Subversion Repositories SmartDukaan

Rev

Rev 16502 | Rev 17294 | 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 urllib2
2
import simplejson as json
3
import pymongo
15270 kshitij.so 4
from dtr.utils.utils import to_java_date, getNlcPoints 
13917 kshitij.so 5
from datetime import datetime, timedelta
13828 kshitij.so 6
import time
14180 kshitij.so 7
from multiprocessing import Pool as ThreadPool
8
from multiprocessing import cpu_count
14254 kshitij.so 9
import optparse
14325 kshitij.so 10
from dtr.storage.MemCache import MemCache
14705 kshitij.so 11
from dtr.utils.utils import getCashBack
15270 kshitij.so 12
import traceback
13
from operator import itemgetter
16085 kshitij.so 14
import chardet
13828 kshitij.so 15
 
16
con = None
17
 
14254 kshitij.so 18
parser = optparse.OptionParser()
19
parser.add_option("-m", "--m", dest="mongoHost",
20
                      default="localhost",
21
                      type="string", help="The HOST where the mongo server is running",
22
                      metavar="mongo_host")
23
 
24
(options, args) = parser.parse_args()
25
 
14325 kshitij.so 26
mc = MemCache(options.mongoHost)
14254 kshitij.so 27
 
16869 kshitij.so 28
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5,'PAYTM.COM':6}
16019 kshitij.so 29
 
13828 kshitij.so 30
headers = { 
31
           'User-agent':'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11',
32
            'Accept' : 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',      
33
            'Accept-Language' : 'en-US,en;q=0.8',                     
34
            'Accept-Charset' : 'ISO-8859-1,utf-8;q=0.7,*;q=0.3'
35
        }
36
 
16429 kshitij.so 37
 
14254 kshitij.so 38
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 39
    global con
40
    if con is None:
41
        print "Establishing connection %s host and port %d" %(host,port)
42
        try:
43
            con = pymongo.MongoClient(host, port)
44
        except Exception, e:
45
            print e
46
            return None
47
    return con
48
 
14180 kshitij.so 49
def populate():
50
    toScrapMap = {}
14132 kshitij.so 51
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
52
    for bestSeller in bestSellers: 
53
        snapdealBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':3}))
54
        for data in snapdealBestSellers:
14180 kshitij.so 55
            if not toScrapMap.has_key(data['_id']):
15270 kshitij.so 56
                data['dealFlag'] = 0
57
                data['dealType'] = 0
14180 kshitij.so 58
                toScrapMap[data['_id']] = data
16174 kshitij.so 59
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':3,'showDeal':1,'totalPoints':{'$gt':-100}}))
14252 kshitij.so 60
    for deal in dealFlagged:
61
        if not toScrapMap.has_key(deal['_id']):
14261 kshitij.so 62
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
15270 kshitij.so 63
            data[0]['dealFlag'] = 0
64
            data[0]['dealType'] = 0
14261 kshitij.so 65
            toScrapMap[deal['_id']] = data[0]
15270 kshitij.so 66
    manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':3}))
67
    for manualDeal in manualDeals:
68
        if not toScrapMap.has_key(manualDeal['sku']):
69
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
70
            if len(data) > 0:
71
                data[0]['dealFlag'] = 1
72
                data[0]['dealType'] = manualDeal['dealType']
73
                toScrapMap[manualDeal['sku']] = data[0]
74
        else:
75
            data = toScrapMap.get(manualDeal['sku'])
76
            data['dealFlag'] = 1
77
            data['dealType'] = manualDeal['dealType']
16502 kshitij.so 78
 
16019 kshitij.so 79
    for val in toScrapMap.values():
80
        updatePrices(val)
81
#    pool = ThreadPool(cpu_count() *2)
82
#    offset = 0
83
#    limit =100
84
#    while(offset<=len(toScrapMap.values())):
85
#        pool.map(updatePrices,toScrapMap.values()[offset:offset+limit])
86
#        offset = offset + limit
87
#    pool.close()
88
#    pool.join()
89
#    print "joining threads at %s"%(str(datetime.now()))
14180 kshitij.so 90
 
