Subversion Repositories SmartDukaan

import pymongo
from dtr.utils.utils import to_java_date
from datetime import datetime, timedelta
from dtr.utils import AmazonPriceOnlyScraper

con = None
scraperAmazon = AmazonPriceOnlyScraper.AmazonScraper()


def get_mongo_connection(host='localhost', port=27017):
    global con
    if con is None:
        print "Establishing connection %s host and port %d" %(host,port)
        try:
            con = pymongo.MongoClient(host, port)
        except Exception, e:
            print e
            return None
    return con

def scrapeAmazon():
    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:
            inStock = 0
            print str(data['identifier'])
            if data['identifier'] is None or len(data['identifier'].strip())==0:
                print "continue"
                continue
            
            try:
                if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
                    print "sku id is already updated",data['_id'] 
                    continue
            except:
                pass
            
            url = "http://www.amazon.in/gp/offer-listing/%s/ref=olp_sort_ps"%(data['identifier'].strip())
            print url
            lowestPrice = 0.0
            lowestPrice = scraperAmazon.read(url)
            print lowestPrice
            if lowestPrice > 0:
                inStock = 1
            print lowestPrice
            print inStock
            if 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}}, 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}}, 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",skuBundleId
    
    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('available_price',pymongo.ASCENDING)]))
    bestPrice = float("inf")
    bestOne = None
    bestSellerPoints = 0
    toUpdate = []
    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'] }, {'$set':{'showDeal':0 }})
            continue
        if similarItem['available_price'] < bestPrice:
            bestOne = similarItem
            bestPrice = similarItem['available_price']
            bestSellerPoints = similarItem['bestSellerPoints']
        elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
            bestOne = similarItem
            bestPrice = similarItem['available_price']
            bestSellerPoints = similarItem['bestSellerPoints']
        else:
            pass
    if bestOne is not None:
        for similarItem in similarItems:
            toUpdate.append(similarItem['_id'])
        toUpdate.remove(bestOne['_id'])
        get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$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()