Subversion Repositories SmartDukaan

import urllib2

import simplejson as json

import pymongo

from dtr.utils.utils import to_java_date, getNlcPoints 

from datetime import datetime, timedelta

import time

from multiprocessing import Pool as ThreadPool

from multiprocessing import cpu_count

import optparse

from dtr.storage.MemCache import MemCache

from dtr.utils.utils import getCashBack, get_mongo_connection, SOURCE_MAP, DEAL_PRIORITY

import traceback

from operator import itemgetter

import chardet

from dtr.utils import HomeShop18Scraper

SOURCE_MAP = {'AMAZON':1,'FLIPKART':2,'SNAPDEAL':3,'SAHOLIC':4, 'SHOPCLUES.COM':5,'PAYTM.COM':6, 'HOMESHOP18.COM':7}

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 getNetPriceForItem(itemId, source_id, category_id ,price):

    cash_back_type = 0

    cash_back = 0

    try:

        cashBack = getCashBack(itemId, source_id, category_id, mc, options.mongoHost)

        if not cashBack or cashBack.get('cash_back_status')!=1:

            cash_back_type = 0

            cash_back = 0 

        else:

            if cashBack['cash_back_type'] in (1,2):

                if cashBack.get('maxCashBack') is not None:

                    if cashBack.get('cash_back_type') ==1 and (float(cashBack.get('cash_back'))*price)/100 > cashBack.get('maxCashBack'):

                        cashBack['cash_back_type'] = 2

                        cashBack['cash_back'] = cashBack['maxCashBack']

                    elif cashBack.get('cash_back_type') ==2 and cashBack.get('cash_back') > cashBack.get('maxCashBack'):

                        cashBack['cash_back'] = cashBack['maxCashBack']

                    else:

                        pass

                cash_back_type = cashBack['cash_back_type']

                cash_back = float(cashBack['cash_back'])

    except Exception as cashBackEx:

        pass

    if cash_back_type ==1:

        return (price - float(cash_back)*price/100)

    elif cash_back_type ==2:

        return (price - cash_back)

    else:

        return price

def populate():

    toScrapMap = {}

    bestSellers = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'rank':{'$gt':0}}))

    for bestSeller in bestSellers: 

        snapdealBestSellers = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'skuBundleId':bestSeller['skuBundleId'],'source_id':7}))

        for data in snapdealBestSellers:

            if not toScrapMap.has_key(data['_id']):

                data['dealFlag'] = 0

                data['dealType'] = 0

                toScrapMap[data['_id']] = data

    dealFlagged = list(get_mongo_connection(host=options.mongoHost).Catalog.Deals.find({'source_id':7,'showDeal':1,'totalPoints':{'$gt':-100}}))

    for deal in dealFlagged:

        if not toScrapMap.has_key(deal['_id']):

            data = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'_id':deal['_id']}))

            data[0]['dealFlag'] = 0

            data[0]['dealType'] = 0

            toScrapMap[deal['_id']] = data[0]

    manualDeals = list(get_mongo_connection(host=options.mongoHost).Catalog.ManualDeals.find({'startDate':{'$lte':to_java_date(datetime.now())},'endDate':{'$gte':to_java_date(datetime.now())},'source_id':7}))

    for manualDeal in manualDeals:

        if not toScrapMap.has_key(manualDeal['sku']):

            data = list(get_mongo_connection(host=options.mongoHost).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']

    for val in toScrapMap.values():

        updatePrices(val)

    print "Done with homeshop pricing"

def updatePrices(data):

    if data.get('ignorePricing') ==1:

        print "Ignored items returning for %d"%(data['_id'])

        return

    if data['source_id']!=7:

        return

    print data['identifier']

    if data['identifier'] is None or len(data['identifier'].strip())==0:

        print "returning"

        return

    try:

        if data['priceUpdatedOn'] > to_java_date(datetime.now() - timedelta(minutes=5)):

            print "sku id is already updated",data['_id'] 

            return

    except:

        pass

    result = None

    try:

        url = 'http://m.homeshop18.com/product.mobi?productId='+str(data['identifier'])

        scraper = HomeShop18Scraper.HomeShop18Scraper()

        result = scraper.read(url)

    except:

        print "Unable to scrape %d"%(data['_id'])

        get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'updatedOn':to_java_date(datetime.now()),'in_stock':0,'priceUpdatedOn':to_java_date(datetime.now())}})

        get_mongo_connection(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':0}})

        return

    inStock = 0

    lowestOfferPrice = 0

    if result is not None:

        lowestOfferPrice = float(result['price']+result['shippingCharge'])

        inStock = result['inStock']

    print lowestOfferPrice

    print inStock

    print "*************"

    if lowestOfferPrice ==0:

        inStock = 0

    if inStock  == 1:

        netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('HOMESHOP18.COM'), data['category_id'], lowestOfferPrice)

