Subversion Repositories SmartDukaan

Rev

Rev 14607 | Rev 14624 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 14607 Rev 14608
Line 40... Line 40...
40 
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, insert=False):
 40 
    def parseOrderRawHtml(self, orderId, subTagId, userId, rawHtml, orderSuccessUrl, insert=False):
41 
        resp = {}
 41 
        resp = {}
42 
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
 42 
        if ORDER_SUCCESS_URL in orderSuccessUrl or THANKYOU_URL in orderSuccessUrl:
43 
            try:
 43 
            try:
44 
                soup = BeautifulSoup(rawHtml)
 44 
                soup = BeautifulSoup(rawHtml)
- 
 
 45 
                try:
45 
                orderUrl = soup.find('div', {"id":"thank-you-box-wrapper"}).div.findAll('div', recursive=False)[1].a['href']
 46 
                    orderUrl = soup.find('div', {"id":"thank-you-box-wrapper"}).div.findAll('div', recursive=False)[1].a['href']
46 
                merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
 47 
                    merchantOrderId = re.findall(r'.*&oid=(.*)&?.*?', orderUrl)[0]
- 
 
 48 
                except:
- 
 
 49 
                    merchantOrderId = soup.find(id="orders-list").div.span.b.text
47 
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
 50 
                order = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl, False)
48 
                order.orderSuccessUrl = ORDER_REDIRECT_URL % (merchantOrderId)
 51 
                order.orderSuccessUrl = ORDER_REDIRECT_URL % (merchantOrderId)
49 
                order.merchantOrderId = merchantOrderId
 52 
                order.merchantOrderId = merchantOrderId
50 
                order.requireDetail = True
 53 
                order.requireDetail = True
51 
                order.closed = None
 54 
                order.closed = None
Line 83... Line 86...
83 
                resp['result'] = 'ORDER_NOT_CREATED'
 86 
                resp['result'] = 'ORDER_NOT_CREATED'
84 
                return resp
87 
                return resp
85
88
86 
    #This should be exposed from api for specific sources
 89 
    #This should be exposed from api for specific sources
87 
    def scrapeStoreOrders(self):
 90 
    def scrapeStoreOrders(self):
- 
 
 91 
#        br = getBrowserObject()
- 
 
 92 
#        orders = self.db.merchantOrder.find({"storeId":1, "closed":False, "subOrders.closed":False, "subOrders.trackingUrl":{"$exists":True}})
- 
 
 93 
#        for merchantOrder in orders:
- 
 
 94 
#            for subOrder in merchantOrder.get("subOrders"):
- 
 
 95 
#                trackingUrl = subOrder.get("trackingUrl")
- 
 
 96 
#                if not subOrder.get("closed") and trackingUrl is not None:
- 
 
 97 
#
- 
 
 98 
#                    self.parseTrackingUrl(br, trackingUrl)
88 
        pass
 99 
        pass
- 
 
 100
- 
 
 101
- 
 
 102
89
103
90 
    def parseOldStlye(self, merchantOrder, soup):
 104 
    def parseOldStlye(self, merchantOrder, soup):
91 
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
 105 
        merchantOrder.orderTrackingUrl = merchantOrder.orderSuccessUrl
92 
        table = soup.body.findAll("table", recursive=False)[1]
 106 
        table = soup.body.findAll("table", recursive=False)[1]
93 
        #print table
 107 
        #print table
Line 317... Line 331...
317 
                                trackingUrl = None
 331 
                                trackingUrl = None
318 
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
 332 
                                for buttonDiv in shipdiv.findAll('span', {'class':'a-button-inner'}):
319 
                                    if buttonDiv.find('a').text.strip()=='Track package':
 333 
                                    if buttonDiv.find('a').text.strip()=='Track package':
320 
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
 334 
                                        trackingUrl = buttonDiv.find('a')['href'].strip()
321 
                                        if not trackingUrl.startswith("http"):
 335 
                                        if not trackingUrl.startswith("http"):
322 
                                            trackingUrl = "https://www.amazon.in/" + trackingUrl
 336 
                                            trackingUrl = "http://www.amazon.in" + trackingUrl
323 
                                        break
 337 
                                        break
324 
                                for prodDiv in productDivs:
 338 
                                for prodDiv in productDivs:
325 
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
 339 
                                    prodDiv.find('div', {'class':'a-fixed-left-grid-inner'})
326 
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
 340 
                                    productTitle = prodDiv.find('div', {'class':'a-fixed-left-grid-inner'}).find("div", {'class':'a-row'}).find('a').text.strip()
327 
                                    imgUrl = prodDiv.find("img")["src"]
 341 
                                    imgUrl = prodDiv.find("img")["src"]
Line 378... Line 392...
378 
                return key
 392 
                return key
379
393
380 
        print "Detailed Status need to be mapped"
 394 
        print "Detailed Status need to be mapped"
381 
        print "Found new order status", detailedStatus
 395 
        print "Found new order status", detailedStatus
382 
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
 396 
        raise ParseException("_getStatusFromDetailedStatus", "Found new order status" + detailedStatus)
- 
 
 397
- 
 
 398 
    def parseTrackingUrl(self, br, trackingUrl):
- 
 
 399 
        response = br.open(trackingUrl)
- 
 
 400 
        page = ungzipResponse(response)
- 
 
 401 
        soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)
- 
 
 402 
        alertContainer = soup.find("a-box-inner a-alert-container")
- 
 
 403
- 
 
 404
383 
 
 405 
 
384 
 
 406 
 
385 
def main():
 407 
def main():
386 
    store = getStore(1)
 408 
    store = getStore(1)
387 
    #store.scrapeStoreOrders()
409 
    #br = getBrowserObject()
388 
    store.parseOrderRawHtml(11112, 'saad', '21232211', readSSh('/home/amit/786.html'), 'https://www.amazon.in/gp/css/summary/edit.html?orderID=12212')
 410 
    #store.parseTrackingUrl(br, "http://www.amazon.in//gp/css/shiptrack/view.html/ref=oh_aui_direct_track_pkg_o00_s00?ie=UTF8&marketplaceOfOrigin=&orderID=171-7613541-7425906&orderingShipmentId=729710930106&packageId=1&ref=")
389 
    #store.trackOrdersForUser(8,'https://www.amazon.in/gp/css/order-history', readSSh('/tmp/User2/2015-03-03 00:29:40.165513'))
 411 
    store.parseOrderRawHtml(2121, "13232", 14, readSSh("/home/amit/917.html"), "https://www.amazon.in/gp/buy/spc/handlers/static-submit-decoupled.html/ref=ox_spc_place_order?ie=UTF8&hasWorkingJavascript=")
390 
    #readSSh('snapdeal.csv')
 - 
 
391 
def readSSh(fileName):
 412 
def readSSh(fileName):
392 
    try:
 413 
    try:
393 
            str1 = open(fileName).read()
 414 
            str1 = open(fileName).read()
394 
            return str1
 415 
            return str1
395 
    except:
 416 
    except: