Subversion Repositories SmartDukaan

'''

Created on 26-Aug-2011

@author: Varun Gupta

'''

import json, sys, os

cmd_folder = os.path.dirname(os.path.abspath(os.environ["HOME"] + "/code/trunk/PyProj/src/shop2020/"))

if cmd_folder not in sys.path:

    sys.path.insert(0, cmd_folder)

from shop2020.clients.CatalogClient import CatalogClient

CHARACTER_ENCODING = 'ISO-8859-1'

class BrandAndModelExtracter:

    def __init__(self):

        try:

            client = CatalogClient().get_client()

            self.brands = client.getAllBrandsByCategory(10001)

        except Exception:

            self.brands = ['Micromax', 'BlackBerry', 'Blackberry', 'Motorola', 'Alcatel', 'Sony Ericsson', 'Apple', \

                      'Spice', 'Nokia', 'HTC', 'Samsung', 'LG', 'Dell', 'Karbonn', 'Lava']

        self.brands.append('Blackberry') #To resolve issue of 'BlackBerry' and 'Blackberry'

    def extract(self, full_name):

        for brand in self.brands:

            if full_name.startswith(brand):  return (brand, full_name.replace(brand, '').strip())

        return ("", full_name)

def isValidRule(rule):

    try:

        if rule is None:

            return False

        elif rule['url'] is None:

            return False

        elif rule['source'] is None:

            return False

        else:

            return True

    except KeyError:

        return False

def getItemsWithTopScore(items):

    filterd_items = []

    top_score = -1.0

    for item in items:

        if item['score'] >= top_score:

            filterd_items.append(item)

            top_score = item['score'] 

        else:

            return filterd_items

    return filterd_items

def isPriceSame(items):

    for i in range(0, items.__len__() - 1):

        if items[i]['price'] != items[i + 1]['price']:    return False

    return True

def getProductClusters(products):

    '''

    Receives a list of products (returned from search results) &

    returns a clustered dictionary, where products are grouped by

    the 'source'

    '''

    clustered_results = {'flipkart': [], 'homeshop18': [], 'infibeam': [], 'letsbuy': []}

    for product in products:

        clustered_results[product['source']].append(product)

    return clustered_results

def getFilteredClustersWithTopScores(product_clusters):

    filtered_cluster = {}

    for source, products in product_clusters.iteritems():

        filtered_cluster[source] = getItemsWithTopScore(products)

    return filtered_cluster

def removePriceFormatting(price_string):

    return price_string.replace('Rs.', '').replace(',', '').strip()

def getDisplayInfo(filtered_cluster):

    display_info = {'flipkart': {}, 'homeshop18': {}, 'infibeam': {}, 'letsbuy': {}}

    for source, products in filtered_cluster.iteritems():

        if len(products) > 0:

            if isPriceSame(products):

                display_info[source]['price'] = products[0]['price']

                display_info[source]['data'] = None

                display_info[source]['url'] = products[0]['url']

                display_info[source]['text'] = removePriceFormatting(products[0]['price'])

                display_info[source]['title'] = products[0]['name']

            else:

                display_info[source]['price'] = None

                display_info[source]['data'] = json.dumps(products)

                display_info[source]['url'] = None

                display_info[source]['text'] = 'Conflict'

        else:

            display_info[source]['price'] = None

            display_info[source]['data'] = None

            display_info[source]['url'] = None

            display_info[source]['text'] = 'Not Found'

    return display_info

def getSynonyms():

    file_path = '/tmp/synonyms.json'

    file = open(file_path, 'r')

    synonyms_json = file.read()

    synonyms = {}

    for key, value in json.loads(synonyms_json).iteritems():

        list_synonyms = []

        if 'MODEL_NAME' in value:

            list_synonyms.extend(value['MODEL_NAME'])

        if 'MODEL_NUMBER' in value:

            list_synonyms.extend(value['MODEL_NUMBER'])

        synonyms[int(key)] = list_synonyms

    return synonyms

if __name__ == '__main__':

    extracter = BrandAndModelExtracter()

#    print extracter.extract('Nokia X5-01 (Pink)')

    print getSynonyms()