Subversion Repositories SmartDukaan

Rev

Rev 14251 | Rev 14262 | 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
14123 kshitij.so 5
from dtr.utils import FlipkartScraper,NewFlipkartScraper
14178 kshitij.so 6
from multiprocessing import Pool as ThreadPool
14172 kshitij.so 7
from multiprocessing import cpu_count
14255 kshitij.so 8
import optparse
13828 kshitij.so 9
 
10
con = None
11
 
14255 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
 
20
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 21
    global con
22
    if con is None:
23
        print "Establishing connection %s host and port %d" %(host,port)
24
        try:
25
            con = pymongo.MongoClient(host, port)
26
        except Exception, e:
27
            print e
28
            return None
29
    return con
30
 
14149 kshitij.so 31
def populate():
32
    toScrapMap = {}
14131 kshitij.so 33
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
34
    for bestSeller in bestSellers: 
14149 kshitij.so 35
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':2}))
36
        for data in amazonBestSellers:
37
            if not toScrapMap.has_key(data['_id']):
38
                toScrapMap[data['_id']] = data
14251 kshitij.so 39
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':2,'showDeal':1,'totalPoints':{'$gt':0}}))
40
    for deal in dealFlagged:
41
        if not toScrapMap.has_key(deal['_id']):
42
            toScrapMap[deal['_id']] = deal
14178 kshitij.so 43
    pool = ThreadPool(cpu_count() *2)
14149 kshitij.so 44
    pool.map(scrapeFlipkart,toScrapMap.values())
45
    pool.close()
46
    pool.join()
14251 kshitij.so 47
    print "joining threads at %s"%(str(datetime.now()))
14149 kshitij.so 48
 
49
def scrapeFlipkart(data):
50
    if data['source_id']!=2:
14157 kshitij.so 51
        return
14149 kshitij.so 52
    retryCount = 0
53
    print str(data['identifier'])
54
    if data['identifier'] is None or len(data['identifier'].strip())==0:
14157 kshitij.so 55
        print "returning in valid identifier"
56
        return
14149 kshitij.so 57
 
58
    try:
59
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
60
            print "sku id is already updated",data['_id'] 
14157 kshitij.so 61
            return
14149 kshitij.so 62
    except:
63
        pass
64
 
65
 
66
    lowestSp = 0
67
    inStock = 0
14157 kshitij.so 68
    scraperFk = FlipkartScraper.FlipkartScraper()
69
    scraperProductPage = NewFlipkartScraper.FlipkartProductPageScraper()
14149 kshitij.so 70
    try:
71
        if data['marketPlaceUrl']!="" or data['marketPlaceUrl'] !="http://www.flipkart.com/ps/%s"%(data['identifier']):
72
            result = scraperProductPage.read(data['marketPlaceUrl'])
73
            if result.get('lowestSp')!=0:
74
                lowestSp = result.get('lowestSp')
75
                inStock = result.get('inStock')
76
    except:
77
        print "Unable to scrape product page ",data['identifier']
78
 
79
 
80
    if lowestSp == 0:
81
        url = "http://www.flipkart.com/ps/%s"%(data['identifier'].strip())
82
        while(retryCount < 3):
14131 kshitij.so 83
            try:
14149 kshitij.so 84
                vendorsData = scraperFk.read(url)
85
                fetched = True
86
                break
87
            except Exception as e:
88
                print "***Retry count ",retryCount 
89
                retryCount+=1
90
                if retryCount == 3:
91
                    fetched = False
92
                print e
93
        if not fetched:
94
            print "Unable to fetch data after multiple tries.Continue for ",data['identifier']
14157 kshitij.so 95
            return
14149 kshitij.so 96
 
97
        sortedVendorsData = []
98
        sortedVendorsData = sorted(vendorsData, key=itemgetter('sellingPrice'))
99
        print "data",sortedVendorsData
100
        lowestSp, iterator = (0,)*2
101
        for vData in sortedVendorsData:
102
            if iterator == 0:
103
                lowestSp = vData['sellingPrice']
104
            break
105
        if lowestSp > 0:
106
            inStock = 1
107
    print lowestSp
108
    print inStock
109
    if lowestSp > 0:
110
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)
111
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock}}, multi=True)
112
    else:
113
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)
114
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)
115
 
116
    try:
117
        recomputeDeal(data['skuBundleId'])
118
    except:
119
        print "Unable to compute deal for ",data['skuBundleId']
13828 kshitij.so 120
 
13915 kshitij.so 121
def recomputeDeal(skuBundleId):
122
    """Lets recompute deal for this bundle"""
123
    print "Recomputing for bundleId",skuBundleId
124
 
125
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
126
    bestPrice = float("inf")
127
    bestOne = None
128
    bestSellerPoints = 0
129
    toUpdate = []
130
    for similarItem in similarItems:
13974 kshitij.so 131
        if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price']:
13915 kshitij.so 132
            get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})
133
            continue
134
        if similarItem['available_price'] < bestPrice:
135
            bestOne = similarItem
136
            bestPrice = similarItem['available_price']
137
            bestSellerPoints = similarItem['bestSellerPoints']
138
        elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
139
            bestOne = similarItem
140
            bestPrice = similarItem['available_price']
141
            bestSellerPoints = similarItem['bestSellerPoints']
142
        else:
143
            pass
144
    if bestOne is not None:
145
        for similarItem in similarItems:
146
            toUpdate.append(similarItem['_id'])
147
        toUpdate.remove(bestOne['_id'])
148
        get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})
149
    if len(toUpdate) > 0:
150
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)
151
 
13828 kshitij.so 152
def main():
14157 kshitij.so 153
    populate()
13828 kshitij.so 154
 
155
if __name__=='__main__':
156
    main()