[mecab] Add new port (#7589)

* [mecab] Add new port

5 files changed, 126 insertions, 0 deletions

diff --git a/ports/mecab/CMakeLists.txt b/ports/mecab/CMakeLists.txt
new file mode 100644
index 000000000..06768b148
--- /dev/null
+++ b/ports/mecab/CMakeLists.txt
@@ -0,0 +1,47 @@
+cmake_minimum_required(VERSION 3.12 FATAL_ERROR)
+
+PROJECT(Mecab)
+                          
+file(GLOB SOURCE_FILE
+	"*.cpp"
+	"*.c"
+)
+file(GLOB HEADERS_FILE
+	"*.h"
+)
+
+message(STATUS "current cmake path: ${CMAKE_CURRENT_SOURCE_DIR}")
+list(REMOVE_ITEM SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/mecab-cost-train.cpp)
+list(REMOVE_ITEM SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/mecab-dict-gen.cpp)
+list(REMOVE_ITEM SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/mecab-dict-index.cpp)
+list(REMOVE_ITEM SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/mecab-system-eval.cpp)
+list(REMOVE_ITEM SOURCE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/mecab-test-gen.cpp)
+#list(REMOVE_ITEM SOURCE_FILE "mecab-cost-train.cpp" "mecab-dict-gen.cpp" "mecab-dict-index.cpp" "mecab-system-eval.cpp" "mecab-test-gen.cpp")
+
+add_library (Mecab ${SOURCE_FILE} ${HEADERS_FILE})
+
+if(UNIX)
+	target_compile_definitions(Mecab PUBLIC -DHAVE_UNISTD_H -DHAVE_FCNTL_H -DHAVE_STDINT_H -DHAVE_SYS_TYPES_H -DHAVE_SYS_STAT_H -DHAVE_DIRENT_H -DDIC_VERSION=102 -DVERSION="@VERSION@" -DPACKAGE="mecab" -DMECAB_DEFAULT_RC="./mecabrc")
+endif(UNIX)
+if(WIN32)
+	target_compile_definitions(Mecab PUBLIC -D_CRT_SECURE_NO_DEPRECATE -DMECAB_USE_THREAD -DDLL_EXPORT -DHAVE_GETENV -DHAVE_WINDOWS_H -DDIC_VERSION=102 -DVERSION="@VERSION@" -DPACKAGE="mecab" -DUNICODE -D_UNICODE -DMECAB_DEFAULT_RC="mecabrc")
+endif(WIN32)
+
+install(TARGETS Mecab
+    RUNTIME DESTINATION bin
+    ARCHIVE DESTINATION lib
+    LIBRARY DESTINATION lib
+)
+include(CMakePackageConfigHelpers)
+set(cmake_package_name Mecab)
+set(generated_dir "${CMAKE_CURRENT_BINARY_DIR}/generated" CACHE INTERNAL "")
+set(cmake_files_install_dir "${CMAKE_INSTALL_LIBDIR}/cmake/${cmake_package_name}")
+set(config_file "${generated_dir}/${cmake_package_name}Config.cmake")
+configure_package_config_file("${CMAKE_CURRENT_SOURCE_DIR}/Config.cmake.in"
+"${config_file}" INSTALL_DESTINATION ${cmake_files_install_dir})
+#install(FILES ${config_file} DESTINATION ${cmake_files_install_dir} CONFIGURATIONS Release)
+install(FILES ${config_file} DESTINATION share/mecab CONFIGURATIONS Release)
+message(STATUS "ccmake_files_install_dir: ${cmake_files_install_dir}")
+
+install(FILES ${HEADERS_FILE} DESTINATION include CONFIGURATIONS Release)
+file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/include)
\ No newline at end of file
diff --git a/ports/mecab/CONTROL b/ports/mecab/CONTROL
new file mode 100644
index 000000000..894d9f1fa
--- /dev/null
+++ b/ports/mecab/CONTROL
@@ -0,0 +1,3 @@
+Source: mecab

