Subversion Repositories SmartDukaan

Rev

Rev 11193 | Rev 11668 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
10503 kshitij.so 1 
import urllib2
2 
from BeautifulSoup import BeautifulSoup
3 
import re
4 
 
5 
class FlipkartScraper:
6 
    def __init__(self):
7 
        self.count_trials = 0
8 
 
9 
    def read(self, url):
10 
        request = urllib2.Request(url)
11 
        request.add_header('User-Agent', 'Mozilla/5.0 (X11; Linux i686) AppleWebKit/535.1 (KHTML, like Gecko) Chrome/13.0.782.218 Safari/535.1')
12 
        opener = urllib2.build_opener()
13 
        response_data = ""
14 
        try:
15 
            response_data = opener.open(request).read()
16 
 
17 
        except urllib2.HTTPError as e:
18 
            print 'ERROR: ', e
19 
            print 'Retrying'
20 
            self.count_trials += 1
21 
 
22 
            if self.count_trials < 3:
23 
                return self.read(url)
24 
 
25 
        self.response_data=response_data
26 
 
27 
    def createData(self):
28 
        page=self.response_data.decode("utf-8")
29 
        self.soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
30 
        return self.scrape(self.soup)
31 
 
32 
 
33 
    def scrape(self,soup):
34 
        info = []
35 
        oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
36 
        for data in oddSeller:
37 
            temp={}
38 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
39 
            temp['sellingPrice']=float(price)
40 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
41 
                sellerName = sellerInfo.find('a').string
42 
                temp['sellerName'] = sellerName
43 
            for metrics in data.find("div",{"class":"fk-text-right"}):
44 
                try:
11217 kshitij.so 45 
                    metric = metrics.findAll('input', {'type': 'submit'})
46 
                except AttributeError:
47 
                    continue
48 
                try:
49 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 50 
                except TypeError:
51 
                    continue
11217 kshitij.so 52 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 53 
                try:
54 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
55 
                except ValueError:
56 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
57 
                temp['buyTrend']=buyTrend
58 
                dataMetric = dataMetrics.split(';')
59 
                sellerCode = dataMetric[0]
60 
                temp['sellerCode']=sellerCode
61 
                temp['sellingPriceMetric'] = float(dataMetric[1])
62 
                temp['shippingTime'] = dataMetric[3]
63 
                temp['sellerScore'] = int(dataMetric[4])
64 
                info.append(temp)
65 
        evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
66 
        for data in evenSeller:
67 
            temp={}
68 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
69 
            temp['sellingPrice']=float(price)
70 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
71 
                sellerName = sellerInfo.find('a').string
72 
                temp['sellerName'] = sellerName
73 
            for metrics in data.find("div",{"class":"fk-text-right"}):
74 
                try:
11217 kshitij.so 75 
                    metric = metrics.findAll('input', {'type': 'submit'})
76 
                except AttributeError:
77 
                    continue
78 
                try:
79 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 80 
                except TypeError:
81 
                    continue
11217 kshitij.so 82 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 83 
                try:
84 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
85 
                except ValueError:
86 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
87 
                temp['buyTrend']=buyTrend
88 
                dataMetric = dataMetrics.split(';')
89 
                temp['sellerCode'] = dataMetric[0]
90 
                temp['sellingPriceMetric'] = float(dataMetric[1])
91 
                temp['shippingTime'] = dataMetric[3]
92 
                temp['sellerScore'] = int(dataMetric[4])
93 
                info.append(temp)
94 
        return info
95 
 
96 
if __name__ == '__main__':
97 
    scraper = FlipkartScraper()
11217 kshitij.so 98 
    scraper.read('http://www.flipkart.com/ps/ACCCZJZNQYSRGH8F')
10503 kshitij.so 99 
    print scraper.createData()
100