Merge pull request mozillazg#110 from mozillazg/typing

增加类型注解

.travis.yml

@@ -25,6 +25,7 @@ install:
 
 script:
   - pre-commit run --all-files
+  - mypy pypinyin
   - tox -e $TOX_ENV
   - python setup.py install
   - pypinyin hello

CHANGELOG.rst

@@ -1,6 +1,11 @@
 Changelog
 ---------
 
+0.28.0 (2017-mm-dd)
++++++++++++++++++++++
+
+* **[New]** 增加类型注解。
+
 
 0.27.0 (2017-10-28)
 +++++++++++++++++++++

MANIFEST.in

@@ -1 +1,2 @@
 include README.rst LICENSE.txt CHANGELOG.rst
+recursive-include pypinyin *.pyi

Makefile

@@ -7,6 +7,7 @@ help:
 	@echo "gen_data         gen pinyin data"
 	@echo "gen_pinyin_dict  gen single hanzi pinyin dict"
 	@echo "gen_phrases_dict gen phrase hanzi pinyin dict"
+	@echo "lint             run lint"
 
 .PHONY: test
 test:
@@ -43,3 +44,7 @@ gen_pinyin_dict:
 .PHONY: gen_phrases_dict
 gen_phrases_dict:
 	python gen_phrases_dict.py phrase-pinyin-data/pinyin.txt pypinyin/phrases_dict.py
+
+.PHONY: lint
+lint:
+	mypy pypinyin

pypinyin/__init__.pyi

@@ -0,0 +1,50 @@
+from typing import Any
+
+from . import compat
+from . import constants
+from . import core
+
+__title__ = ...  # type: Any
+__version__ = ...  # type: Any
+__author__ = ...  # type: Any
+__license__ = ...  # type: Any
+__copyright__ = ...  # type: Any
+__all__ = ...  # type: Any
+
+PY2 = compat.PY2
+
+Style = constants.Style
+STYLE_NORMAL = constants.STYLE_NORMAL
+NORMAL = constants.NORMAL
+STYLE_TONE = constants.STYLE_TONE
+TONE = constants.TONE
+STYLE_TONE2 = constants.STYLE_TONE2
+TONE2 = constants.TONE2
+STYLE_TONE3 = constants.STYLE_TONE3
+TONE3 = constants.TONE3
+STYLE_INITIALS = constants.STYLE_INITIALS
+INITIALS = constants.INITIALS
+STYLE_FIRST_LETTER = constants.STYLE_FIRST_LETTER
+FIRST_LETTER = constants.FIRST_LETTER
+STYLE_FINALS = constants.STYLE_FINALS
+FINALS = constants.FINALS
+STYLE_FINALS_TONE = constants.STYLE_FINALS_TONE
+FINALS_TONE = constants.FINALS_TONE
+STYLE_FINALS_TONE2 = constants.STYLE_FINALS_TONE2
+FINALS_TONE2 = constants.FINALS_TONE2
+STYLE_FINALS_TONE3 = constants.STYLE_FINALS_TONE3
+FINALS_TONE3 = constants.FINALS_TONE3
+STYLE_BOPOMOFO = constants.STYLE_BOPOMOFO
+BOPOMOFO = constants.BOPOMOFO
+STYLE_BOPOMOFO_FIRST = constants.STYLE_BOPOMOFO_FIRST
+BOPOMOFO_FIRST = constants.BOPOMOFO_FIRST
+STYLE_CYRILLIC = constants.STYLE_CYRILLIC
+CYRILLIC = constants.CYRILLIC
+STYLE_CYRILLIC_FIRST = constants.STYLE_CYRILLIC_FIRST
+CYRILLIC_FIRST = constants.CYRILLIC_FIRST
+
+pinyin = core.pinyin
+lazy_pinyin = core.lazy_pinyin
+slug = core.slug
+load_single_dict = core.load_single_dict
+load_phrases_dict = core.load_phrases_dict

pypinyin/__main__.py

@@ -1,8 +1,6 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
-import sys
-
 from pypinyin.runner import main
 
 if __name__ == '__main__':
-    sys.exit(main())
+    main()

