Subversion Repositories SmartDukaan

Rev

Rev 18377 | Rev 18379 | 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
18156 amit.gupta 14
from dtr.sources.flipkart import todict
17307 amit.gupta 15
from dtr.storage.DataService import OrdersRaw, Orders, Order_Parse_Info, \
16986 amit.gupta 16
    All_user_addresses
16874 amit.gupta 17
from dtr.storage.Mongo import getDealRank
17307 amit.gupta 18
from dtr.utils import utils
16874 amit.gupta 19
from dtr.utils.utils import fetchResponseUsingProxy, readSSh
20
from elixir import *
14650 amit.gupta 21
import base64
16874 amit.gupta 22
import dtr
14650 amit.gupta 23
import gzip
14624 amit.gupta 24
import mechanize
14285 amit.gupta 25
import os.path
18376 amit.gupta 26
from pyquery import PyQuery as pq
13774 amit.gupta 27
import re
14033 amit.gupta 28
import time
13809 amit.gupta 29
import traceback
14650 amit.gupta 30
import urllib2
13774 amit.gupta 31
 
32
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 33
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 34
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
14624 amit.gupta 35
AMAZON_AFF_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/listReports'
14650 amit.gupta 36
AMAZON_AFF_FILE_URL = 'https://assoc-datafeeds-eu.amazon.com/datafeed/getReport?filename=saholic-21-orders-report-%s.tsv.gz'
13774 amit.gupta 37
class Store(MStore):
13821 amit.gupta 38
 
17348 amit.gupta 39
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch', 'order received'],
14064 amit.gupta 40
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
17324 amit.gupta 41
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled', 'replacement complete', 'return received'],
17258 amit.gupta 42
                            MStore.ORDER_DELIVERED : ['delivered', 'your package was delivered', 'package was handed directly to customer']
13821 amit.gupta 43
                           }
13774 amit.gupta 44
 
45
    def __init__(self,store_id):
46
        super(Store, self).__init__(store_id)
47
 
48
    def getName(self):
14749 amit.gupta 49
        return "amazon"
13774 amit.gupta 50
 
14945 amit.gupta 51
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, track=False):
17718 amit.gupta 52
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14958 amit.gupta 53
        parseString = "Tracking" if track else "Transacted"
54
        print parseString, "Order Id to be parsed is :", orderId
13796 amit.gupta 55
        resp = {}
14813 amit.gupta 56
        resp['result'] = 'ORDER_NOT_CREATED'
13821 amit.gupta 57
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 58
            try:
17600 amit.gupta 59
                doc = pq(rawHtml)
14608 amit.gupta 60
                try:
17600 amit.gupta 61
                    orderUrl = doc('a.a-touch-link').attr.href
14608 amit.gupta 62
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
63
                except:
17600 amit.gupta 64
                    merchantOrderId = doc("#orders-list>div>span>b").html()
14699 amit.gupta 65
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, True)
14972 amit.gupta 66
                order.orderTrackingUrl = ORDER_REDIRECT_URL % (merchantOrderId)
67
                order.orderSuccessUrl = orderSuccessUrl
14211 amit.gupta 68
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 69
                order.requireDetail = True
14698 amit.gupta 70
                order.status = 'html_required'
14297 amit.gupta 71
                order.closed = None
14312 amit.gupta 72
                if self._saveToOrder(todict(order)):
73
                    resp['result'] = 'ORDER_CREATED'
74
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
75
                    resp["htmlRequired"] = True
76
                    resp['orderId'] = orderId
77
                else:
78
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
79
 
13774 amit.gupta 80
            except:
14813 amit.gupta 81
                #Write all cases here for Order Not created Known
18158 amit.gupta 82
                soup = BeautifulSoup(rawHtml, "html5lib")
14809 amit.gupta 83
                try:
16874 amit.gupta 84
                    if not soup.body:
85
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
17359 amit.gupta 86
                    elif 'Securely redirecting you' in soup.find("h3").text.strip() or soup.find("h3").text.strip()=="Orders":
14813 amit.gupta 87
                        resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
14809 amit.gupta 88
                    else:
14813 amit.gupta 89
                        raise
14809 amit.gupta 90
                except:
14814 amit.gupta 91
                    try:
17356 amit.gupta 92
                        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 93
                            resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
94
                        else:
95
                            raise
96
                    except:
15046 amit.gupta 97
                        try:
17356 amit.gupta 98
                            if soup.find("h2").text.strip() in ['Web page not available','Webpage not available', 'Do you have an Amazon password?']:
15046 amit.gupta 99
                                resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
100
                            else:
101
                                raise
102
                        except:
15565 amit.gupta 103
                            try:
16876 amit.gupta 104
                                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 105
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
16877 amit.gupta 106
                                elif soup.find("b", {'class':'h1'}).text.strip().find("We're sorry") > -1:
107
                                    resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
15565 amit.gupta 108
                                else:
109
                                    raise
110
                            except:
111
                                resp['result'] = 'ORDER_NOT_CREATED_UNKNOWN'
112
 
