Subversion Repositories SmartDukaan

Rev

Rev 4203 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 4203 Rev 5291
Line 23... Line 23...
23 
        self.soup = BeautifulSoup(html)
 23 
        self.soup = BeautifulSoup(html)
24 
        self.phones = None
 24 
        self.phones = None
25
25
26 
    def getPhones(self):
 26 
    def getPhones(self):
27 
        phones = []
 27 
        phones = []
28
- 
 
29 
        for div in self.soup.findAll('div', {'class': 'fk-product-thumb fkp-medium'}):
 28 
        for div in self.soup.findAll('div', {'class': 'fk-product-thumb fkp-medium'}):
30 
            try:
 29 
            try:
31 
                anchor = div.findAll('a', {'class': 'title fk-anchor-link'})[0]
 30 
                anchor = div.find('a', {'class': 'title tpadding5 fk-anchor-link'})
32 
                name = anchor['title'].strip()
 31 
                name = anchor['title'].strip()
33 
                price = None
 32 
                price = None
34 
                product_url = anchor['href'].strip()
 33 
                product_url = anchor['href'].strip()
35 
                in_stock = 0 if div.findAll('b').__len__() > 0 else 1
 34 
                in_stock = 0 if div.findAll('b').__len__() > 0 else 1
36
35
Line 91... Line 90...
91 
    def getDataFromProductPage(self, url):
 90 
    def getDataFromProductPage(self, url):
92 
        html = BaseScraper.read(self, url)
 91 
        html = BaseScraper.read(self, url)
93 
        soup = BeautifulSoup(html)
 92 
        soup = BeautifulSoup(html)
94 
        name = soup.find('h1', {'itemprop': 'name'}).string.strip()
 93 
        name = soup.find('h1', {'itemprop': 'name'}).string.strip()
95 
        price = soup.find('span',{'id': 'fk-mprod-our-id'}).contents[2]
 94 
        price = soup.find('span',{'id': 'fk-mprod-our-id'}).contents[2]
96 
        in_stock = soup.find('div', {'id': 'fk-stock-info-id'}).string.strip()
 95 
        in_stock = 1
97
96
98 
        data = {
 97 
        data = {
99 
            "product_url": str(url),
98 
            "product_url": str(url),
100 
            "source": "flipkart",
99 
            "source": "flipkart",
101 
            "price": price,
100 
            "price": price,
Line 104... Line 103...
104 
        }
 103 
        }
105 
        return data
 104 
        return data
106 
 
 105 
 
107 
if __name__ == '__main__':
 106 
if __name__ == '__main__':
108 
    s = FlipcartScraper()
 107 
    s = FlipcartScraper()
109 
#    data = s.getDataFromProductPage('http://www.flipkart.com/mobiles/micromax/itmd4nf8p5rfhk2y?pid=mobd4nf7rcrckjhn')
 108 
    data = s.getDataFromProductPage('http://www.flipkart.com/samsung-wave-ii-s8530-mobile-phone/p/itmctnexz3gyjfac?pid=MOBCTXB47XCP7Z9X&ref=eca2ea19-cde2-4bfd-a3d8-15cf737c88d3')
110 
#    print data
 109 
    print data
111
110
112 
    s.setUrl('http://www.flipkart.com/mobiles/all/24')
 - 
 
113 
    s.scrape()
 - 
 
114 
    phones = s.getPhones()
 - 
 
115 
    for p in phones: print p
 - 
 
116 
    print s.getNextUrl()
 - 
 
117 

            111
            
#    s.setUrl('http://www.flipkart.com/mobiles/all')

          

          

            -
            
 

            112
            
#    s.scrape()

          

          

            -
            
 

            113
            
#    phones = s.getPhones()

          

          

            -
            
 

            114
            
#    for p in phones: print p

          

          

            -
            
 

            115
            
#    print s.getNextUrl()

          

          

            -
            
 

            116
            

          

          

            118
            

            117