Subversion Repositories SmartDukaan

Rev

Rev 18440 | Rev 18727 | 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
        print "detail" if detailSection else "", "shipment" if shipmentSection else "", "summary" if summarySection else ""
312
 
313
        print pq(detailSection('li .a-row .a-span-last')[0]).text(), pq(detailSection('li .a-row .a-span-last')[1]).text(), pq(detailSection('li .a-row .a-span-last')[2]).text()
314
 
315
        i=-1    
316
        for s in shipmentSection('.a-box-group'):
317
            shipmentGroup = pq(s)
318
            for shipment in shipmentGroup('.a-box'):
319
                i += 1 
320
                if i==0:
321
                    continue
322
                shipment = pq(shipment)
323
                shipmentStatusSection = shipment('.a-section:eq(0)')
324
                productDetails = shipment('.a-section:eq(1)>.a-row')
18380 amit.gupta 325
                print shipmentStatusSection('h3').text(), shipmentStatusSection('p').text(), shipmentStatusSection('span').text()
18376 amit.gupta 326
                for productDetail in productDetails:
327
                    productDetail = pq(productDetail)
328
                    pImg = productDetail.children('div').eq(0)
329
                    pQty = productDetail.children('div').eq(1)
330
                    #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 331
                    productUrl = pImg('a').attr('href'), 
332
                    qty = pQty('.a-row:nth-child(2)')('span').text().split(':')[1].strip()
333
                    price = int(float(pQty('nobr').text().replace('Rs.','').replace(',', '')))
334
                    #subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
18376 amit.gupta 335
 
336
 
337
        raise   
338
 
339
 
340
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
341
        merchantOrder.placedOn
342
        merchantOrder.merchantOrderId
343
        merchantOrder.paidAmount
14464 amit.gupta 344
 
345
    def parseNewStlye(self, merchantOrder, soup):
346
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
347
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 348
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
349
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 350
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
351
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
352
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 353
        try:
354
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
355
        except:
356
            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 357
        priceRows = priceBox.findAll('div', {'class':'a-row'})
358
        subTotal = 0
14566 amit.gupta 359
        shippingPrice = 0   
14464 amit.gupta 360
        promoApplied = 0
361
        for priceRow in priceRows:
362
            if "Item(s) Subtotal:" in str(priceRow):
363
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
364
            elif "Shipping:" in str(priceRow):
365
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
366
            elif "Grand Total:" in str(priceRow):
367
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
368
                merchantOrder.paidAmount = grandPrice
369
            elif "Total:" in str(priceRow):
370
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
371
            elif "Promotion Applied:" in str(priceRow):
372
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
373
        totalPaid = subTotal        
374
        if promoApplied > 0:
375
            totalPaid -= promoApplied
376
            if shippingPrice <= promoApplied:
377
                totalPaid  += shippingPrice
15473 amit.gupta 378
 
17601 amit.gupta 379
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 380
        subOrders = []
381
        merchantOrder.subOrders = subOrders
15473 amit.gupta 382
        closedStatus = True
17601 amit.gupta 383
        subOrders = []
18161 amit.gupta 384
        j=0
14464 amit.gupta 385
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 386
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 387
            try:
388
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
389
            except:
390
                trackingUrl= None
391
 
