Subversion Repositories SmartDukaan

Rev

Rev 17705 | Rev 17715 | 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:
17665 amit.gupta 128
                        print "soup.prettify()"
129
                        print soup.prettify()
17319 amit.gupta 130
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 131
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 132
                    except:
15056 amit.gupta 133
                        try:
17319 amit.gupta 134
                            traceback.print_exc()
135
                            self.parseOldStlye(merchantOrder, soup)
136
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 137
                        except:
17319 amit.gupta 138
                            traceback.print_exc()
15059 amit.gupta 139
                            try:
17319 amit.gupta 140
                                self.parseCancelled(merchantOrder, soup)
141
                                resp['result'] = 'ORDER_CANCELLED'
15059 amit.gupta 142
                            except:
17319 amit.gupta 143
                                try:
17356 amit.gupta 144
                                    if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
17319 amit.gupta 145
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
146
                                    else:
147
                                        raise
148
                                except:
17356 amit.gupta 149
                                    if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
17319 amit.gupta 150
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
151
                                    else:
152
                                        raise
17599 amit.gupta 153
                if resp['result'] == 'DETAIL_NOT_CREATED_KNOWN':
17357 amit.gupta 154
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 155
            except:
17666 amit.gupta 156
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
14813 amit.gupta 157
                print "Error occurred"
15046 amit.gupta 158
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 159
                traceback.print_exc()
15046 amit.gupta 160
        return resp    
13774 amit.gupta 161
 
13868 amit.gupta 162
    #This should be exposed from api for specific sources
163
    def scrapeStoreOrders(self):
14624 amit.gupta 164
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
165
        for merchantOrder in orders:
14981 amit.gupta 166
            executeBulk = False
14624 amit.gupta 167
            try:
168
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
169
                closed = True
170
                map1 = {}
171
                for subOrder in merchantOrder.get("subOrders"):
172
                    if subOrder.get("closed"):
173
                        continue
174
                    elif subOrder.get("trackingUrl") is None:
175
                        closed = False
176
                        continue
177
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
178
                    trackingUrl = subOrder.get("trackingUrl")
179
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 180
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 181
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 182
                    if newOrder:
183
                        executeBulk = True
184
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
185
                        print findMap, "\n", updateMap
186
                        bulk.find(findMap).update({'$set' : updateMap})
187
                        closed = closed and newOrder['closed']
14981 amit.gupta 188
                if executeBulk:
14697 amit.gupta 189
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 190
                    bulk.execute()
14624 amit.gupta 191
            except:
14847 amit.gupta 192
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 193
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 194
                traceback.print_exc()
195
 
14608 amit.gupta 196
 
197
 
14771 amit.gupta 198
    def parserest(self, soup):
199
        print "Hi"
200
        if soup.find('h1'):
201
            print "OK"
14608 amit.gupta 202
 
14464 amit.gupta 203
    def parseOldStlye(self, merchantOrder, soup):
204
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
205
        table = soup.body.findAll("table", recursive=False)[1]
206
        #print table
207
        tables = table.tr.td.findAll("table", recursive=False)
208
        for tr in tables[2].findAll("tr"):
209
            boldElement = tr.td.b
210
            if "Order Placed" in str(boldElement):
211
                merchantOrder.placedOn = boldElement.next_sibling.strip()
212
            if "order number" in str(boldElement):
213
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
214
            if "Order Total" in str(boldElement):
215
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
216
        anchors = table.tr.td.findAll("a", recursive=False)
217
        paymentAnchor = anchors.pop(-1)
218
 
219
        count = 0
220
        subOrders = []
221
        merchantOrder.subOrders = subOrders
222
        counter = 0 
223
        for anchor in anchors:
224
            count += 1
225
            tab = anchor.next_sibling
226
            status = MStore.ORDER_PLACED
227
            subStr = "Delivery #" + str(count) + ":"
228
            if subStr in  tab.find("b").text:
229
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
230
 
