Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed
'''Created on 20-May-2010@author: gaurav'''from scrapy.spider import BaseSpiderfrom scrapy.selector import HtmlXPathSelectorfrom scrapy.http import Requestfrom demo.items import DemoItemfrom scrapy.contrib.spidermiddleware import refererfrom scrapy.http.headers import Headersfrom scrapy.http.request.form import FormRequestfrom scrapy.log import msgfrom scrapy.http.response import Responsefrom datastore import DataAccessorfrom datastore.DataAccessor import DataHelperclass mobstore_price(BaseSpider):def __init__(self):self.domain_name = "mobstorephones"da = DataHelper()for pitem in da.get_allmobstoreurls():self.start_urls.append(pitem.url.strip())def start_requests(self):listreq = []for url1 in self.start_urls:request = Request(url = url1, callback=self.parse)request.headers.setdefault("Referer", "www.themobilestore.in")listreq.append(request)return listreqdef parse(self, response):site = response.urlvatplustax = 0pos1 = pos2 = 0temp = ""pos1 = site.rfind('/')if pos1 != -1:temp = site[pos1+1:len(site)]#pos2 = site.rfind('/',0,pos1-1)#if pos2 > 0:#temp = site[pos2+1:len(site)]pos3 = temp.find('.')temp1 = temp[pos3:len(temp)]name = temp.replace(temp1,"")hxs = HtmlXPathSelector(response)prices = hxs.select('//div[@id ="priceComp"]//tr[2]/td[3]/span/text()')da = DataHelper()for price in prices:name = str(name).strip()price = price.extract()price = str(price).strip()price = price.replace("Rs", "")price = price.replace("/", "")price = price.replace("-", "")price = price.replace(".", "")shown_pr = int(price)final_pr = shown_pr + vatplustaxda.add_new_mobstorephone(name,shown_pr,final_pr)print nameprint final_prprint "\n"'''for i in items:str1 = str(i['title']).strip()print str1amnt = i['price'].replace(",","")amnt = amnt.replace("Rs", "")amnt = amnt.replace("/", "")amnt = amnt.replace("-", "")amnt = amnt.strip()vatplustax = 4*int(amnt)/100pr = int(amnt) + vatplustax#print prda.add_new_univerphone(str1,amnt,pr)'''SPIDER = mobstore_price()