Subversion Repositories SmartDukaan

Rev

Rev 17617 | Rev 17641 | 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
17231 amit.gupta 9
from datetime import datetime, timedelta, date
16874 amit.gupta 10
from dtr.api.Order import process_rejects
14650 amit.gupta 11
from dtr.dao import Order, SubOrder, AmazonAffiliateInfo
14464 amit.gupta 12
from dtr.main import getStore, Store as MStore, ParseException, getBrowserObject, \
14624 amit.gupta 13
    ungzipResponse, tprint
14650 amit.gupta 14
from dtr.sources.flipkart import todict, AFF_REPORT_URL
17307 amit.gupta 15
from dtr.storage.DataService import OrdersRaw, Orders, Order_Parse_Info, \
16986 amit.gupta 16
    All_user_addresses
16874 amit.gupta 17
from dtr.storage.Mongo import getDealRank
17307 amit.gupta 18
from dtr.utils import utils
16874 amit.gupta 19
from dtr.utils.utils import fetchResponseUsingProxy, readSSh
20
from elixir import *
14650 amit.gupta 21
import base64
16874 amit.gupta 22
import dtr
14650 amit.gupta 23
import gzip
14624 amit.gupta 24
import mechanize
14285 amit.gupta 25
import os.path
13774 amit.gupta 26
import re
14033 amit.gupta 27
import time
13809 amit.gupta 28
import traceback
14650 amit.gupta 29
import urllib2
17600 amit.gupta 30
import pyquery
13774 amit.gupta 31
 
32
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 33
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 34
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
14624 amit.gupta 35
AMAZON_AFF_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/listReports'
14650 amit.gupta 36
AMAZON_AFF_FILE_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/getReport?filename=saholic-21-orders-report-%s.tsv.gz'
13774 amit.gupta 37
class Store(MStore):
13821 amit.gupta 38
 
17348 amit.gupta 39
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch', 'order received'],
14064 amit.gupta 40
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
17324 amit.gupta 41
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled', 'replacement complete', 'return received'],
17258 amit.gupta 42
                            MStore.ORDER_DELIVERED : ['delivered', 'your package was delivered', 'package was handed directly to customer']
13821 amit.gupta 43
                           }
13774 amit.gupta 44
 
45
    def __init__(self,store_id):
46
        super(Store, self).__init__(store_id)
47
 
48
    def getName(self):
14749 amit.gupta 49
        return "amazon"
13774 amit.gupta 50
 
14945 amit.gupta 51
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, track=False):
14958 amit.gupta 52
        parseString = "Tracking" if track else "Transacted"
53
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 54
        resp = {}
14813 amit.gupta 55
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 56
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 57
            try:
17600 amit.gupta 58
                from pyquery import PyQuery as pq
59
                doc = pq(rawHtml)
14608 amit.gupta 60
                try:
17600 amit.gupta 61
                    orderUrl = doc('a.a-touch-link').attr.href
14608 amit.gupta 62
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
63
                except:
17600 amit.gupta 64
                    merchantOrderId = doc("#orders-list>div>span>b").html()
14699 amit.gupta 65
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 66
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
67
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 68
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 69
                order.requireDetail = True
14698 amit.gupta 70
                order.status = 'html_required'
14297 amit.gupta 71
                order.closed = None
14312 amit.gupta 72
                if self._saveToOrder(todict(order)):
73
                    resp['result'] = 'ORDER_CREATED'
74
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
75
                    resp["htmlRequired"] = True
76
                    resp['orderId'] = orderId
77
                else:
78
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
79
 
