Subversion Repositories SmartDukaan

Rev

Rev 14325 | Rev 15267 | Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed

import pymongo
from dtr.utils.utils import to_java_date
from datetime import datetime, timedelta
from dtr.utils import AmazonPriceOnlyScraper, AmazonDealScraper
from multiprocessing import Pool as ThreadPool
from multiprocessing import cpu_count
import optparse
from dtr.storage.MemCache import MemCache

con = None

parser = 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 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 populate():
    toScrapMap = {}
    bestSellers = list(get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0}}))
    for bestSeller in bestSellers: 
        amazonBestSellers = list(get_mongo_connection().Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':1}))
        for data in amazonBestSellers:
            if not toScrapMap.has_key(data['_id']):
                toScrapMap[data['_id']] = data
    dealFlagged = list(get_mongo_connection().Catalog.Deals.find({'source_id':1,'showDeal':1,'totalPoints':{'$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'] = 1
                data[0]['dealType'] = manualDeal['dealType']
                toScrapMap[manualDeal['sku']] = data[0]
        else:
            data = toScrapMap.get(manualDeal['sku'])
            data['dealFlag'] = 1
            data['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 = 0
    dealScraping = False
    print str(data['identifier'])
    if data['identifier'] is None or len(data['identifier'].strip())==0:
        return
    
    try:
        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):
            print "sku id is already updated",data['_id'] 
            return
    except:
        pass
    
    url = "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 = True
    except:
        data['dealFlag'] = 0
        data['dealType'] = 0
        
    print url
    lowestPrice = 0.0
    
    if 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 = False
        
    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,'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",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 mc.get("negative_deals") is None:
            populateNegativeDeals()
        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'] }, {'$set':{'showDeal':0 }})
            continue
        if similarItem['available_price'] < bestPrice:
            bestOne = similarItem
            bestPrice = similarItem['available_price']
            bestSellerPoints = similarItem['bestSellerPoints']
        elif similarItem['available_price'] == bestPrice and bestSellerPoints < similarItem['bestSellerPoints']:
            bestOne = similarItem
            bestPrice = similarItem['available_price']
            bestSellerPoints = similarItem['bestSellerPoints']
        else:
            pass
    if bestOne is not None:
        for similarItem in similarItems:
            toUpdate.append(similarItem['_id'])
        toUpdate.remove(bestOne['_id'])
        get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$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():
    populate()
            
if __name__=='__main__':
    main()