Merge branch 'main' into add-morphing

.github/workflows/build_and_deploy.yml

@@ -247,7 +247,7 @@ jobs:
       - name: Raplace resource
         if: inputs.is_production
         shell: bash
-        run: |
+        run:
           mv -f download/resource/core/README.md ./README.md
       - name: Install cargo-binstall
         uses: taiki-e/install-action@cargo-binstall
@@ -320,7 +320,7 @@ jobs:
           cp -v target/${{ matrix.target }}/release/libvoicevox_core_java_api.so java_artifact/ || true
       - name: Code signing (Windows)
         if: startsWith(matrix.os, 'windows') && inputs.code_signing
-        run: |
+        run:
           bash build_util/codesign.bash "artifact/${{ env.ASSET_NAME }}/voicevox_core.dll"
         env:
           ESIGNERCKA_USERNAME: ${{ secrets.ESIGNERCKA_USERNAME }}
@@ -366,44 +366,34 @@ jobs:
     needs: [config, build_and_deploy]
     runs-on: macos-12
     env:
+      IOS_X86_64_PATH: artifact/voicevox_core-x86_64-apple-ios
+      IOS_AARCH64_SIM_PATH: artifact/voicevox_core-aarch64-apple-ios-sim
+      IOS_AARCH64_PATH: artifact/voicevox_core-aarch64-apple-ios
       ASSET_NAME: voicevox_core-ios-xcframework-cpu-${{ needs.config.outputs.version }}
     steps:
       - uses: actions/checkout@v3
       - uses: actions/download-artifact@v2
         with:
           name: voicevox_core-x86_64-apple-ios
-          path: artifact/voicevox_core-x86_64-apple-ios
+          path: ${{ env.IOS_X86_64_PATH }}
       - uses: actions/download-artifact@v2
         with:
           name: voicevox_core-aarch64-apple-ios-sim
-          path: artifact/voicevox_core-aarch64-apple-ios-sim
+          path: ${{ env.IOS_AARCH64_SIM_PATH }}
       - uses: actions/download-artifact@v2
         with:
           name: voicevox_core-aarch64-apple-ios
-          path: artifact/voicevox_core-aarch64-apple-ios
-      - name: Create fat binary
+          path: ${{ env.IOS_AARCH64_PATH }}
+      - name: Create xcframework
+        id: create-xcframework 
         run: |
-          mkdir -p "artifact/voicevox_core-sim"
-          lipo -create "artifact/voicevox_core-x86_64-apple-ios/libvoicevox_core.dylib" "artifact/voicevox_core-aarch64-apple-ios-sim/libvoicevox_core.dylib" -output "artifact/voicevox_core-sim/libvoicevox_core.dylib"
-      - name: Create XCFramework
-        run: |
-          mkdir -p "artifact/${{ env.ASSET_NAME }}"
-          # 必要なファイルだけコピー
-          mkdir -p "Headers-sim"
-          cp -v artifact/voicevox_core-x86_64-apple-ios/voicevox_core.h "Headers-sim"
-          cp -v crates/voicevox_core_c_api/xcframework/Headers/module.modulemap "Headers-sim"
-          mkdir -p "Headers-aarch64"
-          cp -v artifact/voicevox_core-aarch64-apple-ios/voicevox_core.h "Headers-aarch64"
-          cp -v crates/voicevox_core_c_api/xcframework/Headers/module.modulemap "Headers-aarch64"
-          xcodebuild -create-xcframework \
-            -library "artifact/voicevox_core-sim/libvoicevox_core.dylib" \
-            -headers "Headers-sim" \
-            -library "artifact/voicevox_core-aarch64-apple-ios/libvoicevox_core.dylib" \
-            -headers "Headers-aarch64" \
-            -output "artifact/${{ env.ASSET_NAME }}/voicevox_core.xcframework"
+          build_util/make_ios_xcframework.bash
+          echo "output_asset_path=${OUTPUT_ASSET_PATH}" >> "$GITHUB_OUTPUT" 
+        env:
+          OUTPUT_ASSET_PATH: artifact/voicevox_core-ios-xcframework-cpu
       - name: Archive artifact
         run: |
-          cd artifact/${{ env.ASSET_NAME }}
+          cd ${{ steps.create-xcframework.outputs.output_asset_path }}
           7z a "../../${{ env.ASSET_NAME }}.zip" "voicevox_core.xcframework"
       - name: Upload to Release
         if: fromJson(needs.config.outputs.deploy)
@@ -432,7 +422,7 @@ jobs:
       - name: Raplace resource
         if: inputs.is_production
         shell: bash
-        run: |
+        run:
           rm -r ./model; mv download/fat_resource/core/model ./model
       - name: Create artifact
         run: |
@@ -474,7 +464,7 @@ jobs:
       - name: Install cargo-edit
         run: cargo binstall cargo-edit@^0.11 --no-confirm
       - name: set cargo version
-        run: |
+        run:
           cargo set-version "$VERSION" -p voicevox_core_java_api
 
