Merge remote-tracking branch 'origin/develop'

CHANGELOG.rst

@@ -2,6 +2,21 @@ Changelog
 ---------
 
 
+`0.47.1`_ (2022-08-21)
++++++++++++++++++++++++++
+
+* **[Bugfixed]** 修复无声母和韵母的场景下指定 neutral_tone_with_five=True 会返回 ``5`` 作为拼音的问题:
+
+  .. code-block:: python
+
+      # 修复前
+      >>> lazy_pinyin('嗯', style=Style.FINALS_TONE3, neutral_tone_with_five=True)
+      ['5']
+      # 修复后
+      >>> lazy_pinyin('嗯', style=Style.FINALS_TONE3, neutral_tone_with_five=True)
+      ['']
+
+
 `0.47.0`_ (2022-07-30)
 +++++++++++++++++++++++++
 
@@ -993,3 +1008,4 @@ __ https://github.com/mozillazg/python-pinyin/issues/8
 .. _0.45.0: https://github.com/mozillazg/python-pinyin/compare/v0.44.0...v0.45.0
 .. _0.46.0: https://github.com/mozillazg/python-pinyin/compare/v0.45.0...v0.46.0
 .. _0.47.0: https://github.com/mozillazg/python-pinyin/compare/v0.46.0...v0.47.0
+.. _0.47.1: https://github.com/mozillazg/python-pinyin/compare/v0.47.0...v0.47.1

pypinyin/contrib/neutral_tone.py

@@ -56,6 +56,8 @@ def post_convert_style(self, han, orig_pinyin, converted_pinyin,
 
         if pre_data is not None:
             converted_pinyin = pre_data
+        if not converted_pinyin:    # 空字符串
+            return converted_pinyin
         # 有声调，跳过
         if _re_number.search(converted_pinyin):
             return converted_pinyin

pypinyin/converter.py

@@ -52,12 +52,19 @@ def convert(self, words, style, heteronym, errors, strict, **kwargs):
         if RE_HANS.match(words):
             pys = self._phrase_pinyin(words, style=style, heteronym=heteronym,
                                       errors=errors, strict=strict)
-            return _remove_dup_and_empty(pys)
+            post_data = self.post_pinyin(words, heteronym, pys)
+            if post_data is not None:
+                pys = post_data
+
+            pys = self.convert_styles(
+                pys, words, style, heteronym, errors, strict)
+
+        else:
+            py = self.handle_nopinyin(words, style=style, errors=errors,
+                                      heteronym=heteronym, strict=strict)
+            if py:
+                pys.extend(py)
 
-        py = self.handle_nopinyin(words, style=style, errors=errors,
-                                  heteronym=heteronym, strict=strict)
-        if py:
-            pys.extend(py)
         return _remove_dup_and_empty(pys)
 
     def pre_convert_style(self, han, orig_pinyin, style, strict, **kwargs):
@@ -84,7 +91,7 @@ def convert_style(self, han, orig_pinyin, style, strict, **kwargs):
         转换风格前会调用 ``pre_convert_style`` 方法，
         转换后会调用 ``post_convert_style`` 方法。
 
-        :param han: 要处理的汉字
+        :param han: 要处理的单个汉字
         :param orig_pinyin: 汉字对应的原始带声调拼音
         :param style: 拼音风格
         :param strict: 只获取声母或只获取韵母相关拼音风格的返回结果
@@ -251,10 +258,11 @@ def _phrase_pinyin(self, phrase, style, heteronym, errors, strict):
                 py = self._single_pinyin(han, style, heteronym, errors, strict)
                 pinyin_list.extend(py)
 
-        post_data = self.post_pinyin(phrase, heteronym, pinyin_list)
-        if post_data is not None:
-            pinyin_list = post_data
+        return pinyin_list
 
+    def convert_styles(self, pinyin_list, phrase, style, heteronym, errors,
+                       strict, **kwargs):
+        """转换多个汉字的拼音结果的风格"""
         for idx, item in enumerate(pinyin_list):
             han = phrase[idx]
             if heteronym:

pypinyin/converter.pyi

@@ -55,6 +55,11 @@ class DefaultConverter(Converter):
                     pinyin: TPinyinResult,
                     **kwargs: Any) -> Union[TPinyinResult, None]: ...
 
+    def convert_styles(self, pinyin_list: TPinyinResult,
+                       phrase: Text, style: TStyle, heteronym: bool,
+                       errors: TErrors, strict: bool, **kwargs: Any,
+                       ) -> TPinyinResult: ...
+
     def _phrase_pinyin(self, phrase: Text, style: TStyle, heteronym: bool,
                        errors: TErrors, strict: bool
                        ) -> TPinyinResult: ...

tests/test_others.py

@@ -2,6 +2,8 @@
 # -*- coding: utf-8 -*-
 from __future__ import unicode_literals
 
+from pytest import mark
+
 from pypinyin import pinyin, Style, lazy_pinyin, slug
 from pypinyin.seg.simpleseg import simple_seg
 
@@ -49,6 +51,22 @@ def test_issue_266():
     assert lazy_pinyin('呣', style=Style.FINALS, strict=False) == ['m']
 
 
+@mark.parametrize('neutral_tone_with_five', [True, False])
+def test_issue_284(neutral_tone_with_five):
+    assert lazy_pinyin('嗯', style=Style.FINALS_TONE3,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+    assert lazy_pinyin('呣', style=Style.FINALS_TONE3,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+    assert lazy_pinyin('嗯', style=Style.FINALS,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+    assert lazy_pinyin('呣', style=Style.FINALS,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+    assert lazy_pinyin('嗯', style=Style.INITIALS,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+    assert lazy_pinyin('呣', style=Style.INITIALS,
+                       neutral_tone_with_five=neutral_tone_with_five) == ['']
+
+
 if __name__ == '__main__':
     import pytest
     pytest.cmdline.main()