tuna2134/sbv2-v2
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Add User dict module from VOICEVOX

Browse Source
This commit is contained in:
litagin02
2024-02-25 13:14:28 +09:00
parent 760f6c2680
commit 357a3330c8
7 changed files with 917 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

165
text/user_dict/LGPL_LICENSE Normal file
View File

@@ -0,0 +1,165 @@
GNU LESSER GENERAL PUBLIC LICENSE
Version 3, 29 June 2007
Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
This version of the GNU Lesser General Public License incorporates
the terms and conditions of version 3 of the GNU General Public
License, supplemented by the additional permissions listed below.
0. Additional Definitions.
As used herein, "this License" refers to version 3 of the GNU Lesser
General Public License, and the "GNU GPL" refers to version 3 of the GNU
General Public License.
"The Library" refers to a covered work governed by this License,
other than an Application or a Combined Work as defined below.
An "Application" is any work that makes use of an interface provided
by the Library, but which is not otherwise based on the Library.
Defining a subclass of a class defined by the Library is deemed a mode
of using an interface provided by the Library.
A "Combined Work" is a work produced by combining or linking an
Application with the Library. The particular version of the Library
with which the Combined Work was made is also called the "Linked
Version".
The "Minimal Corresponding Source" for a Combined Work means the
Corresponding Source for the Combined Work, excluding any source code
for portions of the Combined Work that, considered in isolation, are
based on the Application, and not on the Linked Version.
The "Corresponding Application Code" for a Combined Work means the
object code and/or source code for the Application, including any data
and utility programs needed for reproducing the Combined Work from the
Application, but excluding the System Libraries of the Combined Work.
1. Exception to Section 3 of the GNU GPL.
You may convey a covered work under sections 3 and 4 of this License
without being bound by section 3 of the GNU GPL.
2. Conveying Modified Versions.
If you modify a copy of the Library, and, in your modifications, a
facility refers to a function or data to be supplied by an Application
that uses the facility (other than as an argument passed when the
facility is invoked), then you may convey a copy of the modified
version:
a) under this License, provided that you make a good faith effort to
ensure that, in the event an Application does not supply the
function or data, the facility still operates, and performs
whatever part of its purpose remains meaningful, or
b) under the GNU GPL, with none of the additional permissions of
this License applicable to that copy.
3. Object Code Incorporating Material from Library Header Files.
The object code form of an Application may incorporate material from
a header file that is part of the Library. You may convey such object
code under terms of your choice, provided that, if the incorporated
material is not limited to numerical parameters, data structure
layouts and accessors, or small macros, inline functions and templates
(ten or fewer lines in length), you do both of the following:
a) Give prominent notice with each copy of the object code that the
Library is used in it and that the Library and its use are
covered by this License.
b) Accompany the object code with a copy of the GNU GPL and this license
document.
4. Combined Works.
You may convey a Combined Work under terms of your choice that,
taken together, effectively do not restrict modification of the
portions of the Library contained in the Combined Work and reverse
engineering for debugging such modifications, if you also do each of
the following:
a) Give prominent notice with each copy of the Combined Work that
the Library is used in it and that the Library and its use are
covered by this License.
b) Accompany the Combined Work with a copy of the GNU GPL and this license
document.
c) For a Combined Work that displays copyright notices during
execution, include the copyright notice for the Library among
these notices, as well as a reference directing the user to the
copies of the GNU GPL and this license document.
d) Do one of the following:
0) Convey the Minimal Corresponding Source under the terms of this
License, and the Corresponding Application Code in a form
suitable for, and under terms that permit, the user to
recombine or relink the Application with a modified version of
the Linked Version to produce a modified Combined Work, in the
manner specified by section 6 of the GNU GPL for conveying
Corresponding Source.
1) Use a suitable shared library mechanism for linking with the
Library. A suitable mechanism is one that (a) uses at run time
a copy of the Library already present on the user's computer
system, and (b) will operate properly with a modified version
of the Library that is interface-compatible with the Linked
Version.
e) Provide Installation Information, but only if you would otherwise
be required to provide such information under section 6 of the
GNU GPL, and only to the extent that such information is
necessary to install and execute a modified version of the
Combined Work produced by recombining or relinking the
Application with a modified version of the Linked Version. (If
you use option 4d0, the Installation Information must accompany
the Minimal Corresponding Source and Corresponding Application
Code. If you use option 4d1, you must provide the Installation
Information in the manner specified by section 6 of the GNU GPL
for conveying Corresponding Source.)
