Subversion Repositories SmartDukaan

Rev

Rev 16355 | Rev 16509 | 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
15267 kshitij.so 2
from dtr.utils.utils import to_java_date, getNlcPoints
13914 kshitij.so 3
from datetime import datetime, timedelta
14308 kshitij.so 4
from dtr.utils import AmazonPriceOnlyScraper, AmazonDealScraper
14174 kshitij.so 5
from multiprocessing import Pool as ThreadPool
14170 kshitij.so 6
from multiprocessing import cpu_count
14256 kshitij.so 7
import optparse
14325 kshitij.so 8
from dtr.storage.MemCache import MemCache
14705 kshitij.so 9
from dtr.utils.utils import getCashBack
15267 kshitij.so 10
import traceback
13828 kshitij.so 11
 
12
con = None
13
 
14256 kshitij.so 14
parser = optparse.OptionParser()
15
parser.add_option("-m", "--m", dest="mongoHost",
16
                      default="localhost",
17
                      type="string", help="The HOST where the mongo server is running",
18
                      metavar="mongo_host")
19
 
20
(options, args) = parser.parse_args()
21
 
14325 kshitij.so 22
mc = MemCache(options.mongoHost)
23
 
15616 kshitij.so 24
 
16019 kshitij.so 25
SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5}
26
 
14256 kshitij.so 27
def get_mongo_connection(host=options.mongoHost, port=27017):
13828 kshitij.so 28
    global con
29
    if con is None:
30
        print "Establishing connection %s host and port %d" %(host,port)
31
        try:
32
            con = pymongo.MongoClient(host, port)
33
        except Exception, e:
34
            print e
35
            return None
36
    return con
37
 
14147 kshitij.so 38
def populate():
39
    toScrapMap = {}
14133 kshitij.so 40
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
41
    for bestSeller in bestSellers: 
42
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
43
        for data in amazonBestSellers:
14147 kshitij.so 44
            if not toScrapMap.has_key(data['_id']):
15267 kshitij.so 45
                data['dealFlag'] = 0
46
                data['dealType'] = 0
14147 kshitij.so 47
                toScrapMap[data['_id']] = data
16173 kshitij.so 48
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':-100}}))
14250 kshitij.so 49
    for deal in dealFlagged:
50
        if not toScrapMap.has_key(deal['_id']):
14260 kshitij.so 51
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
15267 kshitij.so 52
            data[0]['dealFlag'] = 0
53
            data[0]['dealType'] = 0
14260 kshitij.so 54
            toScrapMap[deal['_id']] = data[0]
14308 kshitij.so 55
    manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':1}))
56
    for manualDeal in manualDeals:
57
        if not toScrapMap.has_key(manualDeal['sku']):
58
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
59
            if len(data) > 0:
60
                data[0]['dealFlag'] = 1
61
                data[0]['dealType'] = manualDeal['dealType']
62
                toScrapMap[manualDeal['sku']] = data[0]
63
        else:
64
            data = toScrapMap.get(manualDeal['sku'])
65
            data['dealFlag'] = 1
66
            data['dealType'] = manualDeal['dealType']
14170 kshitij.so 67
    pool = ThreadPool(cpu_count() * 2)
14147 kshitij.so 68
    pool.map(scrapeAmazon,toScrapMap.values())
69
    pool.close()
70
    pool.join()
14250 kshitij.so 71
    print "joining threads at %s"%(str(datetime.now()))
14147 kshitij.so 72
 
73
 
74
def scrapeAmazon(data):
75
    inStock = 0
14308 kshitij.so 76
    dealScraping = False
14147 kshitij.so 77
    print str(data['identifier'])
78
    if data['identifier'] is None or len(data['identifier'].strip())==0:
79
        return
80
 
16501 kshitij.so 81
    if data.get('ignorePricing') ==1:
15616 kshitij.so 82
        print "Ignored items returning for %d"%(data['_id'])
83
        return 
84
 
14147 kshitij.so 85
    try:
86
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
87
            print "sku id is already updated",data['_id'] 
88
            return
89
    except:
90
        pass
91
 
15959 kshitij.so 92
    url = "http://www.amazon.in/gp/aw/ol/%s?o=New&op=1"%(data['identifier'])
14308 kshitij.so 93
 
94
    try:
95
        if data['dealFlag'] ==1 and data['dealType'] ==1:
96
            deal_url = "http://www.amazon.in/dp/%s"%(data['identifier'].strip())
97
            dealScraping = True
98
    except:
99
        data['dealFlag'] = 0
100
        data['dealType'] = 0
101
 
14147 kshitij.so 102
    print url
103
    lowestPrice = 0.0
14308 kshitij.so 104
 
105
    if not dealScraping:
106
        scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
107
        lowestPrice = scraperAmazon.read(url)
108
    else:
109
        dealScraperAmazon = AmazonDealScraper.AmazonScraper()
110
        lowestPrice = dealScraperAmazon.read(deal_url)
111
        if lowestPrice == 0:
112
            scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()
113
            lowestPrice = scraperAmazon.read(url)
114
            dealScraping = False
115
 
