Subversion Repositories SmartDukaan

Rev

Rev 16202 | Rev 16226 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
15509 manish.sha 1
'''
2
Created on Jan 15, 2015
3
 
4
@author: Manish 
5
'''
6
from bs4 import BeautifulSoup
7
from bson.binary import Binary
8
from datetime import datetime, date, timedelta
9
from dtr import main
15791 manish.sha 10
from dtr.dao import AffiliateInfo, Order, SubOrder, ShopCluesAffiliateInfo
15509 manish.sha 11
from dtr.main import getBrowserObject, ScrapeException, getStore, ParseException, \
12
    Store as MStore, ungzipResponse, tprint
13
from dtr.storage import Mongo
14
from dtr.storage.Mongo import getImgSrc
15
from dtr.utils.utils import fetchResponseUsingProxy
16
from pprint import pprint
17
from pymongo import MongoClient
18
import json
19
import pymongo
20
import re
21
import time
22
import traceback
23
import urllib
15791 manish.sha 24
import urllib2
15509 manish.sha 25
from urlparse import urlparse, parse_qs
15791 manish.sha 26
import xml.etree.ElementTree as ET
27
from dtr.storage import MemCache
16097 manish.sha 28
from dtr.storage.Mongo import getDealRank
15509 manish.sha 29
 
30
USERNAME='profittill2@gmail.com'
31
PASSWORD='spice@2020'
15791 manish.sha 32
AFFLIATE_TRASACTIONS_URL = "https://admin.optimisemedia.com/v2/reports/affiliate/leads/leadsummaryexport.aspx?Contact=796881&Country=26&Agency=95&Merchant=420562&Status=-1&Year=%d&Month=%d&Day=%d&EndYear=%d&EndMonth=%d&EndDay=%d&DateType=0&Sort=CompletionDate&Login=1347562DA5E3EFF6FB1561765C47C782&Format=XML&RestrictURL=0"
15509 manish.sha 33
ORDER_TRACK_URL='http://www.shopclues.com/index.php?dispatch=order_lookup.details'
15791 manish.sha 34
ORDER_TRACK_URL_DB='https://sm.shopclues.com/trackOrder?'
15509 manish.sha 35
BASE_URL= 'http://www.shopclues.com'
36
BASE_MURL= 'http://m.shopclues.com'
37
 
38
class Store(MStore):
39
 
40
    '''
41
    This is to map order statuses of our system to order statuses of snapdeal.
42
    And our statuses will change accordingly.
43
 
44
    '''
45
    OrderStatusMap = {
16199 manish.sha 46
                      MStore.ORDER_PLACED : ['payment successful', 'new order - cod confirmation pending', 'processing', 'quality check','on schedule', 'processing - pickup initiated', 'processing - ready to dispatch','processing - procurement delay from merchant','processing - slight procurment delay from merchant',''],
15509 manish.sha 47
                      MStore.ORDER_DELIVERED : ['delivered', 'complete'],
48
                      MStore.ORDER_SHIPPED : ['in transit', 'dispatched','shipped','order handed to courier'],
16154 manish.sha 49
                      MStore.ORDER_CANCELLED : ['payment failed', 'canceled', 'payment declined', 'order on hold - cancellation requested by customer', 'courier returned', 'canceled on customer request', 'canceled by customer','order canceled by customer']
15509 manish.sha 50
                      }
51
    OrderStatusConfirmationMap= {
15791 manish.sha 52
                                 "P" : "Payment Successful",
53
                                 "D" : "Order Declined",
54
                                 "O" : "New Order - COD confirmation Pending"
15509 manish.sha 55
                                 }
56
 
57
    CONF_CB_AMOUNT = MStore.CONF_CB_DISCOUNTED_PRICE
58
 
16111 manish.sha 59
    def __init__(self,store_id):
60
        super(Store, self).__init__(store_id)
61
 
15791 manish.sha 62
    def convertToObj(self,offer):
16202 manish.sha 63
        orderRef = offer['MerchantRef']
64
        if len(orderRef)>15:
65
            orderRef = orderRef[0:len(orderRef)-10]
16203 manish.sha 66
        offer1 = ShopCluesAffiliateInfo(offer['UID'], offer['TransactionTime'], offer['TransactionId'], orderRef,  orderRef, offer['Merchant'], offer['PID'], offer['Product'], float(str(offer['SR'])), float(str(offer['TransactionValue'])), offer['UKey'], offer['ClickTime'], offer['Status'])
15791 manish.sha 67
        return offer1
68
 