5. Combined Libraries.
You may place library facilities that are a work based on the
Library side by side in a single library together with other library
facilities that are not Applications and are not covered by this
License, and convey such a combined library under terms of your
choice, if you do both of the following:
a) Accompany the combined library with a copy of the same work based
on the Library, uncombined with any other library facilities,
conveyed under the terms of this License.
b) Give prominent notice with the combined library that part of it
is a work based on the Library, and explaining where to find the
accompanying uncombined form of the same work.
6. Revised Versions of the GNU Lesser General Public License.
The Free Software Foundation may publish revised and/or new versions
of the GNU Lesser General Public License from time to time. Such new
versions will be similar in spirit to the present version, but may
differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the
Library as you received it specifies that a certain numbered version
of the GNU Lesser General Public License "or any later version"
applies to it, you have the option of following the terms and
conditions either of that published version or of any later version
published by the Free Software Foundation. If the Library as you
received it does not specify a version number of the GNU Lesser
General Public License, you may choose any version of the GNU Lesser
General Public License ever published by the Free Software Foundation.
If the Library as you received it specifies that a proxy can decide
whether future versions of the GNU Lesser General Public License shall
apply, that proxy's public statement of acceptance of any version is
permanent authorization for you to choose that version for the
Library.

16
text/user_dict/README.md Normal file
View File

@@ -0,0 +1,16 @@
このモジュールは、[voicevox_engine](https://github.com/VOICEVOX/voicevox_engine)を使わせていただいています。
引用元:
https://github.com/VOICEVOX/voicevox_engine/tree/709527be089c0410c08e989df95a4a1d78439423/voicevox_engine/user_dict
改変部分は以下のとおりです。
- ファイル名の書き換え、それに伴うimport文の書き換え
- VOICEVOX固有の他のモジュールへの依存のコメントアウト
- mutexを使用している部分のコメントアウト
- 参照しているpyopenjtalkの違いによるメソッド名書き換え
- UserDictWordのmora_countのデフォルト値をNoneに指定
ライセンス: LGPL
[LGPL_LICENSE](LGPL_LICENSE)を参照してください。

462
text/user_dict/__init__.py Normal file
View File

@@ -0,0 +1,462 @@
import json
import sys
import threading
import traceback
from pathlib import Path
from typing import Dict, List, Optional
from uuid import UUID, uuid4
import numpy as np
import pyopenjtalk
from fastapi import HTTPException
from .word_model import UserDictWord, WordTypes
# from ..utility.mutex_utility import mutex_wrapper
# from ..utility.path_utility import engine_root, get_save_dir
from .part_of_speech_data import MAX_PRIORITY, MIN_PRIORITY, part_of_speech_data
from common.constants import USER_DICT_DIR
# root_dir = engine_root()
# save_dir = get_save_dir()
root_dir = Path(USER_DICT_DIR)
save_dir = Path(USER_DICT_DIR)
if not save_dir.is_dir():
save_dir.mkdir(parents=True)
default_dict_path = root_dir / "default.csv" # VOICEVOXデフォルト辞書ファイルのパス
user_dict_path = save_dir / "user_dict.json" # ユーザー辞書ファイルのパス
