Subversion Repositories SmartDukaan

Rev

Rev 18031 | Rev 18156 | 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
18028 amit.gupta 11
from dtr.dao import Order, SubOrder, AmazonAffiliateInfo, obj
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
18028 amit.gupta 26
import pyquery
13774 amit.gupta 27
import re
14033 amit.gupta 28
import time
13809 amit.gupta 29
import traceback
14650 amit.gupta 30
import urllib2
13774 amit.gupta 31
 
32
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 33
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 34
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
14624 amit.gupta 35
AMAZON_AFF_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/listReports'
14650 amit.gupta 36
AMAZON_AFF_FILE_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/getReport?filename=saholic-21-orders-report-%s.tsv.gz'
13774 amit.gupta 37
class Store(MStore):
13821 amit.gupta 38
 
17348 amit.gupta 39
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch', 'order received'],
14064 amit.gupta 40
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
17324 amit.gupta 41
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled', 'replacement complete', 'return received'],
17258 amit.gupta 42
                            MStore.ORDER_DELIVERED : ['delivered', 'your package was delivered', 'package was handed directly to customer']
13821 amit.gupta 43
                           }
13774 amit.gupta 44
 
45
    def __init__(self,store_id):
46
        super(Store, self).__init__(store_id)
47
 
48
    def getName(self):
14749 amit.gupta 49
        return "amazon"
13774 amit.gupta 50
 
14945 amit.gupta 51
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, track=False):
17718 amit.gupta 52
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14958 amit.gupta 53
        parseString = "Tracking" if track else "Transacted"
54
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 55
        resp = {}
14813 amit.gupta 56
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 57
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 58
            try:
17600 amit.gupta 59
                from pyquery import PyQuery as pq
60
                doc = pq(rawHtml)
14608 amit.gupta 61
                try:
17600 amit.gupta 62
                    orderUrl = doc('a.a-touch-link').attr.href
14608 amit.gupta 63
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
64
                except:
17600 amit.gupta 65
                    merchantOrderId = doc("#orders-list>div>span>b").html()
14699 amit.gupta 66
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 67
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
68
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 69
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 70
                order.requireDetail = True
14698 amit.gupta 71
                order.status = 'html_required'
14297 amit.gupta 72
                order.closed = None
14312 amit.gupta 73
                if self._saveToOrder(todict(order)):
74
                    resp['result'] = 'ORDER_CREATED'
75
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
76
                    resp["htmlRequired"] = True
77
                    resp['orderId'] = orderId
78
                else:
79
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
80
 
13774 amit.gupta 81
            except:
14813 amit.gupta 82
                #Write all cases here for Order Not created Known
17600 amit.gupta 83
                soup = BeautifulSoup(rawHtml)
14809 amit.gupta 84
                try:
16874 amit.gupta 85
                    if not soup.body:
86
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
17359 amit.gupta 87
                    elif 'Securely redirecting you' in soup.find("h3").text.strip() or soup.find("h3").text.strip()=="Orders":
14813 amit.gupta 88
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 89
                    else:
14813 amit.gupta 90
                        raise
14809 amit.gupta 91
                except:
14814 amit.gupta 92
                    try:
17356 amit.gupta 93
                        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 94
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
95
                        else:
96
                            raise
97
                    except:
15046 amit.gupta 98
                        try:
17356 amit.gupta 99
                            if soup.find("h2").text.strip() in ['Web page not available','Webpage not available', 'Do you have an Amazon password?']:
15046 amit.gupta 100
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
101
                            else:
102
                                raise
103
                        except:
15565 amit.gupta 104
                            try:
16876 amit.gupta 105
                                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 106
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
16877 amit.gupta 107
                                elif soup.find("b", {'class':'h1'}).text.strip().find("We're sorry") > -1:
108
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
15565 amit.gupta 109
                                else:
110
                                    raise
111
                            except:
112
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
113
 
