Subversion Repositories SmartDukaan

Rev

Rev 14380 | Rev 19240 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
13754 kshitij.so 1 
import urllib2
2 
from BeautifulSoup import BeautifulSoup
3 
import pymongo
4 
import re
18284 kshitij.so 5 
from dtr.utils.utils import to_java_date, fetchResponseUsingProxy
13828 kshitij.so 6 
from datetime import datetime
14257 kshitij.so 7 
import optparse
14379 kshitij.so 8 
import smtplib
9 
from email.mime.text import MIMEText
10 
from email.mime.multipart import MIMEMultipart
18284 kshitij.so 11 
from dtr.utils.utils import fetchResponseUsingProxy
13754 kshitij.so 12 
 
13 
con = None
14257 kshitij.so 14 
parser = optparse.OptionParser()
15 
parser.add_option("-m", "--m", dest="mongoHost",
16 
                      default="localhost",
17 
                      type="string", help="The HOST where the mongo server is running",
18 
                      metavar="mongo_host")
19 
 
20 
(options, args) = parser.parse_args()
21 
 
13754 kshitij.so 22 
bestSellers = []
13828 kshitij.so 23 
now = datetime.now()
14379 kshitij.so 24 
exceptionList = []
13754 kshitij.so 25 
 
26 
class __RankInfo:
27 
 
14379 kshitij.so 28 
    def __init__(self, identifier, rank, category):
13754 kshitij.so 29 
        self.identifier = identifier
30 
        self.rank  = rank
14379 kshitij.so 31 
        self.category = category
13754 kshitij.so 32 
 
14257 kshitij.so 33 
def get_mongo_connection(host=options.mongoHost, port=27017):
13754 kshitij.so 34 
    global con
35 
    if con is None:
36 
        print "Establishing connection %s host and port %d" %(host,port)
37 
        try:
38 
            con = pymongo.MongoClient(host, port)
39 
        except Exception, e:
18284 kshitij.so 40 
 
13754 kshitij.so 41 
            print e
42 
            return None
43 
    return con
44 
 
45 
def getSoupObject(url):
46 
    print "Getting soup object for"
47 
    print url
18284 kshitij.so 48 
    soup = None
49 
    response_data = fetchResponseUsingProxy(url)
50 
    print response_data
51 
    try:
52 
        page=response_data.decode("utf-8")
53 
        soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
54 
    except:
55 
        soup = BeautifulSoup(response_data,convertEntities=BeautifulSoup.HTML_ENTITIES)
56 
    return soup
13754 kshitij.so 57 
 
58 
def scrapeBestSellerMobiles():
59 
    global bestSellers
60 
    rank = 0
61 
    for i in [1, 21, 41, 61, 81]:
62 
        url = "http://www.flipkart.com/mobiles/~bestsellers/pr?p[]=sort=popularity&sid=tyy,4io&start=%d&ajax=true" %(i)
63 
        soup = getSoupObject(url)
64 
        product_divs = soup.findAll('div',{'class':re.compile('.*browse-product')})
65 
        for x in product_divs:
66 
            rank  = rank +1
67 
            print rank,
68 
            print '\t',
69 
            print x['data-pid']
14379 kshitij.so 70 
            r_info = __RankInfo(x['data-pid'].strip(),rank, None)
13754 kshitij.so 71 
            bestSellers.append(r_info)
72 
 
73 
def scrapeBestSellerTablets():
74 
    global bestSellers
75 
    bestSellers = []
76 
    rank = 0
77 
    for i in [1, 21, 41, 61, 81]:
78 
        url = "http://www.flipkart.com/tablets/~bestsellers/pr?p[]=sort=popularity&sid=tyy,hry&start=%d&ajax=true" %(i)
79 
        soup = getSoupObject(url)
80 
        product_divs = soup.findAll('div',{'class':re.compile('.*browse-product')})
81 
        for x in product_divs:
82 
            rank  = rank +1
83 
            print rank,
84 
            print '\t',
85 
            print x['data-pid']
14379 kshitij.so 86 
            r_info = __RankInfo(x['data-pid'].strip(),rank, None)
13754 kshitij.so 87 
            bestSellers.append(r_info)
88 
 
14379 kshitij.so 89 
def commitBestSellers(category):
90 
    global exceptionList
13754 kshitij.so 91 
    for x in bestSellers:
92 
        print x.rank,
93 
        print '\t',
94 
        print x.identifier,
95 
        col = get_mongo_connection().Catalog.MasterData.find({'identifier':x.identifier.strip()})
96 
        print "count sku",
97 
        print '\t',
14379 kshitij.so 98 
        if len(list(col)) == 0:
99 
            x.category = category
100 
            exceptionList.append(x)
101 
        else:
102 
            get_mongo_connection().Catalog.MasterData.update({'identifier':x.identifier.strip()}, {'$set' : {'rank':x.rank,'updatedOn':to_java_date(now)}}, multi=True)
13754 kshitij.so 103 
 
104 
def resetRanks(category):
13828 kshitij.so 105 
    oldRankedItems = get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':2,'category':category})
106 
    for item in oldRankedItems:
107 
        get_mongo_connection().Catalog.MasterData.update({'_id':item['_id']}, {'$set' : {'rank':0,'updatedOn':to_java_date(now)}}, multi=True)
14379 kshitij.so 108 
 
109 
def sendEmail():
110 
    message="""<html>
111 
            <body>
112 
            <h3>Flipkart Best Sellers not in master</h3>
113 
            <table border="1" style="width:100%;">
114 
            <thead>
115 
            <tr><th>Identifier</th>
116 
            <th>Category</th>
117 
            <th>Rank</th>
118 
            </tr></thead>
119 
            <tbody>"""
120 
    for item in exceptionList:
121 
        message+="""<tr>
122 
        <td style="text-align:center">"""+(item.identifier)+"""</td>
123 
        <td style="text-align:center">"""+(item.category)+"""</td>
124 
        <td style="text-align:center">"""+str(item.rank)+"""</td>
125 
        </tr>"""
126 
    message+="""</tbody></table></body></html>"""
127 
    print message
18284 kshitij.so 128 
    recipients = ['kshitij.sood@saholic.com']
129 
    #recipients = ['rajneesh.arora@saholic.com','kshitij.sood@saholic.com','chaitnaya.vats@saholic.com','manoj.kumar@saholic.com']
14379 kshitij.so 130 
    msg = MIMEMultipart()
131 
    msg['Subject'] = "Flipkart Best Sellers" + ' - ' + str(datetime.now())
132 
    msg['From'] = ""
133 
    msg['To'] = ",".join(recipients)
134 
    msg.preamble = "Flipkart Best Sellers" + ' - ' + str(datetime.now())
135 
    html_msg = MIMEText(message, 'html')
136 
    msg.attach(html_msg)
137 
 
138 
    smtpServer = smtplib.SMTP('localhost')
139 
    smtpServer.set_debuglevel(1)
140 
    sender = 'dtr@shop2020.in'
141 
    try:
142 
        smtpServer.sendmail(sender, recipients, msg.as_string())
143 
        print "Successfully sent email"
144 
    except:
145 
        print "Error: unable to send email."
13754 kshitij.so 146 
 
147 
def main():
148 
    scrapeBestSellerMobiles()
149 
    if len(bestSellers) > 0:
150 
        resetRanks('Mobiles')
14379 kshitij.so 151 
        commitBestSellers('MOBILES')
13754 kshitij.so 152 
    scrapeBestSellerTablets()
153 
    if len(bestSellers) > 0:
154 
        resetRanks('Tablets')
14379 kshitij.so 155 
        commitBestSellers('TABLETS')
156 
    sendEmail()
13754 kshitij.so 157 
 
158 
if __name__=='__main__':
159 
    main()