Subversion Repositories SmartDukaan

Rev

Rev 3232 | Rev 4039 | Go to most recent revision | 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
from lucene import File, StandardAnalyzer, IndexWriter, Version, SimpleFSDirectory, Document, Field
7
from Clients import GAEServletClient
8
from PhonePriceDoc import PhonePriceDoc
9
import lucene, Utils
10
 
11
class IndexBuilder:
12
 
3313 varun.gupt 13
    def __init__(self, price_data, new_index = True):
3232 varun.gupt 14
        self.indexDir = "/tmp/lucene-index-dir"
15
        lucene.initVM()
16
        dir = SimpleFSDirectory(File(self.indexDir))
17
        self.analyzer = StandardAnalyzer(Version.LUCENE_30)
3313 varun.gupt 18
        self.writer = IndexWriter(dir, self.analyzer, new_index, IndexWriter.MaxFieldLength(512))
3232 varun.gupt 19
        self.price_data = price_data
20
 
21
    def build(self):
22
        print "Currently there are %d documents in the index..." % self.writer.numDocs()
3313 varun.gupt 23
        count = 0
3232 varun.gupt 24
 
25
        for phone_price in self.price_data:
26
            print phone_price
27
            #doc = PhonePriceDoc(phone_price)
3313 varun.gupt 28
            count += 1
29
 
3232 varun.gupt 30
            brand, name = Utils.extractBrandAndName(str(phone_price['name']))
31
            doc = Document()
32
            doc.add(Field("name", name, Field.Store.YES, Field.Index.ANALYZED))
33
            doc.add(Field("brand", brand, Field.Store.YES, Field.Index.ANALYZED))
34
            doc.add(Field("source", str(phone_price['source']), Field.Store.YES, Field.Index.ANALYZED))
35
            doc.add(Field("price", str(phone_price['price']), Field.Store.YES, Field.Index.NO))
36
            doc.add(Field("in_stock", str(phone_price['in_stock']), Field.Store.YES, Field.Index.NO))
37
            doc.add(Field("url", str(phone_price['url']), Field.Store.YES, Field.Index.NO))
38
 
39
            self.writer.addDocument(doc)
3313 varun.gupt 40
 
3232 varun.gupt 41
        print "Indexed lines from stdin (%d documents in index)" % (self.writer.numDocs())
42
        print "About to optimize index of %d documents..." % self.writer.numDocs()
43
        self.writer.optimize()
44
        print "...done optimizing index of %d documents" % self.writer.numDocs()
45
        print "Closing index of %d documents..." % self.writer.numDocs()
3313 varun.gupt 46
        print "%d docs added" % count
3232 varun.gupt 47
        self.writer.close()
48
 
49
if __name__ == '__main__':
50
    phones = GAEServletClient.getPhonePricesJSON()
3313 varun.gupt 51
#    print phones
52
    indexer = IndexBuilder(price_data = phones, new_index = False)
3232 varun.gupt 53
    indexer.build()
54
#    catalog_client = InventoryClient().get_client()
55
#    items = catalog_client.getAllItems(True)
56
#    print phones
57
#    for item in items:  print item.id, item.brand, item.modelName, item.modelNumber