Subversion Repositories SmartDukaan

Rev

Rev 12199 | Rev 12202 | 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
 
12200 kshitij.so 29
    def createData(self,url):
10503 kshitij.so 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
12200 kshitij.so 34
        print "Soup created from flipkart data for %s" %(url)
35
        return self.scrape(self.soup,url)
10503 kshitij.so 36
 
37
 
12200 kshitij.so 38
    def scrape(self,soup,url):
10503 kshitij.so 39
        info = []
40
        oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
41
        for data in oddSeller:
42
            temp={}
11668 kshitij.so 43
            try:
44
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
45
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
46
                temp['shippingTime']=shippingTime
47
            except:
48
                pass
10503 kshitij.so 49
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
50
            temp['sellingPrice']=float(price)
51
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
52
                sellerName = sellerInfo.find('a').string
53
                temp['sellerName'] = sellerName
54
            for metrics in data.find("div",{"class":"fk-text-right"}):
55
                try:
11217 kshitij.so 56
                    metric = metrics.findAll('input', {'type': 'submit'})
57
                except AttributeError:
58
                    continue
59
                try:
60
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 61
                except TypeError:
62
                    continue
11217 kshitij.so 63
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 64
                try:
65
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
66
                except ValueError:
67
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
68
                temp['buyTrend']=buyTrend
69
                dataMetric = dataMetrics.split(';')
70
                sellerCode = dataMetric[0]
71
                temp['sellerCode']=sellerCode
72
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 73
                if not temp.has_key('shippingTime'):
74
                    print "Populating shipping time from metrics"
75
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 76
                temp['sellerScore'] = int(dataMetric[4])
77
                info.append(temp)
78
        evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
79
        for data in evenSeller:
80
            temp={}
81
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
11668 kshitij.so 82
            try:
83
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
84
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
85
                temp['shippingTime']=shippingTime
86
            except:
87
                pass
10503 kshitij.so 88
            temp['sellingPrice']=float(price)
89
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
90
                sellerName = sellerInfo.find('a').string
91
                temp['sellerName'] = sellerName
92
            for metrics in data.find("div",{"class":"fk-text-right"}):
93
                try:
11217 kshitij.so 94
                    metric = metrics.findAll('input', {'type': 'submit'})
95
                except AttributeError:
96
                    continue
97
                try:
98
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 99
                except TypeError:
100
                    continue
11217 kshitij.so 101
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 102
                try:
103
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
104
                except ValueError:
105
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
106
                temp['buyTrend']=buyTrend
107
                dataMetric = dataMetrics.split(';')
108
                temp['sellerCode'] = dataMetric[0] 
109
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 110
                if not temp.has_key('shippingTime'):
111
                    print "Populating shipping time from metrics"
112
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 113
                temp['sellerScore'] = int(dataMetric[4])
114
                info.append(temp)
12200 kshitij.so 115
        print "Returning Json response from flipkart for %s" %(url)
10503 kshitij.so 116
        return info
117
 
118
if __name__ == '__main__':
119
    scraper = FlipkartScraper()
11967 kshitij.so 120
    scraper.read('http://www.flipkart.com/ps/ACCDSN84XXH5P9WG')
12200 kshitij.so 121
    print scraper.createData('http://www.flipkart.com/ps/ACCDSN84XXH5P9WG')
10503 kshitij.so 122