Subversion Repositories SmartDukaan

Rev

Rev 18727 | Rev 20498 | 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
17307 amit.gupta 14
from dtr.storage.DataService import OrdersRaw, Orders, Order_Parse_Info, \
16986 amit.gupta 15
    All_user_addresses
16874 amit.gupta 16
from dtr.storage.Mongo import getDealRank
17307 amit.gupta 17
from dtr.utils import utils
18440 amit.gupta 18
from dtr.utils.utils import fetchResponseUsingProxy, readSSh, todict
16874 amit.gupta 19
from elixir import *
14650 amit.gupta 20
import base64
16874 amit.gupta 21
import dtr
14650 amit.gupta 22
import gzip
14624 amit.gupta 23
import mechanize
14285 amit.gupta 24
import os.path
18376 amit.gupta 25
from pyquery import PyQuery as pq
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):
17718 amit.gupta 51
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14958 amit.gupta 52
        parseString = "Tracking" if track else "Transacted"
53
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 54
        resp = {}
14813 amit.gupta 55
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 56
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 57
            try:
17600 amit.gupta 58
                doc = pq(rawHtml)
14608 amit.gupta 59
                try:
17600 amit.gupta 60
                    orderUrl = doc('a.a-touch-link').attr.href
14608 amit.gupta 61
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
62
                except:
17600 amit.gupta 63
                    merchantOrderId = doc("#orders-list>div>span>b").html()
14699 amit.gupta 64
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 65
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
66
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 67
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 68
                order.requireDetail = True
14698 amit.gupta 69
                order.status = 'html_required'
14297 amit.gupta 70
                order.closed = None
14312 amit.gupta 71
                if self._saveToOrder(todict(order)):
72
                    resp['result'] = 'ORDER_CREATED'
73
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
74
                    resp["htmlRequired"] = True
75
                    resp['orderId'] = orderId
76
                else:
77
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
78
 
13774 amit.gupta 79
            except:
14813 amit.gupta 80
                #Write all cases here for Order Not created Known
18158 amit.gupta 81
                soup = BeautifulSoup(rawHtml, "html5lib")
14809 amit.gupta 82
                try:
16874 amit.gupta 83
                    if not soup.body:
84
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
17359 amit.gupta 85
                    elif 'Securely redirecting you' in soup.find("h3").text.strip() or soup.find("h3").text.strip()=="Orders":
14813 amit.gupta 86
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 87
                    else:
14813 amit.gupta 88
                        raise
14809 amit.gupta 89
                except:
14814 amit.gupta 90
                    try:
17356 amit.gupta 91
                        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 92
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
93
                        else:
94
                            raise
95
                    except:
15046 amit.gupta 96
                        try:
17356 amit.gupta 97
                            if soup.find("h2").text.strip() in ['Web page not available','Webpage not available', 'Do you have an Amazon password?']:
15046 amit.gupta 98
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
99
                            else:
100
                                raise
101
                        except:
15565 amit.gupta 102
                            try:
16876 amit.gupta 103
                                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 104
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
16877 amit.gupta 105
                                elif soup.find("b", {'class':'h1'}).text.strip().find("We're sorry") > -1:
106
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
15565 amit.gupta 107
                                else:
108
                                    raise
109
                            except:
110
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
111
 
13774 amit.gupta 112
        else:
13781 amit.gupta 113
            try:
14722 amit.gupta 114
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
115
                if mo is not None:
14749 amit.gupta 116
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 117
                    merchantOrder.createdOn = mo.get("createdOn")
118
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
119
                else:
14945 amit.gupta 120
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 121
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
18158 amit.gupta 122
                soup = BeautifulSoup(rawHtml, "html5lib")
17331 amit.gupta 123
                if not soup.body:
17319 amit.gupta 124
                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
125
                else:   
14545 amit.gupta 126
                    try:
17319 amit.gupta 127
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 128
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 129
                    except:
15056 amit.gupta 130
                        try:
17319 amit.gupta 131
                            traceback.print_exc()
18376 amit.gupta 132
                            self.parseAnotherStlye(merchantOrder, pq(rawHtml))
17319 amit.gupta 133
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 134
                        except:
15059 amit.gupta 135
                            try:
18376 amit.gupta 136
                                traceback.print_exc()
137
                                self.parseOldStlye(merchantOrder, soup)
138
                                resp['result'] = 'DETAIL_CREATED'
15059 amit.gupta 139
                            except:
18376 amit.gupta 140
                                traceback.print_exc()
17319 amit.gupta 141
                                try:
18376 amit.gupta 142
                                    self.parseCancelled(merchantOrder, soup)
143
                                    resp['result'] = 'ORDER_CANCELLED'
17319 amit.gupta 144
                                except:
18376 amit.gupta 145
                                    try:
146
                                        if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
147
                                            resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
148
                                        else:
149
                                            raise
150
                                    except:
151
                                        if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
152
                                            resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
153
                                        else:
154
                                            raise
17599 amit.gupta 155
                if resp['result'] == 'DETAIL_NOT_CREATED_KNOWN':
17357 amit.gupta 156
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 157
            except:
17666 amit.gupta 158
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
14813 amit.gupta 159
                print "Error occurred"
15046 amit.gupta 160
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 161
                traceback.print_exc()
15046 amit.gupta 162
        return resp    
13774 amit.gupta 163
 
