Subversion Repositories SmartDukaan

Rev

Rev 17599 | Rev 17601 | 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
 
15474 amit.gupta 323
        shipmentDivs = orderDetailsContainer.find('div', class_='shipment').findAll('div', recursive = False)
14464 amit.gupta 324
        subOrders = []
325
        merchantOrder.subOrders = subOrders
15473 amit.gupta 326
        closedStatus = True
14464 amit.gupta 327
        for shipmentDiv in shipmentDivs:
17361 amit.gupta 328
            try:
329
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
330
            except:
331
                trackingUrl= None
332
 
17347 amit.gupta 333
            deliverySpanTop = None
14464 amit.gupta 334
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
335
            statusDiv = innerBoxes[0]
336
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 337
            try:
338
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
339
                productDivs = innerBoxes[-1].div.div.div.findAll('div', recursive=False)
340
            except:
341
                productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 342
            subOrders = []
343
            merchantOrder.subOrders = subOrders
344
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 345
                deliverySpan = deliverySpanTop
346
                j=i+1
347
                if not deliverySpanTop:
348
                    if i%2==1:
349
                        continue
350
                    j=i/2 + 1
17599 amit.gupta 351
                    deliverySpan = statusDiv.div.div.span
14464 amit.gupta 352
                imgDiv  = productDiv.div.div
353
                detailDiv = imgDiv.find_next_sibling('div')
354
                detailDivs = detailDiv.findAll('div', recursive=False)
355
                arr = detailDivs[0].a.text.strip().split(" of ", 1)
14465 amit.gupta 356
                (productTitle, quantity) = (arr[-1], (1 if len(arr)==1 else int(arr[0])) )
14820 amit.gupta 357
                try:
358
                    unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
359
                except:
360
                    unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
14464 amit.gupta 361
                amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
362
                productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
363
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
364
                subOrder.productCode = productUrl.split('/')[5]
365
                subOrder.unitPrice = unitPrice
17347 amit.gupta 366
                subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
14722 amit.gupta 367
                estDlvyTime = datetime.now()
14566 amit.gupta 368
                if deliverySpan is not None:
14756 amit.gupta 369
                    try:
370
                        subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
371
                        estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
372
                        subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
373
                        estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
374
                    except:
15473 amit.gupta 375
                        if "Delivered on" in deliverySpan.text:
376
                            subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
14756 amit.gupta 377
                        subOrder.estimatedDeliveryDate = "Not available"
15473 amit.gupta 378
 
14722 amit.gupta 379
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
14756 amit.gupta 380
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 381
                subOrder.detailedStatus = subOrderStatus
17599 amit.gupta 382
                status=MStore.ORDER_PLACED
383
                try:
384
                    status = self._getStatusFromDetailedStatus(subOrderStatus)
385
                except:
386
                    try:
387
                        dateString = subOrderStatus.split("Delivered ")[1].strip()
388
                        subOrder.status = MStore.ORDER_DELIVERED
389
                        subOrder.detailedStatus = 'Delivered'
390
                        subOrder.deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
391
                    except:
392
                        try:
393
                            dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
394
                            subOrder.status = MStore.ORDER_SHIPPED
395
                            subOrder.deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
396
                        except:
397
                            print "Unknown status Alert -", status
398
 
14464 amit.gupta 399
                subOrder.deliveryCharges = shippingPrice
17361 amit.gupta 400
                if trackingUrl:
401
                    subOrder.tracingkUrl  = trackingUrl
14466 amit.gupta 402
                subOrder.imgUrl = imgDiv.img["src"]
14464 amit.gupta 403
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid)
15346 amit.gupta 404
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
405
                subOrder.dealRank = dealRank.get('rank')
406
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 407
                cashbackStatus = Store.CB_PENDING
408
                if cashbackAmount <= 0:
409
                    cashbackStatus = Store.CB_NA
410
                subOrder.cashBackStatus = cashbackStatus
411
                subOrder.cashBackAmount = cashbackAmount
412
                if percentage > 0:
413
                    subOrder.cashBackPercentage = percentage
17599 amit.gupta 414
                if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
15473 amit.gupta 415
                    subOrder.status = Store.ORDER_DELIVERED
416
                    subOrder.closed = True
417
                    if subOrder.cashBackStatus == Store.CB_PENDING:
418
                        subOrder.cashBackStatus = Store.CB_APPROVED
419
                elif closedStatus:
420
                    closedStatus= False
14464 amit.gupta 421
                subOrders.append(subOrder)
14698 amit.gupta 422
        merchantOrder.status='success'
15473 amit.gupta 423
        merchantOrder.closed = closedStatus
14464 amit.gupta 424
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 425
 