231
            tab = tab.next_sibling.next_sibling
232
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
233
 
234
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
235
 
236
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
237
            i = -1
238
            for orderItemTr in orderItemTrs:
239
                i += 1
240
                if i%2 == 0:
241
                    continue
242
                counter += 1
243
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
244
 
245
                productUrl = orderItemTr.td.contents[1].a["href"]
246
                productTitle = orderItemTr.td.contents[1].a.text
247
 
248
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
249
 
250
 
251
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
252
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
253
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 254
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
255
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
256
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 257
                subOrder.productCode = productUrl.split('/')[5]
258
                subOrder.detailedStatus = detailedStatus
259
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice*quantity)
15346 amit.gupta 260
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
261
                subOrder.dealRank = dealRank.get('rank')
262
                subOrder.rankDesc = dealRank.get('description')
16283 amit.gupta 263
                subOrder.maxNlc = dealRank.get('maxNlc')
264
                subOrder.minNlc = dealRank.get('minNlc')
265
                subOrder.db = dealRank.get('dp')
266
                subOrder.itemStatus = dealRank.get('status')
14464 amit.gupta 267
                cashbackStatus = Store.CB_PENDING
268
                if cashbackAmount <= 0:
269
                    cashbackStatus = Store.CB_NA
270
                subOrder.cashBackStatus = cashbackStatus
271
                subOrder.cashBackAmount = cashbackAmount
272
                if percentage > 0:
273
                    subOrder.cashBackPercentage = percentage
274
                subOrders.append(subOrder)
275
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
276
        totalAmount = 0
277
        grandAmount = 0
278
        for price in priceList:
279
            labelTd = price.td
280
            if 'Subtotal:' in labelTd.text:
281
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
282
            elif 'Grand Total:' in labelTd.text:
283
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
284
        if grandAmount < totalAmount:
285
            diff = totalAmount - grandAmount
286
            for subOrder in merchantOrder.subOrders:
287
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 288
        merchantOrder.status='success'
14464 amit.gupta 289
        self._updateToOrder(todict(merchantOrder))
290
 
291
    def parseNewStlye(self, merchantOrder, soup):
292
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
293
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 294
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
295
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 296
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
297
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
298
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 299
        try:
300
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
301
        except:
302
            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 303
        priceRows = priceBox.findAll('div', {'class':'a-row'})
304
        subTotal = 0
14566 amit.gupta 305
        shippingPrice = 0   
14464 amit.gupta 306
        promoApplied = 0
307
        for priceRow in priceRows:
308
            if "Item(s) Subtotal:" in str(priceRow):
309
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
310
            elif "Shipping:" in str(priceRow):
311
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
312
            elif "Grand Total:" in str(priceRow):
313
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
314
                merchantOrder.paidAmount = grandPrice
315
            elif "Total:" in str(priceRow):
316
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
317
            elif "Promotion Applied:" in str(priceRow):
318
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
319
        totalPaid = subTotal        
320
        if promoApplied > 0:
321
            totalPaid -= promoApplied
322
            if shippingPrice <= promoApplied:
323
                totalPaid  += shippingPrice
15473 amit.gupta 324
 
17601 amit.gupta 325
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 326
        subOrders = []
327
        merchantOrder.subOrders = subOrders
15473 amit.gupta 328
        closedStatus = True
17601 amit.gupta 329
        subOrders = []
14464 amit.gupta 330
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 331
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 332
            try:
333
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
334
            except:
335
                trackingUrl= None
336
 
