Subversion Repositories SmartDukaan

Rev

Rev 19593 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
19495 kshitij.so 1
from elixir import *
2
from dtr.storage.DataService import FlipkartOrders
3
from dtr.storage import DataService
4
from dtr.utils.utils import get_mongo_connection, get_mongo_connection_dtr_data, \
5
to_java_date, to_py_date
6
from cassandra.cluster import Cluster
7
import optparse
8
from datetime import datetime, timedelta, time
9
 
10
parser = optparse.OptionParser()
11
parser.add_option("-H", "--host", dest="hostname",
12
                      default="localhost",
13
                      type="string", help="The HOST where the DB server is running",
14
                      metavar="host")
15
parser.add_option("-m", "--m", dest="mongoHostDtr",
16
                      default="localhost",
17
                      type="string", help="Host where Catalog/Dtr dbs exist",
18
                      metavar="mongo_host")
19
parser.add_option("-d", "--d", dest="mongoHostDtrData",
20
                      default="localhost",
21
                      type="string", help="Host to dump data",
22
                      metavar="mongo_host_dtr_data")
23
 
24
parser.add_option("-c", "--c", dest="cassandraHost",
25
                      default="localhost",
26
                      type="string", help="Cassandra host",
27
                      metavar="cassandra_host")
28
 
29
(options, args) = parser.parse_args()
30
 
31
dtr_host = options.mongoHostDtr
32
dtr_data_host = options.mongoHostDtrData
33
db_hostname = options.hostname
34
cassandra_host = options.cassandraHost
35
 
36
cluster = Cluster([cassandra_host])
37
session_cassandra = cluster.connect()
38
session_cassandra.set_keyspace("profitmandi")
39
 
40
print dtr_host
41
print db_hostname
42
 
43
 
44
bundleMap = {}
45
 
46
class Analytics:
47
    def __init__(self, skuBundleId, sales, views, date):
48
        self.skuBundleId = skuBundleId
49
        self.sales = sales
50
        self.views = views
51
        self.date = date
52
 
53
def createViewsMap():
54
    global bundleMap
55
    #Datefield - Midnight.
56
    #start - Previous day midnight
57
    #Data stored in cassandra is in utc format
58
    date_field = (datetime.combine((datetime.now()).date(), time.min))
59
    start = to_java_date(date_field -timedelta(days=1))
60
    end = to_java_date(date_field)
61
    utc_start =  to_java_date(datetime.utcfromtimestamp((start)/1000))
62
    utc_end =  to_java_date(datetime.utcfromtimestamp((end)/1000))
63
    query = "select * from events where label='click' and time >='"+str(utc_start)+"' and time < '"+str(utc_end)+"' ALLOW FILTERING"
64
    print query
65
    result = session_cassandra.execute(query)
66
    for i in result:
19650 kshitij.so 67
        try:
68
            sku = int((i.properties)['sku'])
69
        except:
70
            continue
19495 kshitij.so 71
        master = get_mongo_connection(host=dtr_host).Catalog.MasterData.find_one({'_id':sku})
72
        if master is None:
73
            continue
74
        bundleId = master['skuBundleId']
75
        if bundleMap.has_key(bundleId):
76
            obj = bundleMap.get(bundleId)
77
            obj.views = obj.views + 1 
78
        else:
79
            obj = Analytics(bundleId, 0, 1, start)
80
            bundleMap[bundleId] = obj
81
 
