open_jtalk-rsを更新し、caminoを利用 (#745)

Cargo.toml

@@ -12,6 +12,7 @@ async-std = "1.12.0"
 async_zip = "0.0.11"
 binstall-tar = "0.4.39"
 bytes = "1.1.0"
+camino = "1.1.6"
 cbindgen = "0.24.3"
 chrono = { version = "0.4.26", default-features = false }
 clap = "4.0.10"
@@ -87,7 +88,7 @@ rev = "ebb9dcb9b26ee681889b52b6db3b4f642b04a250"
 
 [workspace.dependencies.open_jtalk]
 git = "https://github.com/VOICEVOX/open_jtalk-rs.git"
-rev = "a16714ce16dec76fd0e3041a7acfa484921db3b5"
+rev = "70c76bb54522830e92803038191bf533ba68ce85"
 
 # FIXME: iOS対応のpull request(https://github.com/wesleywiser/process_path/pull/16)がマージされる見込みが無いため
 [workspace.dependencies.process_path]

crates/voicevox_core/Cargo.toml

@@ -11,6 +11,7 @@ directml = ["onnxruntime/directml"]
 [dependencies]
 anyhow.workspace = true
 async_zip = { workspace = true, features = ["full"] }
+camino.workspace = true
 derive-getters.workspace = true
 derive-new.workspace = true
 derive_more.workspace = true

crates/voicevox_core/src/__internal/doctest_fixtures.rs

@@ -1,9 +1,9 @@
-use std::path::Path;
+use camino::Utf8Path;
 
 use crate::{AccelerationMode, InitializeOptions};
 
 pub async fn synthesizer_with_sample_voice_model(
-    open_jtalk_dic_dir: impl AsRef<Path>,
+    open_jtalk_dic_dir: impl AsRef<Utf8Path>,
 ) -> anyhow::Result<crate::tokio::Synthesizer<crate::tokio::OpenJtalk>> {
     let syntesizer = crate::tokio::Synthesizer::new(
         crate::tokio::OpenJtalk::new(open_jtalk_dic_dir).await?,

crates/voicevox_core/src/engine/open_jtalk.rs

@@ -15,11 +15,11 @@ pub trait FullcontextExtractor: Clone + Send + Sync + 'static {
 pub(crate) mod blocking {
     use std::{
         io::Write as _,
-        path::Path,
         sync::{Arc, Mutex},
     };
 
-    use anyhow::anyhow;
+    use anyhow::Context as _;
+    use camino::{Utf8Path, Utf8PathBuf};
     use open_jtalk::{mecab_dict_index, text2mecab, JpCommon, ManagedResource, Mecab, Njd};
     use tempfile::NamedTempFile;
 
@@ -32,12 +32,8 @@ pub(crate) mod blocking {
     pub struct OpenJtalk(pub(super) Arc<Inner>);
 
     impl self::OpenJtalk {
-        pub fn new(open_jtalk_dict_dir: impl AsRef<Path>) -> crate::result::Result<Self> {
-            let dict_dir = open_jtalk_dict_dir
-                .as_ref()
-                .to_str()
-                .unwrap_or_else(|| todo!()) // FIXME: `camino::Utf8Path`を要求するようにする
-                .to_owned();
+        pub fn new(open_jtalk_dict_dir: impl AsRef<Utf8Path>) -> crate::result::Result<Self> {
+            let dict_dir = open_jtalk_dict_dir.as_ref().to_owned();
 
             // FIXME: この`{}`はGitのdiffを抑えるためだけに存在
             {
@@ -47,11 +43,12 @@ pub(crate) mod blocking {
                     jpcommon: ManagedResource::initialize(),
                 };
 
-                let result = resources.mecab.load(&*dict_dir);
-                if !result {
-                    // FIXME: 「システム辞書を読もうとしたけど読めなかった」というエラーをちゃんと用意する
-                    return Err(ErrorRepr::NotLoadedOpenjtalkDict.into());
-                }
+                // FIXME: 「システム辞書を読もうとしたけど読めなかった」というエラーをちゃんと用意する
+                resources
+                    .mecab
+                    .load(&*dict_dir)
+                    .inspect_err(|e| tracing::error!("{e:?}"))
+                    .map_err(|_| ErrorRepr::NotLoadedOpenjtalkDict)?;
 