17347 amit.gupta 392
            deliverySpanTop = None
14464 amit.gupta 393
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
394
            statusDiv = innerBoxes[0]
395
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 396
            try:
397
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
398
            except:
17617 amit.gupta 399
                pass
18030 amit.gupta 400
            #if not deliverySpanTop:
401
            productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 402
            merchantOrder.subOrders = subOrders
403
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 404
                deliverySpan = deliverySpanTop
405
                if not deliverySpanTop:
406
                    if i%2==1:
407
                        continue
18028 amit.gupta 408
                    deliverySpan = productDiv.div.div.span
18156 amit.gupta 409
                    try:
410
                        productDiv = productDivs[i+1]
411
                    except:
412
                        pass
18028 amit.gupta 413
                for  prodRow in productDiv.findAll('div', recursive=False):
414
                    j += 1
415
                    imgDiv  = prodRow.div.div
416
                    detailDiv = imgDiv.find_next_sibling('div')
417
                    detailDivs = detailDiv.findAll('div', recursive=False)
418
                    arr = re.split("^(\d+) of", detailDivs[0].a.text.strip())
419
                    (productTitle, quantity) = (arr[-1], (1 if len(arr)<2 else int(arr[1])) )
14756 amit.gupta 420
                    try:
18028 amit.gupta 421
                        unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
14756 amit.gupta 422
                    except:
18028 amit.gupta 423
                        unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
424
                    amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
425
                    productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
426
                    subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
427
                    subOrder.productCode = productUrl.split('/')[5]
428
                    subOrder.unitPrice = unitPrice
429
                    subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
430
                    estDlvyTime = datetime.now()
431
                    if deliverySpan is not None:
432
                        try:
433
                            subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
434
                            estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
435
                            subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
436
                            estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
437
                        except:
438
                            if "Delivered on" in deliverySpan.text:
439
                                subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
440
                            subOrder.estimatedDeliveryDate = "Not available"
441
 
442
                    createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
443
                    subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
444
                    subOrder.detailedStatus = subOrderStatus
445
                    status=MStore.ORDER_PLACED
17599 amit.gupta 446
                    try:
18028 amit.gupta 447
                        status = self._getStatusFromDetailedStatus(subOrderStatus)
17599 amit.gupta 448
                    except:
449
                        try:
18028 amit.gupta 450
                            dateString = subOrderStatus.split("Delivered ")[1].strip()
451
                            subOrder.status = MStore.ORDER_DELIVERED
452
                            subOrder.detailedStatus = 'Delivered'
453
                            dateString = getDateStringDelivered(dateString)
17631 amit.gupta 454
                            if dateString is not None:
18028 amit.gupta 455
                                subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 456
                        except:
18028 amit.gupta 457
                            try:
458
                                dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
459
                                subOrder.status = MStore.ORDER_SHIPPED
460
                                dateString = getDateStringArriving(dateString)
461
                                if dateString is not None:
462
                                    subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
463
                                else:
464
                                    subOrder.deliveryEstimate = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
465
                            except:
466
                                print "Unknown status Alert -", status
467
 
468
                    subOrder.deliveryCharges = shippingPrice
469
                    if trackingUrl:
18156 amit.gupta 470
                        subOrder.trackingUrl  = trackingUrl
18028 amit.gupta 471
                    subOrder.imgUrl = imgDiv.img["src"]
472
                    (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid/quantity)
473
                    dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
474
                    subOrder.dealRank = dealRank.get('rank')
475
                    subOrder.rankDesc = dealRank.get('description')
18384 amit.gupta 476
                    subOrder.maxNlc = dealRank.get('maxNlc')
477
                    subOrder.minNlc = dealRank.get('minNlc')
478
                    subOrder.db = dealRank.get('dp')
479
                    subOrder.itemStatus = dealRank.get('status')
18028 amit.gupta 480
                    cashbackStatus = Store.CB_PENDING
481
                    if cashbackAmount <= 0:
482
                        cashbackStatus = Store.CB_NA
483
                    subOrder.cashBackStatus = cashbackStatus
484
                    subOrder.cashBackAmount = cashbackAmount*quantity
485
                    if percentage > 0:
486
                        subOrder.cashBackPercentage = percentage
487
                    if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
488
                        subOrder.status = Store.ORDER_DELIVERED
489
                        subOrder.closed = True
490
                        if subOrder.cashBackStatus == Store.CB_PENDING:
491
                            subOrder.cashBackStatus = Store.CB_APPROVED
492
                    elif closedStatus:
493
                        closedStatus= False
494
                    subOrders.append(subOrder)
14698 amit.gupta 495
        merchantOrder.status='success'
15473 amit.gupta 496
        merchantOrder.closed = closedStatus
14464 amit.gupta 497
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 498
 
499
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 500
        try:
501
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
502
            if fonts[0].text == "Important Message":
503
                if fonts[1].text=="This order has been cancelled.":
504
                    merchantOrder.closed = True
505
                    merchantOrder.status = "cancelled"
506
                    merchantOrder.requireDetail = False
507
                    self._updateToOrder(todict(merchantOrder))
508
                    return
509
                else:
510
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 511
            else:
512
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 513
        except:
14549 amit.gupta 514
            orderDetails = soup.body.find(id="orderDetails")
515
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
516
                merchantOrder.closed = True
14698 amit.gupta 517
                merchantOrder.status = "cancelled"
14549 amit.gupta 518
                merchantOrder.requireDetail = False
519
                self._updateToOrder(todict(merchantOrder))
520
            else:
521
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 522
 
13868 amit.gupta 523
    def getTrackingUrls(self, userId):
14074 amit.gupta 524
 
525
        missingOrderUrls = []
526
        missingOrders = self._getMissingOrders({'userId':userId})
527
        for missingOrder in missingOrders:
528
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 529
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 530
        count = len(orders)
13879 amit.gupta 531
        print "count", count
14948 amit.gupta 532
        print "Missing Urls"
533
        print "*************"
534
        print missingOrderUrls
13868 amit.gupta 535
        if count > 0:
14684 amit.gupta 536
            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 537
        else: 
14074 amit.gupta 538
            return missingOrderUrls
13927 amit.gupta 539
 
540
    def trackOrdersForUser(self, userId, url, rawHtml):
17718 amit.gupta 541
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14985 amit.gupta 542
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 543
        if not os.path.exists(directory):
544
            os.makedirs(directory)
14173 amit.gupta 545
 
14945 amit.gupta 546
 
13995 amit.gupta 547
        try:
548
            searchMap = {'userId':userId}
549
            collectionMap = {'merchantOrderId':1}
550
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 551
            datetimeNow = datetime.now()
552
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 553
            print "url----------------", url
17465 amit.gupta 554
            cancelledSummary = False
14320 amit.gupta 555
            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 556
                if url == 'https://www.amazon.in/gp/css/order-history':
557
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
558
                else:   
17465 amit.gupta 559
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
560
                    cancelledSummary = True   
14945 amit.gupta 561
                f = open(filename,'w')
562
                f.write(rawHtml) # python will convert \n to os.linesep
563
                f.close() # you can omit in most cases as the destructor will call if
18158 amit.gupta 564
                soup = BeautifulSoup(rawHtml,'html5lib')
13995 amit.gupta 565
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
566
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
567
                for activeOrder in activeOrders:
17307 amit.gupta 568
                    matched=False
13995 amit.gupta 569
                    for orderEle in allOrders:
17254 amit.gupta 570
                        deliveredOn = None
571
                        deliveryEstimate = None
17258 amit.gupta 572
                        shippingEstimate = None
13995 amit.gupta 573
                        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'})
574
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
575
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 576
                            matched=True
13995 amit.gupta 577
                            closed = True
17465 amit.gupta 578
                            if not cancelledSummary:
579
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
580
                            else:
581
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
582
                                shipments.pop(0)
13995 amit.gupta 583
                            for shipment in shipments:
17307 amit.gupta 584
                                orderStatusDesc = None
13995 amit.gupta 585
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 586
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 587
                                try:
17270 amit.gupta 588
                                    orderStatus = sdivs[0].span.text.strip()
589
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 590
                                except:
17270 amit.gupta 591
                                    try:
592
                                        dateString = orderStatus.split("Delivered ")[1].strip()
593
                                        status = MStore.ORDER_DELIVERED
594
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
595
                                    except:
596
                                        try:
17271 amit.gupta 597
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 598
                                            status = MStore.ORDER_SHIPPED
599
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
600
                                        except:
601
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 602
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 603
                                try:
17270 amit.gupta 604
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
605
                                except:
17271 amit.gupta 606
                                    try:
607
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
608
                                    except:
17307 amit.gupta 609
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 610
 
17271 amit.gupta 611
                                if orderStatusDesc: 
17348 amit.gupta 612
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 613
                                    try:
17348 amit.gupta 614
                                        status = self._getStatusFromDetailedStatus(orderStatus)
615
                                    except:
616
                                        pass
617
                                    try:
17271 amit.gupta 618
                                        if "Dispatch estimate" in orderStatusDesc:
619
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
620
                                        elif "Delivery estimate" in orderStatus:
621
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
622
                                        elif "Arriving" in orderStatus:
623
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
624
                                    except:
625
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
626
                                        closed=False
627
                                        status = None
17258 amit.gupta 628
 
14335 amit.gupta 629
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 630
                                trackingUrl = None
631
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
632
                                    if buttonDiv.find('a').text.strip()=='Track package':
633
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
634
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 635
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 636
                                        break
637
                                for prodDiv in productDivs:
638
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
639
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
640
                                    imgUrl = prodDiv.find("img")["src"]
641
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 642
                                        if subOrder['closed']==True:
643
                                            continue
17230 amit.gupta 644
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 645
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
646
                                            updateMap = {}
647
                                            closedStatus = False
648
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 649
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 650
                                            if status:
17348 amit.gupta 651
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 652
                                                updateMap['subOrders.$.status'] = status 
653
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 654
 
17251 amit.gupta 655
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 656
                                                if deliveredOn:                               
17254 amit.gupta 657
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 658
                                                closedStatus = True
659
                                                updateMap['subOrders.$.closed'] = True
660
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 661
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 662
                                            if status==MStore.ORDER_CANCELLED:     
663
                                                closedStatus = True
664
                                                updateMap['subOrders.$.closed'] = True
665
                                                if cashbackStatus == Store.CB_PENDING:
666
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 667
                                            if status==MStore.ORDER_SHIPPED:
668
                                                if deliveryEstimate:   
669
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 670
                                                if trackingUrl is not None:
671
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 672
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 673
                                            if shippingEstimate:   
674
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 675
                                            if not closedStatus:
676
                                                closed = False
17307 amit.gupta 677
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 678
                                            updateMap['status']='success'
13995 amit.gupta 679
                                            bulk.find(findMap).update({'$set' : updateMap})
680
                                            break
17321 amit.gupta 681
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 682
                            break
683
                    if not matched:
17319 amit.gupta 684
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 685
                        for subOrder in activeOrder['subOrders']:
686
                            if subOrder['closed']==True:
687
                                continue
688
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
689
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
690
                            bulk.find({'orderId': activeOrder['orderId']})
691
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 692
                bulk.execute()
693
                return 'PARSED_SUCCESS'
694
            else:
15563 amit.gupta 695
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
696
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 697
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 698
 
699
                filename = directory + "/" + merchantOrderId   
700
                f = open(filename,'w')
701
                f.write(rawHtml) # python will convert \n to os.linesep
702
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 703
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 704
                print "result", result
15555 amit.gupta 705
                try:
15560 amit.gupta 706
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
707
                    order1.status = result
708
                    order1.rawhtml = rawHtml
17358 amit.gupta 709
                    order1.order_url = url
15555 amit.gupta 710
                    session.commit()
15561 amit.gupta 711
                except:
712
                    traceback.print_exc()
15555 amit.gupta 713
                finally:
714
                    session.close()
14080 amit.gupta 715
                return 'PARSED_SUCCESS'
716
                pass
14008 amit.gupta 717
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 718
        except:
719
            traceback.print_exc()    
720
            return 'PARSED_FAILED'
721
 
722
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 723
        if "ordered from" in detailedStatus.lower():
724
            return MStore.ORDER_PLACED 
725
 
13995 amit.gupta 726
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 727
            if detailedStatus.lower() in value:
13995 amit.gupta 728
                return key
729
 
