Subversion Repositories SmartDukaan

Rev

Rev 18384 | Rev 18725 | 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):
14624 amit.gupta 166
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
167
        for merchantOrder in orders:
14981 amit.gupta 168
            executeBulk = False
14624 amit.gupta 169
            try:
170
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
171
                closed = True
172
                map1 = {}
173
                for subOrder in merchantOrder.get("subOrders"):
174
                    if subOrder.get("closed"):
175
                        continue
176
                    elif subOrder.get("trackingUrl") is None:
177
                        closed = False
178
                        continue
179
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
180
                    trackingUrl = subOrder.get("trackingUrl")
181
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 182
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 183
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 184
                    if newOrder:
185
                        executeBulk = True
186
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
187
                        print findMap, "\n", updateMap
188
                        bulk.find(findMap).update({'$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')
331
                    print pImg('a').attr('href'), pQty('.a-row:nth-child(2)')('span').text().split(':')[1].strip()
332
                    print int(float(pQty('nobr').text().replace('Rs.','').replace(',', '')))
333
 
334
 
335
        raise   
336
 
337
 
338
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
339
        merchantOrder.placedOn
340
        merchantOrder.merchantOrderId
341
        merchantOrder.paidAmount
14464 amit.gupta 342
 
343
    def parseNewStlye(self, merchantOrder, soup):
344
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
345
        orderDetailsContainer = soup.body.find(id="orderDetails")
14566 amit.gupta 346
        divAfterH1 = orderDetailsContainer.h1.next_sibling.next_sibling
347
        orderLeftDiv = divAfterH1.div
14464 amit.gupta 348
        placedOnSpan = orderLeftDiv.find("span", {'class':'order-date-invoice-item'})
349
        merchantOrder.placedOn =placedOnSpan.text.split('Ordered on')[1].strip()
350
        merchantOrder.merchantOrderId = placedOnSpan.next_sibling.next_sibling.text.split('Order#')[1].strip()
15555 amit.gupta 351
        try:
352
            priceBox = divAfterH1.next_sibling.next_sibling.next_sibling.next_sibling.find("div", {"class":"a-box-inner"}).div.div.findAll('div', recursive=False)[-1]
353
        except:
354
            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 355
        priceRows = priceBox.findAll('div', {'class':'a-row'})
356
        subTotal = 0
14566 amit.gupta 357
        shippingPrice = 0   
14464 amit.gupta 358
        promoApplied = 0
359
        for priceRow in priceRows:
360
            if "Item(s) Subtotal:" in str(priceRow):
361
                subTotal = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
362
            elif "Shipping:" in str(priceRow):
363
                shippingPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
364
            elif "Grand Total:" in str(priceRow):
365
                grandPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
366
                merchantOrder.paidAmount = grandPrice
367
            elif "Total:" in str(priceRow):
368
                totalPrice  = int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
369
            elif "Promotion Applied:" in str(priceRow):
370
                promoApplied  += int(float(priceRow.div.next_sibling.next_sibling.span.span.text.replace('Rs.','').replace(',', '')))
371
        totalPaid = subTotal        
372
        if promoApplied > 0:
373
            totalPaid -= promoApplied
374
            if shippingPrice <= promoApplied:
375
                totalPaid  += shippingPrice
15473 amit.gupta 376
 
17601 amit.gupta 377
        shipmentDivs = orderDetailsContainer.findAll('div', class_='shipment')
14464 amit.gupta 378
        subOrders = []
379
        merchantOrder.subOrders = subOrders
15473 amit.gupta 380
        closedStatus = True
17601 amit.gupta 381
        subOrders = []
18161 amit.gupta 382
        j=0
14464 amit.gupta 383
        for shipmentDiv in shipmentDivs:
17601 amit.gupta 384
            shipmentDiv = shipmentDiv.div
17361 amit.gupta 385
            try:
386
                trackingUrl = 'http://www.amazon.in/' + shipmentDiv.find('span', class_='track-package-button').span.a.get('href')
387
            except:
388
                trackingUrl= None
389
 
17347 amit.gupta 390
            deliverySpanTop = None
14464 amit.gupta 391
            innerBoxes = shipmentDiv.findAll('div', recursive = False)
392
            statusDiv = innerBoxes[0]
393
            subOrderStatus = statusDiv.div.span.text.strip()
17347 amit.gupta 394
            try:
395
                deliverySpanTop = statusDiv.div.div.find_next_sibling('div').span
396
            except:
17617 amit.gupta 397
                pass
18030 amit.gupta 398
            #if not deliverySpanTop:
399
            productDivs = innerBoxes[-1].div.div.findAll('div', recursive=False)
14464 amit.gupta 400
            merchantOrder.subOrders = subOrders
401
            for i, productDiv in enumerate(productDivs):
17347 amit.gupta 402
                deliverySpan = deliverySpanTop
403
                if not deliverySpanTop:
404
                    if i%2==1:
405
                        continue
18028 amit.gupta 406
                    deliverySpan = productDiv.div.div.span
18156 amit.gupta 407
                    try:
408
                        productDiv = productDivs[i+1]
409
                    except:
410
                        pass
18028 amit.gupta 411
                for  prodRow in productDiv.findAll('div', recursive=False):
412
                    j += 1
413
                    imgDiv  = prodRow.div.div
414
                    detailDiv = imgDiv.find_next_sibling('div')
415
                    detailDivs = detailDiv.findAll('div', recursive=False)
416
                    arr = re.split("^(\d+) of", detailDivs[0].a.text.strip())
417
                    (productTitle, quantity) = (arr[-1], (1 if len(arr)<2 else int(arr[1])) )
14756 amit.gupta 418
                    try:
18028 amit.gupta 419
                        unitPrice = int(float(detailDivs[2].span.text.replace('Rs. ','').replace(',','')))
14756 amit.gupta 420
                    except:
18028 amit.gupta 421
                        unitPrice = int(float(detailDivs[3].span.text.replace('Rs. ','').replace(',','')))
422
                    amountPaid = int((unitPrice*quantity*totalPaid)/subTotal)
423
                    productUrl = "http://www.amazon.in" + detailDivs[0].a.get('href')
424
                    subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, amountPaid, MStore.ORDER_PLACED, quantity)
425
                    subOrder.productCode = productUrl.split('/')[5]
426
                    subOrder.unitPrice = unitPrice
427
                    subOrder.merchantSubOrderId = str(j) + " of " + merchantOrder.merchantOrderId
428
                    estDlvyTime = datetime.now()
429
                    if deliverySpan is not None:
430
                        try:
431
                            subOrder.estimatedDeliveryDate = deliverySpan.span.text.strip()
432
                            estDate = subOrder.estimatedDeliveryDate.split("-")[0].strip()
433
                            subOrder.estimatedDeliveryInt = int(time.mktime((datetime.strptime(estDate, "%A %d %B %Y")).timetuple()))
434
                            estDlvyTime = datetime.strptime(estDate, "%A %d %B %Y")
435
                        except:
436
                            if "Delivered on" in deliverySpan.text:
437
                                subOrder.deliveredOn = deliverySpan.text.split(":")[1].strip() 
438
                            subOrder.estimatedDeliveryDate = "Not available"
439
 
440
                    createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
441
                    subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
442
                    subOrder.detailedStatus = subOrderStatus
443
                    status=MStore.ORDER_PLACED
17599 amit.gupta 444
                    try:
18028 amit.gupta 445
                        status = self._getStatusFromDetailedStatus(subOrderStatus)
17599 amit.gupta 446
                    except:
447
                        try:
18028 amit.gupta 448
                            dateString = subOrderStatus.split("Delivered ")[1].strip()
449
                            subOrder.status = MStore.ORDER_DELIVERED
450
                            subOrder.detailedStatus = 'Delivered'
451
                            dateString = getDateStringDelivered(dateString)
17631 amit.gupta 452
                            if dateString is not None:
18028 amit.gupta 453
                                subOrder.deliveredOn = datetime.strftime(dateString, '%d-%b-%y')
17599 amit.gupta 454
                        except:
18028 amit.gupta 455
                            try:
456
                                dateString = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
457
                                subOrder.status = MStore.ORDER_SHIPPED
458
                                dateString = getDateStringArriving(dateString)
459
                                if dateString is not None:
460
                                    subOrder.deliveryEstimate = datetime.strftime(dateString, '%d-%b-%y')
461
                                else:
462
                                    subOrder.deliveryEstimate = subOrderStatus.split("Arriving ")[1].split("by")[0].strip()
463
                            except:
464
                                print "Unknown status Alert -", status
465
 
466
                    subOrder.deliveryCharges = shippingPrice
467
                    if trackingUrl:
18156 amit.gupta 468
                        subOrder.trackingUrl  = trackingUrl
18028 amit.gupta 469
                    subOrder.imgUrl = imgDiv.img["src"]
470
                    (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amountPaid/quantity)
471
                    dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
472
                    subOrder.dealRank = dealRank.get('rank')
473
                    subOrder.rankDesc = dealRank.get('description')
18384 amit.gupta 474
                    subOrder.maxNlc = dealRank.get('maxNlc')
475
                    subOrder.minNlc = dealRank.get('minNlc')
476
                    subOrder.db = dealRank.get('dp')
477
                    subOrder.itemStatus = dealRank.get('status')
18028 amit.gupta 478
                    cashbackStatus = Store.CB_PENDING
479
                    if cashbackAmount <= 0:
480
                        cashbackStatus = Store.CB_NA
481
                    subOrder.cashBackStatus = cashbackStatus
482
                    subOrder.cashBackAmount = cashbackAmount*quantity
483
                    if percentage > 0:
484
                        subOrder.cashBackPercentage = percentage
485
                    if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
486
                        subOrder.status = Store.ORDER_DELIVERED
487
                        subOrder.closed = True
488
                        if subOrder.cashBackStatus == Store.CB_PENDING:
489
                            subOrder.cashBackStatus = Store.CB_APPROVED
490
                    elif closedStatus:
491
                        closedStatus= False
492
                    subOrders.append(subOrder)
14698 amit.gupta 493
        merchantOrder.status='success'
15473 amit.gupta 494
        merchantOrder.closed = closedStatus
14464 amit.gupta 495
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 496
 
497
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 498
        try:
499
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
500
            if fonts[0].text == "Important Message":
501
                if fonts[1].text=="This order has been cancelled.":
502
                    merchantOrder.closed = True
503
                    merchantOrder.status = "cancelled"
504
                    merchantOrder.requireDetail = False
505
                    self._updateToOrder(todict(merchantOrder))
506
                    return
507
                else:
508
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 509
            else:
510
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 511
        except:
14549 amit.gupta 512
            orderDetails = soup.body.find(id="orderDetails")
513
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
514
                merchantOrder.closed = True
14698 amit.gupta 515
                merchantOrder.status = "cancelled"
14549 amit.gupta 516
                merchantOrder.requireDetail = False
517
                self._updateToOrder(todict(merchantOrder))
518
            else:
519
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 520
 
13868 amit.gupta 521
    def getTrackingUrls(self, userId):
14074 amit.gupta 522
 
523
        missingOrderUrls = []
524
        missingOrders = self._getMissingOrders({'userId':userId})
525
        for missingOrder in missingOrders:
526
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 527
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 528
        count = len(orders)
13879 amit.gupta 529
        print "count", count
14948 amit.gupta 530
        print "Missing Urls"
531
        print "*************"
532
        print missingOrderUrls
13868 amit.gupta 533
        if count > 0:
14684 amit.gupta 534
            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 535
        else: 
14074 amit.gupta 536
            return missingOrderUrls
13927 amit.gupta 537
 
538
    def trackOrdersForUser(self, userId, url, rawHtml):
17718 amit.gupta 539
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14985 amit.gupta 540
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 541
        if not os.path.exists(directory):
542
            os.makedirs(directory)
14173 amit.gupta 543
 
