| 13831 |
kshitij.so |
1 |
import urllib2
|
|
|
2 |
from BeautifulSoup import BeautifulSoup
|
|
|
3 |
import re
|
|
|
4 |
from sys import exit
|
|
|
5 |
|
|
|
6 |
class FlipkartScraper:
|
|
|
7 |
def __init__(self):
|
|
|
8 |
self.count_trials = 0
|
|
|
9 |
self.redirectCount = 0
|
|
|
10 |
|
|
|
11 |
def read(self, url):
|
|
|
12 |
#url = url.replace('www.flipkart.com','163.53.77.21')
|
|
|
13 |
url = url.replace('www.flipkart.com','163.53.76.55')
|
|
|
14 |
print url
|
|
|
15 |
request = urllib2.Request(url)
|
|
|
16 |
request.add_header('Accept','text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8')
|
|
|
17 |
#request.add_header('Accept-Charset','ISO-8859-1,utf-8;q=0.7,*;q=0.3')
|
|
|
18 |
#request.add_header('Accept-Encoding','gzip,deflate,sdch')
|
|
|
19 |
request.add_header('Accept-Language','en-US,en;q=0.8,hi;q=0.6')
|
|
|
20 |
request.add_header('Connection','keep-alive')
|
|
|
21 |
request.add_header('Cookie','T=TI141257426738726661427143281839817329423126740566618323641725716448; __sonar=7237334677420142002; __gads=ID=c8b82101a0e4f451:T=1412574724:S=ALNI_MbPMbEOZj2nAGjM54z8ZHFMqwTOTQ; FK-CMP-DATA=; SN=2.VI11FB3FB6ED9D4693A796AB8C965B3417.SI802C325AC43444858830E870C4FD3324.VS141257426735693951472.1412576209; VID=2.VI11FB3FB6ED9D4693A796AB8C965B3417.1412576209.VS141257426735693951472; NSID=2.SI802C325AC43444858830E870C4FD3324.1412576209.VI11FB3FB6ED9D4693A796AB8C965B3417; __utma=19769839.709301254.1412574234.1412574234.1412574234.1; __utmb=19769839.23.10.1412574234; __utmc=19769839; __utmz=19769839.1412574234.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none); s_cc=true; gpv_pn=SellerListing%3AMobile%3AMicromax%20Canvas%20Fire%20A093; gpv_pn_t=no%20value; s_sq=%5B%5BB%5D%5D; s_ppv=36')
|
|
|
22 |
request.add_header('Cache-Control','max-age=0')
|
|
|
23 |
request.add_header('Host','www.flipkart.com')
|
|
|
24 |
request.add_header('User-Agent','Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/37.0.2062.120 Safari/537.36')
|
|
|
25 |
response_data = ""
|
|
|
26 |
redirect_url = ""
|
|
|
27 |
try:
|
|
|
28 |
response = urllib2.urlopen(request)
|
|
|
29 |
response_data = response.read()
|
|
|
30 |
print "Fetched response from flipkart for %s" %(url)
|
|
|
31 |
redirect_url = response.url
|
|
|
32 |
|
|
|
33 |
except Exception as e:
|
|
|
34 |
print 'ERROR: ', e
|
|
|
35 |
print 'Retrying'
|
|
|
36 |
self.count_trials += 1
|
|
|
37 |
|
|
|
38 |
if self.count_trials < 3:
|
|
|
39 |
return self.read(url)
|
|
|
40 |
|
|
|
41 |
self.response_data=response_data
|
|
|
42 |
return self.createData(url,redirect_url)
|
|
|
43 |
|
|
|
44 |
def scrapeRedirectedPage(self,soup,redirect_url):
|
|
|
45 |
print soup
|
|
|
46 |
print redirect_url
|
|
|
47 |
t = soup.find("div" , {"class" : "seller-table fk-user-select-none line"})
|
|
|
48 |
print t
|
|
|
49 |
table_rows = t.findAll("tr" , {"class" : re.compile('t-row.*')})
|
|
|
50 |
print table_rows
|
|
|
51 |
for x in table_rows:
|
|
|
52 |
print x
|
|
|
53 |
|
|
|
54 |
def createData(self,url, redirect_url):
|
|
|
55 |
print "Creating soup from flipkart data for %s" %(url)
|
|
|
56 |
#redirect_url = redirect_url.replace('www.flipkart.com','163.53.77.21')
|
|
|
57 |
print "Redirect url is %s"%(redirect_url)
|
|
|
58 |
page=self.response_data.decode("utf-8")
|
|
|
59 |
self.soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
|
|
|
60 |
page = None
|
|
|
61 |
self.response_data = None
|
|
|
62 |
print "Soup created from flipkart data for %s" %(url)
|
|
|
63 |
if (url==redirect_url):
|
|
|
64 |
return self.scrape(self.soup,url)
|
|
|