13868 amit.gupta 164
    #This should be exposed from api for specific sources
165
    def scrapeStoreOrders(self):
18725 amit.gupta 166
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}, "subOrders.login":{"$exists":False}})
14624 amit.gupta 167
        for merchantOrder in orders:
14981 amit.gupta 168
            executeBulk = False
14624 amit.gupta 169
            try:
170
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
171
                closed = True
172
                map1 = {}
173
                for subOrder in merchantOrder.get("subOrders"):
174
                    if subOrder.get("closed"):
175
                        continue
176
                    elif subOrder.get("trackingUrl") is None:
177
                        closed = False
178
                        continue
179
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
180
                    trackingUrl = subOrder.get("trackingUrl")
181
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 182
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 183
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 184
                    if newOrder:
185
                        executeBulk = True
186
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
187
                        print findMap, "\n", updateMap
188
                        bulk.find(findMap).update({'$set' : updateMap})
189
                        closed = closed and newOrder['closed']
14981 amit.gupta 190
                if executeBulk:
14697 amit.gupta 191
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 192
                    bulk.execute()
14624 amit.gupta 193
            except:
14847 amit.gupta 194
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 195
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 196
                traceback.print_exc()
197
 
14608 amit.gupta 198
 
199
 
14771 amit.gupta 200
    def parserest(self, soup):
201
        print "Hi"
202
        if soup.find('h1'):
203
            print "OK"
14608 amit.gupta 204
 
14464 amit.gupta 205
    def parseOldStlye(self, merchantOrder, soup):
206
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
207
        table = soup.body.findAll("table", recursive=False)[1]
208
        #print table
209
        tables = table.tr.td.findAll("table", recursive=False)
210
        for tr in tables[2].findAll("tr"):
211
            boldElement = tr.td.b
212
            if "Order Placed" in str(boldElement):
213
                merchantOrder.placedOn = boldElement.next_sibling.strip()
214
            if "order number" in str(boldElement):
215
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
216
            if "Order Total" in str(boldElement):
217
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
218
        anchors = table.tr.td.findAll("a", recursive=False)
219
        paymentAnchor = anchors.pop(-1)
220
 
221
        count = 0
222
        subOrders = []
223
        merchantOrder.subOrders = subOrders
224
        counter = 0 
225
        for anchor in anchors:
226
            count += 1
227
            tab = anchor.next_sibling
228
            status = MStore.ORDER_PLACED
229
            subStr = "Delivery #" + str(count) + ":"
230
            if subStr in  tab.find("b").text:
231
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
232
 
233
            tab = tab.next_sibling.next_sibling
234
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
235
 
236
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
237
 
238
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
239
            i = -1
240
            for orderItemTr in orderItemTrs:
241
                i += 1
242
                if i%2 == 0:
243
                    continue
244
                counter += 1
245
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
246
 
247
                productUrl = orderItemTr.td.contents[1].a["href"]
248
                productTitle = orderItemTr.td.contents[1].a.text
249
 
