Subversion Repositories SmartDukaan

Rev

Rev 5291 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 5291 Rev 5401
Line 4... Line 4...
4 
@author: Varun Gupta
 4 
@author: Varun Gupta
5 
'''
 5 
'''
6 
from BeautifulSoup import BeautifulSoup
 6 
from BeautifulSoup import BeautifulSoup
7 
from BaseScraper import BaseScraper
 7 
from BaseScraper import BaseScraper
8 
from Utils import removePriceFormatting
 8 
from Utils import removePriceFormatting
- 
 
 9 
from SoupSelect import select
9 
 
 10 
 
10 
class HS18Scraper(BaseScraper):
 11 
class HS18Scraper(BaseScraper):
11 
 
 12 
 
12 
    def __init__(self):
 13 
    def __init__(self):
13 
        self.url = None
 14 
        self.url = None
Line 21... Line 22...
21 
        self.soup = BeautifulSoup(html)
 22 
        self.soup = BeautifulSoup(html)
22
23
23 
    def getPhones(self):
 24 
    def getPhones(self):
24 
        product_prices = []
 25 
        product_prices = []
25
26
26 
        for div in self.soup.findAll('div', {'class': 'product_div'}):
 27 
        for div in select(self.soup, "div.product_div"):#self.soup.findAll('div', {'class': 'product_div'}):
27
28
28 
            anchor = div.find('p', {'class': 'product_title'})('a')[0]
 29 
            anchor = div.find('p', {'class': 'product_title'})('a')[0]
29 
            name = str(anchor['title'].strip())
 30 
            name = str(anchor['title'].strip())
30
31
31 
            if name.endswith(' Mobile Phone'):
 32 
            if name.endswith(' Mobile Phone'):
Line 85... Line 86...
85 
 
 86 
 
86 
 
 87 
 
87 
if __name__ == '__main__':
 88 
if __name__ == '__main__':
88 
    scraper = HS18Scraper()
 89 
    scraper = HS18Scraper()
89 
#    print scraper.getDataFromProductPage('http://www.homeshop18.com/samsung-galaxy-note-n7000-mobile-phone/mobiles-accessories/gsm-handsets/product:16601211/cid:3027/')
 90 
#    print scraper.getDataFromProductPage('http://www.homeshop18.com/samsung-galaxy-note-n7000-mobile-phone/mobiles-accessories/gsm-handsets/product:16601211/cid:3027/')
90 
    scraper.setUrl('http://www.homeshop18.com/gsm-mobiles/categoryid:3027/search:*/start:112/')
 91 
    scraper.setUrl('http://www.homeshop18.com/mobiles/category:14569/')
91 
    scraper.scrape()
 92 
    scraper.scrape()
92 
    products = scraper.getPhones()
 93 
    products = scraper.getPhones()
93 
    print products
 94 
    print products
94 
    print scraper.getNextUrl()
 95 
    print scraper.getNextUrl()
95 

            96