13774 amit.gupta 80
            except:
14813 amit.gupta 81
                #Write all cases here for Order Not created Known
17600 amit.gupta 82
                soup = BeautifulSoup(rawHtml)
14809 amit.gupta 83
                try:
16874 amit.gupta 84
                    if not soup.body:
85
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
17359 amit.gupta 86
                    elif 'Securely redirecting you' in soup.find("h3").text.strip() or soup.find("h3").text.strip()=="Orders":
14813 amit.gupta 87
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 88
                    else:
14813 amit.gupta 89
                        raise
14809 amit.gupta 90
                except:
14814 amit.gupta 91
                    try:
17356 amit.gupta 92
                        if soup.find("h1").text.strip() in ['This is a duplicate order', 'There was a problem with your payment.', 'Your Orders', 'Your Shopping Cart is empty.', 'Select a payment method', 'Edit quantities'] or "Saved for later" in soup.find("h1").text.strip():
14814 amit.gupta 93
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
94
                        else:
95
                            raise
96
                    except:
15046 amit.gupta 97
                        try:
17356 amit.gupta 98
                            if soup.find("h2").text.strip() in ['Web page not available','Webpage not available', 'Do you have an Amazon password?']:
15046 amit.gupta 99
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
100
                            else:
101
                                raise
102
                        except:
15565 amit.gupta 103
                            try:
16876 amit.gupta 104
                                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 or soup.find(id="nav-greeting-name") is not None:
15565 amit.gupta 105
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
16877 amit.gupta 106
                                elif soup.find("b", {'class':'h1'}).text.strip().find("We're sorry") > -1:
107
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
15565 amit.gupta 108
                                else:
109
                                    raise
110
                            except:
111
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
112
 
13774 amit.gupta 113
        else:
13781 amit.gupta 114
            try:
14722 amit.gupta 115
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
116
                if mo is not None:
14749 amit.gupta 117
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 118
                    merchantOrder.createdOn = mo.get("createdOn")
119
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
120
                else:
14945 amit.gupta 121
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 122
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14464 amit.gupta 123
                soup = BeautifulSoup(rawHtml)
17331 amit.gupta 124
                if not soup.body:
17319 amit.gupta 125
                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
126
                else:   
14545 amit.gupta 127
                    try:
17319 amit.gupta 128
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 129
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 130
                    except:
15056 amit.gupta 131
                        try:
17319 amit.gupta 132
                            traceback.print_exc()
133
                            self.parseOldStlye(merchantOrder, soup)
134
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 135
                        except:
17319 amit.gupta 136
                            traceback.print_exc()
15059 amit.gupta 137
                            try:
17319 amit.gupta 138
                                self.parseCancelled(merchantOrder, soup)
139
                                resp['result'] = 'ORDER_CANCELLED'
15059 amit.gupta 140
                            except:
17319 amit.gupta 141
                                try:
17356 amit.gupta 142
                                    if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
17319 amit.gupta 143
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
144
                                    else:
145
                                        raise
146
                                except:
17356 amit.gupta 147
                                    if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
17319 amit.gupta 148
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
149
                                    else:
150
                                        raise
17599 amit.gupta 151
                if resp['result'] == 'DETAIL_NOT_CREATED_KNOWN':
17357 amit.gupta 152
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 153
            except:
17599 amit.gupta 154
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"parse_failed"}})
14813 amit.gupta 155
                print "Error occurred"
15046 amit.gupta 156
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 157
                traceback.print_exc()
15046 amit.gupta 158
        return resp    
13774 amit.gupta 159
 
13868 amit.gupta 160
    #This should be exposed from api for specific sources
161
    def scrapeStoreOrders(self):
14624 amit.gupta 162
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
163
        for merchantOrder in orders:
14981 amit.gupta 164
            executeBulk = False
14624 amit.gupta 165
            try:
166
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
167
                closed = True
168
                map1 = {}
169
                for subOrder in merchantOrder.get("subOrders"):
170
                    if subOrder.get("closed"):
171
                        continue
172
                    elif subOrder.get("trackingUrl") is None:
173
                        closed = False
174
                        continue
175
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
176
                    trackingUrl = subOrder.get("trackingUrl")
177
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 178
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 179
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 180
                    if newOrder:
181
                        executeBulk = True
182
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
183
                        print findMap, "\n", updateMap
184
                        bulk.find(findMap).update({'$set' : updateMap})
185
                        closed = closed and newOrder['closed']
14981 amit.gupta 186
                if executeBulk:
14697 amit.gupta 187
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 188
                    bulk.execute()
14624 amit.gupta 189
            except:
14847 amit.gupta 190
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 191
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 192
                traceback.print_exc()
193
 