13774 amit.gupta 114
        else:
13781 amit.gupta 115
            try:
14722 amit.gupta 116
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
117
                if mo is not None:
14749 amit.gupta 118
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 119
                    merchantOrder.createdOn = mo.get("createdOn")
120
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
121
                else:
14945 amit.gupta 122
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 123
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14464 amit.gupta 124
                soup = BeautifulSoup(rawHtml)
17331 amit.gupta 125
                if not soup.body:
17319 amit.gupta 126
                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
127
                else:   
14545 amit.gupta 128
                    try:
17319 amit.gupta 129
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 130
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 131
                    except:
15056 amit.gupta 132
                        try:
17319 amit.gupta 133
                            traceback.print_exc()
134
                            self.parseOldStlye(merchantOrder, soup)
135
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 136
                        except:
17319 amit.gupta 137
                            traceback.print_exc()
15059 amit.gupta 138
                            try:
17319 amit.gupta 139
                                self.parseCancelled(merchantOrder, soup)
140
                                resp['result'] = 'ORDER_CANCELLED'
15059 amit.gupta 141
                            except:
17319 amit.gupta 142
                                try:
17356 amit.gupta 143
                                    if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
17319 amit.gupta 144
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
145
                                    else:
146
                                        raise
147
                                except:
17356 amit.gupta 148
                                    if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
17319 amit.gupta 149
                                        resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
150
                                    else:
151
                                        raise
17599 amit.gupta 152
                if resp['result'] == 'DETAIL_NOT_CREATED_KNOWN':
17357 amit.gupta 153
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 154
            except:
17666 amit.gupta 155
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
14813 amit.gupta 156
                print "Error occurred"
15046 amit.gupta 157
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 158
                traceback.print_exc()
15046 amit.gupta 159
        return resp    
13774 amit.gupta 160
 
13868 amit.gupta 161
    #This should be exposed from api for specific sources
162
    def scrapeStoreOrders(self):
14624 amit.gupta 163
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
164
        for merchantOrder in orders:
14981 amit.gupta 165
            executeBulk = False
14624 amit.gupta 166
            try:
167
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
168
                closed = True
169
                map1 = {}
170
                for subOrder in merchantOrder.get("subOrders"):
171
                    if subOrder.get("closed"):
172
                        continue
173
                    elif subOrder.get("trackingUrl") is None:
174
                        closed = False
175
                        continue
176
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
177
                    trackingUrl = subOrder.get("trackingUrl")
178
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 179
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 180
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 181
                    if newOrder:
182
                        executeBulk = True
183
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
184
                        print findMap, "\n", updateMap
185
                        bulk.find(findMap).update({'$set' : updateMap})
186
                        closed = closed and newOrder['closed']
14981 amit.gupta 187
                if executeBulk:
14697 amit.gupta 188
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 189
                    bulk.execute()
14624 amit.gupta 190
            except:
14847 amit.gupta 191
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 192
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 193
                traceback.print_exc()
194
 
14608 amit.gupta 195
 
196
 
14771 amit.gupta 197
    def parserest(self, soup):
198
        print "Hi"
199
        if soup.find('h1'):
200
            print "OK"
14608 amit.gupta 201
 
14464 amit.gupta 202
    def parseOldStlye(self, merchantOrder, soup):
203
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
204
        table = soup.body.findAll("table", recursive=False)[1]
205
        #print table
206
        tables = table.tr.td.findAll("table", recursive=False)
207
        for tr in tables[2].findAll("tr"):
208
            boldElement = tr.td.b
209
            if "Order Placed" in str(boldElement):
210
                merchantOrder.placedOn = boldElement.next_sibling.strip()
211
            if "order number" in str(boldElement):
212
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
213
            if "Order Total" in str(boldElement):
214
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
215
        anchors = table.tr.td.findAll("a", recursive=False)
216
        paymentAnchor = anchors.pop(-1)
217
 
218
        count = 0
219
        subOrders = []
220
        merchantOrder.subOrders = subOrders
221
        counter = 0 
222
        for anchor in anchors:
223
            count += 1
224
            tab = anchor.next_sibling
225
            status = MStore.ORDER_PLACED
226
            subStr = "Delivery #" + str(count) + ":"
227
            if subStr in  tab.find("b").text:
228
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
229
 
