Subversion Repositories SmartDukaan

Rev

Rev 16235 | Rev 16250 | 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 = {
16232 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'],
16249 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','canceled - address not shippable']
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 = {}
16230 manish.sha 108
        orderTables = soup.body.findAll("table", {'class':'table product-list'})
109
        orderTable = orderTables[len(orderTables)-1].findAll('tr', recursive=False)
15791 manish.sha 110
        firstRow = orderTable.pop(0)
15949 manish.sha 111
        totalColumns = len(firstRow.find_all('th'))
15791 manish.sha 112
        jsonSubOrdersMap = {}
113
        count = 1
114
 
115
        for val in orderObj['0']['items'].values():
116
            newCount = 0
16110 manish.sha 117
            counts = []
15791 manish.sha 118
            for key in jsonSubOrdersMap.keys():
119
                splitKey = key.split('-')
16112 manish.sha 120
                if orderObj['0']['is_parent_order'] == 'N':
121
                    if str(val['order_id']) == splitKey[0]:
122
                        counts.append(int(splitKey[1]))
123
                else:
124
                    if str(val['child']) == splitKey[0]:
125
                        counts.append(int(splitKey[1]))
16110 manish.sha 126
 
127
            if len(counts) >0:
128
                newCount = max(counts)
16108 manish.sha 129
            count = newCount +1
16112 manish.sha 130
            if orderObj['0']['is_parent_order'] == 'N':
131
                jsonSubOrdersMap[str(val['order_id'])+'-'+str(count)] = val
132
            else:
133
                jsonSubOrdersMap[str(val['child'])+'-'+str(count)] = val
15791 manish.sha 134
 
16110 manish.sha 135
 
15941 manish.sha 136
        print jsonSubOrdersMap.items()
137
 
15791 manish.sha 138
        count = 1
15509 manish.sha 139
        for orderTr in orderTable:
15791 manish.sha 140
            productDetailsSubMap = None
141
            productDetailsSubMap = {}
15509 manish.sha 142
            cols = orderTr.find_all('td')
143
            product_details = cols[0].find_all('a')
144
            #print product_details
145
            productUrl = product_details[0].get('href')
146
            productName = product_details[0].contents[0].strip()
147
            quantity = int(cols[1].text.strip())
15980 manish.sha 148
            sellingPrice = long(cols[2].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 149
            discount = 0
150
            subtotal = 0
151
            if totalColumns == 5:
15949 manish.sha 152
                if cols[3].text.strip()!='-' and 'Rs.' in cols[3].text.strip():
15980 manish.sha 153
                    discount = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 154
                else:
155
                    discount = 0
15980 manish.sha 156
                subtotal = long(cols[4].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 157
            else:
15980 manish.sha 158
                subtotal = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15509 manish.sha 159
            productDetailsSubMap['productUrl']=BASE_MURL+productUrl
160
            productDetailsSubMap['productName']=productName
15791 manish.sha 161
            productDetailsSubMap['subOrderTrackingUrl']=ORDER_TRACK_URL_DB+'order_id=' +str(orderId)+'&email_id='+ orderObj['0']['email']
15509 manish.sha 162
            productDetailsSubMap['sellingPrice']=sellingPrice
163
            productDetailsSubMap['quantity']=quantity
164
            productDetailsSubMap['discount']=discount
165
            productDetailsSubMap['subtotal']=subtotal
15942 manish.sha 166
            key = str(orderId)+'-'+str(count)
167
            print 'SubOrder Map Key--',key
15791 manish.sha 168
            jsonSubOrderDetails = jsonSubOrdersMap.get(str(orderId)+'-'+str(count))
169
            productCode = jsonSubOrderDetails['product_code']
170
            productImgUrl = jsonSubOrderDetails['images']['image_path'][0]
15509 manish.sha 171
            productDetailsSubMap['productCode']=productCode
172
            productDetailsSubMap['imgUrl']=productImgUrl
173
 
15791 manish.sha 174
            productDetailsSubMap['subOrderStatus']=orderStatus
175
            productDetailsSubMap['subOrderStatusTime']=statusTime
176
            productDetailsMap[str(orderId)+'-'+str(count)]=productDetailsSubMap
177
            count = count +1
15509 manish.sha 178
        return productDetailsMap
179
 