pypinyin/constants.pyi

@@ -0,0 +1,72 @@
+from enum import IntEnum, unique
+from typing import Dict, List, Any, Text
+
+PHRASES_DICT = ...  # type: Dict[Text, List[List[Text]]]
+
+PINYIN_DICT = ...  # type: Dict[int, Text]
+
+RE_TONE2 = ...  # type: Any
+
+RE_HANS = ...  # type: Any
+
+
+@unique
+class Style(IntEnum):
+
+    NORMAL = ...
+
+    TONE = ...
+
+    TONE2 = ...
+
+    TONE3 = ...
+
+    INITIALS = ...
+
+    FIRST_LETTER = ...
+
+    FINALS = ...
+
+    FINALS_TONE = ...
+
+    FINALS_TONE2 = ...
+
+    FINALS_TONE3 = ...
+
+    BOPOMOFO = ...
+
+    BOPOMOFO_FIRST = ...
+
+    CYRILLIC = ...
+
+    CYRILLIC_FIRST = ...
+
+
+NORMAL = ...  # type: Style
+STYLE_NORMAL = ...  # type: Style
+TONE = ...  # type: Style
+STYLE_TONE = ...  # type: Style
+TONE2  = ...  # type: Style
+STYLE_TONE2 = ...  # type: Style
+TONE3 = ...  # type: Style
+STYLE_TONE3 = ...  # type: Style
+INITIALS = ...  # type: Style
+STYLE_INITIALS = ...  # type: Style
+FIRST_LETTER  = ...  # type: Style
+STYLE_FIRST_LETTER = ...  # type: Style
+FINALS = ...  # type: Style
+STYLE_FINALS = ...  # type: Style
+FINALS_TONE  = ...  # type: Style
+STYLE_FINALS_TONE = ...  # type: Style
+FINALS_TONE2 = ...  # type: Style
+STYLE_FINALS_TONE2 = ...  # type: Style
+FINALS_TONE3 = ...  # type: Style
+STYLE_FINALS_TONE3 = ...  # type: Style
+BOPOMOFO = ...  # type: Style
+STYLE_BOPOMOFO = ...  # type: Style
+BOPOMOFO_FIRST = ...  # type: Style
+STYLE_BOPOMOFO_FIRST = ...  # type: Style
+CYRILLIC = ...  # type: Style
+STYLE_CYRILLIC = ...  # type: Style
+CYRILLIC_FIRST = ...  # type: Style
+STYLE_CYRILLIC_FIRST = ...  # type: Style

pypinyin/contrib/mmseg.pyi

@@ -0,0 +1,29 @@
+from typing import Iterator
+from typing import Text
+
+
+class Seg(object):
+    """最大正向匹配分词
+
+    :type prefix_set: PrefixSet
+    """
+    def __init__(self, prefix_set: PrefixSet) -> None: ...
+
+    def cut(self, text: Text) -> Iterator[Text]: ...
+
+    def train(self, words: Iterator[Text]) -> None: ...
+
+
+class PrefixSet(object):
+    def __init__(self): ...
+
+    def train(self, word_s: Iterator[Text]) -> None: ...
+
+    def __contains__(self, key: Text) -> bool: ...
+
+
+p_set = ...  # type: PrefixSet
+seg = ...  # type: Seg
+
+
+def retrain(seg_instance: Seg) -> None: ...

pypinyin/core.pyi

@@ -0,0 +1,83 @@
+from typing import List
+from typing import Dict
+from typing import Union
+from typing import Callable
+from typing import Optional
+from typing import Text
+
+from pypinyin.constants import Style
+
+
+TStyle = Style
+TErrors = Union[Callable[[Text], Text], Text]
+TPinyinResult = List[List[Text]]
+
+
+def seg(hans: Text) -> List[Text]: ...
+
+
+def load_single_dict(pinyin_dict: Dict[int, Text],
+                     style: str = ...) -> None: ...
+
+
+def load_phrases_dict(phrases_dict: Dict[Text, List[List[Text]]],
+                      style: str = ...
+                      ) -> None: ...
+
+
+def to_fixed(pinyin: Text, style: TStyle,
+             strict: bool = ...) -> Text: ...
+
+
+def _handle_nopinyin_char(chars: Text, errors: TErrors = ...
+                          ) -> Optional[Text]: ...
+
+
+def handle_nopinyin(chars: Text, errors: TErrors = ...
+                    ) -> List[Text]: ...
+
+
+def single_pinyin(han: Text, style: TStyle, heteronym: bool,
+                  errors: TErrors = ...,
+                  strict: bool = ...
+                  ) -> List[Text]: ...
+
+
+def phrase_pinyin(phrase: Text,
+                  style: TStyle,
+                  heteronym: bool,
+                  errors: TErrors = ...,
+                  strict: bool = ...
+                  ) -> List[List[Text]]: ...
+
+
+def _pinyin(words: Text,
+            style: TStyle,
+            heteronym: bool,
+            errors: TErrors,
+            strict: bool = ...
+            ) -> List[List[Text]]:...
+
+
+def pinyin(hans: Union[List[Text], Text],
+           style: TStyle = ...,
+           heteronym: bool = ...,
+           errors: TErrors = ...,
+           strict: bool = ...
+           ) -> List[List[Text]]: ...
+
+
+def slug(hans: Union[List[Text], Text],
+         style: TStyle = ...,
+         heteronym: bool = ...,
+         separator: Text = ...,
+         errors: TErrors = ...,
+         strict: bool = ...
+         ) -> Text: ...
+
+
+def lazy_pinyin(hans: Union[List[Text], Text],
+                style: TStyle = ...,
+                errors: TErrors = ...,
+                strict: bool = ...
+                ) -> List[Text]: ...