230
            tab = tab.next_sibling.next_sibling
231
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
232
 
233
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
234
 
235
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
236
            i = -1
237
            for orderItemTr in orderItemTrs:
238
                i += 1
239
                if i%2 == 0:
240
                    continue
241
                counter += 1
242
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
243
 
244
                productUrl = orderItemTr.td.contents[1].a["href"]
245
                productTitle = orderItemTr.td.contents[1].a.text
246
 
247
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
248
 
249
 
250
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
251
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
252
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 253
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
254
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
255
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 256
                subOrder.productCode = productUrl.split('/')[5]
257
                subOrder.detailedStatus = detailedStatus
18028 amit.gupta 258
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice)
15346 amit.gupta 259
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
260
                subOrder.dealRank = dealRank.get('rank')
261
                subOrder.rankDesc = dealRank.get('description')
16283 amit.gupta 262
                subOrder.maxNlc = dealRank.get('maxNlc')
263
                subOrder.minNlc = dealRank.get('minNlc')
264
                subOrder.db = dealRank.get('dp')
265
                subOrder.itemStatus = dealRank.get('status')
14464 amit.gupta 266
                cashbackStatus = Store.CB_PENDING
267
                if cashbackAmount <= 0:
268
                    cashbackStatus = Store.CB_NA
269
                subOrder.cashBackStatus = cashbackStatus
18028 amit.gupta 270
                subOrder.cashBackAmount = cashbackAmount*quantity
14464 amit.gupta 271
                if percentage > 0:
272
                    subOrder.cashBackPercentage = percentage
273
                subOrders.append(subOrder)
274
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
275
        totalAmount = 0
276
        grandAmount = 0
277
        for price in priceList:
278
            labelTd = price.td
279
            if 'Subtotal:' in labelTd.text:
280
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
281
            elif 'Grand Total:' in labelTd.text:
282
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
283
        if grandAmount < totalAmount:
284
            diff = totalAmount - grandAmount
285
            for subOrder in merchantOrder.subOrders:
286
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 287
        merchantOrder.status='success'
14464 amit.gupta 288
        self._updateToOrder(todict(merchantOrder))
289
 
290
    def parseNewStlye(self, merchantOrder, soup):
291
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
292
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 293
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
294
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 295
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
296
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
297
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 298
        try:
299
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
300
        except:
301
            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 302
        priceRows = priceBox.findAll('div', {'class':'a-row'})
303
        subTotal = 0
14566 amit.gupta 304
        shippingPrice = 0   
14464 amit.gupta 305
        promoApplied = 0
306
        for priceRow in priceRows:
307
            if "Item(s) Subtotal:" in str(priceRow):
308
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
309
            elif "Shipping:" in str(priceRow):
310
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
311
            elif "Grand Total:" in str(priceRow):
312
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
313
                merchantOrder.paidAmount = grandPrice
314
            elif "Total:" in str(priceRow):
315
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
316
            elif "Promotion Applied:" in str(priceRow):
317
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
318
        totalPaid = subTotal        
319
        if promoApplied > 0:
320
            totalPaid -= promoApplied
321
            if shippingPrice <= promoApplied:
322
                totalPaid  += shippingPrice
15473 amit.gupta 323
 
17601 amit.gupta 324
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 325
        subOrders = []
326
        merchantOrder.subOrders = subOrders
15473 amit.gupta 327
        closedStatus = True
17601 amit.gupta 328
        subOrders = []
14464 amit.gupta 329
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 330
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 331
            try:
332
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
333
            except:
334
                trackingUrl= None
335
 
