Subversion Repositories SmartDukaan

Rev

Rev 14287 | Rev 14289 | 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]
14145 amit.gupta 59
        print token
13569 amit.gupta 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)
14145 amit.gupta 213
            merchantOrder = self._parse(orderId, subTagId, userId, page, orderSuccessUrl)
214
            merchantOrder.orderTrackingUrl = url   
13582 amit.gupta 215
            self._saveToOrder(todict(merchantOrder))
13796 amit.gupta 216
            resp['result'] = 'ORDER_CREATED'
217
            return resp
13582 amit.gupta 218
        except:
219
            print "Error occurred"
13603 amit.gupta 220
            traceback.print_exc()
13796 amit.gupta 221
            resp['result'] = 'PARSE_ERROR'
222
            return resp
13781 amit.gupta 223
 
13569 amit.gupta 224
 
225
        #soup = BeautifulSoup(rawHtml,convertEntities=BeautifulSoup.HTML_ENTITIES)
226
        #soup.find(name, attrs, recursive, text)
13576 amit.gupta 227
 
228
    def _getStatusFromDetailedStatus(self, detailedStatus):
229
        for key, value in Store.OrderStatusMap.iteritems():
14288 amit.gupta 230
            print key, value
13576 amit.gupta 231
            if detailedStatus in value:
232
                return key
14287 amit.gupta 233
            print "Detailed Status need to be mapped", detailedStatus
13576 amit.gupta 234
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
13569 amit.gupta 235
 
13610 amit.gupta 236
 
13569 amit.gupta 237
    def scrapeStoreOrders(self,):
13760 amit.gupta 238
        #collectionMap = {'palcedOn':1}
13576 amit.gupta 239
        orders = self._getActiveOrders()
13730 amit.gupta 240
        print "Found orders", orders
13576 amit.gupta 241
        br = getBrowserObject()
242
        for order in orders:
243
            url = ORDER_TRACK_URL + re.findall('.*(\?.*?)$', order['orderSuccessUrl'],re.IGNORECASE)[0]
244
            response = br.open(url)
13680 amit.gupta 245
            page = ungzipResponse(response)
13576 amit.gupta 246
            #page=page.decode("utf-8")
247
            soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
248
            sections = soup.findAll("section")
13760 amit.gupta 249
            orderEl = sections[1]
250
            orderTrs = orderEl.findAll("tr")
251
 
252
            placedOn = str(orderTrs[0].findAll("td")[1].text)
13576 amit.gupta 253
            sections.pop(0)
254
            sections.pop(0)
255
 
