[tesseract] Update and fix feature training-tools build (#29225)

Co-authored-by: Billy Robert O'Neal III <bion@microsoft.com>

ports/tesseract/fix-debug-postfix.patch

@@ -0,0 +1,33 @@
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 8c6845cb..27c1f4a2 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -812,14 +812,8 @@ set_target_properties(libtesseract
+ set_target_properties(libtesseract
+   PROPERTIES SOVERSION ${VERSION_MAJOR}.${VERSION_MINOR}.${VERSION_PATCH})
+ 
+-if(WIN32)
+-  set_target_properties(libtesseract
+-     PROPERTIES OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR})
+-  set_target_properties(libtesseract
+-     PROPERTIES DEBUG_OUTPUT_NAME tesseract${VERSION_MAJOR}${VERSION_MINOR}d)
+-else()
+-  set_target_properties(libtesseract PROPERTIES OUTPUT_NAME tesseract)
+-endif()
++set_target_properties(libtesseract
++  PROPERTIES OUTPUT_NAME tesseract$<$<BOOL:${WIN32}>:${VERSION_MAJOR}${VERSION_MINOR}$<$<CONFIG:DEBUG>:d>>)
+ 
+ if(SW_BUILD)
+   target_link_libraries(libtesseract PUBLIC org.sw.demo.danbloomberg.leptonica
+@@ -884,8 +878,10 @@ 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
++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.pc INPUT ${CMAKE_CURRENT_BINARY_DIR}/tesseract.pc.in)
+ 
+ configure_package_config_file(
+   cmake/templates/TesseractConfig.cmake.in

ports/tesseract/fix-tools.patch

@@ -0,0 +1,31 @@
+diff --git a/src/training/CMakeLists.txt b/src/training/CMakeLists.txt
+index a05b216c..faab9bc2 100644
+--- a/src/training/CMakeLists.txt
++++ b/src/training/CMakeLists.txt
+@@ -259,13 +259,8 @@ if(ICU_FOUND)
+     target_link_libraries(unicharset_training
+                           PUBLIC common_training org.sw.demo.unicode.icu.i18n)
+   else()
+-    if(${CMAKE_VERSION} VERSION_LESS "3.12.0")
+       target_link_libraries(unicharset_training PUBLIC common_training
+                                                        PkgConfig::ICU)
+-    else()
+-      target_link_libraries(unicharset_training PUBLIC common_training
+-                                                       ${ICU_LIBRARIES})
+-    endif()
+   endif()
+   target_include_directories(unicharset_training
+                              PUBLIC unicharset ${CMAKE_CURRENT_BINARY_DIR})
+@@ -388,11 +383,7 @@ if(ICU_FOUND)
+         target_include_directories(pango_training BEFORE
+                                    PUBLIC ${PANGO_INCLUDE_DIRS})
+         target_compile_definitions(pango_training PUBLIC -DPANGO_ENABLE_ENGINE)
+-        if(${CMAKE_VERSION} VERSION_LESS "3.12.0")
+-          target_link_libraries(pango_training PUBLIC PkgConfig::PANGO)
+-        else()
+-          target_link_libraries(pango_training PUBLIC ${PANGO_LINK_LIBRARIES})
+-        endif()
++        target_link_libraries(pango_training PUBLIC PkgConfig::PANGO)
+       endif()
+     endif()
+     target_include_directories(pango_training