       - name: "Download artifact (android-arm64-cpu)"

build_util/make_ios_xcframework.bash

@@ -0,0 +1,79 @@
+#!/usr/bin/env bash
+set -eu
+
+if [ ! -v IOS_X86_64_PATH ]; then # X86_64用のモジュールのディレクトリ(simulator)
+    echo "IOS_X86_64_PATHが未定義です"
+    exit 1
+fi
+if [ ! -v IOS_AARCH64_SIM_PATH ]; then # AARCH64_SIM用のモジュールのディレクトリ(simulator)
+    echo "IOS_AARCH64_SIM_PATHが未定義です"
+    exit 1
+fi
+if [ ! -v IOS_AARCH64_PATH ]; then # AARCH64用のモジュールのディレクトリ(実機)
+    echo "IOS_AARCH64_PATHが未定義です"
+    exit 1
+fi
+if [ ! -v OUTPUT_ASSET_PATH ]; then # 出力するASSETのディレクトリ
+    echo "OUTPUT_ASSET_PATHが未定義です"
+    exit 1
+fi
+
+echo "* Get original onnxruntime file name from rpath"
+output=$(otool -L "${IOS_AARCH64_PATH}/libvoicevox_core.dylib")
+matched_line=$(echo "$output" | grep "@rpath" | grep "libonnxruntime")
+if [[ $matched_line ]]; then
+    if [[ $matched_line =~ (@rpath/([^ ]+\.dylib)) ]]; then
+        dylib_string=${BASH_REMATCH[2]}
+    else
+        echo "Expected pattern not found in the matched line"
+        echo "$output"
+        exit 1
+    fi
+else
+    echo "No line containing '@rpath' and 'libonnxruntime' found"
+    echo "$output"
+    exit 1
+fi
+echo "Original onnx dylib file name: $dylib_string"
+
+echo "* Copy Framework template"
+arches=("aarch64" "sim")
+artifacts=("${IOS_AARCH64_PATH}" "${IOS_AARCH64_SIM_PATH}")
+for i in "${!arches[@]}"; do
+    arch="${arches[$i]}"
+    artifact="${artifacts[$i]}"
+    echo "* Copy Framework-${arch} template"
+    mkdir -p "Framework-${arch}/voicevox_core.framework/Headers"
+    cp -vr "crates/voicevox_core_c_api/xcframework/Frameworks/${arch}/" "Framework-${arch}/"
+    cp -v "${artifact}/voicevox_core.h" \
+        "Framework-${arch}/voicevox_core.framework/Headers/voicevox_core.h"
+done
+
+echo "* Create dylib"
+# aarch64はdylibをコピー
+cp -v "${IOS_AARCH64_PATH}/libvoicevox_core.dylib" \
+    "Framework-aarch64/voicevox_core.framework/voicevox_core"
+
+# simはx86_64とarrch64を合わせてdylib作成
+lipo -create "${IOS_X86_64_PATH}/libvoicevox_core.dylib" \
+    "${IOS_AARCH64_SIM_PATH}/libvoicevox_core.dylib" \
+    -output "Framework-sim/voicevox_core.framework/voicevox_core"
+
+for arch in "${arches[@]}"; do
+    echo "* Change ${arch} @rpath"
+    # 自身への@rpathを変更
+    install_name_tool -id "@rpath/voicevox_core.framework/voicevox_core" \
+        "Framework-${arch}/voicevox_core.framework/voicevox_core"
+
+    # 依存ライブラリonnxruntimeへの@rpathを変更
+    install_name_tool -change "@rpath/$dylib_string" \
+        "@rpath/onnxruntime.framework/onnxruntime" \
+        "Framework-${arch}/voicevox_core.framework/voicevox_core"
+done
+
+echo "* Create XCFramework"
+mkdir -p "${OUTPUT_ASSET_PATH}"
+xcodebuild -create-xcframework \
+    -framework "Framework-sim/voicevox_core.framework" \
+    -framework "Framework-aarch64/voicevox_core.framework" \
+    -output "${OUTPUT_ASSET_PATH}/voicevox_core.xcframework"

