make aeo prior to iu, fix #8

pinyin/pinyin.py

@@ -1,13 +1,13 @@
 # -*- coding: utf-8 -*-
 
 import os
+import itertools
 import unicodedata
+
 from ._compat import u
 
 __all__ = ['get', 'get_pinyin', 'get_initial']
 
-tonemarks = ["", u("̄"), u("́"), u("̌"), u("̀"), ""]
-
 
 # init pinyin dict
 pinyin_dict = {}
@@ -35,9 +35,10 @@ def _pinyin_generator(chars, format):
         elif format == "numerical":
             pinyin += str(tone)
         elif format == "diacritical":
-            # Find first vowel -- we should put the diacritical mark
-            # just after
-            vowel = pinyin.index(next(x for x in pinyin if x in "aeiou")) + 1
+            # Find first vowel -- where we should put the diacritical mark
+            vowels = itertools.chain((c for c in pinyin if c in "aeo"),
+                                     (c for c in pinyin if c in "iu"))
+            vowel = pinyin.index(next(vowels)) + 1
             pinyin = pinyin[:vowel] + tonemarks[tone] + pinyin[vowel:]
         else:
             error = "Format must be one of: numerical/diacritical/strip"
@@ -65,3 +66,5 @@ def get_initial(s, delimiter=' '):
     """
     initials = (p[0] for p in _pinyin_generator(u(s), format="strip"))
     return delimiter.join(initials)
+
+tonemarks = ["", u("̄"), u("́"), u("̌"), u("̀"), ""]

setup.py

@@ -5,7 +5,7 @@
 
 setup(
     name='pinyin',
-    version='0.3',
+    version='0.3.1',
     description='Translate chinese chars to pinyin based on Mandarin.dat',
     author='Lx Yu',
     author_email='[email protected]',

test_pinyin.py

@@ -41,6 +41,12 @@ def test_get_initial(self):
     def test_mixed_chinese_english_input(self):
         self.assertEqual(pinyin.get('hi你好'), u('hinǐhǎo'))
 
+    def test_correct_diacritical(self):
+        self.assertEqual(pinyin.get("操"), u("cāo"))
+        self.assertEqual(pinyin.get("小"), u("xiǎo"))
+        self.assertEqual(pinyin.get("绝"), u("jué"))
+        self.assertEqual(pinyin.get("被"), u("bèi"))
+
 
 if __name__ == '__main__':
     unittest.main()