| 14844 |
amit.gupta |
1 |
'''
|
|
|
2 |
Created on Apr 15, 2015
|
|
|
3 |
|
|
|
4 |
@author: amit
|
|
|
5 |
'''
|
| 14875 |
amit.gupta |
6 |
from dtr.reports.affiliatereco import SMTP_SERVER, SMTP_PORT, SENDER, PASSWORD
|
| 14844 |
amit.gupta |
7 |
from dtr.reports.usersegmentation import TMP_FILE
|
|
|
8 |
from dtr.sources.spice import todict
|
|
|
9 |
from dtr.utils.utils import fetchResponseUsingProxy, sendNotification
|
|
|
10 |
from email import encoders
|
|
|
11 |
from email.mime.base import MIMEBase
|
|
|
12 |
from email.mime.multipart import MIMEMultipart
|
|
|
13 |
from email.mime.text import MIMEText
|
|
|
14 |
from pymongo.mongo_client import MongoClient
|
|
|
15 |
import json
|
|
|
16 |
import smtplib
|
| 14875 |
amit.gupta |
17 |
import traceback
|
| 14844 |
amit.gupta |
18 |
import xlwt
|
|
|
19 |
|
|
|
20 |
|
|
|
21 |
|
|
|
22 |
|
|
|
23 |
PRODUCT_DETAIL_API = "https://catalog.paytm.com/v1/p/%s"
|
| 14875 |
amit.gupta |
24 |
OFFER_DETAIL_API = "https://paytm.com/papi/v1/promosearch/product/%s/offers"
|
| 14844 |
amit.gupta |
25 |
client = MongoClient('mongodb://localhost:27017/')
|
|
|
26 |
|
|
|
27 |
boldStyle = xlwt.XFStyle()
|
|
|
28 |
f = xlwt.Font()
|
|
|
29 |
f.bold = True
|
|
|
30 |
boldStyle.font = f
|
|
|
31 |
|
|
|
32 |
|
|
|
33 |
def sendmail(email, message, fileName, title):
|
|
|
34 |
if email == "":
|
|
|
35 |
return
|
|
|
36 |
mailServer = smtplib.SMTP(SMTP_SERVER, SMTP_PORT)
|
|
|
37 |
mailServer.ehlo()
|
|
|
38 |
mailServer.starttls()
|
|
|
39 |
mailServer.ehlo()
|
|
|
40 |
|
|
|
41 |
# Create the container (outer) email message.
|
|
|
42 |
msg = MIMEMultipart()
|
|
|
43 |
msg['Subject'] = title
|
|
|
44 |
msg.preamble = title
|
|
|
45 |
html_msg = MIMEText(message, 'html')
|
|
|
46 |
msg.attach(html_msg)
|
|
|
47 |
|
|
|
48 |
fileMsg = MIMEBase('application', 'vnd.ms-excel')
|
|
|
49 |
fileMsg.set_payload(file(TMP_FILE).read())
|
|
|
50 |
encoders.encode_base64(fileMsg)
|
|
|
51 |
fileMsg.add_header('Content-Disposition', 'attachment;filename=' + fileName)
|
|
|
52 |
msg.attach(fileMsg)
|
|
|
53 |
|
|
|
54 |
|
|
|
55 |
email.append('amit.gupta@shop2020.in')
|
|
|
56 |
MAILTO = email
|
|
|
57 |
mailServer.login(SENDER, PASSWORD)
|
|
|
58 |
mailServer.sendmail(PASSWORD, MAILTO, msg.as_string())
|
|
|
59 |
|
|
|
60 |
def scrapepaytm():
|
| 14875 |
amit.gupta |
61 |
global i
|
| 14844 |
amit.gupta |
62 |
db = client.Catalog
|
|
|
63 |
workbook = xlwt.Workbook()
|
|
|
64 |
worksheet = workbook.add_sheet("Sample")
|
| 14875 |
amit.gupta |
65 |
worksheet2 = workbook.add_sheet("Codes")
|
|
|
66 |
add_headers(worksheet, worksheet2)
|
| 14844 |
amit.gupta |
67 |
row = 0
|
| 14875 |
amit.gupta |
68 |
row2 = 0
|
|
|
69 |
offermap = {}
|
| 14844 |
amit.gupta |
70 |
for sku in db.MasterData.find({"source":"paytm.com"}):
|
|
|
71 |
try:
|
|
|
72 |
prodUrl = sku.get("url")
|
|
|
73 |
prodIdentifier = prodUrl.split("?")[0].split("/")[-1]
|
|
|
74 |
apiUrl = PRODUCT_DETAIL_API%(prodIdentifier)
|
|
|
75 |
response = fetchResponseUsingProxy(apiUrl)
|
|
|
76 |
prod = json.loads(response)
|
| 16380 |
amit.gupta |
77 |
if prod.get("error"):
|
|
|
78 |
continue
|
| 14875 |
amit.gupta |
79 |
offerUrl = OFFER_DETAIL_API%(prod['parent_id'])
|
|
|
80 |
response = fetchResponseUsingProxy(offerUrl)
|
| 14879 |
amit.gupta |
81 |
row += 1
|
| 14875 |
amit.gupta |
82 |
offers = json.loads(response)
|
|
|
83 |
codes=[]
|
|
|
84 |
for code in offers.get("codes"):
|
|
|
85 |
offercode = code.get("code")
|
|
|
86 |
row2 += 1
|
|
|
87 |
i =-1
|
|
|
88 |
worksheet2.write(row2, inc(), prod['parent_id'])
|
|
|
89 |
worksheet2.write(row2, inc(), offercode)
|
|
|
90 |
worksheet2.write(row2, inc(), code['offerText'])
|
|
|
91 |
worksheet2.write(row2, inc(), code['terms'])
|
|
|
92 |
worksheet2.write(row2, inc(), code['priority'])
|
|
|
93 |
