Subversion Repositories SmartDukaan

Rev

Rev 16026 | Rev 16175 | 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
 
15653 kshitij.so 25
ignoreItems = []
16019 kshitij.so 26
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5}
15610 kshitij.so 27
 
14255 kshitij.so 28
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 29
    global con
30
    if con is None:
31
        print "Establishing connection %s host and port %d" %(host,port)
32
        try:
33
            con = pymongo.MongoClient(host, port)
34
        except Exception, e:
35
            print e
36
            return None
37
    return con
38
 
14149 kshitij.so 39
def populate():
40
    toScrapMap = {}
14131 kshitij.so 41
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
42
    for bestSeller in bestSellers: 
14149 kshitij.so 43
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':2}))
44
        for data in amazonBestSellers:
45
            if not toScrapMap.has_key(data['_id']):
15269 kshitij.so 46
                data['dealFlag'] = 0
47
                data['dealType'] = 0
48
                data['dealPoints'] = 0
49
                data['manualDealThresholdPrice'] = None
14149 kshitij.so 50
                toScrapMap[data['_id']] = data
14251 kshitij.so 51
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':2,'showDeal':1,'totalPoints':{'$gt':0}}))
52
    for deal in dealFlagged:
53
        if not toScrapMap.has_key(deal['_id']):
14262 kshitij.so 54
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
15269 kshitij.so 55
            data[0]['dealFlag'] = 0
56
            data[0]['dealType'] = 0
57
            data[0]['dealPoints'] = 0
58
            data[0]['manualDealThresholdPrice'] = None
14262 kshitij.so 59
            toScrapMap[deal['_id']] = data[0]
15269 kshitij.so 60
    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}))
61
    for manualDeal in manualDeals:
62
        if not toScrapMap.has_key(manualDeal['sku']):
63
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
64
            if len(data) > 0:
65
                data[0]['dealFlag'] = 1
66
                data[0]['dealType'] = manualDeal['dealType']
67
                data[0]['dealPoints'] = manualDeal['dealPoints']
68
                data[0]['manualDealThresholdPrice'] = manualDeal['dealThresholdPrice']
69
                toScrapMap[manualDeal['sku']] = data[0]
70
        else:
71
            data = toScrapMap.get(manualDeal['sku'])
72
            data['dealFlag'] = 1
73
            data['dealType'] = manualDeal['dealType']
74
            data['dealPoints'] = manualDeal['dealPoints']
75
            data['manualDealThresholdPrice'] = manualDeal['dealThresholdPrice']
14178 kshitij.so 76
    pool = ThreadPool(cpu_count() *2)
14149 kshitij.so 77
    pool.map(scrapeFlipkart,toScrapMap.values())
78
    pool.close()
79
    pool.join()
14251 kshitij.so 80
    print "joining threads at %s"%(str(datetime.now()))
14149 kshitij.so 81
 
82
def scrapeFlipkart(data):
83
    if data['source_id']!=2:
14157 kshitij.so 84
        return
14149 kshitij.so 85
    retryCount = 0
86
    if data['identifier'] is None or len(data['identifier'].strip())==0:
14157 kshitij.so 87
        print "returning in valid identifier"
88
        return
14149 kshitij.so 89
 
15610 kshitij.so 90
    if data['_id'] in ignoreItems:
91
        print "Ignored items returning for %d"%(data['_id'])
92
        return 
93
 
14149 kshitij.so 94
    try:
95
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
15269 kshitij.so 96
            print "sku id is already updated %d" %(data['_id']) 
14157 kshitij.so 97
            return
14149 kshitij.so 98
    except:
99
        pass
100
 
101
 
102
    lowestSp = 0
103
    inStock = 0
15269 kshitij.so 104
    buyBoxPrice = 0
105
    isBuyBox = 0
14157 kshitij.so 106
    scraperFk = FlipkartScraper.FlipkartScraper()
107
    scraperProductPage = NewFlipkartScraper.FlipkartProductPageScraper()
14149 kshitij.so 108
    try:
109
        if data['marketPlaceUrl']!="" or data['marketPlaceUrl'] !="http://www.flipkart.com/ps/%s"%(data['identifier']):
110
            result = scraperProductPage.read(data['marketPlaceUrl'])
111
            if result.get('lowestSp')!=0:
112
                lowestSp = result.get('lowestSp')
113
                inStock = result.get('inStock')
15269 kshitij.so 114
                buyBoxPrice = result.get('buyBoxPrice')
14149 kshitij.so 115
    except:
15269 kshitij.so 116
        print "Unable to scrape product page %s" %(data['identifier'])
14149 kshitij.so 117
 
118
 
119
    if lowestSp == 0:
120
        url = "http://www.flipkart.com/ps/%s"%(data['identifier'].strip())
121
        while(retryCount < 3):
14131 kshitij.so 122
            try:
15269 kshitij.so 123
                vendorsData, buyBoxInfo = (scraperFk.read(url))
14149 kshitij.so 124
                fetched = True
125
                break
126
            except Exception as e:
127
                print "***Retry count ",retryCount 
128
                retryCount+=1
129
                if retryCount == 3:
130
                    fetched = False
131
                print e
132
        if not fetched:
15269 kshitij.so 133
            print "Unable to fetch data after multiple tries.Continue for %s"%(data['identifier'])
14157 kshitij.so 134
            return
14149 kshitij.so 135
 