69
    def _saveToAffiliate(self, offers):
70
        collection = self.db.shopcluesOrderAffiliateInfo
71
        mcollection = self.db.merchantOrder
72
        for offerObj in offers:
73
            offer = self.convertToObj(offerObj)
74
            collection.update({"transactionId":offer.transactionId, "subTagId":offer.subTagId, "payOut":offer.payOut},{"$set":todict(offer)}, upsert=True)
75
            mcollection.update({"subTagId":offer.subTagId, "storeId":self.store_id, "subOrders.missingAff":True}, {"$set":{"subOrders.$.missingAff":False}})
76
 
15940 manish.sha 77
    def scrapeAffiliate(self, startDate=datetime.today() - timedelta(days=10), endDate=datetime.today()):
15791 manish.sha 78
        uri = AFFLIATE_TRASACTIONS_URL%(startDate.year,startDate.month,startDate.day,endDate.year,endDate.month,endDate.day)
79
        root = ET.parse(urllib2.urlopen(uri)).getroot()
80
        if len(root)> 0 and len(root[0])> 0:
81
            offers = []
82
            for child in root[0][0]:
83
                offers.append(child.attrib)
84
            self._saveToAffiliate(offers)
85
 
15509 manish.sha 86
    def _setLastSaleDate(self, saleDate):
87
        self.db.lastSaleDtate.update({'storeId':self.store_id}, {'$set':{'saleDate':saleDate}})
88
 
89
    def getName(self):
90
        return "shopclues"    
91
 
92
    def _getLastSaleDate(self,):
93
        lastDaySaleObj = self.db.lastDaySale.find_one({"storeId":self.store_id})
94
        if lastDaySaleObj is None:
95
            return datetime.min
96
 
97
    def _getStatusFromDetailedStatus(self, detailedStatus):
98
        for key, value in Store.OrderStatusMap.iteritems():
99
            if detailedStatus.lower() in value:
100
                return key
101
        print "Detailed Status need to be mapped", detailedStatus, self.store_id
102
        return None
103
 
15791 manish.sha 104
    def _getSingleSubOrderMap(self, orderId, soup, orderObj):
105
        orderStatus = self.OrderStatusConfirmationMap.get(orderObj['0']['status'])
106
        statusTime= soup.findAll(attrs={'class' : 'sts no_mobile'})[1].contents[2].strip()
15509 manish.sha 107
        productDetailsMap = {}
108
        orderTable = soup.body.find("table", {'class':'table product-list'}).findAll('tr', recursive=False)
15791 manish.sha 109
        firstRow = orderTable.pop(0)
15949 manish.sha 110
        totalColumns = len(firstRow.find_all('th'))
15791 manish.sha 111
        jsonSubOrdersMap = {}
112
        count = 1
113
 
114
        for val in orderObj['0']['items'].values():
115
            newCount = 0
16110 manish.sha 116
            counts = []
15791 manish.sha 117
            for key in jsonSubOrdersMap.keys():
118
                splitKey = key.split('-')
16112 manish.sha 119
                if orderObj['0']['is_parent_order'] == 'N':
120
                    if str(val['order_id']) == splitKey[0]:
121
                        counts.append(int(splitKey[1]))
122
                else:
123
                    if str(val['child']) == splitKey[0]:
124
                        counts.append(int(splitKey[1]))
16110 manish.sha 125
 
126
            if len(counts) >0:
127
                newCount = max(counts)
16108 manish.sha 128
            count = newCount +1
16112 manish.sha 129
            if orderObj['0']['is_parent_order'] == 'N':
130
                jsonSubOrdersMap[str(val['order_id'])+'-'+str(count)] = val
131
            else:
132
                jsonSubOrdersMap[str(val['child'])+'-'+str(count)] = val
15791 manish.sha 133
 
