Subversion Repositories SmartDukaan

Rev

Rev 14312 | Rev 14319 | 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
14162 amit.gupta 9
from datetime import datetime
13774 amit.gupta 10
from dtr.dao import Order, SubOrder
13995 amit.gupta 11
from dtr.main import getStore, Store as MStore, ParseException
13774 amit.gupta 12
from dtr.sources.flipkart import todict
14162 amit.gupta 13
from paramiko import sftp
14
from paramiko.client import SSHClient
15
from paramiko.sftp_client import SFTPClient
14285 amit.gupta 16
import os.path
14162 amit.gupta 17
import paramiko
13774 amit.gupta 18
import re
14033 amit.gupta 19
import time
13809 amit.gupta 20
import traceback
13774 amit.gupta 21
 
22
ORDER_REDIRECT_URL = 'https://www.amazon.in/gp/css/summary/edit.html?orderID=%s'
13810 amit.gupta 23
ORDER_SUCCESS_URL = 'https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html'
13823 amit.gupta 24
THANKYOU_URL = 'https://www.amazon.in/gp/buy/thankyou/handlers/display.html'
13774 amit.gupta 25
class Store(MStore):
13821 amit.gupta 26
 
14032 amit.gupta 27
    orderStatusRegexMap = { MStore.ORDER_PLACED : ['ordered from', 'not yet dispatched','dispatching now', 'preparing for dispatch'],
14064 amit.gupta 28
                            MStore.ORDER_SHIPPED : ['dispatched on','dispatched', 'on the way', 'out for delivery', 'Out for delivery'],
14032 amit.gupta 29
                            MStore.ORDER_CANCELLED : ['return complete', 'refunded', 'cancelled'],
30
                            MStore.ORDER_DELIVERED : ['delivered']
13821 amit.gupta 31
                           }
13774 amit.gupta 32
 
33
    def __init__(self,store_id):
34
        super(Store, self).__init__(store_id)
35
 
36
    def getName(self):
37
        return "flipkart"
38
 
14202 amit.gupta 39
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, insert=False):
13796 amit.gupta 40
        resp = {}
13821 amit.gupta 41
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
13774 amit.gupta 42
            try:
43
                soup = BeautifulSoup(rawHtml)
44
                orderUrl = soup.find('div', {"id":"thank-you-box-wrapper"}).div.findAll('div', recursive=False)[1].a['href']
14202 amit.gupta 45
                merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
14211 amit.gupta 46
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
47
                order.orderSuccessUrl = ORDER_REDIRECT_URL % (merchantOrderId)
48
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 49
                order.requireDetail = True
14297 amit.gupta 50
                order.closed = None
14312 amit.gupta 51
                if self._saveToOrder(todict(order)):
52
                    resp['result'] = 'ORDER_CREATED'
53
                    resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
54
                    resp["htmlRequired"] = True
55
                    resp['orderId'] = orderId
56
                else:
57
                    resp['result'] = 'ORDER_ALREADY_CREATED_IGNORED'
58
 
13796 amit.gupta 59
                return resp
13774 amit.gupta 60
            except:
14312 amit.gupta 61
                resp["result"] = 'ORDER_NOT_CREATED'
13796 amit.gupta 62
                return resp
13774 amit.gupta 63
 
64
        else:
13781 amit.gupta 65
            try:
66
                merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14145 amit.gupta 67
                merchantOrder.orderTrackingUrl = orderSuccessUrl
13781 amit.gupta 68
                soup = BeautifulSoup(rawHtml,from_encoding="utf-8")
13774 amit.gupta 69
 
13781 amit.gupta 70
                table = soup.body.findAll("table", recursive=False)[1]
71
                #print table
72
                tables = table.tr.td.findAll("table", recursive=False)
73
                for tr in tables[2].findAll("tr"):
74
                    boldElement = tr.td.b
75
                    if "Order Placed" in str(boldElement):
76
                        merchantOrder.placedOn = boldElement.next_sibling.strip()
77
                    if "order number" in str(boldElement):
14023 amit.gupta 78
                        merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
13781 amit.gupta 79
                    if "Order Total" in str(boldElement):
13809 amit.gupta 80
                        merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
13781 amit.gupta 81
                anchors = table.tr.td.findAll("a", recursive=False)
14293 amit.gupta 82
                paymentAnchor = anchors.pop(-1)
13774 amit.gupta 83
 
13809 amit.gupta 84
                count = 0
13781 amit.gupta 85
                subOrders = []
86
                merchantOrder.subOrders = subOrders
87
                for anchor in anchors:
13809 amit.gupta 88
                    count += 1
13781 amit.gupta 89
                    tab = anchor.next_sibling
90
                    status = MStore.ORDER_PLACED
91
                    subStr = "Delivery #" + str(count) + ":"
92
                    if subStr in  tab.find("b").text:
93
                        detailedStatus = tab.find("b").text.replace(subStr, '').strip()
94
 
95
                    tab = tab.next_sibling.next_sibling
13813 amit.gupta 96
                    trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
13781 amit.gupta 97
 