136
        sortedVendorsData = []
137
        sortedVendorsData = sorted(vendorsData, key=itemgetter('sellingPrice'))
138
        print "data",sortedVendorsData
139
        lowestSp, iterator = (0,)*2
140
        for vData in sortedVendorsData:
141
            if iterator == 0:
142
                lowestSp = vData['sellingPrice']
143
            break
144
        if lowestSp > 0:
145
            inStock = 1
15269 kshitij.so 146
        if len(buyBoxInfo) > 0:
147
            buyBoxPrice = buyBoxInfo[0].get('sellingPrice')
148
        else:
149
            print "No info about buy box for %d"%(data.get('_id'))
14149 kshitij.so 150
    print lowestSp
151
    print inStock
15269 kshitij.so 152
    if buyBoxPrice is not None and buyBoxPrice == lowestSp:
153
        isBuyBox = 1
14149 kshitij.so 154
    if lowestSp > 0:
15269 kshitij.so 155
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock,'buyBoxFlag':isBuyBox}}, multi=True)
16019 kshitij.so 156
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestSp , 'in_stock':inStock,'codAvailable':data['codAvailable']}}, multi=True)
14149 kshitij.so 157
    else:
15269 kshitij.so 158
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now()),'buyBoxFlag':isBuyBox}}, multi=True)
16019 kshitij.so 159
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'codAvailable':data['codAvailable']}})
14149 kshitij.so 160
 
161
    try:
15269 kshitij.so 162
        recomputeDeal(data)
14149 kshitij.so 163
    except:
15269 kshitij.so 164
        print "Unable to compute deal for %s"%(data['skuBundleId'])
13828 kshitij.so 165
 
15269 kshitij.so 166
def recomputePoints(item, deal):
167
    try:
15344 kshitij.so 168
        if item.get('available_price') == deal['available_price']:
169
            print "No need to compute points for %d , as price is still same" %(item['_id'])
170
            raise
15269 kshitij.so 171
        nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
172
    except:
173
        print traceback.print_exc()
174
        nlcPoints = deal['nlcPoints']
175
    if item['manualDealThresholdPrice'] >= deal['available_price']:
176
        dealPoints = item['dealPoints']
177
    else:
178
        dealPoints = 0
179
    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']}})
180
 
14325 kshitij.so 181
def populateNegativeDeals():
182
    negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
183
    mc.set("negative_deals", negativeDeals, 600)  
184
 
15269 kshitij.so 185
def recomputeDeal(item):
13915 kshitij.so 186
    """Lets recompute deal for this bundle"""
15269 kshitij.so 187
    print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
188
    skuBundleId = item['skuBundleId']
13915 kshitij.so 189
 
190
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
191
    bestPrice = float("inf")
192
    bestOne = None
193
    bestSellerPoints = 0
194
    toUpdate = []
16019 kshitij.so 195
    prepaidBestPrice = float("inf")
196
    prepaidBestOne = None
197
    prepaidBestSellerPoints = 0
13915 kshitij.so 198
    for similarItem in similarItems:
15269 kshitij.so 199
        if similarItem['_id'] == item['_id']:
16019 kshitij.so 200
            try:
201
                recomputePoints(item, similarItem)
202
            except:
203
                traceback.print_exc()
204
        if similarItem['codAvailable'] ==1:
205
            if mc.get("negative_deals") is None:
206
                populateNegativeDeals()
207
            if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price'] or similarItem['_id'] in mc.get("negative_deals"):
208
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
209
                continue
210
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
211
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
212
                continue
213
            if similarItem['available_price'] < bestPrice:
214
                bestOne = similarItem
215
                bestPrice = similarItem['available_price']
216
                bestSellerPoints = similarItem['bestSellerPoints']
217
            elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
218
                bestOne = similarItem
219
                bestPrice = similarItem['available_price']
220
                bestSellerPoints = similarItem['bestSellerPoints']
221
            else:
222
                pass
13915 kshitij.so 223
        else:
16019 kshitij.so 224
            if mc.get("negative_deals") is None:
225
                populateNegativeDeals()
226
            if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price'] or similarItem['_id'] in mc.get("negative_deals"):
227
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
228
                continue
229
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
230
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
231
                continue
232
            if similarItem['available_price'] < prepaidBestPrice:
233
                prepaidBestOne = similarItem
234
                prepaidBestPrice = similarItem['available_price']
235
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
236
            elif similarItem['available_price'] == prepaidBestPrice and prepaidBestSellerPoints < similarItem['bestSellerPoints']:
237
                prepaidBestOne = similarItem
238
                prepaidBestPrice = similarItem['available_price']
239
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
240
            else:
241
                pass
16026 kshitij.so 242
    if bestOne is not None or prepaidBestOne is not None:
13915 kshitij.so 243
        for similarItem in similarItems:
244
            toUpdate.append(similarItem['_id'])
16026 kshitij.so 245
        if bestOne is not None:
246
            toUpdate.remove(bestOne['_id'])
247
            get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
248
        if prepaidBestOne is not None:
16074 kshitij.so 249
            if bestOne is not None:
250
                if prepaidBestOne['available_price'] < bestOne['available_price']: 
251
                    toUpdate.remove(prepaidBestOne['_id'])
252
                    get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
253
            else:
254
                toUpdate.remove(prepaidBestOne['_id'])
255
                get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
13915 kshitij.so 256
    if len(toUpdate) > 0:
16019 kshitij.so 257
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
13915 kshitij.so 258
 
13828 kshitij.so 259
def main():
14157 kshitij.so 260
    populate()
13828 kshitij.so 261
 
262
if __name__=='__main__':
263
    main()