15791 manish.sha 180
    def _getMultiSubOrdersMap(self, orderId, soup, orderObj):
16230 manish.sha 181
        orderTables = soup.body.findAll("table", {'class':'table product-list'})
182
        orderTable = orderTables[len(orderTables)-1].findAll('tr', recursive=False)
16105 manish.sha 183
        firstRow = orderTable.pop(0)
15791 manish.sha 184
        #totalColumns = len(firstRow.find_all('td'))
15509 manish.sha 185
        productDetailsMap = {}
15791 manish.sha 186
        '''
187
        jsonSubOrdersMap = {}
188
        for val in orderObj['0']['items'].values():
189
            jsonSubOrdersMap[val['order_id']] = val
190
        '''
191
        existingOrders = []
15509 manish.sha 192
        for orderTr in orderTable:
15791 manish.sha 193
            subOrderDetailsMap = {}
194
 
15509 manish.sha 195
            cols = orderTr.find_all('td')
196
            product_details = cols[0].find_all('a')
197
            #print product_details
198
            subOrderId= product_details[1].contents[0].strip()
15791 manish.sha 199
            if subOrderId in existingOrders:
200
                continue
201
            else:
202
                existingOrders.append(subOrderId)
203
            #productUrl = product_details[0].get('href')
204
            #productName = product_details[0].contents[0].strip()
205
            subOrderTrackingParsingUrl = product_details[1].get('href')
206
            #subOrderTrackingUrl = subOrderTrackingParsingUrl.split('order_lookup.details&')[1]
207
            '''
15509 manish.sha 208
            quantity = int(cols[1].text.strip())
209
            sellingPrice = long(cols[2].text.strip().replace("Rs.",""))
15791 manish.sha 210
            discount = 0
211
            subtotal = 0
212
            if totalColumns == 5:
213
                if cols[3].text.strip()!='-' or 'Rs.' in cols[3].text.strip():
214
                    discount = long(cols[3].text.strip().replace("Rs.",""))
215
                else:
216
                    discount = 0                    
217
                subtotal = long(cols[4].text.strip().replace("Rs.",""))
218
            else:
219
                subtotal = long(cols[3].text.strip().replace("Rs.",""))
15509 manish.sha 220
            productDetailsSubMap['productUrl']=BASE_MURL+productUrl
221
            productDetailsSubMap['productName']=productName
15791 manish.sha 222
            productDetailsSubMap['subOrderTrackingUrl']=ORDER_TRACK_URL_DB+subOrderTrackingUrl
15509 manish.sha 223
            productDetailsSubMap['sellingPrice']=sellingPrice
224
            productDetailsSubMap['quantity']=quantity
225
            productDetailsSubMap['discount']=discount
226
            productDetailsSubMap['subtotal']=subtotal
15791 manish.sha 227
            jsonSubOrderDetails = jsonSubOrdersMap.get(subOrderId)
228
            productCode = jsonSubOrderDetails['product_code']
229
            productImgUrl = jsonSubOrderDetails['images']['image_path'][0]
15509 manish.sha 230
            productDetailsSubMap['productCode']=productCode
231
            productDetailsSubMap['imgUrl']=productImgUrl
15791 manish.sha 232
            '''
16151 manish.sha 233
            orderTrackingPage = fetchResponseUsingProxy(BASE_URL+subOrderTrackingParsingUrl)
234
            orderTrackingPageSoup = BeautifulSoup(orderTrackingPage)
15791 manish.sha 235
            '''
15509 manish.sha 236
            subOrderStatusList = orderTrackingPageSoup.findAll(attrs={'class' : 'price ord_status'})
237
            subOrderStatus = subOrderStatusList[0].contents[0].strip()
238
            subOrderStatusTime= orderTrackingPageSoup.findAll(attrs={'class' : 'sts no_mobile'})[1].contents[2].strip()
239
            productDetailsSubMap['subOrderStatus'] = subOrderStatus
240
            productDetailsSubMap['subOrderStatusTime'] = subOrderStatusTime
241
            productDetailsSubMap['parentOrderId']=orderId
15791 manish.sha 242
            '''
243
 
244
            subOrderDetailsMap = self._getSingleSubOrderMap(subOrderId, orderTrackingPageSoup, orderObj)
15509 manish.sha 245
 
15791 manish.sha 246
            productDetailsMap = dict(productDetailsMap.items()+subOrderDetailsMap.items())
15509 manish.sha 247
 
