Subversion Repositories SmartDukaan

Rev

Rev 12190 | Rev 12199 | 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)
11967 kshitij.so 31 
        page = None
32 
        self.response_data = None
10503 kshitij.so 33 
        return self.scrape(self.soup)
34 
 
35 
 
36 
    def scrape(self,soup):
37 
        info = []
38 
        oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
39 
        for data in oddSeller:
40 
            temp={}
11668 kshitij.so 41 
            try:
42 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
43 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
44 
                temp['shippingTime']=shippingTime
45 
            except:
46 
                pass
10503 kshitij.so 47 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
48 
            temp['sellingPrice']=float(price)
49 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
50 
                sellerName = sellerInfo.find('a').string
51 
                temp['sellerName'] = sellerName
52 
            for metrics in data.find("div",{"class":"fk-text-right"}):
53 
                try:
11217 kshitij.so 54 
                    metric = metrics.findAll('input', {'type': 'submit'})
55 
                except AttributeError:
56 
                    continue
57 
                try:
58 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 59 
                except TypeError:
60 
                    continue
11217 kshitij.so 61 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 62 
                try:
63 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
64 
                except ValueError:
65 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
66 
                temp['buyTrend']=buyTrend
67 
                dataMetric = dataMetrics.split(';')
68 
                sellerCode = dataMetric[0]
69 
                temp['sellerCode']=sellerCode
70 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 71 
                if not temp.has_key('shippingTime'):
72 
                    print "Populating shipping time from metrics"
73 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 74 
                temp['sellerScore'] = int(dataMetric[4])
75 
                info.append(temp)
76 
        evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
77 
        for data in evenSeller:
78 
            temp={}
79 
            price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
11668 kshitij.so 80 
            try:
81 
                businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
82 
                shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace('  ','-')
83 
                temp['shippingTime']=shippingTime
84 
            except:
85 
                pass
10503 kshitij.so 86 
            temp['sellingPrice']=float(price)
87 
            for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
88 
                sellerName = sellerInfo.find('a').string
89 
                temp['sellerName'] = sellerName
90 
            for metrics in data.find("div",{"class":"fk-text-right"}):
91 
                try:
11217 kshitij.so 92 
                    metric = metrics.findAll('input', {'type': 'submit'})
93 
                except AttributeError:
94 
                    continue
95 
                try:
96 
                    inputTags = metric[0]['data-lst-buytrend']
10503 kshitij.so 97 
                except TypeError:
98 
                    continue
11217 kshitij.so 99 
                dataMetrics = metric[0]['data-listing-metrics']
10503 kshitij.so 100 
                try:
101 
                    buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
102 
                except ValueError:
103 
                    buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
104 
                temp['buyTrend']=buyTrend
105 
                dataMetric = dataMetrics.split(';')
106 
                temp['sellerCode'] = dataMetric[0]
107 
                temp['sellingPriceMetric'] = float(dataMetric[1])
11668 kshitij.so 108 
                if not temp.has_key('shippingTime'):
109 
                    print "Populating shipping time from metrics"
110 
                    temp['shippingTime'] = dataMetric[3]
10503 kshitij.so 111 
                temp['sellerScore'] = int(dataMetric[4])
112 
                info.append(temp)
113 
        return info
114 
 
115 
if __name__ == '__main__':
116 
    scraper = FlipkartScraper()
11967 kshitij.so 117 
    scraper.read('http://www.flipkart.com/ps/ACCDSN84XXH5P9WG')
10503 kshitij.so 118 
    print scraper.createData()
119