250
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
251
 
252
 
253
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
254
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
255
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 256
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
257
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
258
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 259
                subOrder.productCode = productUrl.split('/')[5]
260
                subOrder.detailedStatus = detailedStatus
18028 amit.gupta 261
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice)
15346 amit.gupta 262
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
263
                subOrder.dealRank = dealRank.get('rank')
264
                subOrder.rankDesc = dealRank.get('description')
16283 amit.gupta 265
                subOrder.maxNlc = dealRank.get('maxNlc')
266
                subOrder.minNlc = dealRank.get('minNlc')
267
                subOrder.db = dealRank.get('dp')
268
                subOrder.itemStatus = dealRank.get('status')
14464 amit.gupta 269
                cashbackStatus = Store.CB_PENDING
270
                if cashbackAmount <= 0:
271
                    cashbackStatus = Store.CB_NA
272
                subOrder.cashBackStatus = cashbackStatus
18028 amit.gupta 273
                subOrder.cashBackAmount = cashbackAmount*quantity
14464 amit.gupta 274
                if percentage > 0:
275
                    subOrder.cashBackPercentage = percentage
276
                subOrders.append(subOrder)
277
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
278
        totalAmount = 0
279
        grandAmount = 0
280
        for price in priceList:
281
            labelTd = price.td
282
            if 'Subtotal:' in labelTd.text:
283
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
284
            elif 'Grand Total:' in labelTd.text:
285
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
286
        if grandAmount < totalAmount:
287
            diff = totalAmount - grandAmount
288
            for subOrder in merchantOrder.subOrders:
289
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 290
        merchantOrder.status='success'
14464 amit.gupta 291
        self._updateToOrder(todict(merchantOrder))
18376 amit.gupta 292
 
293
    def parseAnotherStlye(self, merchantOrder, pqobj):
294
        counter=0
295
        detailSection = None
296
        shipmentSection = None
297
        summarySection = None
298
        for el in pqobj('.a-section'):
299
            if "View order details" in pq(el).text():
300
                detailSection =  pq(el)
301
                counter += 1
302
            if "Shipment details" in pq(el).text():
303
                shipmentSection =  pq(el)
304
                counter += 1
305
            if "Order Summary" in pq(el).text():
306
                summarySection = pq(el)
307
                counter += 1
308
            if counter == 3:
309
                break
310
 
311
        i=-1    
312
        for s in shipmentSection('.a-box-group'):
313
            shipmentGroup = pq(s)
314
            for shipment in shipmentGroup('.a-box'):
315
                i += 1 
316
                if i==0:
317
                    continue
318
                shipment = pq(shipment)
319
                shipmentStatusSection = shipment('.a-section:eq(0)')
320
                productDetails = shipment('.a-section:eq(1)>.a-row')
18380 amit.gupta 321
                print shipmentStatusSection('h3').text(), shipmentStatusSection('p').text(), shipmentStatusSection('span').text()
18376 amit.gupta 322
                for productDetail in productDetails:
323
                    productDetail = pq(productDetail)
324
                    pImg = productDetail.children('div').eq(0)
325
                    pQty = productDetail.children('div').eq(1)
326
                    #print pImg('a').attr('href'), pImg('a').attr('title'), pImg('img').attr('src'), pQty('.a-row:nth-child(2)')('span').text().split(':')[1].strip(), pQty('span.currencyINR')
18725 amit.gupta 327
                    productUrl = pImg('a').attr('href'), 
328
                    qty = pQty('.a-row:nth-child(2)')('span').text().split(':')[1].strip()
329
                    price = int(float(pQty('nobr').text().replace('Rs.','').replace(',', '')))
330
                    #subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
18376 amit.gupta 331
        raise   
332
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
333
        merchantOrder.placedOn
334
        merchantOrder.merchantOrderId
335
        merchantOrder.paidAmount
14464 amit.gupta 336
 
337
    def parseNewStlye(self, merchantOrder, soup):
338
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
339
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 340
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
341
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 342
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
343
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
344
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 345
        try:
346
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
347
        except:
348
            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 349
        priceRows = priceBox.findAll('div', {'class':'a-row'})
350
        subTotal = 0
14566 amit.gupta 351
        shippingPrice = 0   
14464 amit.gupta 352
        promoApplied = 0
353
        for priceRow in priceRows:
354
            if "Item(s) Subtotal:" in str(priceRow):
355
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
356
            elif "Shipping:" in str(priceRow):
357
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
358
            elif "Grand Total:" in str(priceRow):
359
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
360
                merchantOrder.paidAmount = grandPrice
361
            elif "Total:" in str(priceRow):
362
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
363
            elif "Promotion Applied:" in str(priceRow):
364
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
365
        totalPaid = subTotal        
366
        if promoApplied > 0:
367
            totalPaid -= promoApplied
368
            if shippingPrice <= promoApplied:
369
                totalPaid  += shippingPrice
15473 amit.gupta 370
 
