def __preHandling(self):
"""
待匹配字符串的清理空白符和语气助词以及大写数字转化的预处理
:return:
"""
self.target = StringPreHandler.delKeyword(self.target,
u"\\s+") # 清理空白符
self.target = StringPreHandler.delKeyword(self.target,
u"[的]+") # 清理语气助词
self.target = StringPreHandler.numberTranslator(self.target) # 大写数字转化
def _filter(self, input_query):
# 这里对于下个周末这种做转化 把个给移除掉
input_query = StringPreHandler.numberTranslator(input_query)
rule = u"[0-9]月[0-9]"
pattern = re.compile(rule)
match = pattern.search(input_query)
if match != None:
index = input_query.find('月')
rule = u"日|号"
pattern = re.compile(rule)
match = pattern.search(input_query[index:])
if match == None:
rule = u"[0-9]月[0-9]+"
pattern = re.compile(rule)
match = pattern.search(input_query)
if match != None:
end = match.span()[1]
input_query = input_query[:end] + '号' + input_query[end:]
rule = u"月"
pattern = re.compile(rule)
match = pattern.search(input_query)
if match == None:
input_query = input_query.replace('个', '')
input_query = input_query.replace('中旬', '15号')
input_query = input_query.replace('傍晚', '午后')
input_query = input_query.replace('大年', '')
input_query = input_query.replace('五一', '劳动节')
input_query = input_query.replace('白天', '早上')
input_query = input_query.replace('：', ':')
return input_query