Subversion Repositories SmartDukaan

Rev

Rev 17368 | Rev 17465 | 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
13774 amit.gupta 30
 
31
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 32
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 33
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
14624 amit.gupta 34
AMAZON_AFF_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/listReports'
14650 amit.gupta 35
AMAZON_AFF_FILE_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/getReport?filename=saholic-21-orders-report-%s.tsv.gz'
13774 amit.gupta 36
class Store(MStore):
13821 amit.gupta 37
 
17348 amit.gupta 38
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch', 'order received'],
14064 amit.gupta 39
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
17324 amit.gupta 40
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled', 'replacement complete', 'return received'],
17258 amit.gupta 41
                            MStore.ORDER_DELIVERED : ['delivered', 'your package was delivered', 'package was handed directly to customer']
13821 amit.gupta 42
                           }
13774 amit.gupta 43
 
44
    def __init__(self,store_id):
45
        super(Store, self).__init__(store_id)
46
 
47
    def getName(self):
14749 amit.gupta 48
        return "amazon"
13774 amit.gupta 49
 
14945 amit.gupta 50
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, track=False):
14958 amit.gupta 51
        parseString = "Tracking" if track else "Transacted"
52
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 53
        resp = {}
14813 amit.gupta 54
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 55
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 56
            try:
57
                soup = BeautifulSoup(rawHtml)
14608 amit.gupta 58
                try:
59
                    orderUrl = soup.find('div', {"id":"thank-you-box-wrapper"}).div.findAll('div', recursive=False)[1].a['href']
60
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
61
                except:
62
                    merchantOrderId = soup.find(id="orders-list").div.span.b.text
14699 amit.gupta 63
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 64
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
65
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 66
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 67
                order.requireDetail = True
14698 amit.gupta 68
                order.status = 'html_required'
14297 amit.gupta 69
                order.closed = None
14312 amit.gupta 70
                if self._saveToOrder(todict(order)):
71
                    resp['result'] = 'ORDER_CREATED'
72
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
73
                    resp["htmlRequired"] = True
74
                    resp['orderId'] = orderId
75
                else:
76
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
77
 
13774 amit.gupta 78
            except:
14813 amit.gupta 79
                #Write all cases here for Order Not created Known
14809 amit.gupta 80
                try:
16874 amit.gupta 81
                    if not soup.body:
82
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
17359 amit.gupta 83
                    elif 'Securely redirecting you' in soup.find("h3").text.strip() or soup.find("h3").text.strip()=="Orders":
14813 amit.gupta 84
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 85
                    else:
14813 amit.gupta 86
                        raise
14809 amit.gupta 87
                except:
14814 amit.gupta 88
                    try:
17356 amit.gupta 89
                        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 90
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
91
                        else:
92
                            raise
93
                    except:
15046 amit.gupta 94
                        try:
17356 amit.gupta 95
                            if soup.find("h2").text.strip() in ['Web page not available','Webpage not available', 'Do you have an Amazon password?']:
15046 amit.gupta 96
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
97
                            else:
98
                                raise
99
                        except:
15565 amit.gupta 100
                            try:
16876 amit.gupta 101
                                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 102
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
16877 amit.gupta 103
                                elif soup.find("b", {'class':'h1'}).text.strip().find("We're sorry") > -1:
104
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
15565 amit.gupta 105
                                else:
106
                                    raise
107
                            except:
108
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
109
 
13774 amit.gupta 110
        else:
13781 amit.gupta 111
            try:
14722 amit.gupta 112
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
113
                if mo is not None:
14749 amit.gupta 114
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 115
                    merchantOrder.createdOn = mo.get("createdOn")
116
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
117
                else:
14945 amit.gupta 118
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 119
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14464 amit.gupta 120
                soup = BeautifulSoup(rawHtml)
17331 amit.gupta 121
                if not soup.body:
17319 amit.gupta 122
                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
123
                else:   
14545 amit.gupta 124
                    try:
17319 amit.gupta 125
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 126
                        resp['result'] = 'DETAIL_CREATED'
17368 amit.gupta 127
                        return
14545 amit.gupta 128
                    except:
15056 amit.gupta 129
                        try:
17319 amit.gupta 130
                            traceback.print_exc()
131
                            self.parseOldStlye(merchantOrder, soup)
132
                            resp['result'] = 'DETAIL_CREATED'
17368 amit.gupta 133
                            return
15056 amit.gupta 134
                        except:
17319 amit.gupta 135
                            traceback.print_exc()
15059 amit.gupta 136
                            try:
17319 amit.gupta 137
                                self.parseCancelled(merchantOrder, soup)
138
                                resp['result'] = 'ORDER_CANCELLED'
15059 amit.gupta 139
                            except:
17319 amit.gupta 140
                                try:
17356 amit.gupta 141
                                    if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
17319 amit.gupta 142
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
143
                                    else:
144
                                        raise
145
                                except:
17356 amit.gupta 146
                                    if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
17319 amit.gupta 147
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
148
                                    else:
149
                                        raise
17357 amit.gupta 150
                order = self.db.merchantOrder.find_one({"orderId":orderId})
151
                if order is not None:
152
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 153
            except:
14813 amit.gupta 154
                order = self.db.merchantOrder.find_one({"orderId":orderId})
155
                if order is not None:
156
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"parse_failed"}})
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
 
