Subversion Repositories SmartDukaan

Rev

Rev 4039 | Rev 4199 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 4039 Rev 4198
Line 3... Line 3...
3 
 
 3 
 
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 
 
 9 
 
9 
class LetsBuyScraper(BaseScraper):
 10 
class LetsBuyScraper(BaseScraper):
10
11
11 
    def __init__(self):
 12 
    def __init__(self):
12 
        BaseScraper.__init__(self)
 13 
        BaseScraper.__init__(self)
Line 24... Line 25...
24 
        phone_prices = []
 25 
        phone_prices = []
25 
 
 26 
 
26 
        for div in self.soup.findAll('div', {'class': "detailbox"}):
 27 
        for div in self.soup.findAll('div', {'class': "detailbox"}):
27 
            name_tag = div('h2')[0]('a')[0]
 28 
            name_tag = div('h2')[0]('a')[0]
28 
            name = name_tag.string.strip()
 29 
            name = name_tag.string.strip()
29 
            price = div.findAll('span', {'class': "text12_stb"})[0].string.strip()
 30 
            price = removePriceFormatting(div.findAll('span', {'class': "text12_stb"})[0].string.strip())
30 
            url = str(name_tag['href'])
 31 
            url = str(name_tag['href'])
31 
            try:
 32 
            try:
- 
 
 33 
                phone_prices.append({
- 
 
 34 
                        "name": str(name),
- 
 
 35 
                        "price": str(price),
- 
 
 36 
                        'source': 'letsbuy',
- 
 
 37 
                        "in_stock": 1,
32 
                phone_prices.append({"name": str(name), "price": str(price), "in_stock": 1, "product_url": str(url)})
 38 
                        "product_url": str(url)
- 
 
 39 
                    })
33 
            except UnicodeEncodeError as e:
 40 
            except UnicodeEncodeError as e:
34 
                print 'Unicode Error', e, name
 41 
                print 'Unicode Error', e, name
35 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
 42 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
36 
                print name_ascii
 43 
                print name_ascii
- 
 
 44 
                phone_prices.append({
- 
 
 45 
                        "name": str(name_ascii),
- 
 
 46 
                        "price": str(price),
- 
 
 47 
                        'source': 'letsbuy',
- 
 
 48 
                        "in_stock": 1,
37 
                phone_prices.append({"name": str(name_ascii), "price": str(price), "in_stock": 1, "product_url": str(url)})
 49 
                        "product_url": str(url)
- 
 
 50 
                    })
38
51
39 
        return phone_prices
 52 
        return phone_prices
40
53
41 
    def getNextUrl(self):
 54 
    def getNextUrl(self):
42 
        next_url = None
 55 
        next_url = None