+Version:0.996

+Description: A morphological analysis engine based on CRF

diff --git a/ports/mecab/Config.cmake.in b/ports/mecab/Config.cmake.in
new file mode 100644
index 000000000..a18cb7d4b
--- /dev/null
+++ b/ports/mecab/Config.cmake.in
@@ -0,0 +1 @@
+@PACKAGE_INIT@
diff --git a/ports/mecab/fix_wpath_unsigned.patch b/ports/mecab/fix_wpath_unsigned.patch
new file mode 100644
index 000000000..d8dcf1509
--- /dev/null
+++ b/ports/mecab/fix_wpath_unsigned.patch
@@ -0,0 +1,40 @@
+diff --git a/mecab/src/common.h b/mecab/src/common.h
+index d1fc459..2951d5a 100644
+--- a/mecab/src/common.h
++++ b/mecab/src/common.h
+@@ -86,7 +86,8 @@
+ #define WPATH_FORCE(path) (MeCab::Utf8ToWide(path).c_str())
+ #define WPATH(path) (path)
+ #else
+-#define WPATH(path) WPATH_FORCE(path)
++#define WPATH_FORCE(path) (MeCab::Utf8ToWide(path).c_str())
++#define WPATH(path) (path)
+ #endif
+ #else
+ #define WPATH_FORCE(path) (path)
+diff --git a/mecab/src/feature_index.cpp b/mecab/src/feature_index.cpp
+index 051bdf8..fdd0145 100644
+--- a/mecab/src/feature_index.cpp
++++ b/mecab/src/feature_index.cpp
+@@ -353,7 +353,7 @@ bool FeatureIndex::buildUnigramFeature(LearnerPath *path,
+               if (!r) goto NEXT;
+               os_ << r;
+             } break;
+-            case 't':  os_ << (size_t)path->rnode->char_type;     break;
++            case 't':  os_ << (int)(size_t)path->rnode->char_type;     break;
+             case 'u':  os_ << ufeature; break;
+             case 'w':
+               if (path->rnode->stat == MECAB_NOR_NODE) {
+diff --git a/mecab/src/writer.cpp b/mecab/src/writer.cpp
+index 0ef6975..f068fa3 100644
+--- a/mecab/src/writer.cpp
++++ b/mecab/src/writer.cpp
+@@ -257,7 +257,7 @@ bool Writer::writeNode(Lattice *lattice,
+             // input sentence
+           case 'S': os->write(lattice->sentence(), lattice->size()); break;
+             // sentence length
+-          case 'L': *os << lattice->size(); break;
++          case 'L': *os << (int)lattice->size(); break;
+             // morph
+           case 'm': os->write(node->surface, node->length); break;
+           case 'M': os->write(reinterpret_cast<const char *>
diff --git a/ports/mecab/portfile.cmake b/ports/mecab/portfile.cmake
new file mode 100644
index 000000000..6383b2d91
--- /dev/null
+++ b/ports/mecab/portfile.cmake
@@ -0,0 +1,35 @@
+if (VCPKG_TARGET_TRIPLET STREQUAL "x64-uwp" OR VCPKG_TARGET_TRIPLET STREQUAL "arm64-windows" OR VCPKG_TARGET_TRIPLET STREQUAL "arm-uwp")

+    message(FATAL_ERROR "mecab does not support on this platform")

+endif()

+

+include(vcpkg_common_functions)

+

+set(MECAB_VERSION 0.996)

+vcpkg_from_github(

+	OUT_SOURCE_PATH SOURCE_PATH

+	REPO taku910/mecab

+	REF master

+	SHA512 2a7f1d159ddca846357b5bcab2d2b5de2e6a27dca4301cdd1cc52c155c352f9c7030b77d1187afe9c0a7f1b131a1acdcc40ee81ce7ba5c0fa6b2325c56676353

+	HEAD_REF master

+	PATCHES

+		fix_wpath_unsigned.patch

+)

+

+message(STATUS "source path is : ${SOURCE_PATH}")

+

+file(COPY ${CMAKE_CURRENT_LIST_DIR}/CMakeLists.txt DESTINATION ${SOURCE_PATH}/mecab/src)

+file(COPY ${CMAKE_CURRENT_LIST_DIR}/Config.cmake.in DESTINATION ${SOURCE_PATH}/mecab/src)

+file(COPY ${SOURCE_PATH}/mecab/COPYING DESTINATION ${SOURCE_PATH}/mecab/src)

+

+message(STATUS "CURRENT_PACKAGES_DIR is : ${CURRENT_PACKAGES_DIR}")

+vcpkg_configure_cmake(

+    SOURCE_PATH ${SOURCE_PATH}/mecab/src

+)

+

+vcpkg_install_cmake()

+vcpkg_copy_pdbs()

+file(COPY ${SOURCE_PATH}/mecab/COPYING DESTINATION ${CURRENT_PACKAGES_DIR}/share/mecab)

+file(RENAME ${CURRENT_PACKAGES_DIR}/share/mecab/COPYING ${CURRENT_PACKAGES_DIR}/share/mecab/copyright)

+

+# Post-build test for cmake libraries

+# vcpkg_test_cmake(PACKAGE_NAME mecab)