Subversion Repositories SmartDukaan

Rev

Rev 17360 | Rev 17362 | 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'
14545 amit.gupta 127
                    except:
15056 amit.gupta 128
                        try:
17319 amit.gupta 129
                            traceback.print_exc()
130
                            self.parseOldStlye(merchantOrder, soup)
131
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 132
                        except:
17319 amit.gupta 133
                            traceback.print_exc()
15059 amit.gupta 134
                            try:
17319 amit.gupta 135
                                self.parseCancelled(merchantOrder, soup)
136
                                resp['result'] = 'ORDER_CANCELLED'
15059 amit.gupta 137
                            except:
17319 amit.gupta 138
                                try:
17356 amit.gupta 139
                                    if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
17319 amit.gupta 140
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
141
                                    else:
142
                                        raise
143
                                except:
17356 amit.gupta 144
                                    if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
17319 amit.gupta 145
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
146
                                    else:
147
                                        raise
17357 amit.gupta 148
                order = self.db.merchantOrder.find_one({"orderId":orderId})
149
                if order is not None:
150
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 151
            except:
14813 amit.gupta 152
                order = self.db.merchantOrder.find_one({"orderId":orderId})
153
                if order is not None:
154
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"parse_failed"}})
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
351
                    deliverySpan = productDiv.div.div.span
352
                    productDiv = productDivs[i+1].div
14464 amit.gupta 353
                imgDiv  = productDiv.div.div
354
                detailDiv = imgDiv.find_next_sibling('div')
355
                detailDivs = detailDiv.findAll('div', recursive=False)
356
                arr = detailDivs[0].a.text.strip().split(" of ", 1)
14465 amit.gupta 357
                (productTitle, quantity) = (arr[-1], (1 if len(arr)==1 else int(arr[0])) )
14820 amit.gupta 358
                try:
359
                    unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
360
                except:
361
                    unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
14464 amit.gupta 362
                amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
363
                productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
364
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
365
                subOrder.productCode = productUrl.split('/')[5]
366
                subOrder.unitPrice = unitPrice
17347 amit.gupta 367
                subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
14722 amit.gupta 368
                estDlvyTime = datetime.now()
14566 amit.gupta 369
                if deliverySpan is not None:
14756 amit.gupta 370
                    try:
371
                        subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
372
                        estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
373
                        subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
374
                        estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
375
                    except:
15473 amit.gupta 376
                        if "Delivered on" in deliverySpan.text:
377
                            subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
14756 amit.gupta 378
                        subOrder.estimatedDeliveryDate = "Not available"
15473 amit.gupta 379
 
14722 amit.gupta 380
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
14756 amit.gupta 381
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 382
                subOrder.detailedStatus = subOrderStatus
383
                subOrder.deliveryCharges = shippingPrice
17361 amit.gupta 384
                if trackingUrl:
385
                    subOrder.tracingkUrl  = trackingUrl
14466 amit.gupta 386
                subOrder.imgUrl = imgDiv.img["src"]
14464 amit.gupta 387
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid)
15346 amit.gupta 388
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
389
                subOrder.dealRank = dealRank.get('rank')
390
                subOrder.rankDesc = dealRank.get('description')
14464 amit.gupta 391
                cashbackStatus = Store.CB_PENDING
392
                if cashbackAmount <= 0:
393
                    cashbackStatus = Store.CB_NA
394
                subOrder.cashBackStatus = cashbackStatus
395
                subOrder.cashBackAmount = cashbackAmount
396
                if percentage > 0:
397
                    subOrder.cashBackPercentage = percentage
15473 amit.gupta 398
                if hasattr(subOrder, 'deliveredOn'):                               
399
                    subOrder.status = Store.ORDER_DELIVERED
400
                    subOrder.closed = True
401
                    if subOrder.cashBackStatus == Store.CB_PENDING:
402
                        subOrder.cashBackStatus = Store.CB_APPROVED
403
                elif closedStatus:
404
                    closedStatus= False
14464 amit.gupta 405
                subOrders.append(subOrder)
14698 amit.gupta 406
        merchantOrder.status='success'
15473 amit.gupta 407
        merchantOrder.closed = closedStatus
14464 amit.gupta 408
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 409
 