15474 amit.gupta 325
        shipmentDivs = orderDetailsContainer.find('div', class_='shipment').findAll('div', recursive = False)
14464 amit.gupta 326
        subOrders = []
327
        merchantOrder.subOrders = subOrders
15473 amit.gupta 328
        closedStatus = True
14464 amit.gupta 329
        for shipmentDiv in shipmentDivs:
17361 amit.gupta 330
            try:
331
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
332
            except:
333
                trackingUrl= None
334
 
17347 amit.gupta 335
            deliverySpanTop = None
14464 amit.gupta 336
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
337
            statusDiv = innerBoxes[0]
338
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 339
            try:
340
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
341
                productDivs = innerBoxes[-1].div.div.div.findAll('div', recursive=False)
342
            except:
343
                productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 344
            subOrders = []
345
            merchantOrder.subOrders = subOrders
346
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 347
                deliverySpan = deliverySpanTop
348
                j=i+1
349
                if not deliverySpanTop:
350
                    if i%2==1:
351
                        continue
352
                    j=i/2 + 1
353
                    deliverySpan = productDiv.div.div.span
354
                    productDiv = productDivs[i+1].div
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
385
                subOrder.deliveryCharges = shippingPrice
17361 amit.gupta 386
                if trackingUrl:
387
                    subOrder.tracingkUrl  = trackingUrl
14466 amit.gupta 388
                subOrder.imgUrl = imgDiv.img["src"]
14464 amit.gupta 389
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid)
15346 amit.gupta 390
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
391
                subOrder.dealRank = dealRank.get('rank')
392
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 393
                cashbackStatus = Store.CB_PENDING
394
                if cashbackAmount <= 0:
395
                    cashbackStatus = Store.CB_NA
396
                subOrder.cashBackStatus = cashbackStatus
397
                subOrder.cashBackAmount = cashbackAmount
398
                if percentage > 0:
399
                    subOrder.cashBackPercentage = percentage
15473 amit.gupta 400
                if hasattr(subOrder, 'deliveredOn'):                               
401
                    subOrder.status = Store.ORDER_DELIVERED
402
                    subOrder.closed = True
403
                    if subOrder.cashBackStatus == Store.CB_PENDING:
404
                        subOrder.cashBackStatus = Store.CB_APPROVED
405
                elif closedStatus:
406
                    closedStatus= False
14464 amit.gupta 407
                subOrders.append(subOrder)
14698 amit.gupta 408
        merchantOrder.status='success'
15473 amit.gupta 409
        merchantOrder.closed = closedStatus
14464 amit.gupta 410
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 411
 
412
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 413
        try:
414
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
415
            if fonts[0].text == "Important Message":
416
                if fonts[1].text=="This order has been cancelled.":
417
                    merchantOrder.closed = True
418
                    merchantOrder.status = "cancelled"
419
                    merchantOrder.requireDetail = False
420
                    self._updateToOrder(todict(merchantOrder))
421
                    return
422
                else:
423
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 424
            else:
425
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 426
        except:
14549 amit.gupta 427
            orderDetails = soup.body.find(id="orderDetails")
428
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
429
                merchantOrder.closed = True
14698 amit.gupta 430
                merchantOrder.status = "cancelled"
14549 amit.gupta 431
                merchantOrder.requireDetail = False
432
                self._updateToOrder(todict(merchantOrder))
433
            else:
434
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 435
 
13868 amit.gupta 436
    def getTrackingUrls(self, userId):
14074 amit.gupta 437
 
438
        missingOrderUrls = []
439
        missingOrders = self._getMissingOrders({'userId':userId})
440
        for missingOrder in missingOrders:
441
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 442
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 443
        count = len(orders)
13879 amit.gupta 444
        print "count", count
14948 amit.gupta 445
        print "Missing Urls"
446
        print "*************"
447
        print missingOrderUrls
13868 amit.gupta 448
        if count > 0:
14684 amit.gupta 449
            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 450
        else: 
14074 amit.gupta 451
            return missingOrderUrls
13927 amit.gupta 452
 
