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 InfibeamScraper(BaseScraper):
 10 
class InfibeamScraper(BaseScraper):
10
11
11 
    def __init__(self):
 12 
    def __init__(self):
12 
        self.url = None
 13 
        self.url = None
Line 25... Line 26...
25 
 
 26 
 
26 
            name = li.findAll('span', {'class': 'title'})[0].string
 27 
            name = li.findAll('span', {'class': 'title'})[0].string
27 
            try:
 28 
            try:
28 
                price = li.findAll('div', {'class': 'price'})[0].findAll('span', {'class': 'normal'})[0].string
 29 
                price = li.findAll('div', {'class': 'price'})[0].findAll('span', {'class': 'normal'})[0].string
29 
            except IndexError:
 30 
            except IndexError:
30 
                price = li.findAll('span', {'class': 'price'})[0].contents[-1].strip()
 31 
                price = removePriceFormatting(li.findAll('span', {'class': 'price'})[0].contents[-1].strip())
31
32
32 
            url = li.findAll('a')[0]['href']
 33 
            url = li.findAll('a')[0]['href']
33
34
34 
            try:
 35 
            try:
- 
 
 36 
                phone_prices.append({
- 
 
 37 
                        'name': str(name),
- 
 
 38 
                        'price': str(price),
- 
 
 39 
                        'source': 'infibeam',
- 
 
 40 
                        'in_stock': 1,
35 
                phone_prices.append({'name': str(name), 'price': str(price), 'in_stock': 1, 'product_url': str(url)})
 41 
                        'product_url': str(url)
- 
 
 42 
                    })
36
43
37 
            except UnicodeEncodeError as e:
 44 
            except UnicodeEncodeError as e:
38 
                print 'Unicode Error', e, name
 45 
                print 'Unicode Error', e, name
39 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
 46 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
40 
                print name_ascii
 47 
                print name_ascii
- 
 
 48 
                phone_prices.append({
- 
 
 49 
                        "name": str(name_ascii),
- 
 
 50 
                        "price": str(price),
- 
 
 51 
                        'source': 'infibeam',
- 
 
 52 
                        "in_stock": 1,
41 
                phone_prices.append({"name": str(name_ascii), "price": str(price), "in_stock": 1, "product_url": str(url)})
 53 
                        "product_url": str(url)
- 
 
 54 
                    })
42
55
43 
        return phone_prices
 56 
        return phone_prices
44
57
45 
    def getNextUrl(self):
 58 
    def getNextUrl(self):
46 
        b = self.soup.findAll('div', {'class': 'resultsSummary'})[0].findAll('b')
 59 
        b = self.soup.findAll('div', {'class': 'resultsSummary'})[0].findAll('b')