| 13828 |
kshitij.so |
1 |
import pymongo
|
|
|
2 |
from dtr.utils.utils import to_java_date
|
| 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
|
| 13828 |
kshitij.so |
9 |
|
|
|
10 |
con = None
|
|
|
11 |
|
| 14256 |
kshitij.so |
12 |
parser = optparse.OptionParser()
|
|
|
13 |
parser.add_option("-m", "--m", dest="mongoHost",
|
|
|
14 |
default="localhost",
|
|
|
15 |
type="string", help="The HOST where the mongo server is running",
|
|
|
16 |
metavar="mongo_host")
|
|
|
17 |
|
|
|
18 |
(options, args) = parser.parse_args()
|
|
|
19 |
|
| 14325 |
kshitij.so |
20 |
mc = MemCache(options.mongoHost)
|
|
|
21 |
|
| 14256 |
kshitij.so |
22 |
def get_mongo_connection(host=options.mongoHost, port=27017):
|
| 13828 |
kshitij.so |
23 |
global con
|
|
|
24 |
if con is None:
|
|
|
25 |
print "Establishing connection %s host and port %d" %(host,port)
|
|
|
26 |
try:
|
|
|
27 |
con = pymongo.MongoClient(host, port)
|
|
|
28 |
except Exception, e:
|
|
|
29 |
print e
|
|
|
30 |
return None
|
|
|
31 |
return con
|
|
|
32 |
|
| 14147 |
kshitij.so |
33 |
def populate():
|
|
|
34 |
toScrapMap = {}
|
| 14133 |
kshitij.so |
35 |
bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
|
|
|
36 |
for bestSeller in bestSellers:
|
|
|
37 |
amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
|
|
|
38 |
for data in amazonBestSellers:
|
| 14147 |
kshitij.so |
39 |
if not toScrapMap.has_key(data['_id']):
|
|
|
40 |
toScrapMap[data['_id']] = data
|
| 14250 |
kshitij.so |
41 |
dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':0}}))
|
|
|
42 |
for deal in dealFlagged:
|
|
|
43 |
if not toScrapMap.has_key(deal['_id']):
|
| 14260 |
kshitij.so |
44 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
|
|
|
45 |
toScrapMap[deal['_id']] = data[0]
|
| 14308 |
kshitij.so |
46 |
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}))
|
|
|
47 |
for manualDeal in manualDeals:
|
|
|
48 |
if not toScrapMap.has_key(manualDeal['sku']):
|
|
|
49 |
data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
|
|
|
50 |
if len(data) > 0:
|
|
|
51 |
data[0]['dealFlag'] = 1
|
|
|
52 |
data[0]['dealType'] = manualDeal['dealType']
|
|
|
53 |
toScrapMap[manualDeal['sku']] = data[0]
|
|
|
54 |
else:
|
|
|
55 |
data = toScrapMap.get(manualDeal['sku'])
|
|
|
56 |
data['dealFlag'] = 1
|
|
|
57 |
data['dealType'] = manualDeal['dealType']
|
| 14170 |
kshitij.so |
58 |
pool = ThreadPool(cpu_count() * 2)
|
| 14147 |
kshitij.so |
59 |
pool.map(scrapeAmazon,toScrapMap.values())
|
|
|
60 |
pool.close()
|
|
|
61 |
pool.join()
|
| 14250 |
kshitij.so |
62 |
print "joining threads at %s"%(str(datetime.now()))
|
| 14147 |
kshitij.so |
63 |
|
|
|
64 |
|
|
|
65 |
def scrapeAmazon(data):
|
|
|
66 |
inStock = 0
|
| 14308 |
kshitij.so |
67 |
dealScraping = False
|
| 14147 |
kshitij.so |
68 |
print str(data['identifier'])
|
|
|
69 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
|
|
70 |
return
|
|
|
71 |
|
|
|
72 |
try:
|
|
|
73 |
if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
|
|
|
74 |
print "sku id is already updated",data['_id']
|
|
|
75 |
return
|
|
|
76 |
except:
|
|
|
77 |
pass
|
|
|
78 |
|
|
|
79 |
url = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'].strip())
|
| 14308 |
kshitij.so |
80 |
|
|
|
81 |
try:
|
|
|
82 |
if data['dealFlag'] ==1 and data['dealType'] ==1:
|
|
|
83 |
deal_url = "http://www.amazon.in/dp/%s"%(data['identifier'].strip())
|
|
|
84 |
dealScraping = True
|
|
|
85 |
except:
|
|
|
86 |