410
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 411
        try:
412
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
413
            if fonts[0].text == "Important Message":
414
                if fonts[1].text=="This order has been cancelled.":
415
                    merchantOrder.closed = True
416
                    merchantOrder.status = "cancelled"
417
                    merchantOrder.requireDetail = False
418
                    self._updateToOrder(todict(merchantOrder))
419
                    return
420
                else:
421
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 422
            else:
423
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 424
        except:
14549 amit.gupta 425
            orderDetails = soup.body.find(id="orderDetails")
426
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
427
                merchantOrder.closed = True
14698 amit.gupta 428
                merchantOrder.status = "cancelled"
14549 amit.gupta 429
                merchantOrder.requireDetail = False
430
                self._updateToOrder(todict(merchantOrder))
431
            else:
432
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 433
 
13868 amit.gupta 434
    def getTrackingUrls(self, userId):
14074 amit.gupta 435
 
436
        missingOrderUrls = []
437
        missingOrders = self._getMissingOrders({'userId':userId})
438
        for missingOrder in missingOrders:
439
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
14683 amit.gupta 440
        orders = self._getActiveOrders({'userId':userId, "subOrders.trackingUrl":{"$exists":False} })
13882 amit.gupta 441
        count = len(orders)
13879 amit.gupta 442
        print "count", count
14948 amit.gupta 443
        print "Missing Urls"
444
        print "*************"
445
        print missingOrderUrls
13868 amit.gupta 446
        if count > 0:
14684 amit.gupta 447
            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 448
        else: 
14074 amit.gupta 449
            return missingOrderUrls
13927 amit.gupta 450
 
451
    def trackOrdersForUser(self, userId, url, rawHtml):
14985 amit.gupta 452
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 453
        if not os.path.exists(directory):
454
            os.makedirs(directory)
14173 amit.gupta 455
 
14945 amit.gupta 456
 
13995 amit.gupta 457
        try:
458
            searchMap = {'userId':userId}
459
            collectionMap = {'merchantOrderId':1}
460
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 461
            datetimeNow = datetime.now()
462
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 463
            print "url----------------", url
14008 amit.gupta 464
 
14320 amit.gupta 465
            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 466
                if url == 'https://www.amazon.in/gp/css/order-history':
467
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
468
                else:   
469
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')   
470
                f = open(filename,'w')
471
                f.write(rawHtml) # python will convert \n to os.linesep
472
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 473
                soup = BeautifulSoup(rawHtml)
474
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
475
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
476
                for activeOrder in activeOrders:
17307 amit.gupta 477
                    matched=False
13995 amit.gupta 478
                    for orderEle in allOrders:
17254 amit.gupta 479
                        deliveredOn = None
480
                        deliveryEstimate = None
17258 amit.gupta 481
                        shippingEstimate = None
13995 amit.gupta 482
                        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'})
483
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
484
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 485
                            matched=True
13995 amit.gupta 486
                            closed = True
17324 amit.gupta 487
                            shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
13995 amit.gupta 488
                            for shipment in shipments:
17307 amit.gupta 489
                                orderStatusDesc = None
13995 amit.gupta 490
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 491
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 492
                                try:
17270 amit.gupta 493
                                    orderStatus = sdivs[0].span.text.strip()
494
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 495
                                except:
17270 amit.gupta 496
                                    try:
497
                                        dateString = orderStatus.split("Delivered ")[1].strip()
498
                                        status = MStore.ORDER_DELIVERED
499
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
500
                                    except:
501
                                        try:
17271 amit.gupta 502
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 503
                                            status = MStore.ORDER_SHIPPED
504
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
505
                                        except:
506
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 507
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 508
                                try:
17270 amit.gupta 509
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
510
                                except:
17271 amit.gupta 511
                                    try:
512
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
513
                                    except:
17307 amit.gupta 514
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 515
 
17271 amit.gupta 516
                                if orderStatusDesc: 
17348 amit.gupta 517
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 518
                                    try:
17348 amit.gupta 519
                                        status = self._getStatusFromDetailedStatus(orderStatus)
520
                                    except:
521
                                        pass
522
                                    try:
17271 amit.gupta 523
                                        if "Dispatch estimate" in orderStatusDesc:
524
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
525
                                        elif "Delivery estimate" in orderStatus:
526
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
527
                                        elif "Arriving" in orderStatus:
528
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
529
                                    except:
530
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
531
                                        closed=False
532
                                        status = None
17258 amit.gupta 533
 