91
 
92
def updatePrices(data):
16502 kshitij.so 93
    if data.get('ignorePricing') ==1:
16429 kshitij.so 94
        print "Ignored items returning for %d"%(data['_id'])
95
        return
14180 kshitij.so 96
    if data['source_id']!=3:
97
        return
98
    print data['identifier']
99
    if data['identifier'] is None or len(data['identifier'].strip())==0:
100
        print "returning"
101
        return
102
 
103
    try:
104
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
105
            print "sku id is already updated",data['_id'] 
106
            return
107
    except:
108
        pass
109
 
15823 kshitij.so 110
    url="http://www.snapdeal.com/acors/json/v2/gvbps?supc=%s&catUrl=&bn=&catId=175&start=0&count=10000"%(data['identifier'].strip())
14180 kshitij.so 111
    print url
112
    lowestOfferPrice = 0
113
    instock = 0
15270 kshitij.so 114
    buyBoxPrice = 0
115
    isBuyBox = 1
116
    stock = 0
15820 kshitij.so 117
    try:
16019 kshitij.so 118
        req = urllib2.Request(url,headers=headers)
119
        response = urllib2.urlopen(req)
16085 kshitij.so 120
        snapdeal_data = response.read()
16087 kshitij.so 121
        response.close()
16019 kshitij.so 122
    except:
123
        print "Unable to scrape %d"%(data['_id'])
16087 kshitij.so 124
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':0,'priceUpdatedOn':to_java_date(datetime.now())}})
125
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':0}})
16019 kshitij.so 126
        return
16085 kshitij.so 127
    encoding =  chardet.detect(snapdeal_data)
16019 kshitij.so 128
    try:
16085 kshitij.so 129
        snapdeal_data = snapdeal_data.decode(encoding.get('encoding'))
15820 kshitij.so 130
    except:
16429 kshitij.so 131
        snapdeal_data = snapdeal_data.decode('latin-1')
16085 kshitij.so 132
    try:
133
        vendorInfo = json.loads(snapdeal_data)
134
    except:
16086 kshitij.so 135
        print "Unable to load json for %d"%(data['_id'])
15820 kshitij.so 136
        return
15270 kshitij.so 137
 
15820 kshitij.so 138
    try:
139
        buyBoxStock = vendorInfo['primaryVendor']['buyableInventory']
140
        if buyBoxStock >0:
141
            buyBoxPrice = vendorInfo['primaryVendor']['sellingPrice']
142
    except:
143
        pass
144
 
145
    sortedVendorsData = sorted(vendorInfo['vendors'], key=itemgetter('sellingPrice'))
146
    for sortedVendorData in sortedVendorsData:
147
        lowestOfferPrice = float(sortedVendorData['sellingPrice'])
148
        try:
149
            stock = sortedVendorData['buyableInventory']
150
        except:
151
            pass
152
        if stock > 0 and lowestOfferPrice > 0:
153
            instock = 1
154
            break
155
    if buyBoxPrice != lowestOfferPrice:
156
        isBuyBox = 0
157
 
14180 kshitij.so 158
    print lowestOfferPrice
159
    print instock
14181 kshitij.so 160
    print "Lowest Offer Price for id %d is %d , stock is %d and stock count is %d" %(data['_id'],lowestOfferPrice,instock,stock)
14180 kshitij.so 161
    print "*************"
162
    if instock  == 1:
15270 kshitij.so 163
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':instock,'buyBoxFlag':isBuyBox}}, multi=True)
16019 kshitij.so 164
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice , 'in_stock':instock,'codAvailable':data['codAvailable']}}, multi=True)
14180 kshitij.so 165
    else:
15270 kshitij.so 166
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':instock,'priceUpdatedOn':to_java_date(datetime.now()),'buyBoxFlag':isBuyBox}}, multi=True)
16019 kshitij.so 167
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':instock,'codAvailable':data['codAvailable']}})
14180 kshitij.so 168
 
169
    try:
15270 kshitij.so 170
        recomputeDeal(data)
14180 kshitij.so 171
    except:
172
        print "Unable to compute deal for ",data['skuBundleId']
173
 
14325 kshitij.so 174
 
175
def populateNegativeDeals():
176
    negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
15270 kshitij.so 177
    mc.set("negative_deals", negativeDeals, 600)
178
 
