Subversion Repositories SmartDukaan

Rev

Rev 13809 | Rev 14145 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
13569 amit.gupta 1
'''
2
Created on Jan 15, 2015
3
 
4
@author: amit
5
'''
13576 amit.gupta 6
from BeautifulSoup import BeautifulSoup
13569 amit.gupta 7
from bson.binary import Binary
13680 amit.gupta 8
from datetime import datetime, date, timedelta
13569 amit.gupta 9
from dtr import main
13576 amit.gupta 10
from dtr.dao import AffiliateInfo, Order, SubOrder
13680 amit.gupta 11
from dtr.main import getBrowserObject, ScrapeException, getStore, ParseException, Store as MStore,\
12
    ungzipResponse
13662 amit.gupta 13
from pprint import pprint
13569 amit.gupta 14
from pymongo import MongoClient
15
import json
16
import pymongo
17
import re
13662 amit.gupta 18
import traceback
13569 amit.gupta 19
import urllib
13603 amit.gupta 20
 
13721 amit.gupta 21
USERNAME='profittill2@gmail.com'
13569 amit.gupta 22
PASSWORD='spice@2020'
23
AFFILIATE_URL='http://affiliate.snapdeal.com'
24
POST_URL='https://api-p03.hasoffers.com/v3/Affiliate_Report.json'
25
ORDER_TRACK_URL='https://m.snapdeal.com/orderSummary'
26
CONFIG_URL='http://affiliate.snapdeal.com/publisher/js/config.php'
27
 
13662 amit.gupta 28
class Store(MStore):
13569 amit.gupta 29
 
30
    '''
31
    This is to map order statuses of our system to order statuses of snapdeal.
32
    And our statuses will change accordingly.
33
 
34
    '''
35
    OrderStatusMap = {
13662 amit.gupta 36
                      MStore.ORDER_PLACED : ['In Progress','N/A'],
37
                      MStore.ORDER_DELIVERED : ['Delivered'],
38
                      MStore.ORDER_SHIPPED : ['In Transit'],
13809 amit.gupta 39
                      MStore.ORDER_CANCELLED : ['Closed For Vendor Reallocation', 'Cancelled', 'Product returned by courier', 'Returned']
13569 amit.gupta 40
                      }
13662 amit.gupta 41
 
42
    CONF_CB_AMOUNT = MStore.CONF_CB_DISCOUNTED_PRICE
13569 amit.gupta 43
    def __init__(self,store_id):
44
        super(Store, self).__init__(store_id)
45
 
46
    def getName(self):
47
        return "snapdeal"
48
 
49
    def scrapeAffiliate(self, startDate=None, endDate=None):
50
        br = getBrowserObject()
51
        br.open(AFFILIATE_URL)
52
        br.select_form(nr=0)
53
        br.form['data[User][password]'] = PASSWORD 
54
        br.form['data[User][email]'] = USERNAME
55
        br.submit()
56
        response = br.open(CONFIG_URL)
57
 
13680 amit.gupta 58
        token =  re.findall('"session_token":"(.*?)"', ungzipResponse(response), re.IGNORECASE)[0]
13569 amit.gupta 59
 
60
        allOffers = self._getAllOffers(br, token)
61
 
13662 amit.gupta 62
        allPyOffers = []
63
        maxSaleDate = self._getLastSaleDate()
64
        newMaxSaleDate = maxSaleDate
65
        for offer in allOffers:
13680 amit.gupta 66
            pyOffer = self.covertToObj(offer).__dict__
67
            allPyOffers.append(pyOffer)
68
            saleDate = datetime.strptime(pyOffer['saleDate'],"%Y-%m-%d %H:%M:%S")
13662 amit.gupta 69
            if maxSaleDate < saleDate:
13721 amit.gupta 70
                self._updateOrdersPayBackStatus({'subTagId':pyOffer['subTagId'], 'saleDate':pyOffer['saleDate']}, {})
13662 amit.gupta 71
                if newMaxSaleDate < saleDate:
72
                    newMaxSaleDate = saleDate
73
 
74
        self._setLastSaleDate(newMaxSaleDate)
13569 amit.gupta 75
        self._saveToAffiliate(allPyOffers)
76
 
13662 amit.gupta 77
    def _setLastSaleDate(self, saleDate):
13680 amit.gupta 78
        self.db.lastSaleDtate.update({'storeId':self.store_id}, {'$set':{'saleDate':saleDate}})
13569 amit.gupta 79
 