426
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 427
        try:
428
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
429
            if fonts[0].text == "Important Message":
430
                if fonts[1].text=="This order has been cancelled.":
431
                    merchantOrder.closed = True
432
                    merchantOrder.status = "cancelled"
433
                    merchantOrder.requireDetail = False
434
                    self._updateToOrder(todict(merchantOrder))
435
                    return
436
                else:
437
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 438
            else:
439
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 440
        except:
14549 amit.gupta 441
            orderDetails = soup.body.find(id="orderDetails")
442
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
443
                merchantOrder.closed = True
14698 amit.gupta 444
                merchantOrder.status = "cancelled"
14549 amit.gupta 445
                merchantOrder.requireDetail = False
446
                self._updateToOrder(todict(merchantOrder))
447
            else:
448
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 449
 
13868 amit.gupta 450
    def getTrackingUrls(self, userId):
14074 amit.gupta 451
 
452
        missingOrderUrls = []
453
        missingOrders = self._getMissingOrders({'userId':userId})
454
        for missingOrder in missingOrders:
455
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 456
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 457
        count = len(orders)
13879 amit.gupta 458
        print "count", count
14948 amit.gupta 459
        print "Missing Urls"
460
        print "*************"
461
        print missingOrderUrls
13868 amit.gupta 462
        if count > 0:
14684 amit.gupta 463
            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 464
        else: 
14074 amit.gupta 465
            return missingOrderUrls
13927 amit.gupta 466
 
467
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 468
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 469
        if not os.path.exists(directory):
470
            os.makedirs(directory)
14173 amit.gupta 471
 
14945 amit.gupta 472
 
13995 amit.gupta 473
        try:
474
            searchMap = {'userId':userId}
475
            collectionMap = {'merchantOrderId':1}
476
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 477
            datetimeNow = datetime.now()
478
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 479
            print "url----------------", url
17465 amit.gupta 480
            cancelledSummary = False
14320 amit.gupta 481
            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 482
                if url == 'https://www.amazon.in/gp/css/order-history':
483
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
484
                else:   
17465 amit.gupta 485
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
486
                    cancelledSummary = True   
14945 amit.gupta 487
                f = open(filename,'w')
488
                f.write(rawHtml) # python will convert \n to os.linesep
489
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 490
                soup = BeautifulSoup(rawHtml)
491
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
492
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
493
                for activeOrder in activeOrders:
17307 amit.gupta 494
                    matched=False
13995 amit.gupta 495
                    for orderEle in allOrders:
17254 amit.gupta 496
                        deliveredOn = None
497
                        deliveryEstimate = None
17258 amit.gupta 498
                        shippingEstimate = None
13995 amit.gupta 499
                        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'})
500
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
501
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 502
                            matched=True
13995 amit.gupta 503
                            closed = True
17465 amit.gupta 504
                            if not cancelledSummary:
505
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
506
                            else:
507
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
508
                                shipments.pop(0)
13995 amit.gupta 509
                            for shipment in shipments:
17307 amit.gupta 510
                                orderStatusDesc = None
13995 amit.gupta 511
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 512
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 513
                                try:
17270 amit.gupta 514
                                    orderStatus = sdivs[0].span.text.strip()
515
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 516
                                except:
17270 amit.gupta 517
                                    try:
518
                                        dateString = orderStatus.split("Delivered ")[1].strip()
519
                                        status = MStore.ORDER_DELIVERED
520
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
521
                                    except:
522
                                        try:
17271 amit.gupta 523
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 524
                                            status = MStore.ORDER_SHIPPED
525
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
526
                                        except:
527
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 528
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 529
                                try:
17270 amit.gupta 530
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
531
                                except:
17271 amit.gupta 532
                                    try:
533
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
534
                                    except:
17307 amit.gupta 535
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 536
 
