| 13828 |
kshitij.so |
1 |
import pymongo
|
|
|
2 |
from dtr.utils.utils import to_java_date
|
|
|
3 |
from datetime import datetime
|
|
|
4 |
from operator import itemgetter
|
|
|
5 |
from shop2020.model.v1.catalog.script import FlipkartScraper
|
| 13832 |
kshitij.so |
6 |
from dtr.utils import AmazonPriceOnlyScraper
|
| 13828 |
kshitij.so |
7 |
|
|
|
8 |
con = None
|
|
|
9 |
now = datetime.now()
|
|
|
10 |
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
|
|
|
11 |
print to_java_date(now)
|
|
|
12 |
|
|
|
13 |
|
|
|
14 |
def get_mongo_connection(host='localhost', port=27017):
|
|
|
15 |
global con
|
|
|
16 |
if con is None:
|
|
|
17 |
print "Establishing connection %s host and port %d" %(host,port)
|
|
|
18 |
try:
|
|
|
19 |
con = pymongo.MongoClient(host, port)
|
|
|
20 |
except Exception, e:
|
|
|
21 |
print e
|
|
|
22 |
return None
|
|
|
23 |
return con
|
|
|
24 |
|
|
|
25 |
def scrapeAmazon():
|
|
|
26 |
amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':1}))
|
|
|
27 |
for data in amazonBestSellers:
|
|
|
28 |
inStock = 0
|
|
|
29 |
print str(data['identifier'])
|
|
|
30 |
if data['identifier'] is None or len(data['identifier'])==0:
|
|
|
31 |
print "continue"
|
|
|
32 |
continue
|
|
|
33 |
|
|
|
34 |
url = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'])
|
|
|
35 |
print url
|
|
|
36 |
lowestPrice = 0.0
|
|
|
37 |
lowestPrice = scraperAmazon.read(url)
|
|
|
38 |
print lowestPrice
|
|
|
39 |
if lowestPrice > 0:
|
|
|
40 |
inStock = 1
|
|
|
41 |
print lowestPrice
|
|
|
42 |
print inStock
|
|
|
43 |
if lowestPrice > 0:
|
|
|
44 |
get_mongo_connection().Catalog.MasterData.update({'identifier':data['identifier']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
45 |
else:
|
|
|
46 |
get_mongo_connection().Catalog.MasterData.update({'identifier':data['identifier']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
|
|
|
47 |
print "+++++++++++++++++++++++++++"
|
|
|
48 |
|
|
|
49 |
def main():
|
|
|
50 |
scrapeAmazon()
|
|
|
51 |
|
|
|
52 |
if __name__=='__main__':
|
|
|
53 |
main()
|