def get_release_date(self, meta, response):
     regx = '//span[@property="v:initialReleaseDate"]/@content'
     data = response.xpath(regx).extract()
     if data:
         release_date = validator.str_to_date(validator.match_date(data[0]))
         if release_date:
             meta['release_date'] = release_date
     return meta
Exemple #2
0
 def set_release_date(self, meta, response):
     regex = '//span[@property="v:initialReleaseDate"]/@content'
     match = response.xpath(regex).get()
     if match:
         release_date = validator.str_to_date(validator.match_date(match))
         if release_date:
             meta["release_date"] = release_date
     return meta
Exemple #3
0
 def get_birth(self, meta, response):
     regx = '//div[@class="info"]/ul/li/text()[preceding-sibling::span[text()="出生日期"]]'
     data = response.xpath(regx).extract()
     print("============get_birth:", data)
     if data:
         meta['birth'] = validator.str_to_date(
             validator.match_date(data[0].strip("\n")))
         if not meta['birth']:
             meta['birth'] = data[0].strip("\n").split(":")[-1]
     return meta