[tesseract] update to 5.2.0 (#26114)

* [tesseract] update to 5.2.0

* x-add-version

* fix windows build error

* version

* fix opencv4 build error

* version

* conflict

* version

* fix opencv build error on uwp

* version

* fix linux build error

* version

* fix

* version

* version

ports/opencv4/0018-fix-depend-tesseract.patch

@@ -0,0 +1,13 @@
+diff --git a/modules/text/cmake/init.cmake b/modules/text/cmake/init.cmake
+index 254757e..73194ec 100644
+--- a/modules/text/cmake/init.cmake
++++ b/modules/text/cmake/init.cmake
+@@ -4,6 +4,8 @@ OCV_OPTION(WITH_TESSERACT "Include Tesseract OCR library support" (NOT CMAKE_CRO
+ if(NOT HAVE_TESSERACT
+     AND (WITH_TESSERACT OR OPENCV_FIND_TESSERACT)
+ )
++  find_package(Tesseract CONFIG REQUIRED)
++  set(Tesseract_LIBRARIES Tesseract::libtesseract)
+   if(NOT Tesseract_FOUND)
+     find_package(Tesseract QUIET)  # Prefer CMake's standard locations (including Tesseract_DIR)
+   endif()

ports/opencv4/portfile.cmake

@@ -177,6 +177,7 @@ if("contrib" IN_LIST FEATURES)
       0013-fix-ceres.patch
       0014-fix-ogre.patch
       0016-fix-freetype-contrib.patch
+      0018-fix-depend-tesseract.patch
   )
   set(BUILD_WITH_CONTRIB_FLAG "-DOPENCV_EXTRA_MODULES_PATH=${CONTRIB_SOURCE_PATH}/modules")
 
@@ -352,6 +353,12 @@ if("qt" IN_LIST FEATURES)
   list(APPEND ADDITIONAL_BUILD_FLAGS "-DCMAKE_AUTOMOC=ON")
 endif()
 
+if("contrib" IN_LIST FEATURES)
+  if(VCPKG_TARGET_IS_UWP)
+    list(APPEND ADDITIONAL_BUILD_FLAGS "-DWITH_TESSERACT=OFF")
+  endif()
+endif()
+
 vcpkg_cmake_configure(
     SOURCE_PATH "${SOURCE_PATH}"
     OPTIONS

ports/opencv4/vcpkg.json

@@ -1,6 +1,7 @@
 {
   "name": "opencv4",
   "version": "4.6.0",
+  "port-version": 1,
   "description": "computer vision library",
   "homepage": "https://github.com/opencv/opencv",
   "license": "Apache-2.0",

ports/tesseract/fix-depend-libarchive.patch

@@ -0,0 +1,14 @@
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index bd2649d..f932913 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -792,7 +792,8 @@ if(OpenCL_FOUND)
+   target_link_libraries(libtesseract PUBLIC OpenCL::OpenCL)
+ endif()
+ if(LibArchive_FOUND)
+-  target_link_libraries(libtesseract PUBLIC ${LibArchive_LIBRARIES})
++  find_package(LibArchive REQUIRED)
++  target_link_libraries(libtesseract PRIVATE LibArchive::LibArchive)
+ endif(LibArchive_FOUND)
+ if(CURL_FOUND)
+   if(NOT CURL_LIBRARIES)

ports/tesseract/fix-tiff-linkage.patch

@@ -1,13 +0,0 @@
-diff --git a/CMakeLists.txt b/CMakeLists.txt
-index f154663..c07b827 100644
---- a/CMakeLists.txt
-+++ b/CMakeLists.txt
-@@ -431,7 +431,7 @@ endif()
- add_executable                  (tesseract ${tesseractmain_src} ${tesseractmain_rsc})
- target_link_libraries           (tesseract libtesseract)
- if (HAVE_TIFFIO_H)
--    target_link_libraries(tesseract tiff)
-+    target_link_libraries(tesseract TIFF::TIFF)
- endif()
- 
- if (OPENMP_BUILD AND UNIX)

ports/tesseract/fix-timeval.patch

@@ -1,14 +0,0 @@
-diff --git a/src/ccutil/ocrclass.h b/src/ccutil/ocrclass.h
-index d39a6dd..3755632 100644
---- a/src/ccutil/ocrclass.h
-+++ b/src/ccutil/ocrclass.h
-@@ -28,6 +28,9 @@
- 
- #include <chrono>
- #include <ctime>
-+#ifndef _WIN32
-+#include <sys/time.h>
-+#endif
- #ifdef _WIN32
- #include <winsock2.h> // for timeval
- #endif

