| 136 |
ashish |
1 |
'''
|
|
|
2 |
Created on 11-May-2010
|
|
|
3 |
|
|
|
4 |
@author: gaurav
|
|
|
5 |
'''
|
|
|
6 |
|
|
|
7 |
from scrapy.spider import BaseSpider
|
|
|
8 |
from scrapy.selector import HtmlXPathSelector
|
|
|
9 |
from scrapy.http import Request
|
|
|
10 |
|
|
|
11 |
from demo.items import DemoItem
|
|
|
12 |
from scrapy.contrib.spidermiddleware import referer
|
|
|
13 |
from scrapy.http.headers import Headers
|
|
|
14 |
from scrapy.http.request.form import FormRequest
|
|
|
15 |
from scrapy.log import msg
|
|
|
16 |
from scrapy.http.response import Response
|
|
|
17 |
|
|
|
18 |
from datastore import DataAccessor
|
|
|
19 |
from datastore.DataAccessor import DataHelper
|
|
|
20 |
|
|
|
21 |
|
|
|
22 |
class vendor_links(BaseSpider):
|
|
|
23 |
domain_name = "vendors"
|
|
|
24 |
start_urls = [
|
|
|
25 |
"http://www.infibeam.com/Mobiles/"
|
|
|
26 |
]
|
|
|
27 |
|
|
|
28 |
def start_requests(self):
|
|
|
29 |
request = Request(url = "http://www.infibeam.com/Mobiles/", callback=self.parse)
|
|
|
30 |
request.headers.setdefault("Referer", "www.google.com/search")
|
|
|
31 |
return [request]
|
|
|
32 |
|
|
|
33 |
def parse(self, response):
|
|
|
34 |
str1 = "www.infibeam.com"
|
|
|
35 |
hxs = HtmlXPathSelector(response)
|
|
|
36 |
vendor_info = hxs.select('//td[h3="Mobiles"]/ul/li')
|
|
|
37 |
print len(vendor_info)
|
|
|
38 |
items = []
|
|
|
39 |
for i in vendor_info:
|
|
|
40 |
item = {}
|
|
|
41 |
item['name'] = i.select('.//a/@title')[0].extract()
|
|
|
42 |
item['site'] = i.select('.//a/@href')[0].extract()
|
|
|
43 |
items.append(item)
|
|
|
44 |
|
|
|
45 |
da = DataHelper()
|
|
|
46 |
for item in items:
|
|
|
47 |
str2 = str1 + str(item['site'])
|
|
|
48 |
da.add_vendor(item['name'], str2)
|
|
|
49 |
print item['name']
|
|
|
50 |
print str2
|
|
|
51 |
|
|
|
52 |
SPIDER = vendor_links()
|