| 185 |
ashish |
1 |
'''
|
|
|
2 |
Created on 20-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 |
|
|
|
19 |
from datastore import DataAccessor
|
|
|
20 |
from datastore.DataAccessor import DataHelper
|
|
|
21 |
|
|
|
22 |
class mobstore_price(BaseSpider):
|
|
|
23 |
|
|
|
24 |
def __init__(self):
|
|
|
25 |
self.domain_name = "mobstorephones"
|
|
|
26 |
da = DataHelper()
|
|
|
27 |
for pitem in da.get_allmobstoreurls():
|
|
|
28 |
self.start_urls.append(pitem.url.strip())
|
|
|
29 |
|
|
|
30 |
def start_requests(self):
|
|
|
31 |
listreq = []
|
|
|
32 |
for url1 in self.start_urls:
|
|
|
33 |
request = Request(url = url1, callback=self.parse)
|
|
|
34 |
request.headers.setdefault("Referer", "www.themobilestore.in")
|
|
|
35 |
listreq.append(request)
|
|
|
36 |
return listreq
|
|
|
37 |
|
|
|
38 |
def parse(self, response):
|
|
|
39 |
site = response.url
|
|
|
40 |
vatplustax = 0
|
|
|
41 |
pos1 = pos2 = 0
|
|
|
42 |
temp = ""
|
|
|
43 |
pos1 = site.rfind('/')
|
|
|
44 |
if pos1 != -1:
|
|
|
45 |
temp = site[pos1+1:len(site)]
|
|
|
46 |
#pos2 = site.rfind('/',0,pos1-1)
|
|
|
47 |
#if pos2 > 0:
|
|
|
48 |
#temp = site[pos2+1:len(site)]
|
|
|
49 |
pos3 = temp.find('.')
|
|
|
50 |
temp1 = temp[pos3:len(temp)]
|
|
|
51 |
name = temp.replace(temp1,"")
|
|
|
52 |
hxs = HtmlXPathSelector(response)
|
|
|
53 |
prices = hxs.select('//div[@id ="priceComp"]//tr[2]/td[3]/span/text()')
|
|
|
54 |
|
|
|
55 |
da = DataHelper()
|
|
|
56 |
for price in prices:
|
|
|
57 |
name = str(name).strip()
|
|
|
58 |
price = price.extract()
|
|
|
59 |
price = str(price).strip()
|
|
|
60 |
price = price.replace("Rs", "")
|
|
|
61 |
price = price.replace("/", "")
|
|
|
62 |
price = price.replace("-", "")
|
|
|
63 |
price = price.replace(".", "")
|
|
|
64 |
shown_pr = int(price)
|
|
|
65 |
final_pr = shown_pr + vatplustax
|
|
|
66 |
da.add_new_mobstorephone(name,shown_pr,final_pr)
|
|
|
67 |
print name
|
|
|
68 |
print final_pr
|
|
|
69 |
print "\n"
|
|
|
70 |
'''
|
|
|
71 |
for i in items:
|
|
|
72 |
str1 = str(i['title']).strip()
|
|
|
73 |
print str1
|
|
|
74 |
amnt = i['price'].replace(",","")
|
|
|
75 |
amnt = amnt.replace("Rs", "")
|
|
|
76 |
amnt = amnt.replace("/", "")
|
|
|
77 |
amnt = amnt.replace("-", "")
|
|
|
78 |
amnt = amnt.strip()
|
|
|
79 |
vatplustax = 4*int(amnt)/100
|
|
|
80 |
pr = int(amnt) + vatplustax
|
|
|
81 |
#print pr
|
|
|
82 |
da.add_new_univerphone(str1,amnt,pr)
|
|
|
83 |
'''
|
|
|
84 |
SPIDER = mobstore_price()
|
|
|
85 |
|