17601 amit.gupta 371
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 372
        subOrders = []
373
        merchantOrder.subOrders = subOrders
15473 amit.gupta 374
        closedStatus = True
17601 amit.gupta 375
        subOrders = []
18161 amit.gupta 376
        j=0
14464 amit.gupta 377
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 378
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 379
            try:
380
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
381
            except:
382
                trackingUrl= None
383
 
17347 amit.gupta 384
            deliverySpanTop = None
14464 amit.gupta 385
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
386
            statusDiv = innerBoxes[0]
387
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 388
            try:
389
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
390
            except:
17617 amit.gupta 391
                pass
18030 amit.gupta 392
            #if not deliverySpanTop:
393
            productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 394
            merchantOrder.subOrders = subOrders
395
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 396
                deliverySpan = deliverySpanTop
397
                if not deliverySpanTop:
398
                    if i%2==1:
399
                        continue
18028 amit.gupta 400
                    deliverySpan = productDiv.div.div.span
18156 amit.gupta 401
                    try:
402
                        productDiv = productDivs[i+1]
403
                    except:
404
                        pass
18028 amit.gupta 405
                for  prodRow in productDiv.findAll('div', recursive=False):
406
                    j += 1
407
                    imgDiv  = prodRow.div.div
408
                    detailDiv = imgDiv.find_next_sibling('div')
409
                    detailDivs = detailDiv.findAll('div', recursive=False)
410
                    arr = re.split("^(\d+) of", detailDivs[0].a.text.strip())
411
                    (productTitle, quantity) = (arr[-1], (1 if len(arr)<2 else int(arr[1])) )
14756 amit.gupta 412
                    try:
18028 amit.gupta 413
                        unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
14756 amit.gupta 414
                    except:
18028 amit.gupta 415
                        unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
416
                    amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
417
                    productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
418
                    subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
419
                    subOrder.productCode = productUrl.split('/')[5]
420
                    subOrder.unitPrice = unitPrice
421
                    subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
422
                    estDlvyTime = datetime.now()
423
                    if deliverySpan is not None:
424
                        try:
425
                            subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
426
                            estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
427
                            subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
428
                            estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
429
                        except:
430
                            if "Delivered on" in deliverySpan.text:
431
                                subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
432
                            subOrder.estimatedDeliveryDate = "Not available"
433
 
434
                    createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
435
                    subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
436
                    subOrder.detailedStatus = subOrderStatus
437
                    status=MStore.ORDER_PLACED
17599 amit.gupta 438
                    try:
18028 amit.gupta 439
                        status = self._getStatusFromDetailedStatus(subOrderStatus)
17599 amit.gupta 440
                    except:
441
                        try:
18028 amit.gupta 442
                            dateString = subOrderStatus.split("Delivered ")[1].strip()
443
                            subOrder.status = MStore.ORDER_DELIVERED
444
                            subOrder.detailedStatus = 'Delivered'
445
                            dateString = getDateStringDelivered(dateString)
17631 amit.gupta 446
                            if dateString is not None:
18028 amit.gupta 447
                                subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 448
                        except:
18028 amit.gupta 449
                            try:
450
                                dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
451
                                subOrder.status = MStore.ORDER_SHIPPED
452
                                dateString = getDateStringArriving(dateString)
453
                                if dateString is not None:
454
                                    subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
455
                                else:
456
                                    subOrder.deliveryEstimate = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
457
                            except:
458
                                print "Unknown status Alert -", status
459
 
