samples/LuceneInAction/lia/analysis/keyword/KeywordAnalyzerTest.py

   1 # ====================================================================
   2 #   Licensed under the Apache License, Version 2.0 (the "License");
   3 #   you may not use this file except in compliance with the License.
   4 #   You may obtain a copy of the License at
   5 #
   6 #       http://www.apache.org/licenses/LICENSE-2.0
   7 #
   8 #   Unless required by applicable law or agreed to in writing, software
   9 #   distributed under the License is distributed on an "AS IS" BASIS,
  10 #   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  11 #   See the License for the specific language governing permissions and
  12 #   limitations under the License.
  13 # ====================================================================
  14
  15 from unittest import TestCase
  16
  17 from lucene import \
  18      IndexWriter, Term, SimpleAnalyzer, PerFieldAnalyzerWrapper, \
  19      RAMDirectory, Document, Field, IndexSearcher, TermQuery, \
  20      QueryParser, Analyzer, StringReader, Token, JavaError, \
  21      Version
  22
  23 from lia.analysis.keyword.KeywordAnalyzer import KeywordAnalyzer
  24 from lia.analysis.keyword.SimpleKeywordAnalyzer import SimpleKeywordAnalyzer
  25
  26
  27 class KeywordAnalyzerTest(TestCase):
  28
  29     def setUp(self):
  30
  31         self.directory = RAMDirectory()
  32         writer = IndexWriter(self.directory, SimpleAnalyzer(), True,
  33                              IndexWriter.MaxFieldLength.UNLIMITED)
  34
  35         doc = Document()
  36         doc.add(Field("partnum", "Q36",
  37                       Field.Store.YES, Field.Index.NOT_ANALYZED))
  38         doc.add(Field("description", "Illidium Space Modulator",
  39                       Field.Store.YES, Field.Index.ANALYZED))
  40         writer.addDocument(doc)
  41         writer.close()
  42
  43         self.searcher = IndexSearcher(self.directory, True)
  44
  45     def testTermQuery(self):
  46
  47         query = TermQuery(Term("partnum", "Q36"))
  48         scoreDocs = self.searcher.search(query, 50).scoreDocs
  49         self.assertEqual(1, len(scoreDocs))
  50
  51     def testBasicQueryParser(self):
  52
  53         analyzer = SimpleAnalyzer()
  54         query = QueryParser(Version.LUCENE_CURRENT, "description",
  55                             analyzer).parse("partnum:Q36 AND SPACE")
  56
  57         scoreDocs = self.searcher.search(query, 50).scoreDocs
  58         self.assertEqual("+partnum:q +space", query.toString("description"),
  59                          "note Q36 -> q")
  60         self.assertEqual(0, len(scoreDocs), "doc not found :(")
  61
  62     def testPerFieldAnalyzer(self):
  63
  64         analyzer = PerFieldAnalyzerWrapper(SimpleAnalyzer())
  65         analyzer.addAnalyzer("partnum", KeywordAnalyzer())
  66
  67         query = QueryParser(Version.LUCENE_CURRENT, "description",
  68                             analyzer).parse("partnum:Q36 AND SPACE")
  69         scoreDocs = self.searcher.search(query, 50).scoreDocs
  70
  71         #self.assertEqual("+partnum:Q36 +space", query.toString("description"))
  72         self.assertEqual(1, len(scoreDocs), "doc found!")