13774 amit.gupta 113
        else:
13781 amit.gupta 114
            try:
14722 amit.gupta 115
                mo = self.db.merchantOrder.find_one({"orderId":orderId})
116
                if mo is not None:
14749 amit.gupta 117
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
14722 amit.gupta 118
                    merchantOrder.createdOn = mo.get("createdOn")
119
                    merchantOrder.createdOnInt =  mo.get("createdOnInt")
120
                else:
14945 amit.gupta 121
                    print "Could not find amazon order with order Id", orderId
14749 amit.gupta 122
                    merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
18158 amit.gupta 123
                soup = BeautifulSoup(rawHtml, "html5lib")
17331 amit.gupta 124
                if not soup.body:
17319 amit.gupta 125
                    resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
126
                else:   
14545 amit.gupta 127
                    try:
17319 amit.gupta 128
                        self.parseNewStlye(merchantOrder, soup)
15046 amit.gupta 129
                        resp['result'] = 'DETAIL_CREATED'
14545 amit.gupta 130
                    except:
15056 amit.gupta 131
                        try:
17319 amit.gupta 132
                            traceback.print_exc()
18376 amit.gupta 133
                            self.parseAnotherStlye(merchantOrder, pq(rawHtml))
17319 amit.gupta 134
                            resp['result'] = 'DETAIL_CREATED'
15056 amit.gupta 135
                        except:
15059 amit.gupta 136
                            try:
18376 amit.gupta 137
                                traceback.print_exc()
138
                                self.parseOldStlye(merchantOrder, soup)
139
                                resp['result'] = 'DETAIL_CREATED'
15059 amit.gupta 140
                            except:
18376 amit.gupta 141
                                traceback.print_exc()
17319 amit.gupta 142
                                try:
18376 amit.gupta 143
                                    self.parseCancelled(merchantOrder, soup)
144
                                    resp['result'] = 'ORDER_CANCELLED'
17319 amit.gupta 145
                                except:
18376 amit.gupta 146
                                    try:
147
                                        if soup.find("h1").text.strip() in ["Your Account"] or soup.find("h1").span.text=="Account":
148
                                            resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
149
                                        else:
150
                                            raise
151
                                    except:
152
                                        if soup.find(id="ap_signin_pagelet_title").find("h1").text.strip()=="Sign In":
153
                                            resp['result'] = 'DETAIL_NOT_CREATED_KNOWN'
154
                                        else:
155
                                            raise
17599 amit.gupta 156
                if resp['result'] == 'DETAIL_NOT_CREATED_KNOWN':
17357 amit.gupta 157
                    self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
13781 amit.gupta 158
            except:
17666 amit.gupta 159
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"status":"html_required"}})
14813 amit.gupta 160
                print "Error occurred"
15046 amit.gupta 161
                resp['result'] = 'DETAIL_NOT_CREATED_UNKNOWN'
14813 amit.gupta 162
                traceback.print_exc()
15046 amit.gupta 163
        return resp    
13774 amit.gupta 164
 
13868 amit.gupta 165
    #This should be exposed from api for specific sources
166
    def scrapeStoreOrders(self):
14624 amit.gupta 167
        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
168
        for merchantOrder in orders:
14981 amit.gupta 169
            executeBulk = False
14624 amit.gupta 170
            try:
171
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
172
                closed = True
173
                map1 = {}
174
                for subOrder in merchantOrder.get("subOrders"):
175
                    if subOrder.get("closed"):
176
                        continue
177
                    elif subOrder.get("trackingUrl") is None:
178
                        closed = False
179
                        continue
180
                    findMap = {"orderId":merchantOrder.get("orderId"), "subOrders.merchantSubOrderId":subOrder.get("merchantSubOrderId")} 
181
                    trackingUrl = subOrder.get("trackingUrl")
182
                    if not map1.has_key(trackingUrl):
16468 amit.gupta 183
                        map1[trackingUrl] = self.parseTrackingUrl(trackingUrl, merchantOrder.get("orderId"))
14624 amit.gupta 184
                    newOrder = map1.get(trackingUrl)
16474 amit.gupta 185
                    if newOrder:
186
                        executeBulk = True
187
                        updateMap = self.getUpdateMap(newOrder, subOrder.get('cashBackStatus'))
188
                        print findMap, "\n", updateMap
189
                        bulk.find(findMap).update({'$set' : updateMap})
190
                        closed = closed and newOrder['closed']
14981 amit.gupta 191
                if executeBulk:
14697 amit.gupta 192
                    bulk.find({"orderId":merchantOrder.get("orderId")}).update({"$set":{"closed":closed, "parseError":False}})
14981 amit.gupta 193
                    bulk.execute()
14624 amit.gupta 194
            except:
14847 amit.gupta 195
                tprint("Could not update " + str(merchantOrder['orderId']) + " For store " + self.getName())
14692 amit.gupta 196
                self.db.merchantOrder.update({"orderId":merchantOrder['orderId']}, {"$set":{"parseError":True}})
14624 amit.gupta 197
                traceback.print_exc()
198
 
14608 amit.gupta 199
 
200
 
14771 amit.gupta 201
    def parserest(self, soup):
202
        print "Hi"
203
        if soup.find('h1'):
204
            print "OK"
14608 amit.gupta 205
 
14464 amit.gupta 206
    def parseOldStlye(self, merchantOrder, soup):
207
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
208
        table = soup.body.findAll("table", recursive=False)[1]
209
        #print table
210
        tables = table.tr.td.findAll("table", recursive=False)
211
        for tr in tables[2].findAll("tr"):
212
            boldElement = tr.td.b
213
            if "Order Placed" in str(boldElement):
214
                merchantOrder.placedOn = boldElement.next_sibling.strip()
215
            if "order number" in str(boldElement):
216
                merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
217
            if "Order Total" in str(boldElement):
218
                merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
219
        anchors = table.tr.td.findAll("a", recursive=False)
220
        paymentAnchor = anchors.pop(-1)
221
 
222
        count = 0
223
        subOrders = []
224
        merchantOrder.subOrders = subOrders
225
        counter = 0 
226
        for anchor in anchors:
227
            count += 1
228
            tab = anchor.next_sibling
229
            status = MStore.ORDER_PLACED
230
            subStr = "Delivery #" + str(count) + ":"
231
            if subStr in  tab.find("b").text:
232
                detailedStatus = tab.find("b").text.replace(subStr, '').strip()
233
 
234
            tab = tab.next_sibling.next_sibling
235
            trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
236
 
237
            estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
238
 
239
            orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
240
            i = -1
241
            for orderItemTr in orderItemTrs:
242
                i += 1
243
                if i%2 == 0:
244
                    continue
245
                counter += 1
246
                quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
247
 
248
                productUrl = orderItemTr.td.contents[1].a["href"]
249
                productTitle = orderItemTr.td.contents[1].a.text
250
 
251
                unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
252
 
253
 
254
                subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
255
                subOrder.merchantSubOrderId = str(counter) + " of " + merchantOrder.merchantOrderId
256
                subOrder.estimatedDeliveryDate = estimatedDelivery
14722 amit.gupta 257
                estDlvyTime = datetime.strptime(estimatedDelivery.split('-')[0].strip(), "%A %d %B %Y")
258
                createdOn = datetime.fromtimestamp(merchantOrder.createdOnInt)
259
                subOrder.trackAfter = int(time.mktime(max(estDlvyTime-timedelta(days=4),createdOn + timedelta(days=3)).timetuple()))
14464 amit.gupta 260
                subOrder.productCode = productUrl.split('/')[5]
261
                subOrder.detailedStatus = detailedStatus
18028 amit.gupta 262
                (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice)
15346 amit.gupta 263
                dealRank = getDealRank(subOrder.productCode, self.store_id, merchantOrder.userId)
264
                subOrder.dealRank = dealRank.get('rank')
265
                subOrder.rankDesc = dealRank.get('description')
16283 amit.gupta 266
                subOrder.maxNlc = dealRank.get('maxNlc')
267
                subOrder.minNlc = dealRank.get('minNlc')
268
                subOrder.db = dealRank.get('dp')
269
                subOrder.itemStatus = dealRank.get('status')
14464 amit.gupta 270
                cashbackStatus = Store.CB_PENDING
271
                if cashbackAmount <= 0:
272
                    cashbackStatus = Store.CB_NA
273
                subOrder.cashBackStatus = cashbackStatus
18028 amit.gupta 274
                subOrder.cashBackAmount = cashbackAmount*quantity
14464 amit.gupta 275
                if percentage > 0:
276
                    subOrder.cashBackPercentage = percentage
277
                subOrders.append(subOrder)
278
        priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
279
        totalAmount = 0
280
        grandAmount = 0
281
        for price in priceList:
282
            labelTd = price.td
283
            if 'Subtotal:' in labelTd.text:
284
                totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
285
            elif 'Grand Total:' in labelTd.text:
286
                grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
287
        if grandAmount < totalAmount:
288
            diff = totalAmount - grandAmount
289
            for subOrder in merchantOrder.subOrders:
290
                subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14698 amit.gupta 291
        merchantOrder.status='success'
14464 amit.gupta 292
        self._updateToOrder(todict(merchantOrder))
18376 amit.gupta 293
 
294
    def parseAnotherStlye(self, merchantOrder, pqobj):
295
        counter=0
296
        detailSection = None
297
        shipmentSection = None
298
        summarySection = None
299
        for el in pqobj('.a-section'):
300
            if "View order details" in pq(el).text():
301
                detailSection =  pq(el)
302
                counter += 1
303
            if "Shipment details" in pq(el).text():
304
                shipmentSection =  pq(el)
305
                counter += 1
306
            if "Order Summary" in pq(el).text():
307
                summarySection = pq(el)
308
                counter += 1
309
            if counter == 3:
310
                break
311
 
312
        print "detail" if detailSection else "", "shipment" if shipmentSection else "", "summary" if summarySection else ""
313
 
314
        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()
315
 
316
        i=-1    
317
        for s in shipmentSection('.a-box-group'):
318
            shipmentGroup = pq(s)
319
            for shipment in shipmentGroup('.a-box'):
320
                i += 1 
321
                if i==0:
322
                    continue
323
                shipment = pq(shipment)
324
                shipmentStatusSection = shipment('.a-section:eq(0)')
325
                productDetails = shipment('.a-section:eq(1)>.a-row')
326
                print shipmentStatusSection('h3').text(), shipmentStatusSection('span').text()
327
                for productDetail in productDetails:
328
                    productDetail = pq(productDetail)
329
                    print productDetail
330
                    pImg = productDetail.children('div').eq(0)
331
                    pQty = productDetail.children('div').eq(1)
332
                    #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')
333
                    print pImg('a').attr('href'), pQty('.a-row:nth-child(2)')('span').text().split(':')[1].strip()
334
                    print int(float(pQty('nobr').text().replace('Rs.','').replace(',', '')))
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')
476
                    cashbackStatus = Store.CB_PENDING
477
                    if cashbackAmount <= 0:
478
                        cashbackStatus = Store.CB_NA
479
                    subOrder.cashBackStatus = cashbackStatus
480
                    subOrder.cashBackAmount = cashbackAmount*quantity
481
                    if percentage > 0:
482
                        subOrder.cashBackPercentage = percentage
483
                    if hasattr(subOrder, 'deliveredOn') or subOrder.status==Store.ORDER_DELIVERED:                               
484
                        subOrder.status = Store.ORDER_DELIVERED
485
                        subOrder.closed = True
486
                        if subOrder.cashBackStatus == Store.CB_PENDING:
487
                            subOrder.cashBackStatus = Store.CB_APPROVED
488
                    elif closedStatus:
489
                        closedStatus= False
490
                    subOrders.append(subOrder)
14698 amit.gupta 491
        merchantOrder.status='success'
15473 amit.gupta 492
        merchantOrder.closed = closedStatus
14464 amit.gupta 493
        self._updateToOrder(todict(merchantOrder))
14545 amit.gupta 494
 
495
    def parseCancelled(self, merchantOrder,soup):
14722 amit.gupta 496
        try:
497
            fonts = soup.body.findAll("table", recursive=False)[1].findAll("font")
498
            if fonts[0].text == "Important Message":
499
                if fonts[1].text=="This order has been cancelled.":
500
                    merchantOrder.closed = True
501
                    merchantOrder.status = "cancelled"
502
                    merchantOrder.requireDetail = False
503
                    self._updateToOrder(todict(merchantOrder))
504
                    return
505
                else:
506
                    raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14545 amit.gupta 507
            else:
508
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14722 amit.gupta 509
        except:
14549 amit.gupta 510
            orderDetails = soup.body.find(id="orderDetails")
511
            if orderDetails is not None and orderDetails.h4.text == "This order has been cancelled.":
512
                merchantOrder.closed = True
14698 amit.gupta 513
                merchantOrder.status = "cancelled"
14549 amit.gupta 514
                merchantOrder.requireDetail = False
515
                self._updateToOrder(todict(merchantOrder))
516
            else:
517
                raise ParseException("parseCancelled", "Found detailed status" + fonts[1].text)
14464 amit.gupta 518
 
13868 amit.gupta 519
    def getTrackingUrls(self, userId):
14074 amit.gupta 520
 
521
        missingOrderUrls = []
522
        missingOrders = self._getMissingOrders({'userId':userId})
523
        for missingOrder in missingOrders:
524
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
17453 amit.gupta 525
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 526
        count = len(orders)
13879 amit.gupta 527
        print "count", count
14948 amit.gupta 528
        print "Missing Urls"
529
        print "*************"
530
        print missingOrderUrls
13868 amit.gupta 531
        if count > 0:
14684 amit.gupta 532
            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 533
        else: 
14074 amit.gupta 534
            return missingOrderUrls
13927 amit.gupta 535
 
536
    def trackOrdersForUser(self, userId, url, rawHtml):
17718 amit.gupta 537
        rawHtml = re.sub(r'[^\x00-\x7F]+',' ', rawHtml)
14985 amit.gupta 538
        directory = "/AmazonTrack/User" + str(userId)
14285 amit.gupta 539
        if not os.path.exists(directory):
540
            os.makedirs(directory)
14173 amit.gupta 541
 
14945 amit.gupta 542
 
13995 amit.gupta 543
        try:
544
            searchMap = {'userId':userId}
545
            collectionMap = {'merchantOrderId':1}
546
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 547
            datetimeNow = datetime.now()
548
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 549
            print "url----------------", url
17465 amit.gupta 550
            cancelledSummary = False
14320 amit.gupta 551
            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 552
                if url == 'https://www.amazon.in/gp/css/order-history':
553
                    filename = directory + "/orderSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
554
                else:   
17465 amit.gupta 555
                    filename = directory + "/cancelledSummary" + datetime.strftime(datetime.now(), '%d-%m:%H:%M:%S')
556
                    cancelledSummary = True   
14945 amit.gupta 557
                f = open(filename,'w')
558
                f.write(rawHtml) # python will convert \n to os.linesep
559
                f.close() # you can omit in most cases as the destructor will call if
18158 amit.gupta 560
                soup = BeautifulSoup(rawHtml,'html5lib')
13995 amit.gupta 561
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
562
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
563
                for activeOrder in activeOrders:
17307 amit.gupta 564
                    matched=False
13995 amit.gupta 565
                    for orderEle in allOrders:
17254 amit.gupta 566
                        deliveredOn = None
567
                        deliveryEstimate = None
17258 amit.gupta 568
                        shippingEstimate = None
13995 amit.gupta 569
                        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'})
570
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
571
                        if merchantOrderId==activeOrder['merchantOrderId']:
17307 amit.gupta 572
                            matched=True
13995 amit.gupta 573
                            closed = True
17465 amit.gupta 574
                            if not cancelledSummary:
575
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?shipment.*?')}, recursive=False)
576
                            else:
577
                                shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
578
                                shipments.pop(0)
13995 amit.gupta 579
                            for shipment in shipments:
17307 amit.gupta 580
                                orderStatusDesc = None
13995 amit.gupta 581
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 582
                                sdivs = shipment.div.div.findAll('div', recursive=False)
17254 amit.gupta 583
                                try:
17270 amit.gupta 584
                                    orderStatus = sdivs[0].span.text.strip()
585
                                    status = self._getStatusFromDetailedStatus(orderStatus)
17254 amit.gupta 586
                                except:
17270 amit.gupta 587
                                    try:
588
                                        dateString = orderStatus.split("Delivered ")[1].strip()
589
                                        status = MStore.ORDER_DELIVERED
590
                                        deliveredOn = datetime.strftime(getDateStringDelivered(dateString), '%d-%b-%y')
591
                                    except:
592
                                        try:
17271 amit.gupta 593
                                            dateString = sdivs[0].span.text.strip().split("Arriving ")[1].split("by")[0].strip()
17270 amit.gupta 594
                                            status = MStore.ORDER_SHIPPED
595
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(dateString), '%d-%b-%y')
596
                                        except:
597
                                            print "Unknown status Alert -", orderStatus
17348 amit.gupta 598
                                print merchantOrderId, "Order Status", orderStatus
17256 amit.gupta 599
                                try:
17270 amit.gupta 600
                                    orderStatusDesc = sdivs[0].findAll('div')[1].div.text.strip()
601
                                except:
17271 amit.gupta 602
                                    try:
603
                                        orderStatusDesc = sdivs[0].findAll('div')[1].text.strip()
604
                                    except:
17307 amit.gupta 605
                                        print "Order Status Description None or empty for", merchantOrderId, "and User", userId
17270 amit.gupta 606
 
17271 amit.gupta 607
                                if orderStatusDesc: 
17348 amit.gupta 608
                                    print merchantOrderId, "Order status desc", orderStatusDesc
17271 amit.gupta 609
                                    try:
17348 amit.gupta 610
                                        status = self._getStatusFromDetailedStatus(orderStatus)
611
                                    except:
612
                                        pass
613
                                    try:
17271 amit.gupta 614
                                        if "Dispatch estimate" in orderStatusDesc:
615
                                            shippingEstimate = orderStatus.split("Dispatch estimate").split("-")[0].strip()
616
                                        elif "Delivery estimate" in orderStatus:
617
                                            deliveryEstimate = orderStatus.split("Delivery estimate").split("-")[0].strip()
618
                                        elif "Arriving" in orderStatus:
619
                                            deliveryEstimate = datetime.strftime(getDateStringArriving(orderStatus.split("Arriving")[1].strip().split("by")[0].strip()), '%d-%b-%y')
620
                                    except:
621
                                        print "Could not find anything relevent for merchantOrder", merchantOrderId, "and User", userId 
622
                                        closed=False
623
                                        status = None
17258 amit.gupta 624
 
14335 amit.gupta 625
                                productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 626
                                trackingUrl = None
627
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
628
                                    if buttonDiv.find('a').text.strip()=='Track package':
629
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
630
                                        if not trackingUrl.startswith("http"):
14608 amit.gupta 631
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
13995 amit.gupta 632
                                        break
633
                                for prodDiv in productDivs:
634
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
635
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
636
                                    imgUrl = prodDiv.find("img")["src"]
637
                                    for subOrder in activeOrder['subOrders']:
17307 amit.gupta 638
                                        if subOrder['closed']==True:
639
                                            continue
17230 amit.gupta 640
                                        if subOrder['productTitle'] in productTitle:
13995 amit.gupta 641
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
642
                                            updateMap = {}
643
                                            closedStatus = False
644
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 645
                                            updateMap['subOrders.$.lastTracked'] = timestamp
17258 amit.gupta 646
                                            if status:
17348 amit.gupta 647
                                                updateMap['subOrders.$.detailedStatus'] = orderStatus
17258 amit.gupta 648
                                                updateMap['subOrders.$.status'] = status 
649
                                                cashbackStatus = subOrder.get("cashBackStatus")
13995 amit.gupta 650
 
17251 amit.gupta 651
                                            if status==MStore.ORDER_DELIVERED:
17271 amit.gupta 652
                                                if deliveredOn:                               
17254 amit.gupta 653
                                                    updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 654
                                                closedStatus = True
655
                                                updateMap['subOrders.$.closed'] = True
656
                                                if cashbackStatus == Store.CB_PENDING:
14607 amit.gupta 657
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_APPROVED
13995 amit.gupta 658
                                            if status==MStore.ORDER_CANCELLED:     
659
                                                closedStatus = True
660
                                                updateMap['subOrders.$.closed'] = True
661
                                                if cashbackStatus == Store.CB_PENDING:
662
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
17254 amit.gupta 663
                                            if status==MStore.ORDER_SHIPPED:
664
                                                if deliveryEstimate:   
665
                                                    updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
13995 amit.gupta 666
                                                if trackingUrl is not None:
667
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
17322 amit.gupta 668
                                                    updateMap['subOrders.$.trackMissing'] = False
17258 amit.gupta 669
                                            if shippingEstimate:   
670
                                                updateMap['subOrders.$.estimatedShippingDate'] = shippingEstimate
13995 amit.gupta 671
                                            if not closedStatus:
672
                                                closed = False
17307 amit.gupta 673
                                            #{"subOrders.closed":False,"subOrders.trackingUrl":{"$exists":False},"subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}
17359 amit.gupta 674
                                            updateMap['status']='success'
13995 amit.gupta 675
                                            bulk.find(findMap).update({'$set' : updateMap})
676
                                            break
17321 amit.gupta 677
                            bulk.find({'orderId': activeOrder['orderId']}).update({"$set":{'closed':closed}})
17307 amit.gupta 678
                            break
679
                    if not matched:
17319 amit.gupta 680
                        updateMap = {'subOrders.$.trackMissing': True}
17307 amit.gupta 681
                        for subOrder in activeOrder['subOrders']:
682
                            if subOrder['closed']==True:
683
                                continue
684
                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId"), 
685
                                       "subOrders.trackAfter":{"$lt":utils.getCurrTimeStamp()}}
686
                            bulk.find({'orderId': activeOrder['orderId']})
687
                            bulk.find(findMap).update({'$set':updateMap})
14080 amit.gupta 688
                bulk.execute()
689
                return 'PARSED_SUCCESS'
690
            else:
15563 amit.gupta 691
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0]
692
                print  "merchantOrderId", merchantOrderId
14085 amit.gupta 693
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14945 amit.gupta 694
 
695
                filename = directory + "/" + merchantOrderId   
696
                f = open(filename,'w')
697
                f.write(rawHtml) # python will convert \n to os.linesep
698
                f.close() # you can omit in most cases as the destructor will call if
15555 amit.gupta 699
                result = self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, True)['result']
15562 amit.gupta 700
                print "result", result
15555 amit.gupta 701
                try:
15560 amit.gupta 702
                    order1 = session.query(OrdersRaw).filter_by(id=merchantOrder['orderId']).first()
703
                    order1.status = result
704
                    order1.rawhtml = rawHtml
17358 amit.gupta 705
                    order1.order_url = url
15555 amit.gupta 706
                    session.commit()
15561 amit.gupta 707
                except:
708
                    traceback.print_exc()
15555 amit.gupta 709
                finally:
710
                    session.close()
14080 amit.gupta 711
                return 'PARSED_SUCCESS'
712
                pass
14008 amit.gupta 713
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 714
        except:
715
            traceback.print_exc()    
716
            return 'PARSED_FAILED'
717
 
718
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 719
        if "ordered from" in detailedStatus.lower():
720
            return MStore.ORDER_PLACED 
721
 
13995 amit.gupta 722
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 723
            if detailedStatus.lower() in value:
13995 amit.gupta 724
                return key
725
 
17253 amit.gupta 726
        print "Detailed Status need to be mapped", "Store:", self.store_id, detailedStatus
13995 amit.gupta 727
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
14624 amit.gupta 728
    def scrapeAffiliate(self, startDate=None, endDate=None):
14650 amit.gupta 729
        br = getBrowserObject()
730
        br.add_password('https://assoc-datafeeds-eu.amazon.com', 'Saholic', 'Fnubyvp')
731
        url = AMAZON_AFF_URL
732
        response = br.open(url)
733
        #get data for past 40 days and store it to mongo
734
        dt = datetime.now()
14701 amit.gupta 735
        dat = dt - timedelta(days=2)
14651 amit.gupta 736
        url = AMAZON_AFF_FILE_URL%(datetime.strftime(dat, "%Y%m%d"))
737
        response = br.open(url)
738
        page = gzip.GzipFile(fileobj=response, mode='rb').read()
739
        j=-1
740
        for row in page.split("\n"):
741
            j += 1
742
            if j== 0 or j==1:
743
                continue
744
            fields = row.split("\t")
745
            if len(fields)>1:
746
                print fields
747
                amazonAffiliate = AmazonAffiliateInfo(fields[0], fields[1], fields[2], fields[3], fields[4], fields[5], fields[6], fields[7], fields[8], fields[9])
748
                print amazonAffiliate
749
                self.db.amazonAffiliateInfo.insert(todict(amazonAffiliate))
750
            else:
751
                break
14650 amit.gupta 752
 