256
            subOrders = sections
257
            bulk = self.db.merchantOrder.initialize_ordered_bulk_op()
258
            for subOrderElement in subOrders:
259
                closed = True
260
                divs = subOrderElement.findAll("div", {"class": "blk lrPad subordrs"})
261
                if len(divs)<=0:
262
                    raise ParseException("subOrder", "Could not Parse suborders for Snapdeal")
13721 amit.gupta 263
                subOrder = None
13760 amit.gupta 264
                breakFlag = False
13576 amit.gupta 265
                for div in divs:
266
                    divStr = str(div)
267
                    divStr = divStr.replace("\n","").replace("\t", "")
268
                    updateMap = {}
269
                    for line in divStr.split("<br />"):
270
                        if "Suborder ID" in line:
13634 amit.gupta 271
                            merchantSubOrderId = re.findall(r'\d+', line)[0]
272
                            #break if suborder is inactive   
13721 amit.gupta 273
                            subOrder =  self._isSubOrderActive(order, merchantSubOrderId)
274
                            if subOrder is None:
13809 amit.gupta 275
                                subOrders = self.parseSubOrder(subOrderElement, placedOn)
14239 amit.gupta 276
                                self.db.merchantOrder.update({"orderId":order['orderId']},{'$push':{"subOrders":{"$each":todict(subOrders)}}})
14241 amit.gupta 277
                                print "Added new suborders to Order id - " + order['orderId']
13760 amit.gupta 278
                                closed = False
14240 amit.gupta 279
                                breakFlag = True
280
                                break
13760 amit.gupta 281
                            elif subOrder['closed']:
282
                                breakFlag = True
13634 amit.gupta 283
                                break
13760 amit.gupta 284
                            else: 
285
                                findMap = {"orderId": order['orderId'], "subOrders.merchantSubOrderId": merchantSubOrderId}
13576 amit.gupta 286
                        elif "Status" in line:
287
                            detailedStatus = re.findall('>(.*?)</span>', line, re.IGNORECASE)[0]
288
                            updateMap["subOrders.$.detailedStatus"] = detailedStatus
289
                            status = self._getStatusFromDetailedStatus(detailedStatus) 
13634 amit.gupta 290
                            closedStatus = status in [Store.ORDER_DELIVERED, Store.ORDER_CANCELLED]
13760 amit.gupta 291
                            updateMap["subOrders.$.status"] = status
292
                            if detailedStatus == 'Closed For Vendor Reallocation':
13770 amit.gupta 293
                                #if it is more than 6hours mark closed.
13760 amit.gupta 294
                                closeAt = subOrder.get("closeAt") 
295
                                if closeAt is None:
296
                                    closeAt = datetime.now() + timedelta(hours=6)
13770 amit.gupta 297
                                    updateMap["subOrders.$.closeAt"] = datetime.strftime(closeAt,"%Y-%m-%d %H:%M:%S")
13760 amit.gupta 298
                                else:
13770 amit.gupta 299
                                    closeAt = datetime.strptime(closeAt,"%Y-%m-%d %H:%M:%S")
13760 amit.gupta 300
                                    if datetime.now() > closeAt:
301
                                        closedStatus = True
302
 
303
 
13634 amit.gupta 304
                            if closedStatus:
305
                                #if status is closed then change the paybackStatus accordingly
13760 amit.gupta 306
                                updateMap["subOrders.$.closed"] = True
307
                                if status == Store.ORDER_DELIVERED:
13721 amit.gupta 308
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
13634 amit.gupta 309
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_APPROVED
13760 amit.gupta 310
                                elif status == Store.ORDER_CANCELLED:
13721 amit.gupta 311
                                    if subOrder.get("cashBackStatus") == Store.CB_PENDING:
13634 amit.gupta 312
                                        updateMap["subOrders.$.cashBackStatus"] = Store.CB_CANCELLED
313
 
314
                            else:
315
                                closed = False
13576 amit.gupta 316
                        elif "Est. Shipping Date" in line:
317
                            estimatedShippingDate = line.split(":")[1].strip()
318
                            updateMap["subOrders.$.estimatedShippingDate"] = estimatedShippingDate
319
                        elif "Est. Delivery Date" in line:
320
                            estimatedDeliveryDate = line.split(":")[1].strip()
321
                            updateMap["subOrders.$.estimatedDeliveryDate"] = estimatedDeliveryDate
322
                        elif "Courier Name" in line:
323
                            courierName = line.split(":")[1].strip()
324
                            updateMap["subOrders.$.courierName"] = courierName
325
                        elif "Tracking No" in line:
326
                            trackingNumber = line.split(":")[1].strip()
327
                            updateMap["subOrders.$.trackingNumber"] = trackingNumber
13760 amit.gupta 328
 
329
                    if breakFlag:
330
                        break
13576 amit.gupta 331
 
332
                    bulk.find(findMap).update({'$set' : updateMap})
13721 amit.gupta 333
                bulk.find({'orderId': order['orderId']}).update({'$set':{'closed': closed}})
13576 amit.gupta 334
            result = bulk.execute()
335
            pprint(result)        
336
 
337
 
13569 amit.gupta 338
    def _saveToAffiliate(self, offers):
13725 amit.gupta 339
        if offers is None or len(offers)==0:
340
            print "no affiliate have been pushed"
341
            return
13576 amit.gupta 342
        collection = self.db.snapdealOrderAffiliateInfo
13569 amit.gupta 343
        try:
344
            collection.insert(offers,continue_on_error=True)
345
        except pymongo.errors.DuplicateKeyError as e:
346
            print e.details
347
 