460
                    subOrder.deliveryCharges = shippingPrice
461
                    if trackingUrl:
18156 amit.gupta 462
                        subOrder.trackingUrl  = trackingUrl
18028 amit.gupta 463
                    subOrder.imgUrl = imgDiv.img["src"]
464
                    (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid/quantity)
465
                    dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
466
                    subOrder.dealRank = dealRank.get('rank')
467
                    subOrder.rankDesc = dealRank.get('description')
18384 amit.gupta 468
                    subOrder.maxNlc = dealRank.get('maxNlc')
469
                    subOrder.minNlc = dealRank.get('minNlc')
470
                    subOrder.db = dealRank.get('dp')
471
                    subOrder.itemStatus = dealRank.get('status')
18028 amit.gupta 472
                    cashbackStatus = Store.CB_PENDING
473
                    if cashbackAmount <= 0:
474
                        cashbackStatus = Store.CB_NA
475
                    subOrder.cashBackStatus = cashbackStatus
476
                    subOrder.cashBackAmount = cashbackAmount*quantity
477
                    if percentage > 0:
478
                        subOrder.cashBackPercentage = percentage
479
                    if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
480
                        subOrder.status = Store.ORDER_DELIVERED
481
                        subOrder.closed = True
482
                        if subOrder.cashBackStatus == Store.CB_PENDING:
483
                            subOrder.cashBackStatus = Store.CB_APPROVED
484
                    elif closedStatus:
485
                        closedStatus= False
486
                    subOrders.append(subOrder)
14698 amit.gupta 487
        merchantOrder.status='success'
15473 amit.gupta 488
        merchantOrder.closed = closedStatus
14464 amit.gupta 489
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 490
 
491
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 492
        try:
493
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
494
            if fonts[0].text == "Important Message":
495
                if fonts[1].text=="This order has been cancelled.":
496
                    merchantOrder.closed = True
497
                    merchantOrder.status = "cancelled"
498
                    merchantOrder.requireDetail = False
499
                    self._updateToOrder(todict(merchantOrder))
500
                    return
501
                else:
502
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 503
            else:
504
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 505
        except:
14549 amit.gupta 506
            orderDetails = soup.body.find(id="orderDetails")
507
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
508
                merchantOrder.closed = True
14698 amit.gupta 509
                merchantOrder.status = "cancelled"
14549 amit.gupta 510
                merchantOrder.requireDetail = False
511
                self._updateToOrder(todict(merchantOrder))
512
            else:
513
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 514
 
13868 amit.gupta 515
    def getTrackingUrls(self, userId):
14074 amit.gupta 516
 
517
        missingOrderUrls = []
518
        missingOrders = self._getMissingOrders({'userId':userId})
519
        for missingOrder in missingOrders:
520
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 521
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 522
        count = len(orders)
13879 amit.gupta 523
        print "count", count
14948 amit.gupta 524
        print "Missing Urls"
525
        print "*************"
526
        print missingOrderUrls
13868 amit.gupta 527
        if count > 0:
14684 amit.gupta 528
            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 529
        else: 
14074 amit.gupta 530
            return missingOrderUrls
13927 amit.gupta 531
 
532
    def trackOrdersForUser(self, userId, url, rawHtml):
17718 amit.gupta 533
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14985 amit.gupta 534
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 535
        if not os.path.exists(directory):
536
            os.makedirs(directory)
14173 amit.gupta 537
 
14945 amit.gupta 538
 
13995 amit.gupta 539
        try:
540
            searchMap = {'userId':userId}
541
            collectionMap = {'merchantOrderId':1}
542
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 543
            datetimeNow = datetime.now()
544
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 545
            print "url----------------", url
17465 amit.gupta 546
            cancelledSummary = False
14320 amit.gupta 547
            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 548
                if url == 'https://www.amazon.in/gp/css/order-history':
549
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
550
                else:   
17465 amit.gupta 551
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
552
                    cancelledSummary = True   
14945 amit.gupta 553
                f = open(filename,'w')
554
                f.write(rawHtml) # python will convert \n to os.linesep
555
                f.close() # you can omit in most cases as the destructor will call if
18158 amit.gupta 556
                soup = BeautifulSoup(rawHtml,'html5lib')
13995 amit.gupta 557
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
558
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
559
                for activeOrder in activeOrders:
17307 amit.gupta 560
                    matched=False
13995 amit.gupta 561
                    for orderEle in allOrders:
17254 amit.gupta 562
                        deliveredOn = None
563
                        deliveryEstimate = None
17258 amit.gupta 564
                        shippingEstimate = None
13995 amit.gupta 565
                        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'})
566
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
567
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 568
                            matched=True
13995 amit.gupta 569
                            closed = True
17465 amit.gupta 570
                            if not cancelledSummary:
571
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
572
                            else:
573
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
574
                                shipments.pop(0)
13995 amit.gupta 575
                            for shipment in shipments:
17307 amit.gupta 576
                                orderStatusDesc = None
13995 amit.gupta 577
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 578
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 579
                                try:
17270 amit.gupta 580
                                    orderStatus = sdivs[0].span.text.strip()
581
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 582
                                except:
17270 amit.gupta 583
                                    try:
584
                                        dateString = orderStatus.split("Delivered ")[1].strip()
585
                                        status = MStore.ORDER_DELIVERED
586
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
587
                                    except:
588
                                        try:
17271 amit.gupta 589
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 590
                                            status = MStore.ORDER_SHIPPED
591
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
592
                                        except:
593
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 594
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 595
                                try:
17270 amit.gupta 596
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
597
                                except:
17271 amit.gupta 598
                                    try:
599
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
600
                                    except:
17307 amit.gupta 601
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 602
 
