[tesseract] port update to 4.1.0 release (#7227)

6 files changed, 65 insertions, 44 deletions

diff --git a/ports/tesseract/CONTROL b/ports/tesseract/CONTROL
index df6e60fec..3f04dcf96 100644
--- a/ports/tesseract/CONTROL
+++ b/ports/tesseract/CONTROL
@@ -1,5 +1,12 @@
 Source: tesseract

-Version: 4.0.0-3

+Version: 4.1.0

 Homepage: https://github.com/tesseract-ocr/tesseract

 Description: An OCR Engine that was developed at HP Labs between 1985 and 1995... and now at Google.

-Build-Depends: leptonica, icu

+Build-Depends: leptonica

+

+Feature: training_tools

+Description: build training tools

+Build-Depends: icu

+

+Feature: cpu_independed

+Description: build on any cpu extension commands support

diff --git a/ports/tesseract/fix-tiff-linkage.patch b/ports/tesseract/fix-tiff-linkage.patch
new file mode 100644
index 000000000..f4720c022
--- /dev/null
+++ b/ports/tesseract/fix-tiff-linkage.patch
@@ -0,0 +1,13 @@
+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} ${tesseractmain_rsc})
+ target_link_libraries           (tesseract libtesseract)
+ if (HAVE_TIFFIO_H)
+-    target_link_libraries(tesseract tiff)
++    target_link_libraries(tesseract TIFF::TIFF)
+ endif()
+ 
+ if (OPENMP_BUILD AND UNIX)
diff --git a/ports/tesseract/leptonica.patch b/ports/tesseract/leptonica.patch
deleted file mode 100644
index 507721d9d..000000000
--- a/ports/tesseract/leptonica.patch
+++ /dev/null
@@ -1,13 +0,0 @@
-diff --git a/cmake/templates/TesseractConfig.cmake.in b/cmake/templates/TesseractConfig.cmake.in
-index 13838361..245ca2bc 100644
---- a/cmake/templates/TesseractConfig.cmake.in
-+++ b/cmake/templates/TesseractConfig.cmake.in
-@@ -21,6 +21,8 @@
- 
- include(${CMAKE_CURRENT_LIST_DIR}/TesseractTargets.cmake)
- 
-+find_package(Leptonica REQUIRED)
-+
- # ======================================================
- #  Version variables:
- # ======================================================
diff --git a/ports/tesseract/portfile.cmake b/ports/tesseract/portfile.cmake
index 9650db44a..dc03a69e7 100644
--- a/ports/tesseract/portfile.cmake
+++ b/ports/tesseract/portfile.cmake
@@ -5,24 +5,41 @@ vcpkg_check_linkage(ONLY_STATIC_LIBRARY)
 vcpkg_from_github(
     OUT_SOURCE_PATH SOURCE_PATH
     REPO tesseract-ocr/tesseract
-    REF 4.0.0
-    SHA512 69e57d4ba1fc43d212fd0fff69a2b5d48a3b37cfee7054fdc083cbb7e04d92317609a32e457229661d70ce8d9b16c9d25e81bfc3861db660dd2c8f292202d447
-    HEAD_REF master
+    REF 4.1.0
+    SHA512 d617f5c5b826640b2871dbe3d7973bcc5e66fafd837921a20e009d683806ed50f0f258aa455019d99fc54f5cb65c2fa0380e3a3c92b39ab0684b8799c730b09d
     PATCHES
-        use-vcpkg-icu.patch
-        ws2-32.patch
-        leptonica.patch
+        fix-tiff-linkage.patch
 )
 
 # The built-in cmake FindICU is better
 file(REMOVE ${SOURCE_PATH}/cmake/FindICU.cmake)
 
+# Handle Static Library Output
+if(VCPKG_LIBRARY_LINKAGE EQUAL "static")
+    list(APPEND OPTIONS_LIST -DSTATIC=ON)
+endif()
+
+# Handle CONTROL
+if("training_tools" IN_LIST FEATURES)
+    list(APPEND OPTIONS_LIST -DBUILD_TRAINING_TOOLS=ON)
+else()
+    list(APPEND OPTIONS_LIST -DBUILD_TRAINING_TOOLS=OFF)
+endif()
+if("independed_architecture" IN_LIST FEATURES)
+    list(APPEND OPTIONS_LIST -DTARGET_ARCHITECTURE=none)
+else()
+    list(APPEND OPTIONS_LIST -DTARGET_ARCHITECTURE=auto)
+endif()
+
 vcpkg_configure_cmake(
     SOURCE_PATH ${SOURCE_PATH}
     PREFER_NINJA
     OPTIONS
         -DSTATIC=ON
         -DUSE_SYSTEM_ICU=True
+	#any value for vcpkg leptonica link cmake branch select
+	-DLeptonica_DIR=YES
+	${OPTIONS_LIST}
 )
 
 vcpkg_install_cmake()
@@ -35,6 +52,26 @@ set(EXTENSION)
 if(WIN32)
     set(EXTENSION ".exe")
 endif()
+
+# copy training tools
+set(TRAINING_TOOLS_DIR ${CURRENT_PACKAGES_DIR}/tools/tesseract/training)
+if("training_tools" IN_LIST FEATURES)
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/ambiguous_words${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/classifier_tester${EXTENSION} 	DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/combine_tessdata${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/cntraining${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/dawg2wordlist${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/mftraining${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/shapeclustering${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/wordlist2dawg${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/combine_lang_model${EXTENSION} 	DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/lstmeval${EXTENSION} 			DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/lstmtraining${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/set_unicharset_properties${EXTENSION} DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/unicharset_extractor${EXTENSION} 	DESTINATION ${TRAINING_TOOLS_DIR})
+    file(COPY ${CURRENT_PACKAGES_DIR}/bin/text2image${EXTENSION} 		DESTINATION ${TRAINING_TOOLS_DIR})
+endif()
+
 file(COPY ${CURRENT_PACKAGES_DIR}/bin/tesseract${EXTENSION} DESTINATION ${CURRENT_PACKAGES_DIR}/tools/tesseract)
 vcpkg_copy_tool_dependencies(${CURRENT_PACKAGES_DIR}/tools/tesseract)
 
diff --git a/ports/tesseract/use-vcpkg-icu.patch b/ports/tesseract/use-vcpkg-icu.patch
deleted file mode 100644
index 9c75726e1..000000000
--- a/ports/tesseract/use-vcpkg-icu.patch
+++ /dev/null
@@ -1,11 +0,0 @@
---- 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)
diff --git a/ports/tesseract/ws2-32.patch b/ports/tesseract/ws2-32.patch
deleted file mode 100644
index 1d3be8189..000000000
--- a/ports/tesseract/ws2-32.patch
+++ /dev/null
@@ -1,12 +0,0 @@
-diff --git a/viewer/svutil.cpp b/viewer/svutil.cpp

-index 34a2286..bc2a7e1 100644

---- a/viewer/svutil.cpp

-+++ b/viewer/svutil.cpp

-@@ -23,6 +23,7 @@

- #include <stdio.h>

- #ifdef _WIN32

- #include <windows.h>

-+#pragma comment(lib, "Ws2_32.lib")

- struct addrinfo {

-   struct sockaddr* ai_addr;

-   int ai_addrlen;