17347 amit.gupta 336
            deliverySpanTop = None
14464 amit.gupta 337
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
338
            statusDiv = innerBoxes[0]
339
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 340
            try:
341
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
342
            except:
17617 amit.gupta 343
                pass
18030 amit.gupta 344
            #if not deliverySpanTop:
345
            productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 346
            merchantOrder.subOrders = subOrders
18028 amit.gupta 347
            j=0
14464 amit.gupta 348
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 349
                deliverySpan = deliverySpanTop
350
                if not deliverySpanTop:
351
                    if i%2==1:
352
                        continue
18028 amit.gupta 353
                    deliverySpan = productDiv.div.div.span
354
                    productDiv = productDivs[i+1]
355
                for  prodRow in productDiv.findAll('div', recursive=False):
356
                    j += 1
357
                    imgDiv  = prodRow.div.div
358
                    detailDiv = imgDiv.find_next_sibling('div')
359
                    detailDivs = detailDiv.findAll('div', recursive=False)
360
                    arr = re.split("^(\d+) of", detailDivs[0].a.text.strip())
361
                    (productTitle, quantity) = (arr[-1], (1 if len(arr)<2 else int(arr[1])) )
14756 amit.gupta 362
                    try:
18028 amit.gupta 363
                        unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
14756 amit.gupta 364
                    except:
18028 amit.gupta 365
                        unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
366
                    amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
367
                    productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
368
                    subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
369
                    subOrder.productCode = productUrl.split('/')[5]
370
                    subOrder.unitPrice = unitPrice
371
                    subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
372
                    estDlvyTime = datetime.now()
373
                    if deliverySpan is not None:
374
                        try:
375
                            subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
376
                            estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
377
                            subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
378
                            estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
379
                        except:
380
                            if "Delivered on" in deliverySpan.text:
381
                                subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
382
                            subOrder.estimatedDeliveryDate = "Not available"
383
 
384
                    createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
385
                    subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
386
                    subOrder.detailedStatus = subOrderStatus
387
                    status=MStore.ORDER_PLACED
17599 amit.gupta 388
                    try:
18028 amit.gupta 389
                        status = self._getStatusFromDetailedStatus(subOrderStatus)
17599 amit.gupta 390
                    except:
391
                        try:
18028 amit.gupta 392
                            dateString = subOrderStatus.split("Delivered ")[1].strip()
393
                            subOrder.status = MStore.ORDER_DELIVERED
394
                            subOrder.detailedStatus = 'Delivered'
395
                            dateString = getDateStringDelivered(dateString)
17631 amit.gupta 396
                            if dateString is not None:
18028 amit.gupta 397
                                subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 398
                        except:
18028 amit.gupta 399
                            try:
400
                                dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
401
                                subOrder.status = MStore.ORDER_SHIPPED
402
                                dateString = getDateStringArriving(dateString)
403
                                if dateString is not None:
404
                                    subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
405
                                else:
406
                                    subOrder.deliveryEstimate = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
407
                            except:
408
                                print "Unknown status Alert -", status
409
 
410
                    subOrder.deliveryCharges = shippingPrice
411
                    if trackingUrl:
18107 amit.gupta 412
                        subOrder.tracingUrl  = trackingUrl
18028 amit.gupta 413
                    subOrder.imgUrl = imgDiv.img["src"]
414
                    (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid/quantity)
415
                    dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
416
                    subOrder.dealRank = dealRank.get('rank')
417
                    subOrder.rankDesc = dealRank.get('description')
418
                    cashbackStatus = Store.CB_PENDING
419
                    if cashbackAmount <= 0:
420
                        cashbackStatus = Store.CB_NA
421
                    subOrder.cashBackStatus = cashbackStatus
422
                    subOrder.cashBackAmount = cashbackAmount*quantity
423
                    if percentage > 0:
424
                        subOrder.cashBackPercentage = percentage
425
                    if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
426
                        subOrder.status = Store.ORDER_DELIVERED
427
                        subOrder.closed = True
428
                        if subOrder.cashBackStatus == Store.CB_PENDING:
429
                            subOrder.cashBackStatus = Store.CB_APPROVED
430
                    elif closedStatus:
431
                        closedStatus= False
432
                    subOrders.append(subOrder)
14698 amit.gupta 433
        merchantOrder.status='success'
15473 amit.gupta 434
        merchantOrder.closed = closedStatus
14464 amit.gupta 435
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 436
 
437
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 438
        try:
439
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
440
            if fonts[0].text == "Important Message":
441
                if fonts[1].text=="This order has been cancelled.":
442
                    merchantOrder.closed = True
443
                    merchantOrder.status = "cancelled"
444
                    merchantOrder.requireDetail = False
445
                    self._updateToOrder(todict(merchantOrder))
446
                    return
447
                else:
448
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 449
            else:
450
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 451
        except:
14549 amit.gupta 452
            orderDetails = soup.body.find(id="orderDetails")
453
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
454
                merchantOrder.closed = True
14698 amit.gupta 455
                merchantOrder.status = "cancelled"
14549 amit.gupta 456
                merchantOrder.requireDetail = False
457
                self._updateToOrder(todict(merchantOrder))
458
            else:
459
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 460
 
13868 amit.gupta 461
    def getTrackingUrls(self, userId):
14074 amit.gupta 462
 
463
        missingOrderUrls = []
464
        missingOrders = self._getMissingOrders({'userId':userId})
465
        for missingOrder in missingOrders:
466
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 467
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 468
        count = len(orders)
13879 amit.gupta 469
        print "count", count
14948 amit.gupta 470
        print "Missing Urls"
471
        print "*************"
472
        print missingOrderUrls
13868 amit.gupta 473
        if count > 0:
14684 amit.gupta 474
            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 475
        else: 
14074 amit.gupta 476
            return missingOrderUrls
13927 amit.gupta 477
 
478
    def trackOrdersForUser(self, userId, url, rawHtml):
17718 amit.gupta 479
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14985 amit.gupta 480
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 481
        if not os.path.exists(directory):
482
            os.makedirs(directory)
14173 amit.gupta 483
 
14945 amit.gupta 484
 
13995 amit.gupta 485
        try:
486
            searchMap = {'userId':userId}
487
            collectionMap = {'merchantOrderId':1}
488
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 489
            datetimeNow = datetime.now()
490
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 491
            print "url----------------", url
17465 amit.gupta 492
            cancelledSummary = False
14320 amit.gupta 493
            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 494
                if url == 'https://www.amazon.in/gp/css/order-history':
495
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
496
                else:   
17465 amit.gupta 497
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
498
                    cancelledSummary = True   
14945 amit.gupta 499
                f = open(filename,'w')
500
                f.write(rawHtml) # python will convert \n to os.linesep
501
                f.close() # you can omit in most cases as the destructor will call if
13995 amit.gupta 502
                soup = BeautifulSoup(rawHtml)
503
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
504
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
505
                for activeOrder in activeOrders:
17307 amit.gupta 506
                    matched=False
13995 amit.gupta 507
                    for orderEle in allOrders:
17254 amit.gupta 508
                        deliveredOn = None
509
                        deliveryEstimate = None
17258 amit.gupta 510
                        shippingEstimate = None
13995 amit.gupta 511
                        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'})
512
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
513
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 514
                            matched=True
13995 amit.gupta 515
                            closed = True
17465 amit.gupta 516
                            if not cancelledSummary:
517
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
518
                            else:
519
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
520
                                shipments.pop(0)
13995 amit.gupta 521
                            for shipment in shipments:
17307 amit.gupta 522
                                orderStatusDesc = None
13995 amit.gupta 523
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 524
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 525
                                try:
17270 amit.gupta 526
                                    orderStatus = sdivs[0].span.text.strip()
527
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 528
                                except:
17270 amit.gupta 529
                                    try:
530
                                        dateString = orderStatus.split("Delivered ")[1].strip()
531
                                        status = MStore.ORDER_DELIVERED
532
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
533
                                    except:
534
                                        try:
17271 amit.gupta 535
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 536
                                            status = MStore.ORDER_SHIPPED
537
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
538
                                        except:
539
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 540
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 541
                                try:
17270 amit.gupta 542
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
543
                                except:
17271 amit.gupta 544
                                    try:
545
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
546
                                    except:
17307 amit.gupta 547
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 548
 
