Subversion Repositories SmartDukaan

Rev

Rev 16501 | Rev 16869 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 16501 Rev 16509
Line 42... Line 42...
42
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
42
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
43
        for data in amazonBestSellers:
43
        for data in amazonBestSellers:
44
            if not toScrapMap.has_key(data['_id']):
44
            if not toScrapMap.has_key(data['_id']):
45
                data['dealFlag'] = 0
45
                data['dealFlag'] = 0
46
                data['dealType'] = 0
46
                data['dealType'] = 0
-
 
47
                data['dealUrl'] = ""
47
                toScrapMap[data['_id']] = data
48
                toScrapMap[data['_id']] = data
48
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':-100}}))
49
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':-100}}))
49
    for deal in dealFlagged:
50
    for deal in dealFlagged:
50
        if not toScrapMap.has_key(deal['_id']):
51
        if not toScrapMap.has_key(deal['_id']):
51
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
52
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))
52
            data[0]['dealFlag'] = 0
53
            data[0]['dealFlag'] = 0
53
            data[0]['dealType'] = 0
54
            data[0]['dealType'] = 0
-
 
55
            data[0]['dealUrl'] = ""
54
            toScrapMap[deal['_id']] = data[0]
56
            toScrapMap[deal['_id']] = data[0]
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}))
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':1}))
56
    for manualDeal in manualDeals:
58
    for manualDeal in manualDeals:
57
        if not toScrapMap.has_key(manualDeal['sku']):
59
        if not toScrapMap.has_key(manualDeal['sku']):
58
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
60
            data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))
59
            if len(data) > 0:
61
            if len(data) > 0:
60
                data[0]['dealFlag'] = 1
62
                data[0]['dealFlag'] = 1
61
                data[0]['dealType'] = manualDeal['dealType']
63
                data[0]['dealType'] = manualDeal['dealType']
-
 
64
                data[0]['dealUrl'] = manualDeal['dealUrl']
62
                toScrapMap[manualDeal['sku']] = data[0]
65
                toScrapMap[manualDeal['sku']] = data[0]
63
        else:
66
        else:
64
            data = toScrapMap.get(manualDeal['sku'])
67
            data = toScrapMap.get(manualDeal['sku'])
65
            data['dealFlag'] = 1
68
            data['dealFlag'] = 1
66
            data['dealType'] = manualDeal['dealType']
69
            data['dealType'] = manualDeal['dealType']
-
 
70
            data['dealUrl'] = manualDeal['dealUrl']
67
    pool = ThreadPool(cpu_count() * 2)
71
    pool = ThreadPool(cpu_count() * 2)
68
    pool.map(scrapeAmazon,toScrapMap.values())
72
    pool.map(scrapeAmazon,toScrapMap.values())
69
    pool.close()
73
    pool.close()
70
    pool.join()
74
    pool.join()
71
    print "joining threads at %s"%(str(datetime.now()))
75
    print "joining threads at %s"%(str(datetime.now()))
Line 91... Line 95...
91
    
95
    
92
    url = "http://www.amazon.in/gp/aw/ol/%s?o=New&op=1"%(data['identifier'])
96
    url = "http://www.amazon.in/gp/aw/ol/%s?o=New&op=1"%(data['identifier'])
93
    
97
    
94
    try:
98
    try:
95
        if data['dealFlag'] ==1 and data['dealType'] ==1:
99
        if data['dealFlag'] ==1 and data['dealType'] ==1:
96
            deal_url = "http://www.amazon.in/dp/%s"%(data['identifier'].strip())
100
            deal_url = data['dealUrl'].strip()
97
            dealScraping = True
101
            dealScraping = True
98
    except:
102
    except:
99
        data['dealFlag'] = 0
103
        data['dealFlag'] = 0
100
        data['dealType'] = 0
104
        data['dealType'] = 0
-
 
105
        data['dealUrl'] = ""
101
        
106
        
102
    print url
107
    print url
103
    lowestPrice = 0.0
108
    lowestPrice = 0.0
104
    
109
    
105
    if not dealScraping:
110
    if not dealScraping: