Subversion Repositories SmartDukaan

Rev

Rev 14379 | Rev 21135 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
13828 kshitij.so 1
import urllib
2
import urllib2
3
from BeautifulSoup import BeautifulSoup
4
import pymongo
5
import re
6
from dtr.utils.utils import to_java_date
7
from datetime import datetime
14257 kshitij.so 8
import optparse
14379 kshitij.so 9
import smtplib
10
from email.mime.text import MIMEText
11
from email.mime.multipart import MIMEMultipart
13828 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
 
13828 kshitij.so 22
bestSellers = []
23
now = datetime.now()
24
mobUrl = "http://www.saholic.com/mobile-phone/10006"
25
tabUrl = "http://www.saholic.com/all-tablets/10010"
26
 
14379 kshitij.so 27
exceptionList = []
28
 
13828 kshitij.so 29
class __RankInfo:
30
 
14379 kshitij.so 31
    def __init__(self, identifier, rank, category):
13828 kshitij.so 32
        self.identifier = identifier
33
        self.rank  = rank
14379 kshitij.so 34
        self.category = category
13828 kshitij.so 35
 
14257 kshitij.so 36
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 37
    global con
38
    if con is None:
39
        print "Establishing connection %s host and port %d" %(host,port)
40
        try:
41
            con = pymongo.MongoClient(host, port)
42
        except Exception, e:
43
            print e
44
            return None
45
    return con
46
 
47
def getSoupObject(url,data):
48
    print "Getting soup object for"
49
    global RETRY_COUNT
50
    RETRY_COUNT = 1 
51
    while RETRY_COUNT < 10:
52
        try:
53
            soup = None
54
            request = urllib2.Request(url,data)
55
            request.add_header('Accept','text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8')
56
            request.add_header('Accept-Language','en-US,en;q=0.8,hi;q=0.6')
57
            request.add_header('Connection','keep-alive')
58
            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')
59
 
60
            response = urllib2.urlopen(request)   
61
            response_data = response.read()
62
            response.close()
63
            try:
64
                page=response_data.decode("utf-8")
65
                soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
66
            except:
67
                soup = BeautifulSoup(response_data,convertEntities=BeautifulSoup.HTML_ENTITIES)
68
            if soup is None:
69
                raise
70
            return soup
71
        except Exception as e:
72
            print e
73
            print "Retrying"
74
            RETRY_COUNT = RETRY_COUNT + 1
75
 
76
def scrapeBestSellerMobiles():
77
    global bestSellers
78
    rank = 1
79
    for i in range(1,6):
80
        query_args = { 'fq':'F_50028:In Stock', 'page':i }
81
        data = urllib.urlencode(query_args)
82
        soup = getSoupObject(mobUrl,data)
83
        titleTags = soup.findAll('div',{'class':'title'})
84
        for titleTag in titleTags:
85
            print titleTag.find('a').text
86
            identifier = getSaholicIdentifier(titleTag.find('a')['href'])
14379 kshitij.so 87
            r_info = __RankInfo(str(identifier), rank ,None)
13828 kshitij.so 88
            bestSellers.append(r_info)
89
            rank = rank + 1
90
 
91
def scrapeBestSellerTablets():
92
    global bestSellers
93
    bestSellers = [] 
94
    rank = 1
95
    for i in range(1,6):
96
        query_args = { 'fq':'F_50028:In Stock', 'page':i }
97
        data = urllib.urlencode(query_args)
98
        soup = getSoupObject(tabUrl, data)
99
        titleTags = soup.findAll('div',{'class':'title'})
100
        for titleTag in titleTags:
101
            print titleTag.find('a').text
102
            identifier = getSaholicIdentifier(titleTag.find('a')['href'])
14379 kshitij.so 103
            r_info = __RankInfo(str(identifier), rank, None)
13828 kshitij.so 104
            bestSellers.append(r_info)
105
            rank = rank + 1
106
 
107
def getSaholicIdentifier(url):
108
    return url[url.rfind('-')+len('-'):]
109
 
110
def resetRanks(category):
111
    oldRankedItems = get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':4,'category':category})
112
    for item in oldRankedItems:
113
        print item['_id']
114
        get_mongo_connection().Catalog.MasterData.update({'_id':item['_id']}, {'$set' : {'rank':0,'updatedOn':to_java_date(now)}}, multi=True)
115
 
14379 kshitij.so 116
def commitBestSellers(category):
117
    global exceptionList
13828 kshitij.so 118
    print "Rank",
119
    print '\t',
120
    print 'Identifier'
121
    for x in bestSellers:
122
        print x.rank,
123
        print '\t',
14057 kshitij.so 124
        print x.identifier,
13828 kshitij.so 125
        col = get_mongo_connection().Catalog.MasterData.find({'identifier':x.identifier})
126
        print "count sku",
127
        print '\t',
14379 kshitij.so 128
        if len(list(col)) == 0:
129
            x.category = category
130
            exceptionList.append(x)
131
        else:
132
            get_mongo_connection().Catalog.MasterData.update({'identifier':x.identifier,'source_id':4}, {'$set' : {'rank':x.rank,'updatedOn':to_java_date(now)}}, multi=True)
13828 kshitij.so 133
 
14379 kshitij.so 134
def sendEmail():
135
    message="""<html>
136
            <body>
137
            <h3>Saholic Best Sellers not in master</h3>
138
            <table border="1" style="width:100%;">
139
            <thead>
140
            <tr><th>Identifier</th>
141
            <th>Category</th>
142
            <th>Rank</th>
143
            </tr></thead>
144
            <tbody>"""
145
    for item in exceptionList:
146
        message+="""<tr>
147
        <td style="text-align:center">"""+(item.identifier)+"""</td>
148
        <td style="text-align:center">"""+(item.category)+"""</td>
149
        <td style="text-align:center">"""+str(item.rank)+"""</td>
150
        </tr>"""
151
    message+="""</tbody></table></body></html>"""
152
    print message
153
    #recipients = ['kshitij.sood@saholic.com']
20172 aman.kumar 154
    recipients = ['rajneesh.arora@saholic.com','kshitij.sood@saholic.com','chaitnaya.vats@saholic.com','ritesh.chauhan@saholic.com','khushal.bhatia@saholic.com']
14379 kshitij.so 155
    msg = MIMEMultipart()
156
    msg['Subject'] = "Saholic Best Sellers" + ' - ' + str(datetime.now())
157
    msg['From'] = ""
158
    msg['To'] = ",".join(recipients)
159
    msg.preamble = "Saholic Best Sellers" + ' - ' + str(datetime.now())
160
    html_msg = MIMEText(message, 'html')
161
    msg.attach(html_msg)
162
 
163
    smtpServer = smtplib.SMTP('localhost')
164
    smtpServer.set_debuglevel(1)
165
    sender = 'dtr@shop2020.in'
166
    try:
167
        smtpServer.sendmail(sender, recipients, msg.as_string())
168
        print "Successfully sent email"
169
    except:
170
        print "Error: unable to send email."        
13828 kshitij.so 171
 
172
def main():
173
    scrapeBestSellerMobiles()
174
    if len(bestSellers) > 0:
175
        resetRanks('Mobiles')
14379 kshitij.so 176
        commitBestSellers('MOBILES')
13828 kshitij.so 177
    scrapeBestSellerTablets()
178
    if len(bestSellers) > 0:
179
        resetRanks('Tablets')
14379 kshitij.so 180
        commitBestSellers('TABLETS')
181
 
182
    sendEmail()
183
 
13828 kshitij.so 184
if __name__=='__main__':
185
    main()