Patch sentencepiece to enable iOS build. (#335)

cmake/externals/sentencepieceproject.cmake

@@ -1,27 +1,36 @@
+# spm is abbreviation of sentencepiece to meet the path length limits on Windows
+
+set(spm_patches)
+if(_ONNXRUNTIME_EMBEDDED)
+  # use protobuf provided by ORT
+  list(APPEND spm_patches "${PROJECT_SOURCE_DIR}/cmake/externals/sentencepieceproject.protobuf.patch")
+endif()
+if(IOS)
+  # fix for iOS build
+  list(APPEND spm_patches "${PROJECT_SOURCE_DIR}/cmake/externals/sentencepieceproject.ios.patch")
+endif()
+
+set(spm_patch_command)
+if(spm_patches)
+  set(spm_patch_command git checkout . && git apply --ignore-space-change --ignore-whitespace ${spm_patches})
+endif()
+
 FetchContent_Declare(
   spm
   GIT_REPOSITORY https://github.com/google/sentencepiece.git
   GIT_TAG v0.1.96
+  PATCH_COMMAND ${spm_patch_command}
 )
-# spm is abbr. of sentencepiece to meet the MAX_PATH compiling requirement on Windows
 FetchContent_GetProperties(spm)
 
 if(NOT spm_POPULATED)
   FetchContent_Populate(spm)
-  # need to patch sentencepiece to use protobuf provided by ort. This is ifdef onnxruntime_BUILD_WEBASSEMBLY
-  # for now but there is no real reason to not use it for all builds.
-  # 'git apply' should be in FetchContent_Declare() but that creates issues in sucessive builds so for now it is using execute_process().
-  if(onnxruntime_BUILD_WEBASSEMBLY)
-    set(SPM_PATCH "${PROJECT_SOURCE_DIR}/cmake/externals/sentencepieceproject.patch")
-    message("-- sentencepiece: patching with ${SPM_PATCH}")
-    execute_process(COMMAND git apply --ignore-space-change --ignore-whitespace ${SPM_PATCH} WORKING_DIRECTORY ${spm_SOURCE_DIR})
-  endif()
   add_subdirectory(${spm_SOURCE_DIR} ${spm_BINARY_DIR} EXCLUDE_FROM_ALL)
   set_target_properties(sentencepiece-static PROPERTIES
     FOLDER externals/google/sentencepiece)
 endif()
 
-if (onnxruntime_BUILD_WEBASSEMBLY)
+if(_ONNXRUNTIME_EMBEDDED)
   set(SPM_USE_BUILTIN_PROTOBUF OFF)
   set(spm_INCLUDE_DIRS
     ${REPO_ROOT}/cmake/external/protobuf/src

cmake/externals/sentencepieceproject.ios.patch

@@ -0,0 +1,28 @@
+diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
+index 8b7fb76..c08b322 100644
+--- a/src/CMakeLists.txt
++++ b/src/CMakeLists.txt
+@@ -285,10 +285,19 @@ endif()
+ list(APPEND SPM_INSTALLTARGETS
+   spm_encode spm_decode spm_normalize spm_train spm_export_vocab)
+ 
+-install(TARGETS ${SPM_INSTALLTARGETS}
+-  RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+-  LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+-  ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR})
++if (CMAKE_SYSTEM_NAME STREQUAL "iOS")
++  install(TARGETS ${SPM_INSTALLTARGETS}
++    BUNDLE DESTINATION ${CMAKE_INSTALL_BINDIR}
++    RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
++    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
++    ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR})
++else()
++  install(TARGETS ${SPM_INSTALLTARGETS}
++    RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
++    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
++    ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR})
++endif()
++
+ install(FILES sentencepiece_trainer.h sentencepiece_processor.h
+   DESTINATION ${CMAKE_INSTALL_INCDIR})
+ if (NOT SPM_USE_BUILTIN_PROTOBUF)

tools/ios/build_xcframework.py

@@ -63,11 +63,6 @@ def build_framework_for_platform_and_arch(
         f"-DCMAKE_OSX_ARCHITECTURES={arch}",
         f"-DCMAKE_XCODE_ATTRIBUTE_CODE_SIGNING_ALLOWED=NO",
         "-DOCOS_BUILD_APPLE_FRAMEWORK=ON",
-        # our version of sentencepiece doesn't support iOS build
-        #   CMake Error at out/ios/RelWithDebInfo/_deps/spm-src/src/CMakeLists.txt:288 (install):
-        #     install TARGETS given no BUNDLE DESTINATION for MACOSX_BUNDLE executable
-        #     target "spm_encode".
-        "-DOCOS_ENABLE_SPM_TOKENIZER=OFF",
         # use OpenCV's CMake toolchain file
         f"-DCMAKE_TOOLCHAIN_FILE={_get_opencv_toolchain_file(platform, opencv_dir)}",
         # required by OpenCV CMake toolchain file