453
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 454
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 455
        if not os.path.exists(directory):
456
            os.makedirs(directory)
14173 amit.gupta 457
 
14945 amit.gupta 458
 
13995 amit.gupta 459
        try:
460
            searchMap = {'userId':userId}
461
            collectionMap = {'merchantOrderId':1}
462
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 463
            datetimeNow = datetime.now()
464
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 465
            print "url----------------", url
14008 amit.gupta 466
 
14320 amit.gupta 467
            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 468
                if url == 'https://www.amazon.in/gp/css/order-history':
469
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
470
                else:   
471
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')   
472
                f = open(filename,'w')
473
                f.write(rawHtml) # python will convert \n to os.linesep
474
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 475
                soup = BeautifulSoup(rawHtml)
476
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
477
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
478
                for activeOrder in activeOrders:
17307 amit.gupta 479
                    matched=False
13995 amit.gupta 480
                    for orderEle in allOrders:
17254 amit.gupta 481
                        deliveredOn = None
482
                        deliveryEstimate = None
17258 amit.gupta 483
                        shippingEstimate = None
13995 amit.gupta 484
                        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'})
485
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
486
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 487
                            matched=True
13995 amit.gupta 488
                            closed = True
17324 amit.gupta 489
                            shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
13995 amit.gupta 490
                            for shipment in shipments:
17307 amit.gupta 491
                                orderStatusDesc = None
13995 amit.gupta 492
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 493
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 494
                                try:
17270 amit.gupta 495
                                    orderStatus = sdivs[0].span.text.strip()
496
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 497
                                except:
17270 amit.gupta 498
                                    try:
499
                                        dateString = orderStatus.split("Delivered ")[1].strip()
500
                                        status = MStore.ORDER_DELIVERED
501
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
502
                                    except:
503
                                        try:
17271 amit.gupta 504
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 505
                                            status = MStore.ORDER_SHIPPED
506
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
507
                                        except:
508
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 509
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 510
                                try:
17270 amit.gupta 511
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
512
                                except:
17271 amit.gupta 513
                                    try:
514
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
515
                                    except:
17307 amit.gupta 516
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 517
 
17271 amit.gupta 518
                                if orderStatusDesc: 
17348 amit.gupta 519
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 520
                                    try:
17348 amit.gupta 521
                                        status = self._getStatusFromDetailedStatus(orderStatus)
522
                                    except:
523
                                        pass
524
                                    try:
17271 amit.gupta 525
                                        if "Dispatch estimate" in orderStatusDesc:
526
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
527
                                        elif "Delivery estimate" in orderStatus:
528
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
529
                                        elif "Arriving" in orderStatus:
530
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
531
                                    except:
532
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
533
                                        closed=False
534
                                        status = None
17258 amit.gupta 535
 
14335 amit.gupta 536
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 537
                                trackingUrl = None
538
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
539
                                    if buttonDiv.find('a').text.strip()=='Track package':
540
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
541
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 542
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 543
                                        break
544
                                for prodDiv in productDivs:
545
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
546
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
547
                                    imgUrl = prodDiv.find("img")["src"]
548
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 549
                                        if subOrder['closed']==True:
550
                                            continue
17230 amit.gupta 551
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 552
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
553
                                            updateMap = {}
554
                                            closedStatus = False
555
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 556
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 557
                                            if status:
17348 amit.gupta 558
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 559
                                                updateMap['subOrders.$.status'] = status 
560
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 561
 
17251 amit.gupta 562
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 563
                                                if deliveredOn:                               
17254 amit.gupta 564
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 565
                                                closedStatus = True
566
                                                updateMap['subOrders.$.closed'] = True
567
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 568
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 569
                                            if status==MStore.ORDER_CANCELLED:     
570
                                                closedStatus = True
571
                                                updateMap['subOrders.$.closed'] = True
572
                                                if cashbackStatus == Store.CB_PENDING:
573
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 574
                                            if status==MStore.ORDER_SHIPPED:
575
                                                if deliveryEstimate:   
576
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 577
                                                if trackingUrl is not None:
578
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 579
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 580
                                            if shippingEstimate:   
581
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 582
                                            if not closedStatus:
583
                                                closed = False
17307 amit.gupta 584
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 585
                                            updateMap['status']='success'
13995 amit.gupta 586
                                            bulk.find(findMap).update({'$set' : updateMap})
587
                                            break
17321 amit.gupta 588
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 589
                            break
590
                    if not matched:
17319 amit.gupta 591
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 592
                        for subOrder in activeOrder['subOrders']:
593
                            if subOrder['closed']==True:
594
                                continue
595
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
596
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
597
                            bulk.find({'orderId': activeOrder['orderId']})
598
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 599
                bulk.execute()
600
                return 'PARSED_SUCCESS'
601
            else:
15563 amit.gupta 602
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
603
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 604
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 605
 