17347 amit.gupta 337
            deliverySpanTop = None
14464 amit.gupta 338
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
339
            statusDiv = innerBoxes[0]
340
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 341
            try:
342
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
343
            except:
17617 amit.gupta 344
                pass
345
            productDivs = innerBoxes[-1].div.div.div.findAll('div', recursive=False)
14464 amit.gupta 346
            merchantOrder.subOrders = subOrders
347
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 348
                deliverySpan = deliverySpanTop
349
                j=i+1
350
                if not deliverySpanTop:
351
                    if i%2==1:
352
                        continue
353
                    j=i/2 + 1
17599 amit.gupta 354
                    deliverySpan = statusDiv.div.div.span
14464 amit.gupta 355
                imgDiv  = productDiv.div.div
356
                detailDiv = imgDiv.find_next_sibling('div')
357
                detailDivs = detailDiv.findAll('div', recursive=False)
358
                arr = detailDivs[0].a.text.strip().split(" of ", 1)
14465 amit.gupta 359
                (productTitle, quantity) = (arr[-1], (1 if len(arr)==1 else int(arr[0])) )
14820 amit.gupta 360
                try:
361
                    unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
362
                except:
363
                    unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
14464 amit.gupta 364
                amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
365
                productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
366
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
367
                subOrder.productCode = productUrl.split('/')[5]
368
                subOrder.unitPrice = unitPrice
17347 amit.gupta 369
                subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
14722 amit.gupta 370
                estDlvyTime = datetime.now()
14566 amit.gupta 371
                if deliverySpan is not None:
14756 amit.gupta 372
                    try:
373
                        subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
374
                        estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
375
                        subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
376
                        estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
377
                    except:
15473 amit.gupta 378
                        if "Delivered on" in deliverySpan.text:
379
                            subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
14756 amit.gupta 380
                        subOrder.estimatedDeliveryDate = "Not available"
15473 amit.gupta 381
 
14722 amit.gupta 382
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
14756 amit.gupta 383
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 384
                subOrder.detailedStatus = subOrderStatus
17599 amit.gupta 385
                status=MStore.ORDER_PLACED
386
                try:
387
                    status = self._getStatusFromDetailedStatus(subOrderStatus)
388
                except:
389
                    try:
390
                        dateString = subOrderStatus.split("Delivered ")[1].strip()
391
                        subOrder.status = MStore.ORDER_DELIVERED
392
                        subOrder.detailedStatus = 'Delivered'
17631 amit.gupta 393
                        dateString = getDateStringDelivered(dateString)
394
                        if dateString is not None:
395
                            subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 396
                    except:
397
                        try:
398
                            dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
399
                            subOrder.status = MStore.ORDER_SHIPPED
17631 amit.gupta 400
                            dateString = getDateStringArriving(dateString)
401
                            if dateString is not None:
402
                                subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
17641 amit.gupta 403
                            else:
404
                                subOrder.deliveryEstimate = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
17599 amit.gupta 405
                        except:
406
                            print "Unknown status Alert -", status
407
 
14464 amit.gupta 408
                subOrder.deliveryCharges = shippingPrice
17361 amit.gupta 409
                if trackingUrl:
410
                    subOrder.tracingkUrl  = trackingUrl
14466 amit.gupta 411
                subOrder.imgUrl = imgDiv.img["src"]
17704 amit.gupta 412
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid/unitPrice)
15346 amit.gupta 413
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
414
                subOrder.dealRank = dealRank.get('rank')
415
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 416
                cashbackStatus = Store.CB_PENDING
417
                if cashbackAmount <= 0:
418
                    cashbackStatus = Store.CB_NA
419
                subOrder.cashBackStatus = cashbackStatus
17704 amit.gupta 420
                subOrder.cashBackAmount = cashbackAmount*quantity
14464 amit.gupta 421
                if percentage > 0:
422
                    subOrder.cashBackPercentage = percentage
17599 amit.gupta 423
                if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
15473 amit.gupta 424
                    subOrder.status = Store.ORDER_DELIVERED
425
                    subOrder.closed = True
426
                    if subOrder.cashBackStatus == Store.CB_PENDING:
427
                        subOrder.cashBackStatus = Store.CB_APPROVED
428
                elif closedStatus:
429
                    closedStatus= False
14464 amit.gupta 430
                subOrders.append(subOrder)
14698 amit.gupta 431
        merchantOrder.status='success'
15473 amit.gupta 432
        merchantOrder.closed = closedStatus
14464 amit.gupta 433
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 434
 
