Subversion Repositories SmartDukaan

Rev

Rev 5291 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
3232 varun.gupt 1
'''
2
Created on 30-Aug-2011
3
 
4
@author: Varun Gupta
5
'''
6
import lucene
7
from lucene import SimpleFSDirectory, File, StandardAnalyzer, IndexSearcher, Version, MultiFieldQueryParser, QueryParser
8
 
9
class Retriever:
10
 
11
    def __init__(self):
5761 amar.kumar 12
        self.indexDir = "/usr/price-comp-dashboard/lucene-index-dir"
3232 varun.gupt 13
        lucene.initVM()
14
 
15
        dir = SimpleFSDirectory(File(self.indexDir))
16
        self.analyzer = StandardAnalyzer(Version.LUCENE_30)
17
        self.searcher = IndexSearcher(dir)
18
        self.MAX_RESULTS = 1000
19
 
3440 varun.gupt 20
    def retrieve(self, model_number, model_name = None, brand = None, synonyms = None):
3232 varun.gupt 21
        if model_number is None:  return
22
        print model_number, model_name, brand
23
 
24
        query_str = ''
25
 
26
        if brand is not None:
27
            query_brand = "brand:%s" % brand
28
 
29
        if model_name is not None:
30
            name_chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_name.split(' '))
31
 
32
            if brand is not None:   query_str += ' AND '
33
 
34
            query_model_name = ' OR '.join(name_chunks)
35
        else:
36
            query_model_name = ''
37
 
38
        if model_number is not None:
39
            chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_number.split(' '))
40
 
41
            for i in range(0, len(chunks)):
42
 
43
                if chunks[i].find('-') > -1:
44
                    chunk_parts = chunks[i].split('-')
45
                    chunks.append(''.join(chunk_parts))
3457 varun.gupt 46
                    chunks.append("(%s)" % ' AND '.join(chunk_parts))
3232 varun.gupt 47
 
48
            query_model_number = (' OR '.join(chunks)).strip()
49
        else:
50
            query_model_number = ''
3440 varun.gupt 51
 
52
        query_synonyms = ' '.join(synonyms) if synonyms is not None else ''
53
 
3232 varun.gupt 54
        if query_model_name:
55
            query_str = "%s AND (%s" % (query_brand, query_model_name)
56
        else:
57
            query_str = "%s AND (" % query_brand
58
 
59
        if query_model_number:
60
            if query_model_name:
3440 varun.gupt 61
                query_str += " OR %s %s)" % (query_model_number, query_synonyms)
3232 varun.gupt 62
            else:
3440 varun.gupt 63
                query_str += " %s %s)" % (query_model_number, query_synonyms)
3232 varun.gupt 64
        else:
3440 varun.gupt 65
            query_str += " %s)" % query_synonyms
3232 varun.gupt 66
 
67
#        query = QueryParser(Version.LUCENE_30, "name", analyzer).parse(user_query)
68
        qp = MultiFieldQueryParser(Version.LUCENE_30, ["name", "brand"], self.analyzer)
69
        query = qp.parse(qp, query_str)
70
        hits = self.searcher.search(query, self.MAX_RESULTS)
5291 varun.gupt 71
        print "Found %d document(s) that matched query '%s':" % (hits.totalHits, query_str)
3232 varun.gupt 72
 
73
        phones = []
74
 
75
        for hit in hits.scoreDocs:
76
            doc = self.searcher.doc(hit.doc)
77
            phones.append({
5291 varun.gupt 78
                    'name': str(doc.get("name")),
79
                    'price': str(doc.get("price")),
80
                    'score': hit.score,
81
                    'in_stock': str(doc.get("in_stock")),
82
                    'url': str(doc.get("url")),
83
                    'source': str(doc.get("source"))
84
                })
85
        print phones
3232 varun.gupt 86
        return phones
87
 
3313 varun.gupt 88
    def retrieveForQuery(self, query_str):
89
        qp = MultiFieldQueryParser(Version.LUCENE_30, ["name", "brand"], self.analyzer)
90
        query = qp.parse(qp, query_str)
91
        hits = self.searcher.search(query, self.MAX_RESULTS)
3232 varun.gupt 92
 
3313 varun.gupt 93
#        print "Found %d document(s) that matched query '%s':" % (hits.totalHits, query)
94
 
95
        phones = []
96
 
97
        for hit in hits.scoreDocs:
98
            doc = self.searcher.doc(hit.doc)
99
            phones.append({
4039 varun.gupt 100
                    'name': str(doc.get("name")),
101
                    'price': str(doc.get("price")),
102
                    'score': hit.score,
103
                    'in_stock': str(doc.get("in_stock")),
104
                    'url': str(doc.get("url")),
105
                    'source': str(doc.get("source"))
106
            })
3313 varun.gupt 107
        return phones
108
 
3232 varun.gupt 109
if __name__ == "__main__":
110
    retriever = Retriever()
4039 varun.gupt 111
#    print retriever.retrieve(model_number="E2230", model_name = "Hero", brand="Samsung")
5291 varun.gupt 112
    docs = retriever.retrieveForQuery('brand:Spice AND (M-6868 OR M6868 OR (M AND 6868) )')
4198 varun.gupt 113
    print len(docs)
114
    print docs
3440 varun.gupt 115
    #    print phone
3232 varun.gupt 116
    '''
117
    catalog_client = CatalogClient().get_client()
118
    items = catalog_client.getAllItems(True)
119
    products = {}
120
 
121
    for item in items:
122
        if item.category in (10002, 10003, 10004, 10005):  products[item.catalogItemId] = item
123
 
124
    for key, item in products.iteritems():
125
        try:
126
            name = "%s %s" % (item.modelName, item.modelNumber)
127
            matching_phones_with_top_score = getItemsWithTopScore(retriever.retrieve(name))
128
            print name, '\n', isPriceSame(matching_phones_with_top_score), matching_phones_with_top_score
129
        except Exception as e:
130
            print e
131
    '''