17271 amit.gupta 603
                                if orderStatusDesc: 
17348 amit.gupta 604
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 605
                                    try:
17348 amit.gupta 606
                                        status = self._getStatusFromDetailedStatus(orderStatus)
607
                                    except:
608
                                        pass
609
                                    try:
17271 amit.gupta 610
                                        if "Dispatch estimate" in orderStatusDesc:
611
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
612
                                        elif "Delivery estimate" in orderStatus:
613
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
614
                                        elif "Arriving" in orderStatus:
615
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
616
                                    except:
617
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
618
                                        closed=False
619
                                        status = None
17258 amit.gupta 620
 
14335 amit.gupta 621
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 622
                                trackingUrl = None
623
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
624
                                    if buttonDiv.find('a').text.strip()=='Track package':
625
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
626
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 627
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 628
                                        break
629
                                for prodDiv in productDivs:
630
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
631
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
632
                                    imgUrl = prodDiv.find("img")["src"]
633
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 634
                                        if subOrder['closed']==True:
635
                                            continue
17230 amit.gupta 636
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 637
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
638
                                            updateMap = {}
639
                                            closedStatus = False
640
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 641
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 642
                                            if status:
17348 amit.gupta 643
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 644
                                                updateMap['subOrders.$.status'] = status 
645
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 646
 
17251 amit.gupta 647
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 648
                                                if deliveredOn:                               
17254 amit.gupta 649
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 650
                                                closedStatus = True
651
                                                updateMap['subOrders.$.closed'] = True
652
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 653
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 654
                                            if status==MStore.ORDER_CANCELLED:     
655
                                                closedStatus = True
656
                                                updateMap['subOrders.$.closed'] = True
657
                                                if cashbackStatus == Store.CB_PENDING:
658
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 659
                                            if status==MStore.ORDER_SHIPPED:
660
                                                if deliveryEstimate:   
661
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 662
                                                if trackingUrl is not None:
663
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 664
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 665
                                            if shippingEstimate:   
666
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 667
                                            if not closedStatus:
668
                                                closed = False
17307 amit.gupta 669
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 670
                                            updateMap['status']='success'
13995 amit.gupta 671
                                            bulk.find(findMap).update({'$set' : updateMap})
672
                                            break
17321 amit.gupta 673
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 674
                            break
675
                    if not matched:
17319 amit.gupta 676
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 677
                        for subOrder in activeOrder['subOrders']:
678
                            if subOrder['closed']==True:
679
                                continue
680
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
681
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
682
                            bulk.find({'orderId': activeOrder['orderId']})
683
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 684
                bulk.execute()
685
                return 'PARSED_SUCCESS'
686
            else:
15563 amit.gupta 687
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
688
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 689
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 690
 
691
                filename = directory + "/" + merchantOrderId   
692
                f = open(filename,'w')
693
                f.write(rawHtml) # python will convert \n to os.linesep
694
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 695
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 696
                print "result", result
15555 amit.gupta 697
                try:
15560 amit.gupta 698
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
699
                    order1.status = result
700
                    order1.rawhtml = rawHtml
17358 amit.gupta 701
                    order1.order_url = url
15555 amit.gupta 702
                    session.commit()
15561 amit.gupta 703
                except:
704
                    traceback.print_exc()
15555 amit.gupta 705
                finally:
706
                    session.close()
14080 amit.gupta 707
                return 'PARSED_SUCCESS'
708
                pass
14008 amit.gupta 709
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 710
        except:
711
            traceback.print_exc()    
712
            return 'PARSED_FAILED'
713
 
714
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 715
        if "ordered from" in detailedStatus.lower():
716
            return MStore.ORDER_PLACED 
717
 
