Subversion Repositories SmartDukaan

Rev

Rev 14285 | Rev 14293 | 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]
13796 amit.gupta 46
                resp["result"] = 'HTML_REQUIRED'
14211 amit.gupta 47
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
48
                order.orderSuccessUrl = ORDER_REDIRECT_URL % (merchantOrderId)
49
                order.merchantOrderId = merchantOrderId
14291 amit.gupta 50
                order.requireDetail = True
14211 amit.gupta 51
                self._saveToOrder(todict(order))
14202 amit.gupta 52
                resp["url"] = ORDER_REDIRECT_URL % (merchantOrderId)
13796 amit.gupta 53
                return resp
13774 amit.gupta 54
            except:
13796 amit.gupta 55
                resp["result"] = 'IGNORED'
56
                return resp
13774 amit.gupta 57
 
58
        else:
13781 amit.gupta 59
            try:
60
                merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14145 amit.gupta 61
                merchantOrder.orderTrackingUrl = orderSuccessUrl
13781 amit.gupta 62
                soup = BeautifulSoup(rawHtml,from_encoding="utf-8")
13774 amit.gupta 63
 
13781 amit.gupta 64
                table = soup.body.findAll("table", recursive=False)[1]
65
                #print table
66
                tables = table.tr.td.findAll("table", recursive=False)
67
                for tr in tables[2].findAll("tr"):
68
                    boldElement = tr.td.b
69
                    if "Order Placed" in str(boldElement):
70
                        merchantOrder.placedOn = boldElement.next_sibling.strip()
71
                    if "order number" in str(boldElement):
14023 amit.gupta 72
                        merchantOrder.merchantOrderId = boldElement.next_sibling.strip()
13781 amit.gupta 73
                    if "Order Total" in str(boldElement):
13809 amit.gupta 74
                        merchantOrder.paidAmount = int(float(boldElement.find('span').contents[-1].replace(',','')))
13781 amit.gupta 75
                anchors = table.tr.td.findAll("a", recursive=False)
76
                anchors.pop(-1)
13774 amit.gupta 77
 
13809 amit.gupta 78
                count = 0
13781 amit.gupta 79
                subOrders = []
80
                merchantOrder.subOrders = subOrders
81
                for anchor in anchors:
13809 amit.gupta 82
                    count += 1
13781 amit.gupta 83
                    tab = anchor.next_sibling
84
                    status = MStore.ORDER_PLACED
85
                    subStr = "Delivery #" + str(count) + ":"
86
                    if subStr in  tab.find("b").text:
87
                        detailedStatus = tab.find("b").text.replace(subStr, '').strip()
88
 
89
                    tab = tab.next_sibling.next_sibling
13813 amit.gupta 90
                    trs = tab.find("table").find('tbody').findAll("tr", recursive = False)
13781 amit.gupta 91
 
92
                    estimatedDelivery = trs[0].td.find("b").next_sibling.strip()
93
 
13813 amit.gupta 94
                    orderItemTrs = trs[1].findAll("td", recursive=False)[1].table.tbody.findAll("tr", recursive = False)
13781 amit.gupta 95
                    i = -1
96
                    for orderItemTr in orderItemTrs:
97
                        i += 1
98
                        if i%2 == 0:
99
                            continue
13809 amit.gupta 100
                        quantity =  int(re.findall(r'\d+', orderItemTr.td.contents[0])[0])
13781 amit.gupta 101
 
13809 amit.gupta 102
                        productUrl = orderItemTr.td.contents[1].a["href"]
103
                        productTitle = orderItemTr.td.contents[1].a.text
104
 
105
                        unitPrice = int(float(orderItemTr.findAll('td')[1].span.text.replace('Rs. ','').replace(',','')))
106
 
13781 amit.gupta 107
 
13809 amit.gupta 108
                        subOrder = SubOrder(productTitle, productUrl, merchantOrder.placedOn, unitPrice*quantity, status, quantity)
109
                        subOrder.merchantSubOrderId = orderItemTr.find("input")["name"]
110
                        subOrder.estimatedDeliveryDate = estimatedDelivery
111
                        subOrder.productCode = productUrl.split('/')[5]
112
                        subOrder.detailedStatus = detailedStatus
113
                        (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, unitPrice*quantity)
114
                        cashbackStatus = Store.CB_PENDING
115
                        if cashbackAmount <= 0:
116
                            cashbackStatus = Store.CB_NA
117
                        subOrder.cashBackStatus = cashbackStatus
118
                        subOrder.cashBackAmount = cashbackAmount
119
                        if percentage > 0:
120
                            subOrder.cashBackPercentage = percentage
121
                        subOrders.append(subOrder)
14080 amit.gupta 122
                if insert:
123
                    self._saveToOrder(todict(merchantOrder))
124
                else:
125
                    self._updateToOrder(todict(merchantOrder))
13796 amit.gupta 126
                resp['result'] = 'ORDER_CREATED'
127
                return resp    
13781 amit.gupta 128
            except:
13809 amit.gupta 129
                print "Error occurred"
130
                traceback.print_exc()
13796 amit.gupta 131
                resp['result'] = 'PARSE_ERROR'
132
                return resp    
13774 amit.gupta 133
 