compiled_dict_path = save_dir / "user.dic" # コンパイル済み辞書ファイルのパス
# # 同時書き込みの制御
# mutex_user_dict = threading.Lock()
# mutex_openjtalk_dict = threading.Lock()
# @mutex_wrapper(mutex_user_dict)
def _write_to_json(user_dict: Dict[str, UserDictWord], user_dict_path: Path) -> None:
"""
ユーザー辞書ファイルへのユーザー辞書データ書き込み
Parameters
----------
user_dict : Dict[str, UserDictWord]
ユーザー辞書データ
user_dict_path : Path
ユーザー辞書ファイルのパス
"""
converted_user_dict = {}
for word_uuid, word in user_dict.items():
word_dict = word.dict()
word_dict["cost"] = _priority2cost(
word_dict["context_id"], word_dict["priority"]
)
del word_dict["priority"]
converted_user_dict[word_uuid] = word_dict
# 予めjsonに変換できることを確かめる
user_dict_json = json.dumps(converted_user_dict, ensure_ascii=False)
# ユーザー辞書ファイルへの書き込み
user_dict_path.write_text(user_dict_json, encoding="utf-8")
# @mutex_wrapper(mutex_openjtalk_dict)
def update_dict(
default_dict_path: Path = default_dict_path,
user_dict_path: Path = user_dict_path,
compiled_dict_path: Path = compiled_dict_path,
) -> None:
"""
辞書の更新
Parameters
----------
default_dict_path : Path
デフォルト辞書ファイルのパス
user_dict_path : Path
ユーザー辞書ファイルのパス
compiled_dict_path : Path
コンパイル済み辞書ファイルのパス
"""
random_string = uuid4()
tmp_csv_path = compiled_dict_path.with_suffix(
f".dict_csv-{random_string}.tmp"
) # csv形式辞書データの一時保存ファイル
tmp_compiled_path = compiled_dict_path.with_suffix(
f".dict_compiled-{random_string}.tmp"
) # コンパイル済み辞書データの一時保存ファイル
try:
# 辞書.csvを作成
csv_text = ""
# デフォルト辞書データの追加
if not default_dict_path.is_file():
print("Warning: Cannot find default dictionary.", file=sys.stderr)
return
default_dict = default_dict_path.read_text(encoding="utf-8")
if default_dict == default_dict.rstrip():
default_dict += "\n"
csv_text += default_dict
# ユーザー辞書データの追加
user_dict = read_dict(user_dict_path=user_dict_path)
for word_uuid in user_dict:
word = user_dict[word_uuid]
csv_text += (
"{surface},{context_id},{context_id},{cost},{part_of_speech},"
+ "{part_of_speech_detail_1},{part_of_speech_detail_2},"
+ "{part_of_speech_detail_3},{inflectional_type},"
+ "{inflectional_form},{stem},{yomi},{pronunciation},"
+ "{accent_type}/{mora_count},{accent_associative_rule}\n"
).format(
surface=word.surface,
context_id=word.context_id,
cost=_priority2cost(word.context_id, word.priority),
part_of_speech=word.part_of_speech,
part_of_speech_detail_1=word.part_of_speech_detail_1,
part_of_speech_detail_2=word.part_of_speech_detail_2,
part_of_speech_detail_3=word.part_of_speech_detail_3,
inflectional_type=word.inflectional_type,
inflectional_form=word.inflectional_form,
stem=word.stem,
yomi=word.yomi,
pronunciation=word.pronunciation,
accent_type=word.accent_type,
mora_count=word.mora_count,
accent_associative_rule=word.accent_associative_rule,
)
# 辞書データを辞書.csv へ一時保存
tmp_csv_path.write_text(csv_text, encoding="utf-8")
# 辞書.csvをOpenJTalk用にコンパイル
# pyopenjtalk.create_user_dict(str(tmp_csv_path), str(tmp_compiled_path))
pyopenjtalk.mecab_dict_index(str(tmp_csv_path), str(tmp_compiled_path))
if not tmp_compiled_path.is_file():
raise RuntimeError("辞書のコンパイル時にエラーが発生しました。")
# コンパイル済み辞書の置き換え・読み込み
pyopenjtalk.unset_user_dict()
tmp_compiled_path.replace(compiled_dict_path)
if compiled_dict_path.is_file():
