| 17076 |
manish.sha |
1 |
import urllib2
|
|
|
2 |
import simplejson as json
|
|
|
3 |
import pymongo
|
|
|
4 |
from dtr.utils.utils import to_java_date, getNlcPoints
|
|
|
5 |
from datetime import datetime, timedelta
|
|
|
6 |
import time
|
|
|
7 |
from multiprocessing import Pool as ThreadPool
|
|
|
8 |
from multiprocessing import cpu_count
|
|
|
9 |
import optparse
|
|
|
10 |
from dtr.storage.MemCache import MemCache
|
| 20347 |
kshitij.so |
11 |
from dtr.utils.utils import getCashBack, get_mongo_connection, SOURCE_MAP, DEAL_PRIORITY
|
| 17076 |
manish.sha |
12 |
import traceback
|
|
|
13 |
from operator import itemgetter
|
|
|
14 |
import chardet
|
|
|
15 |
from dtr.utils import HomeShop18Scraper
|
|
|
16 |
|
|
|
17 |
|
| 19191 |
kshitij.so |
18 |
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5,'PAYTM.COM':6, 'HOMESHOP18.COM':7}
|
|
|
19 |
|
| 17076 |
manish.sha |
20 |
con = None
|
|
|
21 |
|
|
|
22 |
parser = optparse.OptionParser()
|
|
|
23 |
parser.add_option("-m", "--m", dest="mongoHost",
|
|
|
24 |
default="localhost",
|
|
|
25 |
type="string", help="The HOST where the mongo server is running",
|
|
|
26 |
metavar="mongo_host")
|
|
|
27 |
|
|
|
28 |
(options, args) = parser.parse_args()
|
|
|
29 |
|
|
|
30 |
mc = MemCache(options.mongoHost)
|
|
|
31 |
|
| 19191 |
kshitij.so |
32 |
def getNetPriceForItem(itemId, source_id, category_id ,price):
|
|
|
33 |
cash_back_type = 0
|
|
|
34 |
cash_back = 0
|
|
|
35 |
try:
|
|
|
36 |
cashBack = getCashBack(itemId, source_id, category_id, mc, options.mongoHost)
|
|
|
37 |
if not cashBack or cashBack.get('cash_back_status')!=1:
|
|
|
38 |
cash_back_type = 0
|
|
|
39 |
cash_back = 0
|
|
|
40 |
|
|
|
41 |
else:
|
|
|
42 |
if cashBack['cash_back_type'] in (1,2):
|
|
|
43 |
|
|
|
44 |
if cashBack.get('maxCashBack') is not None:
|
|
|
45 |
|
|
|
46 |
if cashBack.get('cash_back_type') ==1 and (float(cashBack.get('cash_back'))*price)/100 > cashBack.get('maxCashBack'):
|
|
|
47 |
cashBack['cash_back_type'] = 2
|
|
|
48 |
cashBack['cash_back'] = cashBack['maxCashBack']
|
|
|
49 |
elif cashBack.get('cash_back_type') ==2 and cashBack.get('cash_back') > cashBack.get('maxCashBack'):
|
|
|
50 |
cashBack['cash_back'] = cashBack['maxCashBack']
|
|
|
51 |
else:
|
|
|
52 |
pass
|
|
|
53 |
|
|
|
54 |
|
|
|
55 |
|
|
|
56 |
cash_back_type = cashBack['cash_back_type']
|
|
|
57 |
cash_back = float(cashBack['cash_back'])
|
|
|
58 |
except Exception as cashBackEx:
|
|
|
59 |
pass
|
|
|
60 |
|
|
|
61 |
if cash_back_type ==1:
|
|
|
62 |
return (price - float(cash_back)*price/100)
|
|
|
63 |
elif cash_back_type ==2:
|
|
|
64 |
return (price - cash_back)
|
|
|
65 |
else:
|
|
|
66 |
return price
|
|
|
67 |
|
|
|
68 |
|
| 17076 |
manish.sha |
69 |
def populate():
|
|
|
70 |
toScrapMap = {}
|
| 17107 |
manish.sha |
71 |
bestSellers = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'rank':{'$gt':0}}))
|
| 17076 |
manish.sha |
72 |
for bestSeller in bestSellers:
|
| 17107 |
manish.sha |
73 |
snapdealBestSellers = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':7}))
|
| 17076 |
manish.sha |
74 |
for data in snapdealBestSellers:
|
|
|
75 |
if not toScrapMap.has_key(data['_id']):
|
|
|
76 |
data['dealFlag'] = 0
|
|
|
77 |
data['dealType'] = 0
|
|
|
78 |
toScrapMap[data['_id']] = data
|
| 17107 |
manish.sha |
79 |
dealFlagged = list(get_mongo_connection(host=options.mongoHost).Catalog.Deals.find({'source_id':7,'showDeal':1,'totalPoints':{'$gt':-100}}))
|
| 17076 |
manish.sha |
80 |
for deal in dealFlagged:
|
|
|
81 |
if not toScrapMap.has_key(deal['_id']):
|
| 17107 |
manish.sha |
82 |
data = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'_id':deal['_id']}))
|
| 17076 |
manish.sha |
83 |
data[0]['dealFlag'] = 0
|
|
|
84 |
data[0]['dealType'] = 0
|
|
|
85 |
toScrapMap[deal['_id']] = data[0]
|
| 17107 |
manish.sha |
86 |
manualDeals = list(get_mongo_connection(host=options.mongoHost).Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':7}))
|
| 17076 |
manish.sha |
87 |
for manualDeal in manualDeals:
|
|
|
88 |