248
        return productDetailsMap
249
 
250
    def updateCashbackInSubOrders(self, subOrders):
251
        for subOrder in subOrders:
252
            cashbackStatus = Store.CB_NA
253
            cashbackAmount = 0
254
            percentage = 0
255
            amount = subOrder.amountPaid
256
            if amount > 0:
257
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amount)
258
                if cashbackAmount > 0:
259
                    cashbackStatus = Store.CB_PENDING
260
            subOrder.cashBackStatus = cashbackStatus
261
            subOrder.cashBackAmount = cashbackAmount
262
            subOrder.cashBackPercentage = percentage
263
        return subOrders
264
 
15947 manish.sha 265
    def _parseOrders(self, orderId, mpOrderId, subTagId, userId, page, orderSuccessUrl, orderObj):
15509 manish.sha 266
        soup = BeautifulSoup(page)
267
        productDetailsMap = {}
268
        paymentFields = soup.findAll(attrs={'class' : 'box_paymentcalculations_row'})
269
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
15948 manish.sha 270
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("\r","").replace("  ","")
15509 manish.sha 271
        ordersubtotal=0
272
        ordercluebucks=0
273
        ordershippingcost=0
274
        ordertotal=0
275
 
276
        for val in paymentFields:
277
            for value in val.contents:
278
                if value is not None:
279
                    if 'div' in str(value).strip():
280
                        if 'Subtotal' in value.text.strip():
15981 manish.sha 281
                            ordersubtotal = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 282
                            print 'Order SubTotal:- ',ordersubtotal
15509 manish.sha 283
                        if 'Shipping Cost' in value.text.strip():
15981 manish.sha 284
                            ordershippingcost = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 285
                            print 'Shipping Cost:- ',ordershippingcost
15509 manish.sha 286
                        if 'Clue' in value.text.strip():
15981 manish.sha 287
                            ordercluebucks = long(val.contents[3].text.strip().replace("Rs.","").replace(',',''))
16165 manish.sha 288
                            print 'Clue bucks:- ',ordercluebucks
15509 manish.sha 289
                        if 'Total' in value.text.strip():
15791 manish.sha 290
                            ordertotal = val.contents[3].text.strip().replace("Rs.","")
291
                            ordertotal = ordertotal.replace(',','')
16165 manish.sha 292
                            print 'Order Total:- ',ordertotal
15791 manish.sha 293
 
294
        if orderObj['0']['is_parent_order'] == 'N':
15947 manish.sha 295
            productDetailsMap = self._getSingleSubOrderMap(mpOrderId, soup, orderObj)
15509 manish.sha 296
        else:
15947 manish.sha 297
            productDetailsMap = self._getMultiSubOrdersMap(mpOrderId, soup, orderObj)
15509 manish.sha 298
 
15791 manish.sha 299
        merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
15509 manish.sha 300
        merchantOrder.placedOn = placedOn
15791 manish.sha 301
        merchantOrder.merchantOrderId = orderObj['0']['order_id']
15509 manish.sha 302
        merchantOrder.paidAmount = ordertotal
303
        totalOrdersAmount = 0
304
        totalDiscount = 0
305
        subOrders= []
306
        for key in productDetailsMap:
307
            subOrderDetail = productDetailsMap.get(key)
308
            totalOrdersAmount = totalOrdersAmount  + (subOrderDetail['sellingPrice'] * subOrderDetail['quantity'])
309
            totalDiscount = totalDiscount + (subOrderDetail['discount'] * subOrderDetail['quantity'])
310
            subOrder = SubOrder(subOrderDetail['productName'], subOrderDetail['productUrl'], placedOn, subOrderDetail['subtotal'])
311
            subOrder.merchantSubOrderId = key
312
            subOrder.detailedStatus = subOrderDetail['subOrderStatus']
313
            subOrder.imgUrl = subOrderDetail['imgUrl']
314
            subOrder.offerDiscount = subOrderDetail['discount'] * subOrderDetail['quantity']
315
            subOrder.unitPrice = subOrderDetail['sellingPrice']
316
            subOrder.productCode = subOrderDetail['productCode']
317
            subOrder.amountPaid = subOrderDetail['subtotal']
318
            subOrder.quantity = subOrderDetail['quantity']
15791 manish.sha 319
            subOrder.tracingkUrl = subOrderDetail['subOrderTrackingUrl']
16097 manish.sha 320
            dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
321
            subOrder.dealRank = dealRank.get('rank')
322
            subOrder.rankDesc = dealRank.get('description')
15509 manish.sha 323
            subOrders.append(subOrder)
324
        merchantOrder.totalAmount = totalOrdersAmount
325
        merchantOrder.discountApplied = totalDiscount
326
        merchantOrder.deliveryCharges = ordershippingcost
327
        merchantOrder.subOrders = self.updateCashbackInSubOrders(subOrders)
328
 