14608 amit.gupta 194
 
195
 
14771 amit.gupta 196
    def parserest(self, soup):
197
        print "Hi"
198
        if soup.find('h1'):
199
            print "OK"
14608 amit.gupta 200
 
14464 amit.gupta 201
    def parseOldStlye(self, merchantOrder, soup):
202
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
203
        table = soup.body.findAll("table", recursive=False)[1]
204
        #print table
205
        tables = table.tr.td.findAll("table", recursive=False)
206
        for tr in tables[2].findAll("tr"):
207
            boldElement = tr.td.b
208
            if "Order Placed" in str(boldElement):
209
                merchantOrder.placedOn = boldElement.next_sibling.strip()
210
            if "order number" in str(boldElement):
211
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
212
            if "Order Total" in str(boldElement):
213
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
214
        anchors = table.tr.td.findAll("a", recursive=False)
215
        paymentAnchor = anchors.pop(-1)
216
 
217
        count = 0
218
        subOrders = []
219
        merchantOrder.subOrders = subOrders
220
        counter = 0 
221
        for anchor in anchors:
222
            count += 1
223
            tab = anchor.next_sibling
224
            status = MStore.ORDER_PLACED
225
            subStr = "Delivery #" + str(count) + ":"
226
            if subStr in  tab.find("b").text:
227
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
228
 
229
            tab = tab.next_sibling.next_sibling
230
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
231
 
232
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
233
 
234
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
235
            i = -1
236
            for orderItemTr in orderItemTrs:
237
                i += 1
238
                if i%2 == 0:
239
                    continue
240
                counter += 1
241
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
242
 
243
                productUrl = orderItemTr.td.contents[1].a["href"]
244
                productTitle = orderItemTr.td.contents[1].a.text
245
 
246
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
247
 
248
 
249
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
250
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
251
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 252
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
253
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
254
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 255
                subOrder.productCode = productUrl.split('/')[5]
256
                subOrder.detailedStatus = detailedStatus
257
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice*quantity)
15346 amit.gupta 258
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
259
                subOrder.dealRank = dealRank.get('rank')
260
                subOrder.rankDesc = dealRank.get('description')
16283 amit.gupta 261
                subOrder.maxNlc = dealRank.get('maxNlc')
262
                subOrder.minNlc = dealRank.get('minNlc')
263
                subOrder.db = dealRank.get('dp')
264
                subOrder.itemStatus = dealRank.get('status')
14464 amit.gupta 265
                cashbackStatus = Store.CB_PENDING
266
                if cashbackAmount <= 0:
267
                    cashbackStatus = Store.CB_NA
268
                subOrder.cashBackStatus = cashbackStatus
269
                subOrder.cashBackAmount = cashbackAmount
270
                if percentage > 0:
271
                    subOrder.cashBackPercentage = percentage
272
                subOrders.append(subOrder)
273
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
274
        totalAmount = 0
275
        grandAmount = 0
276
        for price in priceList:
277
            labelTd = price.td
278
            if 'Subtotal:' in labelTd.text:
279
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
280
            elif 'Grand Total:' in labelTd.text:
281
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
282
        if grandAmount < totalAmount:
283
            diff = totalAmount - grandAmount
284
            for subOrder in merchantOrder.subOrders:
285
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 286
        merchantOrder.status='success'
14464 amit.gupta 287
        self._updateToOrder(todict(merchantOrder))
288
 
289
    def parseNewStlye(self, merchantOrder, soup):
290
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
291
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 292
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
293
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 294
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
295
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
296
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 297
        try:
298
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
299
        except:
300
            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 301
        priceRows = priceBox.findAll('div', {'class':'a-row'})
302
        subTotal = 0
14566 amit.gupta 303
        shippingPrice = 0   
14464 amit.gupta 304
        promoApplied = 0
305
        for priceRow in priceRows:
306
            if "Item(s) Subtotal:" in str(priceRow):
307
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
308
            elif "Shipping:" in str(priceRow):
309
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
310
            elif "Grand Total:" in str(priceRow):
311
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
312
                merchantOrder.paidAmount = grandPrice
313
            elif "Total:" in str(priceRow):
314
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
315
            elif "Promotion Applied:" in str(priceRow):
316
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
317
        totalPaid = subTotal        
318
        if promoApplied > 0:
319
            totalPaid -= promoApplied
320
            if shippingPrice <= promoApplied:
321
                totalPaid  += shippingPrice
15473 amit.gupta 322
 
17601 amit.gupta 323
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 324
        subOrders = []
325
        merchantOrder.subOrders = subOrders
15473 amit.gupta 326
        closedStatus = True
17601 amit.gupta 327
        subOrders = []
14464 amit.gupta 328
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 329
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 330
            try:
331
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
332
            except:
333
                trackingUrl= None
334
 
17347 amit.gupta 335
            deliverySpanTop = None
14464 amit.gupta 336
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
337
            statusDiv = innerBoxes[0]
338
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 339
            try:
340
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
341
            except:
17617 amit.gupta 342
                pass
343
            productDivs = innerBoxes[-1].div.div.div.findAll('div', recursive=False)
14464 amit.gupta 344
            merchantOrder.subOrders = subOrders
345
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 346
                deliverySpan = deliverySpanTop
347
                j=i+1
348
                if not deliverySpanTop:
349
                    if i%2==1:
350
                        continue
351
                    j=i/2 + 1
17599 amit.gupta 352
                    deliverySpan = statusDiv.div.div.span
14464 amit.gupta 353
                imgDiv  = productDiv.div.div
354
                detailDiv = imgDiv.find_next_sibling('div')
355
                detailDivs = detailDiv.findAll('div', recursive=False)
356
                arr = detailDivs[0].a.text.strip().split(" of ", 1)
14465 amit.gupta 357
                (productTitle, quantity) = (arr[-1], (1 if len(arr)==1 else int(arr[0])) )
14820 amit.gupta 358
                try:
359
                    unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
360
                except:
361
                    unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
14464 amit.gupta 362
                amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
363
                productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
364
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
365
                subOrder.productCode = productUrl.split('/')[5]
366
                subOrder.unitPrice = unitPrice
17347 amit.gupta 367
                subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
14722 amit.gupta 368
                estDlvyTime = datetime.now()
14566 amit.gupta 369
                if deliverySpan is not None:
14756 amit.gupta 370
                    try:
371
                        subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
372
                        estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
373
                        subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
374
                        estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
375
                    except:
15473 amit.gupta 376
                        if "Delivered on" in deliverySpan.text:
377
                            subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
14756 amit.gupta 378
                        subOrder.estimatedDeliveryDate = "Not available"
15473 amit.gupta 379
 
14722 amit.gupta 380
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
14756 amit.gupta 381
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 382
                subOrder.detailedStatus = subOrderStatus
17599 amit.gupta 383
                status=MStore.ORDER_PLACED
384
                try:
385
                    status = self._getStatusFromDetailedStatus(subOrderStatus)
386
                except:
387
                    try:
388
                        dateString = subOrderStatus.split("Delivered ")[1].strip()
389
                        subOrder.status = MStore.ORDER_DELIVERED
390
                        subOrder.detailedStatus = 'Delivered'
17631 amit.gupta 391
                        dateString = getDateStringDelivered(dateString)
392
                        if dateString is not None:
393
                            subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 394
                    except:
395
                        try:
396
                            dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
397
                            subOrder.status = MStore.ORDER_SHIPPED
17631 amit.gupta 398
                            dateString = getDateStringArriving(dateString)
399
                            if dateString is not None:
400
                                subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 401
                        except:
402
                            print "Unknown status Alert -", status
403
 
