Rev 13976 | Rev 14133 | 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 AmazonPriceOnlyScrapercon = Nonenow = datetime.now()scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()print to_java_date(now)def get_mongo_connection(host='localhost', 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 scrapeAmazon():amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':1}))for data in amazonBestSellers:inStock = 0print str(data['identifier'])if data['identifier'] is None or len(data['identifier'].strip())==0:print "continue"continuetry:if data['priceUpdatedOn'] > to_java_date(now - timedelta(minutes=5)):print "sku id is already updated",data['_id']continueexcept:passurl = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'].strip())print urllowestPrice = 0.0lowestPrice = scraperAmazon.read(url)print lowestPriceif lowestPrice > 0:inStock = 1print lowestPriceprint inStockif lowestPrice > 0:get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice,'updatedOn':to_java_date(now),'priceUpdatedOn':to_java_date(now),'in_stock':inStock}}, multi=True)get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestPrice , 'in_stock':inStock}}, multi=True)else:get_mongo_connection().Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(now),'in_stock':inStock,'priceUpdatedOn':to_java_date(now)}}, multi=True)get_mongo_connection().Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock}}, multi=True)try:recomputeDeal(data['skuBundleId'])except:print "Unable to compute deal for ",data['skuBundleId']def recomputeDeal(skuBundleId):"""Lets recompute deal for this bundle"""print "Recomputing for bundleId",skuBundleIdsimilarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))bestPrice = float("inf")bestOne = NonebestSellerPoints = 0toUpdate = []for similarItem in similarItems:if similarItem['in_stock'] == 0 or similarItem['maxprice'] is None or similarItem['maxprice'] < similarItem['available_price']:get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$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():scrapeAmazon()if __name__=='__main__':main()