# pyopenjtalk.set_user_dict(str(compiled_dict_path.resolve(strict=True)))
pyopenjtalk.update_global_jtalk_with_user_dict(str(compiled_dict_path))
except Exception as e:
print("Error: Failed to update dictionary.", file=sys.stderr)
traceback.print_exc(file=sys.stderr)
raise e
finally:
# 後処理
if tmp_csv_path.exists():
tmp_csv_path.unlink()
if tmp_compiled_path.exists():
tmp_compiled_path.unlink()
# @mutex_wrapper(mutex_user_dict)
def read_dict(user_dict_path: Path = user_dict_path) -> Dict[str, UserDictWord]:
"""
ユーザー辞書の読み出し
Parameters
----------
user_dict_path : Path
ユーザー辞書ファイルのパス
Returns
-------
result : Dict[str, UserDictWord]
ユーザー辞書
"""
# 指定ユーザー辞書が存在しない場合、空辞書を返す
if not user_dict_path.is_file():
return {}
with user_dict_path.open(encoding="utf-8") as f:
result: Dict[str, UserDictWord] = {}
for word_uuid, word in json.load(f).items():
# cost2priorityで変換を行う際にcontext_idが必要となるが、
# 0.12以前の辞書は、context_idがハードコーディングされていたためにユーザー辞書内に保管されていない
# ハードコーディングされていたcontext_idは固有名詞を意味するものなので、固有名詞のcontext_idを補完する
if word.get("context_id") is None:
word["context_id"] = part_of_speech_data[
WordTypes.PROPER_NOUN
].context_id
word["priority"] = _cost2priority(word["context_id"], word["cost"])
del word["cost"]
result[str(UUID(word_uuid))] = UserDictWord(**word)
return result
def _create_word(
surface: str,
pronunciation: str,
accent_type: int,
word_type: Optional[WordTypes] = None,
priority: Optional[int] = None,
) -> UserDictWord:
"""
単語オブジェクトの生成
Parameters
----------
surface : str
単語情報
pronunciation : str
単語情報
accent_type : int
単語情報
word_type : Optional[WordTypes]
品詞
priority : Optional[int]
優先度
Returns
-------
: UserDictWord
単語オブジェクト
"""
if word_type is None:
word_type = WordTypes.PROPER_NOUN
if word_type not in part_of_speech_data.keys():
raise HTTPException(status_code=422, detail="不明な品詞です")
if priority is None:
priority = 5
if not MIN_PRIORITY <= priority <= MAX_PRIORITY:
raise HTTPException(status_code=422, detail="優先度の値が無効です")
pos_detail = part_of_speech_data[word_type]
return UserDictWord(
surface=surface,
context_id=pos_detail.context_id,
priority=priority,
part_of_speech=pos_detail.part_of_speech,
part_of_speech_detail_1=pos_detail.part_of_speech_detail_1,
part_of_speech_detail_2=pos_detail.part_of_speech_detail_2,
part_of_speech_detail_3=pos_detail.part_of_speech_detail_3,
inflectional_type="*",
inflectional_form="*",
stem="*",
yomi=pronunciation,
pronunciation=pronunciation,
accent_type=accent_type,
accent_associative_rule="*",
)
def apply_word(
surface: str,
pronunciation: str,
accent_type: int,
word_type: Optional[WordTypes] = None,
priority: Optional[int] = None,
user_dict_path: Path = user_dict_path,
compiled_dict_path: Path = compiled_dict_path,
) -> str:
"""
新規単語の追加
Parameters
----------
surface : str
単語情報
pronunciation : str
単語情報
accent_type : int
単語情報
word_type : Optional[WordTypes]
品詞
priority : Optional[int]
優先度
user_dict_path : Path
ユーザー辞書ファイルのパス
compiled_dict_path : Path
コンパイル済み辞書ファイルのパス
Returns
-------
word_uuid : UserDictWord
追加された単語に発行されたUUID
"""
# 新規単語の追加による辞書データの更新
word = _create_word(
surface=surface,
pronunciation=pronunciation,
accent_type=accent_type,
word_type=word_type,
priority=priority,
)
user_dict = read_dict(user_dict_path=user_dict_path)