16110 manish.sha 134
 
15941 manish.sha 135
        print jsonSubOrdersMap.items()
136
 
15791 manish.sha 137
        count = 1
15509 manish.sha 138
        for orderTr in orderTable:
15791 manish.sha 139
            productDetailsSubMap = None
140
            productDetailsSubMap = {}
15509 manish.sha 141
            cols = orderTr.find_all('td')
142
            product_details = cols[0].find_all('a')
143
            #print product_details
144
            productUrl = product_details[0].get('href')
145
            productName = product_details[0].contents[0].strip()
146
            quantity = int(cols[1].text.strip())
15980 manish.sha 147
            sellingPrice = long(cols[2].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 148
            discount = 0
149
            subtotal = 0
150
            if totalColumns == 5:
15949 manish.sha 151
                if cols[3].text.strip()!='-' and 'Rs.' in cols[3].text.strip():
15980 manish.sha 152
                    discount = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 153
                else:
154
                    discount = 0
15980 manish.sha 155
                subtotal = long(cols[4].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 156
            else:
15980 manish.sha 157
                subtotal = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15509 manish.sha 158
            productDetailsSubMap['productUrl']=BASE_MURL+productUrl
159
            productDetailsSubMap['productName']=productName
15791 manish.sha 160
            productDetailsSubMap['subOrderTrackingUrl']=ORDER_TRACK_URL_DB+'order_id=' +str(orderId)+'&email_id='+ orderObj['0']['email']
15509 manish.sha 161
            productDetailsSubMap['sellingPrice']=sellingPrice
162
            productDetailsSubMap['quantity']=quantity
163
            productDetailsSubMap['discount']=discount
164
            productDetailsSubMap['subtotal']=subtotal
15942 manish.sha 165
            key = str(orderId)+'-'+str(count)
166
            print 'SubOrder Map Key--',key
15791 manish.sha 167
            jsonSubOrderDetails = jsonSubOrdersMap.get(str(orderId)+'-'+str(count))
168
            productCode = jsonSubOrderDetails['product_code']
169
            productImgUrl = jsonSubOrderDetails['images']['image_path'][0]
15509 manish.sha 170
            productDetailsSubMap['productCode']=productCode
171
            productDetailsSubMap['imgUrl']=productImgUrl
172
 
15791 manish.sha 173
            productDetailsSubMap['subOrderStatus']=orderStatus
174
            productDetailsSubMap['subOrderStatusTime']=statusTime
175
            productDetailsMap[str(orderId)+'-'+str(count)]=productDetailsSubMap
176
            count = count +1
15509 manish.sha 177
        return productDetailsMap
178
 
15791 manish.sha 179
    def _getMultiSubOrdersMap(self, orderId, soup, orderObj):
15509 manish.sha 180
        orderTable = soup.body.find("table", {'class':'table product-list'}).findAll('tr', recursive=False)
16105 manish.sha 181
        firstRow = orderTable.pop(0)
15791 manish.sha 182
        #totalColumns = len(firstRow.find_all('td'))
15509 manish.sha 183
        productDetailsMap = {}
15791 manish.sha 184
        '''
185
        jsonSubOrdersMap = {}
186
        for val in orderObj['0']['items'].values():
187
            jsonSubOrdersMap[val['order_id']] = val
188
        '''
189
        existingOrders = []
15509 manish.sha 190
        for orderTr in orderTable:
15791 manish.sha 191
            subOrderDetailsMap = {}
192
 
15509 manish.sha 193
            cols = orderTr.find_all('td')
194
            product_details = cols[0].find_all('a')
195
            #print product_details
196
            subOrderId= product_details[1].contents[0].strip()
15791 manish.sha 197
            if subOrderId in existingOrders:
198
                continue
199
            else:
200
                existingOrders.append(subOrderId)
201
            #productUrl = product_details[0].get('href')
202
            #productName = product_details[0].contents[0].strip()
203
            subOrderTrackingParsingUrl = product_details[1].get('href')
204
            #subOrderTrackingUrl = subOrderTrackingParsingUrl.split('order_lookup.details&')[1]
205
            '''
15509 manish.sha 206
            quantity = int(cols[1].text.strip())
207
            sellingPrice = long(cols[2].text.strip().replace("Rs.",""))
15791 manish.sha 208
            discount = 0
209
            subtotal = 0
210
            if totalColumns == 5:
211
                if cols[3].text.strip()!='-' or 'Rs.' in cols[3].text.strip():
212
                    discount = long(cols[3].text.strip().replace("Rs.",""))
213
                else:
214
                    discount = 0                    
215
                subtotal = long(cols[4].text.strip().replace("Rs.",""))
216
            else:
217
                subtotal = long(cols[3].text.strip().replace("Rs.",""))
15509 manish.sha 218
            productDetailsSubMap['productUrl']=BASE_MURL+productUrl
219
            productDetailsSubMap['productName']=productName
15791 manish.sha 220
            productDetailsSubMap['subOrderTrackingUrl']=ORDER_TRACK_URL_DB+subOrderTrackingUrl
15509 manish.sha 221
            productDetailsSubMap['sellingPrice']=sellingPrice
222
            productDetailsSubMap['quantity']=quantity
223
            productDetailsSubMap['discount']=discount
224
            productDetailsSubMap['subtotal']=subtotal
15791 manish.sha 225
            jsonSubOrderDetails = jsonSubOrdersMap.get(subOrderId)
226
            productCode = jsonSubOrderDetails['product_code']
227
            productImgUrl = jsonSubOrderDetails['images']['image_path'][0]
15509 manish.sha 228
            productDetailsSubMap['productCode']=productCode
229
            productDetailsSubMap['imgUrl']=productImgUrl
15791 manish.sha 230
            '''
16151 manish.sha 231
            orderTrackingPage = fetchResponseUsingProxy(BASE_URL+subOrderTrackingParsingUrl)
232
            orderTrackingPageSoup = BeautifulSoup(orderTrackingPage)
15791 manish.sha 233
            '''
15509 manish.sha 234
            subOrderStatusList = orderTrackingPageSoup.findAll(attrs={'class' : 'price ord_status'})
235
            subOrderStatus = subOrderStatusList[0].contents[0].strip()
236
            subOrderStatusTime= orderTrackingPageSoup.findAll(attrs={'class' : 'sts no_mobile'})[1].contents[2].strip()
237
            productDetailsSubMap['subOrderStatus'] = subOrderStatus
238
            productDetailsSubMap['subOrderStatusTime'] = subOrderStatusTime
239
            productDetailsSubMap['parentOrderId']=orderId
15791 manish.sha 240
            '''
241
 
242
            subOrderDetailsMap = self._getSingleSubOrderMap(subOrderId, orderTrackingPageSoup, orderObj)
15509 manish.sha 243
 
15791 manish.sha 244
            productDetailsMap = dict(productDetailsMap.items()+subOrderDetailsMap.items())
15509 manish.sha 245
 
246
        return productDetailsMap
247
 
248
    def updateCashbackInSubOrders(self, subOrders):
249
        for subOrder in subOrders:
250
            cashbackStatus = Store.CB_NA
251
            cashbackAmount = 0
252
            percentage = 0
253
            amount = subOrder.amountPaid
254
            if amount > 0:
255
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amount)
256
                if cashbackAmount > 0:
257
                    cashbackStatus = Store.CB_PENDING
258
            subOrder.cashBackStatus = cashbackStatus
259
            subOrder.cashBackAmount = cashbackAmount
260
            subOrder.cashBackPercentage = percentage
261
        return subOrders
262
 
15947 manish.sha 263
    def _parseOrders(self, orderId, mpOrderId, subTagId, userId, page, orderSuccessUrl, orderObj):
15509 manish.sha 264
        soup = BeautifulSoup(page)
265
        productDetailsMap = {}
266
        paymentFields = soup.findAll(attrs={'class' : 'box_paymentcalculations_row'})
267
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
15948 manish.sha 268
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("\r","").replace("  ","")
15509 manish.sha 269
        ordersubtotal=0
270
        ordercluebucks=0
271
        ordershippingcost=0
272
        ordertotal=0
273
 
274
        for val in paymentFields:
275
            for value in val.contents:
276
                if value is not None:
277
                    if 'div' in str(value).strip():
278
                        if 'Subtotal' in value.text.strip():
15981 manish.sha 279
                            ordersubtotal = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 280
                            print 'Order SubTotal:- ',ordersubtotal
15509 manish.sha 281
                        if 'Shipping Cost' in value.text.strip():
15981 manish.sha 282
                            ordershippingcost = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 283
                            print 'Shipping Cost:- ',ordershippingcost
15509 manish.sha 284
                        if 'Clue' in value.text.strip():
15981 manish.sha 285
                            ordercluebucks = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 286
                            print 'Clue bucks:- ',ordercluebucks
15509 manish.sha 287
                        if 'Total' in value.text.strip():
15791 manish.sha 288
                            ordertotal = val.contents[3].text.strip().replace("Rs.","")
289
                            ordertotal = ordertotal.replace(',','')
16165 manish.sha 290
                            print 'Order Total:- ',ordertotal
15791 manish.sha 291
 
292
        if orderObj['0']['is_parent_order'] == 'N':
15947 manish.sha 293
            productDetailsMap = self._getSingleSubOrderMap(mpOrderId, soup, orderObj)
15509 manish.sha 294
        else:
15947 manish.sha 295
            productDetailsMap = self._getMultiSubOrdersMap(mpOrderId, soup, orderObj)
15509 manish.sha 296
 
