Subversion Repositories SmartDukaan

Rev

Rev 12363 | Rev 12402 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 12363 Rev 12396
Line 46... Line 46...
46 
                    tag.replaceWith(s)
 46 
                    tag.replaceWith(s)
47 
            x.close()
 47 
            x.close()
48 
            sellerCount=0
 48 
            sellerCount=0
49 
            info = []
 49 
            info = []
50 
            sellerData = soup.findAll("div" , {"class" : "a-row a-spacing-mini olpOffer"})
 50 
            sellerData = soup.findAll("div" , {"class" : "a-row a-spacing-mini olpOffer"})
- 
 
 51 
            dataLength = len(sellerData)
51 
            for data in sellerData:
 52 
            for data in sellerData:
52 
                tempMap={}
 53 
                tempMap={}
53 
                price = data.find('span', attrs={'class' : re.compile('.*olpOfferPrice*')}).find('span').text
 54 
                price = data.find('span', attrs={'class' : re.compile('.*olpOfferPrice*')}).find('span').text
54 
                unitCost = float(price.replace("Rs.","").replace(",",""))
 55 
                unitCost = float(price.replace("Rs.","").replace(",",""))
55 
                shippingCost = data.find('p', attrs={'class' : re.compile('.*olpShippingInfo*')}).find('span').text
 56 
                shippingCost = data.find('p', attrs={'class' : re.compile('.*olpShippingInfo*')}).find('span').text
Line 91... Line 92...
91 
                    tempMap['sellerPrice'] = unitCost+shippingCost
 92 
                    tempMap['sellerPrice'] = unitCost+shippingCost
92 
                if sellerCount==3:
 93 
                if sellerCount==3:
93 
                    tempMap['sellerName'] = store.strip()
 94 
                    tempMap['sellerName'] = store.strip()
94 
                    tempMap['sellerPrice'] = unitCost+shippingCost
 95 
                    tempMap['sellerPrice'] = unitCost+shippingCost
95 
                info.append(tempMap)
96 
                info.append(tempMap)
96 
                if sellerCount==3:
 97 
                if sellerCount==3 or sellerCount==dataLength:
97 
                    returnMap[asin] = info
98 
                    returnMap[asin] = info
98 
                    break
 99 
                    break
99 
        if findStore:
 100 
        if findStore:
100 
            return self.findStoreFront(returnMap)
 101 
            return self.findStoreFront(returnMap)
101 
        else:
 102 
        else:
Line 153... Line 154...
153 
        #return soup.title.string
 154 
        #return soup.title.string
154
155
155 
 
 156 
 
156 
if __name__ == '__main__':
 157 
if __name__ == '__main__':
157 
    urls=[]
 158 
    urls=[]
158 
    urls.append("http://amazon.in/gp/offer-listing/B00CE2LQSW/ref=olp_sort_ps")
 159 
    urls.append("http://amazon.in/gp/offer-listing/B003SNIN9Q/ref=olp_sort_ps")
159 
    urls.append("http://amazon.in/gp/offer-listing")
 160 
    urls.append("http://amazon.in/gp/offer-listing")
160 
#    asin = []
 161 
#    asin = []
161 
#    for a in amazonlisted:
 162 
#    for a in amazonlisted:
162 
#        asin.append(a.asin)
 163 
#        asin.append(a.asin)
163 
#        urls.append('http://www.amazon.in/gp/offer-listing/'+str(a.asin)+'/ref=olp_sort_ps')
 164 
#        urls.append('http://www.amazon.in/gp/offer-listing/'+str(a.asin)+'/ref=olp_sort_ps')
164 
#        if len(urls)==50:
 165 
#        if len(urls)==50:
165 
#            break
 166 
#            break
166 
    print urls
 167 
    print urls
167 
    scraper = AmazonAsyncScraper()
 168 
    scraper = AmazonAsyncScraper()
168 
    'http://www.amazon.in/gp/offer-listing/B00CE2LQSW/ref=olp_sort_ps'
 169 
    'http://www.amazon.in/gp/offer-listing/B003SNIN9Q/ref=olp_sort_ps'
169 
    print len(urls)
 170 
    print len(urls)
170 
    x = scraper.read(urls,True)
 171 
    x = scraper.read(urls,True)
171 
    print x
 172 
    print x
172 
    print "##################"
 173 
    print "##################"
173 
#    fetched = x.items()
 174 
#    fetched = x.items()