435
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 436
        try:
437
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
438
            if fonts[0].text == "Important Message":
439
                if fonts[1].text=="This order has been cancelled.":
440
                    merchantOrder.closed = True
441
                    merchantOrder.status = "cancelled"
442
                    merchantOrder.requireDetail = False
443
                    self._updateToOrder(todict(merchantOrder))
444
                    return
445
                else:
446
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 447
            else:
448
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 449
        except:
14549 amit.gupta 450
            orderDetails = soup.body.find(id="orderDetails")
451
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
452
                merchantOrder.closed = True
14698 amit.gupta 453
                merchantOrder.status = "cancelled"
14549 amit.gupta 454
                merchantOrder.requireDetail = False
455
                self._updateToOrder(todict(merchantOrder))
456
            else:
457
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 458
 
13868 amit.gupta 459
    def getTrackingUrls(self, userId):
14074 amit.gupta 460
 
461
        missingOrderUrls = []
462
        missingOrders = self._getMissingOrders({'userId':userId})
463
        for missingOrder in missingOrders:
464
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 465
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 466
        count = len(orders)
13879 amit.gupta 467
        print "count", count
14948 amit.gupta 468
        print "Missing Urls"
469
        print "*************"
470
        print missingOrderUrls
13868 amit.gupta 471
        if count > 0:
14684 amit.gupta 472
            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 473
        else: 
14074 amit.gupta 474
            return missingOrderUrls
13927 amit.gupta 475
 
476
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 477
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 478
        if not os.path.exists(directory):
479
            os.makedirs(directory)
14173 amit.gupta 480
 
14945 amit.gupta 481
 
13995 amit.gupta 482
        try:
483
            searchMap = {'userId':userId}
484
            collectionMap = {'merchantOrderId':1}
485
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 486
            datetimeNow = datetime.now()
487
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 488
            print "url----------------", url
17465 amit.gupta 489
            cancelledSummary = False
14320 amit.gupta 490
            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 491
                if url == 'https://www.amazon.in/gp/css/order-history':
492
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
493
                else:   
17465 amit.gupta 494
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
495
                    cancelledSummary = True   
14945 amit.gupta 496
                f = open(filename,'w')
497
                f.write(rawHtml) # python will convert \n to os.linesep
498
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 499
                soup = BeautifulSoup(rawHtml)
500
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
501
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
502
                for activeOrder in activeOrders:
17307 amit.gupta 503
                    matched=False
13995 amit.gupta 504
                    for orderEle in allOrders:
17254 amit.gupta 505
                        deliveredOn = None
506
                        deliveryEstimate = None
17258 amit.gupta 507
                        shippingEstimate = None
13995 amit.gupta 508
                        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'})
509
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
510
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 511
                            matched=True
13995 amit.gupta 512
                            closed = True
17465 amit.gupta 513
                            if not cancelledSummary:
514
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
515
                            else:
516
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
517
                                shipments.pop(0)
13995 amit.gupta 518
                            for shipment in shipments:
17307 amit.gupta 519
                                orderStatusDesc = None
13995 amit.gupta 520
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 521
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 522
                                try:
17270 amit.gupta 523
                                    orderStatus = sdivs[0].span.text.strip()
524
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 525
                                except:
17270 amit.gupta 526
                                    try:
527
                                        dateString = orderStatus.split("Delivered ")[1].strip()
528
                                        status = MStore.ORDER_DELIVERED
529
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
530
                                    except:
531
                                        try:
17271 amit.gupta 532
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 533
                                            status = MStore.ORDER_SHIPPED
534
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
535
                                        except:
536
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 537
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 538
                                try:
17270 amit.gupta 539
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
540
                                except:
17271 amit.gupta 541
                                    try:
542
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
543
                                    except:
17307 amit.gupta 544
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 545
 