15791 manish.sha 297
        merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
15509 manish.sha 298
        merchantOrder.placedOn = placedOn
15791 manish.sha 299
        merchantOrder.merchantOrderId = orderObj['0']['order_id']
15509 manish.sha 300
        merchantOrder.paidAmount = ordertotal
301
        totalOrdersAmount = 0
302
        totalDiscount = 0
303
        subOrders= []
304
        for key in productDetailsMap:
305
            subOrderDetail = productDetailsMap.get(key)
306
            totalOrdersAmount = totalOrdersAmount  + (subOrderDetail['sellingPrice'] * subOrderDetail['quantity'])
307
            totalDiscount = totalDiscount + (subOrderDetail['discount'] * subOrderDetail['quantity'])
308
            subOrder = SubOrder(subOrderDetail['productName'], subOrderDetail['productUrl'], placedOn, subOrderDetail['subtotal'])
309
            subOrder.merchantSubOrderId = key
310
            subOrder.detailedStatus = subOrderDetail['subOrderStatus']
311
            subOrder.imgUrl = subOrderDetail['imgUrl']
312
            subOrder.offerDiscount = subOrderDetail['discount'] * subOrderDetail['quantity']
313
            subOrder.unitPrice = subOrderDetail['sellingPrice']
314
            subOrder.productCode = subOrderDetail['productCode']
315
            subOrder.amountPaid = subOrderDetail['subtotal']
316
            subOrder.quantity = subOrderDetail['quantity']
15791 manish.sha 317
            subOrder.tracingkUrl = subOrderDetail['subOrderTrackingUrl']
16097 manish.sha 318
            dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
319
            subOrder.dealRank = dealRank.get('rank')
320
            subOrder.rankDesc = dealRank.get('description')
15509 manish.sha 321
            subOrders.append(subOrder)
322
        merchantOrder.totalAmount = totalOrdersAmount
323
        merchantOrder.discountApplied = totalDiscount
324
        merchantOrder.deliveryCharges = ordershippingcost
325
        merchantOrder.subOrders = self.updateCashbackInSubOrders(subOrders)
326
 
327
        return merchantOrder
328
 
329
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl):
330
        resp = {}
331
        try:
15791 manish.sha 332
            rawHtmlSoup = BeautifulSoup(rawHtml)
333
            emailId = None
334
            orderObj = None
335
            merchantOrderId = None
336
            for script in rawHtmlSoup.find_all('script'):
337
                if 'orderConfirmation' in script.text:
338
                    orderObj = json.loads(script.text.strip().split('\n')[0].split('orderConfirmation = ')[1].split(';')[0])
339
                    if merchantOrderId is None:
340
                        merchantOrderId = orderObj['0']['order_id']
341
                        emailId = orderObj['0']['email']
16197 manish.sha 342
            if orderObj is None or merchantOrderId is None:
343
                resp['result'] = 'ORDER_NON_PARSEABLE'
344
                return resp
15509 manish.sha 345
            br = getBrowserObject()
15791 manish.sha 346
            url = ORDER_TRACK_URL +'&order_id=' +str(merchantOrderId)+'&email_id='+ emailId
15509 manish.sha 347
            page = br.open(url)
348
            headers = str(page.info()).split('\n')
349
            page = ungzipResponse(page)
350
            jsonResponse = None
351
            for header in headers:
352
                header = header.split(':')
353
                if header[0] == 'Content-Type' and 'json' in header[1]:
354
                    jsonResponse = json.loads(page)
355
            if jsonResponse is not None:
356
                page = jsonResponse['text']
357
 
15947 manish.sha 358
            merchantOrder = self._parseOrders(orderId, merchantOrderId, subTagId, userId, page, orderSuccessUrl, orderObj)
15509 manish.sha 359
 
15948 manish.sha 360
            merchantOrder.orderTrackingUrl = ORDER_TRACK_URL_DB + 'order_id=' +str(merchantOrderId)+'&email_id='+ emailId
15509 manish.sha 361
 
362
            if self._saveToOrder(todict(merchantOrder)):
363
                resp['result'] = 'ORDER_CREATED'
364
            else:
365
                resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
366
 
367
            return resp
368
        except:
369
            print "Error occurred"
370
            traceback.print_exc()
371
            resp['result'] = 'ORDER_NOT_CREATED'
16120 manish.sha 372
            return resp
15509 manish.sha 373
 
