| 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):
|
|
|
12 |
self.indexDir = "/tmp/lucene-index-dir"
|
|
|
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 |
|
|
|
20 |
|
|
|
21 |
def retrieve(self, model_number, model_name = None, brand = None):
|
|
|
22 |
|
|
|
23 |
if model_number is None: return
|
|
|
24 |
print model_number, model_name, brand
|
|
|
25 |
|
|
|
26 |
query_str = ''
|
|
|
27 |
|
|
|
28 |
if brand is not None:
|
|
|
29 |
query_brand = "brand:%s" % brand
|
|
|
30 |
|
|
|
31 |
if model_name is not None:
|
|
|
32 |
name_chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_name.split(' '))
|
|
|
33 |
|
|
|
34 |
if brand is not None: query_str += ' AND '
|
|
|
35 |
|
|
|
36 |
query_model_name = ' OR '.join(name_chunks)
|
|
|
37 |
else:
|
|
|
38 |
query_model_name = ''
|
|
|
39 |
|
|
|
40 |
if model_number is not None:
|
|
|
41 |
chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_number.split(' '))
|
|
|
42 |
|
|
|
43 |
for i in range(0, len(chunks)):
|
|
|
44 |
|
|
|
45 |
if chunks[i].find('-') > -1:
|
|
|
46 |
chunk_parts = chunks[i].split('-')
|
|
|
47 |
chunks.append(''.join(chunk_parts))
|
|
|
48 |
chunks.append(' '.join(chunk_parts))
|
|
|
49 |
|
|
|
50 |
query_model_number = (' OR '.join(chunks)).strip()
|
|
|
51 |
else:
|
|
|
52 |
query_model_number = ''
|
|
|
53 |
|
|
|
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:
|
|
|
61 |
query_str += " OR %s)" % query_model_number
|
|
|
62 |
else:
|
|
|
63 |
query_str += " %s)" % query_model_number
|
|
|
64 |
else:
|
|
|
65 |
query_str += " )"
|
|
|
66 |
|
|
|
67 |
print query_str
|
|
|
68 |
# query = QueryParser(Version.LUCENE_30, "name", analyzer).parse(user_query)
|
|
|
69 |
qp = MultiFieldQueryParser(Version.LUCENE_30, ["name", "brand"], self.analyzer)
|
|
|
70 |
query = qp.parse(qp, query_str)
|
|
|
71 |
hits = self.searcher.search(query, self.MAX_RESULTS)
|
|
|
72 |
|
|
|
73 |
# print "Found %d document(s) that matched query '%s':" % (hits.totalHits, query)
|
|
|
74 |
|
|
|
75 |
phones = []
|
|
|
76 |
|
|
|
77 |
for hit in hits.scoreDocs:
|
|
|
78 |
doc = self.searcher.doc(hit.doc)
|
|
|
79 |
phones.append({
|
|
|
80 |
"name": str(doc.get("name")),
|
|
|
81 |
"price": str(doc.get("price")),
|
|
|
82 |
'score': hit.score,
|
|
|
83 |
'in_stock': str(doc.get("in_stock")),
|
|
|
84 |
'url': str(doc.get("url"))
|
|
|
85 |
})
|
|
|
86 |
return phones
|
|
|
87 |
|
|
|
88 |
|
|
|
89 |
if __name__ == "__main__":
|
|
|
90 |
retriever = Retriever()
|
|
|
91 |
print retriever.retrieve(model_number="M-6350", brand="Spice")
|
|
|
92 |
'''
|
|
|
93 |
catalog_client = CatalogClient().get_client()
|
|
|
94 |
items = catalog_client.getAllItems(True)
|
|
|
95 |
products = {}
|
|
|
96 |
|
|
|
97 |
for item in items:
|
|
|
98 |
if item.category in (10002, 10003, 10004, 10005): products[item.catalogItemId] = item
|
|
|
99 |
|
|
|
100 |
for key, item in products.iteritems():
|
|
|
101 |
try:
|
|
|
102 |
name = "%s %s" % (item.modelName, item.modelNumber)
|
|
|
103 |
matching_phones_with_top_score = getItemsWithTopScore(retriever.retrieve(name))
|
|
|
104 |
print name, '\n', isPriceSame(matching_phones_with_top_score), matching_phones_with_top_score
|
|
|
105 |
except Exception as e:
|
|
|
106 |
print e
|
|
|
107 |
'''
|