def addpinyin(text):
ltext = list(text)
textcount=len(text)
pinyin = []#用过的拼音
index = []#用过的索引
pinyincount=int(textcount*0.5)#最小添加拼音的字数,1个字
print textcount,pinyincount
for i in range(pinyincount):
j = random.randint(0,textcount-1)#随机抽取一个文字的索引
if findindex(j,index):continue
index.append(j)
s = ltext[j]#随机抽取一个文字
if s.strip()=="":continue
if s.strip().isdigit():continue
p = hanzi2pinyin(s)#获取这个字的拼音
if p in pinyin:continue
pinyin.append(p)#添加到已用拼音列表中
news = "%s(%s)" % (s,hanzi2pinyin(s))
ltext[j] = news
return "".join(ltext)
def assign_pinyin(person, propertyName, phoneticPropertyName):
name = person.valueForProperty_(propertyName)
if name and contain_cjk_char(name):
pname = person.valueForProperty_(phoneticPropertyName)
if not reset and pname:
return None
name_py = pinyin.hanzi2pinyin(name).capitalize()
person.setValue_forProperty_(name_py, phoneticPropertyName)
return (name, name_py)
return None
def assign_pinyin(person, propertyName, phoneticPropertyName):
name = person.valueForProperty_(propertyName)
if name and contain_cjk_char(name):
pname = person.valueForProperty_(phoneticPropertyName)
if not args.reset and pname:
return None
name_py = pinyin.hanzi2pinyin(name).capitalize()
if not args.dryrun:
person.setValue_forProperty_(name_py, phoneticPropertyName)
return (name, name_py)
return None
def testMixedText(self):
assert hanzi2pinyin(u'公共的 模 asdf!') == u'gonggongde mo asdf!'
def testEnglish(self):
assert hanzi2pinyin(u'this is a english') == u'this is a english'
def testChinese(self):
assert hanzi2pinyin(u'公共的模') == u'gonggongdemo'
def testMixedText(self):
assert hanzi2pinyin(u'公共的 模 asdf!') == u'gonggongde mo asdf!'
def testEnglish(self):
assert hanzi2pinyin(u'this is a english') == u'this is a english'
def testChinese(self):
assert hanzi2pinyin(u'公共的模') == u'gonggongdemo'