if not toScrapMap.has_key(manualDeal['sku']):
|
| 17107 |
manish.sha |
89 |
data = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'_id':manualDeal['sku']}))
|
| 17076 |
manish.sha |
90 |
if len(data) > 0:
|
|
|
91 |
data[0]['dealFlag'] = 1
|
|
|
92 |
data[0]['dealType'] = manualDeal['dealType']
|
|
|
93 |
toScrapMap[manualDeal['sku']] = data[0]
|
|
|
94 |
else:
|
|
|
95 |
data = toScrapMap.get(manualDeal['sku'])
|
|
|
96 |
data['dealFlag'] = 1
|
|
|
97 |
data['dealType'] = manualDeal['dealType']
|
|
|
98 |
|
|
|
99 |
for val in toScrapMap.values():
|
|
|
100 |
updatePrices(val)
|
|
|
101 |
|
|
|
102 |
def updatePrices(data):
|
|
|
103 |
if data.get('ignorePricing') ==1:
|
|
|
104 |
print "Ignored items returning for %d"%(data['_id'])
|
|
|
105 |
return
|
|
|
106 |
if data['source_id']!=7:
|
|
|
107 |
return
|
|
|
108 |
print data['identifier']
|
|
|
109 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
|
|
110 |
print "returning"
|
|
|
111 |
return
|
|
|
112 |
|
|
|
113 |
try:
|
|
|
114 |
if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
|
|
|
115 |
print "sku id is already updated",data['_id']
|
|
|
116 |
return
|
|
|
117 |
except:
|
|
|
118 |
pass
|
|
|
119 |
|
|
|
120 |
result = None
|
|
|
121 |
try:
|
|
|
122 |
url = 'http://m.homeshop18.com/product.mobi?productId='+str(data['identifier'])
|
|
|
123 |
scraper = HomeShop18Scraper.HomeShop18Scraper()
|
|
|
124 |
result = scraper.read(url)
|
|
|
125 |
except:
|
|
|
126 |
print "Unable to scrape %d"%(data['_id'])
|
| 17107 |
manish.sha |
127 |
get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':0,'priceUpdatedOn':to_java_date(datetime.now())}})
|
|
|
128 |
get_mongo_connection(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':0}})
|
| 17076 |
manish.sha |
129 |
return
|
|
|
130 |
|
|
|
131 |
inStock = 0
|
|
|
132 |
|
|
|
133 |
lowestOfferPrice = 0
|
|
|
134 |
if result is not None:
|
|
|
135 |
lowestOfferPrice = float(result['price']+result['shippingCharge'])
|
|
|
136 |
inStock = result['inStock']
|
|
|
137 |
|
|
|
138 |
print lowestOfferPrice
|
|
|
139 |
print inStock
|
|
|
140 |
print "*************"
|
|
|
141 |
|
|
|
142 |
if lowestOfferPrice ==0:
|
|
|
143 |
inStock = 0
|
|
|
144 |
|
|
|
145 |
if inStock == 1:
|
| 19191 |
kshitij.so |
146 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('HOMESHOP18.COM'), data['category_id'], lowestOfferPrice)
|
| 17107 |
manish.sha |
147 |
get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)
|
| 19191 |
kshitij.so |
148 |
get_mongo_connection(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice , 'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)
|
| 17076 |
manish.sha |
149 |
else:
|
|
|
150 |
lowestOfferPrice = data['available_price']
|
| 19191 |
kshitij.so |
151 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('HOMESHOP18.COM'), data['category_id'], lowestOfferPrice)
|
| 17107 |
manish.sha |
152 |
get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)
|
| 19191 |
kshitij.so |
153 |
get_mongo_connection(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)
|
| 17076 |
manish.sha |
154 |
|
|
|
155 |
try:
|
|
|
156 |
recomputeDeal(data)
|
|
|
157 |
except:
|
|
|
158 |
print "Unable to compute deal for ",data['skuBundleId']
|
|
|
159 |
|
|
|
160 |
def populateNegativeDeals():
|
| 17107 |
manish.sha |
161 |
negativeDeals = get_mongo_connection(host=options.mongoHost).Catalog.NegativeDeals.find().distinct('sku')
|
| 17076 |
manish.sha |
162 |
mc.set("negative_deals", negativeDeals, 600)
|
|
|
163 |
|
|
|
164 |
def recomputeDeal(item):
|
|
|
165 |
"""Lets recompute deal for this bundle"""
|
|
|
166 |
print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
|
|
|
167 |
skuBundleId = item['skuBundleId']
|
|
|
168 |
|
| 19191 |
kshitij.so |
169 |
similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('netPriceAfterCashBack',pymongo.ASCENDING)]))
|
| 17076 |
manish.sha |
170 |
bestPrice = float("inf")
|
|
|
171 |
bestOne = None
|
|
|
172 |
bestSellerPoints = 0
|
|
|
173 |
toUpdate = []
|
|
|
174 |
prepaidBestPrice = float("inf")
