Subversion Repositories SmartDukaan

Rev

Rev 13974 | Rev 14123 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
13828 kshitij.so 1
import pymongo
2
from dtr.utils.utils import to_java_date
13915 kshitij.so 3
from datetime import datetime, timedelta
13828 kshitij.so 4
from operator import itemgetter
13833 kshitij.so 5
from dtr.utils import FlipkartScraper
13828 kshitij.so 6
 
7
con = None
8
now = datetime.now()
9
print to_java_date(now)
10
scraperFk = FlipkartScraper.FlipkartScraper()
11
 
12
def get_mongo_connection(host='localhost', port=27017):
13
    global con
14
    if con is None:
15
        print "Establishing connection %s host and port %d" %(host,port)
16
        try:
17
            con = pymongo.MongoClient(host, port)
18
        except Exception, e:
19
            print e
20
            return None
21
    return con
22
 
23
def scrapeFlipkart():
24
    flipkartBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':2}))
25
    for data in flipkartBestSellers:
26
        inStock = 0
13841 kshitij.so 27
        retryCount = 0
13828 kshitij.so 28
        print str(data['identifier'])
13845 kshitij.so 29
        if data['identifier'] is None or len(data['identifier'].strip())==0:
13828 kshitij.so 30
            print "continue"
31
            continue
32
 
13915 kshitij.so 33
        try:
13979 kshitij.so 34
            if data['priceUpdatedOn'] > to_java_date(now - timedelta(minutes=5)):
13915 kshitij.so 35
                print "sku id is already updated",data['_id'] 
36
                continue
37
        except:
38
            pass
39
 
13876 kshitij.so 40
        url = "http://www.flipkart.com/ps/%s"%(data['identifier'].strip())
13841 kshitij.so 41
        while(retryCount < 3):
42
            try:
43
                vendorsData = scraperFk.read(url)
44
                fetched = True
13848 kshitij.so 45
                break
13841 kshitij.so 46
            except Exception as e:
13842 kshitij.so 47
                print "***Retry count ",retryCount 
13841 kshitij.so 48
                retryCount+=1
49
                if retryCount == 3:
50
                    fetched = False
51
                print e
52
        if not fetched:
53
            print "Unable to fetch data after multiple tries.Continue for ",data['identifier']
54
            continue
13915 kshitij.so 55
 
13828 kshitij.so 56
        sortedVendorsData = []
57
        sortedVendorsData = sorted(vendorsData, key=itemgetter('sellingPrice'))
58
        print "data",sortedVendorsData
59
        lowestSp, iterator = (0,)*2
60
        for vData in sortedVendorsData:
61
            if iterator == 0:
62
                lowestSp = vData['sellingPrice']
63
            break
64
        if lowestSp > 0:
65
            inStock = 1
66
        print lowestSp
67
        print inStock
68
        if lowestSp > 0:
13974 kshitij.so 69
            get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(now),'priceUpdatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)
13915 kshitij.so 70
            get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock}}, multi=True)
13828 kshitij.so 71
        else:
13979 kshitij.so 72
            get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':inStock,'priceUpdatedOn':to_java_date(now)}}, multi=True)
13915 kshitij.so 73
            get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)
13828 kshitij.so 74
 
13915 kshitij.so 75
        try:
76
            recomputeDeal(data['skuBundleId'])
77
        except:
78
            print "Unable to compute deal for ",data['skuBundleId']
13828 kshitij.so 79
 
13915 kshitij.so 80
def recomputeDeal(skuBundleId):
81
    """Lets recompute deal for this bundle"""
82
    print "Recomputing for bundleId",skuBundleId
83
 
84
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
85
    bestPrice = float("inf")
86
    bestOne = None
87
    bestSellerPoints = 0
88
    toUpdate = []
89
    for similarItem in similarItems:
13974 kshitij.so 90
        if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price']:
13915 kshitij.so 91
            get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})
92
            continue
93
        if similarItem['available_price'] < bestPrice:
94
            bestOne = similarItem
95
            bestPrice = similarItem['available_price']
96
            bestSellerPoints = similarItem['bestSellerPoints']
97
        elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
98
            bestOne = similarItem
99
            bestPrice = similarItem['available_price']
100
            bestSellerPoints = similarItem['bestSellerPoints']
101
        else:
102
            pass
103
    if bestOne is not None:
104
        for similarItem in similarItems:
105
            toUpdate.append(similarItem['_id'])
106
        toUpdate.remove(bestOne['_id'])
107
        get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})
108
    if len(toUpdate) > 0:
109
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)
110
 
13828 kshitij.so 111
def main():
112
    scrapeFlipkart()
113
 
114
if __name__=='__main__':
115
    main()