samples/LuceneInAction/lia/advsearching/FilterTest.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 lia.common.LiaTestCase import LiaTestCase
  16
  17 from lucene import Integer, \
  18      IndexSearcher, Term, TermQuery, MatchAllDocsQuery, \
  19      BooleanQuery, BooleanClause, CachingWrapperFilter, \
  20      TermRangeFilter, NumericRangeFilter, FieldCacheRangeFilter, \
  21      FieldCacheTermsFilter, QueryWrapperFilter, PrefixFilter
  22
  23
  24 class FilterTest(LiaTestCase):
  25
  26     def setUp(self):
  27
  28         super(FilterTest, self).setUp()
  29
  30         self.allBooks = MatchAllDocsQuery()
  31         self.searcher = IndexSearcher(self.directory, True)
  32         scoreDocs = self.searcher.search(self.allBooks, 50).scoreDocs
  33         self.numAllBooks = len(scoreDocs)
  34
  35     def testTermRangeFilter(self):
  36
  37         filter = TermRangeFilter("title2", "d", "j", True, True)
  38         scoreDocs = self.searcher.search(self.allBooks, filter, 50).scoreDocs
  39         self.assertEqual(3, len(scoreDocs))
  40
  41     def testNumericDateFilter(self):
  42
  43         filter = NumericRangeFilter.newIntRange("pubmonth",
  44                                                 Integer(198805),
  45                                                 Integer(198810),
  46                                                 True, True)
  47         scoreDocs = self.searcher.search(self.allBooks, filter, 50).scoreDocs
  48         self.assertEqual(1, len(scoreDocs))
  49
  50     def testFieldCacheRangeFilter(self):
  51
  52         filter = FieldCacheRangeFilter.newStringRange("title2", "d", "j",
  53                                                       True, True)
  54         scoreDocs = self.searcher.search(self.allBooks, filter, 50).scoreDocs
  55         self.assertEqual(3, len(scoreDocs))
  56
  57         filter = FieldCacheRangeFilter.newIntRange("pubmonth",
  58                                                    Integer(198805),
  59                                                    Integer(198810),
  60                                                    True, True)
  61         scoreDocs = self.searcher.search(self.allBooks, filter, 50).scoreDocs
  62         self.assertEqual(1, len(scoreDocs))
  63
  64     def testFieldCacheTermsFilter(self):
  65
  66         filter = FieldCacheTermsFilter("category",
  67                                        ["/health/alternative/chinese",
  68                                         "/technology/computers/ai",
  69                                         "/technology/computers/programming"])
  70         scoreDocs = self.searcher.search(self.allBooks, filter, 50).scoreDocs
  71         self.assertEqual(7, len(scoreDocs), "expected 7 hits")
  72
  73     def testQueryWrapperFilter(self):
  74
  75         categoryQuery = TermQuery(Term("category", "/philosophy/eastern"))
  76         categoryFilter = QueryWrapperFilter(categoryQuery)
  77         scoreDocs = self.searcher.search(self.allBooks, categoryFilter, 50).scoreDocs
  78         self.assertEqual(1, len(scoreDocs), "only tao te ching")
  79
  80     def testSpanQueryFilter(self):
  81
  82         categoryQuery = TermQuery(Term("category", "/philosophy/eastern"))
  83         categoryFilter = QueryWrapperFilter(categoryQuery)
  84         scoreDocs = self.searcher.search(self.allBooks, categoryFilter, 50).scoreDocs
  85         self.assertEqual(1, len(scoreDocs), "only tao te ching")
  86
  87     def testFilterAlternative(self):
  88
  89         categoryQuery = TermQuery(Term("category", "/philosophy/eastern"))
  90
  91         constrainedQuery = BooleanQuery()
  92         constrainedQuery.add(self.allBooks, BooleanClause.Occur.MUST)
  93         constrainedQuery.add(categoryQuery, BooleanClause.Occur.MUST)
  94
  95         scoreDocs = self.searcher.search(constrainedQuery, 50).scoreDocs
  96         self.assertEqual(1, len(scoreDocs), "only tao te ching")
  97
  98     def testPrefixFilter(self):
  99
 100         prefixFilter = PrefixFilter(Term("category", "/technology/computers"))
 101         scoreDocs = self.searcher.search(self.allBooks, prefixFilter, 50).scoreDocs
 102         self.assertEqual(8, len(scoreDocs),
 103                          "only /technology/computers/* books")
 104
 105     def testCachingWrapper(self):
 106
 107         filter = TermRangeFilter("title2", "d", "j", True, True)
 108         cachingFilter = CachingWrapperFilter(filter)
 109         scoreDocs = self.searcher.search(self.allBooks, cachingFilter, 50).scoreDocs
 110         self.assertEqual(3, len(scoreDocs))