        get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice,'updatedOn':to_java_date(datetime.now()),'priceUpdatedOn':to_java_date(datetime.now()),'in_stock':inStock}}, multi=True)

        get_mongo_connection(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'available_price':lowestOfferPrice , 'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)

    else:

        lowestOfferPrice = data['available_price']

        netPriceAfterCashBack = getNetPriceForItem(data['_id'], SOURCE_MAP.get('HOMESHOP18.COM'), data['category_id'], lowestOfferPrice)

        get_mongo_connection(host=options.mongoHost).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(host=options.mongoHost).Catalog.Deals.update({'_id':data['_id']}, {'$set' : {'in_stock':inStock,'codAvailable':data['codAvailable'],'netPriceAfterCashBack':netPriceAfterCashBack}}, multi=True)

    try:

        recomputeDeal(data)

    except:

        print "Unable to compute deal for ",data['skuBundleId']

def populateNegativeDeals():

    negativeDeals = get_mongo_connection(host=options.mongoHost).Catalog.NegativeDeals.find().distinct('sku')

    mc.set("negative_deals", negativeDeals, 600)

def recomputeDeal(item):

    """Lets recompute deal for this bundle"""

    print "Recomputing for bundleId %d" %(item.get('skuBundleId'))

    skuBundleId = item['skuBundleId']

    similarItems = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':skuBundleId}).sort([('netPriceAfterCashBack',pymongo.ASCENDING)]))

    bestPrice = float("inf")

    bestOne = None

    #bestSellerPoints = 0

    toUpdate = []

    prepaidBestPrice = float("inf")

    prepaidBestOne = None

    #prepaidBestSellerPoints = 0

    for similarItem in similarItems:

        if similarItem['codAvailable'] ==1:

            if mc.get("negative_deals") is None:

                populateNegativeDeals()

            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):

                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})

                continue

            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:

                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})

                continue

            if similarItem.get('netPriceAfterCashBack') < bestPrice:

                bestOne = similarItem

                bestPrice = similarItem.get('netPriceAfterCashBack')

                #bestSellerPoints = similarItem['bestSellerPoints']

            elif similarItem.get('netPriceAfterCashBack') == bestPrice:

                try:

                    if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(prepaidBestOne['source_id']))):

                        continue

                except:

                    traceback.print_exc()

                bestOne = similarItem

                bestPrice = similarItem.get('netPriceAfterCashBack')

                #bestSellerPoints = similarItem['bestSellerPoints']

            else:

                pass

        else:

            if mc.get("negative_deals") is None:

                populateNegativeDeals()

            if similarItem['in_stock'] == 0  or similarItem['_id'] in mc.get("negative_deals"):

                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0, 'prepaidDeal':0 }})

                continue

            if similarItem['source_id'] == SOURCE_MAP.get('SHOPCLUES.COM') and similarItem['rank']==0:

                get_mongo_connection().Catalog.Deals.update({ '_id' : similarItem['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':0 }})

                continue

            if similarItem.get('netPriceAfterCashBack') < prepaidBestPrice:

                prepaidBestOne = similarItem

                prepaidBestPrice = similarItem.get('netPriceAfterCashBack')

                #prepaidBestSellerPoints = similarItem['bestSellerPoints']

            elif similarItem.get('netPriceAfterCashBack') == prepaidBestPrice:

                try:

                    if (DEAL_PRIORITY.index(int(similarItem['source_id'])) > DEAL_PRIORITY.index(int(bestOne['source_id']))):

                        continue

                except:

                    traceback.print_exc()

                prepaidBestOne = similarItem

                prepaidBestPrice = similarItem.get('netPriceAfterCashBack')

                #prepaidBestSellerPoints = similarItem['bestSellerPoints']

            else:

                pass

    if bestOne is not None or prepaidBestOne is not None:

        for similarItem in similarItems:

            toUpdate.append(similarItem['_id'])

        if bestOne is not None:

            toUpdate.remove(bestOne['_id'])

            get_mongo_connection().Catalog.Deals.update({ '_id' : bestOne['_id'] }, {'$set':{'showDeal':1,'prepaidDeal':0 }})

        if prepaidBestOne is not None:

            if bestOne is not None:

                if prepaidBestOne.get('netPriceAfterCashBack') < bestOne.get('netPriceAfterCashBack'): 

                    toUpdate.remove(prepaidBestOne['_id'])

                    get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})

            else:

                toUpdate.remove(prepaidBestOne['_id'])

                get_mongo_connection().Catalog.Deals.update({ '_id' : prepaidBestOne['_id'] }, {'$set':{'showDeal':0,'prepaidDeal':1 }})

    if len(toUpdate) > 0:

        get_mongo_connection().Catalog.Deals.update({ '_id' : { "$in": toUpdate } }, {'$set':{'showDeal':0,'prepaidDeal':0 }},upsert=False, multi=True)

def main():

    populate()

if __name__=='__main__':

    main()