13868 amit.gupta 134
    #This should be exposed from api for specific sources
135
    def scrapeStoreOrders(self):
136
        pass
137
 
138
    def getTrackingUrls(self, userId):
14074 amit.gupta 139
 
140
        missingOrderUrls = []
141
        missingOrders = self._getMissingOrders({'userId':userId})
142
        for missingOrder in missingOrders:
143
            missingOrderUrls.append(ORDER_REDIRECT_URL%(missingOrder['merchantOrderId']))
13879 amit.gupta 144
        orders = self._getActiveOrders({'userId':userId})
13882 amit.gupta 145
        count = len(orders)
13879 amit.gupta 146
        print "count", count
13868 amit.gupta 147
        if count > 0:
14074 amit.gupta 148
            return missingOrderUrls + ['https://www.amazon.in/gp/css/order-history', 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled']
13959 amit.gupta 149
        else: 
14074 amit.gupta 150
            return missingOrderUrls
13927 amit.gupta 151
 
152
    def trackOrdersForUser(self, userId, url, rawHtml):
14285 amit.gupta 153
        directory = "/tmp/User" + str(userId)
154
        if not os.path.exists(directory):
155
            os.makedirs(directory)
156
        f = open(directory + "/" + str(datetime.now()),'w')
14173 amit.gupta 157
        f.write(rawHtml) # python will convert \n to os.linesep
158
        f.close() # you can omit in most cases as the destructor will call if
159
 
13995 amit.gupta 160
        try:
161
            searchMap = {'userId':userId}
162
            collectionMap = {'merchantOrderId':1}
163
            activeOrders = self._getActiveOrders(searchMap, collectionMap)
14033 amit.gupta 164
            datetimeNow = datetime.now()
165
            timestamp = int(time.mktime(datetimeNow.timetuple()))
14009 amit.gupta 166
            print "url----------------", url
14008 amit.gupta 167
 
13995 amit.gupta 168
            if url == 'https://www.amazon.in/gp/css/order-history' or url == 'https://www.amazon.in/gp/css/order-history/?orderFilter=cancelled':
169
                soup = BeautifulSoup(rawHtml)
170
                allOrders = soup.find(id="ordersContainer").findAll('div', {'class':'a-box-group a-spacing-base order'})
171
                bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
172
                for activeOrder in activeOrders:
173
                    for orderEle in allOrders:
174
                        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'})
175
                        merchantOrderId = orderdiv.find('span', {'class':'a-color-secondary value'}).text.strip()
176
                        if merchantOrderId==activeOrder['merchantOrderId']:
177
                            closed = True
178
                            shipments = orderEle.findAll('div',{'class':re.compile('.*?a-box.*?')}, recursive=False)
179
                            shipments.pop(0)
180
                            for shipment in shipments:
181
                                shipdiv = shipment.find('div', {'class':'a-box-inner'})
14050 amit.gupta 182
                                sdivs = shipment.div.div.findAll('div', recursive=False)
14065 amit.gupta 183
                                orderStatus = sdivs[0].span.text.strip()
14050 amit.gupta 184
                                status = self._getStatusFromDetailedStatus(orderStatus)
185
                                if status == MStore.ORDER_DELIVERED:
186
                                    deliveredOn = sdivs[1].find('span').text
187
                                    deliveredOn = deliveredOn.split(":")[1].strip()
188
                                else:
14065 amit.gupta 189
                                    deliveryestimatespan = sdivs[0].find('span', {'class':'a-color-success'})
14050 amit.gupta 190
                                    deliveryEstimate = None
191
                                    if deliveryestimatespan is not None:
192
                                        deliveryEstimate = deliveryestimatespan.find('span', {'class':'a-text-bold'}).text.strip()
193
                                    productDivs = shipdiv.find('div', {'class':re.compile('.*?a-spacing-top-medium.*?')}).find('div', {'class':'a-row'}).findAll('div', recursive=False)
13995 amit.gupta 194
                                trackingUrl = None
195
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
196
                                    if buttonDiv.find('a').text.strip()=='Track package':
197
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
198
                                        if not trackingUrl.startswith("http"):
199
                                            trackingUrl = "https://www.amazon.in/" + trackingUrl
200
                                        break
201
                                for prodDiv in productDivs:
202
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
203
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
204
                                    imgUrl = prodDiv.find("img")["src"]
205
                                    for subOrder in activeOrder['subOrders']:
206
                                        if subOrder['productTitle'] == productTitle:
207
                                            findMap = {"orderId": activeOrder['orderId'], "subOrders.merchantSubOrderId": subOrder.get("merchantSubOrderId")}
208
                                            updateMap = {}
209
                                            closedStatus = False
210
                                            updateMap['subOrders.$.imgUrl'] = imgUrl
14033 amit.gupta 211
                                            updateMap['subOrders.$.lastTracked'] = timestamp
13995 amit.gupta 212
                                            updateMap['subOrders.$.detailedStatus'] = orderStatus
213
                                            cashbackStatus = subOrder.get("cashBackStatus")
214
                                            updateMap['subOrders.$.status'] = status 
215
 