82
def createSalesMap():
83
    DataService.initialize(db_hostname=db_hostname)
84
    date_field = (datetime.combine((datetime.now()).date(), time.min))
85
    start = to_java_date(date_field -timedelta(days=1))
86
    end = to_java_date(date_field)
87
    all_orders = get_mongo_connection(host=dtr_host).Dtr.merchantOrder.find({'createdOnInt':{"$gte":start/1000,"$lt":end/1000}})
88
    for a in all_orders:
89
        store = a['storeId']
90
        if a.get('subOrders') is None:
91
            continue
92
        subOrders = a['subOrders']
93
        for s in subOrders:
94
            if s.get('productCode') is None:
95
                continue
96
            productCode = str(s['productCode'])
97
            qty = s['quantity']
98
            if store!=3:
99
                master = get_mongo_connection(host=dtr_host).Catalog.MasterData.find_one({'identifier':productCode,'source_id':store})
100
            else:
101
                master = get_mongo_connection(host=dtr_host).Catalog.MasterData.find_one({'secondaryIdentifier':productCode,'source_id':store})
102
            if master is None:
103
                continue
104
            skuBundleId = master.get('skuBundleId')
105
            if bundleMap.has_key(skuBundleId):
106
                obj = bundleMap.get(skuBundleId)
107
                obj.sales = obj.sales + qty 
108
            else:
109
                obj = Analytics(skuBundleId, qty, 0, start)
110
                bundleMap[skuBundleId] = obj
111
    fk_orders = session.query(FlipkartOrders).filter(FlipkartOrders.created> to_py_date(start)).filter(FlipkartOrders.created< to_py_date(end)).all()
112
    for fk_order in fk_orders:
113
        if fk_order.catalogId is None:
114
            continue
115
        skuBundleId = fk_order.catalogId
116
        qty = fk_order.quantity
117
        if bundleMap.has_key(skuBundleId):
118
            obj = bundleMap.get(skuBundleId)
119
            obj.sales = obj.sales + qty 
120
        else:
121
            obj = Analytics(skuBundleId, qty, 0, start)
122
            bundleMap[skuBundleId] = obj
123
    session.close()
124
 
125
def commit():
126
    for v in bundleMap.values():
127
        get_mongo_connection_dtr_data().Analytics.ViewSalesData.insert(v.__dict__)
128
 
19591 kshitij.so 129
def assignBestSellerRanks():
130
    skuBundleRankMap = {}
131
    date_field = (datetime.combine((datetime.now()).date(), time.min))
132
    start = to_java_date(date_field -timedelta(days=30))
133
    salesData = get_mongo_connection_dtr_data().Analytics.ViewSalesData.find({'date':{"$gte":start}},{'skuBundleId':1,'sales':1})
134
    for data in salesData:
135
        if skuBundleRankMap.has_key(data['skuBundleId']):
136
            skuBundleRankMap[data['skuBundleId']] = skuBundleRankMap.get(data['skuBundleId']) + data['sales']
137
        else:
138
            skuBundleRankMap[data['skuBundleId']] = data['sales']
139
    sorted_skuBundleRankMap = sorted(skuBundleRankMap.items(), key=lambda x: x[1],reverse=True)
19593 kshitij.so 140
    get_mongo_connection(host=dtr_host).Catalog.MasterData.update({'rank':{"$gt":0},'category_id':6},{"$set":{'rank':0}},upsert=False,multi=True)
19591 kshitij.so 141
    rank = 1
142
    for i in sorted_skuBundleRankMap:
143
        if rank > 100:
144
            return
145
        master = get_mongo_connection(host=dtr_host).Catalog.MasterData.find_one({'skuBundleId':i[0]})
146
        if master is None or master['category_id']!=6:
147
            continue
19592 kshitij.so 148
        get_mongo_connection(host=dtr_host).Catalog.MasterData.update({'skuBundleId':master['skuBundleId']},{"$set":{"rank":rank,'updatedOn':to_java_date(datetime.now())}},upsert=False,multi=True)
19591 kshitij.so 149
        rank = rank + 1
150
 
151
 
19495 kshitij.so 152
def main():
153
    createViewsMap()
154
    createSalesMap()
155
    commit()
19591 kshitij.so 156
    assignBestSellerRanks()
19495 kshitij.so 157
 
158
 
159
if __name__ == '__main__':
160
    main()