Subversion Repositories SmartDukaan

Rev

Rev 12256 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 12256 Rev 12275
Line 69... Line 69...
69 
            sellerColumn =  data.find('p', attrs={'class' : re.compile('.*olpSellerName*')})
 69 
            sellerColumn =  data.find('p', attrs={'class' : re.compile('.*olpSellerName*')})
70 
            store=""
 70 
            store=""
71 
            if self.findStore:
 71 
            if self.findStore:
72 
                print "Seller info ",sellerColumn
 72 
                print "Seller info ",sellerColumn
73 
                x = sellerColumn.find('a')['href']
 73 
                x = sellerColumn.find('a')['href']
74 
                print "&&&&"
 74 
                print x
- 
 
 75 
                temp =  sellerColumn.find('a')
75 
                storeUrl = x
 76 
                store = temp.text
76 
                store = self.findStoreFront(storeUrl)
 77 
                if len(store)==0:
77 
                try:
 78 
                    storeUrl = x
78 
                    ind = store.index("@ Amazon.in")
 79 
                    dom_in = storeUrl.find("www.amazon.in")
- 
 
 80 
                    if dom_in ==-1:
79 
                    store = store[0:ind].strip()
 81 
                        storeUrl="http://amazon.in"+storeUrl
80 
                except:
 82 
                    store = self.findStoreFront(storeUrl)
81 
                    try:
 83 
                    try:
82 
                        ind = store.split(":")
 84 
                        ind = store.index("@ Amazon.in")
83 
                        store = ind[1].strip()
 85 
                        store = store[0:ind].strip()
84 
                    except:
 86 
                    except:
- 
 
 87 
                        try:
- 
 
 88 
                            ind = store.split(":")
- 
 
 89 
                            store = ind[1].strip()
- 
 
 90 
                        except:
85 
                        store =""
 91 
                            store =""
86 
            ratingColumn = data.find('p', attrs={'class' : 'a-spacing-small'}).find('a').contents[0]
 92 
            ratingColumn = data.find('p', attrs={'class' : 'a-spacing-small'}).find('a').contents[0]
87 
            print "Rating info ",ratingColumn
 93 
            print "Rating info ",ratingColumn
88 
            print "***********************"
 94 
            print "***********************"
89 
            return unitCost+shippingCost,store
 95 
            return unitCost+shippingCost,store
90
96
Line 108... Line 114...
108 
        return soup.title.string
 114 
        return soup.title.string
109
115
110 
 
 116 
 
111 
if __name__ == '__main__':
 117 
if __name__ == '__main__':
112 
    scraper = AmazonScraper()
 118 
    scraper = AmazonScraper()
113 
    scraper.read('http://www.amazon.in/gp/offer-listing/B001D0ROGO/ref=olp_sort_ps',True)
 119 
    scraper.read('http://www.amazon.in/gp/offer-listing/B006PB44NM/ref=olp_sort_ps',True)
114 
    print scraper.createData()
 120 
    print scraper.createData()
115
121
116 

            122