17253 amit.gupta 730
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 731
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 732
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 733
        br = getBrowserObject()
734
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
735
        url = AMAZON_AFF_URL
736
        response = br.open(url)
737
        #get data for past 40 days and store it to mongo
738
        dt = datetime.now()
14701 amit.gupta 739
        dat = dt - timedelta(days=2)
14651 amit.gupta 740
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
741
        response = br.open(url)
742
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
743
        j=-1
744
        for row in page.split("\n"):
745
            j += 1
746
            if j== 0 or j==1:
747
                continue
748
            fields = row.split("\t")
749
            if len(fields)>1:
750
                print fields
751
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
752
                print amazonAffiliate
753
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
754
            else:
755
                break
14650 amit.gupta 756
 
14608 amit.gupta 757
 
16468 amit.gupta 758
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 759
        print trackingUrl
14624 amit.gupta 760
        subOrder = {}
14749 amit.gupta 761
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 762
        status = MStore.ORDER_SHIPPED
763
        #print page
14624 amit.gupta 764
        soup = BeautifulSoup(page)
16468 amit.gupta 765
        header1 = soup.find("h1") 
766
        if header1:
767
            if header1.text=="Sign In":
768
                print "Login page is displayed for order id", orderId
17307 amit.gupta 769
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
18725 amit.gupta 770
                subOrder['login'] = True
14624 amit.gupta 771
                return subOrder
16468 amit.gupta 772
        try:
16474 amit.gupta 773
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 774
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
775
            try:
776
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
777
            except:
778
                displayStatus = detailedStatus
779
            print displayStatus
780
            if detailedStatus.lower().find("delivered")>=0:
781
                print detailedStatus
782
                displayStatus = "Delivered"
783
                status = "Delivered"
784
                try:
785
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
786
                except:
787
                    pass
788
 
789
            elif detailedStatus.lower() == 'returned':
790
                status = 'Cancelled'
791
            subOrder['status'] = status
792
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 793
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 794
 
795
        except:
17350 amit.gupta 796
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 797
            print "failed to parse", orderId
798
            traceback.print_exc()
799
 
14624 amit.gupta 800
        return subOrder
801
 
802
 
803
 
16980 amit.gupta 804
    def parseInfo(self,):
805
        from pyquery import PyQuery as pq
16986 amit.gupta 806
        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 807
        try:
808
            for order in orders:
809
                try:
810
                    doc = pq(order.rawhtml)
811
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
812
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 813
                    orderInfo = All_user_addresses()
16980 amit.gupta 814
                    orderInfo.address = lists[-3].text
815
                    orderInfo.user_id = order.user_id
16986 amit.gupta 816
                    orderInfo.source = 'order' 
817
                    #orderInfo.order_id = order.id
818
                    #orderInfo.email = None
819
                    #orderInfo.name = lists[0].text 
820
                    #orderInfo.mobile = None
16980 amit.gupta 821
                    adSplit = lists[-2].text.split(",")
822
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
823
                    if match:
824
                        items = match.groups()
16986 amit.gupta 825
                    orderInfo.city = adSplit[0].strip()
826
                    orderInfo.pincode = items[1].strip()
827
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 828
                    session.commit()
829
                except:
830
                    session.rollback()
831
                    continue
832
        finally:
833
            session.close()
14608 amit.gupta 834
 
13927 amit.gupta 835
 
836
 
13774 amit.gupta 837
def main():
16874 amit.gupta 838
    store = getStore(1)
18376 amit.gupta 839
    #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 840
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
18378 amit.gupta 841
    session.close()
18377 amit.gupta 842
    for o in orders:
843
        try:
844
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
845
        finally:
846
            session.close()
17631 amit.gupta 847
    #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 848
    #store.trackOrdersForUser(46195, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515', readSSh('/home/amit/amazon.html'))
17231 amit.gupta 849
 
850
def getSummaryFile(directory):
851
    date1 = datetime(2015,1,1)
852
    finalFile = None
853
    try:
854
        for file in os.listdir(directory):
855
            if file.startswith("orderSummary"):
856
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
857
                if date2 > date1:
858
                    date1 = date2
859
                    finalFile=file
860
    except:
861
        print "Missing directory"
17251 amit.gupta 862
    return finalFile 
17231 amit.gupta 863
 