pypinyin/phonetic_symbol.pyi

@@ -0,0 +1,5 @@
+from typing import Dict, Text
+
+
+phonetic_symbol = ...  # type: Dict[Text, Text]
+phonetic_symbol_reverse = ...  # type: Dict[Text, Text]

pypinyin/phrases_dict.pyi

@@ -0,0 +1,3 @@
+from typing import Dict, List, Text
+
+phrases_dict = ...  # type: Dict[Text, List[List[Text]]]

pypinyin/pinyin_dict.pyi

@@ -0,0 +1,3 @@
+from typing import Dict, Text
+
+pinyin_dict = ...  # type: Dict[int, Text]

pypinyin/runner.py

@@ -3,6 +3,7 @@
 
 from __future__ import unicode_literals
 from argparse import ArgumentParser
+import logging
 import sys
 
 import pypinyin
@@ -81,7 +82,6 @@ def get_parser():
 
 
 def main():
-    import logging
     # 禁用除 CRITICAL 外的日志消息
     logging.disable(logging.CRITICAL)
 

pypinyin/runner.pyi

@@ -0,0 +1,12 @@
+from argparse import ArgumentParser
+from typing import Union, Text, ByteString
+
+
+class NullWriter(object):
+    def write(self, string: Union[Text, ByteString]) -> None: ...
+
+
+def get_parser() -> ArgumentParser: ...
+
+
+def main() -> None: ...

pypinyin/standard.pyi

@@ -0,0 +1,18 @@
+from typing import Text
+
+def convert_zero_consonant(pinyin: Text) -> Text: ...
+
+
+def convert_uv(pinyin: Text) -> Text: ...
+
+
+def convert_iou(pinyin: Text) -> Text: ...
+
+
+def convert_uei(pinyin: Text) -> Text: ...
+
+
+def convert_uen(pinyin: Text) -> Text: ...
+
+
+def convert_finals(pinyin: Text) -> Text: ...

pypinyin/style/__init__.pyi

@@ -0,0 +1,20 @@
+# -*- coding: utf-8 -*-
+from typing import Any, Optional, Callable, Dict, Text, Union
+
+from pypinyin.constants import Style
+
+TStyle = Style
+TRegisterFunc = Optional[Callable[[Text, Dict[Any, Any]], Optional[Text]]]
+TWrapperFunc = Optional[Callable[[Text, Dict[Any, Any]], Optional[Text]]]
+
+_registry = {}  # type: Dict[Union[TStyle, int, str, Any], TRegisterFunc]
+
+
+def convert(pinyin: Text, style: TStyle, strict: bool,
+            default: Optional[Text] = ..., **kwargs: Any) -> Any: ...
+
+
+def register(style: Union[TStyle, int, str, Any],
+             func: TRegisterFunc = ...) -> TWrapperFunc: ...
+
+def auto_discover() -> None: ...

pypinyin/style/_constants.pyi

@@ -0,0 +1,19 @@
+from typing import Any, List, Dict, Text
+
+
+_INITIALS = ...  # type: List[Text]
+
+_INITIALS_NOT_STRICT = ...  # type: List[Text]
+
+
+PHONETIC_SYMBOL_DICT = ...  # type: Dict[Text, Text]
+
+RE_PHONETIC_SYMBOL = ...  # type : Any
+
+
+RE_TONE2 = ...  # type : Any
+
+RE_TONE3 = ...  # type : Any
+
+
+RE_NUMBER = ...  # type: Any