17271 amit.gupta 549
                                if orderStatusDesc: 
17348 amit.gupta 550
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 551
                                    try:
17348 amit.gupta 552
                                        status = self._getStatusFromDetailedStatus(orderStatus)
553
                                    except:
554
                                        pass
555
                                    try:
17271 amit.gupta 556
                                        if "Dispatch estimate" in orderStatusDesc:
557
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
558
                                        elif "Delivery estimate" in orderStatus:
559
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
560
                                        elif "Arriving" in orderStatus:
561
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
562
                                    except:
563
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
564
                                        closed=False
565
                                        status = None
17258 amit.gupta 566
 
14335 amit.gupta 567
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 568
                                trackingUrl = None
569
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
570
                                    if buttonDiv.find('a').text.strip()=='Track package':
571
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
572
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 573
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 574
                                        break
575
                                for prodDiv in productDivs:
576
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
577
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
578
                                    imgUrl = prodDiv.find("img")["src"]
579
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 580
                                        if subOrder['closed']==True:
581
                                            continue
17230 amit.gupta 582
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 583
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
584
                                            updateMap = {}
585
                                            closedStatus = False
586
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 587
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 588
                                            if status:
17348 amit.gupta 589
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 590
                                                updateMap['subOrders.$.status'] = status 
591
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 592
 
17251 amit.gupta 593
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 594
                                                if deliveredOn:                               
17254 amit.gupta 595
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 596
                                                closedStatus = True
597
                                                updateMap['subOrders.$.closed'] = True
598
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 599
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 600
                                            if status==MStore.ORDER_CANCELLED:     
601
                                                closedStatus = True
602
                                                updateMap['subOrders.$.closed'] = True
603
                                                if cashbackStatus == Store.CB_PENDING:
604
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 605
                                            if status==MStore.ORDER_SHIPPED:
606
                                                if deliveryEstimate:   
607
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 608
                                                if trackingUrl is not None:
609
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 610
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 611
                                            if shippingEstimate:   
612
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 613
                                            if not closedStatus:
614
                                                closed = False
17307 amit.gupta 615
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 616
                                            updateMap['status']='success'
13995 amit.gupta 617
                                            bulk.find(findMap).update({'$set' : updateMap})
618
                                            break
17321 amit.gupta 619
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 620
                            break
621
                    if not matched:
17319 amit.gupta 622
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 623
                        for subOrder in activeOrder['subOrders']:
624
                            if subOrder['closed']==True:
625
                                continue
626
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
627
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
628
                            bulk.find({'orderId': activeOrder['orderId']})
629
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 630
                bulk.execute()
631
                return 'PARSED_SUCCESS'
632
            else:
15563 amit.gupta 633
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
634
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 635
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 636
 
637
                filename = directory + "/" + merchantOrderId   
638
                f = open(filename,'w')
639
                f.write(rawHtml) # python will convert \n to os.linesep
640
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 641
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 642
                print "result", result
15555 amit.gupta 643
                try:
15560 amit.gupta 644
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
645
                    order1.status = result
646
                    order1.rawhtml = rawHtml
17358 amit.gupta 647
                    order1.order_url = url
15555 amit.gupta 648
                    session.commit()
15561 amit.gupta 649
                except:
650
                    traceback.print_exc()
15555 amit.gupta 651
                finally:
652
                    session.close()
14080 amit.gupta 653
                return 'PARSED_SUCCESS'
654
                pass
14008 amit.gupta 655
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 656
        except:
657
            traceback.print_exc()    
658
            return 'PARSED_FAILED'
659
 
660
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 661
        if "ordered from" in detailedStatus.lower():
662
            return MStore.ORDER_PLACED 
663
 
13995 amit.gupta 664
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 665
            if detailedStatus.lower() in value:
13995 amit.gupta 666
                return key
667
 