17271 amit.gupta 546
                                if orderStatusDesc: 
17348 amit.gupta 547
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 548
                                    try:
17348 amit.gupta 549
                                        status = self._getStatusFromDetailedStatus(orderStatus)
550
                                    except:
551
                                        pass
552
                                    try:
17271 amit.gupta 553
                                        if "Dispatch estimate" in orderStatusDesc:
554
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
555
                                        elif "Delivery estimate" in orderStatus:
556
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
557
                                        elif "Arriving" in orderStatus:
558
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
559
                                    except:
560
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
561
                                        closed=False
562
                                        status = None
17258 amit.gupta 563
 
14335 amit.gupta 564
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 565
                                trackingUrl = None
566
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
567
                                    if buttonDiv.find('a').text.strip()=='Track package':
568
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
569
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 570
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 571
                                        break
572
                                for prodDiv in productDivs:
573
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
574
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
575
                                    imgUrl = prodDiv.find("img")["src"]
576
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 577
                                        if subOrder['closed']==True:
578
                                            continue
17230 amit.gupta 579
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 580
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
581
                                            updateMap = {}
582
                                            closedStatus = False
583
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 584
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 585
                                            if status:
17348 amit.gupta 586
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 587
                                                updateMap['subOrders.$.status'] = status 
588
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 589
 
17251 amit.gupta 590
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 591
                                                if deliveredOn:                               
17254 amit.gupta 592
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 593
                                                closedStatus = True
594
                                                updateMap['subOrders.$.closed'] = True
595
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 596
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 597
                                            if status==MStore.ORDER_CANCELLED:     
598
                                                closedStatus = True
599
                                                updateMap['subOrders.$.closed'] = True
600
                                                if cashbackStatus == Store.CB_PENDING:
601
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 602
                                            if status==MStore.ORDER_SHIPPED:
603
                                                if deliveryEstimate:   
604
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 605
                                                if trackingUrl is not None:
606
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 607
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 608
                                            if shippingEstimate:   
609
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 610
                                            if not closedStatus:
611
                                                closed = False
17307 amit.gupta 612
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 613
                                            updateMap['status']='success'
13995 amit.gupta 614
                                            bulk.find(findMap).update({'$set' : updateMap})
615
                                            break
17321 amit.gupta 616
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 617
                            break
618
                    if not matched:
17319 amit.gupta 619
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 620
                        for subOrder in activeOrder['subOrders']:
621
                            if subOrder['closed']==True:
622
                                continue
623
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
624
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
625
                            bulk.find({'orderId': activeOrder['orderId']})
626
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 627
                bulk.execute()
628
                return 'PARSED_SUCCESS'
629
            else:
15563 amit.gupta 630
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
631
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 632
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 633
 
634
                filename = directory + "/" + merchantOrderId   
635
                f = open(filename,'w')
636
                f.write(rawHtml) # python will convert \n to os.linesep
637
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 638
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 639
                print "result", result
15555 amit.gupta 640
                try:
15560 amit.gupta 641
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
642
                    order1.status = result
643
                    order1.rawhtml = rawHtml
17358 amit.gupta 644
                    order1.order_url = url
15555 amit.gupta 645
                    session.commit()
15561 amit.gupta 646
                except:
647
                    traceback.print_exc()
15555 amit.gupta 648
                finally:
649
                    session.close()
14080 amit.gupta 650
                return 'PARSED_SUCCESS'
651
                pass
14008 amit.gupta 652
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 653
        except:
654
            traceback.print_exc()    
655
            return 'PARSED_FAILED'
656
 
657
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 658
        if "ordered from" in detailedStatus.lower():
659
            return MStore.ORDER_PLACED 
660
 