14464 amit.gupta 404
                subOrder.deliveryCharges = shippingPrice
17361 amit.gupta 405
                if trackingUrl:
406
                    subOrder.tracingkUrl  = trackingUrl
14466 amit.gupta 407
                subOrder.imgUrl = imgDiv.img["src"]
14464 amit.gupta 408
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid)
15346 amit.gupta 409
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
410
                subOrder.dealRank = dealRank.get('rank')
411
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 412
                cashbackStatus = Store.CB_PENDING
413
                if cashbackAmount <= 0:
414
                    cashbackStatus = Store.CB_NA
415
                subOrder.cashBackStatus = cashbackStatus
416
                subOrder.cashBackAmount = cashbackAmount
417
                if percentage > 0:
418
                    subOrder.cashBackPercentage = percentage
17599 amit.gupta 419
                if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
15473 amit.gupta 420
                    subOrder.status = Store.ORDER_DELIVERED
421
                    subOrder.closed = True
422
                    if subOrder.cashBackStatus == Store.CB_PENDING:
423
                        subOrder.cashBackStatus = Store.CB_APPROVED
424
                elif closedStatus:
425
                    closedStatus= False
14464 amit.gupta 426
                subOrders.append(subOrder)
14698 amit.gupta 427
        merchantOrder.status='success'
15473 amit.gupta 428
        merchantOrder.closed = closedStatus
14464 amit.gupta 429
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 430
 
431
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 432
        try:
433
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
434
            if fonts[0].text == "Important Message":
435
                if fonts[1].text=="This order has been cancelled.":
436
                    merchantOrder.closed = True
437
                    merchantOrder.status = "cancelled"
438
                    merchantOrder.requireDetail = False
439
                    self._updateToOrder(todict(merchantOrder))
440
                    return
441
                else:
442
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 443
            else:
444
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 445
        except:
14549 amit.gupta 446
            orderDetails = soup.body.find(id="orderDetails")
447
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
448
                merchantOrder.closed = True
14698 amit.gupta 449
                merchantOrder.status = "cancelled"
14549 amit.gupta 450
                merchantOrder.requireDetail = False
451
                self._updateToOrder(todict(merchantOrder))
452
            else:
453
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 454
 
13868 amit.gupta 455
    def getTrackingUrls(self, userId):
14074 amit.gupta 456
 
457
        missingOrderUrls = []
458
        missingOrders = self._getMissingOrders({'userId':userId})
459
        for missingOrder in missingOrders:
460
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 461
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 462
        count = len(orders)
13879 amit.gupta 463
        print "count", count
14948 amit.gupta 464
        print "Missing Urls"
465
        print "*************"
466
        print missingOrderUrls
13868 amit.gupta 467
        if count > 0:
14684 amit.gupta 468
            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 469
        else: 
14074 amit.gupta 470
            return missingOrderUrls
13927 amit.gupta 471
 
472
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 473
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 474
        if not os.path.exists(directory):
475
            os.makedirs(directory)
14173 amit.gupta 476
 
14945 amit.gupta 477
 
13995 amit.gupta 478
        try:
479
            searchMap = {'userId':userId}
480
            collectionMap = {'merchantOrderId':1}
481
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 482
            datetimeNow = datetime.now()
483
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 484
            print "url----------------", url
17465 amit.gupta 485
            cancelledSummary = False
14320 amit.gupta 486
            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 487
                if url == 'https://www.amazon.in/gp/css/order-history':
488
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
489
                else:   
17465 amit.gupta 490
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
491
                    cancelledSummary = True   
14945 amit.gupta 492
                f = open(filename,'w')
493
                f.write(rawHtml) # python will convert \n to os.linesep
494
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 495
                soup = BeautifulSoup(rawHtml)
496
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
497
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
498
                for activeOrder in activeOrders:
17307 amit.gupta 499
                    matched=False
13995 amit.gupta 500
                    for orderEle in allOrders:
17254 amit.gupta 501
                        deliveredOn = None
502
                        deliveryEstimate = None
17258 amit.gupta 503
                        shippingEstimate = None
13995 amit.gupta 504
                        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'})
505
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
506
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 507
                            matched=True
13995 amit.gupta 508
                            closed = True
17465 amit.gupta 509
                            if not cancelledSummary:
510
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
511
                            else:
512
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
513
                                shipments.pop(0)
13995 amit.gupta 514
                            for shipment in shipments:
17307 amit.gupta 515
                                orderStatusDesc = None
13995 amit.gupta 516
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 517
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 518
                                try:
17270 amit.gupta 519
                                    orderStatus = sdivs[0].span.text.strip()
520
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 521
                                except:
17270 amit.gupta 522
                                    try:
523
                                        dateString = orderStatus.split("Delivered ")[1].strip()
524
                                        status = MStore.ORDER_DELIVERED
525
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
526
                                    except:
527
                                        try:
17271 amit.gupta 528
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 529
                                            status = MStore.ORDER_SHIPPED
530
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
531
                                        except:
532
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 533
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 534
                                try:
17270 amit.gupta 535
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
536
                                except:
17271 amit.gupta 537
                                    try:
538
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
539
                                    except:
17307 amit.gupta 540
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 541
 