17271 amit.gupta 537
                                if orderStatusDesc: 
17348 amit.gupta 538
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 539
                                    try:
17348 amit.gupta 540
                                        status = self._getStatusFromDetailedStatus(orderStatus)
541
                                    except:
542
                                        pass
543
                                    try:
17271 amit.gupta 544
                                        if "Dispatch estimate" in orderStatusDesc:
545
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
546
                                        elif "Delivery estimate" in orderStatus:
547
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
548
                                        elif "Arriving" in orderStatus:
549
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
550
                                    except:
551
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
552
                                        closed=False
553
                                        status = None
17258 amit.gupta 554
 
14335 amit.gupta 555
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 556
                                trackingUrl = None
557
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
558
                                    if buttonDiv.find('a').text.strip()=='Track package':
559
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
560
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 561
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 562
                                        break
563
                                for prodDiv in productDivs:
564
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
565
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
566
                                    imgUrl = prodDiv.find("img")["src"]
567
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 568
                                        if subOrder['closed']==True:
569
                                            continue
17230 amit.gupta 570
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 571
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
572
                                            updateMap = {}
573
                                            closedStatus = False
574
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 575
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 576
                                            if status:
17348 amit.gupta 577
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 578
                                                updateMap['subOrders.$.status'] = status 
579
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 580
 
17251 amit.gupta 581
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 582
                                                if deliveredOn:                               
17254 amit.gupta 583
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 584
                                                closedStatus = True
585
                                                updateMap['subOrders.$.closed'] = True
586
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 587
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 588
                                            if status==MStore.ORDER_CANCELLED:     
589
                                                closedStatus = True
590
                                                updateMap['subOrders.$.closed'] = True
591
                                                if cashbackStatus == Store.CB_PENDING:
592
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 593
                                            if status==MStore.ORDER_SHIPPED:
594
                                                if deliveryEstimate:   
595
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 596
                                                if trackingUrl is not None:
597
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 598
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 599
                                            if shippingEstimate:   
600
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 601
                                            if not closedStatus:
602
                                                closed = False
17307 amit.gupta 603
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 604
                                            updateMap['status']='success'
13995 amit.gupta 605
                                            bulk.find(findMap).update({'$set' : updateMap})
606
                                            break
17321 amit.gupta 607
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 608
                            break
609
                    if not matched:
17319 amit.gupta 610
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 611
                        for subOrder in activeOrder['subOrders']:
612
                            if subOrder['closed']==True:
613
                                continue
614
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
615
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
616
                            bulk.find({'orderId': activeOrder['orderId']})
617
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 618
                bulk.execute()
619
                return 'PARSED_SUCCESS'
620
            else:
15563 amit.gupta 621
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
622
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 623
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 624
 
625
                filename = directory + "/" + merchantOrderId   
626
                f = open(filename,'w')
627
                f.write(rawHtml) # python will convert \n to os.linesep
628
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 629
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 630
                print "result", result
15555 amit.gupta 631
                try:
15560 amit.gupta 632
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
633
                    order1.status = result
634
                    order1.rawhtml = rawHtml
17358 amit.gupta 635
                    order1.order_url = url
15555 amit.gupta 636
                    session.commit()
15561 amit.gupta 637
                except:
638
                    traceback.print_exc()
15555 amit.gupta 639
                finally:
640
                    session.close()
14080 amit.gupta 641
                return 'PARSED_SUCCESS'
642
                pass
14008 amit.gupta 643
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 644
        except:
645
            traceback.print_exc()    
646
            return 'PARSED_FAILED'
647
 
648
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 649
        if "ordered from" in detailedStatus.lower():
650
            return MStore.ORDER_PLACED 
651
 