13995 amit.gupta 718
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 719
            if detailedStatus.lower() in value:
13995 amit.gupta 720
                return key
721
 
17253 amit.gupta 722
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 723
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 724
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 725
        br = getBrowserObject()
726
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
727
        url = AMAZON_AFF_URL
728
        response = br.open(url)
729
        #get data for past 40 days and store it to mongo
730
        dt = datetime.now()
14701 amit.gupta 731
        dat = dt - timedelta(days=2)
14651 amit.gupta 732
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
733
        response = br.open(url)
734
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
735
        j=-1
736
        for row in page.split("\n"):
737
            j += 1
738
            if j== 0 or j==1:
739
                continue
740
            fields = row.split("\t")
741
            if len(fields)>1:
742
                print fields
743
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
744
                print amazonAffiliate
745
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
746
            else:
747
                break
14650 amit.gupta 748
 
14608 amit.gupta 749
 
16468 amit.gupta 750
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 751
        print trackingUrl
14624 amit.gupta 752
        subOrder = {}
14749 amit.gupta 753
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 754
        status = MStore.ORDER_SHIPPED
755
        #print page
14624 amit.gupta 756
        soup = BeautifulSoup(page)
16468 amit.gupta 757
        header1 = soup.find("h1") 
758
        if header1:
18727 amit.gupta 759
            if header1.text=="Sign In" or header1.text.strip()=="Login":
16468 amit.gupta 760
                print "Login page is displayed for order id", orderId
17307 amit.gupta 761
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
18725 amit.gupta 762
                subOrder['login'] = True
14624 amit.gupta 763
                return subOrder
16468 amit.gupta 764
        try:
16474 amit.gupta 765
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 766
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
767
            try:
768
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
769
            except:
770
                displayStatus = detailedStatus
771
            print displayStatus
772
            if detailedStatus.lower().find("delivered")>=0:
773
                print detailedStatus
774
                displayStatus = "Delivered"
775
                status = "Delivered"
776
                try:
777
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
778
                except:
779
                    pass
780
 
781
            elif detailedStatus.lower() == 'returned':
782
                status = 'Cancelled'
783
            subOrder['status'] = status
784
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 785
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 786
 
787
        except:
17350 amit.gupta 788
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 789
            print "failed to parse", orderId
790
            traceback.print_exc()
791
 
14624 amit.gupta 792
        return subOrder
793
 
794
 
795
 
16980 amit.gupta 796
    def parseInfo(self,):
797
        from pyquery import PyQuery as pq
16986 amit.gupta 798
        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 799
        try:
800
            for order in orders:
801
                try:
802
                    doc = pq(order.rawhtml)
803
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
804
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 805
                    orderInfo = All_user_addresses()
16980 amit.gupta 806
                    orderInfo.address = lists[-3].text
807
                    orderInfo.user_id = order.user_id
16986 amit.gupta 808
                    orderInfo.source = 'order' 
809
                    #orderInfo.order_id = order.id
810
                    #orderInfo.email = None
811
                    #orderInfo.name = lists[0].text 
812
                    #orderInfo.mobile = None
16980 amit.gupta 813
                    adSplit = lists[-2].text.split(",")
814
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
815
                    if match:
816
                        items = match.groups()
16986 amit.gupta 817
                    orderInfo.city = adSplit[0].strip()
818
                    orderInfo.pincode = items[1].strip()
819
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 820
                    session.commit()
821
                except:
822
                    session.rollback()
823
                    continue
824
        finally:
825
            session.close()
14608 amit.gupta 826
 
13927 amit.gupta 827
 
828
 
13774 amit.gupta 829
def main():
16874 amit.gupta 830
    store = getStore(1)
18376 amit.gupta 831
    #store.parseOrderRawHtml("444444", '123', 14, readSSh('/home/amit/80444.html'), 'https://www.amazon.in/gp/css/summary/edit.html?orderID=402-1449548-9569169')
18031 amit.gupta 832
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
18378 amit.gupta 833
    session.close()
18377 amit.gupta 834
    for o in orders:
835
        try:
836
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
837
        finally:
838
            session.close()
17631 amit.gupta 839
    #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 840
    #store.trackOrdersForUser(46195, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515', readSSh('/home/amit/amazon.html'))
17231 amit.gupta 841
 