14608 amit.gupta 753
 
16468 amit.gupta 754
    def parseTrackingUrl(self, trackingUrl, orderId):
16474 amit.gupta 755
        print trackingUrl
14624 amit.gupta 756
        subOrder = {}
14749 amit.gupta 757
        page = fetchResponseUsingProxy(trackingUrl)
16468 amit.gupta 758
        status = MStore.ORDER_SHIPPED
759
        #print page
14624 amit.gupta 760
        soup = BeautifulSoup(page)
16468 amit.gupta 761
        header1 = soup.find("h1") 
762
        if header1:
763
            if header1.text=="Sign In":
764
                print "Login page is displayed for order id", orderId
17307 amit.gupta 765
                self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})
14624 amit.gupta 766
                return subOrder
16468 amit.gupta 767
        try:
16474 amit.gupta 768
            print "Tracking page is displayed for order id", orderId
16468 amit.gupta 769
            detailedStatus = soup.find("div", {"class":"top"}).span.text.strip()
770
            try:
771
                displayStatus = soup.find("div",{"class":"a-column a-span12 shipment-status-content"}).span.text.strip()
772
            except:
773
                displayStatus = detailedStatus
774
            print displayStatus
775
            if detailedStatus.lower().find("delivered")>=0:
776
                print detailedStatus
777
                displayStatus = "Delivered"
778
                status = "Delivered"
779
                try:
780
                    subOrder["deliveredOn"] = detailedStatus.split("on")[1].strip()
781
                except:
782
                    pass
783
 
784
            elif detailedStatus.lower() == 'returned':
785
                status = 'Cancelled'
786
            subOrder['status'] = status
787
            subOrder['detailedStatus'] = displayStatus 
17350 amit.gupta 788
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":False}})    
16468 amit.gupta 789
 
790
        except:
17350 amit.gupta 791
            self.db.merchantOrder.update({"orderId":orderId}, {"$set":{"trackError":True}})    
16468 amit.gupta 792
            print "failed to parse", orderId
793
            traceback.print_exc()
794
 
14624 amit.gupta 795
        return subOrder
796
 
797
 
798
 
16980 amit.gupta 799
    def parseInfo(self,):
800
        from pyquery import PyQuery as pq
16986 amit.gupta 801
        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 802
        try:
803
            for order in orders:
804
                try:
805
                    doc = pq(order.rawhtml)
806
                    #a1= " ".join(["" if not div.text else div.text.replace("\t","").replace("\n","").replace(" ", "") for div in pq(doc('article')[-1])('div')])
807
                    lists = doc('ul.displayAddressUL li')
16986 amit.gupta 808
                    orderInfo = All_user_addresses()
16980 amit.gupta 809
                    orderInfo.address = lists[-3].text
810
                    orderInfo.user_id = order.user_id
16986 amit.gupta 811
                    orderInfo.source = 'order' 
812
                    #orderInfo.order_id = order.id
813
                    #orderInfo.email = None
814
                    #orderInfo.name = lists[0].text 
815
                    #orderInfo.mobile = None
16980 amit.gupta 816
                    adSplit = lists[-2].text.split(",")
817
                    match = re.match(r"([a-z ]+)([0-9]+)", adSplit[1], re.I)
818
                    if match:
819
                        items = match.groups()
16986 amit.gupta 820
                    orderInfo.city = adSplit[0].strip()
821
                    orderInfo.pincode = items[1].strip()
822
                    orderInfo.state = items[0].strip().title()
16980 amit.gupta 823
                    session.commit()
824
                except:
825
                    session.rollback()
826
                    continue
827
        finally:
828
            session.close()
14608 amit.gupta 829
 
13927 amit.gupta 830
 
831
 
13774 amit.gupta 832
def main():
16874 amit.gupta 833
    store = getStore(1)
18376 amit.gupta 834
    #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 835
    orders = list(session.query(OrdersRaw).filter_by(status = 'DETAIL_NOT_CREATED_UNKNOWN').filter(OrdersRaw.id > 61071).all())
18378 amit.gupta 836
    session.close()
18377 amit.gupta 837
    for o in orders:
838
        try:
839
            store.trackOrdersForUser(o.id, o.order_url, o.rawhtml)
840
        finally:
841
            session.close()
17631 amit.gupta 842
    #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 843
    #store.trackOrdersForUser(46195, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=404-4294022-1187515', readSSh('/home/amit/amazon.html'))
17231 amit.gupta 844
 
845
def getSummaryFile(directory):
846
    date1 = datetime(2015,1,1)
847
    finalFile = None
848
    try:
849
        for file in os.listdir(directory):
850
            if file.startswith("orderSummary"):
851
                date2 = datetime.strptime("2015-" + file.split("orderSummary")[1].split(":")[0], "%Y-%d-%m")
852
                if date2 > date1:
853
                    date1 = date2
854
                    finalFile=file
855
    except:
856
        print "Missing directory"
17251 amit.gupta 857
    return finalFile 
17231 amit.gupta 858
 
