修复 m̄ ê̄ ế ê̌ ề 这几个音无法转换为不含声调结果的问题

CHANGELOG.rst

@@ -1,6 +1,12 @@
 Changelog
 ---------
 
+`0.35.4`_ (2019-07-xx)
++++++++++++++++++++++++
+
+* **[Bugfixed]** 修复 ``m̄`` ``ê̄``  ``ế`` ``ê̌`` ``ề`` 这几个音无法转换为不含声调结果的问题。
+
+
 `0.35.3`_ (2019-05-11)
 ++++++++++++++++++++++++
 
@@ -795,3 +801,4 @@ __ https://github.com/mozillazg/python-pinyin/issues/8
 .. _0.35.1: https://github.com/mozillazg/python-pinyin/compare/v0.35.0...v0.35.1
 .. _0.35.2: https://github.com/mozillazg/python-pinyin/compare/v0.35.1...v0.35.2
 .. _0.35.3: https://github.com/mozillazg/python-pinyin/compare/v0.35.2...v0.35.3
+.. _0.35.4: https://github.com/mozillazg/python-pinyin/compare/v0.35.3...v0.35.4

pypinyin/phonetic_symbol.py

@@ -41,7 +41,13 @@
     "ň": "n3",
     "ǹ": "n4",
 
+    "m̄": "m1",  # len('m̄') == 2
     "ḿ": "m2",
     "m̀": "m4",  # len("m̀") == 2
+
+    "ê̄": "ê1",  # len('ê̄') == 2
+    "ế": "ê2",
+    "ê̌": "ê3",  # len('ê̌') == 2
+    "ề": "ê4",
 }
 phonetic_symbol_reverse = dict((v, k) for k, v in phonetic_symbol.items())

pypinyin/style/_constants.py

@@ -11,6 +11,11 @@
 
 # 带声调字符与数字表示声调的对应关系
 PHONETIC_SYMBOL_DICT = phonetic_symbol.phonetic_symbol.copy()
+PHONETIC_SYMBOL_DICT_KEY_LENGTH_NOT_ONE = dict(
+    (k, v)
+    for k, v in PHONETIC_SYMBOL_DICT.items()
+    if len(k) > 1
+)
 # 匹配带声调字符的正则表达式
 RE_PHONETIC_SYMBOL = re.compile(
     r'[{0}]'.format(

pypinyin/style/_constants.pyi

@@ -7,6 +7,7 @@ _INITIALS_NOT_STRICT = ...  # type: List[Text]
 
 
 PHONETIC_SYMBOL_DICT = ...  # type: Dict[Text, Text]
+PHONETIC_SYMBOL_DICT_KEY_LENGTH_NOT_ONE = ...  # type: Dict[Text, Text]
 
 RE_PHONETIC_SYMBOL = ...  # type : Any
 

pypinyin/style/_utils.py

@@ -5,6 +5,7 @@
 from pypinyin.style._constants import (
     _INITIALS, _INITIALS_NOT_STRICT,
     RE_PHONETIC_SYMBOL, PHONETIC_SYMBOL_DICT,
+    PHONETIC_SYMBOL_DICT_KEY_LENGTH_NOT_ONE,
     RE_NUMBER
 )
 
@@ -57,24 +58,23 @@ def _replace(match):
         return PHONETIC_SYMBOL_DICT[symbol]
 
     # 替换拼音中的带声调字符
-    return RE_PHONETIC_SYMBOL.sub(_replace, pinyin).replace('m̀', 'm4')
+    value = RE_PHONETIC_SYMBOL.sub(_replace, pinyin)
+    for symbol, to in PHONETIC_SYMBOL_DICT_KEY_LENGTH_NOT_ONE.items():
+        value = value.replace(symbol, to)
+
+    return value
 
 
 def replace_symbol_to_no_symbol(pinyin):
     """把带声调字符替换为没有声调的字符"""
-    def _replace(match):
-        symbol = match.group(0)  # 带声调的字符
-        # 去掉声调: a1 -> a
-        return RE_NUMBER.sub(r'', PHONETIC_SYMBOL_DICT[symbol])
-
-    # 替换拼音中的带声调字符
-    return RE_PHONETIC_SYMBOL.sub(_replace, pinyin).replace('m̀', 'm')
+    value = replace_symbol_to_number(pinyin)
+    return RE_NUMBER.sub('', value)
 
 
 def has_finals(pinyin):
     """判断是否有韵母"""
-    # 鼻音: 'ḿ', 'm̀', 'ń', 'ň', 'ǹ ' 没有韵母
-    for symbol in ['ḿ', 'm̀', 'ń', 'ň', 'ǹ']:
+    # 鼻音: 'm̄', 'ḿ', 'm̀', 'ń', 'ň', 'ǹ ' 没有韵母
+    for symbol in ['m̄', 'ḿ', 'm̀', 'ń', 'ň', 'ǹ', 'ê̄', 'ế', 'ê̌', 'ề']:
         if symbol in pinyin:
             return False
 

tests/test_pinyin.py

@@ -9,7 +9,7 @@
     pinyin, slug, lazy_pinyin, load_single_dict,
     load_phrases_dict, NORMAL, TONE, TONE2, TONE3, INITIALS,
     FIRST_LETTER, FINALS, FINALS_TONE, FINALS_TONE2, FINALS_TONE3,
-    BOPOMOFO, BOPOMOFO_FIRST, CYRILLIC, CYRILLIC_FIRST
+    BOPOMOFO, BOPOMOFO_FIRST, CYRILLIC, CYRILLIC_FIRST, Style
 )
 from pypinyin.compat import SUPPORT_UCS4
 from pypinyin.core import seg
@@ -545,6 +545,20 @@ def test_m4():
         han, heteronym=True, style=FINALS_TONE3) == [['m2', 'ou2', 'm4']]
 
 
+@pytest.mark.parametrize('han,style,expect', [
+    ['呣', Style.TONE, ['ḿ', 'm̀']],
+    ['呣', Style.TONE2, ['m2', 'm4']],
+    ['嘸', Style.TONE, ['m̄', 'ḿ']],
+    ['嘸', Style.TONE2, ['m1', 'm2']],
+    ['誒', Style.TONE, ['ê̄', 'ế', 'ê̌', 'ề']],
+    ['誒', Style.TONE2, ['ê1', 'ê2', 'ê3', 'ê4']],
+])
+def test_m_e(han, style, expect):
+    result = pinyin(han, style=style, heteronym=True)
+    assert len(result) == 1
+    assert (set(result[0]) & set(expect)) == set(expect)
+
+
 if __name__ == '__main__':
     import pytest
     pytest.cmdline.main()