329
        return merchantOrder
330
 
331
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl):
332
        resp = {}
333
        try:
15791 manish.sha 334
            rawHtmlSoup = BeautifulSoup(rawHtml)
335
            emailId = None
336
            orderObj = None
337
            merchantOrderId = None
338
            for script in rawHtmlSoup.find_all('script'):
339
                if 'orderConfirmation' in script.text:
340
                    orderObj = json.loads(script.text.strip().split('\n')[0].split('orderConfirmation = ')[1].split(';')[0])
341
                    if merchantOrderId is None:
342
                        merchantOrderId = orderObj['0']['order_id']
343
                        emailId = orderObj['0']['email']
16197 manish.sha 344
            if orderObj is None or merchantOrderId is None:
16226 manish.sha 345
                orderIdDiv = rawHtmlSoup.body.find("div", {'class':'conf_succes'})
346
                if orderIdDiv is None:
347
                    orderIdArt = rawHtmlSoup.body.find("article", {'class':'white'})
348
                    if orderIdArt is None:
349
                        resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
350
                        return resp
351
                    else:
352
                        orderIdDiv= orderIdArt.find('p', recursive=False)
353
                        orderIdVal = str(orderIdDiv.text.split(' : ')[1])
354
                        print "Opening Shopclues Login Page"
355
                        login_url = "https://sm.shopclues.com/login"
356
                        br1 = login(login_url)
357
                        orderDetailPage = br1.open("https://sm.shopclues.com/orderconfirmation?order_id="+orderIdVal+"&status=P")
358
                        orderDetailPage = ungzipResponse(orderDetailPage)
359
                        rawHtmlSoup = BeautifulSoup(orderDetailPage)
360
                        for script in rawHtmlSoup.find_all('script'):
361
                            if 'orderConfirmation' in script.text:
362
                                orderObj = json.loads(script.text.strip().split('\n')[0].split('orderConfirmation = ')[1].split(';')[0])
363
                                if merchantOrderId is None:
364
                                    merchantOrderId = orderObj['0']['order_id']
365
                                    emailId = orderObj['0']['email']
366
                        logoutpage = br1.open("https://sm.shopclues.com/logout")
367
                        print br1.geturl()
368
                else:
369
                    orderIdDiv= orderIdDiv.find('div', recursive=False)
370
                    orderIdVal = str(orderIdDiv.span.text)
371
                    print "Opening Shopclues Login Page"
372
                    login_url = "https://sm.shopclues.com/login"
373
                    br1 = login(login_url)
374
                    orderDetailPage = br1.open("https://sm.shopclues.com/orderconfirmation?order_id="+orderIdVal+"&status=P")
375
                    orderDetailPage = ungzipResponse(orderDetailPage)
376
                    rawHtmlSoup = BeautifulSoup(orderDetailPage)
377
                    for script in rawHtmlSoup.find_all('script'):
378
                        if 'orderConfirmation' in script.text:
379
                            orderObj = json.loads(script.text.strip().split('\n')[0].split('orderConfirmation = ')[1].split(';')[0])
380
                            if merchantOrderId is None:
381
                                merchantOrderId = orderObj['0']['order_id']
382
                                emailId = orderObj['0']['email']
383
                    logoutpage = br1.open("https://sm.shopclues.com/logout")
384
                    print br1.geturl()
385
 
15509 manish.sha 386
            br = getBrowserObject()
15791 manish.sha 387
            url = ORDER_TRACK_URL +'&order_id=' +str(merchantOrderId)+'&email_id='+ emailId
15509 manish.sha 388
            page = br.open(url)
389
            headers = str(page.info()).split('\n')
390
            page = ungzipResponse(page)
391
            jsonResponse = None
392
            for header in headers:
393
                header = header.split(':')
394
                if header[0] == 'Content-Type' and 'json' in header[1]:
395
                    jsonResponse = json.loads(page)
396
            if jsonResponse is not None:
397
                page = jsonResponse['text']
398
 
15947 manish.sha 399
            merchantOrder = self._parseOrders(orderId, merchantOrderId, subTagId, userId, page, orderSuccessUrl, orderObj)
15509 manish.sha 400
 