14945 amit.gupta 544
 
13995 amit.gupta 545
        try:
546
            searchMap = {'userId':userId}
547
            collectionMap = {'merchantOrderId':1}
548
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 549
            datetimeNow = datetime.now()
550
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 551
            print "url----------------", url
17465 amit.gupta 552
            cancelledSummary = False
14320 amit.gupta 553
            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 554
                if url == 'https://www.amazon.in/gp/css/order-history':
555
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
556
                else:   
17465 amit.gupta 557
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
558
                    cancelledSummary = True   
14945 amit.gupta 559
                f = open(filename,'w')
560
                f.write(rawHtml) # python will convert \n to os.linesep
561
                f.close() # you can omit in most cases as the destructor will call if
18158 amit.gupta 562
                soup = BeautifulSoup(rawHtml,'html5lib')
13995 amit.gupta 563
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
564
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
565
                for activeOrder in activeOrders:
17307 amit.gupta 566
                    matched=False
13995 amit.gupta 567
                    for orderEle in allOrders:
17254 amit.gupta 568
                        deliveredOn = None
569
                        deliveryEstimate = None
17258 amit.gupta 570
                        shippingEstimate = None
13995 amit.gupta 571
                        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'})
572
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
573
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 574
                            matched=True
13995 amit.gupta 575
                            closed = True
17465 amit.gupta 576
                            if not cancelledSummary:
577
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
578
                            else:
579
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
580
                                shipments.pop(0)
13995 amit.gupta 581
                            for shipment in shipments:
17307 amit.gupta 582
                                orderStatusDesc = None
13995 amit.gupta 583
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 584
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 585
                                try:
17270 amit.gupta 586
                                    orderStatus = sdivs[0].span.text.strip()
587
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 588
                                except:
17270 amit.gupta 589
                                    try:
590
                                        dateString = orderStatus.split("Delivered ")[1].strip()
591
                                        status = MStore.ORDER_DELIVERED
592
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
593
                                    except:
594
                                        try:
17271 amit.gupta 595
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 596
                                            status = MStore.ORDER_SHIPPED
597
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
598
                                        except:
599
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 600
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 601
                                try:
17270 amit.gupta 602
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
603
                                except:
17271 amit.gupta 604
                                    try:
605
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
606
                                    except:
17307 amit.gupta 607
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 608
 
17271 amit.gupta 609
                                if orderStatusDesc: 
17348 amit.gupta 610
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 611
                                    try:
17348 amit.gupta 612
                                        status = self._getStatusFromDetailedStatus(orderStatus)
613
                                    except:
614
                                        pass
615
                                    try:
17271 amit.gupta 616
                                        if "Dispatch estimate" in orderStatusDesc:
617
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
618
                                        elif "Delivery estimate" in orderStatus:
619
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
620
                                        elif "Arriving" in orderStatus:
621
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
622
                                    except:
623
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
624
                                        closed=False
625
                                        status = None
17258 amit.gupta 626
 
14335 amit.gupta 627
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 628
                                trackingUrl = None
629
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
630
                                    if buttonDiv.find('a').text.strip()=='Track package':
631
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
632
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 633
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 634
                                        break
635
                                for prodDiv in productDivs:
636
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
637
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
638
                                    imgUrl = prodDiv.find("img")["src"]
639
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 640
                                        if subOrder['closed']==True:
641
                                            continue
17230 amit.gupta 642
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 643
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
644
                                            updateMap = {}
645
                                            closedStatus = False
646
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 647
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 648
                                            if status:
17348 amit.gupta 649
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 650
                                                updateMap['subOrders.$.status'] = status 
651
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 652
 
17251 amit.gupta 653
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 654
                                                if deliveredOn:                               
17254 amit.gupta 655
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 656
                                                closedStatus = True
657
                                                updateMap['subOrders.$.closed'] = True
658
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 659
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 660
                                            if status==MStore.ORDER_CANCELLED:     
661
                                                closedStatus = True
662
                                                updateMap['subOrders.$.closed'] = True
663
                                                if cashbackStatus == Store.CB_PENDING:
664
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 665
                                            if status==MStore.ORDER_SHIPPED:
666
                                                if deliveryEstimate:   
667
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 668
                                                if trackingUrl is not None:
669
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 670
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 671
                                            if shippingEstimate:   
672
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 673
                                            if not closedStatus:
674
                                                closed = False
17307 amit.gupta 675
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 676
                                            updateMap['status']='success'
13995 amit.gupta 677
                                            bulk.find(findMap).update({'$set' : updateMap})
678
                                            break
17321 amit.gupta 679
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 680
                            break
681
                    if not matched:
17319 amit.gupta 682
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 683
                        for subOrder in activeOrder['subOrders']:
684
                            if subOrder['closed']==True:
685
                                continue
686
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
687
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
688
                            bulk.find({'orderId': activeOrder['orderId']})
689
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 690
                bulk.execute()
691
                return 'PARSED_SUCCESS'
692
            else:
15563 amit.gupta 693
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
694
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 695
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 696
 
697
                filename = directory + "/" + merchantOrderId   
698
                f = open(filename,'w')
699
                f.write(rawHtml) # python will convert \n to os.linesep
700
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 701
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 702
                print "result", result
15555 amit.gupta 703
                try:
15560 amit.gupta 704
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
705
                    order1.status = result
706
                    order1.rawhtml = rawHtml
17358 amit.gupta 707
                    order1.order_url = url
15555 amit.gupta 708
                    session.commit()
15561 amit.gupta 709
                except:
710
                    traceback.print_exc()
15555 amit.gupta 711
                finally:
712
                    session.close()
14080 amit.gupta 713
                return 'PARSED_SUCCESS'
714
                pass
14008 amit.gupta 715
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 716
        except:
717
            traceback.print_exc()    
718
            return 'PARSED_FAILED'
719
 
