def calendarParse(self, response): item = calendarItem() # print("response['meta']:\t",response.meta) item['house_id'] = response.meta['house_id'] item['response'] = response.body.decode('utf8') if len(item['response']) == 17221 and "arg2" in response.meta: print("********************") if len(item['response']) == 17221: headers = {('X-Airbnb-API-Key', 'd306zoyjsyarp7ifhu67rjxn52tv0t20') } arg1 = re.search("arg1='([^']+)'", item['response']).group(1) # print(arg1) _0x23a392 = unsbox(arg1) arg2 = 'acw_sc__v2=' + hexXor(_0x23a392) + ";" meta = { 'house_id': response.meta['house_id'], 'url': response.meta['url'], "handle_httpstatus_all": True, "arg2": arg2, "last_proxy": response.meta['proxy'] } # print("errMeta:",meta,response.meta) yield Request(url=response.meta['url'], callback=self.calendarParse, errback=self.calendarErrback, meta=meta, dont_filter=True, headers=headers) # with open("/{}.html".format(response.meta['house_id']),"w") as f : # f.write(response.body.decode('utf8')) yield item
def calendarParse(self, response): item = calendarItem() # print('\n',response.request.headers,'\n',response.request.url,'\n',response.request.cookies) item['house_id'] = response.meta['house_id'] item['response'] = response.body.decode('utf8') yield item
def calendarParse(self, response): item = calendarItem() item['house_id'] = response.meta['house_id'] item['response'] = response.body.decode('utf8') yield item