data['dealFlag'] = 0
|
|
|
87 |
data['dealType'] = 0
|
|
|
88 |
|
| 14147 |
kshitij.so |
89 |
print url
|
|
|
90 |
lowestPrice = 0.0
|
| 14308 |
kshitij.so |
91 |
|
|
|
92 |
if not dealScraping:
|
|
|
93 |
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
|
|
|
94 |
lowestPrice = scraperAmazon.read(url)
|
|
|
95 |
else:
|
|
|
96 |
dealScraperAmazon = AmazonDealScraper.AmazonScraper()
|
|
|
97 |
lowestPrice = dealScraperAmazon.read(deal_url)
|
|
|
98 |
if lowestPrice == 0:
|
|
|
99 |
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
|
|
|
100 |
lowestPrice = scraperAmazon.read(url)
|
|
|
101 |
dealScraping = False
|
|
|
102 |
|
| 14147 |
kshitij.so |
103 |
print lowestPrice
|
|
|
104 |
if lowestPrice > 0:
|
|
|
105 |
inStock = 1
|
|
|
106 |
print lowestPrice
|
|
|
107 |
print inStock
|
|
|
108 |
if lowestPrice > 0:
|
|
|
109 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)
|
| 14308 |
kshitij.so |
110 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock,'dealType':data['dealType']}}, multi=True)
|
| 14147 |
kshitij.so |
111 |
else:
|
|
|
112 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)
|
| 14308 |
kshitij.so |
113 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'dealType':data['dealType']}}, multi=True)
|
| 14147 |
kshitij.so |
114 |
|
|
|
115 |
try:
|
|
|
116 |
recomputeDeal(data['skuBundleId'])
|
|
|
117 |
except:
|
|
|
118 |
print "Unable to compute deal for ",data['skuBundleId']
|
| 13914 |
kshitij.so |
119 |
|
| 14325 |
kshitij.so |
120 |
def populateNegativeDeals():
|
|
|
121 |
negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
|
|
|
122 |
mc.set("negative_deals", negativeDeals, 600)
|
| 13828 |
kshitij.so |
123 |
|
| 13914 |
kshitij.so |
124 |
def recomputeDeal(skuBundleId):
|
|
|
125 |
"""Lets recompute deal for this bundle"""
|
|
|
126 |
print "Recomputing for bundleId",skuBundleId
|
|
|
127 |
|
|
|
128 |
similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
|
|
|
129 |
bestPrice = float("inf")
|
|
|
130 |
bestOne = None
|
|
|
131 |
bestSellerPoints = 0
|
|
|
132 |
toUpdate = []
|
|
|
133 |
for similarItem in similarItems:
|
| 14325 |
kshitij.so |
134 |
if mc.get("negativeDeals") is None:
|
|
|
135 |
populateNegativeDeals()
|
|
|
136 |
if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price'] or similarItem['_id'] in mc.get("negativeDeals"):
|
| 13914 |
kshitij.so |
137 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})
|
|
|
138 |
continue
|
|
|
139 |
if similarItem['available_price'] < bestPrice:
|
|
|
140 |
bestOne = similarItem
|
|
|
141 |
bestPrice = similarItem['available_price']
|
|
|
142 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
143 |
elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
|
|
|
144 |
bestOne = similarItem
|
|
|
145 |
bestPrice = similarItem['available_price']
|
|
|
146 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
147 |
else:
|
|
|
148 |
pass
|
|
|
149 |
if bestOne is not None:
|
|
|
150 |
for similarItem in similarItems:
|
|
|
151 |
toUpdate.append(similarItem['_id'])
|
|
|
152 |
toUpdate.remove(bestOne['_id'])
|
|
|
153 |
get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})
|
|
|
154 |
if len(toUpdate) > 0:
|
|
|
155 |
get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)
|
|
|
156 |
|
| 13828 |
kshitij.so |
157 |
def main():
|
| 14147 |
kshitij.so |
158 |
populate()
|
| 13828 |
kshitij.so |
159 |
|
|
|
160 |
if __name__=='__main__':
|
|
|
161 |
main()
|