| 13829 |
kshitij.so |
1 |
import urllib2
|
|
|
2 |
import simplejson as json
|
|
|
3 |
import pymongo
|
|
|
4 |
from dtr.utils.utils import to_java_date
|
| 13918 |
kshitij.so |
5 |
from datetime import datetime, timedelta
|
| 13829 |
kshitij.so |
6 |
from operator import itemgetter
|
|
|
7 |
from dtr.utils.AmazonPriceOnlyScraper import AmazonScraper
|
| 13830 |
kshitij.so |
8 |
from dtr.utils import FlipkartScraper
|
| 13829 |
kshitij.so |
9 |
|
|
|
10 |
con = None
|
|
|
11 |
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3}
|
|
|
12 |
scraperFk = FlipkartScraper.FlipkartScraper()
|
|
|
13 |
scraperAmazon = AmazonScraper()
|
|
|
14 |
|
|
|
15 |
headers = {
|
|
|
16 |
'User-agent':'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11',
|
|
|
17 |
'Accept' : 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
|
|
|
18 |
'Accept-Language' : 'en-US,en;q=0.8',
|
|
|
19 |
'Accept-Charset' : 'ISO-8859-1,utf-8;q=0.7,*;q=0.3'
|
|
|
20 |
}
|
|
|
21 |
|
|
|
22 |
def get_mongo_connection(host='localhost', port=27017):
|
|
|
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 |
|
|
|
33 |
def returnLatestPrice(data, source_id):
|
| 13918 |
kshitij.so |
34 |
now = datetime.now()
|
| 13829 |
kshitij.so |
35 |
if source_id == 1:
|
|
|
36 |
try:
|
| 13918 |
kshitij.so |
37 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
| 13829 |
kshitij.so |
38 |
return {}
|
| 13918 |
kshitij.so |
39 |
|
|
|
40 |
try:
|
| 13919 |
kshitij.so |
41 |
if data['updatedOn'] > to_java_date(now - timedelta(minutes=5)):
|
| 13918 |
kshitij.so |
42 |
print "sku id is already updated",data['_id']
|
|
|
43 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':1,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
44 |
except:
|
|
|
45 |
pass
|
|
|
46 |
|
|
|
47 |
|
| 13829 |
kshitij.so |
48 |
url = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'])
|
|
|
49 |
lowestPrice = 0.0
|
|
|
50 |
lowestPrice = scraperAmazon.read(url)
|
|
|
51 |
inStock = 0
|
|
|
52 |
if lowestPrice > 0:
|
|
|
53 |
inStock = 1
|
|
|
54 |
if lowestPrice > 0:
|
| 13918 |
kshitij.so |
55 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
56 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock}}, multi=True)
|
| 13829 |
kshitij.so |
57 |
else:
|
| 13918 |
kshitij.so |
58 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':0}}, multi=True)
|
|
|
59 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':0}}, multi=True)
|
|
|
60 |
|
|
|
61 |
try:
|
|
|
62 |
recomputeDeal(data['skuBundleId'])
|
|
|
63 |
except:
|
|
|
64 |
print "Unable to compute deal for ",data['skuBundleId']
|
|
|
65 |
|
| 13829 |
kshitij.so |
66 |
return {'_id':data['_id'],'available_price':lowestPrice,'in_stock':inStock,'source_id':1,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
67 |
except:
|
|
|
68 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':1,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
69 |
|
|
|
70 |
elif source_id ==3:
|
|
|
71 |
try:
|
| 13918 |
kshitij.so |
72 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
| 13829 |
kshitij.so |
73 |
return {}
|
| 13918 |
kshitij.so |
74 |
|
|
|
75 |
try:
|
| 13919 |
kshitij.so |
76 |
if data['updatedOn'] > to_java_date(now - timedelta(minutes=5)):
|
| 13918 |
kshitij.so |
77 |
print "sku id is already updated",data['_id']
|
|
|
78 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':3,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
| 13919 |
kshitij.so |
79 |
|
|
|
80 |
except Exception as e:
|
|
|
81 |
print e
|
| 13918 |
kshitij.so |
82 |
pass
|
|
|
83 |
|
|
|
84 |
|
| 13829 |
kshitij.so |
85 |
url="http://www.snapdeal.com/acors/json/gvbps?supc=%s&catId=175&sort=sellingPrice"%(data['identifier'])
|
|
|
86 |
req = urllib2.Request(url,headers=headers)
|
|
|
87 |
response = urllib2.urlopen(req)
|
|
|
88 |
json_input = response.read()
|
|
|
89 |
vendorInfo = json.loads(json_input)
|
|
|
90 |
lowestOfferPrice = 0
|
|
|
91 |
inStock = 0
|
|
|
92 |
for vendor in vendorInfo:
|
|
|
93 |
lowestOfferPrice = float(vendor['sellingPrice'])
|
|
|
94 |
stock = vendor['buyableInventory']
|
|
|
95 |
if stock > 0 and lowestOfferPrice > 0:
|
|
|
96 |
inStock = 1
|
|
|
97 |
break
|
|
|
98 |
|
|
|
99 |
print lowestOfferPrice
|
|
|
100 |
print inStock
|
|
|
101 |
print "*************"
|
|
|
102 |
if inStock == 1:
|
| 13918 |
kshitij.so |
103 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice,'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
104 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice , 'in_stock':inStock}}, multi=True)
|
| 13829 |
kshitij.so |
105 |
else:
|
| 13918 |
kshitij.so |
106 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
107 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)
|
|
|
108 |
|
|
|
109 |
try:
|
|
|
110 |
recomputeDeal(data['skuBundleId'])
|
|
|
111 |
except:
|
|
|
112 |
print "Unable to compute deal for ",data['skuBundleId']
