Subversion Repositories SmartDukaan

Rev

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

Rev 4199 Rev 5291
Line 21... Line 21...
21 
        self.soup = BeautifulSoup(html)
 21 
        self.soup = BeautifulSoup(html)
22
22
23 
    def getPhones(self):
 23 
    def getPhones(self):
24 
        phone_prices = []
 24 
        phone_prices = []
25 
        for li in self.soup.findAll('ul', {'class': 'srch_result portrait'})[0]('li'):
 25 
        for li in self.soup.findAll('ul', {'class': 'srch_result portrait'})[0]('li'):
26 
 
 26
27 
            name = li.findAll('span', {'class': 'title'})[0].string
 27 
            name = li.find('span', {'class': 'title'}).contents[1].strip()
28 
            try:
 28 
            try:
29 
                price = li.findAll('div', {'class': 'price'})[0].findAll('span', {'class': 'normal'})[0].string
 29 
                price = li.find('div', {'class': 'price'}).find('span', {'class': 'normal'}).string
30 
            except IndexError:
 30 
            except IndexError:
- 
 
 31 
                price = removePriceFormatting(li.find('span', {'class': 'price'}).contents[-1].strip())
- 
 
 32 
            except AttributeError:
31 
                price = removePriceFormatting(li.findAll('span', {'class': 'price'})[0].contents[-1].strip())
 33 
                price = removePriceFormatting(li.find('span', {'class': 'price'}).contents[-1].strip())
32
34
33 
            url = li.findAll('a')[0]['href']
 35 
            url = li.findAll('a')[0]['href']
34
36
35 
            try:
 37 
            try:
36 
                phone_prices.append({
 38 
                phone_prices.append({
37 
                        'name': str(name),
39 
                        'name': str(name),
38 
                        'price': str(price),
 40 
                        'price': removePriceFormatting(str(price)),
39 
                        'source': 'infibeam',
41 
                        'source': 'infibeam',
40 
                        'in_stock': 1,
42 
                        'in_stock': 1,
41 
                        'product_url': str(url)
 43 
                        'product_url': str(url)
42 
                    })
 44 
                    })
43
45
Line 45... Line 47...
45 
                print 'Unicode Error', e, name
 47 
                print 'Unicode Error', e, name
46 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
 48 
                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])
47 
                print name_ascii
 49 
                print name_ascii
48 
                phone_prices.append({
 50 
                phone_prices.append({
49 
                        "name": str(name_ascii),
51 
                        "name": str(name_ascii),
50 
                        "price": str(price),
 52 
                        "price": removePriceFormatting(str(price)),
51 
                        'source': 'infibeam',
53 
                        'source': 'infibeam',
52 
                        "in_stock": 1,
54 
                        "in_stock": 1,
53 
                        "product_url": str(url)
 55 
                        "product_url": str(url)
54 
                    })
 56 
                    })
55
57
Line 78... Line 80...
78 
        }
 80 
        }
79 
        return data
 81 
        return data
80 
 
 82 
 
81 
if __name__ == '__main__':
 83 
if __name__ == '__main__':
82 
    s = InfibeamScraper()
 84 
    s = InfibeamScraper()
83 
    print s.getDataFromProductPage('http://www.infibeam.com/Mobiles/i-HTC-EVO-3D-Android-Smartphone/P-E-M-HTC-EVO-3D.html?id=Black')
 - 
 
84 
#    s.setUrl('http://www.infibeam.com/Mobiles/search?page=17')
 - 
 
85 
#    s.scrape()
 - 
 
86 
#    products = s.getPhones()
 - 
 
87 
#    print products
 - 
 
88 

            85
            
#    print s.getDataFromProductPage('http://www.infibeam.com/Mobiles/i-HTC-EVO-3D-Android-Smartphone/P-E-M-HTC-EVO-3D.html?id=Black')

          

          

            -
            
 

            86
            
    s.setUrl('http://www.infibeam.com/Mobiles/search?page=5')

          

          

            -
            
 

            87
            
    s.scrape()

          

          

            -
            
 

            88
            
    products = s.getPhones()

          

          

            -
            
 

            89
            
    

          

          

            -
            
 

            90
            
    print products

          

          

            -
            
 

            91
            
    print s.getNextUrl()

          

          

            -
            
 

            92
            

          

          

            89
            

            93