98
                    estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
99
 
13813 amit.gupta 100
                    orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
13781 amit.gupta 101
                    i = -1
102
                    for orderItemTr in orderItemTrs:
103
                        i += 1
104
                        if i%2 == 0:
105
                            continue
13809 amit.gupta 106
                        quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
13781 amit.gupta 107
 
13809 amit.gupta 108
                        productUrl = orderItemTr.td.contents[1].a["href"]
109
                        productTitle = orderItemTr.td.contents[1].a.text
110
 
111
                        unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
112
 
13781 amit.gupta 113
 
13809 amit.gupta 114
                        subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
115
                        subOrder.merchantSubOrderId = orderItemTr.find("input")["name"]
116
                        subOrder.estimatedDeliveryDate = estimatedDelivery
117
                        subOrder.productCode = productUrl.split('/')[5]
118
                        subOrder.detailedStatus = detailedStatus
119
                        (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice*quantity)
120
                        cashbackStatus = Store.CB_PENDING
121
                        if cashbackAmount <= 0:
122
                            cashbackStatus = Store.CB_NA
123
                        subOrder.cashBackStatus = cashbackStatus
124
                        subOrder.cashBackAmount = cashbackAmount
125
                        if percentage > 0:
126
                            subOrder.cashBackPercentage = percentage
127
                        subOrders.append(subOrder)
14293 amit.gupta 128
                priceList = paymentAnchor.next_sibling.next_sibling.next_sibling.table.table.tbody.tbody.tbody.findAll('tr', recursive=False)
129
                totalAmount = 0
130
                grandAmount = 0
131
                for price in priceList:
132
                    labelTd = price.td
133
                    if 'Subtotal:' in labelTd.text:
14314 amit.gupta 134
                        totalAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
14293 amit.gupta 135
                    elif 'Grand Total:' in labelTd.text:
14314 amit.gupta 136
                        grandAmount += int(float(labelTd.next_sibling.next_sibling.find('span').contents[-1].replace(',','')))
14293 amit.gupta 137
                if grandAmount < totalAmount:
138
                    diff = totalAmount - grandAmount
139
                    for subOrder in merchantOrder.subOrders:
14294 amit.gupta 140
                        subOrder.amountPaid -= int(diff*(1-subOrder.amountPaid/totalAmount))
14080 amit.gupta 141
                else:
142
                    self._updateToOrder(todict(merchantOrder))
14312 amit.gupta 143
                    resp['result'] = 'ORDER_CREATED'
144
                    return resp    
13781 amit.gupta 145
            except:
13809 amit.gupta 146
                print "Error occurred"
147
                traceback.print_exc()
14312 amit.gupta 148
                resp['result'] = 'ORDER_NOT_CREATED'
13796 amit.gupta 149
                return resp    
13774 amit.gupta 150
 
13868 amit.gupta 151
    #This should be exposed from api for specific sources
152
    def scrapeStoreOrders(self):
153
        pass
154
 
155
    def getTrackingUrls(self, userId):
14074 amit.gupta 156
 
157
        missingOrderUrls = []
158
        missingOrders = self._getMissingOrders({'userId':userId})
159
        for missingOrder in missingOrders:
160
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
13879 amit.gupta 161
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 162
        count = len(orders)
13879 amit.gupta 163
        print "count", count
13868 amit.gupta 164
        if count > 0:
14074 amit.gupta 165
            return missingOrderUrls + ['https://www.amazon.in/gp/css/order-history', 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled']
13959 amit.gupta 166
        else: 
14074 amit.gupta 167
            return missingOrderUrls
13927 amit.gupta 168
 
169
    def trackOrdersForUser(self, userId, url, rawHtml):
14285 amit.gupta 170
        directory = "/tmp/User" + str(userId)
171
        if not os.path.exists(directory):
172
            os.makedirs(directory)
14298 amit.gupta 173
        filename = directory + "/" + str(datetime.now())
174
        print "filename---", filename
175
        f = open(filename,'w')
14173 amit.gupta 176
        f.write(rawHtml) # python will convert \n to os.linesep
177
        f.close() # you can omit in most cases as the destructor will call if
178
 
13995 amit.gupta 179
        try:
180
            searchMap = {'userId':userId}
181
            collectionMap = {'merchantOrderId':1}
182
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 183
            datetimeNow = datetime.now()
184
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 185
            print "url----------------", url
14008 amit.gupta 186
 