worksheet2.write(row2, inc(), code['valid_upto'])
|
|
|
94 |
codes.append(offercode)
|
|
|
95 |
if not offermap.has_key(offercode):
|
|
|
96 |
offermap[offercode] = code
|
|
|
97 |
|
|
|
98 |
|
|
|
99 |
allCodes = ",".join(codes)
|
|
|
100 |
i=-1
|
| 14844 |
amit.gupta |
101 |
worksheet.write(row, inc(), prod["product_id"])
|
|
|
102 |
worksheet.write(row, inc(), prod['parent_id'])
|
|
|
103 |
worksheet.write(row, inc(), prod['instock'])
|
|
|
104 |
worksheet.write(row, inc(), prod['pay_type_supported']['COD'])
|
| 14878 |
amit.gupta |
105 |
worksheet.write(row, inc(), prod['name'])
|
| 14844 |
amit.gupta |
106 |
worksheet.write(row, inc(), prod['actual_price'])
|
|
|
107 |
worksheet.write(row, inc(), prod['offer_price'])
|
|
|
108 |
sellerValues = prod['other_sellers'].get("values")
|
| 14875 |
amit.gupta |
109 |
if sellerValues is not None and len(sellerValues) > 0:
|
|
|
110 |
lowestPrice = sellerValues[0]['offer_price']
|
|
|
111 |
if lowestPrice >= prod['offer_price']:
|
|
|
112 |
worksheet.write(row, inc(), prod['offer_price'])
|
|
|
113 |
else:
|
|
|
114 |
worksheet.write(row, inc(), lowestPrice)
|
| 14844 |
amit.gupta |
115 |
else:
|
| 14875 |
amit.gupta |
116 |
worksheet.write(row, inc(), "")
|
|
|
117 |
|
|
|
118 |
worksheet.write(row, inc(), prod['promo_text'])
|
|
|
119 |
worksheet.write(row, inc(), allCodes)
|
| 14844 |
amit.gupta |
120 |
except:
|
| 14875 |
amit.gupta |
121 |
traceback.print_exc()
|
|
|
122 |
|
|
|
123 |
row = 0
|
|
|
124 |
# for key, val in offermap.iteritems():
|
|
|
125 |
# row += 1
|
|
|
126 |
# i =-1
|
|
|
127 |
# worksheet2.write(row, inc(), key)
|
|
|
128 |
# worksheet2.write(row, inc(), val['offerText'])
|
|
|
129 |
# worksheet2.write(row, inc(), val['terms'])
|
|
|
130 |
# worksheet2.write(row, inc(), val['priority'])
|
|
|
131 |
# worksheet2.write(row, inc(), val['valid_upto'])
|
| 14877 |
amit.gupta |
132 |
workbook.save("/paytmcatalog.xls")
|
| 14875 |
amit.gupta |
133 |
def add_headers(worksheet, worksheet2):
|
| 14844 |
amit.gupta |
134 |
global i
|
| 14875 |
amit.gupta |
135 |
i=-1
|
| 14844 |
amit.gupta |
136 |
worksheet.write(0, inc(), 'productid')
|
|
|
137 |
worksheet.write(0, inc(), 'parentid')
|
|
|
138 |
worksheet.write(0, inc(), 'instock')
|
|
|
139 |
worksheet.write(0, inc(), 'cod supported')
|
| 14875 |
amit.gupta |
140 |
worksheet.write(0, inc(), 'name')
|
| 14844 |
amit.gupta |
141 |
worksheet.write(0, inc(), 'actual price')
|
|
|
142 |
worksheet.write(0, inc(), 'box price')
|
|
|
143 |
worksheet.write(0,inc(), 'lowest price')
|
|
|
144 |
worksheet.write(0,inc(), 'promo text')
|
| 14875 |
amit.gupta |
145 |
worksheet.write(0,inc(), 'promocodes')
|
|
|
146 |
i=-1
|
|
|
147 |
worksheet2.write(0, inc(), 'productid')
|
|
|
148 |
worksheet2.write(0, inc(), 'code')
|
|
|
149 |
worksheet2.write(0, inc(), 'offer_text')
|
|
|
150 |
worksheet2.write(0, inc(), 'terms')
|
|
|
151 |
worksheet2.write(0, inc(), 'priority')
|
|
|
152 |
worksheet2.write(0, inc(), 'valid_upto')
|
| 14844 |
amit.gupta |
153 |
|
| 14875 |
amit.gupta |
154 |
|
|
|
155 |
def inc():
|
|
|
156 |
global i
|
|
|
157 |
i+=1
|
|
|
158 |
return i
|
| 14844 |
amit.gupta |
159 |
|
|
|
160 |
|
|
|
161 |
def main():
|
| 14875 |
amit.gupta |
162 |
scrapepaytm()
|
|
|
163 |
#refunds = db.refund.findOne({"batch":1428949802})
|
| 14844 |
amit.gupta |
164 |
#refunds = [{"timestamp" : "2015-04-14 00:00:02", "userId" : 2, "batch" : 1428949802, "userAmount" : 847 }]
|
| 14875 |
amit.gupta |
165 |
#for ref in refunds:
|
|
|
166 |
#sendNotification([ref.get("userId")], 'Batch Credit', 'Cashback Credited', 'Rs.%s was added to your wallet yesterday'%(ref.get("userAmount")), 'url', 'http://api.profittill.com/cashbacks/mine?user_id=%s'%(ref.get("userId")))
|
| 14844 |
amit.gupta |
167 |
|
|
|
168 |
if __name__ == '__main__':
|
|
|
169 |
main()
|