13995 amit.gupta 652
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 653
            if detailedStatus.lower() in value:
13995 amit.gupta 654
                return key
655
 
17253 amit.gupta 656
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 657
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 658
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 659
        br = getBrowserObject()
660
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
661
        url = AMAZON_AFF_URL
662
        response = br.open(url)
663
        #get data for past 40 days and store it to mongo
664
        dt = datetime.now()
14701 amit.gupta 665
        dat = dt - timedelta(days=2)
14651 amit.gupta 666
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
667
        response = br.open(url)
668
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
669
        j=-1
670
        for row in page.split("\n"):
671
            j += 1
672
            if j== 0 or j==1:
673
                continue
674
            fields = row.split("\t")
675
            if len(fields)>1:
676
                print fields
677
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
678
                print amazonAffiliate
679
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
680
            else:
681
                break
14650 amit.gupta 682
 
14608 amit.gupta 683
 
16468 amit.gupta 684
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 685
        print trackingUrl
14624 amit.gupta 686
        subOrder = {}
14749 amit.gupta 687
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 688
        status = MStore.ORDER_SHIPPED
689
        #print page
14624 amit.gupta 690
        soup = BeautifulSoup(page)
16468 amit.gupta 691
        header1 = soup.find("h1") 
692
        if header1:
693
            if header1.text=="Sign In":
694
                print "Login page is displayed for order id", orderId
17307 amit.gupta 695
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 696
                return subOrder
16468 amit.gupta 697
        try:
16474 amit.gupta 698
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 699
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
700
            try:
701
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
702
            except:
703
                displayStatus = detailedStatus
704
            print displayStatus
705
            if detailedStatus.lower().find("delivered")>=0:
706
                print detailedStatus
707
                displayStatus = "Delivered"
708
                status = "Delivered"
709
                try:
710
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
711
                except:
712
                    pass
713
 
714
            elif detailedStatus.lower() == 'returned':
715
                status = 'Cancelled'
716
            subOrder['status'] = status
717
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 718
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 719
 
720
        except:
17350 amit.gupta 721
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 722
            print "failed to parse", orderId
723
            traceback.print_exc()
724
 
14624 amit.gupta 725
        return subOrder
726
 
727
 
728
 
16980 amit.gupta 729
    def parseInfo(self,):
730
        from pyquery import PyQuery as pq
16986 amit.gupta 731
        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 732
        try:
733
            for order in orders:
734
                try:
735
                    doc = pq(order.rawhtml)
736
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
737
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 738
                    orderInfo = All_user_addresses()
16980 amit.gupta 739
                    orderInfo.address = lists[-3].text
740
                    orderInfo.user_id = order.user_id
16986 amit.gupta 741
                    orderInfo.source = 'order' 
742
                    #orderInfo.order_id = order.id
743
                    #orderInfo.email = None
744
                    #orderInfo.name = lists[0].text 
745
                    #orderInfo.mobile = None
16980 amit.gupta 746
                    adSplit = lists[-2].text.split(",")
747
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
748
                    if match:
749
                        items = match.groups()
16986 amit.gupta 750
                    orderInfo.city = adSplit[0].strip()
751
                    orderInfo.pincode = items[1].strip()
752
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 753
                    session.commit()
754
                except:
755
                    session.rollback()
756
                    continue
757
        finally:
758
            session.close()
14608 amit.gupta 759
 
13927 amit.gupta 760
 
761
 
13774 amit.gupta 762
def main():
16874 amit.gupta 763
    store = getStore(1)
17465 amit.gupta 764
#    for order in store.db.merchantOrder.find({"orderId":{"$gt":47644}, "storeId":3}):
765
#        try:
766
#            orderId=order.get("orderId")
767
#            o = session.query(OrdersRaw).filter_by(id = orderId).one()
768
#            o.status = 'ORDER_CREATED'
769
#            session.commit()
770
#        finally:
771
#            session.close()
17600 amit.gupta 772
    #store.trackOrdersForUser(10466, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-7498756-0837158', readSSh('/AmazonTrack/User10466/403-7498756-0837158'))
773
    #store.trackOrdersForUser(10466, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-2539406-8617932', readSSh('/AmazonTrack/User10466/403-2539406-8617932'))
774
    store.parseOrderRawHtml(58599, 'sub123', 9981, readSSh('/home/amit/sample.html'), 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html/ref=ox_spc_place_order?ie=UTF8&fromAnywhere=1&hasWorkingJav')
17231 amit.gupta 775
 
