| 13828 |
kshitij.so |
1 |
import pymongo
|
| 15267 |
kshitij.so |
2 |
from dtr.utils.utils import to_java_date, getNlcPoints
|
| 13914 |
kshitij.so |
3 |
from datetime import datetime, timedelta
|
| 14308 |
kshitij.so |
4 |
from dtr.utils import AmazonPriceOnlyScraper, AmazonDealScraper
|
| 14174 |
kshitij.so |
5 |
from multiprocessing import Pool as ThreadPool
|
| 14170 |
kshitij.so |
6 |
from multiprocessing import cpu_count
|
| 14256 |
kshitij.so |
7 |
import optparse
|
| 14325 |
kshitij.so |
8 |
from dtr.storage.MemCache import MemCache
|
| 14705 |
kshitij.so |
9 |
from dtr.utils.utils import getCashBack
|
| 15267 |
kshitij.so |
10 |
import traceback
|
| 13828 |
kshitij.so |
11 |
|
|
|
12 |
con = None
|
|
|
13 |
|
| 14256 |
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 |
|
| 14325 |
kshitij.so |
22 |
mc = MemCache(options.mongoHost)
|
|
|
23 |
|
| 15616 |
kshitij.so |
24 |
|
| 16869 |
kshitij.so |
25 |
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5,'PAYTM.COM':6}
|
| 16019 |
kshitij.so |
26 |
|
| 14256 |
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 |
|
| 19185 |
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 |
|
| 14147 |
kshitij.so |
74 |
def populate():
|
|
|
75 |
toScrapMap = {}
|
| 14133 |
kshitij.so |
76 |
bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
|
|
|
77 |
for bestSeller in bestSellers:
|
|
|
78 |
amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
|
|
|
79 |
for data in amazonBestSellers:
|
| 14147 |
kshitij.so |
80 |
if not toScrapMap.has_key(data['_id']):
|
| 15267 |
kshitij.so |
81 |
data['dealFlag'] = 0
|
|
|
82 |
data['dealType'] = 0
|
| 16509 |
kshitij.so |
83 |
data['dealUrl'] = ""
|
| 14147 |
kshitij.so |
84 |
toScrapMap[data['_id']] = data
|
| 16173 |
kshitij.so |
85 |
dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':-100}}))
|
| 14250 |
kshitij.so |
86 |
for deal in dealFlagged:
|
|
|
87 |
if not toScrapMap.has_key(deal['_id']):
|
| 14260 |
kshitij.so |
88 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
|
| 15267 |
kshitij.so |
89 |
data[0]['dealFlag'] = 0
|
|
|
90 |
data[0]['dealType'] = 0
|
| 16509 |
kshitij.so |
91 |
data[0]['dealUrl'] = ""
|
| 14260 |
kshitij.so |
92 |
toScrapMap[deal['_id']] = data[0]
|
| 14308 |
kshitij.so |
93 |
manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':1}))
|
|
|
94 |
for manualDeal in manualDeals:
|
|
|
95 |
if not toScrapMap.has_key(manualDeal['sku']):
|
|
|
96 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
|
|
|
97 |
if len(data) > 0:
|
|
|
98 |
data[0]['dealFlag'] = 1
|
|
|
99 |
data[0]['dealType'] = manualDeal['dealType']
|
| 16509 |
kshitij.so |
100 |
data[0]['dealUrl'] = manualDeal['dealUrl']
|
| 14308 |
kshitij.so |
101 |
toScrapMap[manualDeal['sku']] = data[0]
|
|
|
102 |
else:
|
|
|
103 |
data = toScrapMap.get(manualDeal['sku'])
|
|
|
104 |
data['dealFlag'] = 1
|
|
|
105 |
data['dealType'] = manualDeal['dealType']
|
| 16509 |
kshitij.so |
106 |
data['dealUrl'] = manualDeal['dealUrl']
|
| 14170 |
kshitij.so |
107 |
pool = ThreadPool(cpu_count() * 2)
|
| 14147 |
kshitij.so |
108 |
pool.map(scrapeAmazon,toScrapMap.values())
|
|
|
109 |
pool.close()
|
|
|
110 |
pool.join()
|
| 14250 |
kshitij.so |
111 |
print "joining threads at %s"%(str(datetime.now()))
|
| 14147 |
kshitij.so |
112 |
|
|
|
113 |
|
|
|
114 |
def scrapeAmazon(data):
|
|
|
115 |
inStock = 0
|
| 14308 |
kshitij.so |
116 |
dealScraping = False
|
| 14147 |
kshitij.so |
117 |
print str(data['identifier'])
|
|
|
118 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
|
|
119 |
return
|
|
|
120 |
|
| 16501 |
kshitij.so |
121 |
if data.get('ignorePricing') ==1:
|
| 15616 |
kshitij.so |
122 |
print "Ignored items returning for %d"%(data['_id'])
|
|
|
123 |
return
|
|
|
124 |
|
| 14147 |
kshitij.so |
125 |
try:
|
|
|
126 |
if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
|
|
|
127 |
print "sku id is already updated",data['_id']
|
|
|
128 |
return
|
|
|
129 |
except:
|
|
|
130 |
pass
|
|
|
131 |
|
| 15959 |
kshitij.so |
132 |
