Subversion Repositories SmartDukaan

from elasticsearch import Elasticsearch

from dtr.utils.utils import get_mongo_connection

import optparse

from pymongo import ASCENDING

parser = optparse.OptionParser()

parser.add_option("-m", "--m", dest="mongoHost",

                      default="localhost",

                      type="string", help="The HOST where the mongo server is running",

                      metavar="HOST")

parser.add_option("-e", "--e", dest="elastic_search_host",

                      default="localhost",

                      type="string", help="The HOST where the elastic server is running",

                      metavar="HOST")

parser.add_option("-p", "--p", dest="elastic_search_port",

                      default="9200",

                      type="string", help="The PORT where the elastic server is running",

                      metavar="HOST")

(options, args) = parser.parse_args()

subCategoryList = [19, 20, 27, 29, 33, 28]

es = Elasticsearch([{'host': options.elastic_search_host, 'port': options.elastic_search_port}])

xstr = lambda s: s or ""

class __SkuInfo:

    def __init__(self, id, title, category_id, subCategoryId, subCategory, internalRank):

        self.id = id

        self.title = title

        self.category_id = category_id

        self.subCategoryId = subCategoryId

        self.subCategory = subCategory

        self.internalRank = internalRank

def main():

    added = []

    items = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'subCategoryId':{"$in":subCategoryList}}))

    for item in items:

        if item['skuBundleId'] in added:

            continue

        deal_obj = list(get_mongo_connection().Catalog.Deals.find({'skuBundleId':item['skuBundleId'],'internalRank':{"$gt":0}}).sort([('internalRank',ASCENDING)]))

        if len(deal_obj) == 0:

            internalRank = 999999

        else:

            internalRank = deal_obj[0]['internalRank']

        title = xstr(item['brand'])+" "+xstr(item['model_name'])

        s_info = __SkuInfo(int(item['skuBundleId']),title,int(item['category_id']),int(item['subCategoryId']),item['subCategory'], internalRank)

        es.index(index='my_index', doc_type='my_type', id=s_info.id,body=s_info.__dict__)

        added.append(int(item['skuBundleId']))

def validateListings():

    offset , limit = 0, 100

    body = {

            "query" : {

        "match_all" : {}

        }

            }

    toDelete = []

    while(True):

        result = es.search("my_index", "my_type", body,from_=offset,size=limit)

        print result

        if len(result['hits']['hits']) > 0:

            for x in result['hits']['hits']:

                skuBundleId =x['_source']['id']

                subCategoryId = x['_source']['subCategoryId']

                category_id = x['_source']['category_id']

                exist = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'skuBundleId':skuBundleId}))

                if len(exist) ==0:

                    toDelete.append(skuBundleId)

                else:

                    for item in exist:

                        if item['subCategoryId']!=subCategoryId or item['category_id'] !=category_id:

                            if skuBundleId not in toDelete:

                                print "Deleting item ",skuBundleId

                                toDelete.append(skuBundleId)

            offset = offset+limit

        else:

            break

    for id in toDelete:

        print "Deleting id ",id

        es.delete(index='my_index', doc_type='my_type', id=id)

if __name__ == '__main__':

    main()

    validateListings()