Subversion Repositories SmartDukaan

Rev

Rev 15569 | Rev 16283 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
13774 amit.gupta 1
# coding=utf-8
13569 amit.gupta 2
'''
3
Created on Jan 15, 2015
4
 
5
@author: amit
6
'''
13774 amit.gupta 7
from base64 import encode
8
from bs4 import BeautifulSoup
14650 amit.gupta 9
from datetime import datetime, timedelta
10
from dtr.dao import Order, SubOrder, AmazonAffiliateInfo
14464 amit.gupta 11
from dtr.main import getStore, Store as MStore, ParseException, getBrowserObject, \
14624 amit.gupta 12
    ungzipResponse, tprint
14650 amit.gupta 13
from dtr.sources.flipkart import todict, AFF_REPORT_URL
14749 amit.gupta 14
from dtr.utils.utils import fetchResponseUsingProxy
14162 amit.gupta 15
from paramiko import sftp
16
from paramiko.client import SSHClient
17
from paramiko.sftp_client import SFTPClient
14650 amit.gupta 18
import base64
19
import gzip
14624 amit.gupta 20
import mechanize
14285 amit.gupta 21
import os.path
14162 amit.gupta 22
import paramiko
13774 amit.gupta 23
import re
14033 amit.gupta 24
import time
13809 amit.gupta 25
import traceback
14650 amit.gupta 26
import urllib2
15346 amit.gupta 27
from dtr.storage.Mongo import getDealRank
15555 amit.gupta 28
from dtr.storage.DataService import OrdersRaw
29
from elixir import *
13774 amit.gupta 30
 
31
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 32
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 33
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
14624 amit.gupta 34
AMAZON_AFF_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/listReports'
14650 amit.gupta 35
AMAZON_AFF_FILE_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/getReport?filename=saholic-21-orders-report-%s.tsv.gz'
13774 amit.gupta 36
class Store(MStore):
13821 amit.gupta 37
 
14032 amit.gupta 38
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch'],
14064 amit.gupta 39
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
14032 amit.gupta 40
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled'],
41
                            MStore.ORDER_DELIVERED : ['delivered']
13821 amit.gupta 42
                           }
13774 amit.gupta 43
 
44
    def __init__(self,store_id):
45
        super(Store, self).__init__(store_id)
46
 
47
    def getName(self):
14749 amit.gupta 48
        return "amazon"
13774 amit.gupta 49
 
14945 amit.gupta 50
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, track=False):
14958 amit.gupta 51
        parseString = "Tracking" if track else "Transacted"
52
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 53
        resp = {}
14813 amit.gupta 54
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 55
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 56
            try:
57
                soup = BeautifulSoup(rawHtml)
14608 amit.gupta 58
                try:
59
                    orderUrl = soup.find('div', {"id":"thank-you-box-wrapper"}).div.findAll('div', recursive=False)[1].a['href']
60
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
61
                except:
62
                    merchantOrderId = soup.find(id="orders-list").div.span.b.text
14699 amit.gupta 63
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 64
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
65
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 66
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 67
                order.requireDetail = True
14698 amit.gupta 68
                order.status = 'html_required'
14297 amit.gupta 69
                order.closed = None
14312 amit.gupta 70
                if self._saveToOrder(todict(order)):
71
                    resp['result'] = 'ORDER_CREATED'
72
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
73
                    resp["htmlRequired"] = True
74
                    resp['orderId'] = orderId
75
                else:
76
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
77
 
13774 amit.gupta 78
            except:
14813 amit.gupta 79
                #Write all cases here for Order Not created Known
14809 amit.gupta 80
                try:
14815 amit.gupta 81
                    if 'Securely redirecting you' in soup.h3.text.strip():
14813 amit.gupta 82
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 83
                    else:
14813 amit.gupta 84
                        raise
14809 amit.gupta 85
                except:
14814 amit.gupta 86
                    try:
14815 amit.gupta 87
                        if soup.h1.text.strip() in ['This is a duplicate order', 'There was a problem with your payment.']:
14814 amit.gupta 88
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
89
                        else:
90
                            raise
91
                    except:
15046 amit.gupta 92
                        try:
93
                            if soup.h2.text.strip() == 'Web page not available':
94
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
95
                            else:
96
                                raise
97
                        except:
15565 amit.gupta 98
                            try:
15569 amit.gupta 99
                                if soup.find(id="loading-spinner-img") is not None or soup.find(id="anonCarousel1") is not None or soup.find(id="ap_signin_pagelet_title") is not None:
15565 amit.gupta 100
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
101
                                else:
102
                                    raise
103
                            except:
104
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
105
 
13774 amit.gupta 106
        else:
13781 amit.gupta 107
            try:
14722 amit.gupta 108
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
109
                if mo is not None:
14749 amit.gupta 110
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 111
                    merchantOrder.createdOn = mo.get("createdOn")
112
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
113
                else:
14945 amit.gupta 114
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 115
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14464 amit.gupta 116
                soup = BeautifulSoup(rawHtml)
117
                try:
14961 amit.gupta 118
                    self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 119
                    resp['result'] = 'DETAIL_CREATED'
14464 amit.gupta 120
                except:
14545 amit.gupta 121
                    try:
14566 amit.gupta 122
                        traceback.print_exc()
14961 amit.gupta 123
                        self.parseOldStlye(merchantOrder, soup)
15046 amit.gupta 124
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 125
                    except:
14566 amit.gupta 126
                        traceback.print_exc()
15056 amit.gupta 127
                        try:
128
                            self.parseCancelled(merchantOrder, soup)
129
                            resp['result'] = 'ORDER_CANCELLED'
130
                        except:
15059 amit.gupta 131
                            try:
15570 amit.gupta 132
                                if soup.h1.span.text=="Account" or soup.h1=="Your Account":
15059 amit.gupta 133
                                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
134
                                else:
135
                                    raise
136
                            except:
137
                                if soup.find(id="ap_signin_pagelet_title").h1.text.strip()=="Sign In":
138
                                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
139
                                else:
140
                                    raise
13781 amit.gupta 141
            except:
14813 amit.gupta 142
                order = self.db.merchantOrder.find_one({"orderId":orderId})
143
                if order is not None:
144
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"parse_failed"}})
145
                print "Error occurred"
15046 amit.gupta 146
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 147
                traceback.print_exc()
15046 amit.gupta 148
        return resp    
13774 amit.gupta 149
 
13868 amit.gupta 150
    #This should be exposed from api for specific sources
151
    def scrapeStoreOrders(self):
14624 amit.gupta 152
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
153
        for merchantOrder in orders:
14981 amit.gupta 154
            executeBulk = False
14624 amit.gupta 155
            try:
156
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
157
                closed = True
158
                map1 = {}
159
                for subOrder in merchantOrder.get("subOrders"):
160
                    if subOrder.get("closed"):
161
                        continue
162
                    elif subOrder.get("trackingUrl") is None:
163
                        closed = False
164
                        continue
14981 amit.gupta 165
                    executeBulk = True
14624 amit.gupta 166
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
167
                    trackingUrl = subOrder.get("trackingUrl")
168
                    if not map1.has_key(trackingUrl):
14749 amit.gupta 169
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl)
14624 amit.gupta 170
                    newOrder = map1.get(trackingUrl)
171
                    newOrder['cashBackStatus'] = subOrder.get('cashBackStatus')
172
                    updateMap = self.getUpdateMap(newOrder)
173
                    print findMap, "\n", updateMap
174
                    bulk.find(findMap).update({'$set' : updateMap})
175
                    closed = closed and newOrder['closed']
14981 amit.gupta 176
                if executeBulk:
14697 amit.gupta 177
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 178
                    bulk.execute()
14624 amit.gupta 179
            except:
14847 amit.gupta 180
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 181
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 182
                traceback.print_exc()
183
 
14608 amit.gupta 184
 
185
 
14771 amit.gupta 186
    def parserest(self, soup):
187
        print "Hi"
188
        if soup.find('h1'):
189
            print "OK"
14608 amit.gupta 190
 