word_uuid = str(uuid4())
user_dict[word_uuid] = word
# 更新された辞書データの保存と適用
_write_to_json(user_dict, user_dict_path)
update_dict(user_dict_path=user_dict_path, compiled_dict_path=compiled_dict_path)
return word_uuid
def rewrite_word(
word_uuid: str,
surface: str,
pronunciation: str,
accent_type: int,
word_type: Optional[WordTypes] = None,
priority: Optional[int] = None,
user_dict_path: Path = user_dict_path,
compiled_dict_path: Path = compiled_dict_path,
) -> None:
"""
既存単語の上書き更新
Parameters
----------
word_uuid : str
単語UUID
surface : str
単語情報
pronunciation : str
単語情報
accent_type : int
単語情報
word_type : Optional[WordTypes]
品詞
priority : Optional[int]
優先度
user_dict_path : Path
ユーザー辞書ファイルのパス
compiled_dict_path : Path
コンパイル済み辞書ファイルのパス
"""
word = _create_word(
surface=surface,
pronunciation=pronunciation,
accent_type=accent_type,
word_type=word_type,
priority=priority,
)
# 既存単語の上書きによる辞書データの更新
user_dict = read_dict(user_dict_path=user_dict_path)
if word_uuid not in user_dict:
raise HTTPException(
status_code=422, detail="UUIDに該当するワードが見つかりませんでした"
)
user_dict[word_uuid] = word
# 更新された辞書データの保存と適用
_write_to_json(user_dict, user_dict_path)
update_dict(user_dict_path=user_dict_path, compiled_dict_path=compiled_dict_path)
def delete_word(
word_uuid: str,
user_dict_path: Path = user_dict_path,
compiled_dict_path: Path = compiled_dict_path,
) -> None:
"""
単語の削除
Parameters
----------
word_uuid : str
単語UUID
user_dict_path : Path
ユーザー辞書ファイルのパス
compiled_dict_path : Path
コンパイル済み辞書ファイルのパス
"""
# 既存単語の削除による辞書データの更新
user_dict = read_dict(user_dict_path=user_dict_path)
if word_uuid not in user_dict:
raise HTTPException(
status_code=422, detail="IDに該当するワードが見つかりませんでした"
)
del user_dict[word_uuid]
# 更新された辞書データの保存と適用
_write_to_json(user_dict, user_dict_path)
update_dict(user_dict_path=user_dict_path, compiled_dict_path=compiled_dict_path)
def import_user_dict(
dict_data: Dict[str, UserDictWord],
override: bool = False,
user_dict_path: Path = user_dict_path,
default_dict_path: Path = default_dict_path,
compiled_dict_path: Path = compiled_dict_path,
) -> None:
"""
ユーザー辞書のインポート
Parameters
----------
dict_data : Dict[str, UserDictWord]
インポートするユーザー辞書のデータ
override : bool
重複したエントリがあった場合、上書きするかどうか
user_dict_path : Path
ユーザー辞書ファイルのパス
default_dict_path : Path
デフォルト辞書ファイルのパス
compiled_dict_path : Path
コンパイル済み辞書ファイルのパス
"""
# インポートする辞書データのバリデーション
for word_uuid, word in dict_data.items():
UUID(word_uuid)
assert isinstance(word, UserDictWord)
for pos_detail in part_of_speech_data.values():
if word.context_id == pos_detail.context_id:
assert word.part_of_speech == pos_detail.part_of_speech
assert (
word.part_of_speech_detail_1 == pos_detail.part_of_speech_detail_1
)
assert (
word.part_of_speech_detail_2 == pos_detail.part_of_speech_detail_2
)
assert (
word.part_of_speech_detail_3 == pos_detail.part_of_speech_detail_3
)
assert (
word.accent_associative_rule in pos_detail.accent_associative_rules
)
break
else:
raise ValueError("対応していない品詞です")
# 既存辞書の読み出し
old_dict = read_dict(user_dict_path=user_dict_path)
# 辞書データの更新
# 重複エントリの上書き
if override:
new_dict = {**old_dict, **dict_data}
# 重複エントリの保持
else:
new_dict = {**dict_data, **old_dict}