14335 amit.gupta 534
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 535
                                trackingUrl = None
536
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
537
                                    if buttonDiv.find('a').text.strip()=='Track package':
538
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
539
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 540
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 541
                                        break
542
                                for prodDiv in productDivs:
543
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
544
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
545
                                    imgUrl = prodDiv.find("img")["src"]
546
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 547
                                        if subOrder['closed']==True:
548
                                            continue
17230 amit.gupta 549
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 550
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
551
                                            updateMap = {}
552
                                            closedStatus = False
553
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 554
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 555
                                            if status:
17348 amit.gupta 556
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 557
                                                updateMap['subOrders.$.status'] = status 
558
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 559
 
17251 amit.gupta 560
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 561
                                                if deliveredOn:                               
17254 amit.gupta 562
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 563
                                                closedStatus = True
564
                                                updateMap['subOrders.$.closed'] = True
565
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 566
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 567
                                            if status==MStore.ORDER_CANCELLED:     
568
                                                closedStatus = True
569
                                                updateMap['subOrders.$.closed'] = True
570
                                                if cashbackStatus == Store.CB_PENDING:
571
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 572
                                            if status==MStore.ORDER_SHIPPED:
573
                                                if deliveryEstimate:   
574
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 575
                                                if trackingUrl is not None:
576
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 577
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 578
                                            if shippingEstimate:   
579
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 580
                                            if not closedStatus:
581
                                                closed = False
17307 amit.gupta 582
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 583
                                            updateMap['status']='success'
13995 amit.gupta 584
                                            bulk.find(findMap).update({'$set' : updateMap})
585
                                            break
17321 amit.gupta 586
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 587
                            break
588
                    if not matched:
17319 amit.gupta 589
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 590
                        for subOrder in activeOrder['subOrders']:
591
                            if subOrder['closed']==True:
592
                                continue
593
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
594
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
595
                            bulk.find({'orderId': activeOrder['orderId']})
596
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 597
                bulk.execute()
598
                return 'PARSED_SUCCESS'
599
            else:
15563 amit.gupta 600
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
601
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 602
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 603
 
604
                filename = directory + "/" + merchantOrderId   
605
                f = open(filename,'w')
606
                f.write(rawHtml) # python will convert \n to os.linesep
607
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 608
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 609
                print "result", result
15555 amit.gupta 610
                try:
15560 amit.gupta 611
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
612
                    order1.status = result
613
                    order1.rawhtml = rawHtml
17358 amit.gupta 614
                    order1.order_url = url
15555 amit.gupta 615
                    session.commit()
15561 amit.gupta 616
                except:
617
                    traceback.print_exc()
15555 amit.gupta 618
                finally:
619
                    session.close()
14080 amit.gupta 620
                return 'PARSED_SUCCESS'
621
                pass
14008 amit.gupta 622
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 623
        except:
624
            traceback.print_exc()    
625
            return 'PARSED_FAILED'
626
 
627
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 628
        if "ordered from" in detailedStatus.lower():
629
            return MStore.ORDER_PLACED 
630
 
13995 amit.gupta 631
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 632
            if detailedStatus.lower() in value:
13995 amit.gupta 633
                return key
634
 
17253 amit.gupta 635
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 636
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 637
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 638
        br = getBrowserObject()
639
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
640
        url = AMAZON_AFF_URL
641
        response = br.open(url)
642
        #get data for past 40 days and store it to mongo
643
        dt = datetime.now()
14701 amit.gupta 644
        dat = dt - timedelta(days=2)
14651 amit.gupta 645
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
646
        response = br.open(url)
647
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
648
        j=-1
649
        for row in page.split("\n"):
650
            j += 1
651
            if j== 0 or j==1:
652
                continue
653
            fields = row.split("\t")
654
            if len(fields)>1:
655
                print fields
656
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
657
                print amazonAffiliate
658
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
659
            else:
660
                break
14650 amit.gupta 661
 
14608 amit.gupta 662
 
16468 amit.gupta 663
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 664
        print trackingUrl
14624 amit.gupta 665
        subOrder = {}
14749 amit.gupta 666
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 667
        status = MStore.ORDER_SHIPPED
668
        #print page
14624 amit.gupta 669
        soup = BeautifulSoup(page)
16468 amit.gupta 670
        header1 = soup.find("h1") 
671
        if header1:
672
            if header1.text=="Sign In":
673
                print "Login page is displayed for order id", orderId
17307 amit.gupta 674
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 675
                return subOrder
16468 amit.gupta 676
        try:
16474 amit.gupta 677
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 678
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
679
            try:
680
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
681
            except:
682
                displayStatus = detailedStatus
683
            print displayStatus
684
            if detailedStatus.lower().find("delivered")>=0:
685
                print detailedStatus
686
                displayStatus = "Delivered"
687
                status = "Delivered"
688
                try:
689
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
690
                except:
691
                    pass
692
 
