Subversion Repositories SmartDukaan

'''

Created on 30-Aug-2011

@author: Varun Gupta

'''

import lucene

from lucene import SimpleFSDirectory, File, StandardAnalyzer, IndexSearcher, Version, MultiFieldQueryParser, QueryParser

class Retriever:

    def __init__(self):

        self.indexDir = "/tmp/lucene-index-dir"

        lucene.initVM()

        dir = SimpleFSDirectory(File(self.indexDir))

        self.analyzer = StandardAnalyzer(Version.LUCENE_30)

        self.searcher = IndexSearcher(dir)

        self.MAX_RESULTS = 1000

    def retrieve(self, model_number, model_name = None, brand = None, synonyms = None):

        if model_number is None:  return

        print model_number, model_name, brand

        query_str = ''

        if brand is not None:

            query_brand = "brand:%s" % brand

        if model_name is not None:

            name_chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_name.split(' '))

            if brand is not None:   query_str += ' AND '

            query_model_name = ' OR '.join(name_chunks)

        else:

            query_model_name = ''

        if model_number is not None:

            chunks = filter(lambda chunk: len(chunk.strip()) > 0, model_number.split(' '))

            for i in range(0, len(chunks)):

                if chunks[i].find('-') > -1:

                    chunk_parts = chunks[i].split('-')

                    chunks.append(''.join(chunk_parts))

                    chunks.append(' '.join(chunk_parts))

            query_model_number = (' OR '.join(chunks)).strip()

        else:

            query_model_number = ''

        query_synonyms = ' '.join(synonyms) if synonyms is not None else ''

        if query_model_name:

            query_str = "%s AND (%s" % (query_brand, query_model_name)

        else:

            query_str = "%s AND (" % query_brand

        if query_model_number:

            if query_model_name:

                query_str += " OR %s %s)" % (query_model_number, query_synonyms)

            else:

                query_str += " %s %s)" % (query_model_number, query_synonyms)

        else:

            query_str += " %s)" % query_synonyms

        print query_str

#        query = QueryParser(Version.LUCENE_30, "name", analyzer).parse(user_query)

        qp = MultiFieldQueryParser(Version.LUCENE_30, ["name", "brand"], self.analyzer)

        query = qp.parse(qp, query_str)

        hits = self.searcher.search(query, self.MAX_RESULTS)

#        print "Found %d document(s) that matched query '%s':" % (hits.totalHits, query)

        phones = []

        for hit in hits.scoreDocs:

            doc = self.searcher.doc(hit.doc)

            phones.append({

                           "name": str(doc.get("name")),

                           "price": str(doc.get("price")),

                           'score': hit.score,

                           'in_stock': str(doc.get("in_stock")),

                           'url': str(doc.get("url")),

                           'source': str(doc.get("source"))

                           })

        return phones

    def retrieveForQuery(self, query_str):

        qp = MultiFieldQueryParser(Version.LUCENE_30, ["name", "brand"], self.analyzer)

        query = qp.parse(qp, query_str)

        hits = self.searcher.search(query, self.MAX_RESULTS)

#        print "Found %d document(s) that matched query '%s':" % (hits.totalHits, query)

        phones = []

        for hit in hits.scoreDocs:

            doc = self.searcher.doc(hit.doc)

            phones.append({

                        'name': str(doc.get("name")),

                        'price': str(doc.get("price")),

                        'score': hit.score,

                        'in_stock': str(doc.get("in_stock")),

                        'url': str(doc.get("url")),

                        'source': str(doc.get("source"))

                    })

        return phones

if __name__ == "__main__":

    retriever = Retriever()

#    print retriever.retrieve(model_number="9300", model_name = 'Curve 3G', brand="BlackBerry")

    print retriever.retrieveForQuery('brand:Lava')

    #    print phone

    '''

    catalog_client = CatalogClient().get_client()

    items = catalog_client.getAllItems(True)

    products = {}

    for item in items:

        if item.category in (10002, 10003, 10004, 10005):  products[item.catalogItemId] = item

    for key, item in products.iteritems():

        try:

            name = "%s %s" % (item.modelName, item.modelNumber)

            matching_phones_with_top_score = getItemsWithTopScore(retriever.retrieve(name))

            print name, '\n', isPriceSame(matching_phones_with_top_score), matching_phones_with_top_score

        except Exception as e:

            print e

    '''