842
def getSummaryFile(directory):
843
    date1 = datetime(2015,1,1)
844
    finalFile = None
845
    try:
846
        for file in os.listdir(directory):
847
            if file.startswith("orderSummary"):
848
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
849
                if date2 > date1:
850
                    date1 = date2
851
                    finalFile=file
852
    except:
853
        print "Missing directory"
17251 amit.gupta 854
    return finalFile 
17231 amit.gupta 855
 
15555 amit.gupta 856
 
857
def parseDetailNotCreated():
858
    try:
859
        store=getStore(1)
860
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 861
        session.close()
15555 amit.gupta 862
        for order in orders:
863
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
864
 
865
    finally:
866
        session.close()
17251 amit.gupta 867
 
868
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 869
    print dateString
17251 amit.gupta 870
    if dateString.lower()=='today':
871
        return date.today()
872
    if dateString.lower()=='yesterday':
873
        return date.today() - timedelta(days=1)
874
    try:
875
        return datetime.strptime(dateString, '%d-%b-%y')
876
    except:
877
        try:
878
            #get Closest Date from today
879
            curDate = date.today()
880
            curTime = datetime(curDate.year, curDate.month, curDate.day)
881
            curYear = curDate.year
882
            prevYear = curYear - 1
17254 amit.gupta 883
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
884
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 885
            if dateMax <= curTime:
886
                return dateMax
887
            else:
888
                return dateMin 
889
        except:
17254 amit.gupta 890
            try:
891
                days_of_week = ['sunday','monday','tuesday','wednesday',
892
                            'thursday','friday','saturday']
893
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
894
                if deltaDays <= 0:
895
                    deltaDays= deltaDays + 7
896
                curDate = curDate - timedelta(days=deltaDays)
897
                print datetime.strftime(curDate, '%d-%b-%y')
898
                return curDate
899
            except:
900
                print "could not parse"
901
                return None
15565 amit.gupta 902
 
17254 amit.gupta 903
def getDateStringArriving(dateString='Thursday'):
904
    print dateString
905
    if dateString.lower()=='today':
906
        return date.today()
907
    if dateString.lower()=='tomorrow':
908
        return date.today() + timedelta(days=1)
909
    try:
910
        return datetime.strptime(dateString, '%d-%b-%y')
911
    except:
912
        try:
913
            #get Closest Date from today
914
            curDate = date.today()
915
            curTime = datetime(curDate.year, curDate.month, curDate.day)
916
            curYear = curDate.year
917
            nextYear = curYear + 1
918
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
919
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
920
            if dateMin >= curTime:
921
                return dateMin
922
            else:
923
                return dateMax 
924
        except:
925
            try:
926
                days_of_week = ['sunday','monday','tuesday','wednesday',
927
                            'thursday','friday','saturday']
928
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
929
                if deltaDays < 0:
930
                    deltaDays= deltaDays + 7
931
                curDate = curDate + timedelta(days=deltaDays)
932
                return curDate
933
            except:
934
                print "Could not parse"
935
                return None 
936
 
18028 amit.gupta 937
def main1():
938
    store = getStore(1)
939
    for merchantOrder in store.db.merchantOrder.find({"subOrders":{"$elemMatch":{"cashBackStatus":"Not Applicable", "cashBackPercentage":{"$gt":0}}}}):
940
        mo = obj(merchantOrder)
941
        for subOrder in mo.subOrders:
942
            subOrder.closed=False 
943
        print "orderId", mo.orderId
944
        store.populateDerivedFields(mo, False)
945
        store.db.merchantOrder.update({"orderId":mo.orderId}, {"$set":todict(mo)})
946
        break
947
 
948
 
15565 amit.gupta 949
def parseOrderNotCreated():
950
    try:
951
        store=getStore(1)
952
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
953
        session.close()
954
        for order in orders:
15567 amit.gupta 955
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
956
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
957
            order1.status = result
958
            session.commit()
15565 amit.gupta 959
    finally:
960
        session.close()
15555 amit.gupta 961
 
13774 amit.gupta 962
if __name__ == '__main__':
17599 amit.gupta 963
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
964
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
18156 amit.gupta 965
#    readSSh("/AmazonTrack/User5525/171-0333104-6169933")
18159 amit.gupta 966
    main()
967
    #store=getStore(1)
968
    #store.trackOrdersForUser(5525, "https://www.amazon.in/gp/css/summary/edit.html?orderID=171-0333104-6169933", readSSh("/AmazonTrack/User5525/171-0333104-6169933"))