sinaSpider.py 文件源码

def parseHome(self,response):
        if len(response.body) > 50:
            print "###########################"
            print "Fetch Home Success"
            print "###########################"
            infos = json.loads(response.body)
            if infos.get('cards', ''):
                cards = infos['cards']
                for card in cards:
                    if card['card_type'] == 6:
                        print '========================================='
                        #????ID??????
                        ori_ID = re.findall(r'\d+',card['actionlog']['oid'])[0]
                        ori_url = 'https://m.weibo.cn/api/container/getIndex?containerid={ori_id}_-_WEIBO_SECOND_PROFILE_WEIBO_ORI&type=uid&page_type=03&value={value}'.format(
                            ori_id = ori_ID,value=response.meta['ID']
                        )
                        print 'ori_ID:',ori_ID
                        yield Request(url=ori_url, meta={'ID': response.meta["ID"],'ori_id': ori_ID, 'owner':response.meta['owner']},
                                      callback=self.parseTweets, dont_filter=True)