make spm build for static ort-web build (#293)

cmake/externals/sentencepieceproject.cmake

@@ -8,14 +8,32 @@ 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()
 
-set(spm_INCLUDE_DIRS
-    ${spm_SOURCE_DIR}/third_party/protobuf-lite
+if (onnxruntime_BUILD_WEBASSEMBLY)
+  set(SPM_USE_BUILTIN_PROTOBUF OFF)
+  set(spm_INCLUDE_DIRS
+    ${REPO_ROOT}/cmake/external/protobuf/src
     ${spm_SOURCE_DIR}/src/builtin_pb
     ${spm_SOURCE_DIR}/third_party
     ${spm_SOURCE_DIR}/src
-    )
+)
+else()
+  set(spm_INCLUDE_DIRS
+      ${spm_SOURCE_DIR}/third_party/protobuf-lite
+      ${spm_SOURCE_DIR}/src/builtin_pb
+      ${spm_SOURCE_DIR}/third_party
+      ${spm_SOURCE_DIR}/src
+      )
+endif()

cmake/externals/sentencepieceproject.patch

@@ -0,0 +1,41 @@
+diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
+index 8b7fb76..3aa0f1f 100644
+--- a/src/CMakeLists.txt
++++ b/src/CMakeLists.txt
+@@ -69,12 +69,23 @@ if (SPM_USE_BUILTIN_PROTOBUF)
+   include_directories(${CMAKE_CURRENT_SOURCE_DIR}/../third_party/protobuf-lite)
+   include_directories(builtin_pb)
+ else()
+-  find_package(Protobuf REQUIRED)
++  include_directories(${REPO_ROOT}/cmake/external/protobuf/src)
++  if(EXISTS "${ONNX_CUSTOM_PROTOC_EXECUTABLE}")
++    set(PROTOC_EXECUTABLE ${ONNX_CUSTOM_PROTOC_EXECUTABLE})
++  else()
++    set(PROTOC_EXECUTABLE ${CMAKE_BINARY_DIR}/external/protobuf/cmake/protoc)
++  endif()
++  execute_process(COMMAND ${PROTOC_EXECUTABLE} --cpp_out=${CMAKE_CURRENT_SOURCE_DIR}/builtin_pb --proto_path=${CMAKE_CURRENT_SOURCE_DIR} sentencepiece_model.proto)
++  execute_process(COMMAND ${PROTOC_EXECUTABLE} --cpp_out=${CMAKE_CURRENT_SOURCE_DIR}/builtin_pb --proto_path=${CMAKE_CURRENT_SOURCE_DIR} sentencepiece.proto)
+   include_directories(${Protobuf_INCLUDE_DIRS})
+-  protobuf_generate_cpp(SPM_PROTO_SRCS SPM_PROTO_HDRS sentencepiece.proto)
+-  protobuf_generate_cpp(SPM_MODEL_PROTO_SRCS SPM_MODEL_PROTO_HDRS sentencepiece_model.proto)
+   set(PROTOBUF_LITE_SRCS "")
++  set(PROTOBUF_LITE_LIBRARY "")
+   include_directories(${PROTOBUF_INCLUDE_DIR})
++  include_directories(builtin_pb)
++  set(SPM_PROTO_HDRS builtin_pb/sentencepiece.pb.h)
++  set(SPM_PROTO_SRCS builtin_pb/sentencepiece.pb.cc)
++  set(SPM_MODEL_PROTO_HDRS builtin_pb/sentencepiece_model.pb.h)
++  set(SPM_MODEL_PROTO_SRCS builtin_pb/sentencepiece_model.pb.cc)
+ endif()
+ 
+ include_directories(${CMAKE_CURRENT_BINARY_DIR})
+diff --git a/third_party/CMakeLists.txt b/third_party/CMakeLists.txt
+index d00ecba..b355ccc 100644
+--- a/third_party/CMakeLists.txt
++++ b/third_party/CMakeLists.txt
+@@ -1,4 +1,4 @@
+-include_directories(absl/strings darts_clone esaxx protobuf-lite)
++include_directories(absl/strings darts_clone esaxx)
+ 
+ 
+