15791 manish.sha 374
    def parseSingleSubOrder(self, soup, emailId, subOrderId):
375
        orderStatusList = soup.findAll(attrs={'class' : 'price ord_status'})
376
        subOrderStatus = orderStatusList[0].contents[0].strip()
15509 manish.sha 377
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
15948 manish.sha 378
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("\r","").replace("  ","")    
15509 manish.sha 379
        orderTable = soup.body.find("table", {'class':'table product-list'}).findAll('tr', recursive=False)
15791 manish.sha 380
        firstRow = orderTable.pop(0)
15949 manish.sha 381
        totalColumns = len(firstRow.find_all('th'))
15791 manish.sha 382
        subOrders =[]
383
        count =1   
15509 manish.sha 384
        for orderTr in orderTable:
385
            cols = orderTr.find_all('td')
386
            product_details = cols[0].find_all('a')
387
            #print product_details
388
            productUrl = product_details[0].get('href')
389
            productName = product_details[0].contents[0].strip()
390
            quantity = int(cols[1].text.strip())
15980 manish.sha 391
            sellingPrice = long(cols[2].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 392
            discount = 0
393
            subtotal = 0
394
            if totalColumns == 5:
15949 manish.sha 395
                if cols[3].text.strip()!='-' and 'Rs.' in cols[3].text.strip():
15980 manish.sha 396
                    discount = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 397
                else:
398
                    discount = 0
15980 manish.sha 399
                subtotal = long(cols[4].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 400
            else:
15980 manish.sha 401
                subtotal = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
16151 manish.sha 402
            productPage = fetchResponseUsingProxy(BASE_MURL+productUrl)
403
            productPageSoup = BeautifulSoup(productPage)
15791 manish.sha 404
            productCode = productPageSoup.find('input', {'type':'hidden'})['value']
15509 manish.sha 405
            allproductImageTags = productPageSoup.findAll(attrs={'class' : 'pd-image'})
406
            productImgUrl = ''
407
            if allproductImageTags is not None and len(allproductImageTags)>0:
408
                productImgUrl= allproductImageTags[0].get('style').split("background:url('")[1].split("')no-repeat center")[0].strip()
409
 
410
            subOrder = SubOrder(productName, productUrl, placedOn, subtotal)
15791 manish.sha 411
            subOrder.merchantSubOrderId = str(subOrderId)+'-'+str(count)
15509 manish.sha 412
            subOrder.detailedStatus = subOrderStatus
413
            subOrder.imgUrl = productImgUrl
414
            subOrder.offerDiscount = discount*quantity
415
            subOrder.unitPrice = sellingPrice
416
            subOrder.productCode = productCode
417
            subOrder.amountPaid = subtotal
418
            subOrder.quantity = quantity
15791 manish.sha 419
            subOrder.tracingkUrl = ORDER_TRACK_URL_DB +  'order_id=' +subOrderId+'&email_id='+ emailId
15509 manish.sha 420
            subOrders.append(subOrder)
15791 manish.sha 421
            count = count +1
15509 manish.sha 422
 
423
        subOrders = self.updateCashbackInSubOrders(subOrders)
15791 manish.sha 424
        return subOrders
15509 manish.sha 425
 
15791 manish.sha 426
    def parseMultiSubOrders(self, soup, emailId):
15509 manish.sha 427
        orderTable = soup.body.find("table", {'class':'table product-list'}).findAll('tr', recursive=False)
16145 manish.sha 428
        firstRow = orderTable.pop(0)
15791 manish.sha 429
        '''
430
        totalColumns = len(firstRow.find_all('td'))
431
 
15509 manish.sha 432
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
15791 manish.sha 433
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("  ","")
434
        '''    
15509 manish.sha 435
        subOrders = []
15791 manish.sha 436
        existingOrders = []
15509 manish.sha 437
        for orderTr in orderTable:
438
            cols = orderTr.find_all('td')
439
            product_details = cols[0].find_all('a')
440
            #print product_details
441
            subOrderId= product_details[1].contents[0].strip()
15791 manish.sha 442
            if subOrderId in existingOrders:
443
                continue
444
            else:
445
                existingOrders.append(subOrderId)
446
            '''
15509 manish.sha 447
            productUrl = product_details[0].get('href')
448
            productName = product_details[0].contents[0].strip()
15791 manish.sha 449
            '''
450
            subOrderTrackingParsingUrl = product_details[1].get('href')
451
            #subOrderTrackingUrl = subOrderTrackingParsingUrl.split('order_lookup.details&')[1]
452
            '''
15509 manish.sha 453
            quantity = int(cols[1].text.strip())
454
            sellingPrice = long(cols[2].text.strip().replace("Rs.",""))
15791 manish.sha 455
            discount = 0
456
            subtotal = 0
457
            if totalColumns == 5:
458
                if cols[3].text.strip()!='-' or 'Rs.' in cols[3].text.strip():
459
                    discount = long(cols[3].text.strip().replace("Rs.",""))
460
                else:
461
                    discount = 0
462
                subtotal = long(cols[4].text.strip().replace("Rs.",""))
463
            else:
464
                subtotal = long(cols[3].text.strip().replace("Rs.",""))
465
 