13662 amit.gupta 80
 
81
 
82
    def _getLastSaleDate(self,):
83
        lastDaySaleObj = self.db.lastDaySale.find_one({"storeId":self.store_id})
84
        if lastDaySaleObj is None:
85
            return datetime.min
86
 
13760 amit.gupta 87
    def _parse(self, orderId, subTagId, userId, page, orderSuccessUrl):
13662 amit.gupta 88
 
13760 amit.gupta 89
        #page=page.decode("utf-8")
90
        soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
91
        #orderHead = soup.find(name, attrs, recursive, text)
92
        sections = soup.findAll("section")
93
 
94
        #print sections
95
 
96
        order = sections[1]
97
        orderTrs = order.findAll("tr")
98
 
99
        placedOn = str(orderTrs[0].findAll("td")[1].text)
100
 
101
        #Pop two section elements
102
        sections.pop(0) 
103
        sections.pop(0)
104
        subOrders = sections
105
 
106
 
107
        merchantSubOrders = []
108
 
109
        merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
14023 amit.gupta 110
        merchantOrder.merchantOrderId = re.findall(r'\d+', str(soup.find("div", {"class":"deals_heading"})))[1]
13760 amit.gupta 111
        for orderTr in orderTrs:
112
            orderTrString = str(orderTr)
113
            if "Total Amount" in orderTrString:
114
                merchantOrder.totalAmount = re.findall(r'\d+', orderTrString)[0]
115
            elif "Delivery Charges" in orderTrString:
116
                merchantOrder.deliveryCharges = re.findall(r'\d+', orderTrString)[0]
117
            elif "Discount Applied" in orderTrString:
118
                merchantOrder.discountApplied = re.findall(r'\d+', orderTrString)[0]
119
            elif "Paid Amount" in orderTrString:
120
                merchantOrder.paidAmount = re.findall(r'\d+', orderTrString)[0]
121
 
122
        for subOrderElement in subOrders:
13809 amit.gupta 123
            subOrders = self.parseSubOrder(subOrderElement, placedOn)                           
124
            merchantSubOrders.extend(subOrders)   
13760 amit.gupta 125
 
126
        merchantOrder.subOrders = merchantSubOrders
127
        return merchantOrder
128
 
129
    def parseSubOrder(self, subOrderElement, placedOn):
13809 amit.gupta 130
        subOrders = []
13760 amit.gupta 131
        productUrl = str(subOrderElement.find("a")['href'])
132
        subTable = subOrderElement.find("table", {"class":"lrPad"})
133
        subTrs = subTable.findAll("tr")
134
        unitPrice=None
135
        offerDiscount = None
136
        deliveryCharges = None
137
        amountPaid = None
138
        for subTr in subTrs:
139
            subTrString = str(subTr)
140
            if "Unit Price" in subTrString:
141
                unitPrice = re.findall(r'\d+', subTrString)[0]
142
            if "Quantity" in subTrString:
143
                qty = re.findall(r'\d+', subTrString)[0]
144
            elif "Offer Discount" in subTrString:
145
                offerDiscount =   re.findall(r'\d+', subTrString)[0]
146
            elif "Delivery Charges" in subTrString:
147
                deliveryCharges =   re.findall(r'\d+', subTrString)[0]
148
            elif "Subtotal" in subTrString:
149
                if int(qty) > 0:
150
                    amountPaid =   str(int(re.findall(r'\d+', subTrString)[0])/int(qty))
151
                else:
152
                    amountPaid =   "0"
153
        if self.CONF_CB_AMOUNT == MStore.CONF_CB_SELLING_PRICE or offerDiscount is None:
154
            amount = int(unitPrice)
155
        else:
156
            amount = int(unitPrice) - int(offerDiscount)
157
 
158
        divs = subOrderElement.findAll("div", {"class": "blk lrPad subordrs"})
159
        if len(divs)<=0:
160
            raise ParseException("subOrder", "Could not Parse suborders for Snapdeal")
161
 
162
        for div in divs:
163
            productTitle = str(subOrderElement.find("a").text)
164
            productUrl = "http://m.snapdeal.com/" + productUrl 
165
            subOrder = SubOrder(productTitle, productUrl, placedOn, amountPaid)
166
 