ports/tesseract/portfile.cmake

@@ -5,42 +5,32 @@ endif()
 vcpkg_from_github(
     OUT_SOURCE_PATH SOURCE_PATH
     REPO tesseract-ocr/tesseract
-    REF 5ad5325a0aa8effc47ca033625b6a51682f82767 #v5.2.0
-    SHA512 c6ed442c9deb28772aeb918142dab08d5b55eeeeccb0c1d3f13cf51bb72af227afb7f14c19a5c8db40d6a7b8cfeccb3af08a78adfcd7431e4a06f65372709ceb
-    PATCHES ${tesseract_patch}
+    REF 080da83cc51c4ef8b324a7e03146fe0bd7e0944b #5.3.0
+    SHA512 77f7e69ca220edb51f0d1e21fae67288759bbefb6868203cd095c4457b16d7319d78cd47dd8e72be3da5aabb357f5f649b8da7fc3f2263faedecf10f556eb431
+    PATCHES
+        ${tesseract_patch}
+        fix-tools.patch # See https://github.com/tesseract-ocr/tesseract/pull/4006
+        fix-debug-postfix.patch # See https://github.com/tesseract-ocr/tesseract/pull/4008
 )
 
-# The built-in cmake FindICU is better
-file(REMOVE "${SOURCE_PATH}/cmake/FindICU.cmake")
-
-string(COMPARE EQUAL "${VCPKG_LIBRARY_LINKAGE}" "static" BUILD_STATIC)
-
 vcpkg_check_features(OUT_FEATURE_OPTIONS FEATURE_OPTIONS
     FEATURES
         training-tools  BUILD_TRAINING_TOOLS
 )
 
-if("cpu-independed" IN_LIST FEATURES)
-    set(TARGET_ARCHITECTURE none)
-else()
-    set(TARGET_ARCHITECTURE auto)
-endif()
-
 vcpkg_cmake_configure(
     SOURCE_PATH "${SOURCE_PATH}"
     OPTIONS
         ${FEATURE_OPTIONS}
-        -DSTATIC=${BUILD_STATIC}
         -DUSE_SYSTEM_ICU=True
-        -DCMAKE_DISABLE_FIND_PACKAGE_LibArchive=OFF
+        -DCMAKE_REQUIRE_FIND_PACKAGE_LibArchive=ON
+        -DCMAKE_REQUIRE_FIND_PACKAGE_CURL=ON
+        -DCMAKE_REQUIRE_FIND_PACKAGE_Leptonica=ON
         -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()
@@ -58,10 +48,6 @@ 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" "-ltesseract52" "-ltesseract52d")
-endif()
 vcpkg_fixup_pkgconfig()
 
 if("training-tools" IN_LIST FEATURES)
@@ -70,6 +56,7 @@ if("training-tools" IN_LIST FEATURES)
         cntraining dawg2wordlist mftraining shapeclustering
         wordlist2dawg combine_lang_model lstmeval lstmtraining
         set_unicharset_properties unicharset_extractor text2image
+        merge_unicharsets
     )
     vcpkg_copy_tools(TOOL_NAMES ${TRAINING_TOOLS} AUTO_CLEAN)
 endif()

ports/tesseract/vcpkg.json

@@ -1,15 +1,20 @@
 {
   "name": "tesseract",
-  "version": "5.2.0",
-  "port-version": 3,
+  "version": "5.3.0",
   "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": "!uwp",
   "dependencies": [
-    "curl",
+    {
+      "name": "curl",
+      "default-features": false
+    },
     "leptonica",
-    "libarchive",
+    {
+      "name": "libarchive",
+      "default-features": false
+    },
     {
       "name": "vcpkg-cmake",
       "host": true
@@ -20,9 +25,6 @@
     }
   ],
   "features": {
-    "cpu-independed": {
-      "description": "build on any cpu extension commands support"
-    },
     "training-tools": {
       "description": "build training tools",
       "dependencies": [

versions/baseline.json

@@ -7573,8 +7573,8 @@
       "port-version": 0
     },
     "tesseract": {
-      "baseline": "5.2.0",
-      "port-version": 3
+      "baseline": "5.3.0",
+      "port-version": 0
     },
     "tfhe": {
       "baseline": "1.0.1",

versions/t-/tesseract.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "9536858220896569da2d0e66fa3f6a4b95621186",
+      "version": "5.3.0",
+      "port-version": 0
+    },
     {
       "git-tree": "106efd2adc79ebdcf63488c2cc930b035f01535d",
       "version": "5.2.0",