                 Ok(Self(Arc::new(Inner {
                     resources: Mutex::new(resources),
@@ -124,7 +121,7 @@ pub(crate) mod blocking {
 
     pub(super) struct Inner {
         resources: std::sync::Mutex<Resources>,
-        dict_dir: String, // FIXME: `camino::Utf8PathBuf`にする
+        dict_dir: Utf8PathBuf,
     }
 
     impl Inner {
@@ -145,34 +142,37 @@ pub(crate) mod blocking {
                     NamedTempFile::new().map_err(|e| ErrorRepr::UseUserDict(e.into()))?;
                 let temp_dict_path = temp_dict.into_temp_path();
 
+                // FIXME: `.unwrap()`ではなく、エラーとして回収する
+                let temp_csv_path = Utf8Path::from_path(temp_csv_path.as_ref()).unwrap();
+                let temp_dict_path = Utf8Path::from_path(temp_dict_path.as_ref()).unwrap();
+
                 // Mecabでユーザー辞書をコンパイル
                 // TODO: エラー（SEGV）が出るパターンを把握し、それをRust側で防ぐ。
                 mecab_dict_index(&[
                     "mecab-dict-index",
                     "-d",
-                    &self.dict_dir,
+                    self.dict_dir.as_ref(),
                     "-u",
-                    temp_dict_path.to_str().unwrap(),
+                    temp_dict_path.as_ref(),
                     "-f",
                     "utf-8",
                     "-t",
                     "utf-8",
-                    temp_csv_path.to_str().unwrap(),
+                    temp_csv_path.as_ref(),
                     "-q",
                 ]);
 
-                self.load_with_userdic(Some(temp_dict_path.as_ref()))
+                self.load_with_userdic(Some(temp_dict_path))
             }
         }
-        fn load_with_userdic(&self, dict_path: Option<&Path>) -> crate::result::Result<()> {
+        fn load_with_userdic(&self, dict_path: Option<&Utf8Path>) -> crate::result::Result<()> {
             let Resources { mecab, .. } = &mut *self.resources.lock().unwrap();
 
-            let result = mecab.load_with_userdic(self.dict_dir.as_ref(), dict_path);
-
-            if !result {
-                return Err(ErrorRepr::UseUserDict(anyhow!("辞書を読み込めませんでした。")).into());
-            }
-            Ok(())
+            mecab
+                .load_with_userdic(self.dict_dir.as_ref(), dict_path)
+                .context("辞書を読み込めませんでした。")
+                .map_err(ErrorRepr::UseUserDict)
+                .map_err(Into::into)
         }
     }
 
@@ -188,7 +188,7 @@ pub(crate) mod blocking {
 }
 
 pub(crate) mod tokio {
-    use std::path::Path;
+    use camino::Utf8Path;
 
     use super::FullcontextExtractor;
 
@@ -197,7 +197,7 @@ pub(crate) mod tokio {
     pub struct OpenJtalk(super::blocking::OpenJtalk);
 
     impl self::OpenJtalk {
-        pub async fn new(open_jtalk_dict_dir: impl AsRef<Path>) -> crate::result::Result<Self> {
+        pub async fn new(open_jtalk_dict_dir: impl AsRef<Utf8Path>) -> crate::result::Result<Self> {
             let open_jtalk_dict_dir = open_jtalk_dict_dir.as_ref().to_owned();
             let blocking =
                 crate::task::asyncify(|| super::blocking::OpenJtalk::new(open_jtalk_dict_dir))

crates/voicevox_core_c_api/Cargo.toml

@@ -18,6 +18,7 @@ directml = ["voicevox_core/directml"]
 [dependencies]
 anstream = { workspace = true, default-features = false, features = ["auto"] }
 anstyle-query.workspace = true
+camino.workspace = true
 chrono = { workspace = true, default-features = false, features = ["clock"] }
 colorchoice.workspace = true
 cstr.workspace = true

crates/voicevox_core_c_api/src/c_impls.rs

@@ -1,11 +1,12 @@
 use std::{ffi::CString, path::Path};
 