693
            elif detailedStatus.lower() == 'returned':
694
                status = 'Cancelled'
695
            subOrder['status'] = status
696
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 697
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 698
 
699
        except:
17350 amit.gupta 700
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 701
            print "failed to parse", orderId
702
            traceback.print_exc()
703
 
14624 amit.gupta 704
        return subOrder
705
 
706
 
707
 
16980 amit.gupta 708
    def parseInfo(self,):
709
        from pyquery import PyQuery as pq
16986 amit.gupta 710
        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 711
        try:
712
            for order in orders:
713
                try:
714
                    doc = pq(order.rawhtml)
715
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
716
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 717
                    orderInfo = All_user_addresses()
16980 amit.gupta 718
                    orderInfo.address = lists[-3].text
719
                    orderInfo.user_id = order.user_id
16986 amit.gupta 720
                    orderInfo.source = 'order' 
721
                    #orderInfo.order_id = order.id
722
                    #orderInfo.email = None
723
                    #orderInfo.name = lists[0].text 
724
                    #orderInfo.mobile = None
16980 amit.gupta 725
                    adSplit = lists[-2].text.split(",")
726
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
727
                    if match:
728
                        items = match.groups()
16986 amit.gupta 729
                    orderInfo.city = adSplit[0].strip()
730
                    orderInfo.pincode = items[1].strip()
731
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 732
                    session.commit()
733
                except:
734
                    session.rollback()
735
                    continue
736
        finally:
737
            session.close()
14608 amit.gupta 738
 
13927 amit.gupta 739
 
740
 
13774 amit.gupta 741
def main():
17251 amit.gupta 742
    #getSummaryFile("/AmazonTrack/User8703/")
16874 amit.gupta 743
    #process_rejects("/home/amit/Downloads/RejectCashbackTemplate.xlsx")
16283 amit.gupta 744
    #str1 = readSSh("/AmazonTrack/User2466/orderSummary28-06:13:25:08")
15046 amit.gupta 745
    #readSSh("/tmp/User2/404-2225153-7073122")
16874 amit.gupta 746
    #for orderId, trackingUrl in all.iteritems():
747
    store = getStore(1)
17361 amit.gupta 748
    store.parseOrderRawHtml(46328, '3442342424334346', 9722, readSSh("/home/amit/46328"), "https://www.amazon.in/gp/css/summary/edit.html?orderID=403-4276089-1251522")
16874 amit.gupta 749
     #   print store.parseTrackingUrl(trackingUrl, orderId)
17357 amit.gupta 750
    #store.scrapeStoreOrders()
14650 amit.gupta 751
#    store.scrapeStoreOrders()
17230 amit.gupta 752
    #store.parseInfo()
17331 amit.gupta 753
    #store.trackOrdersForUser(4355, "", readSSh("~/4355"))
14813 amit.gupta 754
    #readSSh("/tmp/User211/2015-04-12 10:32:41.905765")
14703 amit.gupta 755
    #store.scrapeAffiliate()
15565 amit.gupta 756
    #parseDetailNotCreated()
16283 amit.gupta 757
    #parseOrderNotCreated()
17331 amit.gupta 758
#    for userId in store.db.merchantOrder.find({"storeId":1, "subOrders.closed":False}).distinct("userId"):
759
#        userId = int(userId)
760
#        directory = "/AmazonTrack/User" + str(userId) + "/"
761
#        summaryFile=getSummaryFile(directory)
762
#        if summaryFile is None:
763
#            print "No summary for", userId
764
#        else:
765
#            print userId, directory+summaryFile
766
#            store.trackOrdersForUser(userId, 'https://www.amazon.in/gp/css/order-history', readSSh(directory + summaryFile))
17358 amit.gupta 767
#"storeId":1, "status":"parse_failed", "subOrders":{"$exists":False}
17361 amit.gupta 768
    for order in store.db.merchantOrder.find({"subOrders":{"$exists":True}, "subOrders.trackingUrl":{"$exists":False}}):
17334 amit.gupta 769
        try:
17358 amit.gupta 770
            orderId=order.get("orderId")
17335 amit.gupta 771
            o = session.query(OrdersRaw).filter_by(id = orderId).one()
17361 amit.gupta 772
            store.parseOrderRawHtml(o.id, o.sub_tag, o.user_id, o.rawhtml, order['orderTrackingUrl'])
17334 amit.gupta 773
        except:
17335 amit.gupta 774
            print "Could not parse", orderId
17360 amit.gupta 775
        finally:
776
            session.close()
17231 amit.gupta 777
    #store.trackOrdersForUser(8703, 'https://www.amazon.in/gp/css/order-history', readSSh('/home/amit/Downloads/orderSummary06-10_12_15_54'))
778
 