15555 amit.gupta 864
 
865
def parseDetailNotCreated():
866
    try:
867
        store=getStore(1)
868
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 869
        session.close()
15555 amit.gupta 870
        for order in orders:
871
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
872
 
873
    finally:
874
        session.close()
17251 amit.gupta 875
 
876
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 877
    print dateString
17251 amit.gupta 878
    if dateString.lower()=='today':
879
        return date.today()
880
    if dateString.lower()=='yesterday':
881
        return date.today() - timedelta(days=1)
882
    try:
883
        return datetime.strptime(dateString, '%d-%b-%y')
884
    except:
885
        try:
886
            #get Closest Date from today
887
            curDate = date.today()
888
            curTime = datetime(curDate.year, curDate.month, curDate.day)
889
            curYear = curDate.year
890
            prevYear = curYear - 1
17254 amit.gupta 891
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
892
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 893
            if dateMax <= curTime:
894
                return dateMax
895
            else:
896
                return dateMin 
897
        except:
17254 amit.gupta 898
            try:
899
                days_of_week = ['sunday','monday','tuesday','wednesday',
900
                            'thursday','friday','saturday']
901
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
902
                if deltaDays <= 0:
903
                    deltaDays= deltaDays + 7
904
                curDate = curDate - timedelta(days=deltaDays)
905
                print datetime.strftime(curDate, '%d-%b-%y')
906
                return curDate
907
            except:
908
                print "could not parse"
909
                return None
15565 amit.gupta 910
 
17254 amit.gupta 911
def getDateStringArriving(dateString='Thursday'):
912
    print dateString
913
    if dateString.lower()=='today':
914
        return date.today()
915
    if dateString.lower()=='tomorrow':
916
        return date.today() + timedelta(days=1)
917
    try:
918
        return datetime.strptime(dateString, '%d-%b-%y')
919
    except:
920
        try:
921
            #get Closest Date from today
922
            curDate = date.today()
923
            curTime = datetime(curDate.year, curDate.month, curDate.day)
924
            curYear = curDate.year
925
            nextYear = curYear + 1
926
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
927
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
928
            if dateMin >= curTime:
929
                return dateMin
930
            else:
931
                return dateMax 
932
        except:
933
            try:
934
                days_of_week = ['sunday','monday','tuesday','wednesday',
935
                            'thursday','friday','saturday']
936
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
937
                if deltaDays < 0:
938
                    deltaDays= deltaDays + 7
939
                curDate = curDate + timedelta(days=deltaDays)
940
                return curDate
941
            except:
942
                print "Could not parse"
943
                return None 
944
 
18028 amit.gupta 945
def main1():
946
    store = getStore(1)
947
    for merchantOrder in store.db.merchantOrder.find({"subOrders":{"$elemMatch":{"cashBackStatus":"Not Applicable", "cashBackPercentage":{"$gt":0}}}}):
948
        mo = obj(merchantOrder)
949
        for subOrder in mo.subOrders:
950
            subOrder.closed=False 
951
        print "orderId", mo.orderId
952
        store.populateDerivedFields(mo, False)
953
        store.db.merchantOrder.update({"orderId":mo.orderId}, {"$set":todict(mo)})
954
        break
955
 
956
 
15565 amit.gupta 957
def parseOrderNotCreated():
958
    try:
959
        store=getStore(1)
960
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
961
        session.close()
962
        for order in orders:
15567 amit.gupta 963
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
964
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
965
            order1.status = result
966
            session.commit()
15565 amit.gupta 967
    finally:
968
        session.close()
15555 amit.gupta 969
 
13774 amit.gupta 970
if __name__ == '__main__':
17599 amit.gupta 971
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
972
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
18156 amit.gupta 973
#    readSSh("/AmazonTrack/User5525/171-0333104-6169933")
18159 amit.gupta 974
    main()
975
    #store=getStore(1)
976
    #store.trackOrdersForUser(5525, "https://www.amazon.in/gp/css/summary/edit.html?orderID=171-0333104-6169933", readSSh("/AmazonTrack/User5525/171-0333104-6169933"))