Subversion Repositories SmartDukaan

Rev

Rev 14705 | Rev 15344 | 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
15269 kshitij.so 2
from dtr.utils.utils import to_java_date, getNlcPoints
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
14325 kshitij.so 9
from dtr.storage.MemCache import MemCache
14705 kshitij.so 10
from dtr.utils.utils import getCashBack
15269 kshitij.so 11
import traceback
13828 kshitij.so 12
 
13
con = None
14
 
14255 kshitij.so 15
parser = optparse.OptionParser()
16
parser.add_option("-m", "--m", dest="mongoHost",
17
                      default="localhost",
18
                      type="string", help="The HOST where the mongo server is running",
19
                      metavar="mongo_host")
20
 
21
(options, args) = parser.parse_args()
22
 
14325 kshitij.so 23
mc = MemCache(options.mongoHost)
24
 
14255 kshitij.so 25
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 26
    global con
27
    if con is None:
28
        print "Establishing connection %s host and port %d" %(host,port)
29
        try:
30
            con = pymongo.MongoClient(host, port)
31
        except Exception, e:
32
            print e
33
            return None
34
    return con
35
 
14149 kshitij.so 36
def populate():
37
    toScrapMap = {}
14131 kshitij.so 38
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
39
    for bestSeller in bestSellers: 
14149 kshitij.so 40
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':2}))
41
        for data in amazonBestSellers:
42
            if not toScrapMap.has_key(data['_id']):
15269 kshitij.so 43
                data['dealFlag'] = 0
44
                data['dealType'] = 0
45
                data['dealPoints'] = 0
46
                data['manualDealThresholdPrice'] = None
14149 kshitij.so 47
                toScrapMap[data['_id']] = data
14251 kshitij.so 48
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':2,'showDeal':1,'totalPoints':{'$gt':0}}))
49
    for deal in dealFlagged:
50
        if not toScrapMap.has_key(deal['_id']):
14262 kshitij.so 51
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
15269 kshitij.so 52
            data[0]['dealFlag'] = 0
53
            data[0]['dealType'] = 0
54
            data[0]['dealPoints'] = 0
55
            data[0]['manualDealThresholdPrice'] = None
14262 kshitij.so 56
            toScrapMap[deal['_id']] = data[0]
15269 kshitij.so 57
    manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':2}))
58
    for manualDeal in manualDeals:
59
        if not toScrapMap.has_key(manualDeal['sku']):
60
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
61
            if len(data) > 0:
62
                data[0]['dealFlag'] = 1
63
                data[0]['dealType'] = manualDeal['dealType']
64
                data[0]['dealPoints'] = manualDeal['dealPoints']
65
                data[0]['manualDealThresholdPrice'] = manualDeal['dealThresholdPrice']
66
                toScrapMap[manualDeal['sku']] = data[0]
67
        else:
68
            data = toScrapMap.get(manualDeal['sku'])
69
            data['dealFlag'] = 1
70
            data['dealType'] = manualDeal['dealType']
71
            data['dealPoints'] = manualDeal['dealPoints']
72
            data['manualDealThresholdPrice'] = manualDeal['dealThresholdPrice']
14178 kshitij.so 73
    pool = ThreadPool(cpu_count() *2)
14149 kshitij.so 74
    pool.map(scrapeFlipkart,toScrapMap.values())
75
    pool.close()
76
    pool.join()
14251 kshitij.so 77
    print "joining threads at %s"%(str(datetime.now()))
14149 kshitij.so 78
 
79
def scrapeFlipkart(data):
80
    if data['source_id']!=2:
14157 kshitij.so 81
        return
14149 kshitij.so 82
    retryCount = 0
83
    if data['identifier'] is None or len(data['identifier'].strip())==0:
14157 kshitij.so 84
        print "returning in valid identifier"
85
        return
14149 kshitij.so 86
 
87
    try:
88
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
15269 kshitij.so 89
            print "sku id is already updated %d" %(data['_id']) 
14157 kshitij.so 90
            return
14149 kshitij.so 91
    except:
92
        pass
93
 
94
 
95
    lowestSp = 0
96
    inStock = 0
15269 kshitij.so 97
    buyBoxPrice = 0
98
    isBuyBox = 0
14157 kshitij.so 99
    scraperFk = FlipkartScraper.FlipkartScraper()
100
    scraperProductPage = NewFlipkartScraper.FlipkartProductPageScraper()
14149 kshitij.so 101
    try:
102
        if data['marketPlaceUrl']!="" or data['marketPlaceUrl'] !="http://www.flipkart.com/ps/%s"%(data['identifier']):
103
            result = scraperProductPage.read(data['marketPlaceUrl'])
104
            if result.get('lowestSp')!=0:
105
                lowestSp = result.get('lowestSp')
106
                inStock = result.get('inStock')
15269 kshitij.so 107
                buyBoxPrice = result.get('buyBoxPrice')
14149 kshitij.so 108
    except:
15269 kshitij.so 109
        print "Unable to scrape product page %s" %(data['identifier'])
14149 kshitij.so 110
 