779
def getSummaryFile(directory):
780
    date1 = datetime(2015,1,1)
781
    finalFile = None
782
    try:
783
        for file in os.listdir(directory):
784
            if file.startswith("orderSummary"):
785
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
786
                if date2 > date1:
787
                    date1 = date2
788
                    finalFile=file
789
    except:
790
        print "Missing directory"
17251 amit.gupta 791
    return finalFile 
17231 amit.gupta 792
 
15555 amit.gupta 793
 
794
def parseDetailNotCreated():
795
    try:
796
        store=getStore(1)
797
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 798
        session.close()
15555 amit.gupta 799
        for order in orders:
800
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
801
 
802
    finally:
803
        session.close()
17251 amit.gupta 804
 
805
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 806
    print dateString
17251 amit.gupta 807
    if dateString.lower()=='today':
808
        return date.today()
809
    if dateString.lower()=='yesterday':
810
        return date.today() - timedelta(days=1)
811
    try:
812
        return datetime.strptime(dateString, '%d-%b-%y')
813
    except:
814
        try:
815
            #get Closest Date from today
816
            curDate = date.today()
817
            curTime = datetime(curDate.year, curDate.month, curDate.day)
818
            curYear = curDate.year
819
            prevYear = curYear - 1
17254 amit.gupta 820
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
821
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 822
            if dateMax <= curTime:
823
                return dateMax
824
            else:
825
                return dateMin 
826
        except:
17254 amit.gupta 827
            try:
828
                days_of_week = ['sunday','monday','tuesday','wednesday',
829
                            'thursday','friday','saturday']
830
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
831
                if deltaDays <= 0:
832
                    deltaDays= deltaDays + 7
833
                curDate = curDate - timedelta(days=deltaDays)
834
                print datetime.strftime(curDate, '%d-%b-%y')
835
                return curDate
836
            except:
837
                print "could not parse"
838
                return None
15565 amit.gupta 839
 
17254 amit.gupta 840
def getDateStringArriving(dateString='Thursday'):
841
    print dateString
842
    if dateString.lower()=='today':
843
        return date.today()
844
    if dateString.lower()=='tomorrow':
845
        return date.today() + timedelta(days=1)
846
    try:
847
        return datetime.strptime(dateString, '%d-%b-%y')
848
    except:
849
        try:
850
            #get Closest Date from today
851
            curDate = date.today()
852
            curTime = datetime(curDate.year, curDate.month, curDate.day)
853
            curYear = curDate.year
854
            nextYear = curYear + 1
855
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
856
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
857
            if dateMin >= curTime:
858
                return dateMin
859
            else:
860
                return dateMax 
861
        except:
862
            try:
863
                days_of_week = ['sunday','monday','tuesday','wednesday',
864
                            'thursday','friday','saturday']
865
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
866
                if deltaDays < 0:
867
                    deltaDays= deltaDays + 7
868
                curDate = curDate + timedelta(days=deltaDays)
869
                return curDate
870
            except:
871
                print "Could not parse"
872
                return None 
873
 
15565 amit.gupta 874
def parseOrderNotCreated():
875
    try:
876
        store=getStore(1)
877
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
878
        session.close()
879
        for order in orders:
15567 amit.gupta 880
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
881
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
882
            order1.status = result
883
            session.commit()
15565 amit.gupta 884
 
885
    finally:
886
        session.close()
15555 amit.gupta 887
 
13774 amit.gupta 888
if __name__ == '__main__':
17331 amit.gupta 889
#        readSSh("~/46417.html")
17321 amit.gupta 890
        main()
17254 amit.gupta 891
        #getDateStringArriving()