Merge branch 'main' into change-liberate-voicevox-core

 #891 で入れたtodoも解決する。

crates/voicevox_core/Cargo.toml

@@ -25,11 +25,11 @@ camino.workspace = true
 const_format.workspace = true
 derive-getters.workspace = true
 derive-new.workspace = true
-derive_more = { workspace = true, features = ["add", "deref", "display", "from", "from_str"] }
+derive_more = { workspace = true, features = ["add", "deref", "display", "from", "from_str", "index"] }
 duplicate.workspace = true
 easy-ext.workspace = true
 educe.workspace = true
-enum-map.workspace = true
+enum-map = { workspace = true, features = ["serde"] }
 fs-err.workspace = true
 futures-io.workspace = true
 futures-lite.workspace = true

crates/voicevox_core/src/infer/domains/talk.rs

@@ -3,6 +3,7 @@ use std::{collections::BTreeSet, sync::LazyLock};
 use enum_map::Enum;
 use macros::{InferenceInputSignature, InferenceOperation, InferenceOutputSignature};
 use ndarray::{Array0, Array1, Array2};
+use serde::Deserialize;
 
 use crate::{manifest::TalkManifest, StyleType};
 
@@ -23,7 +24,8 @@ impl InferenceDomain for TalkDomain {
     }
 }
 
-#[derive(Clone, Copy, Enum, InferenceOperation)]
+#[derive(Clone, Copy, Deserialize, Enum, InferenceOperation)]
+#[serde(rename_all = "snake_case")]
 #[inference_operation(
     type Domain = TalkDomain;
 )]

crates/voicevox_core/src/manifest.rs

@@ -5,9 +5,9 @@ use std::{
 };
 
 use derive_getters::Getters;
-use derive_more::Deref;
+use derive_more::{Deref, Index};
 use derive_new::new;
-use macros::IndexForFields;
+use enum_map::EnumMap;
 use serde::{de, Deserialize, Deserializer, Serialize};
 use serde_with::{serde_as, DisplayFromStr};
 
@@ -81,21 +81,12 @@ pub struct Manifest {
 
 pub(crate) type ManifestDomains = inference_domain_map_values!(for<D> Option<D::Manifest>);
 
-#[derive(Deserialize, IndexForFields)]
+#[derive(Index, Deserialize)]
 #[cfg_attr(test, derive(Default))]
-#[index_for_fields(TalkOperation)]
 pub(crate) struct TalkManifest {
-    #[index_for_fields(TalkOperation::PredictDuration)]
-    pub(crate) predict_duration: ModelFile,
-
-    #[index_for_fields(TalkOperation::PredictIntonation)]
-    pub(crate) predict_intonation: ModelFile,
-
-    #[index_for_fields(TalkOperation::GenerateFullIntermediate)]
-    pub(crate) generate_full_intermediate: ModelFile,
-
-    #[index_for_fields(TalkOperation::RenderAudioSegment)]
-    pub(crate) render_audio_segment: ModelFile,
+    #[index]
+    #[serde(flatten)]
+    filenames: EnumMap<TalkOperation, ModelFile>,
 
     #[serde(default)]
     pub(crate) style_id_to_inner_voice_id: StyleIdToInnerVoiceId,

crates/voicevox_core/src/voice_model.rs

@@ -11,7 +11,7 @@ use std::{
 use anyhow::{anyhow, Context as _};
 use derive_more::From;
 use easy_ext::ext;
-use enum_map::{enum_map, EnumMap};
+use enum_map::{Enum, EnumMap};
 use futures_io::{AsyncBufRead, AsyncRead, AsyncSeek};
 use futures_util::future::{OptionFuture, TryFutureExt as _};
 use itertools::Itertools as _;
@@ -23,7 +23,7 @@ use crate::{
     asyncs::{Async, Mutex as _},
     error::{LoadModelError, LoadModelErrorKind, LoadModelResult},
     infer::{
-        domains::{inference_domain_map_values, InferenceDomainMap, TalkDomain, TalkOperation},
+        domains::{inference_domain_map_values, InferenceDomainMap, TalkDomain},
         InferenceDomain,
     },
     manifest::{Manifest, ManifestDomains, ModelFile, ModelFileType, StyleIdToInnerVoiceId},
@@ -128,7 +128,7 @@ impl<A: Async> Inner<A> {
 
         let header = VoiceModelHeader::new(manifest, metas, path)?.into();
 