720
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 721
        if "ordered from" in detailedStatus.lower():
722
            return MStore.ORDER_PLACED 
723
 
13995 amit.gupta 724
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 725
            if detailedStatus.lower() in value:
13995 amit.gupta 726
                return key
727
 
17253 amit.gupta 728
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 729
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 730
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 731
        br = getBrowserObject()
732
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
733
        url = AMAZON_AFF_URL
734
        response = br.open(url)
735
        #get data for past 40 days and store it to mongo
736
        dt = datetime.now()
14701 amit.gupta 737
        dat = dt - timedelta(days=2)
14651 amit.gupta 738
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
739
        response = br.open(url)
740
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
741
        j=-1
742
        for row in page.split("\n"):
743
            j += 1
744
            if j== 0 or j==1:
745
                continue
746
            fields = row.split("\t")
747
            if len(fields)>1:
748
                print fields
749
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
750
                print amazonAffiliate
751
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
752
            else:
753
                break
14650 amit.gupta 754
 
14608 amit.gupta 755
 
16468 amit.gupta 756
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 757
        print trackingUrl
14624 amit.gupta 758
        subOrder = {}
14749 amit.gupta 759
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 760
        status = MStore.ORDER_SHIPPED
761
        #print page
14624 amit.gupta 762
        soup = BeautifulSoup(page)
16468 amit.gupta 763
        header1 = soup.find("h1") 
764
        if header1:
765
            if header1.text=="Sign In":
766
                print "Login page is displayed for order id", orderId
17307 amit.gupta 767
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 768
                return subOrder
16468 amit.gupta 769
        try:
16474 amit.gupta 770
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 771
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
772
            try:
773
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
774
            except:
775
                displayStatus = detailedStatus
776
            print displayStatus
777
            if detailedStatus.lower().find("delivered")>=0:
778
                print detailedStatus
779
                displayStatus = "Delivered"
780
                status = "Delivered"
781
                try:
782
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
783
                except:
784
                    pass
785
 
786
            elif detailedStatus.lower() == 'returned':
787
                status = 'Cancelled'
788
            subOrder['status'] = status
789
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 790
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 791
 
792
        except:
17350 amit.gupta 793
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 794
            print "failed to parse", orderId
795
            traceback.print_exc()
796
 
14624 amit.gupta 797
        return subOrder
798
 
799
 
800
 
16980 amit.gupta 801
    def parseInfo(self,):
802
        from pyquery import PyQuery as pq
16986 amit.gupta 803
        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 804
        try:
805
            for order in orders:
806
                try:
807
                    doc = pq(order.rawhtml)
808
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
809
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 810
                    orderInfo = All_user_addresses()
16980 amit.gupta 811
                    orderInfo.address = lists[-3].text
812
                    orderInfo.user_id = order.user_id
16986 amit.gupta 813
                    orderInfo.source = 'order' 
814
                    #orderInfo.order_id = order.id
815
                    #orderInfo.email = None
816
                    #orderInfo.name = lists[0].text 
817
                    #orderInfo.mobile = None
16980 amit.gupta 818
                    adSplit = lists[-2].text.split(",")
819
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
820
                    if match:
821
                        items = match.groups()
16986 amit.gupta 822
                    orderInfo.city = adSplit[0].strip()
823
                    orderInfo.pincode = items[1].strip()
824
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 825
                    session.commit()
826
                except:
827
                    session.rollback()
828
                    continue
829
        finally:
830
            session.close()
14608 amit.gupta 831
 
13927 amit.gupta 832
 
833
 
13774 amit.gupta 834
def main():
16874 amit.gupta 835
    store = getStore(1)
18376 amit.gupta 836
    #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 837
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
18378 amit.gupta 838
    session.close()
18377 amit.gupta 839
    for o in orders:
840
        try:
841
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
842
        finally:
843
            session.close()
17631 amit.gupta 844
    #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 845
    #store.trackOrdersForUser(46195, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515', readSSh('/home/amit/amazon.html'))
17231 amit.gupta 846
 
