Subversion Repositories SmartDukaan

Rev

Rev 11217 | Rev 11967 | 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
11668 kshitij.so 4 
from sys import exit
10503 kshitij.so 5 
 
6 
class FlipkartScraper:
7 
    def __init__(self):
8 
        self.count_trials = 0
9 
 
10 
    def read(self, url):
11 
        request = urllib2.Request(url)
12 
        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')
13 
        opener = urllib2.build_opener()
14 
        response_data = ""
15 
        try:
16 
            response_data = opener.open(request).read()
17 
 
18 
        except urllib2.HTTPError as e:
19 
            print 'ERROR: ', e
20 
            print 'Retrying'
21 
            self.count_trials += 1
22 
 
23 
            if self.count_trials < 3:
24 
                return self.read(url)
25 
 
26 
        self.response_data=response_data
27 
 
28 
    def createData(self):
29 
        page=self.response_data.decode("utf-8")
30 
        self.soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
31 
        return self.scrape(self.soup)
32 
 
33 
 
34 
    def scrape(self,soup):
35 
        info = []
36 
        oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
37 
        for data in oddSeller:
38 
            temp={}
11668 kshitij.so 39 
            try:
40 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
41 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
42 
                temp['shippingTime']=shippingTime
43 
            except:
44 
                pass
10503 kshitij.so 45 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
46 
            temp['sellingPrice']=float(price)
47 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
48 
                sellerName = sellerInfo.find('a').string
49 
                temp['sellerName'] = sellerName
50 
            for metrics in data.find("div",{"class":"fk-text-right"}):
51 
                try:
11217 kshitij.so 52 
                    metric = metrics.findAll('input', {'type': 'submit'})
53 
                except AttributeError:
54 
                    continue
55 
                try:
56 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 57 
                except TypeError:
58 
                    continue
11217 kshitij.so 59 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 60 
                try:
61 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
62 
                except ValueError:
63 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
64 
                temp['buyTrend']=buyTrend
65 
                dataMetric = dataMetrics.split(';')
66 
                sellerCode = dataMetric[0]
67 
                temp['sellerCode']=sellerCode
68 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 69 
                if not temp.has_key('shippingTime'):
70 
                    print "Populating shipping time from metrics"
71 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 72 
                temp['sellerScore'] = int(dataMetric[4])
73 
                info.append(temp)
74 
        evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
75 
        for data in evenSeller:
76 
            temp={}
77 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
11668 kshitij.so 78 
            try:
79 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
80 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
81 
                temp['shippingTime']=shippingTime
82 
            except:
83 
                pass
10503 kshitij.so 84 
            temp['sellingPrice']=float(price)
85 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
86 
                sellerName = sellerInfo.find('a').string
87 
                temp['sellerName'] = sellerName
88 
            for metrics in data.find("div",{"class":"fk-text-right"}):
89 
                try:
11217 kshitij.so 90 
                    metric = metrics.findAll('input', {'type': 'submit'})
91 
                except AttributeError:
92 
                    continue
93 
                try:
94 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 95 
                except TypeError:
96 
                    continue
11217 kshitij.so 97 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 98 
                try:
99 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
100 
                except ValueError:
101 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
102 
                temp['buyTrend']=buyTrend
103 
                dataMetric = dataMetrics.split(';')
104 
                temp['sellerCode'] = dataMetric[0]
105 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 106 
                if not temp.has_key('shippingTime'):
107 
                    print "Populating shipping time from metrics"
108 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 109 
                temp['sellerScore'] = int(dataMetric[4])
110 
                info.append(temp)
111 
        return info
112 
 
113 
if __name__ == '__main__':
114 
    scraper = FlipkartScraper()
11668 kshitij.so 115 
    scraper.read('http://www.flipkart.com/ps/ACCDNKARGGKYVGZZ')
10503 kshitij.so 116 
    print scraper.createData()
117