13995 amit.gupta 661
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 662
            if detailedStatus.lower() in value:
13995 amit.gupta 663
                return key
664
 
17253 amit.gupta 665
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 666
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 667
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 668
        br = getBrowserObject()
669
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
670
        url = AMAZON_AFF_URL
671
        response = br.open(url)
672
        #get data for past 40 days and store it to mongo
673
        dt = datetime.now()
14701 amit.gupta 674
        dat = dt - timedelta(days=2)
14651 amit.gupta 675
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
676
        response = br.open(url)
677
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
678
        j=-1
679
        for row in page.split("\n"):
680
            j += 1
681
            if j== 0 or j==1:
682
                continue
683
            fields = row.split("\t")
684
            if len(fields)>1:
685
                print fields
686
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
687
                print amazonAffiliate
688
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
689
            else:
690
                break
14650 amit.gupta 691
 
14608 amit.gupta 692
 
16468 amit.gupta 693
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 694
        print trackingUrl
14624 amit.gupta 695
        subOrder = {}
14749 amit.gupta 696
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 697
        status = MStore.ORDER_SHIPPED
698
        #print page
14624 amit.gupta 699
        soup = BeautifulSoup(page)
16468 amit.gupta 700
        header1 = soup.find("h1") 
701
        if header1:
702
            if header1.text=="Sign In":
703
                print "Login page is displayed for order id", orderId
17307 amit.gupta 704
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 705
                return subOrder
16468 amit.gupta 706
        try:
16474 amit.gupta 707
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 708
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
709
            try:
710
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
711
            except:
712
                displayStatus = detailedStatus
713
            print displayStatus
714
            if detailedStatus.lower().find("delivered")>=0:
715
                print detailedStatus
716
                displayStatus = "Delivered"
717
                status = "Delivered"
718
                try:
719
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
720
                except:
721
                    pass
722
 
723
            elif detailedStatus.lower() == 'returned':
724
                status = 'Cancelled'
725
            subOrder['status'] = status
726
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 727
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 728
 
729
        except:
17350 amit.gupta 730
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 731
            print "failed to parse", orderId
732
            traceback.print_exc()
733
 
14624 amit.gupta 734
        return subOrder
735
 
736
 
737
 
16980 amit.gupta 738
    def parseInfo(self,):
739
        from pyquery import PyQuery as pq
16986 amit.gupta 740
        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 741
        try:
742
            for order in orders:
743
                try:
744
                    doc = pq(order.rawhtml)
745
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
746
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 747
                    orderInfo = All_user_addresses()
16980 amit.gupta 748
                    orderInfo.address = lists[-3].text
749
                    orderInfo.user_id = order.user_id
16986 amit.gupta 750
                    orderInfo.source = 'order' 
751
                    #orderInfo.order_id = order.id
752
                    #orderInfo.email = None
753
                    #orderInfo.name = lists[0].text 
754
                    #orderInfo.mobile = None
16980 amit.gupta 755
                    adSplit = lists[-2].text.split(",")
756
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
757
                    if match:
758
                        items = match.groups()
16986 amit.gupta 759
                    orderInfo.city = adSplit[0].strip()
760
                    orderInfo.pincode = items[1].strip()
761
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 762
                    session.commit()
763
                except:
764
                    session.rollback()
765
                    continue
766
        finally:
767
            session.close()
14608 amit.gupta 768
 
13927 amit.gupta 769
 
770
 
13774 amit.gupta 771
def main():
16874 amit.gupta 772
    store = getStore(1)
17641 amit.gupta 773
    #[59433, 59472, 59487, 59503, 59545, 59549, 59573, 59591]
17709 amit.gupta 774
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
775
    session.close()
17704 amit.gupta 776
    for o in orders:
17602 amit.gupta 777
        try:
17670 amit.gupta 778
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
17602 amit.gupta 779
        finally:
780
            session.close()
17631 amit.gupta 781
    #store.trackOrdersForUser(10466, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-7498756-0837158', readSSh('/AmazonTrack/User10466/403-7498756-0837158'))
782
    #store.trackOrdersForUser(9031, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-0812780-9465164', readSSh('/home/amit/sample.html'))
17231 amit.gupta 783
 
