WebSVN – SmartDukaan – Diff – /trunk/PyDTR/src/dtr/sources/snapdeal.py

 #        '>': '&lt;',
 #        '&': '&amp;',
 #        '"': '&quot;', # should be escaped in attributes
 #        "'": '&apos;'    # should be escaped in attributes
 #        }
+        secondryIdentiferSupcMap = {}
+        for scriptTag in pq.items("script"):
+            if  "var reqData1 =" in scriptTag.text():
+                match = re.search("(\[.*?\])",scriptTag.text(), re.DOTALL)
+                a = match.group(1)
+                if a:
+                    for mapElement in json.loads(a):
+                        secondryIdentiferSupcMap[mapElement["pog_id"]] = mapElement["supc"]
+                    break
         jsonValue = pq("#orderJSON").attr("value")
         jsonValue.replace("&quot;", '"')
         jsonValue.replace("&amp;", '&')
         jsonValue.replace("&gt;", '>')
         jsonValue.replace("&lt;", '<')
             try:
                 supcElement = pq(supcElement)
                 title = supcElement('div.order-heading').text().strip()
                 productUrl = supcElement.attr("data-href")
                 imgUrl = supcElement.find('img').attr('src')
-                try:
-                    supc = self.catalogdb.MasterData.find_one({"secondaryIdentifier":productUrl.split("/")[-1], "source_id":self.store_id})
+                secondaryIdentifier = productUrl.split("/")[-1]
-                    if supc is None:
+                if secondryIdentiferSupcMap.has_key(secondaryIdentifier):
-                        raise
-                    else:
-                        supc = supc['identifier']
+                    supc = secondryIdentiferSupcMap[secondaryIdentifier]
-                except:
+                else:
-                    try:
-                        supc = imgUrl.split('-')[-3]
-                    except:
-                        supc = self.db.sdIdentifiers.find_one({"secondaryIdentifier":productUrl.split("/")[-1]})
+                    supc = self.catalogdb.MasterData.find_one({"secondaryIdentifier": secondaryIdentifier, "source_id":self.store_id})
-                        if supc is None:
-                            raise
-                        else:
+                    if supc:
-                            supc = supc['identifier']
+                        supc = supc['identifier']
                 supcMap[supc] = {'title':title, 'imgUrl':imgUrl, 'productUrl':productUrl}
             except:
                 pass
         return json.loads(jsonValue)
             try:
                 if pq("title").text()=="Webpage not available":
                     raise
                 orderJSON = self.getOrderJSON(pq, supcMap)
             except:
+                traceback.print_exc()
                 resp['result'] = 'ORDER_NOT_CREATED_KNOWN'
                 return resp
                 '''page =fetchResponseUsingProxy(url)
                 try:
                     merchantOrder = self._parseB(orderId, subTagId, userId, page, orderSuccessUrl)
         #soup = BeautifulSoup(rawHtml,convertEntities=BeautifulSoup.HTML_ENTITIES)
         #soup.find(name, attrs, recursive, text)
     def _parseC(self, orderId, subTagId, userId, supcMap, orderJSON, orderSuccessUrl):
-        print orderJSON
+        print json.dumps(orderJSON)
         merchantOrder = Order(orderId, userId, subTagId, self.store_id, orderSuccessUrl)
         placedOn = datetime.strftime(utils.fromTimeStamp(orderJSON['created']/1000), "%a, %d %b, %Y")
         merchantOrder.placedOn = placedOn
         merchantOrder.merchantOrderId = orderJSON['code']
         merchantOrder.paidAmount = orderJSON['paidAmount']
         merchantOrder.deliveryCharges = orderJSON['shippingCharges']
         merchantOrder.closed= False
         merchantSubOrders = []
         for s in orderJSON['suborders']:
-            print s
-            if not supcMap.has_key(s['supcCode']):
-                skuData = Mongo.get_mongo_connection().Catalog.MasterData.find_one({'identifier':s['supcCode'], 'source_id':self.store_id})
-                if skuData is None:
-                    url = "http://www.snapdeal.com/search?keyword=%s"%s['supcCode']
-                    html = utils.fetchResponseUsingProxy(url)
-                    html = html.replace("&#13;", "")
-                    pq = PyQuery(html)
-                    tag = pq('div.product-tuple-listing')
-                    for resultDiv in tag:
-                        pq_resultDiv = pq(resultDiv)
-                        if (pq_resultDiv.attr['data-defaultsupcforfmcg']).strip() == s['supcCode'].strip():
-                            productUrl =  pq_resultDiv('div.product-tuple-image').children('a').attr('href')
-                            imgUrl =  pq_resultDiv('div.product-tuple-image').children('a').children('img').attr('src')
-                            title = pq_resultDiv('p.product-title').text()
-                            break
-                else:
-                    title = skuData['product_name']
-                    productUrl = skuData['marketPlaceUrl']
-                    imgUrl = skuData['thumbnail']
-                supcMap[s['supcCode']] = {'title':title, 'imgUrl':imgUrl, 'productUrl':productUrl}
             map1 = supcMap[s['supcCode']]
             amountPaid = s['paidAmount']
             productTitle = map1['title']
             productUrl = map1['productUrl']
 def main():
     #print todict([1,2,"3"])
     store = getStore(3)
-    store.scrapeStoreOrders()
+    #store.scrapeStoreOrders()
-    #store.parseOrderRawHtml(332222, "3232311", 2, readSSh("/home/amit/sample1.html"), "    https://m.snapdeal.com/purchaseMobileComplete?code=a8b3420d4c5bc248ea887df6c9b3a724&order=10478372453")
+    store.parseOrderRawHtml(332222, "3232311", 2, readSSh("/home/amit/sample.html"), "https://m.snapdeal.com/purchaseMobileComplete?code=1a011639e72588db39169df568654620&order=17772748329&sdInstant=false")
     #store.scrapeAffiliate()
     #https://m.snapdeal.com/purchaseMobileComplete?code=3fbc8a02a1c4d3c4e906f46886de0464&order=5808451506
     #https://m.snapdeal.com/purchaseMobileComplete?code=9f4dfa49ff08a16d04c5e4bf519506fc&order=9611672826
 #    orders = list(session.query(OrdersRaw).filter_by(store_id=3).filter_by(status='ORDER_NOT_CREATED').all())

Subversion Repositories SmartDukaan

(root)/trunk/PyDTR/src/dtr/sources/snapdeal.py – Rev 20472 → 20694