def parse_news(self, response):
ld = NewsLoader(item=NewsItem(),response=response)
ld.add_value('url',response.url)
ld.add_css('title','div.main .hd>h1::text')
ld.add_css('srcurl','div.main .hd .tit-bar .ll .color-a-1 a[href]::attr(href)')
ld.add_css('src','div.main .hd .tit-bar .ll .color-a-1 a[href]::text')
ld.add_css('datetime','div.main .hd .tit-bar .article-time::text')
ld.add_value('channel',u'qq')
ld.add_value('comment_id',response.body_as_unicode(),re='cmt_id\s?=\s?([\d\w]+)')
yield ld.load_item()
cmturl = '/%s/comment?commentid=0&reqnum=10' % ld.get_output_value('comment_id')
cmturl = self.cmturl + cmturl
yield scrapy.Request(cmturl,self.parse_comment)
def parse_news(self, response):
ld = NewsLoader(item=NewsItem(), response=response)
ld.add_value('url', response.url)
ld.add_css('title', 'div.main .hd>h1::text')
ld.add_css('srcurl',
'div.main .hd .tit-bar .ll .color-a-1 a[href]::attr(href)')
ld.add_css('src', 'div.main .hd .tit-bar .ll .color-a-1 a[href]::text')
ld.add_css('datetime', 'div.main .hd .tit-bar .article-time::text')
ld.add_value('channel', u'qq')
ld.add_value('comment_id',
response.body_as_unicode(),
re='cmt_id\s?=\s?([\d\w]+)')
yield ld.load_item()
cmturl = '/%s/comment?commentid=0&reqnum=10' % ld.get_output_value(
'comment_id')
cmturl = self.cmturl + cmturl
yield scrapy.Request(cmturl, self.parse_comment)