crates/test_util/src/lib.rs

@@ -22,7 +22,7 @@ pub const OPEN_JTALK_DIC_DIR: &str = concat!(
     "/data/open_jtalk_dic_utf_8-1.11"
 );
 
-pub const EXAMPLE_DATA_JSON: &str = include_str!(concat!(
+const EXAMPLE_DATA_JSON: &str = include_str!(concat!(
     env!("CARGO_MANIFEST_DIR"),
     "/data/example_data.json"
 ));

crates/voicevox_core/src/engine/acoustic_feature_extractor.rs

@@ -61,7 +61,7 @@ static PHONEME_MAP: Lazy<HashMap<&str, i64>> = Lazy::new(|| {
 });
 
 #[derive(Debug, Clone, PartialEq, new, Default, Getters)]
-pub struct OjtPhoneme {
+pub(crate) struct OjtPhoneme {
     phoneme: String,
     #[allow(dead_code)]
     start: f32,
@@ -70,23 +70,23 @@ pub struct OjtPhoneme {
 }
 
 impl OjtPhoneme {
-    pub fn num_phoneme() -> usize {
+    pub(crate) fn num_phoneme() -> usize {
         PHONEME_MAP.len()
     }
 
-    pub fn space_phoneme() -> String {
+    fn space_phoneme() -> String {
         "pau".into()
     }
 
-    pub fn phoneme_id(&self) -> i64 {
+    pub(crate) fn phoneme_id(&self) -> i64 {
         if self.phoneme.is_empty() {
             -1
         } else {
             *PHONEME_MAP.get(&self.phoneme.as_str()).unwrap()
         }
     }
 
-    pub fn convert(phonemes: &[OjtPhoneme]) -> Vec<OjtPhoneme> {
+    pub(crate) fn convert(phonemes: &[OjtPhoneme]) -> Vec<OjtPhoneme> {
         let mut phonemes = phonemes.to_owned();
         if let Some(first_phoneme) = phonemes.first_mut() {
             if first_phoneme.phoneme.contains("sil") {

crates/voicevox_core/src/engine/kana_parser.rs

@@ -163,7 +163,7 @@ pub(crate) fn parse_kana(text: &str) -> KanaParseResult<Vec<AccentPhraseModel>>
     Ok(parsed_result)
 }
 
-pub fn create_kana(accent_phrases: &[AccentPhraseModel]) -> String {
+pub(crate) fn create_kana(accent_phrases: &[AccentPhraseModel]) -> String {
     let mut text = String::new();
     for phrase in accent_phrases {
         let moras = phrase.moras();

crates/voicevox_core/src/engine/mora_list.rs

@@ -186,7 +186,7 @@ pub(super) const MORA_LIST_MINIMUM: &[[&str; 3]] = &[
     ["ア", "", "a"],
 ];
 
-pub fn mora2text(mora: &str) -> &str {
+pub(crate) fn mora2text(mora: &str) -> &str {
     for &[text, consonant, vowel] in MORA_LIST_MINIMUM {
         if mora.len() >= consonant.len()
             && &mora[..consonant.len()] == consonant

crates/voicevox_core/src/infer/status.rs

@@ -31,14 +31,14 @@ pub(crate) struct Status<R: InferenceRuntime, D: InferenceDomain> {
 }
 
 impl<R: InferenceRuntime, D: InferenceDomain> Status<R, D> {
-    pub fn new(session_options: EnumMap<D::Operation, InferenceSessionOptions>) -> Self {
+    pub(crate) fn new(session_options: EnumMap<D::Operation, InferenceSessionOptions>) -> Self {
         Self {
             loaded_models: Default::default(),
             session_options,
         }
     }
 
