public virtual void QueriesTest(Query query, int[] expDocNrs) { TopScoreDocCollector collector = TopScoreDocCollector.Create(1000, false); Searcher.Search(query, null, collector); ScoreDoc[] hits1 = collector.GetTopDocs().ScoreDocs; collector = TopScoreDocCollector.Create(1000, true); Searcher.Search(query, null, collector); ScoreDoc[] hits2 = collector.GetTopDocs().ScoreDocs; Assert.AreEqual(MulFactor * collector.TotalHits, BigSearcher.Search(query, 1).TotalHits); CheckHits.CheckHitsQuery(query, hits1, hits2, expDocNrs); }
public virtual void VerifyNrHits(Query q, int expected) { // bs1 ScoreDoc[] h = s.Search(q, null, 1000).ScoreDocs; if (expected != h.Length) { PrintHits(TestName, h, s); } Assert.AreEqual(expected, h.Length, "result count"); //System.out.println("TEST: now check"); // bs2 TopScoreDocCollector collector = TopScoreDocCollector.Create(1000, true); s.Search(q, collector); ScoreDoc[] h2 = collector.GetTopDocs().ScoreDocs; if (expected != h2.Length) { PrintHits(TestName, h2, s); } Assert.AreEqual(expected, h2.Length, "result count (bs2)"); QueryUtils.Check( #if FEATURE_INSTANCE_TESTDATA_INITIALIZATION this, #endif Random, q, s); }
public virtual void TestBooleanSpanQuery() { bool failed = false; int hits = 0; Directory directory = NewDirectory(); Analyzer indexerAnalyzer = new MockAnalyzer(Random); IndexWriterConfig config = new IndexWriterConfig(TEST_VERSION_CURRENT, indexerAnalyzer); IndexWriter writer = new IndexWriter(directory, config); string FIELD = "content"; Document d = new Document(); d.Add(new TextField(FIELD, "clockwork orange", Field.Store.YES)); writer.AddDocument(d); writer.Dispose(); IndexReader indexReader = DirectoryReader.Open(directory); IndexSearcher searcher = NewSearcher(indexReader); BooleanQuery query = new BooleanQuery(); SpanQuery sq1 = new SpanTermQuery(new Term(FIELD, "clockwork")); SpanQuery sq2 = new SpanTermQuery(new Term(FIELD, "clckwork")); query.Add(sq1, Occur.SHOULD); query.Add(sq2, Occur.SHOULD); TopScoreDocCollector collector = TopScoreDocCollector.Create(1000, true); searcher.Search(query, collector); hits = collector.GetTopDocs().ScoreDocs.Length; foreach (ScoreDoc scoreDoc in collector.GetTopDocs().ScoreDocs) { Console.WriteLine(scoreDoc.Doc); } indexReader.Dispose(); Assert.AreEqual(failed, false, "Bug in boolean query composed of span queries"); Assert.AreEqual(hits, 1, "Bug in boolean query composed of span queries"); directory.Dispose(); }
/// <summary> /// Expert: Low-level search implementation. Finds the top <code>n</code> /// hits for <c>query</c>. /// /// <para/>Applications should usually call <see cref="IndexSearcher.Search(Query,int)"/> or /// <see cref="IndexSearcher.Search(Query,Filter,int)"/> instead. </summary> /// <exception cref="BooleanQuery.TooManyClausesException"> If a query would exceed /// <see cref="BooleanQuery.MaxClauseCount"/> clauses. </exception> protected virtual TopDocs Search(IList <AtomicReaderContext> leaves, Weight weight, ScoreDoc after, int nDocs) { // single thread int limit = reader.MaxDoc; if (limit == 0) { limit = 1; } nDocs = Math.Min(nDocs, limit); TopScoreDocCollector collector = TopScoreDocCollector.Create(nDocs, after, !weight.ScoresDocsOutOfOrder); Search(leaves, weight, collector); return(collector.GetTopDocs()); }
public virtual void VerifyNrHits(Query q, int expected) { // bs1 ScoreDoc[] h = s.Search(q, null, 1000).ScoreDocs; if (expected != h.Length) { PrintHits(TestName, h, s); } Assert.AreEqual(expected, h.Length, "result count"); //System.out.println("TEST: now check"); // bs2 TopScoreDocCollector collector = TopScoreDocCollector.Create(1000, true); s.Search(q, collector); ScoreDoc[] h2 = collector.GetTopDocs().ScoreDocs; if (expected != h2.Length) { PrintHits(TestName, h2, s); } Assert.AreEqual(expected, h2.Length, "result count (bs2)"); QueryUtils.Check(Random(), q, s, Similarity); }
