| 13828 |
kshitij.so |
1 |
import pymongo
|
| 20347 |
kshitij.so |
2 |
from dtr.utils.utils import to_java_date, getNlcPoints, DEAL_PRIORITY
|
| 13915 |
kshitij.so |
3 |
from datetime import datetime, timedelta
|
| 13828 |
kshitij.so |
4 |
from operator import itemgetter
|
| 14123 |
kshitij.so |
5 |
from dtr.utils import FlipkartScraper,NewFlipkartScraper
|
| 14178 |
kshitij.so |
6 |
from multiprocessing import Pool as ThreadPool
|
| 14172 |
kshitij.so |
7 |
from multiprocessing import cpu_count
|
| 14255 |
kshitij.so |
8 |
import optparse
|
| 14325 |
kshitij.so |
9 |
from dtr.storage.MemCache import MemCache
|
| 14705 |
kshitij.so |
10 |
from dtr.utils.utils import getCashBack
|
| 15269 |
kshitij.so |
11 |
import traceback
|
| 13828 |
kshitij.so |
12 |
|
|
|
13 |
con = None
|
|
|
14 |
|
| 14255 |
kshitij.so |
15 |
parser = optparse.OptionParser()
|
|
|
16 |
parser.add_option("-m", "--m", dest="mongoHost",
|
|
|
17 |
default="localhost",
|
|
|
18 |
type="string", help="The HOST where the mongo server is running",
|
|
|
19 |
metavar="mongo_host")
|
|
|
20 |
|
|
|
21 |
(options, args) = parser.parse_args()
|
|
|
22 |
|
| 14325 |
kshitij.so |
23 |
mc = MemCache(options.mongoHost)
|
|
|
24 |
|
| 16869 |
kshitij.so |
25 |
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5,'PAYTM.COM':6}
|
| 15610 |
kshitij.so |
26 |
|
| 14255 |
kshitij.so |
27 |
def get_mongo_connection(host=options.mongoHost, port=27017):
|
| 13828 |
kshitij.so |
28 |
global con
|
|
|
29 |
if con is None:
|
|
|
30 |
print "Establishing connection %s host and port %d" %(host,port)
|
|
|
31 |
try:
|
|
|
32 |
con = pymongo.MongoClient(host, port)
|
|
|
33 |
except Exception, e:
|
|
|
34 |
print e
|
|
|
35 |
return None
|
|
|
36 |
return con
|
|
|
37 |
|
| 19187 |
kshitij.so |
38 |
def getNetPriceForItem(itemId, source_id, category_id ,price):
|
|
|
39 |
cash_back_type = 0
|
|
|
40 |
cash_back = 0
|
|
|
41 |
try:
|
|
|
42 |
cashBack = getCashBack(itemId, source_id, category_id, mc, options.mongoHost)
|
|
|
43 |
if not cashBack or cashBack.get('cash_back_status')!=1:
|
|
|
44 |
cash_back_type = 0
|
|
|
45 |
cash_back = 0
|
|
|
46 |
|
|
|
47 |
else:
|
|
|
48 |
if cashBack['cash_back_type'] in (1,2):
|
|
|
49 |
|
|
|
50 |
if cashBack.get('maxCashBack') is not None:
|
|
|
51 |
|
|
|
52 |
if cashBack.get('cash_back_type') ==1 and (float(cashBack.get('cash_back'))*price)/100 > cashBack.get('maxCashBack'):
|
|
|
53 |
cashBack['cash_back_type'] = 2
|
|
|
54 |
cashBack['cash_back'] = cashBack['maxCashBack']
|
|
|
55 |
elif cashBack.get('cash_back_type') ==2 and cashBack.get('cash_back') > cashBack.get('maxCashBack'):
|
|
|
56 |
cashBack['cash_back'] = cashBack['maxCashBack']
|
|
|
57 |
else:
|
|
|
58 |
pass
|
|
|
59 |
|
|
|
60 |
|
|
|
61 |
|
|
|
62 |
cash_back_type = cashBack['cash_back_type']
|
|
|
63 |
cash_back = float(cashBack['cash_back'])
|
|
|
64 |
except Exception as cashBackEx:
|
|
|
65 |
pass
|
|
|
66 |
|
|
|
67 |
if cash_back_type ==1:
|
|
|
68 |
return (price - float(cash_back)*price/100)
|
|
|
69 |
elif cash_back_type ==2:
|
|
|
70 |
return (price - cash_back)
|
|
|
71 |
else:
|
|
|
72 |
return price
|
|
|
73 |
|
|
|
74 |
|
| 14149 |
kshitij.so |
75 |
def populate():
|
|
|
76 |
toScrapMap = {}
|
| 14131 |
kshitij.so |
77 |
bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
|
|
|
78 |
for bestSeller in bestSellers:
|
| 14149 |
kshitij.so |
79 |
amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':2}))
|
|
|
80 |
for data in amazonBestSellers:
