Subversion Repositories SmartDukaan

Rev

Rev 12191 | Rev 12200 | 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()
12199 kshitij.so 17 
            print "Fetched response from flipkart for %s" %(url)
10503 kshitij.so 18 
 
19 
        except urllib2.HTTPError as e:
20 
            print 'ERROR: ', e
21 
            print 'Retrying'
22 
            self.count_trials += 1
23 
 
24 
            if self.count_trials < 3:
25 
                return self.read(url)
26 
 
27 
        self.response_data=response_data
28 
 
29 
    def createData(self):
30 
        page=self.response_data.decode("utf-8")
31 
        self.soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
11967 kshitij.so 32 
        page = None
33 
        self.response_data = None
10503 kshitij.so 34 
        return self.scrape(self.soup)
35 
 
36 
 
37 
    def scrape(self,soup):
38 
        info = []
39 
        oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
40 
        for data in oddSeller:
41 
            temp={}
11668 kshitij.so 42 
            try:
43 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
44 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
45 
                temp['shippingTime']=shippingTime
46 
            except:
47 
                pass
10503 kshitij.so 48 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
49 
            temp['sellingPrice']=float(price)
50 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
51 
                sellerName = sellerInfo.find('a').string
52 
                temp['sellerName'] = sellerName
53 
            for metrics in data.find("div",{"class":"fk-text-right"}):
54 
                try:
11217 kshitij.so 55 
                    metric = metrics.findAll('input', {'type': 'submit'})
56 
                except AttributeError:
57 
                    continue
58 
                try:
59 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 60 
                except TypeError:
61 
                    continue
11217 kshitij.so 62 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 63 
                try:
64 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
65 
                except ValueError:
66 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
67 
                temp['buyTrend']=buyTrend
68 
                dataMetric = dataMetrics.split(';')
69 
                sellerCode = dataMetric[0]
70 
                temp['sellerCode']=sellerCode
71 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 72 
                if not temp.has_key('shippingTime'):
73 
                    print "Populating shipping time from metrics"
74 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 75 
                temp['sellerScore'] = int(dataMetric[4])
76 
                info.append(temp)
77 
        evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
78 
        for data in evenSeller:
79 
            temp={}
80 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
11668 kshitij.so 81 
            try:
82 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
83 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
84 
                temp['shippingTime']=shippingTime
85 
            except:
86 
                pass
10503 kshitij.so 87 
            temp['sellingPrice']=float(price)
88 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
89 
                sellerName = sellerInfo.find('a').string
90 
                temp['sellerName'] = sellerName
91 
            for metrics in data.find("div",{"class":"fk-text-right"}):
92 
                try:
11217 kshitij.so 93 
                    metric = metrics.findAll('input', {'type': 'submit'})
94 
                except AttributeError:
95 
                    continue
96 
                try:
97 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 98 
                except TypeError:
99 
                    continue
11217 kshitij.so 100 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 101 
                try:
102 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
103 
                except ValueError:
104 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
105 
                temp['buyTrend']=buyTrend
106 
                dataMetric = dataMetrics.split(';')
107 
                temp['sellerCode'] = dataMetric[0]
108 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 109 
                if not temp.has_key('shippingTime'):
110 
                    print "Populating shipping time from metrics"
111 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 112 
                temp['sellerScore'] = int(dataMetric[4])
113 
                info.append(temp)
114 
        return info
115 
 
116 
if __name__ == '__main__':
117 
    scraper = FlipkartScraper()
11967 kshitij.so 118 
    scraper.read('http://www.flipkart.com/ps/ACCDSN84XXH5P9WG')
10503 kshitij.so 119 
    print scraper.createData()
120