+use camino::Utf8Path;
 use voicevox_core::{InitializeOptions, Result, VoiceModelId};
 
 use crate::{helpers::CApiResult, OpenJtalkRc, VoicevoxSynthesizer, VoicevoxVoiceModel};
 
 impl OpenJtalkRc {
-    pub(crate) fn new(open_jtalk_dic_dir: impl AsRef<Path>) -> Result<Self> {
+    pub(crate) fn new(open_jtalk_dic_dir: impl AsRef<Utf8Path>) -> Result<Self> {
         Ok(Self {
             open_jtalk: voicevox_core::blocking::OpenJtalk::new(open_jtalk_dic_dir)?,
         })

crates/voicevox_core_python_api/Cargo.toml

@@ -11,6 +11,7 @@ crate-type = ["cdylib"]
 directml = ["voicevox_core/directml"]
 
 [dependencies]
+camino.workspace = true
 easy-ext.workspace = true
 log.workspace = true
 pyo3 = { workspace = true, features = ["abi3-py38", "extension-module"] }

crates/voicevox_core_python_api/src/convert.rs

@@ -1,5 +1,6 @@
 use std::{error::Error as _, future::Future, iter, path::PathBuf};
 
+use camino::Utf8PathBuf;
 use easy_ext::ext;
 use pyo3::{
     exceptions::{PyException, PyValueError},
@@ -38,10 +39,12 @@ pub fn from_acceleration_mode(ob: &PyAny) -> PyResult<AccelerationMode> {
     }
 }
 
-pub fn from_utf8_path(ob: &PyAny) -> PyResult<String> {
+// FIXME: `VoiceModel`や`UserDict`についてはこれではなく、`PathBuf::extract`を直接使うようにする
+pub fn from_utf8_path(ob: &PyAny) -> PyResult<Utf8PathBuf> {
     PathBuf::extract(ob)?
         .into_os_string()
         .into_string()
+        .map(Utf8PathBuf::from)
         .map_err(|s| PyValueError::new_err(format!("{s:?} cannot be encoded to UTF-8")))
 }
 

crates/voicevox_core_python_api/src/lib.rs

@@ -6,6 +6,7 @@ use self::convert::{
     to_py_user_dict_word, to_py_uuid, to_pydantic_dataclass, to_pydantic_voice_model_meta,
     to_rust_user_dict_word, to_rust_uuid, VoicevoxCoreResultExt as _,
 };
+use camino::Utf8PathBuf;
 use easy_ext::ext;
 use log::debug;
 use pyo3::{
@@ -115,7 +116,7 @@ impl VoiceModel {
     #[staticmethod]
     fn from_path(
         py: Python<'_>,
-        #[pyo3(from_py_with = "from_utf8_path")] path: String,
+        #[pyo3(from_py_with = "from_utf8_path")] path: Utf8PathBuf,
     ) -> PyResult<&PyAny> {
         pyo3_asyncio::tokio::future_into_py(py, async move {
             let model = voicevox_core::tokio::VoiceModel::from_path(path).await;
@@ -146,7 +147,7 @@ impl OpenJtalk {
     #[allow(clippy::new_ret_no_self)]
     #[staticmethod]
     fn new(
-        #[pyo3(from_py_with = "from_utf8_path")] open_jtalk_dict_dir: String,
+        #[pyo3(from_py_with = "from_utf8_path")] open_jtalk_dict_dir: Utf8PathBuf,
         py: Python<'_>,
     ) -> PyResult<&PyAny> {
         pyo3_asyncio::tokio::future_into_py(py, async move {
@@ -637,6 +638,7 @@ impl UserDict {
 mod blocking {
     use std::sync::Arc;
 
+    use camino::Utf8PathBuf;
     use pyo3::{
         pyclass, pymethods,
         types::{IntoPyDict as _, PyBytes, PyDict, PyList},
@@ -661,7 +663,7 @@ mod blocking {
         #[staticmethod]
         fn from_path(
             py: Python<'_>,
-            #[pyo3(from_py_with = "crate::convert::from_utf8_path")] path: String,
+            #[pyo3(from_py_with = "crate::convert::from_utf8_path")] path: Utf8PathBuf,
         ) -> PyResult<Self> {
             let model = voicevox_core::blocking::VoiceModel::from_path(path).into_py_result(py)?;
             Ok(Self { model })
@@ -688,7 +690,7 @@ mod blocking {
     impl OpenJtalk {
         #[new]
         fn new(
-            #[pyo3(from_py_with = "super::from_utf8_path")] open_jtalk_dict_dir: String,
+            #[pyo3(from_py_with = "super::from_utf8_path")] open_jtalk_dict_dir: Utf8PathBuf,
             py: Python<'_>,
         ) -> PyResult<Self> {
             let open_jtalk =