# 更新された辞書データの保存と適用
_write_to_json(user_dict=new_dict, user_dict_path=user_dict_path)
update_dict(
default_dict_path=default_dict_path,
user_dict_path=user_dict_path,
compiled_dict_path=compiled_dict_path,
)
def _search_cost_candidates(context_id: int) -> List[int]:
for value in part_of_speech_data.values():
if value.context_id == context_id:
return value.cost_candidates
raise HTTPException(status_code=422, detail="品詞IDが不正です")
def _cost2priority(context_id: int, cost: int) -> int:
assert -32768 <= cost <= 32767
cost_candidates = _search_cost_candidates(context_id)
# cost_candidatesの中にある値で最も近い値を元にpriorityを返す
# 参考: https://qiita.com/Krypf/items/2eada91c37161d17621d
# この関数とpriority2cost関数によって、辞書ファイルのcostを操作しても最も近いpriorityのcostに上書きされる
return MAX_PRIORITY - np.argmin(np.abs(np.array(cost_candidates) - cost)).item()
def _priority2cost(context_id: int, priority: int) -> int:
assert MIN_PRIORITY <= priority <= MAX_PRIORITY
cost_candidates = _search_cost_candidates(context_id)
return cost_candidates[MAX_PRIORITY - priority]

144
text/user_dict/part_of_speech_data.py Normal file
View File

@@ -0,0 +1,144 @@
from typing import Dict
from .word_model import (
USER_DICT_MAX_PRIORITY,
USER_DICT_MIN_PRIORITY,
PartOfSpeechDetail,
WordTypes,
)
MIN_PRIORITY = USER_DICT_MIN_PRIORITY
MAX_PRIORITY = USER_DICT_MAX_PRIORITY
part_of_speech_data: Dict[WordTypes, PartOfSpeechDetail] = {
WordTypes.PROPER_NOUN: PartOfSpeechDetail(
part_of_speech="名詞",
part_of_speech_detail_1="固有名詞",
part_of_speech_detail_2="一般",
part_of_speech_detail_3="*",
context_id=1348,
cost_candidates=[
-988,
3488,
4768,
6048,
7328,
8609,
8734,
8859,
8984,
9110,
14176,
],
accent_associative_rules=[
"*",
"C1",
"C2",
"C3",
"C4",
"C5",
],
),
WordTypes.COMMON_NOUN: PartOfSpeechDetail(
part_of_speech="名詞",
part_of_speech_detail_1="一般",
part_of_speech_detail_2="*",
part_of_speech_detail_3="*",
context_id=1345,
cost_candidates=[
-4445,
49,
1473,
2897,
4321,
5746,
6554,
7362,
8170,
8979,
15001,
],
accent_associative_rules=[
"*",
"C1",
"C2",
"C3",
"C4",
"C5",
],
),
WordTypes.VERB: PartOfSpeechDetail(
part_of_speech="動詞",
part_of_speech_detail_1="自立",
part_of_speech_detail_2="*",
part_of_speech_detail_3="*",
context_id=642,
cost_candidates=[
3100,
6160,
6360,
6561,
6761,
6962,
7414,
7866,
8318,
8771,
13433,
],
accent_associative_rules=[
"*",
],
),
WordTypes.ADJECTIVE: PartOfSpeechDetail(
part_of_speech="形容詞",
part_of_speech_detail_1="自立",
part_of_speech_detail_2="*",
part_of_speech_detail_3="*",
context_id=20,
cost_candidates=[
1527,
3266,
3561,
3857,
4153,
4449,
5149,
5849,
6549,
7250,
10001,
],
accent_associative_rules=[
"*",
],
),
WordTypes.SUFFIX: PartOfSpeechDetail(
part_of_speech="名詞",
part_of_speech_detail_1="接尾",
part_of_speech_detail_2="一般",
part_of_speech_detail_3="*",
context_id=1358,
cost_candidates=[
4399,
5373,
6041,
6710,
7378,
8047,
9440,
10834,
12228,
13622,
15847,
],
accent_associative_rules=[
"*",
"C1",
"C2",
"C3",
"C4",
"C5",
],
),
}

124
text/user_dict/word_model.py Normal file
View File

@@ -0,0 +1,124 @@
from enum import Enum
from re import findall, fullmatch
from typing import List, Optional
from pydantic import BaseModel, Field, validator
USER_DICT_MIN_PRIORITY = 0
USER_DICT_MAX_PRIORITY = 10
class UserDictWord(BaseModel):
