cloned from 'bitbucket', 2013/08/13

1 files changed, 70 insertions, 0 deletions

diff --git a/97suifangqa/apps/utils/xpinyin/__init__.py b/97suifangqa/apps/utils/xpinyin/__init__.py
new file mode 100644
index 0000000..291c10a
--- /dev/null
+++ b/97suifangqa/apps/utils/xpinyin/__init__.py
@@ -0,0 +1,70 @@
+# -*- coding: utf-8 -*-
+import os.path
+
+
+class Pinyin(object):
+    """translate chinese hanzi to pinyin by python, inspired by flyerhzm’s
+    `chinese\_pinyin`_ gem
+
+    usage
+    -----
+    ::
+        In [1]: from xpinyin import Pinyin
+        In [2]: p = Pinyin()
+        In [3]: p.get_pinyin(u"上海")
+        Out[3]: 'shang-hai'
+        In [4]: p.get_initials(u"上")
+        Out[4]: 'S'
+    请输入utf8编码汉字
+    .. _chinese\_pinyin: https://github.com/flyerhzm/chinese_pinyin
+    """
+
+    data_path = os.path.join(os.path.dirname(os.path.abspath(__file__)),
+                             'Mandarin.dat')
+
+    def __init__(self, data_path=data_path):
+        self.dict = {}
+        for line in open(data_path):
+            k, v = line.split('\t')
+            self.dict[k] = v
+
+    def get_pinyin(self, chars=u'你好', splitter=u'-'):
+        result = []
+        flag = 1
+        for char in chars:
+            key = "%X" % ord(char)
+            try:
+                result.append(self.dict[key].split(" ")[0].strip()[:-1]
+                              .lower())
+                flag = 1
+            except KeyError:
+                if flag:
+                    result.append(char)
+                else:
+                    result[-1] += char
+                flag = 0
+
+        return splitter.join(result)
+
+    # def get_initials(self, char=u'你'):
+    #     try:
+    #         return self.dict["%X" % ord(char)].split(" ")[0][0]
+    #     except KeyError:
+    #         return char
+
+    def get_initial(self, chars=u'你好'):
+        _str = u""
+        ch = chars[0]
+        key = "%X" % ord(ch)
+        try:
+            _str += self.dict[key].split(" ")[0][0]
+        except KeyError:
+            _str += ch
+        return _str
+
+    def get_py(self, chars=u'你好'):
+        _str = u""
+        for ch in chars:
+            _str += self.get_initial(ch)
+        return _str
+