776
def getSummaryFile(directory):
777
    date1 = datetime(2015,1,1)
778
    finalFile = None
779
    try:
780
        for file in os.listdir(directory):
781
            if file.startswith("orderSummary"):
782
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
783
                if date2 > date1:
784
                    date1 = date2
785
                    finalFile=file
786
    except:
787
        print "Missing directory"
17251 amit.gupta 788
    return finalFile 
17231 amit.gupta 789
 
15555 amit.gupta 790
 
791
def parseDetailNotCreated():
792
    try:
793
        store=getStore(1)
794
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 795
        session.close()
15555 amit.gupta 796
        for order in orders:
797
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
798
 
799
    finally:
800
        session.close()
17251 amit.gupta 801
 
802
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 803
    print dateString
17251 amit.gupta 804
    if dateString.lower()=='today':
805
        return date.today()
806
    if dateString.lower()=='yesterday':
807
        return date.today() - timedelta(days=1)
808
    try:
809
        return datetime.strptime(dateString, '%d-%b-%y')
810
    except:
811
        try:
812
            #get Closest Date from today
813
            curDate = date.today()
814
            curTime = datetime(curDate.year, curDate.month, curDate.day)
815
            curYear = curDate.year
816
            prevYear = curYear - 1
17254 amit.gupta 817
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
818
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 819
            if dateMax <= curTime:
820
                return dateMax
821
            else:
822
                return dateMin 
823
        except:
17254 amit.gupta 824
            try:
825
                days_of_week = ['sunday','monday','tuesday','wednesday',
826
                            'thursday','friday','saturday']
827
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
828
                if deltaDays <= 0:
829
                    deltaDays= deltaDays + 7
830
                curDate = curDate - timedelta(days=deltaDays)
831
                print datetime.strftime(curDate, '%d-%b-%y')
832
                return curDate
833
            except:
834
                print "could not parse"
835
                return None
15565 amit.gupta 836
 
17254 amit.gupta 837
def getDateStringArriving(dateString='Thursday'):
838
    print dateString
839
    if dateString.lower()=='today':
840
        return date.today()
841
    if dateString.lower()=='tomorrow':
842
        return date.today() + timedelta(days=1)
843
    try:
844
        return datetime.strptime(dateString, '%d-%b-%y')
845
    except:
846
        try:
847
            #get Closest Date from today
848
            curDate = date.today()
849
            curTime = datetime(curDate.year, curDate.month, curDate.day)
850
            curYear = curDate.year
851
            nextYear = curYear + 1
852
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
853
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
854
            if dateMin >= curTime:
855
                return dateMin
856
            else:
857
                return dateMax 
858
        except:
859
            try:
860
                days_of_week = ['sunday','monday','tuesday','wednesday',
861
                            'thursday','friday','saturday']
862
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
863
                if deltaDays < 0:
864
                    deltaDays= deltaDays + 7
865
                curDate = curDate + timedelta(days=deltaDays)
866
                return curDate
867
            except:
868
                print "Could not parse"
869
                return None 
870
 
15565 amit.gupta 871
def parseOrderNotCreated():
872
    try:
873
        store=getStore(1)
874
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
875
        session.close()
876
        for order in orders:
15567 amit.gupta 877
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
878
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
879
            order1.status = result
880
            session.commit()
15565 amit.gupta 881
 
882
    finally:
883
        session.close()
15555 amit.gupta 884
 
13774 amit.gupta 885
if __name__ == '__main__':
17599 amit.gupta 886
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
887
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
888
#    readSSh("/AmazonTrack/User10466/403-7498756-0837158")
889
    main()