167
            subOrder.amountPaid = amountPaid
168
            subOrder.deliveryCharges = deliveryCharges
169
            subOrder.offerDiscount = offerDiscount
13809 amit.gupta 170
            subOrder.unitPrice = int(unitPrice)
13760 amit.gupta 171
            subOrder.productCode = re.findall(r'\d+$', productUrl)[0]
172
            (cashbackAmount, percentage) = self.getCashbackAmount(subOrder.productCode, amount)
13770 amit.gupta 173
            cashbackStatus = Store.CB_PENDING
13760 amit.gupta 174
            if cashbackAmount <= 0:
175
                cashbackStatus = Store.CB_NA
176
            subOrder.cashBackStatus = cashbackStatus
177
            subOrder.cashBackAmount = cashbackAmount
178
            if percentage > 0:
179
                subOrder.cashBackPercentage = percentage
180
 
181
 
182
            trackAnchor = div.find("a")   
183
            if trackAnchor is not None:
184
                subOrder.tracingkUrl = str(trackAnchor['href'])
185
 
186
            divStr = str(div)
187
            divStr = divStr.replace("\n","").replace("\t", "")
188
 
189
            for line in divStr.split("<br />"):
190
                if "Suborder ID" in line:
191
                    subOrder.merchantSubOrderId = re.findall(r'\d+', line)[0]   
192
                elif "Status" in line:
193
                    subOrder.detailedStatus = re.findall('>(.*?)</span>', line, re.IGNORECASE)[0]
194
                elif "Est. Shipping Date" in line:
195
                    subOrder.estimatedShippingDate = line.split(":")[1].strip()
196
                elif "Est. Delivery Date" in line:
197
                    subOrder.estimatedDeliveryDate = line.split(":")[1].strip()
198
                elif "Courier Name" in line:
199
                    subOrder.courierName = line.split(":")[1].strip()
200
                elif "Tracking No" in line:
201
                    subOrder.trackingNumber = line.split(":")[1].strip()
13809 amit.gupta 202
            subOrders.append(subOrder)
203
        return subOrders
13760 amit.gupta 204
 
13576 amit.gupta 205
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl):
13760 amit.gupta 206
                    #print merchantOrder
13796 amit.gupta 207
        resp = {}
13582 amit.gupta 208
        try:
209
            br = getBrowserObject()
210
            url = ORDER_TRACK_URL + re.findall('.*(\?.*?)$', orderSuccessUrl,re.IGNORECASE)[0]
13791 amit.gupta 211
            page = br.open(url)
212
            page = ungzipResponse(page)
13760 amit.gupta 213
            merchantOrder = self._parse(orderId, subTagId, userId, page, orderSuccessUrl)   
13582 amit.gupta 214
            self._saveToOrder(todict(merchantOrder))
13796 amit.gupta 215
            resp['result'] = 'ORDER_CREATED'
216
            return resp
13582 amit.gupta 217
        except:
218
            print "Error occurred"
13603 amit.gupta 219
            traceback.print_exc()
13796 amit.gupta 220
            resp['result'] = 'PARSE_ERROR'
221
            return resp
13781 amit.gupta 222
 
13569 amit.gupta 223
 
224
        #soup = BeautifulSoup(rawHtml,convertEntities=BeautifulSoup.HTML_ENTITIES)
225
        #soup.find(name, attrs, recursive, text)
13576 amit.gupta 226
 
227
    def _getStatusFromDetailedStatus(self, detailedStatus):
228
        for key, value in Store.OrderStatusMap.iteritems():
229
            if detailedStatus in value:
230
                return key
13662 amit.gupta 231
            print "Detailed Status need to be mapped"
13576 amit.gupta 232
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
13569 amit.gupta 233
 
13610 amit.gupta 234
 
13569 amit.gupta 235
    def scrapeStoreOrders(self,):
13760 amit.gupta 236
        #collectionMap = {'palcedOn':1}
13576 amit.gupta 237
        orders = self._getActiveOrders()
13730 amit.gupta 238
        print "Found orders", orders
13576 amit.gupta 239
        br = getBrowserObject()
240
        for order in orders:
241
            url = ORDER_TRACK_URL + re.findall('.*(\?.*?)$', order['orderSuccessUrl'],re.IGNORECASE)[0]
242
            response = br.open(url)
13680 amit.gupta 243
            page = ungzipResponse(response)
13576 amit.gupta 244
            #page=page.decode("utf-8")
245
            soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
246
            sections = soup.findAll("section")
13760 amit.gupta 247
            orderEl = sections[1]
248
            orderTrs = orderEl.findAll("tr")
249
 
250
            placedOn = str(orderTrs[0].findAll("td")[1].text)
13576 amit.gupta 251
            sections.pop(0)
252
            sections.pop(0)
253
 