ports/tesseract/portfile.cmake

@@ -1,11 +1,13 @@
+if(NOT VCPKG_TARGET_IS_WINDOWS)
+    set(tesseract_patch fix-depend-libarchive.patch)
+endif()
+
 vcpkg_from_github(
     OUT_SOURCE_PATH SOURCE_PATH
     REPO tesseract-ocr/tesseract
-    REF 4.1.1
-    SHA512 017723a2268be789fe98978eed02fd294968cc8050dde376dee026f56f2b99df42db935049ae5e72c4519a920e263b40af1a6a40d9942e66608145b3131a71a2
-    PATCHES
-        fix-tiff-linkage.patch
-        fix-timeval.patch # Remove this patch in the next update
+    REF 5ad5325a0aa8effc47ca033625b6a51682f82767 #v5.2.0
+    SHA512 c6ed442c9deb28772aeb918142dab08d5b55eeeeccb0c1d3f13cf51bb72af227afb7f14c19a5c8db40d6a7b8cfeccb3af08a78adfcd7431e4a06f65372709ceb
+    PATCHES ${tesseract_patch}
 )
 
 # The built-in cmake FindICU is better
@@ -30,29 +32,34 @@ vcpkg_cmake_configure(
         ${FEATURE_OPTIONS}
         -DSTATIC=${BUILD_STATIC}
         -DUSE_SYSTEM_ICU=True
-        -DCMAKE_DISABLE_FIND_PACKAGE_LibArchive=ON
+        -DCMAKE_DISABLE_FIND_PACKAGE_LibArchive=OFF
         -DCMAKE_DISABLE_FIND_PACKAGE_OpenCL=ON
         -DLeptonica_DIR=YES
         -DTARGET_ARCHITECTURE=${TARGET_ARCHITECTURE}
+        -DSW_BUILD=OFF
+    MAYBE_UNUSED_VARIABLES
+        CMAKE_DISABLE_FIND_PACKAGE_OpenCL
+        STATIC
+        TARGET_ARCHITECTURE
 )
 
 vcpkg_cmake_install()
 vcpkg_copy_pdbs()
 
-vcpkg_cmake_config_fixup(CONFIG_PATH cmake)
+vcpkg_cmake_config_fixup(CONFIG_PATH lib/cmake/tesseract)
 
 vcpkg_replace_string("${CURRENT_PACKAGES_DIR}/share/tesseract/TesseractConfig.cmake"
-    "find_package(Leptonica REQUIRED)"
+    "find_dependency(Leptonica)"
 [[
-find_package(Leptonica REQUIRED)
-find_package(LibArchive REQUIRED)
+find_dependency(Leptonica)
+find_dependency(LibArchive)
 ]]
 )
 
 vcpkg_copy_tools(TOOL_NAMES tesseract AUTO_CLEAN)
 
 if(NOT VCPKG_BUILD_TYPE OR VCPKG_BUILD_TYPE STREQUAL "debug")
-    vcpkg_replace_string("${CURRENT_PACKAGES_DIR}/debug/lib/pkgconfig/tesseract.pc" "-ltesseract41" "-ltesseract41d")
+    vcpkg_replace_string("${CURRENT_PACKAGES_DIR}/debug/lib/pkgconfig/tesseract.pc" "-ltesseract52" "-ltesseract52d")
 endif()
 vcpkg_fixup_pkgconfig()
 
@@ -68,6 +75,7 @@ endif()
 
 
 file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/include")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/share")
 
 # Handle copyright
 file(INSTALL "${SOURCE_PATH}/LICENSE" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}" RENAME copyright)

ports/tesseract/vcpkg.json

@@ -1,9 +1,9 @@
 {
   "name": "tesseract",
-  "version": "4.1.1",
-  "port-version": 11,
+  "version": "5.2.0",
   "description": "An OCR Engine that was developed at HP Labs between 1985 and 1995... and now at Google.",
   "homepage": "https://github.com/tesseract-ocr/tesseract",
+  "license": "Apache-2.0",
   "supports": "!(arm & (osx | linux))",
   "dependencies": [
     "leptonica",

versions/baseline.json

@@ -5206,7 +5206,7 @@
     },
     "opencv4": {
       "baseline": "4.6.0",
-      "port-version": 0
+      "port-version": 1
     },
     "opendnp3": {
       "baseline": "3.1.1",
@@ -7045,8 +7045,8 @@
       "port-version": 0
     },
     "tesseract": {
-      "baseline": "4.1.1",
-      "port-version": 11
+      "baseline": "5.2.0",
+      "port-version": 0
     },
     "tfhe": {
       "baseline": "1.0.1",

versions/o-/opencv4.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "a6c9bded8fc4ad22bea6869caecf2881c8da2d41",
+      "version": "4.6.0",
+      "port-version": 1
+    },
     {
       "git-tree": "1ff543272c91a32d242ad5320e9ca63b29b34ed2",
       "version": "4.6.0",

versions/t-/tesseract.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "5ea8fcd243e8aa38e1a83eaa1c745fc53a3464cc",
+      "version": "5.2.0",
+      "port-version": 0
+    },
     {
       "git-tree": "24c99e73eaf335a9abedcfd42163c4968ca07ec3",
       "version": "4.1.1",