15509 manish.sha 466
            br = getBrowserObject()
467
            productPage = br.open(BASE_MURL+productUrl)
468
            productPageHeaders = str(productPage.info()).split('\n')
469
            productPage = ungzipResponse(productPage)
470
            jsonProductResponse = None
471
            for header in productPageHeaders:
472
                header = header.split(':')
473
                if header[0] == 'Content-Type' and 'json' in header[1]:
474
                    jsonProductResponse = json.loads(productPage)
475
            productPageSoup= None
476
            if jsonProductResponse is not None:
477
                productPageSoup = BeautifulSoup(jsonProductResponse['text'])
478
            else:
479
                productPageSoup = BeautifulSoup(productPage)
15791 manish.sha 480
            productCode = productPageSoup.find('input', {'type':'hidden'})['value']
15509 manish.sha 481
            allproductImageTags = productPageSoup.findAll(attrs={'class' : 'pd-image'})
482
            productImgUrl = ''
483
            if allproductImageTags is not None and len(allproductImageTags)>0:
484
                productImgUrl= allproductImageTags[0].get('style').split("background:url('")[1].split("')no-repeat center")[0].strip()
485
            productDetailsSubMap['productCode']=productCode
486
            productDetailsSubMap['imgUrl']=productImgUrl
15791 manish.sha 487
            '''
16151 manish.sha 488
            orderTrackingPage = fetchResponseUsingProxy(BASE_URL+subOrderTrackingParsingUrl)
489
            orderTrackingPageSoup = BeautifulSoup(orderTrackingPage)
15791 manish.sha 490
            subOrdersDetails = self.parseSingleSubOrder(orderTrackingPageSoup, emailId, subOrderId)
491
            subOrders = list(set(subOrders + subOrdersDetails))
15509 manish.sha 492
 
15791 manish.sha 493
        return subOrders
15509 manish.sha 494
 
495
    def scrapeStoreOrders(self,):
496
        #collectionMap = {'palcedOn':1}
497
        searchMap = {}
498
        collectionMap = {"orderTrackingUrl":1}
499
        orders = self._getActiveOrders(searchMap,collectionMap)
500
        for order in orders:
501
            print "Order", self.store_name, order['orderId'], order['orderTrackingUrl']
16138 manish.sha 502
            url = ORDER_TRACK_URL +'&'+ order['orderTrackingUrl'].split('trackOrder?')[1]
16141 manish.sha 503
            print url
15509 manish.sha 504
            page = fetchResponseUsingProxy(url)
505
            soup = BeautifulSoup(page)
506
            bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
507
            closed = True
16144 manish.sha 508
            orderIdSpan = soup.body.find("span", {'class':'price ord_no'})
509
            if orderIdSpan is not None:
16143 manish.sha 510
                orderStatusList = soup.findAll(attrs={'class' : 'price ord_status'})
511
                if orderStatusList is not None and len(orderStatusList)>0:
512
                    subOrderId = soup.findAll(attrs={'class':'price ord_no'})[0].text.strip()
513
                    orderStatus = orderStatusList[0].contents[0].strip()
514
                    orderTable = soup.body.find("table", {'class':'table product-list'}).findAll('tr', recursive=False)
515
                    orderTable.pop(0)
516
                    count = 1
517
                    while count <= len(orderTable):
16160 manish.sha 518
                        subbulk = self.db.merchantOrder.initialize_ordered_bulk_op()
16152 manish.sha 519
                        print 'Sub Order Id', str(subOrderId)+'-'+str(count)
16143 manish.sha 520
                        subOrder =  self._isSubOrderActive(order, str(subOrderId)+'-'+str(count))
521
                        if subOrder is None:
522
                            try:
523
                                subOrders = self.parseSingleSubOrder(soup, order['orderTrackingUrl'].split('email_id=')[1], subOrderId)
524
                                for subOrder in subOrders:
525
                                    if subOrder is None:
526
                                        continue
527
                                    self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":{"$each":todict([subOrder])}}})
528
                                    print "Added new suborders to Order id - ", order['orderId']
529
                                    closed = False
530
                            except:
531
                                pass
532
                            continue
533
                        elif subOrder['closed']:
534
                            continue
535
                        else:
16163 manish.sha 536
                            findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": str(subOrderId)+'-'+str(count)}
16143 manish.sha 537
                            updateMap = {}
538
                            updateMap["subOrders.$.detailedStatus"] = orderStatus
539
                            status = self._getStatusFromDetailedStatus(orderStatus) 
540
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
541
                            if status is not None:
542
                                updateMap["subOrders.$.status"] = status
543
                            if closedStatus:
544
                                #if status is closed then change the paybackStatus accordingly
16155 manish.sha 545
                                print 'Order Closed'
16143 manish.sha 546
                                updateMap["subOrders.$.closed"] = True
547
                                if status == Store.ORDER_DELIVERED:
548
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
549
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
550
                                elif status == Store.ORDER_CANCELLED:
551
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
552
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
553
                            else:
15791 manish.sha 554
                                closed = False
16155 manish.sha 555
                                print 'Order not Closed'
16161 manish.sha 556
 