|
|
|
81 |
if not toScrapMap.has_key(data['_id']):
|
| 15269 |
kshitij.so |
82 |
data['dealFlag'] = 0
|
|
|
83 |
data['dealType'] = 0
|
| 14149 |
kshitij.so |
84 |
toScrapMap[data['_id']] = data
|
| 16175 |
kshitij.so |
85 |
dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':2,'showDeal':1,'totalPoints':{'$gt':-100}}))
|
| 14251 |
kshitij.so |
86 |
for deal in dealFlagged:
|
|
|
87 |
if not toScrapMap.has_key(deal['_id']):
|
| 14262 |
kshitij.so |
88 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
|
| 15269 |
kshitij.so |
89 |
data[0]['dealFlag'] = 0
|
|
|
90 |
data[0]['dealType'] = 0
|
| 14262 |
kshitij.so |
91 |
toScrapMap[deal['_id']] = data[0]
|
| 15269 |
kshitij.so |
92 |
manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':2}))
|
|
|
93 |
for manualDeal in manualDeals:
|
|
|
94 |
if not toScrapMap.has_key(manualDeal['sku']):
|
|
|
95 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
|
|
|
96 |
if len(data) > 0:
|
|
|
97 |
data[0]['dealFlag'] = 1
|
|
|
98 |
data[0]['dealType'] = manualDeal['dealType']
|
|
|
99 |
toScrapMap[manualDeal['sku']] = data[0]
|
|
|
100 |
else:
|
|
|
101 |
data = toScrapMap.get(manualDeal['sku'])
|
|
|
102 |
data['dealFlag'] = 1
|
|
|
103 |
data['dealType'] = manualDeal['dealType']
|
| 14178 |
kshitij.so |
104 |
pool = ThreadPool(cpu_count() *2)
|
| 14149 |
kshitij.so |
105 |
pool.map(scrapeFlipkart,toScrapMap.values())
|
|
|
106 |
pool.close()
|
|
|
107 |
pool.join()
|
| 14251 |
kshitij.so |
108 |
print "joining threads at %s"%(str(datetime.now()))
|
| 14149 |
kshitij.so |
109 |
|
|
|
110 |
def scrapeFlipkart(data):
|
|
|
111 |
if data['source_id']!=2:
|
| 14157 |
kshitij.so |
112 |
return
|
| 14149 |
kshitij.so |
113 |
retryCount = 0
|
|
|
114 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
| 14157 |
kshitij.so |
115 |
print "returning in valid identifier"
|
|
|
116 |
return
|
| 14149 |
kshitij.so |
117 |
|
| 16503 |
kshitij.so |
118 |
if data.get('ignorePricing') ==1:
|
| 15610 |
kshitij.so |
119 |
print "Ignored items returning for %d"%(data['_id'])
|
|
|
120 |
return
|
|
|
121 |
|
| 14149 |
kshitij.so |
122 |
try:
|
|
|
123 |
if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
|
| 15269 |
kshitij.so |
124 |
print "sku id is already updated %d" %(data['_id'])
|
| 14157 |
kshitij.so |
125 |
return
|
| 14149 |
kshitij.so |
126 |
except:
|
|
|
127 |
pass
|
|
|
128 |
|
|
|
129 |
|
|
|
130 |
lowestSp = 0
|
|
|
131 |
inStock = 0
|
| 15269 |
kshitij.so |
132 |
buyBoxPrice = 0
|
|
|
133 |
isBuyBox = 0
|
| 14157 |
kshitij.so |
134 |
scraperProductPage = NewFlipkartScraper.FlipkartProductPageScraper()
|
| 14149 |
kshitij.so |
135 |
try:
|
| 17264 |
kshitij.so |
136 |
result = scraperProductPage.read(data['identifier'])
|
|
|
137 |
if result.get('lowestSp')!=0:
|
|
|
138 |
lowestSp = result.get('lowestSp')
|
|
|
139 |
inStock = result.get('inStock')
|
|
|
140 |
buyBoxPrice = result.get('buyBoxPrice')
|
| 14149 |
kshitij.so |
141 |
except:
|
| 15269 |
kshitij.so |
142 |
print "Unable to scrape product page %s" %(data['identifier'])
|
| 17264 |
kshitij.so |
143 |
return
|
| 14149 |
kshitij.so |
144 |
|
|
|
145 |
|
|
|
146 |
print lowestSp
|
|
|
147 |
print inStock
|
| 15269 |
kshitij.so |
148 |
if buyBoxPrice is not None and buyBoxPrice == lowestSp:
|
|
|
149 |
isBuyBox = 1
|
| 14149 |
kshitij.so |
150 |
if lowestSp > 0:
|
| 19187 |
kshitij.so |
151 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('FLIPKART'), data['category_id'], lowestSp)
|
| 15269 |
kshitij.so |
152 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock,'buyBoxFlag':isBuyBox}}, multi=True)
|
| 19187 |
kshitij.so |
153 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)
|
| 14149 |
kshitij.so |
154 |
else:
|
| 19187 |
kshitij.so |
155 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('FLIPKART'), data['category_id'], data['available_price'])
|
| 15269 |
kshitij.so |
156 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now()),'buyBoxFlag':isBuyBox}}, multi=True)
|
| 19187 |
kshitij.so |
157 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}})
|
| 14149 |
kshitij.so |
158 |
|
|
|
159 |
try:
|
| 15269 |
kshitij.so |
160 |
recomputeDeal(data)
|
| 14149 |
kshitij.so |
161 |
except:
|
| 15269 |
kshitij.so |
162 |
print "Unable to compute deal for %s"%(data['skuBundleId'])
|
| 13828 |
kshitij.so |
163 |
|
| 16503 |
kshitij.so |
164 |
#def recomputePoints(item, deal):
|
|
|
165 |
# try:
|
|
|
166 |
# if item.get('available_price') == deal['available_price']:
|
|
|
167 |
# print "No need to compute points for %d , as price is still same" %(item['_id'])
|
|
|
168 |
# raise
|
|
|
169 |
# nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
|
|
|
170 |
# except:
|
|
|
171 |
# print traceback.print_exc()
|
|
|
172 |
# nlcPoints = deal['nlcPoints']
|
|
|
173 |
#
|
|
|
174 |
# bundleDealPoints = list(get_mongo_connection().Catalog.DealPoints.find({'skuBundleId':item['skuBundleId'],'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())}}))
|
|
|
175 |
# if len(bundleDealPoints) > 0:
|
|
|
176 |
# item['manualDealThresholdPrice'] = bundleDealPoints[0]['dealThresholdPrice']
|
|
|
177 |
# dealPoints = bundleDealPoints[0]['dealPoints']
|
|
|
178 |
# else:
|
|
|
179 |
# dealPoints = 0
|
|
|
180 |
# item['manualDealThresholdPrice'] = None
|
|
|
181 |
#
|
|
|
182 |
# get_mongo_connection().Catalog.Deals.update({'_id':deal['_id']},{"$set":{'totalPoints':deal['totalPoints'] - deal['nlcPoints'] + nlcPoints - deal['dealPoints'] +dealPoints , 'nlcPoints': nlcPoints, 'dealPoints': dealPoints, 'manualDealThresholdPrice': item['manualDealThresholdPrice']}})
|
| 15269 |
kshitij.so |
183 |
|
| 14325 |
kshitij.so |
184 |
def populateNegativeDeals():
|
|
|
185 |
negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
|
|
|
186 |
mc.set("negative_deals", negativeDeals, 600)
|
|
|
187 |
|
| 15269 |
kshitij.so |
188 |
def recomputeDeal(item):
|
| 13915 |
kshitij.so |
189 |
"""Lets recompute deal for this bundle"""
|
| 15269 |
kshitij.so |
190 |
print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
|
|
|
191 |
skuBundleId = item['skuBundleId']
|
| 13915 |
kshitij.so |
192 |
|
| 19187 |
kshitij.so |
193 |
similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('netPriceAfterCashBack',pymongo.ASCENDING)]))
|
| 13915 |
kshitij.so |
194 |
bestPrice = float("inf")
|
|
|
195 |
bestOne = None
|
|
|
196 |
bestSellerPoints = 0
|
|
|
197 |
toUpdate = []
|
| 16019 |
kshitij.so |
198 |
prepaidBestPrice = float("inf")
|
|
|
199 |
prepaidBestOne = None
|
|
|
200 |
prepaidBestSellerPoints = 0
|
| 13915 |
kshitij.so |
201 |
for similarItem in similarItems:
|
| 16019 |
kshitij.so |
202 |
if similarItem['codAvailable'] ==1:
|
|
|
203 |
if mc.get("negative_deals") is None:
|
|
|
204 |
populateNegativeDeals()
|
| 16175 |
kshitij.so |
205 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 16019 |
kshitij.so |
206 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
|
|
207 |
continue
|
|
