def parse(self, response): response_selector = Selector(response) next_link = response_selector.select(u'//span[@class="next"]/a/@href').extract()[0] log.msg("jinxp next link is %s" % next_link, level=log.DEBUG) if next_link: next_link = clean_url(response.url,next_link,response.encoding) yield Request(url=next_link, callback=self.parse) for detail_link in response_selector.select(u'//table[@class="olt"]/tr[@class=""]/td[@class="title"]/a/@href').extract(): if detail_link: detail_link = clean_url(response.url,detail_link,response.encoding) yield Request(url=detail_link, callback=self.parse_detail)
def parse(self, response): sel = Selector(response) for detail_link in sel.xpath('//ul[@class="browseLettersLinks"]/li/a/@href').extract(): if detail_link: detail_link = clean_url(response.url, detail_link, response.encoding) yield Request(url=detail_link, callback=self.parse_detail)