14464 amit.gupta 191
    def parseOldStlye(self, merchantOrder, soup):
192
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
193
        table = soup.body.findAll("table", recursive=False)[1]
194
        #print table
195
        tables = table.tr.td.findAll("table", recursive=False)
196
        for tr in tables[2].findAll("tr"):
197
            boldElement = tr.td.b
198
            if "Order Placed" in str(boldElement):
199
                merchantOrder.placedOn = boldElement.next_sibling.strip()
200
            if "order number" in str(boldElement):
201
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
202
            if "Order Total" in str(boldElement):
203
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
204
        anchors = table.tr.td.findAll("a", recursive=False)
205
        paymentAnchor = anchors.pop(-1)
206
 
207
        count = 0
208
        subOrders = []
209
        merchantOrder.subOrders = subOrders
210
        counter = 0 
211
        for anchor in anchors:
212
            count += 1
213
            tab = anchor.next_sibling
214
            status = MStore.ORDER_PLACED
215
            subStr = "Delivery #" + str(count) + ":"
216
            if subStr in  tab.find("b").text:
217
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
218
 
219
            tab = tab.next_sibling.next_sibling
220
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
221
 
222
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
223
 
224
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
225
            i = -1
226
            for orderItemTr in orderItemTrs:
227
                i += 1
228
                if i%2 == 0:
229
                    continue
230
                counter += 1
231
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
232
 
233
                productUrl = orderItemTr.td.contents[1].a["href"]
234
                productTitle = orderItemTr.td.contents[1].a.text
235
 
236
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
237
 
238
 
239
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
240
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
241
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 242
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
243
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
244
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 245
                subOrder.productCode = productUrl.split('/')[5]
246
                subOrder.detailedStatus = detailedStatus
247
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice*quantity)
15346 amit.gupta 248
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
249
                subOrder.dealRank = dealRank.get('rank')
250
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 251
                cashbackStatus = Store.CB_PENDING
252
                if cashbackAmount <= 0:
253
                    cashbackStatus = Store.CB_NA
254
                subOrder.cashBackStatus = cashbackStatus
255
                subOrder.cashBackAmount = cashbackAmount
256
                if percentage > 0:
257
                    subOrder.cashBackPercentage = percentage
258
                subOrders.append(subOrder)
259
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
260
        totalAmount = 0
261
        grandAmount = 0
262
        for price in priceList:
263
            labelTd = price.td
264
            if 'Subtotal:' in labelTd.text:
265
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
266
            elif 'Grand Total:' in labelTd.text:
267
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
268
        if grandAmount < totalAmount:
269
            diff = totalAmount - grandAmount
270
            for subOrder in merchantOrder.subOrders:
271
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 272
        merchantOrder.status='success'
14464 amit.gupta 273
        self._updateToOrder(todict(merchantOrder))
274
 
275
    def parseNewStlye(self, merchantOrder, soup):
276
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
277
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 278
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
279
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 280
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
281
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
282
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 283
        try:
284
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
285
        except:
286
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box a-last"}).div.div.findAll('div', recursive=False)[-1]
14464 amit.gupta 287
        priceRows = priceBox.findAll('div', {'class':'a-row'})
288
        subTotal = 0
14566 amit.gupta 289
        shippingPrice = 0   
14464 amit.gupta 290
        promoApplied = 0
291
        for priceRow in priceRows:
292
            if "Item(s) Subtotal:" in str(priceRow):
293
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
294
            elif "Shipping:" in str(priceRow):
295
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
296
            elif "Grand Total:" in str(priceRow):
297
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
298
                merchantOrder.paidAmount = grandPrice
299
            elif "Total:" in str(priceRow):
300
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
301
            elif "Promotion Applied:" in str(priceRow):
302
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
303
        totalPaid = subTotal        
304
        if promoApplied > 0:
305
            totalPaid -= promoApplied
306
            if shippingPrice <= promoApplied:
307
                totalPaid  += shippingPrice
15473 amit.gupta 308
 
15474 amit.gupta 309
        shipmentDivs = orderDetailsContainer.find('div', class_='shipment').findAll('div', recursive = False)
14464 amit.gupta 310
        subOrders = []
311
        merchantOrder.subOrders = subOrders
312
        i=1
15473 amit.gupta 313
        closedStatus = True
14464 amit.gupta 314
        for shipmentDiv in shipmentDivs:
315
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
316
            statusDiv = innerBoxes[0]
317
            subOrderStatus = statusDiv.div.span.text.strip()
14566 amit.gupta 318
            deliverySpan = statusDiv.div.div.find_next_sibling('div').span
14464 amit.gupta 319
            productDivs = innerBoxes[-1].div.div.div.findAll('div', recursive=False)
320
            subOrders = []
321
            merchantOrder.subOrders = subOrders
322
            for i, productDiv in enumerate(productDivs):
323
                i +=1 
324
                imgDiv  = productDiv.div.div
325
                detailDiv = imgDiv.find_next_sibling('div')
326
                detailDivs = detailDiv.findAll('div', recursive=False)
327
                arr = detailDivs[0].a.text.strip().split(" of ", 1)
14465 amit.gupta 328
                (productTitle, quantity) = (arr[-1], (1 if len(arr)==1 else int(arr[0])) )
14820 amit.gupta 329
                try:
330
                    unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
331
                except:
332
                    unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
14464 amit.gupta 333
                amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
334
                productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
335
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
336
                subOrder.productCode = productUrl.split('/')[5]
337
                subOrder.unitPrice = unitPrice
14566 amit.gupta 338
                subOrder.merchantSubOrderId = str(i) + " of " + merchantOrder.merchantOrderId
14722 amit.gupta 339
                estDlvyTime = datetime.now()
14566 amit.gupta 340
                if deliverySpan is not None:
14756 amit.gupta 341
                    try:
342
                        subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
343
                        estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
344
                        subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
345
                        estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
346
                    except:
15473 amit.gupta 347
                        if "Delivered on" in deliverySpan.text:
348
                            subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
14756 amit.gupta 349
                        subOrder.estimatedDeliveryDate = "Not available"
15473 amit.gupta 350
 
14722 amit.gupta 351
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
14756 amit.gupta 352
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 353
                subOrder.detailedStatus = subOrderStatus
354
                subOrder.deliveryCharges = shippingPrice
14466 amit.gupta 355
                subOrder.imgUrl = imgDiv.img["src"]
14464 amit.gupta 356
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid)
15346 amit.gupta 357
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
358
                subOrder.dealRank = dealRank.get('rank')
359
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 360
                cashbackStatus = Store.CB_PENDING
361
                if cashbackAmount <= 0:
362
                    cashbackStatus = Store.CB_NA
363
                subOrder.cashBackStatus = cashbackStatus
364
                subOrder.cashBackAmount = cashbackAmount
365
                if percentage > 0:
366
                    subOrder.cashBackPercentage = percentage
15473 amit.gupta 367
                if hasattr(subOrder, 'deliveredOn'):                               
368
                    subOrder.status = Store.ORDER_DELIVERED
369
                    subOrder.closed = True
370
                    if subOrder.cashBackStatus == Store.CB_PENDING:
371
                        subOrder.cashBackStatus = Store.CB_APPROVED
372
                elif closedStatus:
373
                    closedStatus= False
14464 amit.gupta 374
                subOrders.append(subOrder)
14698 amit.gupta 375
        merchantOrder.status='success'
15473 amit.gupta 376
        merchantOrder.closed = closedStatus
14464 amit.gupta 377
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 378
 
379
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 380
        try:
381
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
382
            if fonts[0].text == "Important Message":
383
                if fonts[1].text=="This order has been cancelled.":
384
                    merchantOrder.closed = True
385
                    merchantOrder.status = "cancelled"
386
                    merchantOrder.requireDetail = False
387
                    self._updateToOrder(todict(merchantOrder))
388
                    return
389
                else:
390
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 391
            else:
392
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 393
        except:
14549 amit.gupta 394
            orderDetails = soup.body.find(id="orderDetails")
395
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
396
                merchantOrder.closed = True
14698 amit.gupta 397
                merchantOrder.status = "cancelled"
14549 amit.gupta 398
                merchantOrder.requireDetail = False
399
                self._updateToOrder(todict(merchantOrder))
400
            else:
401
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 402
 