-    pub fn insert_model(
+    pub(crate) fn insert_model(
         &self,
         model_header: &VoiceModelHeader,
         model_bytes: &EnumMap<D::Operation, Vec<u8>>,
@@ -64,30 +64,30 @@ impl<R: InferenceRuntime, D: InferenceDomain> Status<R, D> {
         Ok(())
     }
 
-    pub fn unload_model(&self, voice_model_id: &VoiceModelId) -> Result<()> {
+    pub(crate) fn unload_model(&self, voice_model_id: &VoiceModelId) -> Result<()> {
         self.loaded_models.lock().unwrap().remove(voice_model_id)
     }
 
-    pub fn metas(&self) -> VoiceModelMeta {
+    pub(crate) fn metas(&self) -> VoiceModelMeta {
         self.loaded_models.lock().unwrap().metas()
     }
 
     pub(crate) fn ids_for(&self, style_id: StyleId) -> Result<(VoiceModelId, ModelInnerId)> {
         self.loaded_models.lock().unwrap().ids_for(style_id)
     }
 
-    pub fn is_loaded_model(&self, voice_model_id: &VoiceModelId) -> bool {
+    pub(crate) fn is_loaded_model(&self, voice_model_id: &VoiceModelId) -> bool {
         self.loaded_models
             .lock()
             .unwrap()
             .contains_voice_model(voice_model_id)
     }
 
-    pub fn is_loaded_model_by_style_id(&self, style_id: StyleId) -> bool {
+    pub(crate) fn is_loaded_model_by_style_id(&self, style_id: StyleId) -> bool {
         self.loaded_models.lock().unwrap().contains_style(style_id)
     }
 
-    pub fn validate_speaker_id(&self, style_id: StyleId) -> bool {
+    pub(crate) fn validate_speaker_id(&self, style_id: StyleId) -> bool {
         self.is_loaded_model_by_style_id(style_id)
     }
 

crates/voicevox_core/src/lib.rs

@@ -12,13 +12,13 @@ mod numerics;
 mod result;
 mod synthesizer;
 mod task;
+mod text_analyzer;
 mod user_dict;
 mod version;
 mod voice_model;
 
 pub mod __internal;
 pub mod blocking;
-pub mod text_analyzer;
 pub mod tokio;
 
 #[cfg(test)]

crates/voicevox_core/src/test_util.rs

@@ -2,7 +2,7 @@ use std::path::PathBuf;
 
 use crate::Result;
 
-pub async fn open_default_vvm_file() -> crate::tokio::VoiceModel {
+pub(crate) async fn open_default_vvm_file() -> crate::tokio::VoiceModel {
     crate::tokio::VoiceModel::from_path(
         ::test_util::convert_zip_vvm(
             PathBuf::from(env!("CARGO_WORKSPACE_DIR"))

crates/voicevox_core/src/text_analyzer.rs

@@ -3,13 +3,13 @@ use crate::{
     AccentPhraseModel, FullcontextExtractor, Result,
 };
 
-pub trait TextAnalyzer {
+pub(crate) trait TextAnalyzer {
     fn analyze(&self, text: &str) -> Result<Vec<AccentPhraseModel>>;
 }
 
 /// AquesTalk風記法からAccentPhraseの配列を生成するTextAnalyzer
 #[derive(Clone)]
-pub struct KanaAnalyzer;
+pub(crate) struct KanaAnalyzer;
 
 impl TextAnalyzer for KanaAnalyzer {
     fn analyze(&self, text: &str) -> Result<Vec<AccentPhraseModel>> {
@@ -22,10 +22,10 @@ impl TextAnalyzer for KanaAnalyzer {
 
 /// OpenJtalkからAccentPhraseの配列を生成するTextAnalyzer
 #[derive(Clone)]
-pub struct OpenJTalkAnalyzer<O>(O);
+pub(crate) struct OpenJTalkAnalyzer<O>(O);
 
 impl<O> OpenJTalkAnalyzer<O> {
-    pub fn new(open_jtalk: O) -> Self {
+    pub(crate) fn new(open_jtalk: O) -> Self {
         Self(open_jtalk)
     }
 }