13995 amit.gupta 187
            if url == 'https://www.amazon.in/gp/css/order-history' or url == 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled':
188
                soup = BeautifulSoup(rawHtml)
189
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
190
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
191
                for activeOrder in activeOrders:
192
                    for orderEle in allOrders:
193
                        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'})
194
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
195
                        if merchantOrderId==activeOrder['merchantOrderId']:
196
                            closed = True
197
                            shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
198
                            shipments.pop(0)
199
                            for shipment in shipments:
200
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 201
                                sdivs = shipment.div.div.findAll('div', recursive=False)
14065 amit.gupta 202
                                orderStatus = sdivs[0].span.text.strip()
14050 amit.gupta 203
                                status = self._getStatusFromDetailedStatus(orderStatus)
204
                                if status == MStore.ORDER_DELIVERED:
14299 amit.gupta 205
                                    deliveredOn = sdivs[0].findAll('span')[-1].text.strip()
14050 amit.gupta 206
                                    deliveredOn = deliveredOn.split(":")[1].strip()
207
                                else:
14065 amit.gupta 208
                                    deliveryestimatespan = sdivs[0].find('span', {'class':'a-color-success'})
14050 amit.gupta 209
                                    deliveryEstimate = None
210
                                    if deliveryestimatespan is not None:
211
                                        deliveryEstimate = deliveryestimatespan.find('span', {'class':'a-text-bold'}).text.strip()
212
                                    productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 213
                                trackingUrl = None
214
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
215
                                    if buttonDiv.find('a').text.strip()=='Track package':
216
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
217
                                        if not trackingUrl.startswith("http"):
218
                                            trackingUrl = "https://www.amazon.in/" + trackingUrl
219
                                        break
220
                                for prodDiv in productDivs:
221
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
222
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
223
                                    imgUrl = prodDiv.find("img")["src"]
224
                                    for subOrder in activeOrder['subOrders']:
225
                                        if subOrder['productTitle'] == productTitle:
226
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
227
                                            updateMap = {}
228
                                            closedStatus = False
229
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 230
                                            updateMap['subOrders.$.lastTracked'] = timestamp
13995 amit.gupta 231
                                            updateMap['subOrders.$.detailedStatus'] = orderStatus
232
                                            cashbackStatus = subOrder.get("cashBackStatus")
233
                                            updateMap['subOrders.$.status'] = status 
234
 
235
                                            if status==MStore.ORDER_DELIVERED:                               
14291 amit.gupta 236
                                                updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 237
                                                closedStatus = True
238
                                                updateMap['subOrders.$.closed'] = True
239
                                                if cashbackStatus == Store.CB_PENDING:
240
                                                    updateMap['subOrders.$.cashbackStatus'] = Store.CB_APPROVED
241
                                            if status==MStore.ORDER_CANCELLED:     
242
                                                closedStatus = True
243
                                                updateMap['subOrders.$.closed'] = True
244
                                                if cashbackStatus == Store.CB_PENDING:
245
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
246
                                            if status==MStore.ORDER_SHIPPED:   
247
                                                updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
248
                                                if trackingUrl is not None:
249
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
250
                                            if not closedStatus:
251
                                                closed = False
252
                                            bulk.find(findMap).update({'$set' : updateMap})
253
                                            break
254
                            bulk.find({'orderId': activeOrder['orderId']}).update({'$set':{'closed': closed}})            
14080 amit.gupta 255
                bulk.execute()
256
                return 'PARSED_SUCCESS'
257
            else:
14086 amit.gupta 258
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0] 
14085 amit.gupta 259
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14312 amit.gupta 260
                self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, False)['result']
14080 amit.gupta 261
                return 'PARSED_SUCCESS'
262
                pass
14008 amit.gupta 263
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 264
        except:
265
            traceback.print_exc()    
266
            return 'PARSED_FAILED'
267
 
268
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 269
        if "ordered from" in detailedStatus.lower():
270
            return MStore.ORDER_PLACED 
271
 
13995 amit.gupta 272
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 273
            if detailedStatus.lower() in value:
13995 amit.gupta 274
                return key
275
 
276
        print "Detailed Status need to be mapped"
14062 amit.gupta 277
        print "Found new order status", detailedStatus
13995 amit.gupta 278
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
13927 amit.gupta 279
 
280
 
13774 amit.gupta 281
def main():
282
    store = getStore(1)
283
 
14299 amit.gupta 284
    #store.parseOrderRawHtml(1, 'saad', '2123221', readSSh('/tmp/User8/2015-03-02 12:51:59.545557'), 'https://www.amazon.in/gp/css/summary/edit.html?orderID=12212')
285
    store.trackOrdersForUser(8,'https://www.amazon.in/gp/css/order-history', readSSh('/tmp/User8/2015-03-02 12:51:59.545557'))
13774 amit.gupta 286
 
14162 amit.gupta 287
def readSSh(fileName):
288
    try:
289
            str1 = open(fileName).read()
290
            return str1
291
    except:
292
        ssh_client = SSHClient()
293
        str1 = ""
294
        ssh_client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
295
        ssh_client.connect('104.200.25.40', 22, 'root', 'ecip$dtrMay2014')
296
        sftp_client = ssh_client.open_sftp()
297
        try:
14299 amit.gupta 298
            if not os.path.exists(os.path.dirname(fileName)):
299
                os.makedirs(os.path.dirname(fileName))
14162 amit.gupta 300
            sftp_client.get(fileName, fileName)
301
            try:
302
                str1 = open(fileName).read()
303
                return str1
304
            finally:
305
                pass
306
        except:
307
            "could not read"
308
        return str1
13774 amit.gupta 309
 
310
 
311
if __name__ == '__main__':
14284 amit.gupta 312
        main()