package org.apache.lucene.search.function; /** * Licensed to the Apache Software Foundation (ASF) under one or more * contributor license agreements. See the NOTICE file distributed with * this work for additional information regarding copyright ownership. * The ASF licenses this file to You under the Apache License, Version 2.0 * (the "License"); you may not use this file except in compliance with * the License. You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ import org.apache.lucene.index.CorruptIndexException; import org.apache.lucene.index.IndexReader; import org.apache.lucene.search.IndexSearcher; import org.apache.lucene.search.Query; import org.apache.lucene.search.QueryUtils; import org.apache.lucene.search.ScoreDoc; import org.apache.lucene.search.TopDocs; /** * Test search based on OrdFieldSource and ReverseOrdFieldSource. * <p> * Tests here create an index with a few documents, each having * an indexed "id" field. * The ord values of this field are later used for scoring. * <p> * The order tests use Hits to verify that docs are ordered as expected. * <p> * The exact score tests use TopDocs top to verify the exact score. */ public class TestOrdValues extends FunctionTestSetup { /* @override constructor */ public TestOrdValues(String name) { super(name, false); } /** Test OrdFieldSource */ public void testOrdFieldRank () throws CorruptIndexException, Exception { doTestRank(ID_FIELD,true); } /** Test ReverseOrdFieldSource */ public void testReverseOrdFieldRank () throws CorruptIndexException, Exception { doTestRank(ID_FIELD,false); } // Test that queries based on reverse/ordFieldScore scores correctly private void doTestRank (String field, boolean inOrder) throws CorruptIndexException, Exception { IndexSearcher s = new IndexSearcher(dir, true); ValueSource vs; if (inOrder) { vs = new OrdFieldSource(field); } else { vs = new ReverseOrdFieldSource(field); } Query q = new ValueSourceQuery(vs); log("test: "+q); QueryUtils.check(q,s); ScoreDoc[] h = s.search(q, null, 1000).scoreDocs; assertEquals("All docs should be matched!",N_DOCS,h.length); String prevID = inOrder ? "IE" // greater than all ids of docs in this test ("ID0001", etc.) : "IC"; // smaller than all ids of docs in this test ("ID0001", etc.) for (int i=0; i<h.length; i++) { String resID = s.doc(h[i].doc).get(ID_FIELD); log(i+". score="+h[i].score+" - "+resID); log(s.explain(q,h[i].doc)); if (inOrder) { assertTrue("res id "+resID+" should be < prev res id "+prevID, resID.compareTo(prevID)<0); } else { assertTrue("res id "+resID+" should be > prev res id "+prevID, resID.compareTo(prevID)>0); } prevID = resID; } } /** Test exact score for OrdFieldSource */ public void testOrdFieldExactScore () throws CorruptIndexException, Exception { doTestExactScore(ID_FIELD,true); } /** Test exact score for ReverseOrdFieldSource */ public void testReverseOrdFieldExactScore () throws CorruptIndexException, Exception { doTestExactScore(ID_FIELD,false); } // Test that queries based on reverse/ordFieldScore returns docs with expected score. private void doTestExactScore (String field, boolean inOrder) throws CorruptIndexException, Exception { IndexSearcher s = new IndexSearcher(dir, true); ValueSource vs; if (inOrder) { vs = new OrdFieldSource(field); } else { vs = new ReverseOrdFieldSource(field); } Query q = new ValueSourceQuery(vs); TopDocs td = s.search(q,null,1000); assertEquals("All docs should be matched!",N_DOCS,td.totalHits); ScoreDoc sd[] = td.scoreDocs; for (int i=0; i<sd.length; i++) { float score = sd[i].score; String id = s.getIndexReader().document(sd[i].doc).get(ID_FIELD); log("-------- "+i+". Explain doc "+id); log(s.explain(q,sd[i].doc)); float expectedScore = N_DOCS-i; assertEquals("score of result "+i+" shuould be "+expectedScore+" != "+score, expectedScore, score, TEST_SCORE_TOLERANCE_DELTA); String expectedId = inOrder ? id2String(N_DOCS-i) // in-order ==> larger values first : id2String(i+1); // reverse ==> smaller values first assertTrue("id of result "+i+" shuould be "+expectedId+" != "+score, expectedId.equals(id)); } } /** Test caching OrdFieldSource */ public void testCachingOrd () throws CorruptIndexException, Exception { doTestCaching(ID_FIELD,true); } /** Test caching for ReverseOrdFieldSource */ public void tesCachingReverseOrd () throws CorruptIndexException, Exception { doTestCaching(ID_FIELD,false); } // Test that values loaded for FieldScoreQuery are cached properly and consumes the proper RAM resources. private void doTestCaching (String field, boolean inOrder) throws CorruptIndexException, Exception { IndexSearcher s = new IndexSearcher(dir, true); Object innerArray = null; boolean warned = false; // print warning once for (int i=0; i<10; i++) { ValueSource vs; if (inOrder) { vs = new OrdFieldSource(field); } else { vs = new ReverseOrdFieldSource(field); } ValueSourceQuery q = new ValueSourceQuery(vs); ScoreDoc[] h = s.search(q, null, 1000).scoreDocs; try { assertEquals("All docs should be matched!",N_DOCS,h.length); IndexReader[] readers = s.getIndexReader().getSequentialSubReaders(); for(int j = 0; j < readers.length; j++) { IndexReader reader = readers[j]; if (i==0) { innerArray = q.valSrc.getValues(reader).getInnerArray(); } else { log(i+". compare: "+innerArray+" to "+q.valSrc.getValues(reader).getInnerArray()); assertSame("field values should be cached and reused!", innerArray, q.valSrc.getValues(reader).getInnerArray()); } } } catch (UnsupportedOperationException e) { if (!warned) { System.err.println("WARNING: "+testName()+" cannot fully test values of "+q); warned = true; } } } ValueSource vs; ValueSourceQuery q; ScoreDoc[] h; // verify that different values are loaded for a different field String field2 = INT_FIELD; assertFalse(field.equals(field2)); // otherwise this test is meaningless. if (inOrder) { vs = new OrdFieldSource(field2); } else { vs = new ReverseOrdFieldSource(field2); } q = new ValueSourceQuery(vs); h = s.search(q, null, 1000).scoreDocs; assertEquals("All docs should be matched!",N_DOCS,h.length); IndexReader[] readers = s.getIndexReader().getSequentialSubReaders(); for (int j = 0; j < readers.length; j++) { IndexReader reader = readers[j]; try { log("compare (should differ): " + innerArray + " to " + q.valSrc.getValues(reader).getInnerArray()); assertNotSame( "different values shuold be loaded for a different field!", innerArray, q.valSrc.getValues(reader).getInnerArray()); } catch (UnsupportedOperationException e) { if (!warned) { System.err.println("WARNING: " + testName() + " cannot fully test values of " + q); warned = true; } } } // verify new values are reloaded (not reused) for a new reader s = new IndexSearcher(dir, true); if (inOrder) { vs = new OrdFieldSource(field); } else { vs = new ReverseOrdFieldSource(field); } q = new ValueSourceQuery(vs); h = s.search(q, null, 1000).scoreDocs; assertEquals("All docs should be matched!",N_DOCS,h.length); readers = s.getIndexReader().getSequentialSubReaders(); for (int j = 0; j < readers.length; j++) { IndexReader reader = readers[j]; try { log("compare (should differ): " + innerArray + " to " + q.valSrc.getValues(reader).getInnerArray()); assertNotSame( "cached field values should not be reused if reader as changed!", innerArray, q.valSrc.getValues(reader).getInnerArray()); } catch (UnsupportedOperationException e) { if (!warned) { System.err.println("WARNING: " + testName() + " cannot fully test values of " + q); warned = true; } } } } private String testName() { return getClass().getName()+"."+getName(); } }