15948 manish.sha 401
            merchantOrder.orderTrackingUrl = ORDER_TRACK_URL_DB + 'order_id=' +str(merchantOrderId)+'&email_id='+ emailId
15509 manish.sha 402
 
403
            if self._saveToOrder(todict(merchantOrder)):
404
                resp['result'] = 'ORDER_CREATED'
405
            else:
406
                resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
407
 
408
            return resp
409
        except:
410
            print "Error occurred"
411
            traceback.print_exc()
412
            resp['result'] = 'ORDER_NOT_CREATED'
16120 manish.sha 413
            return resp
15509 manish.sha 414
 
15791 manish.sha 415
    def parseSingleSubOrder(self, soup, emailId, subOrderId):
416
        orderStatusList = soup.findAll(attrs={'class' : 'price ord_status'})
417
        subOrderStatus = orderStatusList[0].contents[0].strip()
15509 manish.sha 418
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
16230 manish.sha 419
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("\r","").replace("  ","")
420
        orderTables = soup.body.findAll("table", {'class':'table product-list'})
421
        orderTable = orderTables[len(orderTables)-1].findAll('tr', recursive=False)    
15791 manish.sha 422
        firstRow = orderTable.pop(0)
15949 manish.sha 423
        totalColumns = len(firstRow.find_all('th'))
15791 manish.sha 424
        subOrders =[]
425
        count =1   
15509 manish.sha 426
        for orderTr in orderTable:
427
            cols = orderTr.find_all('td')
428
            product_details = cols[0].find_all('a')
429
            #print product_details
430
            productUrl = product_details[0].get('href')
431
            productName = product_details[0].contents[0].strip()
432
            quantity = int(cols[1].text.strip())
15980 manish.sha 433
            sellingPrice = long(cols[2].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 434
            discount = 0
435
            subtotal = 0
436
            if totalColumns == 5:
15949 manish.sha 437
                if cols[3].text.strip()!='-' and 'Rs.' in cols[3].text.strip():
15980 manish.sha 438
                    discount = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 439
                else:
440
                    discount = 0
15980 manish.sha 441
                subtotal = long(cols[4].text.strip().replace("Rs.","").replace(',',''))
15791 manish.sha 442
            else:
15980 manish.sha 443
                subtotal = long(cols[3].text.strip().replace("Rs.","").replace(',',''))
16151 manish.sha 444
            productPage = fetchResponseUsingProxy(BASE_MURL+productUrl)
445
            productPageSoup = BeautifulSoup(productPage)
15791 manish.sha 446
            productCode = productPageSoup.find('input', {'type':'hidden'})['value']
15509 manish.sha 447
            allproductImageTags = productPageSoup.findAll(attrs={'class' : 'pd-image'})
448
            productImgUrl = ''
449
            if allproductImageTags is not None and len(allproductImageTags)>0:
450
                productImgUrl= allproductImageTags[0].get('style').split("background:url('")[1].split("')no-repeat center")[0].strip()
451
 
452
            subOrder = SubOrder(productName, productUrl, placedOn, subtotal)
15791 manish.sha 453
            subOrder.merchantSubOrderId = str(subOrderId)+'-'+str(count)
15509 manish.sha 454
            subOrder.detailedStatus = subOrderStatus
455
            subOrder.imgUrl = productImgUrl
456
            subOrder.offerDiscount = discount*quantity
457
            subOrder.unitPrice = sellingPrice
458
            subOrder.productCode = productCode
459
            subOrder.amountPaid = subtotal
460
            subOrder.quantity = quantity
15791 manish.sha 461
            subOrder.tracingkUrl = ORDER_TRACK_URL_DB +  'order_id=' +subOrderId+'&email_id='+ emailId
15509 manish.sha 462
            subOrders.append(subOrder)
15791 manish.sha 463
            count = count +1
15509 manish.sha 464
 
465
        subOrders = self.updateCashbackInSubOrders(subOrders)
15791 manish.sha 466
        return subOrders
15509 manish.sha 467
 
15791 manish.sha 468
    def parseMultiSubOrders(self, soup, emailId):
16230 manish.sha 469
        orderTables = soup.body.findAll("table", {'class':'table product-list'})
470
        orderTable = orderTables[len(orderTables)-1].findAll('tr', recursive=False)
16145 manish.sha 471
        firstRow = orderTable.pop(0)
15791 manish.sha 472
        '''
473
        totalColumns = len(firstRow.find_all('td'))
474
 