17271 amit.gupta 542
                                if orderStatusDesc: 
17348 amit.gupta 543
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 544
                                    try:
17348 amit.gupta 545
                                        status = self._getStatusFromDetailedStatus(orderStatus)
546
                                    except:
547
                                        pass
548
                                    try:
17271 amit.gupta 549
                                        if "Dispatch estimate" in orderStatusDesc:
550
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
551
                                        elif "Delivery estimate" in orderStatus:
552
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
553
                                        elif "Arriving" in orderStatus:
554
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
555
                                    except:
556
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
557
                                        closed=False
558
                                        status = None
17258 amit.gupta 559
 
14335 amit.gupta 560
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 561
                                trackingUrl = None
562
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
563
                                    if buttonDiv.find('a').text.strip()=='Track package':
564
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
565
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 566
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 567
                                        break
568
                                for prodDiv in productDivs:
569
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
570
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
571
                                    imgUrl = prodDiv.find("img")["src"]
572
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 573
                                        if subOrder['closed']==True:
574
                                            continue
17230 amit.gupta 575
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 576
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
577
                                            updateMap = {}
578
                                            closedStatus = False
579
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 580
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 581
                                            if status:
17348 amit.gupta 582
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 583
                                                updateMap['subOrders.$.status'] = status 
584
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 585
 
17251 amit.gupta 586
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 587
                                                if deliveredOn:                               
17254 amit.gupta 588
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 589
                                                closedStatus = True
590
                                                updateMap['subOrders.$.closed'] = True
591
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 592
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 593
                                            if status==MStore.ORDER_CANCELLED:     
594
                                                closedStatus = True
595
                                                updateMap['subOrders.$.closed'] = True
596
                                                if cashbackStatus == Store.CB_PENDING:
597
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 598
                                            if status==MStore.ORDER_SHIPPED:
599
                                                if deliveryEstimate:   
600
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 601
                                                if trackingUrl is not None:
602
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 603
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 604
                                            if shippingEstimate:   
605
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 606
                                            if not closedStatus:
607
                                                closed = False
17307 amit.gupta 608
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 609
                                            updateMap['status']='success'
13995 amit.gupta 610
                                            bulk.find(findMap).update({'$set' : updateMap})
611
                                            break
17321 amit.gupta 612
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 613
                            break
614
                    if not matched:
17319 amit.gupta 615
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 616
                        for subOrder in activeOrder['subOrders']:
617
                            if subOrder['closed']==True:
618
                                continue
619
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
620
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
621
                            bulk.find({'orderId': activeOrder['orderId']})
622
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 623
                bulk.execute()
624
                return 'PARSED_SUCCESS'
625
            else:
15563 amit.gupta 626
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
627
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 628
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 629
 
630
                filename = directory + "/" + merchantOrderId   
631
                f = open(filename,'w')
632
                f.write(rawHtml) # python will convert \n to os.linesep
633
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 634
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 635
                print "result", result
15555 amit.gupta 636
                try:
15560 amit.gupta 637
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
638
                    order1.status = result
639
                    order1.rawhtml = rawHtml
17358 amit.gupta 640
                    order1.order_url = url
15555 amit.gupta 641
                    session.commit()
15561 amit.gupta 642
                except:
643
                    traceback.print_exc()
15555 amit.gupta 644
                finally:
645
                    session.close()
14080 amit.gupta 646
                return 'PARSED_SUCCESS'
647
                pass
14008 amit.gupta 648
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 649
        except:
650
            traceback.print_exc()    
651
            return 'PARSED_FAILED'
652
 