|
|
|
113 |
|
| 13829 |
kshitij.so |
114 |
return {'_id':data['_id'],'available_price':lowestOfferPrice,'in_stock':inStock,'source_id':3,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
115 |
except:
|
|
|
116 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':3,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
117 |
|
|
|
118 |
elif source_id == 2:
|
|
|
119 |
try:
|
| 13918 |
kshitij.so |
120 |
if data['identifier'] is None or len(data['identifier'].strip())==0:
|
| 13829 |
kshitij.so |
121 |
return {}
|
| 13918 |
kshitij.so |
122 |
|
|
|
123 |
try:
|
| 13919 |
kshitij.so |
124 |
if data['updatedOn'] > to_java_date(now - timedelta(minutes=5)):
|
| 13918 |
kshitij.so |
125 |
print "sku id is already updated",data['_id']
|
|
|
126 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':2,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
127 |
except:
|
|
|
128 |
pass
|
|
|
129 |
|
|
|
130 |
|
| 13829 |
kshitij.so |
131 |
url = "http://www.flipkart.com/ps/%s"%(data['identifier'])
|
|
|
132 |
vendorsData = scraperFk.read(url)
|
|
|
133 |
sortedVendorsData = []
|
|
|
134 |
sortedVendorsData = sorted(vendorsData, key=itemgetter('sellingPrice'))
|
|
|
135 |
print "data",sortedVendorsData
|
| 13918 |
kshitij.so |
136 |
lowestSp = 0
|
|
|
137 |
inStock = 0
|
| 13829 |
kshitij.so |
138 |
for vData in sortedVendorsData:
|
| 13918 |
kshitij.so |
139 |
lowestSp = vData['sellingPrice']
|
| 13829 |
kshitij.so |
140 |
break
|
|
|
141 |
if lowestSp > 0:
|
|
|
142 |
inStock = 1
|
|
|
143 |
print lowestSp
|
|
|
144 |
print inStock
|
|
|
145 |
if lowestSp > 0:
|
| 13918 |
kshitij.so |
146 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
147 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock}}, multi=True)
|
| 13829 |
kshitij.so |
148 |
else:
|
| 13918 |
kshitij.so |
149 |
get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
150 |
get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)
|
|
|
151 |
|
|
|
152 |
|
|
|
153 |
try:
|
|
|
154 |
recomputeDeal(data['skuBundleId'])
|
|
|
155 |
except:
|
|
|
156 |
print "Unable to compute deal for ",data['skuBundleId']
|
|
|
157 |
|
| 13829 |
kshitij.so |
158 |
return {'_id':data['_id'],'available_price':lowestSp,'in_stock':inStock,'source_id':2,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
159 |
except:
|
|
|
160 |
return {'_id':data['_id'],'available_price':data['available_price'],'in_stock':data['inStock'],'source_id':2,'source_product_name':data['source_product_name'],'marketPlaceUrl':data['marketPlaceUrl'],'thumbnail':data['thumbnail']}
|
|
|
161 |
else:
|
|
|
162 |
return {}
|
|
|
163 |
|
| 13918 |
kshitij.so |
164 |
|
|
|
165 |
def recomputeDeal(skuBundleId):
|
|
|
166 |
"""Lets recompute deal for this bundle"""
|
|
|
167 |
print "Recomputing for bundleId",skuBundleId
|
| 13829 |
kshitij.so |
168 |
|
| 13918 |
kshitij.so |
169 |
similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
|
|
|
170 |
bestPrice = float("inf")
|
|
|
171 |
bestOne = None
|
|
|
172 |
bestSellerPoints = 0
|
|
|
173 |
toUpdate = []
|
|
|
174 |
for similarItem in similarItems:
|
|
|
175 |
if similarItem['in_stock'] == 0:
|
|
|
176 |
get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})
|
|
|
177 |
continue
|
|
|
178 |
if similarItem['available_price'] < bestPrice:
|
|
|
179 |
bestOne = similarItem
|
|
|
180 |
bestPrice = similarItem['available_price']
|
|
|
181 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
182 |
elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
|
|
|
183 |
bestOne = similarItem
|
|
|
184 |
bestPrice = similarItem['available_price']
|
|
|
185 |
bestSellerPoints = similarItem['bestSellerPoints']
|
|
|
186 |
else:
|
|
|
187 |
pass
|
|
|
188 |
if bestOne is not None:
|
|
|
189 |
for similarItem in similarItems:
|
|
|
190 |
toUpdate.append(similarItem['_id'])
|
|
|
191 |
toUpdate.remove(bestOne['_id'])
|
|
|
192 |
get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})
|
|
|
193 |
if len(toUpdate) > 0:
|
|
|
194 |
get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)
|
| 13829 |
kshitij.so |
195 |
|
|
|
196 |
def getLatestPrice(skuBundleId, source_id):
|
|
|
197 |
temp = []
|
|
|
198 |
itemIds = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':skuBundleId,'source_id' : source_id}))
|
|
|
199 |
for item in itemIds:
|
|
|
200 |
temp.append(returnLatestPrice(item, source_id))
|
|
|
201 |
return temp
|
|
|
202 |
|
| 13864 |
kshitij.so |
203 |
def getLatestPriceById(id):
|
|
|
204 |
item = list(get_mongo_connection().Catalog.MasterData.find({'_id':id}))
|
| 13866 |
kshitij.so |
205 |
return returnLatestPrice(item[0], item[0]['source_id'])
|
| 13829 |
kshitij.so |
206 |
|
|
|
207 |
|
|
|
208 |
def main():
|
| 13866 |
kshitij.so |
209 |
print getLatestPriceById(7)
|
| 13829 |
kshitij.so |
210 |
|
|
|
211 |
if __name__=='__main__':
|
|
|
212 |
main()
|