15509 manish.sha 475
        orderDateList = soup.findAll(attrs={'class':'price ord_date'})
15791 manish.sha 476
        placedOn= orderDateList[0].text.strip().replace("\t","").replace("\n","").replace("  ","")
477
        '''    
15509 manish.sha 478
        subOrders = []
15791 manish.sha 479
        existingOrders = []
15509 manish.sha 480
        for orderTr in orderTable:
481
            cols = orderTr.find_all('td')
482
            product_details = cols[0].find_all('a')
483
            #print product_details
484
            subOrderId= product_details[1].contents[0].strip()
15791 manish.sha 485
            if subOrderId in existingOrders:
486
                continue
487
            else:
488
                existingOrders.append(subOrderId)
489
            '''
15509 manish.sha 490
            productUrl = product_details[0].get('href')
491
            productName = product_details[0].contents[0].strip()
15791 manish.sha 492
            '''
493
            subOrderTrackingParsingUrl = product_details[1].get('href')
494
            #subOrderTrackingUrl = subOrderTrackingParsingUrl.split('order_lookup.details&')[1]
495
            '''
15509 manish.sha 496
            quantity = int(cols[1].text.strip())
497
            sellingPrice = long(cols[2].text.strip().replace("Rs.",""))
15791 manish.sha 498
            discount = 0
499
            subtotal = 0
500
            if totalColumns == 5:
501
                if cols[3].text.strip()!='-' or 'Rs.' in cols[3].text.strip():
502
                    discount = long(cols[3].text.strip().replace("Rs.",""))
503
                else:
504
                    discount = 0
505
                subtotal = long(cols[4].text.strip().replace("Rs.",""))
506
            else:
507
                subtotal = long(cols[3].text.strip().replace("Rs.",""))
508
 
15509 manish.sha 509
            br = getBrowserObject()
510
            productPage = br.open(BASE_MURL+productUrl)
511
            productPageHeaders = str(productPage.info()).split('\n')
512
            productPage = ungzipResponse(productPage)
513
            jsonProductResponse = None
514
            for header in productPageHeaders:
515
                header = header.split(':')
516
                if header[0] == 'Content-Type' and 'json' in header[1]:
517
                    jsonProductResponse = json.loads(productPage)
518
            productPageSoup= None
519
            if jsonProductResponse is not None:
520
                productPageSoup = BeautifulSoup(jsonProductResponse['text'])
521
            else:
522
                productPageSoup = BeautifulSoup(productPage)
15791 manish.sha 523
            productCode = productPageSoup.find('input', {'type':'hidden'})['value']
15509 manish.sha 524
            allproductImageTags = productPageSoup.findAll(attrs={'class' : 'pd-image'})
525
            productImgUrl = ''
526
            if allproductImageTags is not None and len(allproductImageTags)>0:
527
                productImgUrl= allproductImageTags[0].get('style').split("background:url('")[1].split("')no-repeat center")[0].strip()
528
            productDetailsSubMap['productCode']=productCode
529
            productDetailsSubMap['imgUrl']=productImgUrl
15791 manish.sha 530
            '''
16151 manish.sha 531
            orderTrackingPage = fetchResponseUsingProxy(BASE_URL+subOrderTrackingParsingUrl)
532
            orderTrackingPageSoup = BeautifulSoup(orderTrackingPage)
15791 manish.sha 533
            subOrdersDetails = self.parseSingleSubOrder(orderTrackingPageSoup, emailId, subOrderId)
534
            subOrders = list(set(subOrders + subOrdersDetails))
15509 manish.sha 535
 
15791 manish.sha 536
        return subOrders
15509 manish.sha 537
 
538
    def scrapeStoreOrders(self,):
539
        #collectionMap = {'palcedOn':1}
540
        searchMap = {}
541
        collectionMap = {"orderTrackingUrl":1}
542
        orders = self._getActiveOrders(searchMap,collectionMap)
543
        for order in orders:
544
            print "Order", self.store_name, order['orderId'], order['orderTrackingUrl']
16138 manish.sha 545
            url = ORDER_TRACK_URL +'&'+ order['orderTrackingUrl'].split('trackOrder?')[1]
16141 manish.sha 546
            print url
16249 manish.sha 547
            page = None
548
            retry = 1
549
            while retry <=3: 
550
                try: 
551
                    page = fetchResponseUsingProxy(url)
552
                    break
553
                except:
554
                    retry = retry + 1
555
 
