Subversion Repositories SmartDukaan

'''
Created on 27-May-2010

@author: gaurav
'''


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 naaptol_spider(BaseSpider):
   
    def __init__(self): 
       self.domain_name = "naaptolsites"
       str1 = "http://www.naaptol.com/sitemap.xml"
       self.start_urls.append(str1)
    
    
    def start_requests(self):
        da = DataHelper()
        da.add_supplier(self.domain_name, "www.naaptol.com")
        listreq = []
        for url1 in self.start_urls:
            request = Request(url = str(url1), callback=self.parse)
            request.headers.setdefault("Referer", "http://www.naaptol.com/sitemap.xml")
            listreq.append(request)
        return listreq
    
    def parse(self, response):
        da = DataHelper()
        hxs = HtmlXPathSelector(response)
        phone_urls = hxs.select('//url/loc/text()')
        for i in phone_urls:
            site = i.extract()
            pos1 = pos2 = pos3 = 0
            temp =""
            pos1 = site.rfind('/')
            if pos1 != -1:
                pos2 = site.rfind('/',0,pos1-1)
            if pos2 != -1:    
                pos3 = site.rfind('/',0,pos2-1)
            if pos3 > 0:
                temp = site[pos3+1:pos1]
            if temp == "mobile_phones/pdas_and_smartphones" or temp == "mobile_phones/gsm_handsets" or temp == "mobile_phones/cdma_handsets":
                da.add_naaptolurl(site)
                #print str(ct) + " " + site
                #print "\n"
                #ct = ct +1    
                           
SPIDER = naaptol_spider()