"""
辞書のコンパイルに使われる情報
"""
surface: str = Field(title="表層形")
priority: int = Field(
title="優先度", ge=USER_DICT_MIN_PRIORITY, le=USER_DICT_MAX_PRIORITY
)
context_id: int = Field(title="文脈ID", default=1348)
part_of_speech: str = Field(title="品詞")
part_of_speech_detail_1: str = Field(title="品詞細分類1")
part_of_speech_detail_2: str = Field(title="品詞細分類2")
part_of_speech_detail_3: str = Field(title="品詞細分類3")
inflectional_type: str = Field(title="活用型")
inflectional_form: str = Field(title="活用形")
stem: str = Field(title="原形")
yomi: str = Field(title="読み")
pronunciation: str = Field(title="発音")
accent_type: int = Field(title="アクセント型")
mora_count: Optional[int] = Field(title="モーラ数", default=None)
accent_associative_rule: str = Field(title="アクセント結合規則")
class Config:
validate_assignment = True
@validator("surface")
def convert_to_zenkaku(cls, surface):
return surface.translate(
str.maketrans(
"".join(chr(0x21 + i) for i in range(94)),
"".join(chr(0xFF01 + i) for i in range(94)),
)
)
@validator("pronunciation", pre=True)
def check_is_katakana(cls, pronunciation):
if not fullmatch(r"[ァ-ヴー]+", pronunciation):
raise ValueError("発音は有効なカタカナでなくてはいけません。")
sutegana = ["ァ", "ィ", "ゥ", "ェ", "ォ", "ャ", "ュ", "ョ", "ヮ", "ッ"]
for i in range(len(pronunciation)):
if pronunciation[i] in sutegana:
# 「キャット」のように、捨て仮名が連続する可能性が考えられるので、
# 「ッ」に関しては「ッ」そのものが連続している場合と、「ッ」の後にほかの捨て仮名が連続する場合のみ無効とする
if i < len(pronunciation) - 1 and (
pronunciation[i + 1] in sutegana[:-1]
or (
pronunciation[i] == sutegana[-1]
and pronunciation[i + 1] == sutegana[-1]
)
):
raise ValueError("無効な発音です。(捨て仮名の連続)")
if pronunciation[i] == "ヮ":
if i != 0 and pronunciation[i - 1] not in ["ク", "グ"]:
raise ValueError(
"無効な発音です。(「くゎ」「ぐゎ」以外の「ゎ」の使用)"
)
return pronunciation
@validator("mora_count", pre=True, always=True)
def check_mora_count_and_accent_type(cls, mora_count, values):
if "pronunciation" not in values or "accent_type" not in values:
# 適切な場所でエラーを出すようにする
return mora_count
if mora_count is None:
rule_others = (
"[イ][ェ]|[ヴ][ャュョ]|[トド][ゥ]|[テデ][ィャュョ]|[デ][ェ]|[クグ][ヮ]"
)
rule_line_i = "[キシチニヒミリギジビピ][ェャュョ]"
rule_line_u = "[ツフヴ][ァ]|[ウスツフヴズ][ィ]|[ウツフヴ][ェォ]"
rule_one_mora = "[ァ-ヴー]"
mora_count = len(
findall(
f"(?:{rule_others}|{rule_line_i}|{rule_line_u}|{rule_one_mora})",
values["pronunciation"],
)
)
if not 0 <= values["accent_type"] <= mora_count:
raise ValueError(
"誤ったアクセント型です({})。 expect: 0 <= accent_type <= {}".format(
values["accent_type"], mora_count
)
)
return mora_count
class PartOfSpeechDetail(BaseModel):
"""
品詞ごとの情報
"""
part_of_speech: str = Field(title="品詞")
part_of_speech_detail_1: str = Field(title="品詞細分類1")
part_of_speech_detail_2: str = Field(title="品詞細分類2")
part_of_speech_detail_3: str = Field(title="品詞細分類3")
# context_idは辞書の左・右文脈IDのこと
# https://github.com/VOICEVOX/open_jtalk/blob/427cfd761b78efb6094bea3c5bb8c968f0d711ab/src/mecab-naist-jdic/_left-id.def # noqa
context_id: int = Field(title="文脈ID")
cost_candidates: List[int] = Field(title="コストのパーセンタイル")
accent_associative_rules: List[str] = Field(title="アクセント結合規則の一覧")
class WordTypes(str, Enum):
"""
fastapiでword_type引数を検証する時に使用するクラス
"""
PROPER_NOUN = "PROPER_NOUN"
COMMON_NOUN = "COMMON_NOUN"
VERB = "VERB"
ADJECTIVE = "ADJECTIVE"
SUFFIX = "SUFFIX"