13868 amit.gupta 403
    def getTrackingUrls(self, userId):
14074 amit.gupta 404
 
405
        missingOrderUrls = []
406
        missingOrders = self._getMissingOrders({'userId':userId})
407
        for missingOrder in missingOrders:
408
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
14683 amit.gupta 409
        orders = self._getActiveOrders({'userId':userId, "subOrders.trackingUrl":{"$exists":False} })
13882 amit.gupta 410
        count = len(orders)
13879 amit.gupta 411
        print "count", count
14948 amit.gupta 412
        print "Missing Urls"
413
        print "*************"
414
        print missingOrderUrls
13868 amit.gupta 415
        if count > 0:
14684 amit.gupta 416
            return missingOrderUrls + ['https://www.amazon.in/gp/css/order-history', 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled', 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled&startIndex=10']
13959 amit.gupta 417
        else: 
14074 amit.gupta 418
            return missingOrderUrls
13927 amit.gupta 419
 
420
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 421
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 422
        if not os.path.exists(directory):
423
            os.makedirs(directory)
14173 amit.gupta 424
 
14945 amit.gupta 425
 
13995 amit.gupta 426
        try:
427
            searchMap = {'userId':userId}
428
            collectionMap = {'merchantOrderId':1}
429
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 430
            datetimeNow = datetime.now()
431
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 432
            print "url----------------", url
14008 amit.gupta 433
 
14320 amit.gupta 434
            if url == 'https://www.amazon.in/gp/css/order-history' or 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled' in url:
14945 amit.gupta 435
                if url == 'https://www.amazon.in/gp/css/order-history':
436
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
437
                else:   
438
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')   
439
                f = open(filename,'w')
440
                f.write(rawHtml) # python will convert \n to os.linesep
441
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 442
                soup = BeautifulSoup(rawHtml)
443
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
444
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
445
                for activeOrder in activeOrders:
446
                    for orderEle in allOrders:
447
                        orderdiv = orderEle.find('div', {'class':'a-box a-color-offset-background order-info'}).find('div', {'class':'a-fixed-right-grid-col actions a-col-right'})
448
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
449
                        if merchantOrderId==activeOrder['merchantOrderId']:
450
                            closed = True
451
                            shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
452
                            shipments.pop(0)
453
                            for shipment in shipments:
454
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 455
                                sdivs = shipment.div.div.findAll('div', recursive=False)
14065 amit.gupta 456
                                orderStatus = sdivs[0].span.text.strip()
14050 amit.gupta 457
                                status = self._getStatusFromDetailedStatus(orderStatus)
458
                                if status == MStore.ORDER_DELIVERED:
14299 amit.gupta 459
                                    deliveredOn = sdivs[0].findAll('span')[-1].text.strip()
14962 amit.gupta 460
                                    try:
461
                                        deliveredOn = deliveredOn.split(":")[1].strip()
462
                                    except:
463
                                        deliveredOn = "" 
14335 amit.gupta 464
                                deliveryestimatespan = sdivs[0].find('span', {'class':'a-color-success'})
465
                                deliveryEstimate = None
466
                                if deliveryestimatespan is not None:
467
                                    deliveryEstimate = deliveryestimatespan.find('span', {'class':'a-text-bold'}).text.strip()
468
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 469
                                trackingUrl = None
470
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
471
                                    if buttonDiv.find('a').text.strip()=='Track package':
472
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
473
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 474
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 475
                                        break
476
                                for prodDiv in productDivs:
477
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
478
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
479
                                    imgUrl = prodDiv.find("img")["src"]
480
                                    for subOrder in activeOrder['subOrders']:
481
                                        if subOrder['productTitle'] == productTitle:
482
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
483
                                            updateMap = {}
484
                                            closedStatus = False
485
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 486
                                            updateMap['subOrders.$.lastTracked'] = timestamp
13995 amit.gupta 487
                                            updateMap['subOrders.$.detailedStatus'] = orderStatus
488
                                            cashbackStatus = subOrder.get("cashBackStatus")
489
                                            updateMap['subOrders.$.status'] = status 
490
 