15509 manish.sha 556
            soup = BeautifulSoup(page)
557
            bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
558
            closed = True
16144 manish.sha 559
            orderIdSpan = soup.body.find("span", {'class':'price ord_no'})
560
            if orderIdSpan is not None:
16143 manish.sha 561
                orderStatusList = soup.findAll(attrs={'class' : 'price ord_status'})
562
                if orderStatusList is not None and len(orderStatusList)>0:
563
                    subOrderId = soup.findAll(attrs={'class':'price ord_no'})[0].text.strip()
564
                    orderStatus = orderStatusList[0].contents[0].strip()
16230 manish.sha 565
                    orderTables = soup.body.findAll("table", {'class':'table product-list'})
16229 manish.sha 566
                    orderTable = orderTables[len(orderTables)-1].findAll('tr', recursive=False)
16143 manish.sha 567
                    orderTable.pop(0)
568
                    count = 1
569
                    while count <= len(orderTable):
16160 manish.sha 570
                        subbulk = self.db.merchantOrder.initialize_ordered_bulk_op()
16152 manish.sha 571
                        print 'Sub Order Id', str(subOrderId)+'-'+str(count)
16143 manish.sha 572
                        subOrder =  self._isSubOrderActive(order, str(subOrderId)+'-'+str(count))
573
                        if subOrder is None:
574
                            try:
575
                                subOrders = self.parseSingleSubOrder(soup, order['orderTrackingUrl'].split('email_id=')[1], subOrderId)
576
                                for subOrder in subOrders:
577
                                    if subOrder is None:
578
                                        continue
579
                                    self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":{"$each":todict([subOrder])}}})
580
                                    print "Added new suborders to Order id - ", order['orderId']
581
                                    closed = False
582
                            except:
16227 manish.sha 583
                                traceback.print_exc()
16143 manish.sha 584
                                pass
16227 manish.sha 585
                            count = count+1
16143 manish.sha 586
                            continue
587
                        elif subOrder['closed']:
16226 manish.sha 588
                            count = count+1
16143 manish.sha 589
                            continue
590
                        else:
16163 manish.sha 591
                            findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": str(subOrderId)+'-'+str(count)}
16143 manish.sha 592
                            updateMap = {}
593
                            updateMap["subOrders.$.detailedStatus"] = orderStatus
594
                            status = self._getStatusFromDetailedStatus(orderStatus) 
595
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
596
                            if status is not None:
597
                                updateMap["subOrders.$.status"] = status
598
                            if closedStatus:
599
                                #if status is closed then change the paybackStatus accordingly
16155 manish.sha 600
                                print 'Order Closed'
16143 manish.sha 601
                                updateMap["subOrders.$.closed"] = True
602
                                if status == Store.ORDER_DELIVERED:
603
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
604
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
605
                                elif status == Store.ORDER_CANCELLED:
606
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
607
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
608
                            else:
15791 manish.sha 609
                                closed = False
16155 manish.sha 610
                                print 'Order not Closed'
16161 manish.sha 611
 
16158 manish.sha 612
                            subbulk.find(findMap).update({'$set' : updateMap})
613
                            subresult = subbulk.execute()
614
                            tprint(subresult)
16164 manish.sha 615
                        count = count +1
16156 manish.sha 616
                    bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed,"parseError":False}})
617
                    result = bulk.execute()
618
                    tprint(result)
16143 manish.sha 619
 
620
                else:
621
                    subOrdersList = self.parseMultiSubOrders(soup, order['orderTrackingUrl'].split('email_id=')[1])
622
                    for subOrderObj in subOrdersList:
16159 manish.sha 623
                        subbulk1 = self.db.merchantOrder.initialize_ordered_bulk_op()
16143 manish.sha 624
                        subOrder =  self._isSubOrderActive(order, subOrderObj.merchantSubOrderId)
625
                        if subOrder is None:
626
                            self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":{"$each":todict([subOrderObj])}}})
627
                            print "Added new suborders to Order id - ", order['orderId']
628
                            closed = False
629
                            continue
630
                        elif subOrder['closed']:
631
                            continue
15791 manish.sha 632
                        else:
16143 manish.sha 633
                            findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": subOrderId}
634
                            updateMap = {}
635
                            updateMap["subOrders.$.detailedStatus"] = subOrderObj.detailedStatus
636
                            status = self._getStatusFromDetailedStatus(orderStatus) 
637
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
638
                            if status is not None:
639
                                updateMap["subOrders.$.status"] = status
640
                            if closedStatus:
641
                                #if status is closed then change the paybackStatus accordingly
642
                                updateMap["subOrders.$.closed"] = True
643
                                if status == Store.ORDER_DELIVERED:
644
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
645
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
646
                                elif status == Store.ORDER_CANCELLED:
647
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
648
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
649
                            else:
650
                                closed = False
16159 manish.sha 651
                            subbulk1.find(findMap).update({'$set' : updateMap})
652
                            subresult1 = subbulk1.execute()
653
                            tprint(subresult1)
16156 manish.sha 654
                    bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed,"parseError":False}})
655
                    result = bulk.execute()
656
                    tprint(result)
15509 manish.sha 657
            else:
16143 manish.sha 658
                print 'Soup Object not found for this Order', order['orderId'], order['orderTrackingUrl']
659
                continue      
15791 manish.sha 660
 