254
            subOrders = sections
255
            bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
256
            for subOrderElement in subOrders:
257
                closed = True
258
                divs = subOrderElement.findAll("div", {"class": "blk lrPad subordrs"})
259
                if len(divs)<=0:
260
                    raise ParseException("subOrder", "Could not Parse suborders for Snapdeal")
13721 amit.gupta 261
                subOrder = None
13760 amit.gupta 262
                breakFlag = False
13576 amit.gupta 263
                for div in divs:
264
                    divStr = str(div)
265
                    divStr = divStr.replace("\n","").replace("\t", "")
266
                    updateMap = {}
267
                    for line in divStr.split("<br />"):
268
                        if "Suborder ID" in line:
13634 amit.gupta 269
                            merchantSubOrderId = re.findall(r'\d+', line)[0]
270
                            #break if suborder is inactive   
13721 amit.gupta 271
                            subOrder =  self._isSubOrderActive(order, merchantSubOrderId)
272
                            if subOrder is None:
13809 amit.gupta 273
                                subOrders = self.parseSubOrder(subOrderElement, placedOn)
274
                                self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":todict(subOrders)}})
13760 amit.gupta 275
                                print "Added new suborder with subOrder Id:", subOrder.merchantSubOrderId
276
                                closed = False
277
                                return
278
                            elif subOrder['closed']:
279
                                breakFlag = True
13634 amit.gupta 280
                                break
13760 amit.gupta 281
                            else: 
282
                                findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": merchantSubOrderId}
13576 amit.gupta 283
                        elif "Status" in line:
284
                            detailedStatus = re.findall('>(.*?)</span>', line, re.IGNORECASE)[0]
285
                            updateMap["subOrders.$.detailedStatus"] = detailedStatus
286
                            status = self._getStatusFromDetailedStatus(detailedStatus) 
13634 amit.gupta 287
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
13760 amit.gupta 288
                            updateMap["subOrders.$.status"] = status
289
                            if detailedStatus == 'Closed For Vendor Reallocation':
13770 amit.gupta 290
                                #if it is more than 6hours mark closed.
13760 amit.gupta 291
                                closeAt = subOrder.get("closeAt") 
292
                                if closeAt is None:
293
                                    closeAt = datetime.now() + timedelta(hours=6)
13770 amit.gupta 294
                                    updateMap["subOrders.$.closeAt"] = datetime.strftime(closeAt,"%Y-%m-%d %H:%M:%S")
13760 amit.gupta 295
                                else:
13770 amit.gupta 296
                                    closeAt = datetime.strptime(closeAt,"%Y-%m-%d %H:%M:%S")
13760 amit.gupta 297
                                    if datetime.now() > closeAt:
298
                                        closedStatus = True
299
 
300
 
13634 amit.gupta 301
                            if closedStatus:
302
                                #if status is closed then change the paybackStatus accordingly
13760 amit.gupta 303
                                updateMap["subOrders.$.closed"] = True
304
                                if status == Store.ORDER_DELIVERED:
13721 amit.gupta 305
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
13634 amit.gupta 306
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
13760 amit.gupta 307
                                elif status == Store.ORDER_CANCELLED:
13721 amit.gupta 308
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
13634 amit.gupta 309
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
310
 
311
                            else:
312
                                closed = False
13576 amit.gupta 313
                        elif "Est. Shipping Date" in line:
314
                            estimatedShippingDate = line.split(":")[1].strip()
315
                            updateMap["subOrders.$.estimatedShippingDate"] = estimatedShippingDate
316
                        elif "Est. Delivery Date" in line:
317
                            estimatedDeliveryDate = line.split(":")[1].strip()
318
                            updateMap["subOrders.$.estimatedDeliveryDate"] = estimatedDeliveryDate
319
                        elif "Courier Name" in line:
320
                            courierName = line.split(":")[1].strip()
321
                            updateMap["subOrders.$.courierName"] = courierName
322
                        elif "Tracking No" in line:
323
                            trackingNumber = line.split(":")[1].strip()
324
                            updateMap["subOrders.$.trackingNumber"] = trackingNumber
13760 amit.gupta 325
 
326
                    if breakFlag:
327
                        break
13576 amit.gupta 328
 
329
                    bulk.find(findMap).update({'$set' : updateMap})
13721 amit.gupta 330
                bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed}})
13576 amit.gupta 331
            result = bulk.execute()
332
            pprint(result)        
333
 