491
                                            if status==MStore.ORDER_DELIVERED:                               
14291 amit.gupta 492
                                                updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 493
                                                closedStatus = True
494
                                                updateMap['subOrders.$.closed'] = True
495
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 496
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 497
                                            if status==MStore.ORDER_CANCELLED:     
498
                                                closedStatus = True
499
                                                updateMap['subOrders.$.closed'] = True
500
                                                if cashbackStatus == Store.CB_PENDING:
501
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
502
                                            if status==MStore.ORDER_SHIPPED:   
503
                                                updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
504
                                                if trackingUrl is not None:
505
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
506
                                            if not closedStatus:
507
                                                closed = False
508
                                            bulk.find(findMap).update({'$set' : updateMap})
509
                                            break
510
                            bulk.find({'orderId': activeOrder['orderId']}).update({'$set':{'closed': closed}})            
14080 amit.gupta 511
                bulk.execute()
512
                return 'PARSED_SUCCESS'
513
            else:
15563 amit.gupta 514
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
515
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 516
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 517
 
518
                filename = directory + "/" + merchantOrderId   
519
                f = open(filename,'w')
520
                f.write(rawHtml) # python will convert \n to os.linesep
521
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 522
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 523
                print "result", result
15555 amit.gupta 524
                try:
15560 amit.gupta 525
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
526
                    order1.status = result
527
                    order1.rawhtml = rawHtml
15555 amit.gupta 528
                    session.commit()
15561 amit.gupta 529
                except:
530
                    traceback.print_exc()
15555 amit.gupta 531
                finally:
532
                    session.close()
14080 amit.gupta 533
                return 'PARSED_SUCCESS'
534
                pass
14008 amit.gupta 535
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 536
        except:
537
            traceback.print_exc()    
538
            return 'PARSED_FAILED'
539
 
540
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 541
        if "ordered from" in detailedStatus.lower():
542
            return MStore.ORDER_PLACED 
543
 
13995 amit.gupta 544
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 545
            if detailedStatus.lower() in value:
13995 amit.gupta 546
                return key
547
 
14676 amit.gupta 548
        print "Detailed Status need to be mapped", "Store:", self.store_id
13995 amit.gupta 549
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 550
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 551
        br = getBrowserObject()
552
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
553
        url = AMAZON_AFF_URL
554
        response = br.open(url)
555
        #get data for past 40 days and store it to mongo
556
        dt = datetime.now()
14701 amit.gupta 557
        dat = dt - timedelta(days=2)
14651 amit.gupta 558
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
559
        response = br.open(url)
560
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
561
        j=-1
562
        for row in page.split("\n"):
563
            j += 1
564
            if j== 0 or j==1:
565
                continue
566
            fields = row.split("\t")
567
            if len(fields)>1:
568
                print fields
569
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
570
                print amazonAffiliate
571
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
572
            else:
573
                break
14650 amit.gupta 574
 
14608 amit.gupta 575
 
14749 amit.gupta 576
    def parseTrackingUrl(self, trackingUrl):
14624 amit.gupta 577
        subOrder = {}
14749 amit.gupta 578
        page = fetchResponseUsingProxy(trackingUrl)
14624 amit.gupta 579
        soup = BeautifulSoup(page)
580
        alertContainer = soup.find("div", {"class":"a-box-inner a-alert-container"})
581
        if alertContainer is not None:
582
            statusText = alertContainer.h4.span.text
583
            detailedStatus = statusText.split(":")[0].strip()
584
            subOrder['detailedStatus'] = detailedStatus
585
            if detailedStatus.lower() == "in transit":
586
                pass
587
            elif detailedStatus.lower() == "undeliverable":
588
                subOrder['status'] = MStore.ORDER_CANCELLED
589
                return subOrder
590
        else:
591
            summaryLeft = soup.find(id="summaryLeft")
592
            statusText = summaryLeft.h2.text
593
            detailedStatus = statusText.split(":")[0].strip()
594
            subOrder['detailedStatus'] = detailedStatus
595
            if detailedStatus.lower() == "delivered":
596
                subOrder['deliveredOn'] = summaryLeft.findAll("span")[-2].text.strip()
597
                subOrder['status'] = MStore.ORDER_DELIVERED
598
                return subOrder
599
            else:
600
                subOrder['expectedDelivery'] = summaryLeft.findAll("span")[-1].text.strip()
601
        return subOrder
602
 