216
                                            if status==MStore.ORDER_DELIVERED:                               
14291 amit.gupta 217
                                                updateMap['subOrders.$.deliveredOn'] = deliveredOn
13995 amit.gupta 218
                                                closedStatus = True
219
                                                updateMap['subOrders.$.closed'] = True
220
                                                if cashbackStatus == Store.CB_PENDING:
221
                                                    updateMap['subOrders.$.cashbackStatus'] = Store.CB_APPROVED
222
                                            if status==MStore.ORDER_CANCELLED:     
223
                                                closedStatus = True
224
                                                updateMap['subOrders.$.closed'] = True
225
                                                if cashbackStatus == Store.CB_PENDING:
226
                                                    updateMap['subOrders.$.cashBackStatus'] = Store.CB_CANCELLED
227
                                            if status==MStore.ORDER_SHIPPED:   
228
                                                updateMap['subOrders.$.estimatedDeliveryDate'] = deliveryEstimate
229
                                                if trackingUrl is not None:
230
                                                    updateMap['subOrders.$.trackingUrl'] = trackingUrl
231
                                            if not closedStatus:
232
                                                closed = False
233
                                            bulk.find(findMap).update({'$set' : updateMap})
234
                                            break
235
                            bulk.find({'orderId': activeOrder['orderId']}).update({'$set':{'closed': closed}})            
14080 amit.gupta 236
                bulk.execute()
237
                return 'PARSED_SUCCESS'
238
            else:
14086 amit.gupta 239
                merchantOrderId = re.findall(r'https://www.amazon.in/gp/css/summary/edit.html\?orderID=(.*)?', url, re.IGNORECASE)[0] 
14085 amit.gupta 240
                merchantOrder = self.db.merchantOrder.find_one({"merchantOrderId":merchantOrderId})
14080 amit.gupta 241
                self.parseOrderRawHtml(merchantOrder['orderId'], merchantOrder['subTagId'], merchantOrder['userId'], rawHtml, url, False)
242
                return 'PARSED_SUCCESS'
243
                pass
14008 amit.gupta 244
            return 'PARSED_SUCCESS_NO_ORDERS'
13995 amit.gupta 245
        except:
246
            traceback.print_exc()    
247
            return 'PARSED_FAILED'
248
 
249
    def _getStatusFromDetailedStatus(self, detailedStatus):
14061 amit.gupta 250
        if "ordered from" in detailedStatus.lower():
251
            return MStore.ORDER_PLACED 
252
 
13995 amit.gupta 253
        for key, value in self.orderStatusRegexMap.iteritems():
14032 amit.gupta 254
            if detailedStatus.lower() in value:
13995 amit.gupta 255
                return key
256
 
257
        print "Detailed Status need to be mapped"
14062 amit.gupta 258
        print "Found new order status", detailedStatus
13995 amit.gupta 259
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
13927 amit.gupta 260
 
261
 
13774 amit.gupta 262
def main():
263
    store = getStore(1)
264
 
14162 amit.gupta 265
    #store.parseOrderRawHtml(1, 'saad', '212321', htmlstr, 'https://www.amazon.in/gp/css/summary/edit.html?orderID=12212')
266
    store.trackOrdersForUser(2,'https://www.amazon.in/gp/css/order-history', readSSh('/tmp/order2015-02-26 11:20:18.816942'))
13774 amit.gupta 267
 
14162 amit.gupta 268
def readSSh(fileName):
269
    try:
270
            str1 = open(fileName).read()
271
            return str1
272
    except:
273
        ssh_client = SSHClient()
274
        str1 = ""
275
        ssh_client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
276
        ssh_client.connect('104.200.25.40', 22, 'root', 'ecip$dtrMay2014')
277
        sftp_client = ssh_client.open_sftp()
278
        try:
279
            sftp_client.get(fileName, fileName)
280
            try:
281
                str1 = open(fileName).read()
282
                return str1
283
            finally:
284
                pass
285
        except:
286
            "could not read"
287
        return str1
13774 amit.gupta 288
 
289
 
290
if __name__ == '__main__':
14284 amit.gupta 291
        main()