17253 amit.gupta 668
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 669
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 670
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 671
        br = getBrowserObject()
672
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
673
        url = AMAZON_AFF_URL
674
        response = br.open(url)
675
        #get data for past 40 days and store it to mongo
676
        dt = datetime.now()
14701 amit.gupta 677
        dat = dt - timedelta(days=2)
14651 amit.gupta 678
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
679
        response = br.open(url)
680
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
681
        j=-1
682
        for row in page.split("\n"):
683
            j += 1
684
            if j== 0 or j==1:
685
                continue
686
            fields = row.split("\t")
687
            if len(fields)>1:
688
                print fields
689
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
690
                print amazonAffiliate
691
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
692
            else:
693
                break
14650 amit.gupta 694
 
14608 amit.gupta 695
 
16468 amit.gupta 696
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 697
        print trackingUrl
14624 amit.gupta 698
        subOrder = {}
14749 amit.gupta 699
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 700
        status = MStore.ORDER_SHIPPED
701
        #print page
14624 amit.gupta 702
        soup = BeautifulSoup(page)
16468 amit.gupta 703
        header1 = soup.find("h1") 
704
        if header1:
705
            if header1.text=="Sign In":
706
                print "Login page is displayed for order id", orderId
17307 amit.gupta 707
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 708
                return subOrder
16468 amit.gupta 709
        try:
16474 amit.gupta 710
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 711
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
712
            try:
713
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
714
            except:
715
                displayStatus = detailedStatus
716
            print displayStatus
717
            if detailedStatus.lower().find("delivered")>=0:
718
                print detailedStatus
719
                displayStatus = "Delivered"
720
                status = "Delivered"
721
                try:
722
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
723
                except:
724
                    pass
725
 
726
            elif detailedStatus.lower() == 'returned':
727
                status = 'Cancelled'
728
            subOrder['status'] = status
729
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 730
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 731
 
732
        except:
17350 amit.gupta 733
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 734
            print "failed to parse", orderId
735
            traceback.print_exc()
736
 
14624 amit.gupta 737
        return subOrder
738
 
739
 
740
 
16980 amit.gupta 741
    def parseInfo(self,):
742
        from pyquery import PyQuery as pq
16986 amit.gupta 743
        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 744
        try:
745
            for order in orders:
746
                try:
747
                    doc = pq(order.rawhtml)
748
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
749
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 750
                    orderInfo = All_user_addresses()
16980 amit.gupta 751
                    orderInfo.address = lists[-3].text
752
                    orderInfo.user_id = order.user_id
16986 amit.gupta 753
                    orderInfo.source = 'order' 
754
                    #orderInfo.order_id = order.id
755
                    #orderInfo.email = None
756
                    #orderInfo.name = lists[0].text 
757
                    #orderInfo.mobile = None
16980 amit.gupta 758
                    adSplit = lists[-2].text.split(",")
759
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
760
                    if match:
761
                        items = match.groups()
16986 amit.gupta 762
                    orderInfo.city = adSplit[0].strip()
763
                    orderInfo.pincode = items[1].strip()
764
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 765
                    session.commit()
766
                except:
767
                    session.rollback()
768
                    continue
769
        finally:
770
            session.close()
14608 amit.gupta 771
 
13927 amit.gupta 772
 
773
 
13774 amit.gupta 774
def main():
16874 amit.gupta 775
    store = getStore(1)
18031 amit.gupta 776
    #store.parseOrderRawHtml("444444", '123', 14, readSSh('/home/amit/amazon.html'), 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515')
17641 amit.gupta 777
    #[59433, 59472, 59487, 59503, 59545, 59549, 59573, 59591]
18031 amit.gupta 778
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
779
    session.close()
780
    for o in orders:
781
        try:
782
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
783
        finally:
784
            session.close()
17631 amit.gupta 785
    #store.trackOrdersForUser(10466, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=403-7498756-0837158', readSSh('/AmazonTrack/User10466/403-7498756-0837158'))
18030 amit.gupta 786
    #store.trackOrdersForUser(46195, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515', readSSh('/home/amit/amazon.html'))
17231 amit.gupta 787
 
788
def getSummaryFile(directory):
789
    date1 = datetime(2015,1,1)
790
    finalFile = None
791
    try:
792
        for file in os.listdir(directory):
793
            if file.startswith("orderSummary"):
794
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
795
                if date2 > date1:
796
                    date1 = date2
797
                    finalFile=file
798
    except:
799
        print "Missing directory"
17251 amit.gupta 800
    return finalFile 
17231 amit.gupta 801
 
