def get_official_site(self, meta, response):
regx = '//a[preceding-sibling::span[text()="官方网站:"]][following-si\
bling::br]/@href'
data = response.xpath(regx).extract()
if data:
meta['official_site'] = validator.process_url(data[0])
return meta
def get_poster(self, meta, response):
regx = '//a[@class="nbgnbg"]/img/@src'
data = response.xpath(regx).extract()
if data:
url = validator.process_url(data[0])
print("Poster Url:", url, sep=" ")
meta['poster_url'] = url
return meta
def set_official_site(self, meta, response):
regex = '//a[preceding-sibling::span[text()="官方网站:"]][following-si\
bling::br]/@href'
match = response.xpath(regex).get()
if match:
meta["official_site"] = validator.process_url(match)
return meta