15555 amit.gupta 859
 
860
def parseDetailNotCreated():
861
    try:
862
        store=getStore(1)
863
        orders = session.query(OrdersRaw).filter_by(status='DETAIL_NOT_CREATED_UNKNOWN').all()
15558 amit.gupta 864
        session.close()
15555 amit.gupta 865
        for order in orders:
866
            store.trackOrdersForUser(order.id, order.order_url, order.rawhtml)
867
 
868
    finally:
869
        session.close()
17251 amit.gupta 870
 
871
def getDateStringDelivered(dateString='Monday'):
17254 amit.gupta 872
    print dateString
17251 amit.gupta 873
    if dateString.lower()=='today':
874
        return date.today()
875
    if dateString.lower()=='yesterday':
876
        return date.today() - timedelta(days=1)
877
    try:
878
        return datetime.strptime(dateString, '%d-%b-%y')
879
    except:
880
        try:
881
            #get Closest Date from today
882
            curDate = date.today()
883
            curTime = datetime(curDate.year, curDate.month, curDate.day)
884
            curYear = curDate.year
885
            prevYear = curYear - 1
17254 amit.gupta 886
            dateMax = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
887
            dateMin = datetime.strptime(dateString + " " + str(prevYear), "%A, %d %b %Y")
17251 amit.gupta 888
            if dateMax <= curTime:
889
                return dateMax
890
            else:
891
                return dateMin 
892
        except:
17254 amit.gupta 893
            try:
894
                days_of_week = ['sunday','monday','tuesday','wednesday',
895
                            'thursday','friday','saturday']
896
                deltaDays = curDate.isoweekday() - days_of_week.index(dateString.lower())
897
                if deltaDays <= 0:
898
                    deltaDays= deltaDays + 7
899
                curDate = curDate - timedelta(days=deltaDays)
900
                print datetime.strftime(curDate, '%d-%b-%y')
901
                return curDate
902
            except:
903
                print "could not parse"
904
                return None
15565 amit.gupta 905
 
17254 amit.gupta 906
def getDateStringArriving(dateString='Thursday'):
907
    print dateString
908
    if dateString.lower()=='today':
909
        return date.today()
910
    if dateString.lower()=='tomorrow':
911
        return date.today() + timedelta(days=1)
912
    try:
913
        return datetime.strptime(dateString, '%d-%b-%y')
914
    except:
915
        try:
916
            #get Closest Date from today
917
            curDate = date.today()
918
            curTime = datetime(curDate.year, curDate.month, curDate.day)
919
            curYear = curDate.year
920
            nextYear = curYear + 1
921
            dateMin = datetime.strptime(dateString + " " + str(curYear), "%A, %d %b %Y")
922
            dateMax = datetime.strptime(dateString + " " + str(nextYear), "%A, %d %b %Y")
923
            if dateMin >= curTime:
924
                return dateMin
925
            else:
926
                return dateMax 
927
        except:
928
            try:
929
                days_of_week = ['sunday','monday','tuesday','wednesday',
930
                            'thursday','friday','saturday']
931
                deltaDays = days_of_week.index(dateString.lower()) - curDate.isoweekday()
932
                if deltaDays < 0:
933
                    deltaDays= deltaDays + 7
934
                curDate = curDate + timedelta(days=deltaDays)
935
                return curDate
936
            except:
937
                print "Could not parse"
938
                return None 
939
 
18028 amit.gupta 940
def main1():
941
    store = getStore(1)
942
    for merchantOrder in store.db.merchantOrder.find({"subOrders":{"$elemMatch":{"cashBackStatus":"Not Applicable", "cashBackPercentage":{"$gt":0}}}}):
943
        mo = obj(merchantOrder)
944
        for subOrder in mo.subOrders:
945
            subOrder.closed=False 
946
        print "orderId", mo.orderId
947
        store.populateDerivedFields(mo, False)
948
        store.db.merchantOrder.update({"orderId":mo.orderId}, {"$set":todict(mo)})
949
        break
950
 
951
 
15565 amit.gupta 952
def parseOrderNotCreated():
953
    try:
954
        store=getStore(1)
955
        orders = session.query(OrdersRaw).filter_by(status='ORDER_NOT_CREATED_UNKNOWN').all()
956
        session.close()
957
        for order in orders:
15567 amit.gupta 958
            result = store.parseOrderRawHtml(order.id, order.sub_tag, order.user_id, order.rawhtml, order.order_url)['result']
959
            order1 = session.query(OrdersRaw).filter_by(id=order.id).first()
960
            order1.status = result
961
            session.commit()
15565 amit.gupta 962
    finally:
963
        session.close()
15555 amit.gupta 964
 
13774 amit.gupta 965
if __name__ == '__main__':
17599 amit.gupta 966
#    readSSh("/AmazonTrack/User2942/402-2467356-7564367")
967
#    readSSh("/AmazonTrack/User10466/orderSummary18-11:21:32:36")
18156 amit.gupta 968
#    readSSh("/AmazonTrack/User5525/171-0333104-6169933")
18159 amit.gupta 969
    main()
970
    #store=getStore(1)
971
    #store.trackOrdersForUser(5525, "https://www.amazon.in/gp/css/summary/edit.html?orderID=171-0333104-6169933", readSSh("/AmazonTrack/User5525/171-0333104-6169933"))