784
def getSummaryFile(directory):
785
    date1 = datetime(2015,1,1)
786
    finalFile = None
787
    try:
788
        for file in os.listdir(directory):
789
            if file.startswith("orderSummary"):
790
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
791
                if date2 > date1:
792
                    date1 = date2
793
                    finalFile=file
794
    except:
795
        print "Missing directory"
17251 amit.gupta 796
    return finalFile 
17231 amit.gupta 797
 
15555 amit.gupta 798
 
799
def parseDetailNotCreated():
800
    try:
801
        store=getStore(1)
802
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 803
        session.close()
15555 amit.gupta 804
        for order in orders:
805
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
806
 
807
    finally:
808
        session.close()
17251 amit.gupta 809
 
810
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 811
    print dateString
17251 amit.gupta 812
    if dateString.lower()=='today':
813
        return date.today()
814
    if dateString.lower()=='yesterday':
815
        return date.today() - timedelta(days=1)
816
    try:
817
        return datetime.strptime(dateString, '%d-%b-%y')
818
    except:
819
        try:
820
            #get Closest Date from today
821
            curDate = date.today()
822
            curTime = datetime(curDate.year, curDate.month, curDate.day)
823
            curYear = curDate.year
824
            prevYear = curYear - 1
17254 amit.gupta 825
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
826
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 827
            if dateMax <= curTime:
828
                return dateMax
829
            else:
830
                return dateMin 
831
        except:
17254 amit.gupta 832
            try:
833
                days_of_week = ['sunday','monday','tuesday','wednesday',
834
                            'thursday','friday','saturday']
835
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
836
                if deltaDays <= 0:
837
                    deltaDays= deltaDays + 7
838
                curDate = curDate - timedelta(days=deltaDays)
839
                print datetime.strftime(curDate, '%d-%b-%y')
840
                return curDate
841
            except:
842
                print "could not parse"
843
                return None
15565 amit.gupta 844
 
17254 amit.gupta 845
def getDateStringArriving(dateString='Thursday'):
846
    print dateString
847
    if dateString.lower()=='today':
848
        return date.today()
849
    if dateString.lower()=='tomorrow':
850
        return date.today() + timedelta(days=1)
851
    try:
852
        return datetime.strptime(dateString, '%d-%b-%y')
853
    except:
854
        try:
855
            #get Closest Date from today
856
            curDate = date.today()
857
            curTime = datetime(curDate.year, curDate.month, curDate.day)
858
            curYear = curDate.year
859
            nextYear = curYear + 1
860
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
861
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
862
            if dateMin >= curTime:
863
                return dateMin
864
            else:
865
                return dateMax 
866
        except:
867
            try:
868
                days_of_week = ['sunday','monday','tuesday','wednesday',
869
                            'thursday','friday','saturday']
870
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
871
                if deltaDays < 0:
872
                    deltaDays= deltaDays + 7
873
                curDate = curDate + timedelta(days=deltaDays)
874
                return curDate
875
            except:
876
                print "Could not parse"
877
                return None 
878
 
15565 amit.gupta 879
def parseOrderNotCreated():
880
    try:
881
        store=getStore(1)
882
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
883
        session.close()
884
        for order in orders:
15567 amit.gupta 885
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
886
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
887
            order1.status = result
888
            session.commit()
15565 amit.gupta 889
    finally:
890
        session.close()
15555 amit.gupta 891
 
13774 amit.gupta 892
if __name__ == '__main__':
17599 amit.gupta 893
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
894
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
17665 amit.gupta 895
    #readSSh("/AmazonTrack/User2482/orderSummary20-11:09:17:47")
896
    main()