url = "http://www.amazon.in/gp/aw/ol/%s?o=New&op=1"%(data['identifier'])
|
| 14308 |
kshitij.so |
133 |
|
|
|
134 |
try:
|
|
|
135 |
if data['dealFlag'] ==1 and data['dealType'] ==1:
|
| 16509 |
kshitij.so |
136 |
deal_url = data['dealUrl'].strip()
|
| 14308 |
kshitij.so |
137 |
dealScraping = True
|
|
|
138 |
except:
|
|
|
139 |
data['dealFlag'] = 0
|
|
|
140 |
data['dealType'] = 0
|
| 16509 |
kshitij.so |
141 |
data['dealUrl'] = ""
|
| 14308 |
kshitij.so |
142 |
|
| 14147 |
kshitij.so |
143 |
print url
|
|
|
144 |
lowestPrice = 0.0
|
| 14308 |
kshitij.so |
145 |
|
|
|
146 |
if not dealScraping:
|
|
|
147 |
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
|
|
|
148 |
lowestPrice = scraperAmazon.read(url)
|
|
|
149 |
else:
|
|
|
150 |
dealScraperAmazon = AmazonDealScraper.AmazonScraper()
|
|
|
151 |
lowestPrice = dealScraperAmazon.read(deal_url)
|
|
|
152 |
if lowestPrice == 0:
|
|
|
153 |
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
|
|
|
154 |
lowestPrice = scraperAmazon.read(url)
|
|
|
155 |
dealScraping = False
|
|
|
156 |
|
| 14147 |
kshitij.so |
157 |
print lowestPrice
|
|
|
158 |
if lowestPrice > 0:
|
|
|
159 |
inStock = 1
|
| 19185 |
kshitij.so |
160 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('AMAZON'), data['category_id'], lowestPrice)
|
|
|
161 |
else:
|
|
|
162 |
netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('AMAZON'), data['category_id'], data['available_price'])
|
|
|
163 |
|
| 14147 |
kshitij.so |
164 |
print lowestPrice
|
|
|
165 |
print inStock
|
|
|
166 |
if lowestPrice > 0:
|
|
|
167 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)
|
| 19185 |
kshitij.so |
168 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock,'dealType':data['dealType'],'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)
|
| 14147 |
kshitij.so |
169 |
else:
|
|
|
170 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)
|
| 19185 |
kshitij.so |
171 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'dealType':data['dealType'],'codAvailable':data['codAvailable'], 'netPriceAfterCashBack':netPriceAfterCashBack}})
|
| 14147 |
kshitij.so |
172 |
|
|
|
173 |
try:
|
| 15267 |
kshitij.so |
174 |
recomputeDeal(data)
|
| 14147 |
kshitij.so |
175 |
except:
|
|
|
176 |
print "Unable to compute deal for ",data['skuBundleId']
|
| 13914 |
kshitij.so |
177 |
|
| 14325 |
kshitij.so |
178 |
def populateNegativeDeals():
|
|
|
179 |
negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
|
| 15267 |
kshitij.so |
180 |
mc.set("negative_deals", negativeDeals, 600)
|
| 13828 |
kshitij.so |
181 |
|
| 16501 |
kshitij.so |
182 |
#def recomputePoints(item, deal):
|
|
|
183 |
# try:
|
|
|
184 |
# if item.get('available_price') == deal['available_price']:
|
|
|
185 |
# print "No need to compute points for %d , as price is still same" %(item['_id'])
|
|
|
186 |
# raise
|
|
|
187 |
# nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
|
|
|
188 |
# except:
|
|
|
189 |
# traceback.print_exc()
|
|
|
190 |
# nlcPoints = deal['nlcPoints']
|
|
|
191 |
#
|
|
|
192 |
# 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())}}))
|
|
|
193 |
# if len(bundleDealPoints) > 0:
|
|
|
194 |
# item['manualDealThresholdPrice'] = bundleDealPoints[0]['dealThresholdPrice']
|
|
|
195 |
# dealPoints = bundleDealPoints[0]['dealPoints']
|
|
|
196 |
# else:
|
|
|
197 |
# dealPoints = 0
|
|
|
198 |
# item['manualDealThresholdPrice'] = None
|
|
|
199 |
#
|
|
|
200 |
# get_mongo_connection().Catalog.Deals.update({'_id':deal['_id']},{"$set":{'totalPoints':deal['totalPoints'] - deal['nlcPoints'] + nlcPoints - deal['dealPoints'] +dealPoints , 'nlcPoints': nlcPoints, 'dealPoints': dealPoints, 'manualDealThresholdPrice': item['manualDealThresholdPrice']}})
|
| 15267 |
kshitij.so |
201 |
|
|
|
202 |
|
|
|
203 |
|
|
|
204 |
def recomputeDeal(item):
|
| 13914 |
kshitij.so |
205 |
"""Lets recompute deal for this bundle"""
|
| 16019 |
kshitij.so |
206 |
print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
|
| 15267 |
kshitij.so |
207 |
skuBundleId = item['skuBundleId']
|
| 13914 |
kshitij.so |
208 |
|