653
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 654
        if "ordered from" in detailedStatus.lower():
655
            return MStore.ORDER_PLACED 
656
 
13995 amit.gupta 657
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 658
            if detailedStatus.lower() in value:
13995 amit.gupta 659
                return key
660
 
17253 amit.gupta 661
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 662
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 663
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 664
        br = getBrowserObject()
665
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
666
        url = AMAZON_AFF_URL
667
        response = br.open(url)
668
        #get data for past 40 days and store it to mongo
669
        dt = datetime.now()
14701 amit.gupta 670
        dat = dt - timedelta(days=2)
14651 amit.gupta 671
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
672
        response = br.open(url)
673
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
674
        j=-1
675
        for row in page.split("\n"):
676
            j += 1
677
            if j== 0 or j==1:
678
                continue
679
            fields = row.split("\t")
680
            if len(fields)>1:
681
                print fields
682
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
683
                print amazonAffiliate
684
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
685
            else:
686
                break
14650 amit.gupta 687
 
14608 amit.gupta 688
 
16468 amit.gupta 689
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 690
        print trackingUrl
14624 amit.gupta 691
        subOrder = {}
14749 amit.gupta 692
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 693
        status = MStore.ORDER_SHIPPED
694
        #print page
14624 amit.gupta 695
        soup = BeautifulSoup(page)
16468 amit.gupta 696
        header1 = soup.find("h1") 
697
        if header1:
698
            if header1.text=="Sign In":
699
                print "Login page is displayed for order id", orderId
17307 amit.gupta 700
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 701
                return subOrder
16468 amit.gupta 702
        try:
16474 amit.gupta 703
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 704
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
705
            try:
706
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
707
            except:
708
                displayStatus = detailedStatus
709
            print displayStatus
710
            if detailedStatus.lower().find("delivered")>=0:
711
                print detailedStatus
712
                displayStatus = "Delivered"
713
                status = "Delivered"
714
                try:
715
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
716
                except:
717
                    pass
718
 
719
            elif detailedStatus.lower() == 'returned':
720
                status = 'Cancelled'
721
            subOrder['status'] = status
722
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 723
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 724
 
725
        except:
17350 amit.gupta 726
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 727
            print "failed to parse", orderId
728
            traceback.print_exc()
729
 
14624 amit.gupta 730
        return subOrder
731
 
732
 
733
 
16980 amit.gupta 734
    def parseInfo(self,):
735
        from pyquery import PyQuery as pq
16986 amit.gupta 736
        orders = list(session.query(Orders).filter_by(store_id=self.store_id).filter_by(status='DETAIL_CREATED').group_by(Orders.user_id).all())
16980 amit.gupta 737
        try:
738
            for order in orders:
739
                try:
740
                    doc = pq(order.rawhtml)
741
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
742
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 743
                    orderInfo = All_user_addresses()
16980 amit.gupta 744
                    orderInfo.address = lists[-3].text
745
                    orderInfo.user_id = order.user_id
16986 amit.gupta 746
                    orderInfo.source = 'order' 
747
                    #orderInfo.order_id = order.id
748
                    #orderInfo.email = None
749
                    #orderInfo.name = lists[0].text 
750
                    #orderInfo.mobile = None
16980 amit.gupta 751
                    adSplit = lists[-2].text.split(",")
752
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
753
                    if match:
754
                        items = match.groups()
16986 amit.gupta 755
                    orderInfo.city = adSplit[0].strip()
756
                    orderInfo.pincode = items[1].strip()
757
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 758
                    session.commit()
759
                except:
760
                    session.rollback()
761
                    continue
762
        finally:
763
            session.close()
14608 amit.gupta 764
 
13927 amit.gupta 765
 
766
 
13774 amit.gupta 767
def main():
16874 amit.gupta 768
    store = getStore(1)
17617 amit.gupta 769
 