606
                filename = directory + "/" + merchantOrderId   
607
                f = open(filename,'w')
608
                f.write(rawHtml) # python will convert \n to os.linesep
609
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 610
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 611
                print "result", result
15555 amit.gupta 612
                try:
15560 amit.gupta 613
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
614
                    order1.status = result
615
                    order1.rawhtml = rawHtml
17358 amit.gupta 616
                    order1.order_url = url
15555 amit.gupta 617
                    session.commit()
15561 amit.gupta 618
                except:
619
                    traceback.print_exc()
15555 amit.gupta 620
                finally:
621
                    session.close()
14080 amit.gupta 622
                return 'PARSED_SUCCESS'
623
                pass
14008 amit.gupta 624
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 625
        except:
626
            traceback.print_exc()    
627
            return 'PARSED_FAILED'
628
 
629
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 630
        if "ordered from" in detailedStatus.lower():
631
            return MStore.ORDER_PLACED 
632
 
13995 amit.gupta 633
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 634
            if detailedStatus.lower() in value:
13995 amit.gupta 635
                return key
636
 
17253 amit.gupta 637
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 638
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 639
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 640
        br = getBrowserObject()
641
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
642
        url = AMAZON_AFF_URL
643
        response = br.open(url)
644
        #get data for past 40 days and store it to mongo
645
        dt = datetime.now()
14701 amit.gupta 646
        dat = dt - timedelta(days=2)
14651 amit.gupta 647
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
648
        response = br.open(url)
649
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
650
        j=-1
651
        for row in page.split("\n"):
652
            j += 1
653
            if j== 0 or j==1:
654
                continue
655
            fields = row.split("\t")
656
            if len(fields)>1:
657
                print fields
658
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
659
                print amazonAffiliate
660
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
661
            else:
662
                break
14650 amit.gupta 663
 
14608 amit.gupta 664
 
16468 amit.gupta 665
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 666
        print trackingUrl
14624 amit.gupta 667
        subOrder = {}
14749 amit.gupta 668
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 669
        status = MStore.ORDER_SHIPPED
670
        #print page
14624 amit.gupta 671
        soup = BeautifulSoup(page)
16468 amit.gupta 672
        header1 = soup.find("h1") 
673
        if header1:
674
            if header1.text=="Sign In":
675
                print "Login page is displayed for order id", orderId
17307 amit.gupta 676
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 677
                return subOrder
16468 amit.gupta 678
        try:
16474 amit.gupta 679
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 680
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
681
            try:
682
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
683
            except:
684
                displayStatus = detailedStatus
685
            print displayStatus
686
            if detailedStatus.lower().find("delivered")>=0:
687
                print detailedStatus
688
                displayStatus = "Delivered"
689
                status = "Delivered"
690
                try:
691
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
692
                except:
693
                    pass
694
 
695
            elif detailedStatus.lower() == 'returned':
696
                status = 'Cancelled'
697
            subOrder['status'] = status
698
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 699
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 700
 
701
        except:
17350 amit.gupta 702
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 703
            print "failed to parse", orderId
704
            traceback.print_exc()
705
 
14624 amit.gupta 706
        return subOrder
707
 
708
 
709
 
16980 amit.gupta 710
    def parseInfo(self,):
711
        from pyquery import PyQuery as pq
16986 amit.gupta 712
        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 713
        try:
714
            for order in orders:
715
                try:
716
                    doc = pq(order.rawhtml)
717
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
718
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 719
                    orderInfo = All_user_addresses()
16980 amit.gupta 720
                    orderInfo.address = lists[-3].text
721
                    orderInfo.user_id = order.user_id
16986 amit.gupta 722
                    orderInfo.source = 'order' 
723
                    #orderInfo.order_id = order.id
724
                    #orderInfo.email = None
725
                    #orderInfo.name = lists[0].text 
726
                    #orderInfo.mobile = None
16980 amit.gupta 727
                    adSplit = lists[-2].text.split(",")
728
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
729
                    if match:
730
                        items = match.groups()
16986 amit.gupta 731
                    orderInfo.city = adSplit[0].strip()
732
                    orderInfo.pincode = items[1].strip()
733
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 734
                    session.commit()
735
                except:
736
                    session.rollback()
737
                    continue
738
        finally:
739
            session.close()
14608 amit.gupta 740
 
13927 amit.gupta 741
 
742
 
13774 amit.gupta 743
def main():
16874 amit.gupta 744
    store = getStore(1)
17453 amit.gupta 745
    for order in store.db.merchantOrder.find({"orderId":{"$gt":47644}, "storeId":3}):
17334 amit.gupta 746
        try:
17358 amit.gupta 747
            orderId=order.get("orderId")
17335 amit.gupta 748
            o = session.query(OrdersRaw).filter_by(id = orderId).one()
17453 amit.gupta 749
            o.status = 'ORDER_CREATED'
17362 amit.gupta 750
            session.commit()
17360 amit.gupta 751
        finally:
752
            session.close()
17231 amit.gupta 753
    #store.trackOrdersForUser(8703, 'https://www.amazon.in/gp/css/order-history', readSSh('/home/amit/Downloads/orderSummary06-10_12_15_54'))
754
 
755
def getSummaryFile(directory):
756
    date1 = datetime(2015,1,1)
757
    finalFile = None
758
    try:
759
        for file in os.listdir(directory):
760
            if file.startswith("orderSummary"):
761
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
762
                if date2 > date1:
763
                    date1 = date2
764
                    finalFile=file
765
    except:
766
        print "Missing directory"
17251 amit.gupta 767
    return finalFile 
17231 amit.gupta 768
 
15555 amit.gupta 769
 
770
def parseDetailNotCreated():
771
    try:
772
        store=getStore(1)
773
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 774
        session.close()
15555 amit.gupta 775
        for order in orders:
776
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
777
 
778
    finally:
779
        session.close()
17251 amit.gupta 780
 
781
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 782
    print dateString
17251 amit.gupta 783
    if dateString.lower()=='today':
784
        return date.today()
785
    if dateString.lower()=='yesterday':
786
        return date.today() - timedelta(days=1)
787
    try:
788
        return datetime.strptime(dateString, '%d-%b-%y')
789
    except:
790
        try:
791
            #get Closest Date from today
792
            curDate = date.today()
793
            curTime = datetime(curDate.year, curDate.month, curDate.day)
794
            curYear = curDate.year
795
            prevYear = curYear - 1
17254 amit.gupta 796
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
797
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 798
            if dateMax <= curTime:
799
                return dateMax
800
            else:
801
                return dateMin 
802
        except:
17254 amit.gupta 803
            try:
804
                days_of_week = ['sunday','monday','tuesday','wednesday',
805
                            'thursday','friday','saturday']
806
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
807
                if deltaDays <= 0:
808
                    deltaDays= deltaDays + 7
809
                curDate = curDate - timedelta(days=deltaDays)
810
                print datetime.strftime(curDate, '%d-%b-%y')
811
                return curDate
812
            except:
813
                print "could not parse"
814
                return None
15565 amit.gupta 815
 
17254 amit.gupta 816
def getDateStringArriving(dateString='Thursday'):
817
    print dateString
818
    if dateString.lower()=='today':
819
        return date.today()
820
    if dateString.lower()=='tomorrow':
821
        return date.today() + timedelta(days=1)
822
    try:
823
        return datetime.strptime(dateString, '%d-%b-%y')
824
    except:
825
        try:
826
            #get Closest Date from today
827
            curDate = date.today()
828
            curTime = datetime(curDate.year, curDate.month, curDate.day)
829
            curYear = curDate.year
830
            nextYear = curYear + 1
831
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
832
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
833
            if dateMin >= curTime:
834
                return dateMin
835
            else:
836
                return dateMax 
837
        except:
838
            try:
839
                days_of_week = ['sunday','monday','tuesday','wednesday',
840
                            'thursday','friday','saturday']
841
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
842
                if deltaDays < 0:
843
                    deltaDays= deltaDays + 7
844
                curDate = curDate + timedelta(days=deltaDays)
845
                return curDate
846
            except:
847
                print "Could not parse"
848
                return None 
849
 
15565 amit.gupta 850
def parseOrderNotCreated():
851
    try:
852
        store=getStore(1)
853
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
854
        session.close()
855
        for order in orders:
15567 amit.gupta 856
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
857
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
858
            order1.status = result
859
            session.commit()
15565 amit.gupta 860
 
861
    finally:
862
        session.close()
15555 amit.gupta 863
 
13774 amit.gupta 864
if __name__ == '__main__':
17331 amit.gupta 865
#        readSSh("~/46417.html")
17321 amit.gupta 866
        main()
17254 amit.gupta 867
        #getDateStringArriving()