16502 kshitij.so 179
#def recomputePoints(item, deal):
180
#    try:
181
#        if item.get('available_price') == deal['available_price']:
182
#            print "No need to compute points for %d , as price is still same" %(item['_id'])
183
#            raise
184
#        nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
185
#    except:
186
#        print traceback.print_exc()
187
#        nlcPoints = deal['nlcPoints']
188
#    
189
#    bundleDealPoints = list(get_mongo_connection().Catalog.DealPoints.find({'skuBundleId':item['skuBundleId'],'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())}}))
190
#    if len(bundleDealPoints) > 0:
191
#        item['manualDealThresholdPrice'] = bundleDealPoints[0]['dealThresholdPrice']
192
#        dealPoints = bundleDealPoints[0]['dealPoints']
193
#    else:
194
#        dealPoints = 0
195
#        item['manualDealThresholdPrice'] = None
196
#        
197
#    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']}})
15270 kshitij.so 198
 
14180 kshitij.so 199
 
15270 kshitij.so 200
def recomputeDeal(item):
13917 kshitij.so 201
    """Lets recompute deal for this bundle"""
16019 kshitij.so 202
    print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
15270 kshitij.so 203
    skuBundleId = item['skuBundleId']
13917 kshitij.so 204
 
205
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
206
    bestPrice = float("inf")
207
    bestOne = None
208
    bestSellerPoints = 0
209
    toUpdate = []
16019 kshitij.so 210
    prepaidBestPrice = float("inf")
211
    prepaidBestOne = None
212
    prepaidBestSellerPoints = 0
13917 kshitij.so 213
    for similarItem in similarItems:
16019 kshitij.so 214
        if similarItem['codAvailable'] ==1:
215
            if mc.get("negative_deals") is None:
216
                populateNegativeDeals()
16174 kshitij.so 217
            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):
16019 kshitij.so 218
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
219
                continue
220
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
221
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
222
                continue
223
            if similarItem['available_price'] < bestPrice:
224
                bestOne = similarItem
225
                bestPrice = similarItem['available_price']
226
                bestSellerPoints = similarItem['bestSellerPoints']
227
            elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
228
                bestOne = similarItem
229
                bestPrice = similarItem['available_price']
230
                bestSellerPoints = similarItem['bestSellerPoints']
231
            else:
232
                pass
13917 kshitij.so 233
        else:
16019 kshitij.so 234
            if mc.get("negative_deals") is None:
235
                populateNegativeDeals()
16174 kshitij.so 236
            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):
16019 kshitij.so 237
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
238
                continue
239
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
240
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
241
                continue
16869 kshitij.so 242
            if similarItem['source_id'] == SOURCE_MAP.get('PAYTM.COM'):
243
                similarItem['available_price'] = similarItem['gross_price']
16019 kshitij.so 244
            if similarItem['available_price'] < prepaidBestPrice:
245
                prepaidBestOne = similarItem
246
                prepaidBestPrice = similarItem['available_price']
247
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
248
            elif similarItem['available_price'] == prepaidBestPrice and prepaidBestSellerPoints < similarItem['bestSellerPoints']:
249
                prepaidBestOne = similarItem
250
                prepaidBestPrice = similarItem['available_price']
251
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
252
            else:
253
                pass
16026 kshitij.so 254
    if bestOne is not None or prepaidBestOne is not None:
13917 kshitij.so 255
        for similarItem in similarItems:
256
            toUpdate.append(similarItem['_id'])
16026 kshitij.so 257
        if bestOne is not None:
258
            toUpdate.remove(bestOne['_id'])
259
            get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
260
        if prepaidBestOne is not None:
16070 kshitij.so 261
            if bestOne is not None:
262
                if prepaidBestOne['available_price'] < bestOne['available_price']: 
263
                    toUpdate.remove(prepaidBestOne['_id'])
264
                    get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
265
            else:
266
                toUpdate.remove(prepaidBestOne['_id'])
267
                get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
13917 kshitij.so 268
    if len(toUpdate) > 0:
16019 kshitij.so 269
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
16026 kshitij.so 270
 
13828 kshitij.so 271
def main():
14180 kshitij.so 272
    populate()
13828 kshitij.so 273
 
274
if __name__=='__main__':
275
    main()