603
 
604
 
14608 amit.gupta 605
 
606
 
13927 amit.gupta 607
 
608
 
13774 amit.gupta 609
def main():
15346 amit.gupta 610
    #str1 = readSSh("/AmazonTrack/User533/403-3748792-1516356")
15046 amit.gupta 611
    #readSSh("/tmp/User2/404-2225153-7073122")
14962 amit.gupta 612
 
15565 amit.gupta 613
    store = getStore(1)
14756 amit.gupta 614
    #store.scrapeStoreOrders()
14650 amit.gupta 615
#    br = mechanize.Browser()
616
#    br.addheaders = [('User-agent','Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11'),
617
#                     ('Accept', 'text/html,application/xhtml+xml,application/json,application/xml;q=0.9,*/*;q=0.8'),
618
#                     ('Accept-Encoding', 'gzip,deflate,sdch'),                  
619
#                     ('Accept-Language', 'en-US,en;q=0.8'),                     
620
#                     ('Accept-Charset', 'ISO-8859-1,utf-8;q=0.7,*;q=0.3')]
621
#    store.scrapeStoreOrders()
15565 amit.gupta 622
    store.parseOrderRawHtml(5322, "13232", 14, readSSh("/home/amit/5322"), "https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html/ref=ox_spc_place_order?ie=UTF8&fromAnywhere=1&hasWorkingJav")
15555 amit.gupta 623
    #store.trackOrdersForUser(4355, "https://www.amazon.in/gp/css/summary/edit.html?orderID=171-4824011-7090713", readSSh("~/4355"))
14813 amit.gupta 624
    #readSSh("/tmp/User211/2015-04-12 10:32:41.905765")
14703 amit.gupta 625
    #store.scrapeAffiliate()
15565 amit.gupta 626
    #parseDetailNotCreated()
627
    parseOrderNotCreated()
15555 amit.gupta 628
 
629
def parseDetailNotCreated():
630
    try:
631
        store=getStore(1)
632
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 633
        session.close()
15555 amit.gupta 634
        for order in orders:
635
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
636
 
637
    finally:
638
        session.close()
15565 amit.gupta 639
 
640
def parseOrderNotCreated():
641
    try:
642
        store=getStore(1)
643
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
644
        session.close()
645
        for order in orders:
15567 amit.gupta 646
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
647
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
648
            order1.status = result
649
            session.commit()
15565 amit.gupta 650
 
651
    finally:
652
        session.close()
15555 amit.gupta 653
 
14162 amit.gupta 654
def readSSh(fileName):
655
    try:
15046 amit.gupta 656
        str1 = open(fileName).read()
657
        return str1
14162 amit.gupta 658
    except:
659
        ssh_client = SSHClient()
660
        str1 = ""
661
        ssh_client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
15056 amit.gupta 662
        ssh_client.connect('dtr', 22, 'root', 'ecip$dtrMay2014')
14162 amit.gupta 663
        sftp_client = ssh_client.open_sftp()
664
        try:
14299 amit.gupta 665
            if not os.path.exists(os.path.dirname(fileName)):
666
                os.makedirs(os.path.dirname(fileName))
14162 amit.gupta 667
            sftp_client.get(fileName, fileName)
668
            try:
669
                str1 = open(fileName).read()
670
                return str1
671
            finally:
672
                pass
673
        except:
674
            "could not read"
675
        return str1
13774 amit.gupta 676
 
677
 
678
if __name__ == '__main__':
14284 amit.gupta 679
        main()