847
def getSummaryFile(directory):
848
    date1 = datetime(2015,1,1)
849
    finalFile = None
850
    try:
851
        for file in os.listdir(directory):
852
            if file.startswith("orderSummary"):
853
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
854
                if date2 > date1:
855
                    date1 = date2
856
                    finalFile=file
857
    except:
858
        print "Missing directory"
17251 amit.gupta 859
    return finalFile 
17231 amit.gupta 860
 
15555 amit.gupta 861
 
862
def parseDetailNotCreated():
863
    try:
864
        store=getStore(1)
865
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 866
        session.close()
15555 amit.gupta 867
        for order in orders:
868
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
869
 
870
    finally:
871
        session.close()
17251 amit.gupta 872
 
873
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 874
    print dateString
17251 amit.gupta 875
    if dateString.lower()=='today':
876
        return date.today()
877
    if dateString.lower()=='yesterday':
878
        return date.today() - timedelta(days=1)
879
    try:
880
        return datetime.strptime(dateString, '%d-%b-%y')
881
    except:
882
        try:
883
            #get Closest Date from today
884
            curDate = date.today()
885
            curTime = datetime(curDate.year, curDate.month, curDate.day)
886
            curYear = curDate.year
887
            prevYear = curYear - 1
17254 amit.gupta 888
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
889
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 890
            if dateMax <= curTime:
891
                return dateMax
892
            else:
893
                return dateMin 
894
        except:
17254 amit.gupta 895
            try:
896
                days_of_week = ['sunday','monday','tuesday','wednesday',
897
                            'thursday','friday','saturday']
898
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
899
                if deltaDays <= 0:
900
                    deltaDays= deltaDays + 7
901
                curDate = curDate - timedelta(days=deltaDays)
902
                print datetime.strftime(curDate, '%d-%b-%y')
903
                return curDate
904
            except:
905
                print "could not parse"
906
                return None
15565 amit.gupta 907
 
17254 amit.gupta 908
def getDateStringArriving(dateString='Thursday'):
909
    print dateString
910
    if dateString.lower()=='today':
911
        return date.today()
912
    if dateString.lower()=='tomorrow':
913
        return date.today() + timedelta(days=1)
914
    try:
915
        return datetime.strptime(dateString, '%d-%b-%y')
916
    except:
917
        try:
918
            #get Closest Date from today
919
            curDate = date.today()
920
            curTime = datetime(curDate.year, curDate.month, curDate.day)
921
            curYear = curDate.year
922
            nextYear = curYear + 1
923
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
924
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
925
            if dateMin >= curTime:
926
                return dateMin
927
            else:
928
                return dateMax 
929
        except:
930
            try:
931
                days_of_week = ['sunday','monday','tuesday','wednesday',
932
                            'thursday','friday','saturday']
933
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
934
                if deltaDays < 0:
935
                    deltaDays= deltaDays + 7
936
                curDate = curDate + timedelta(days=deltaDays)
937
                return curDate
938
            except:
939
                print "Could not parse"
940
                return None 
941
 
18028 amit.gupta 942
def main1():
943
    store = getStore(1)
944
    for merchantOrder in store.db.merchantOrder.find({"subOrders":{"$elemMatch":{"cashBackStatus":"Not Applicable", "cashBackPercentage":{"$gt":0}}}}):
945
        mo = obj(merchantOrder)
946
        for subOrder in mo.subOrders:
947
            subOrder.closed=False 
948
        print "orderId", mo.orderId
949
        store.populateDerivedFields(mo, False)
950
        store.db.merchantOrder.update({"orderId":mo.orderId}, {"$set":todict(mo)})
951
        break
952
 
953
 
15565 amit.gupta 954
def parseOrderNotCreated():
955
    try:
956
        store=getStore(1)
957
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
958
        session.close()
959
        for order in orders:
15567 amit.gupta 960
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
961
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
962
            order1.status = result
963
            session.commit()
15565 amit.gupta 964
    finally:
965
        session.close()
15555 amit.gupta 966
 
13774 amit.gupta 967
if __name__ == '__main__':
17599 amit.gupta 968
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
969
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
18156 amit.gupta 970
#    readSSh("/AmazonTrack/User5525/171-0333104-6169933")
18159 amit.gupta 971
    main()
972
    #store=getStore(1)
973
    #store.trackOrdersForUser(5525, "https://www.amazon.in/gp/css/summary/edit.html?orderID=171-0333104-6169933", readSSh("/AmazonTrack/User5525/171-0333104-6169933"))