Rev 14328 | Rev 15341 | Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed
import pymongofrom dtr.utils.utils import to_java_datefrom datetime import datetime, timedeltafrom dtr.utils import AmazonPriceOnlyScraper, AmazonDealScraperfrom multiprocessing import Pool as ThreadPoolfrom multiprocessing import cpu_countimport optparsefrom dtr.storage.MemCache import MemCachefrom dtr.utils.utils import getCashBackcon = Noneparser = optparse.OptionParser()parser.add_option("-m", "--m", dest="mongoHost",default="localhost",type="string", help="The HOST where the mongo server is running",metavar="mongo_host")(options, args) = parser.parse_args()mc = MemCache(options.mongoHost)def get_mongo_connection(host=options.mongoHost, port=27017):global conif con is None:print "Establishing connection %s host and port %d" %(host,port)try:con = pymongo.MongoClient(host, port)except Exception, e:print ereturn Nonereturn condef populate():toScrapMap = {}bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))for bestSeller in bestSellers:amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))for data in amazonBestSellers:if not toScrapMap.has_key(data['_id']):toScrapMap[data['_id']] = datadealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$gt':0}}))for deal in dealFlagged:if not toScrapMap.has_key(deal['_id']):data = list(get_mongo_connection().Catalog.MasterData.find({'_id':deal['_id']}))toScrapMap[deal['_id']] = data[0]manualDeals = list(get_mongo_connection().Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':1}))for manualDeal in manualDeals:if not toScrapMap.has_key(manualDeal['sku']):data = list(get_mongo_connection().Catalog.MasterData.find({'_id':manualDeal['sku']}))if len(data) > 0:data[0]['dealFlag'] = 1data[0]['dealType'] = manualDeal['dealType']toScrapMap[manualDeal['sku']] = data[0]else:data = toScrapMap.get(manualDeal['sku'])data['dealFlag'] = 1data['dealType'] = manualDeal['dealType']pool = ThreadPool(cpu_count() * 2)pool.map(scrapeAmazon,toScrapMap.values())pool.close()pool.join()print "joining threads at %s"%(str(datetime.now()))def scrapeAmazon(data):inStock = 0dealScraping = Falseprint str(data['identifier'])if data['identifier'] is None or len(data['identifier'].strip())==0:returntry:if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):print "sku id is already updated",data['_id']returnexcept:passurl = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'].strip())try:if data['dealFlag'] ==1 and data['dealType'] ==1:deal_url = "http://www.amazon.in/dp/%s"%(data['identifier'].strip())dealScraping = Trueexcept:data['dealFlag'] = 0data['dealType'] = 0print urllowestPrice = 0.0if not dealScraping:scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()lowestPrice = scraperAmazon.read(url)else:dealScraperAmazon = AmazonDealScraper.AmazonScraper()lowestPrice = dealScraperAmazon.read(deal_url)if lowestPrice == 0:scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()lowestPrice = scraperAmazon.read(url)dealScraping = Falseprint lowestPriceif lowestPrice > 0:inStock = 1print lowestPriceprint inStockif lowestPrice > 0:get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock,'dealType':data['dealType']}}, multi=True)else:get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':inStock,'priceUpdatedOn':to_java_date(datetime.now())}}, multi=True)get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'dealType':data['dealType']}}, multi=True)try:recomputeDeal(data['skuBundleId'])except:print "Unable to compute deal for ",data['skuBundleId']def populateNegativeDeals():negativeDeals = get_mongo_connection().Catalog.NegativeDeals.find().distinct('sku')mc.set("negative_deals", negativeDeals, 600)def recomputeDeal(skuBundleId):"""Lets recompute deal for this bundle"""print "Recomputing for bundleId",skuBundleIdsimilarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))bestPrice = float("inf")bestOne = NonebestSellerPoints = 0toUpdate = []for similarItem in similarItems:if mc.get("negative_deals") is None:populateNegativeDeals()try:cashBack = getCashBack(similarItem['_id'], similarItem['source_id'], similarItem['category_id'], mc, options.mongoHost)if not cashBack or cashBack.get('cash_back_status')!=1:passelse:if cashBack['cash_back_type'] ==1:similarItem['available_price'] = similarItem['available_price'] - similarItem['available_price'] * float(cashBack['cash_back'])/100elif cashBack['cash_back_type'] ==2:similarItem['available_price'] = similarItem['available_price'] - float(cashBack['cash_back'])else:passexcept Exception as cashBackEx:print cashBackExprint "Error calculating cashback."if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price'] or similarItem['_id'] in mc.get("negative_deals"):get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0 }})continueif similarItem['available_price'] < bestPrice:bestOne = similarItembestPrice = similarItem['available_price']bestSellerPoints = similarItem['bestSellerPoints']elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:bestOne = similarItembestPrice = similarItem['available_price']bestSellerPoints = similarItem['bestSellerPoints']else:passif bestOne is not None:for similarItem in similarItems:toUpdate.append(similarItem['_id'])toUpdate.remove(bestOne['_id'])get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1 }})if len(toUpdate) > 0:get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0 }},upsert=False, multi=True)def main():populate()if __name__=='__main__':main()