|
208 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
|
|
209 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
|
|
210 |
continue
|
| 19187 |
kshitij.so |
211 |
if similarItem.get('netPriceAfterCashBack') < bestPrice:
|
| 16019 |
kshitij.so |
212 |
bestOne = similarItem
|
| 19187 |
kshitij.so |
213 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
214 |
bestSellerPoints = similarItem['bestSellerPoints']
|
| 20347 |
kshitij.so |
215 |
elif similarItem.get('netPriceAfterCashBack') == bestPrice:
|
|
|
216 |
|
|
|
217 |
try:
|
|
|
218 |
if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(bestOne['source_id']))):
|
|
|
219 |
continue
|
|
|
220 |
except:
|
|
|
221 |
traceback.print_exc()
|
|
|
222 |
|
| 16019 |
kshitij.so |
223 |
bestOne = similarItem
|
| 19187 |
kshitij.so |
224 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
225 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
226 |
else:
|
|
|
227 |
pass
|
| 13915 |
kshitij.so |
228 |
else:
|
| 16019 |
kshitij.so |
229 |
if mc.get("negative_deals") is None:
|
|
|
230 |
populateNegativeDeals()
|
| 16175 |
kshitij.so |
231 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 16019 |
kshitij.so |
232 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
|
|
233 |
continue
|
|
|
234 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
|
|
235 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
|
|
236 |
continue
|
| 19187 |
kshitij.so |
237 |
if similarItem.get('netPriceAfterCashBack') < prepaidBestPrice:
|
| 16019 |
kshitij.so |
238 |
prepaidBestOne = similarItem
|
| 19187 |
kshitij.so |
239 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
240 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
| 20347 |
kshitij.so |
241 |
elif similarItem.get('netPriceAfterCashBack') == prepaidBestPrice:
|
|
|
242 |
|
|
|
243 |
try:
|
|
|
244 |
if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(bestOne['source_id']))):
|
|
|
245 |
continue
|
|
|
246 |
except:
|
|
|
247 |
traceback.print_exc()
|
|
|
248 |
|
| 16019 |
kshitij.so |
249 |
prepaidBestOne = similarItem
|
| 19187 |
kshitij.so |
250 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
251 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
|
|
252 |
else:
|
|
|
253 |
pass
|
| 16026 |
kshitij.so |
254 |
if bestOne is not None or prepaidBestOne is not None:
|
| 13915 |
kshitij.so |
255 |
for similarItem in similarItems:
|
|
|
256 |
toUpdate.append(similarItem['_id'])
|
| 16026 |
kshitij.so |
257 |
if bestOne is not None:
|
|
|
258 |
toUpdate.remove(bestOne['_id'])
|
|
|
259 |
get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
|
|
|
260 |
if prepaidBestOne is not None:
|
| 16074 |
kshitij.so |
261 |
if bestOne is not None:
|
| 19187 |
kshitij.so |
262 |
if prepaidBestOne.get('netPriceAfterCashBack') < bestOne.get('netPriceAfterCashBack'):
|
| 16074 |
kshitij.so |
263 |
toUpdate.remove(prepaidBestOne['_id'])
|
|
|
264 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
|
|
265 |
else:
|
|
|
266 |
toUpdate.remove(prepaidBestOne['_id'])
|
|
|
267 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
| 13915 |
kshitij.so |
268 |
if len(toUpdate) > 0:
|
| 16019 |
kshitij.so |
269 |
get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
|
| 13915 |
kshitij.so |
270 |
|
| 19187 |
kshitij.so |
271 |
|
| 13828 |
kshitij.so |
272 |
def main():
|
| 14157 |
kshitij.so |
273 |
populate()
|
| 13828 |
kshitij.so |
274 |
|
|
|
275 |
if __name__=='__main__':
|
|
|
276 |
main()
|