14147 kshitij.so 116
    print lowestPrice
117
    if lowestPrice > 0:
118
        inStock = 1
119
    print lowestPrice
120
    print inStock
121
    if lowestPrice > 0:
122
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)
16019 kshitij.so 123
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock,'dealType':data['dealType'],'codAvailable':data['codAvailable']}}, multi=True)
14147 kshitij.so 124
    else:
125
        get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)
16019 kshitij.so 126
        get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'dealType':data['dealType'],'codAvailable':data['codAvailable']}})
14147 kshitij.so 127
 
128
    try:
15267 kshitij.so 129
        recomputeDeal(data)
14147 kshitij.so 130
    except:
131
        print "Unable to compute deal for ",data['skuBundleId']    
13914 kshitij.so 132
 
14325 kshitij.so 133
def populateNegativeDeals():
134
    negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')
15267 kshitij.so 135
    mc.set("negative_deals", negativeDeals, 600)
13828 kshitij.so 136
 
16501 kshitij.so 137
#def recomputePoints(item, deal):
138
#    try:
139
#        if item.get('available_price') == deal['available_price']:
140
#            print "No need to compute points for %d , as price is still same" %(item['_id'])
141
#            raise
142
#        nlcPoints = getNlcPoints(item, deal['minNlc'], deal['maxNlc'], deal['available_price'])
143
#    except:
144
#        traceback.print_exc()
145
#        nlcPoints = deal['nlcPoints']
146
#    
147
#    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())}}))
148
#    if len(bundleDealPoints) > 0:
149
#        item['manualDealThresholdPrice'] = bundleDealPoints[0]['dealThresholdPrice']
150
#        dealPoints = bundleDealPoints[0]['dealPoints']
151
#    else:
152
#        dealPoints = 0
153
#        item['manualDealThresholdPrice'] = None
154
#    
155
#    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']}})
15267 kshitij.so 156
 
157
 
158
 
159
def recomputeDeal(item):
13914 kshitij.so 160
    """Lets recompute deal for this bundle"""
16019 kshitij.so 161
    print "Recomputing for bundleId %d" %(item.get('skuBundleId'))
15267 kshitij.so 162
    skuBundleId = item['skuBundleId']
13914 kshitij.so 163
 
164
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
165
    bestPrice = float("inf")
166
    bestOne = None
167
    bestSellerPoints = 0
168
    toUpdate = []
16019 kshitij.so 169
    prepaidBestPrice = float("inf")
170
    prepaidBestOne = None
171
    prepaidBestSellerPoints = 0
13914 kshitij.so 172
    for similarItem in similarItems:
16019 kshitij.so 173
        if similarItem['codAvailable'] ==1:
174
            if mc.get("negative_deals") is None:
175
                populateNegativeDeals()
16173 kshitij.so 176
            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):
16019 kshitij.so 177
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
178
                continue
179
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
180
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
181
                continue
182
            if similarItem['available_price'] < bestPrice:
183
                bestOne = similarItem
184
                bestPrice = similarItem['available_price']
185
                bestSellerPoints = similarItem['bestSellerPoints']
186
            elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
187
                bestOne = similarItem
188
                bestPrice = similarItem['available_price']
189
                bestSellerPoints = similarItem['bestSellerPoints']
190
            else:
191
                pass
13914 kshitij.so 192
        else:
16019 kshitij.so 193
            if mc.get("negative_deals") is None:
194
                populateNegativeDeals()
16173 kshitij.so 195
            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):
16019 kshitij.so 196
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})
197
                continue
198
            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:
199
                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})
200
                continue
201
            if similarItem['available_price'] < prepaidBestPrice:
202
                prepaidBestOne = similarItem
203
                prepaidBestPrice = similarItem['available_price']
204
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
205
            elif similarItem['available_price'] == prepaidBestPrice and prepaidBestSellerPoints < similarItem['bestSellerPoints']:
206
                prepaidBestOne = similarItem
207
                prepaidBestPrice = similarItem['available_price']
208
                prepaidBestSellerPoints = similarItem['bestSellerPoints']
209
            else:
210
                pass
16026 kshitij.so 211
    if bestOne is not None or prepaidBestOne is not None:
13914 kshitij.so 212
        for similarItem in similarItems:
213
            toUpdate.append(similarItem['_id'])
16026 kshitij.so 214
        if bestOne is not None:
215
            toUpdate.remove(bestOne['_id'])
216
            get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})
217
        if prepaidBestOne is not None:
16073 kshitij.so 218
            if bestOne is not None:
219
                if prepaidBestOne['available_price'] < bestOne['available_price']: 
220
                    toUpdate.remove(prepaidBestOne['_id'])
221
                    get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
222
            else:
223
                toUpdate.remove(prepaidBestOne['_id'])
224
                get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})
13914 kshitij.so 225
    if len(toUpdate) > 0:
16019 kshitij.so 226
        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)
13914 kshitij.so 227
 
13828 kshitij.so 228
def main():
14147 kshitij.so 229
    populate()
13828 kshitij.so 230
 
231
if __name__=='__main__':
232
    main()