15555 amit.gupta 802
 
803
def parseDetailNotCreated():
804
    try:
805
        store=getStore(1)
806
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 807
        session.close()
15555 amit.gupta 808
        for order in orders:
809
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
810
 
811
    finally:
812
        session.close()
17251 amit.gupta 813
 
814
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 815
    print dateString
17251 amit.gupta 816
    if dateString.lower()=='today':
817
        return date.today()
818
    if dateString.lower()=='yesterday':
819
        return date.today() - timedelta(days=1)
820
    try:
821
        return datetime.strptime(dateString, '%d-%b-%y')
822
    except:
823
        try:
824
            #get Closest Date from today
825
            curDate = date.today()
826
            curTime = datetime(curDate.year, curDate.month, curDate.day)
827
            curYear = curDate.year
828
            prevYear = curYear - 1
17254 amit.gupta 829
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
830
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 831
            if dateMax <= curTime:
832
                return dateMax
833
            else:
834
                return dateMin 
835
        except:
17254 amit.gupta 836
            try:
837
                days_of_week = ['sunday','monday','tuesday','wednesday',
838
                            'thursday','friday','saturday']
839
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
840
                if deltaDays <= 0:
841
                    deltaDays= deltaDays + 7
842
                curDate = curDate - timedelta(days=deltaDays)
843
                print datetime.strftime(curDate, '%d-%b-%y')
844
                return curDate
845
            except:
846
                print "could not parse"
847
                return None
15565 amit.gupta 848
 
17254 amit.gupta 849
def getDateStringArriving(dateString='Thursday'):
850
    print dateString
851
    if dateString.lower()=='today':
852
        return date.today()
853
    if dateString.lower()=='tomorrow':
854
        return date.today() + timedelta(days=1)
855
    try:
856
        return datetime.strptime(dateString, '%d-%b-%y')
857
    except:
858
        try:
859
            #get Closest Date from today
860
            curDate = date.today()
861
            curTime = datetime(curDate.year, curDate.month, curDate.day)
862
            curYear = curDate.year
863
            nextYear = curYear + 1
864
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
865
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
866
            if dateMin >= curTime:
867
                return dateMin
868
            else:
869
                return dateMax 
870
        except:
871
            try:
872
                days_of_week = ['sunday','monday','tuesday','wednesday',
873
                            'thursday','friday','saturday']
874
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
875
                if deltaDays < 0:
876
                    deltaDays= deltaDays + 7
877
                curDate = curDate + timedelta(days=deltaDays)
878
                return curDate
879
            except:
880
                print "Could not parse"
881
                return None 
882
 
18028 amit.gupta 883
def main1():
884
    store = getStore(1)
885
    for merchantOrder in store.db.merchantOrder.find({"subOrders":{"$elemMatch":{"cashBackStatus":"Not Applicable", "cashBackPercentage":{"$gt":0}}}}):
886
        mo = obj(merchantOrder)
887
        for subOrder in mo.subOrders:
888
            subOrder.closed=False 
889
        print "orderId", mo.orderId
890
        store.populateDerivedFields(mo, False)
891
        store.db.merchantOrder.update({"orderId":mo.orderId}, {"$set":todict(mo)})
892
        break
893
 
894
 
15565 amit.gupta 895
def parseOrderNotCreated():
896
    try:
897
        store=getStore(1)
898
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
899
        session.close()
900
        for order in orders:
15567 amit.gupta 901
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
902
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
903
            order1.status = result
904
            session.commit()
15565 amit.gupta 905
    finally:
906
        session.close()
15555 amit.gupta 907
 
13774 amit.gupta 908
if __name__ == '__main__':
17599 amit.gupta 909
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
910
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
17665 amit.gupta 911
    #readSSh("/AmazonTrack/User2482/orderSummary20-11:09:17:47")
18107 amit.gupta 912
    #main()
913
    store=getStore(1)
914
    store.scrapeStoreOrders()