Subversion Repositories SmartDukaan

Rev

Rev 4199 | Rev 5377 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 4199 Rev 5291
Line 32... Line 32...
32 
        for brand in self.brands:
 32 
        for brand in self.brands:
33 
            if full_name.startswith(brand):  return (brand, full_name.replace(brand, '').strip())
 33 
            if full_name.startswith(brand):  return (brand, full_name.replace(brand, '').strip())
34
34
35 
        return ("", full_name)
 35 
        return ("", full_name)
36 
 
 36 
 
- 
 
 37 
class DuplicateMappingTracker:
- 
 
 38
- 
 
 39 
    def __init__(self):
- 
 
 40 
        self.mapping = {}
- 
 
 41
- 
 
 42 
    def track(self, url, entity_id):
- 
 
 43 
        if url in self.mapping:
- 
 
 44 
            self.mapping[url].append(entity_id)
- 
 
 45 
        else:
- 
 
 46 
            self.mapping[url] = [entity_id]
- 
 
 47
- 
 
 48 
    def getDuplicateMappings(self):
- 
 
 49
- 
 
 50 
        duplicate_mappings = {}
- 
 
 51
- 
 
 52 
        for url, entities in duplicate_mappings.iteritems():
- 
 
 53 
            if len(entities) > 1:
- 
 
 54 
                duplicate_mappings[url] = entities
- 
 
 55
- 
 
 56 
        return duplicate_mappings
- 
 
 57
37 
def getURLSource(url):
 58 
def getURLSource(url):
38 
    try:
 59 
    try:
39 
        return str(url.split('.')[1].strip())
 60 
        return str(url.split('.')[1].strip())
40 
    except Exception:
 61 
    except Exception:
41 
        return None
 62 
        return None
Line 80... Line 101...
80 
    '''
 101 
    '''
81 
    Receives a list of products (returned from search results) &
 102 
    Receives a list of products (returned from search results) &
82 
    returns a clustered dictionary, where products are grouped by
 103 
    returns a clustered dictionary, where products are grouped by
83 
    the 'source'
 104 
    the 'source'
84 
    '''
 105 
    '''
85 
    clustered_results = {'adexmart': [], 'flipkart': [], 'homeshop18': [], 'infibeam': [], 'letsbuy': []}
 106 
    clustered_results = {'adexmart': [], 'flipkart': [], 'homeshop18': [], 'infibeam': [], 'snapdeal': []}
86
107
87 
    for product in products:
 108 
    for product in products:
88 
        clustered_results[product['source']].append(product)
 109 
        clustered_results[product['source']].append(product)
89
110
90 
    return clustered_results
 111 
    return clustered_results
Line 110... Line 131...
110 
            'letsbuy': 'http://www.letsbuy.com/advanced_search_result.php?cPath=254&keywords=$$'
 131 
            'letsbuy': 'http://www.letsbuy.com/advanced_search_result.php?cPath=254&keywords=$$'
111 
        }
 132 
        }
112 
        return search_urls[source].replace('$$', name)
 133 
        return search_urls[source].replace('$$', name)
113 
 
 134 
 
114 
def getDisplayInfo(filtered_cluster, product_name):
 135 
def getDisplayInfo(filtered_cluster, product_name):
115 
    display_info = {'adexmart': {}, 'flipkart': {}, 'homeshop18': {}, 'infibeam': {}, 'letsbuy': {}}
 136 
    display_info = {'adexmart': {}, 'flipkart': {}, 'homeshop18': {}, 'infibeam': {}, 'snapdeal': {}}
116
137
117 
    for source, products in filtered_cluster.iteritems():
 138 
    for source, products in filtered_cluster.iteritems():
118
139
119 
        if len(products) > 0:
 140 
        if len(products) > 0:
120 
            if isPriceSame(products):
 141 
            if isPriceSame(products):