Subversion Repositories SmartDukaan

from scrapy.spider import BaseSpider

from scrapy.selector import HtmlXPathSelector

from scrapy.http import Request

from demo.items import DemoItem

from scrapy.contrib.spidermiddleware import referer

from scrapy.http.headers import Headers

from scrapy.http.request.form import FormRequest

from scrapy.log import msg

from scrapy.http.response import Response

from datastore import DataAccessor

from datastore.DataAccessor import DataHelper

class vendor_links(BaseSpider):

    domain_name = "univercellvendors"

    start_urls = [

          "http://www.univercell.in/mobiles/populateStore.action"

    ]

    def start_requests(self):

        request = Request(url = "http://www.univercell.in/mobiles/populateStore.action", callback=self.parse)

        request.headers.setdefault("Referer", "www.google.com/search")

        return [request]

    def parse(self, response):

        str1 = "http://www.univercell.in"

        hxs = HtmlXPathSelector(response)

        vendor_info = hxs.select('//div[@id="mobilesTab"]/table/tr[1]/td/table/tr')

        print len(vendor_info)

        items = []

        for i in vendor_info:

            item = {}

            item['name'] = i.select('.//a/text()')[0].extract()

            temp = i.select('.//a/@href')[0].extract()

            a = str(temp).find(";")

            b = str(temp).find("?")

            temp1 = str(temp)[a:b]

            temp2 = str(temp).replace(temp1,"")

            item['site'] =  str(temp2).replace("populate","rePopulate")

            items.append(item)

        da = DataHelper()

        for item in items:

            str2 = str1 + str(item['site'])

            da.add_univervendor(item['name'].strip(), str2)

            print item['name']

            print str2

SPIDER = vendor_links()