65 |
else:
|
|
|
66 |
print self.redirectCount
|
|
|
67 |
self.redirectCount+=1
|
|
|
68 |
if self.redirectCount >5:
|
|
|
69 |
raise
|
|
|
70 |
return self.read(url)
|
|
|
71 |
|
|
|
72 |
|
|
|
73 |
|
|
|
74 |
|
|
|
75 |
def scrape(self,soup,url):
|
|
|
76 |
print "Inside json creator for %s" %(url)
|
|
|
77 |
info = []
|
|
|
78 |
oddSeller = soup.findAll("div" , {"class" : "line seller-item odd "})
|
|
|
79 |
for data in oddSeller:
|
|
|
80 |
temp={}
|
|
|
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
|
|
|
87 |
price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.').strip()
|
|
|
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:
|
|
|
94 |
metric = metrics.findAll('input', {'type': 'submit'})
|
|
|
95 |
except AttributeError:
|
|
|
96 |
continue
|
|
|
97 |
try:
|
|
|
98 |
inputTags = metric[0]['data-lst-buytrend']
|
|
|
99 |
except TypeError:
|
|
|
100 |
continue
|
|
|
101 |
dataMetrics = metric[0]['data-listing-metrics']
|
|
|
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 |
sellerCode = dataMetric[0]
|
|
|
109 |
temp['sellerCode']=sellerCode
|
|
|
110 |
temp['sellingPriceMetric'] = float(dataMetric[1])
|
|
|
111 |
if not temp.has_key('shippingTime'):
|
|
|
112 |
print "Populating shipping time from metrics"
|
|
|
113 |
temp['shippingTime'] = dataMetric[3]
|
|
|
114 |
temp['sellerScore'] = int(dataMetric[4])
|
|
|
115 |
info.append(temp)
|
|
|
116 |
evenSeller = soup.findAll("div" , {"class" : "line seller-item even "})
|
|
|
117 |
for data in evenSeller:
|
|
|
118 |
temp={}
|
|
|
119 |
price = data.find('span', attrs={'class' : re.compile('pxs-final-price.*')}).string.strip('Rs.')
|
|
|
120 |
try:
|
|
|
121 |
businessDays = data.find('span', attrs={'class' : re.compile('fk-deliverable.*')})
|
|
|
122 |
shippingTime = businessDays.find('span', attrs={'class' : re.compile('fk-bold')}).string.replace('to','').replace('business days.','').strip().replace(' ','-')
|
|
|
123 |
temp['shippingTime']=shippingTime
|
|
|
124 |
except:
|
|
|
125 |
pass
|
|
|
126 |
temp['sellingPrice']=float(price)
|
|
|
127 |
for sellerInfo in data.findAll("div",{"class":re.compile(".*seller-info*")}):
|
|
|
128 |
sellerName = sellerInfo.find('a').string
|
|
|
129 |
temp['sellerName'] = sellerName
|
|
|
130 |
for metrics in data.find("div",{"class":"fk-text-right"}):
|
|
|
131 |
try:
|
|
|
132 |
metric = metrics.findAll('input', {'type': 'submit'})
|
|
|
133 |
except AttributeError:
|
|
|
134 |
continue
|
|
|
135 |
try:
|
|
|
136 |
inputTags = metric[0]['data-lst-buytrend']
|
|
|
137 |
except TypeError:
|
|
|
138 |
continue
|
|
|
139 |
dataMetrics = metric[0]['data-listing-metrics']
|
|
|
140 |
try:
|
|
|
141 |
buyTrend = inputTags[0:str(inputTags).index('NWSR')].replace('_','')
|
|
|
142 |
except ValueError:
|
|
|
143 |
buyTrend = inputTags[0:str(inputTags).index('WSR')].replace('_','')
|
|
|
144 |
temp['buyTrend']=buyTrend
|
|
|
145 |
dataMetric = dataMetrics.split(';')
|
|
|
146 |
temp['sellerCode'] = dataMetric[0]
|
|
|
147 |
temp['sellingPriceMetric'] = float(dataMetric[1])
|
|
|
148 |
if not temp.has_key('shippingTime'):
|
|
|
149 |
print "Populating shipping time from metrics"
|
|
|
150 |
temp['shippingTime'] = dataMetric[3]
|
|
|
151 |
temp['sellerScore'] = int(dataMetric[4])
|
|
|
152 |
info.append(temp)
|
|
|
153 |
print info
|
|
|
154 |
print "Returning Json response from flipkart for %s" %(url)
|
|
|
155 |
return info
|
|
|
156 |
|
|
|
157 |
if __name__ == '__main__':
|
|
|
158 |
scraper = FlipkartScraper()
|
|
|
159 |
scraper.read('http://www.flipkart.com/ps/MOBDZB3Q8WJNKVHG')
|