17631 amit.gupta 770
    for orderId in [59433]:
17602 amit.gupta 771
        try:
772
            o = session.query(OrdersRaw).filter_by(id = orderId).one()
773
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
774
        finally:
775
            session.close()
17631 amit.gupta 776
    #store.trackOrdersForUser(10466, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-7498756-0837158', readSSh('/AmazonTrack/User10466/403-7498756-0837158'))
777
    #store.trackOrdersForUser(9031, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-0812780-9465164', readSSh('/home/amit/sample.html'))
17231 amit.gupta 778
 
779
def getSummaryFile(directory):
780
    date1 = datetime(2015,1,1)
781
    finalFile = None
782
    try:
783
        for file in os.listdir(directory):
784
            if file.startswith("orderSummary"):
785
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
786
                if date2 > date1:
787
                    date1 = date2
788
                    finalFile=file
789
    except:
790
        print "Missing directory"
17251 amit.gupta 791
    return finalFile 
17231 amit.gupta 792
 
15555 amit.gupta 793
 
794
def parseDetailNotCreated():
795
    try:
796
        store=getStore(1)
797
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 798
        session.close()
15555 amit.gupta 799
        for order in orders:
800
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
801
 
802
    finally:
803
        session.close()
17251 amit.gupta 804
 
805
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 806
    print dateString
17251 amit.gupta 807
    if dateString.lower()=='today':
808
        return date.today()
809
    if dateString.lower()=='yesterday':
810
        return date.today() - timedelta(days=1)
811
    try:
812
        return datetime.strptime(dateString, '%d-%b-%y')
813
    except:
814
        try:
815
            #get Closest Date from today
816
            curDate = date.today()
817
            curTime = datetime(curDate.year, curDate.month, curDate.day)
818
            curYear = curDate.year
819
            prevYear = curYear - 1
17254 amit.gupta 820
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
821
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 822
            if dateMax <= curTime:
823
                return dateMax
824
            else:
825
                return dateMin 
826
        except:
17254 amit.gupta 827
            try:
828
                days_of_week = ['sunday','monday','tuesday','wednesday',
829
                            'thursday','friday','saturday']
830
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
831
                if deltaDays <= 0:
832
                    deltaDays= deltaDays + 7
833
                curDate = curDate - timedelta(days=deltaDays)
834
                print datetime.strftime(curDate, '%d-%b-%y')
835
                return curDate
836
            except:
837
                print "could not parse"
838
                return None
15565 amit.gupta 839
 
17254 amit.gupta 840
def getDateStringArriving(dateString='Thursday'):
841
    print dateString
842
    if dateString.lower()=='today':
843
        return date.today()
844
    if dateString.lower()=='tomorrow':
845
        return date.today() + timedelta(days=1)
846
    try:
847
        return datetime.strptime(dateString, '%d-%b-%y')
848
    except:
849
        try:
850
            #get Closest Date from today
851
            curDate = date.today()
852
            curTime = datetime(curDate.year, curDate.month, curDate.day)
853
            curYear = curDate.year
854
            nextYear = curYear + 1
855
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
856
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
857
            if dateMin >= curTime:
858
                return dateMin
859
            else:
860
                return dateMax 
861
        except:
862
            try:
863
                days_of_week = ['sunday','monday','tuesday','wednesday',
864
                            'thursday','friday','saturday']
865
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
866
                if deltaDays < 0:
867
                    deltaDays= deltaDays + 7
868
                curDate = curDate + timedelta(days=deltaDays)
869
                return curDate
870
            except:
871
                print "Could not parse"
872
                return None 
873
 
15565 amit.gupta 874
def parseOrderNotCreated():
875
    try:
876
        store=getStore(1)
877
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
878
        session.close()
879
        for order in orders:
15567 amit.gupta 880
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
881
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
882
            order1.status = result
883
            session.commit()
15565 amit.gupta 884
    finally:
885
        session.close()
15555 amit.gupta 886
 
13774 amit.gupta 887
if __name__ == '__main__':
17599 amit.gupta 888
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
889
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
890
#    readSSh("/AmazonTrack/User10466/403-7498756-0837158")
891
    main()