEG -DFZ_ENABLE_JS=0 -DFZ_ENABLE_ICC=0 -DMEMENTO_MUPDF_HACKS -DFZ_ENABLE_DOCX_OUTPUT=0 -DFZ_ENABLE_ODT_OUTPUT=0)
 else()
   target_compile_definitions(libmupdf PRIVATE -DHAVE_PTHREAD=1 -DSHARE_JPEG -DFZ_ENABLE_JS=0 -DFZ_ENABLE_ICC=0 -DMEMENTO_MUPDF_HACKS -DFZ_ENABLE_DOCX_OUTPUT=0 -DFZ_ENABLE_ODT_OUTPUT=0)
 endif()
+
+
+if (ENABLE_OCR)
+  target_compile_definitions(libmupdf PRIVATE -DHAVE_TESSERACT=1 -DHAVE_LEPTONICA=1)
+else()
+  target_compile_definitions(libmupdf PRIVATE -DOCR_DISABLED=1)
+endif()
+
 target_include_directories(libmupdf
   PUBLIC
     include
@@ -45,6 +64,14 @@ target_include_directories(libmupdf
     ${JPEG_INCLUDE_DIR}
     ${HARFBUZZ_INCLUDE}
 )
+
+if (ENABLE_OCR)
+target_include_directories(libmupdf
+  PRIVATE
+    ${Leptonica_INCLUDE_DIRS}
+)
+endif()
+
 target_link_libraries(libmupdf PRIVATE
   openjp2
   freetype

diff --git a/ports/libmupdf/portfile.cmake b/ports/libmupdf/portfile.cmake
@@ -12,11 +12,18 @@ vcpkg_from_github(
 
 file(COPY "${CMAKE_CURRENT_LIST_DIR}/CMakeLists.txt" DESTINATION "${SOURCE_PATH}")
 
+vcpkg_check_features(
+    OUT_FEATURE_OPTIONS FEATURE_OPTIONS
+    FEATURES
+        ocr ENABLE_OCR
+)
+
 vcpkg_cmake_configure(
     SOURCE_PATH "${SOURCE_PATH}"
     DISABLE_PARALLEL_CONFIGURE
     OPTIONS
         -DBUILD_EXAMPLES=OFF
+	${FEATURE_OPTIONS}
 )
 
 vcpkg_cmake_install()

diff --git a/ports/libmupdf/vcpkg.json b/ports/libmupdf/vcpkg.json
@@ -1,6 +1,7 @@
 {
   "name": "libmupdf",
   "version": "1.22.1",
+  "port-version": 1,
   "description": "a lightweight PDF, XPS, and E-book library",
   "homepage": "https://github.com/ArtifexSoftware/mupdf",
   "license": "AGPL-3.0-only",
@@ -19,5 +20,14 @@
       "host": true
     },
     "zlib"
-  ]
+  ],
+  "features": {
+    "ocr": {
+      "description": "Enable OCR support",
+      "dependencies": [
+        "leptonica",
+        "tesseract"
+      ]
+    }
+  }
 }
diff --git a/versions/baseline.json b/versions/baseline.json
@@ -4330,7 +4330,7 @@
     },
     "libmupdf": {
       "baseline": "1.22.1",
-      "port-version": 0
+      "port-version": 1
     },
     "libmysql": {
       "baseline": "8.0.32",

diff --git a/versions/l-/libmupdf.json b/versions/l-/libmupdf.json
@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "06c0e6bccfaf2ab9f729bfc79318da68cb4897dc",
+      "version": "1.22.1",
+      "port-version": 1
+    },
     {
       "git-tree": "c1684ff15eed2c86ba792de4e92d21b2ea58c188",
       "version": "1.22.1",