661
    '''
662
    def getTrackingUrls(self, userId):
663
        missingOrderUrls = ['https://sm.shopclues.com/myaccount','https://sm.shopclues.com/myorders']
664
        return missingOrderUrls
15509 manish.sha 665
 
15791 manish.sha 666
    def parseMyProfileForEmailId(self, userId, url, rawhtml):
667
        profileSoup = BeautifulSoup(rawhtml)
668
        if profileSoup.find('input', {'name':'user_email'}) is not None:
669
            emailId = profileSoup.find('input', {'name':'user_email'})['value']
670
            if emailId is not None and emailId.strip() !='':
671
                mc = MemCache()
672
                mc.set(str(userId), emailId, 600)
673
                return 'EMAIL_SET_SUCCESS'
674
            else:
675
                return 'EMAIL_NOT_FOUND'
676
        else:
677
            return 'EMAIL_NOT_FOUND'
15509 manish.sha 678
 
15791 manish.sha 679
    def parseMyOrdersForEmailId(self, userId, url, rawhtml):
680
        myOrdersPageSoup = BeautifulSoup(rawhtml)
681
        if myOrdersPageSoup.find('a', {'class':'detail'}) is not None:
682
            emailId = myOrdersPageSoup.find('a', {'class':'detail'})['href'].split('&')[1].split('=')[1]
683
            mc = MemCache()
684
            mc.set(str(userId), emailId, 600)
685
            return 'EMAIL_SET_SUCCESS'
686
        else:
687
            return 'EMAIL_NOT_FOUND'
688
    '''        
689
 
690
 
15509 manish.sha 691
def todict(obj, classkey=None):
692
    if isinstance(obj, dict):
693
        data = {}
694
        for (k, v) in obj.items():
695
            data[k] = todict(v, classkey)
696
        return data
697
    elif hasattr(obj, "_ast"):
698
        return todict(obj._ast())
699
    elif hasattr(obj, "__iter__"):
700
        return [todict(v, classkey) for v in obj]
701
    elif hasattr(obj, "__dict__"):
702
        data = dict([(key, todict(value, classkey)) 
703
            for key, value in obj.__dict__.iteritems() 
704
            if not callable(value) and not key.startswith('_')])
705
        if classkey is not None and hasattr(obj, "__class__"):
706
            data[classkey] = obj.__class__.__name__
707
        return data
708
    else:
16226 manish.sha 709
        return obj
710
 
711
def login(url):
712
    br = getBrowserObject()
16235 manish.sha 713
    br.set_proxies({"http": "us-il.proxymesh.com:31280"})
16226 manish.sha 714
    br.open(url)
715
    response = br.open(url)
716
    ungzipResponseBr(response, br)
717
    #html = response.read()
718
    #print html
719
    br.select_form(nr=0)
720
    br.form['user'] = "imanthetester@gmail.com"
721
    br.form['password'] = "$Dl8uk"
722
    response = br.submit()
723
    print "********************"
724
    print "Attempting to Login"
725
    print "********************"
726
    #ungzipResponse(response, br)
727
    return br
15791 manish.sha 728
 
16226 manish.sha 729
def ungzipResponseBr(r,b):
730
    headers = r.info()
731
    if headers['Content-Encoding']=='gzip':
732
        import gzip
733
        print "********************"
734
        print "Deflating gzip response"
735
        print "********************"
736
        gz = gzip.GzipFile(fileobj=r, mode='rb')
737
        html = gz.read()
738
        gz.close()
739
        headers["Content-type"] = "text/html; charset=utf-8"
740
        r.set_data( html )
741
        b.set_response(r)    
742
 
15509 manish.sha 743