334
 
13569 amit.gupta 335
    def _saveToAffiliate(self, offers):
13725 amit.gupta 336
        if offers is None or len(offers)==0:
337
            print "no affiliate have been pushed"
338
            return
13576 amit.gupta 339
        collection = self.db.snapdealOrderAffiliateInfo
13569 amit.gupta 340
        try:
341
            collection.insert(offers,continue_on_error=True)
342
        except pymongo.errors.DuplicateKeyError as e:
343
            print e.details
344
 
345
 
346
    def _getAllOffers(self, br, token):
347
        allOffers = []
348
        nextPage = 1  
349
        while True:
350
            data = getPostData(token, nextPage)
351
            response = br.open(POST_URL, data)
13680 amit.gupta 352
            rmap = json.loads(ungzipResponse(response))
13569 amit.gupta 353
            if rmap is not None:
354
                rmap = rmap['response']
355
                if rmap is not None and len(rmap['errors'])==0:
356
                    allOffers += rmap['data']['data']
357
            nextPage += 1
358
            if rmap['data']['pageCount']<nextPage:
359
                break
360
 
361
        return allOffers
362
 
363
    def covertToObj(self,offer):
364
        offerData = offer['Stat']
365
        offer1 = AffiliateInfo(offerData['affiliate_info1'], self.store_id, offerData['conversion_status'], offerData['ad_id'], 
366
                              offerData['datetime'], offerData['payout'], offer['Offer']['name'], offerData['ip'], offerData['conversion_sale_amount'])
13680 amit.gupta 367
 
13569 amit.gupta 368
        return offer1
369
def getPostData(token, page = 1, limit= 20, startDate=None, endDate=None):
13680 amit.gupta 370
    endDate=date.today() + timedelta(days=1)
371
    startDate=endDate - timedelta(days=31)
13569 amit.gupta 372
 
373
    parameters = (
374
        ("page",str(page)),
375
        ("limit",str(limit)),
376
        ("fields[]","Stat.offer_id"),
377
        ("fields[]","Stat.datetime"),
378
        ("fields[]","Offer.name"),
379
        ("fields[]","Stat.conversion_status"),
380
        ("fields[]","Stat.conversion_sale_amount"),
381
        ("fields[]","Stat.payout"),
382
        ("fields[]","Stat.ip"),
383
        ("fields[]","Stat.ad_id"),
384
        ("fields[]","Stat.affiliate_info1"),
385
        ("sort[Stat.datetime]","desc"),
386
        ("filters[Stat.date][conditional]","BETWEEN"),
387
        ("filters[Stat.date][values][]",startDate.strftime('%Y-%m-%d')),
388
        ("filters[Stat.date][values][]",endDate.strftime('%Y-%m-%d')),
389
        ("data_start",startDate.strftime('%Y-%m-%d')),
390
        ("data_end",endDate.strftime('%Y-%m-%d')),
391
        ("Method","getConversions"),
392
        ("NetworkId","jasper"),
393
        ("SessionToken",token),
394
    )
395
    #Encode the parameters
396
    return urllib.urlencode(parameters)
397
 
398
def main():
13634 amit.gupta 399
 
13569 amit.gupta 400
    store = getStore(3)
13760 amit.gupta 401
    store.scrapeStoreOrders()
13662 amit.gupta 402
    #store._isSubOrderActive(8, "5970688907")
13760 amit.gupta 403
    #store.scrapeAffiliate()
13576 amit.gupta 404
    #store.parseOrderRawHtml(12345, "subtagId", 122323,  "html", 'https://m.snapdeal.com/purchaseMobileComplete?code=1f4166d13ea799b65aa9dea68b3e9e70&order=4509499363')
13569 amit.gupta 405
 
406
 
407
if __name__ == '__main__':
408
    main()
13576 amit.gupta 409
 
410
def todict(obj, classkey=None):
411
    if isinstance(obj, dict):
412
        data = {}
413
        for (k, v) in obj.items():
414
            data[k] = todict(v, classkey)
415
        return data
416
    elif hasattr(obj, "_ast"):
417
        return todict(obj._ast())
418
    elif hasattr(obj, "__iter__"):
419
        return [todict(v, classkey) for v in obj]
420
    elif hasattr(obj, "__dict__"):
421
        data = dict([(key, todict(value, classkey)) 
422
            for key, value in obj.__dict__.iteritems() 
423
            if not callable(value) and not key.startswith('_')])
424
        if classkey is not None and hasattr(obj, "__class__"):
425
            data[classkey] = obj.__class__.__name__
426
        return data
427
    else:
428
        return obj