-        InnerTryBuilder {
+        return InnerTryBuilder {
             header,
             inference_model_entries_builder: |header| {
                 let VoiceModelHeader { manifest, .. } = &**header;
@@ -139,23 +139,9 @@ impl<A: Async> Inner<A> {
                         talk: |talk| {
                             talk.as_ref()
                                 .map(|manifest| {
-                                    let indices = enum_map! {
-                                        TalkOperation::PredictDuration => {
-                                            find_entry_index(&manifest.predict_duration.filename)?
-                                        }
-                                        TalkOperation::PredictIntonation => {
-                                            find_entry_index(&manifest.predict_intonation.filename)?
-                                        }
-                                        TalkOperation::GenerateFullIntermediate => {
-                                            find_entry_index(
-                                                &manifest.generate_full_intermediate.filename,
-                                            )?
-                                        }
-                                        TalkOperation::RenderAudioSegment => find_entry_index(
-                                            &manifest.render_audio_segment.filename,
-                                        )?,
-                                    };
-
+                                    let indices = EnumMap::from_fn(|k| &manifest[k]).try_map(
+                                        |_, ModelFile { filename, .. }| find_entry_index(filename),
+                                    )?;
                                     Ok(InferenceModelEntry { indices, manifest })
                                 })
                                 .transpose()
@@ -174,7 +160,26 @@ impl<A: Async> Inner<A> {
             },
             zip: zip.into_inner().into_inner().into(),
         }
-        .try_build()
+        .try_build();
+
+        #[ext]
+        impl<K: Enum, V> EnumMap<K, V> {
+            fn try_map<V2, E>(
+                self,
+                f: impl FnMut(K, V) -> Result<V2, E>,
+            ) -> Result<EnumMap<K, V2>, E> {
+                let mut elems = self
+                    .map(f)
+                    .into_iter()
+                    .map(|(_, r)| r.map(Some))
+                    .collect::<Result<Vec<_>, _>>()?;
+
+                Ok(EnumMap::<K, _>::from_fn(|key| {
+                    let key = key.into_usize();
+                    elems[key].take().expect("each `key` should be distinct")
+                }))
+            }
+        }
     }
 
     fn id(&self) -> VoiceModelId {

crates/voicevox_core_macros/src/lib.rs

@@ -3,7 +3,6 @@
 mod extract;
 mod inference_domain;
 mod inference_domains;
-mod manifest;
 
 use syn::parse_macro_input;
 
@@ -103,38 +102,6 @@ pub fn derive_inference_output_signature(
     from_syn(inference_domain::derive_inference_output_signature(input))
 }
 
-/// 構造体のフィールドを取得できる`std::ops::Index`の実装を導出する。
-///
-/// # Example
-///
-/// ```
-/// use macros::IndexForFields;
-///
-/// #[derive(IndexForFields)]
-/// #[index_for_fields(TalkOperation)]
-/// pub(crate) struct TalkManifest {
-///     #[index_for_fields(TalkOperation::PredictDuration)]
-///     pub(crate) predict_duration_filename: Arc<str>,
-///
-///     #[index_for_fields(TalkOperation::PredictIntonation)]
-///     pub(crate) predict_intonation_filename: Arc<str>,
-///
-///     #[index_for_fields(TalkOperation::GenerateFullIntermediate)]
-///     pub(crate) generate_full_intermediate_filename: Arc<str>,
-///
-///     #[index_for_fields(TalkOperation::RenderAudioSegment)]
-///     pub(crate) render_audio_segment_filename: Arc<str>,
-///
-///     // …
-/// }
-/// ```
-#[cfg(not(doctest))]
-#[proc_macro_derive(IndexForFields, attributes(index_for_fields))]
-pub fn derive_index_for_fields(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
-    let input = &parse_macro_input!(input);
-    from_syn(manifest::derive_index_for_fields(input))
-}
-
 /// # Example
 ///
 /// ```