|
|
|
175 |
prepaidBestOne = None
|
|
|
176 |
prepaidBestSellerPoints = 0
|
|
|
177 |
for similarItem in similarItems:
|
|
|
178 |
if similarItem['codAvailable'] ==1:
|
|
|
179 |
if mc.get("negative_deals") is None:
|
|
|
180 |
populateNegativeDeals()
|
|
|
181 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 19191 |
kshitij.so |
182 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
| 17076 |
manish.sha |
183 |
continue
|
|
|
184 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
| 19191 |
kshitij.so |
185 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
| 17076 |
manish.sha |
186 |
continue
|
| 19191 |
kshitij.so |
187 |
if similarItem.get('netPriceAfterCashBack') < bestPrice:
|
| 17076 |
manish.sha |
188 |
bestOne = similarItem
|
| 19191 |
kshitij.so |
189 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 17076 |
manish.sha |
190 |
bestSellerPoints = similarItem['bestSellerPoints']
|
| 20347 |
kshitij.so |
191 |
elif similarItem.get('netPriceAfterCashBack') == bestPrice:
|
|
|
192 |
|
|
|
193 |
try:
|
|
|
194 |
if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(bestOne['source_id']))):
|
|
|
195 |
continue
|
|
|
196 |
except:
|
|
|
197 |
traceback.print_exc()
|
|
|
198 |
|
| 17076 |
manish.sha |
199 |
bestOne = similarItem
|
| 19191 |
kshitij.so |
200 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 17076 |
manish.sha |
201 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
202 |
else:
|
|
|
203 |
pass
|
|
|
204 |
else:
|
|
|
205 |
if mc.get("negative_deals") is None:
|
|
|
206 |
populateNegativeDeals()
|
|
|
207 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 19191 |
kshitij.so |
208 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
| 17076 |
manish.sha |
209 |
continue
|
|
|
210 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
| 19191 |
kshitij.so |
211 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
| 17076 |
manish.sha |
212 |
continue
|
| 19191 |
kshitij.so |
213 |
if similarItem.get('netPriceAfterCashBack') < prepaidBestPrice:
|
| 17076 |
manish.sha |
214 |
prepaidBestOne = similarItem
|
| 19191 |
kshitij.so |
215 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 17076 |
manish.sha |
216 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
| 20347 |
kshitij.so |
217 |
elif similarItem.get('netPriceAfterCashBack') == prepaidBestPrice:
|
|
|
218 |
|
|
|
219 |
try:
|
|
|
220 |
if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(bestOne['source_id']))):
|
|
|
221 |
continue
|
|
|
222 |
except:
|
|
|
223 |
traceback.print_exc()
|
|
|
224 |
|
| 17076 |
manish.sha |
225 |
prepaidBestOne = similarItem
|
| 19191 |
kshitij.so |
226 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 17076 |
manish.sha |
227 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
|
|
228 |
else:
|
|
|
229 |
pass
|
|
|
230 |
if bestOne is not None or prepaidBestOne is not None:
|
|
|
231 |
for similarItem in similarItems:
|
|
|
232 |
toUpdate.append(similarItem['_id'])
|
|
|
233 |
if bestOne is not None:
|
|
|
234 |
toUpdate.remove(bestOne['_id'])
|
| 19191 |
kshitij.so |
235 |
get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
|
| 17076 |
manish.sha |
236 |
if prepaidBestOne is not None:
|
|
|
237 |
if bestOne is not None:
|
| 19191 |
kshitij.so |
238 |
if prepaidBestOne.get('netPriceAfterCashBack') < bestOne.get('netPriceAfterCashBack'):
|
| 17076 |
manish.sha |
239 |
toUpdate.remove(prepaidBestOne['_id'])
|
| 19191 |
kshitij.so |
240 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
| 17076 |
manish.sha |
241 |
else:
|
|
|
242 |
toUpdate.remove(prepaidBestOne['_id'])
|
| 19191 |
kshitij.so |
243 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
| 17076 |
manish.sha |
244 |
if len(toUpdate) > 0:
|
| 19191 |
kshitij.so |
245 |
get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
|
| 17076 |
manish.sha |
246 |
|
|
|
247 |
|
|
|
248 |
def main():
|
|
|
249 |
populate()
|
|
|
250 |
|
|
|
251 |
if __name__=='__main__':
|
|
|
252 |
main()
|
|
|
253 |
|
|
|
254 |
|
|
|
255 |
|
|
|
256 |
|