| 19185 |
kshitij.so |
209 |
similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('netPriceAfterCashBack',pymongo.ASCENDING)]))
|
| 13914 |
kshitij.so |
210 |
bestPrice = float("inf")
|
|
|
211 |
bestOne = None
|
|
|
212 |
bestSellerPoints = 0
|
|
|
213 |
toUpdate = []
|
| 16019 |
kshitij.so |
214 |
prepaidBestPrice = float("inf")
|
|
|
215 |
prepaidBestOne = None
|
|
|
216 |
prepaidBestSellerPoints = 0
|
| 13914 |
kshitij.so |
217 |
for similarItem in similarItems:
|
| 16019 |
kshitij.so |
218 |
if similarItem['codAvailable'] ==1:
|
|
|
219 |
if mc.get("negative_deals") is None:
|
|
|
220 |
populateNegativeDeals()
|
| 16173 |
kshitij.so |
221 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 16019 |
kshitij.so |
222 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
|
|
223 |
continue
|
|
|
224 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
|
|
225 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
|
|
226 |
continue
|
| 19185 |
kshitij.so |
227 |
if similarItem.get('netPriceAfterCashBack') < bestPrice:
|
| 16019 |
kshitij.so |
228 |
bestOne = similarItem
|
| 19185 |
kshitij.so |
229 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
230 |
bestSellerPoints = similarItem['bestSellerPoints']
|
| 19185 |
kshitij.so |
231 |
elif similarItem.get('netPriceAfterCashBack') == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
|
| 16019 |
kshitij.so |
232 |
bestOne = similarItem
|
| 19185 |
kshitij.so |
233 |
bestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
234 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
235 |
else:
|
|
|
236 |
pass
|
| 13914 |
kshitij.so |
237 |
else:
|
| 16019 |
kshitij.so |
238 |
if mc.get("negative_deals") is None:
|
|
|
239 |
populateNegativeDeals()
|
| 16173 |
kshitij.so |
240 |
if similarItem['in_stock'] == 0 or similarItem['_id'] in mc.get("negative_deals"):
|
| 16019 |
kshitij.so |
241 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
|
|
|
242 |
continue
|
|
|
243 |
if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
|
|
|
244 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
|
|
|
245 |
continue
|
| 19185 |
kshitij.so |
246 |
if similarItem.get('netPriceAfterCashBack') < prepaidBestPrice:
|
| 16019 |
kshitij.so |
247 |
prepaidBestOne = similarItem
|
| 19185 |
kshitij.so |
248 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
249 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
| 19185 |
kshitij.so |
250 |
elif similarItem.get('netPriceAfterCashBack') == prepaidBestPrice and prepaidBestSellerPoints < similarItem['bestSellerPoints']:
|
| 16019 |
kshitij.so |
251 |
prepaidBestOne = similarItem
|
| 19185 |
kshitij.so |
252 |
prepaidBestPrice = similarItem.get('netPriceAfterCashBack')
|
| 16019 |
kshitij.so |
253 |
prepaidBestSellerPoints = similarItem['bestSellerPoints']
|
|
|
254 |
else:
|
|
|
255 |
pass
|
| 16026 |
kshitij.so |
256 |
if bestOne is not None or prepaidBestOne is not None:
|
| 13914 |
kshitij.so |
257 |
for similarItem in similarItems:
|
|
|
258 |
toUpdate.append(similarItem['_id'])
|
| 16026 |
kshitij.so |
259 |
if bestOne is not None:
|
|
|
260 |
toUpdate.remove(bestOne['_id'])
|
|
|
261 |
get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
|
|
|
262 |
if prepaidBestOne is not None:
|
| 16073 |
kshitij.so |
263 |
if bestOne is not None:
|
| 19185 |
kshitij.so |
264 |
if prepaidBestOne.get('netPriceAfterCashBack') < bestOne.get('netPriceAfterCashBack'):
|
| 16073 |
kshitij.so |
265 |
toUpdate.remove(prepaidBestOne['_id'])
|
|
|
266 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
|
|
267 |
else:
|
|
|
268 |
toUpdate.remove(prepaidBestOne['_id'])
|
|
|
269 |
get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
|
| 13914 |
kshitij.so |
270 |
if len(toUpdate) > 0:
|
| 16019 |
kshitij.so |
271 |
get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
|
| 13914 |
kshitij.so |
272 |
|
| 13828 |
kshitij.so |
273 |
def main():
|
| 14147 |
kshitij.so |
274 |
populate()
|
| 13828 |
kshitij.so |
275 |
|
|
|
276 |
if __name__=='__main__':
|
|
|
277 |
main()
|