111
 
112
    if lowestSp == 0:
113
        url = "http://www.flipkart.com/ps/%s"%(data['identifier'].strip())
114
        while(retryCount < 3):
14131 kshitij.so 115
            try:
15269 kshitij.so 116
                vendorsData, buyBoxInfo = (scraperFk.read(url))
14149 kshitij.so 117
                fetched = True
118
                break
119
            except Exception as e:
120
                print "***Retry count ",retryCount 
121
                retryCount+=1
122
                if retryCount == 3:
123
                    fetched = False
124
                print e
125
        if not fetched:
15269 kshitij.so 126
            print "Unable to fetch data after multiple tries.Continue for %s"%(data['identifier'])
14157 kshitij.so 127
            return
14149 kshitij.so 128
 
129
        sortedVendorsData = []
130
        sortedVendorsData = sorted(vendorsData, key=itemgetter('sellingPrice'))
131
        print "data",sortedVendorsData
132
        lowestSp, iterator = (0,)*2
133
        for vData in sortedVendorsData:
134
            if iterator == 0:
135
                lowestSp = vData['sellingPrice']
136
            break
137
        if lowestSp > 0:
138
            inStock = 1
15269 kshitij.so 139
        if len(buyBoxInfo) > 0:
140
            buyBoxPrice = buyBoxInfo[0].get('sellingPrice')
141
        else:
142
            print "No info about buy box for %d"%(data.get('_id'))
14149 kshitij.so 143
    print lowestSp
144
    print inStock
15269 kshitij.so 145
    if buyBoxPrice is not None and buyBoxPrice == lowestSp:
146
        isBuyBox = 1
14149 kshitij.so 147
    if lowestSp > 0:
15269 kshitij.so 148
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock,'buyBoxFlag':isBuyBox}}, multi=True)
14149 kshitij.so 149
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock}}, multi=True)
150
    else:
15269 kshitij.so 151
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now()),'buyBoxFlag':isBuyBox}}, multi=True)
14149 kshitij.so 152
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)
153
 
154
    try:
15269 kshitij.so 155
        recomputeDeal(data)
14149 kshitij.so 156
    except:
15269 kshitij.so 157
        print "Unable to compute deal for %s"%(data['skuBundleId'])
13828 kshitij.so 158
 
15269 kshitij.so 159
def recomputePoints(item, deal):
160
    try:
161
        nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
162
    except:
163
        print traceback.print_exc()
164
        nlcPoints = deal['nlcPoints']
165
    if item['manualDealThresholdPrice'] >= deal['available_price']:
166
        dealPoints = item['dealPoints']
167
    else:
168
        dealPoints = 0
169
    get_mongo_connection().Catalog.Deals.update({'_id':deal['_id']},{"$set":{'totalPoints':deal['totalPoints'] - deal['nlcPoints'] + nlcPoints - deal['dealPoints'] +dealPoints , 'nlcPoints': nlcPoints, 'dealPoints': dealPoints, 'manualDealThresholdPrice': item['manualDealThresholdPrice']}})
170
 
171
 
172
 
14325 kshitij.so 173
def populateNegativeDeals():
174
    negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
175
    mc.set("negative_deals", negativeDeals, 600)  
176
 
15269 kshitij.so 177
def recomputeDeal(item):
13915 kshitij.so 178
    """Lets recompute deal for this bundle"""
15269 kshitij.so 179
    print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
180
    skuBundleId = item['skuBundleId']
13915 kshitij.so 181
 
182
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
183
    bestPrice = float("inf")
184
    bestOne = None
185
    bestSellerPoints = 0
186
    toUpdate = []
187
    for similarItem in similarItems:
14329 kshitij.so 188
        if mc.get("negative_deals") is None:
14325 kshitij.so 189
            populateNegativeDeals()
15269 kshitij.so 190
#        try:
191
#            cashBack = getCashBack(similarItem['_id'], similarItem['source_id'], similarItem['category_id'], mc, options.mongoHost)
192
#            if not cashBack or cashBack.get('cash_back_status')!=1:
193
#                pass
194
#            else:
195
#                if cashBack['cash_back_type'] ==1:
196
#                    similarItem['available_price'] = similarItem['available_price'] - similarItem['available_price'] * float(cashBack['cash_back'])/100
197
#                elif cashBack['cash_back_type'] ==2:
198
#                    similarItem['available_price'] = similarItem['available_price'] - float(cashBack['cash_back'])
199
#                else:
200
#                    pass
201
#        except Exception as cashBackEx:
202
#            print cashBackEx
203
#            print "Error calculating cashback."
204
        if similarItem['_id'] == item['_id']:
205
            recomputePoints(item, similarItem)
14329 kshitij.so 206
        if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price'] or similarItem['_id'] in mc.get("negative_deals"):
13915 kshitij.so 207
            get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})
208
            continue
209
        if similarItem['available_price'] < bestPrice:
210
            bestOne = similarItem
211
            bestPrice = similarItem['available_price']
212
            bestSellerPoints = similarItem['bestSellerPoints']
213
        elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
214
            bestOne = similarItem
215
            bestPrice = similarItem['available_price']
216
            bestSellerPoints = similarItem['bestSellerPoints']
217
        else:
218
            pass
219
    if bestOne is not None:
220
        for similarItem in similarItems:
221
            toUpdate.append(similarItem['_id'])
222
        toUpdate.remove(bestOne['_id'])
223
        get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})
224
    if len(toUpdate) > 0:
225
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)
226
 
13828 kshitij.so 227
def main():
14157 kshitij.so 228
    populate()
13828 kshitij.so 229
 
230
if __name__=='__main__':
231
    main()