348
 
349
    def _getAllOffers(self, br, token):
350
        allOffers = []
351
        nextPage = 1  
352
        while True:
353
            data = getPostData(token, nextPage)
354
            response = br.open(POST_URL, data)
13680 amit.gupta 355
            rmap = json.loads(ungzipResponse(response))
13569 amit.gupta 356
            if rmap is not None:
357
                rmap = rmap['response']
358
                if rmap is not None and len(rmap['errors'])==0:
359
                    allOffers += rmap['data']['data']
360
            nextPage += 1
361
            if rmap['data']['pageCount']<nextPage:
362
                break
363
 
364
        return allOffers
365
 
366
    def covertToObj(self,offer):
367
        offerData = offer['Stat']
368
        offer1 = AffiliateInfo(offerData['affiliate_info1'], self.store_id, offerData['conversion_status'], offerData['ad_id'], 
369
                              offerData['datetime'], offerData['payout'], offer['Offer']['name'], offerData['ip'], offerData['conversion_sale_amount'])
13680 amit.gupta 370
 
13569 amit.gupta 371
        return offer1
372
def getPostData(token, page = 1, limit= 20, startDate=None, endDate=None):
13680 amit.gupta 373
    endDate=date.today() + timedelta(days=1)
374
    startDate=endDate - timedelta(days=31)
13569 amit.gupta 375
 
376
    parameters = (
377
        ("page",str(page)),
378
        ("limit",str(limit)),
379
        ("fields[]","Stat.offer_id"),
380
        ("fields[]","Stat.datetime"),
381
        ("fields[]","Offer.name"),
382
        ("fields[]","Stat.conversion_status"),
383
        ("fields[]","Stat.conversion_sale_amount"),
384
        ("fields[]","Stat.payout"),
385
        ("fields[]","Stat.ip"),
386
        ("fields[]","Stat.ad_id"),
387
        ("fields[]","Stat.affiliate_info1"),
388
        ("sort[Stat.datetime]","desc"),
389
        ("filters[Stat.date][conditional]","BETWEEN"),
390
        ("filters[Stat.date][values][]",startDate.strftime('%Y-%m-%d')),
391
        ("filters[Stat.date][values][]",endDate.strftime('%Y-%m-%d')),
392
        ("data_start",startDate.strftime('%Y-%m-%d')),
393
        ("data_end",endDate.strftime('%Y-%m-%d')),
394
        ("Method","getConversions"),
395
        ("NetworkId","jasper"),
396
        ("SessionToken",token),
397
    )
398
    #Encode the parameters
399
    return urllib.urlencode(parameters)
400
 
401
def main():
14239 amit.gupta 402
    print todict([1,2,"3"])
403
    #store = getStore(3)
404
    #store.scrapeStoreOrders()
13662 amit.gupta 405
    #store._isSubOrderActive(8, "5970688907")
13760 amit.gupta 406
    #store.scrapeAffiliate()
13576 amit.gupta 407
    #store.parseOrderRawHtml(12345, "subtagId", 122323,  "html", 'https://m.snapdeal.com/purchaseMobileComplete?code=1f4166d13ea799b65aa9dea68b3e9e70&order=4509499363')
13569 amit.gupta 408
 
409
 
13576 amit.gupta 410
 
411
def todict(obj, classkey=None):
412
    if isinstance(obj, dict):
413
        data = {}
414
        for (k, v) in obj.items():
415
            data[k] = todict(v, classkey)
416
        return data
417
    elif hasattr(obj, "_ast"):
418
        return todict(obj._ast())
419
    elif hasattr(obj, "__iter__"):
420
        return [todict(v, classkey) for v in obj]
421
    elif hasattr(obj, "__dict__"):
422
        data = dict([(key, todict(value, classkey)) 
423
            for key, value in obj.__dict__.iteritems() 
424
            if not callable(value) and not key.startswith('_')])
425
        if classkey is not None and hasattr(obj, "__class__"):
426
            data[classkey] = obj.__class__.__name__
427
        return data
428
    else:
429
        return obj
14239 amit.gupta 430
 
431
if __name__ == '__main__':
432
    main()