修复ju, qu, xu, yu, yi和wu的韵母

pypinyin/__init__.py

@@ -196,6 +196,12 @@ def initial(pinyin):
             return i
     return ''
 
+U_FINALS_EXCEPTIONS_MAP = {
+   u'ū': u'ǖ',
+   u'ú': u'ǘ',
+   u'ǔ': u'ǚ',
+   u'ù': u'ǜ',
+}
 
 def final(pinyin):
     """获取单个拼音中的韵母.
@@ -207,9 +213,30 @@ def final(pinyin):
     """
     initial_ = initial(pinyin) or None
     if not initial_:
-        return pinyin
+        return no_initial_final(pinyin)
+    # 特例 j/q/x
+    m = re.match(u'^(j|q|x)(ū|ú|ǔ|ù)$', pinyin)
+    if m:
+        return (U_FINALS_EXCEPTIONS_MAP[m.group(2)])
+    pinyin = re.sub('^(j|q|x)u(\d?)$', r'\1v\2', pinyin)
+
     return ''.join(pinyin.split(initial_, 1))
 
+def no_initial_final(pinyin):
+    # 特例 y/w
+    if pinyin.startswith('y'):
+        if pinyin.startswith('yu'):
+            pinyin = 'v' + pinyin[2:]
+        elif pinyin.startswith('yi'):
+            pinyin = pinyin[1:]
+        else:                
+            pinyin = 'i' + pinyin[1:]
+    if pinyin.startswith('w'):
+        if pinyin.startswith('wu'):
+            pinyin = pinyin[1:]
+        else:
+            pinyin = 'u' + pinyin[1:]
+    return pinyin    
 
 def toFixed(pinyin, style):
     """根据拼音风格格式化带声调的拼音.

tests/test_pinyin.py

@@ -222,14 +222,34 @@ def test_simple_seg():
     ],
     # 误把 yu 放到声母列表了
     ['鱼', {'style': TONE2}, ['yu2']],
-    ['鱼', {'style': FINALS}, ['yu']],
+    ['鱼', {'style': FINALS}, ['v']],    
     ['雨', {'style': TONE2}, ['yu3']],
-    ['雨', {'style': FINALS}, ['yu']],
+    ['雨', {'style': FINALS}, ['v']],    
     ['元', {'style': TONE2}, ['yua2n']],
-    ['元', {'style': FINALS}, ['yuan']],
-    # y, w 也不是拼音
+    ['元', {'style': FINALS}, ['van']],
+
+    # y, w 也不是拼音, yu的韵母是v, yi的韵母是i, wu的韵母是u
     ['呀', {'style': INITIALS}, ['']],
+    ['呀', {'style': TONE2}, ['ya1']],
+    ['呀', {'style': FINALS}, ['ia']],    
     ['无', {'style': INITIALS}, ['']],
+    ['无', {'style': TONE2}, ['wu2']],
+    ['无', {'style': FINALS}, ['u']],
+    ['衣', {'style': TONE2}, ['yi1']],
+    ['衣', {'style': FINALS}, ['i']],    
+    ['万', {'style': TONE2}, ['wa4n']],
+    ['万', {'style': FINALS}, ['uan']],    
+    # ju, qu, xu 的韵母应该是 v
+    ['具', {'style': FINALS_TONE}, ['ǜ']],
+    ['具', {'style': FINALS_TONE2}, ['v4']],
+    ['具', {'style': FINALS}, ['v']],
+    ['取', {'style': FINALS_TONE}, ['ǚ']],    
+    ['取', {'style': FINALS_TONE2}, ['v3']],
+    ['取', {'style': FINALS}, ['v']],    
+    ['徐', {'style': FINALS_TONE}, ['ǘ']],
+    ['徐', {'style': FINALS_TONE2}, ['v2']],
+    ['徐', {'style': FINALS}, ['v']],
+
 ]