16158 manish.sha 557
                            subbulk.find(findMap).update({'$set' : updateMap})
558
                            subresult = subbulk.execute()
559
                            tprint(subresult)
16164 manish.sha 560
                        count = count +1
16156 manish.sha 561
                    bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed,"parseError":False}})
562
                    result = bulk.execute()
563
                    tprint(result)
16143 manish.sha 564
 
565
                else:
566
                    subOrdersList = self.parseMultiSubOrders(soup, order['orderTrackingUrl'].split('email_id=')[1])
567
                    for subOrderObj in subOrdersList:
16159 manish.sha 568
                        subbulk1 = self.db.merchantOrder.initialize_ordered_bulk_op()
16143 manish.sha 569
                        subOrder =  self._isSubOrderActive(order, subOrderObj.merchantSubOrderId)
570
                        if subOrder is None:
571
                            self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":{"$each":todict([subOrderObj])}}})
572
                            print "Added new suborders to Order id - ", order['orderId']
573
                            closed = False
574
                            continue
575
                        elif subOrder['closed']:
576
                            continue
15791 manish.sha 577
                        else:
16143 manish.sha 578
                            findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": subOrderId}
579
                            updateMap = {}
580
                            updateMap["subOrders.$.detailedStatus"] = subOrderObj.detailedStatus
581
                            status = self._getStatusFromDetailedStatus(orderStatus) 
582
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
583
                            if status is not None:
584
                                updateMap["subOrders.$.status"] = status
585
                            if closedStatus:
586
                                #if status is closed then change the paybackStatus accordingly
587
                                updateMap["subOrders.$.closed"] = True
588
                                if status == Store.ORDER_DELIVERED:
589
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
590
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
591
                                elif status == Store.ORDER_CANCELLED:
592
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
593
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
594
                            else:
595
                                closed = False
16159 manish.sha 596
                            subbulk1.find(findMap).update({'$set' : updateMap})
597
                            subresult1 = subbulk1.execute()
598
                            tprint(subresult1)
16156 manish.sha 599
                    bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed,"parseError":False}})
600
                    result = bulk.execute()
601
                    tprint(result)
15509 manish.sha 602
            else:
16143 manish.sha 603
                print 'Soup Object not found for this Order', order['orderId'], order['orderTrackingUrl']
604
                continue      
15791 manish.sha 605
 
606
    '''
607
    def getTrackingUrls(self, userId):
608
        missingOrderUrls = ['https://sm.shopclues.com/myaccount','https://sm.shopclues.com/myorders']
609
        return missingOrderUrls
15509 manish.sha 610
 
15791 manish.sha 611
    def parseMyProfileForEmailId(self, userId, url, rawhtml):
612
        profileSoup = BeautifulSoup(rawhtml)
613
        if profileSoup.find('input', {'name':'user_email'}) is not None:
614
            emailId = profileSoup.find('input', {'name':'user_email'})['value']
615
            if emailId is not None and emailId.strip() !='':
616
                mc = MemCache()
617
                mc.set(str(userId), emailId, 600)
618
                return 'EMAIL_SET_SUCCESS'
619
            else:
620
                return 'EMAIL_NOT_FOUND'
621
        else:
622
            return 'EMAIL_NOT_FOUND'
15509 manish.sha 623
 
15791 manish.sha 624
    def parseMyOrdersForEmailId(self, userId, url, rawhtml):
625
        myOrdersPageSoup = BeautifulSoup(rawhtml)
626
        if myOrdersPageSoup.find('a', {'class':'detail'}) is not None:
627
            emailId = myOrdersPageSoup.find('a', {'class':'detail'})['href'].split('&')[1].split('=')[1]
628
            mc = MemCache()
629
            mc.set(str(userId), emailId, 600)
630
            return 'EMAIL_SET_SUCCESS'
631
        else:
632
            return 'EMAIL_NOT_FOUND'
633
    '''        
634
 
635
 
15509 manish.sha 636
def todict(obj, classkey=None):
637
    if isinstance(obj, dict):
638
        data = {}
639
        for (k, v) in obj.items():
640
            data[k] = todict(v, classkey)
641
        return data
642
    elif hasattr(obj, "_ast"):
643
        return todict(obj._ast())
644
    elif hasattr(obj, "__iter__"):
645
        return [todict(v, classkey) for v in obj]
646
    elif hasattr(obj, "__dict__"):
647
        data = dict([(key, todict(value, classkey)) 
648
            for key, value in obj.__dict__.iteritems() 
649
            if not callable(value) and not key.startswith('_')])
650
        if classkey is not None and hasattr(obj, "__class__"):
651
            data[classkey] = obj.__class__.__name__
652
        return data
653
    else:
15791 manish.sha 654
        return obj    
655
 
15509 manish.sha 656