internal virtual void TestSort(bool useFrom, bool VERBOSE) { IndexReader reader = null; Directory dir = null; if (!VERBOSE) { Console.WriteLine("Verbosity disabled. Enable manually if needed."); } int numDocs = VERBOSE ? AtLeast(50) : AtLeast(1000); //final int numDocs = AtLeast(50); string[] tokens = new string[] { "a", "b", "c", "d", "e" }; if (VERBOSE) { Console.WriteLine("TEST: make index"); } { dir = NewDirectory(); RandomIndexWriter w = new RandomIndexWriter( #if FEATURE_INSTANCE_TESTDATA_INITIALIZATION this, #endif Random, dir); // w.setDoRandomForceMerge(false); // w.w.getConfig().SetMaxBufferedDocs(AtLeast(100)); string[] content = new string[AtLeast(20)]; for (int contentIDX = 0; contentIDX < content.Length; contentIDX++) { StringBuilder sb = new StringBuilder(); int numTokens = TestUtil.NextInt32(Random, 1, 10); for (int tokenIDX = 0; tokenIDX < numTokens; tokenIDX++) { sb.Append(tokens[Random.Next(tokens.Length)]).Append(' '); } content[contentIDX] = sb.ToString(); } for (int docIDX = 0; docIDX < numDocs; docIDX++) { Document doc = new Document(); doc.Add(NewStringField("string", TestUtil.RandomRealisticUnicodeString(Random), Field.Store.NO)); doc.Add(NewTextField("text", content[Random.Next(content.Length)], Field.Store.NO)); doc.Add(new SingleField("float", (float)Random.NextDouble(), Field.Store.NO)); int intValue; if (Random.Next(100) == 17) { intValue = int.MinValue; } else if (Random.Next(100) == 17) { intValue = int.MaxValue; } else { intValue = Random.Next(); } doc.Add(new Int32Field("int", intValue, Field.Store.NO)); if (VERBOSE) { Console.WriteLine(" doc=" + doc); } w.AddDocument(doc); } reader = w.GetReader(); w.Dispose(); } // NOTE: sometimes reader has just one segment, which is // important to test IndexSearcher searcher = NewSearcher(reader); IndexReaderContext ctx = searcher.TopReaderContext; ShardSearcher[] subSearchers; int[] docStarts; if (ctx is AtomicReaderContext) { subSearchers = new ShardSearcher[1]; docStarts = new int[1]; subSearchers[0] = new ShardSearcher((AtomicReaderContext)ctx, ctx); docStarts[0] = 0; } else { CompositeReaderContext compCTX = (CompositeReaderContext)ctx; int size = compCTX.Leaves.Count; subSearchers = new ShardSearcher[size]; docStarts = new int[size]; int docBase = 0; for (int searcherIDX = 0; searcherIDX < subSearchers.Length; searcherIDX++) { AtomicReaderContext leave = compCTX.Leaves[searcherIDX]; subSearchers[searcherIDX] = new ShardSearcher(leave, compCTX); docStarts[searcherIDX] = docBase; docBase += leave.Reader.MaxDoc; } } IList <SortField> sortFields = new JCG.List <SortField>(); sortFields.Add(new SortField("string", SortFieldType.STRING, true)); sortFields.Add(new SortField("string", SortFieldType.STRING, false)); sortFields.Add(new SortField("int", SortFieldType.INT32, true)); sortFields.Add(new SortField("int", SortFieldType.INT32, false)); sortFields.Add(new SortField("float", SortFieldType.SINGLE, true)); sortFields.Add(new SortField("float", SortFieldType.SINGLE, false)); sortFields.Add(new SortField(null, SortFieldType.SCORE, true)); sortFields.Add(new SortField(null, SortFieldType.SCORE, false)); sortFields.Add(new SortField(null, SortFieldType.DOC, true)); sortFields.Add(new SortField(null, SortFieldType.DOC, false)); for (int iter = 0; iter < 1000 * RandomMultiplier; iter++) { // TODO: custom FieldComp... Query query = new TermQuery(new Term("text", tokens[Random.Next(tokens.Length)])); Sort sort; if (Random.Next(10) == 4) { // Sort by score sort = null; } else { SortField[] randomSortFields = new SortField[TestUtil.NextInt32(Random, 1, 3)]; for (int sortIDX = 0; sortIDX < randomSortFields.Length; sortIDX++) { randomSortFields[sortIDX] = sortFields[Random.Next(sortFields.Count)]; } sort = new Sort(randomSortFields); } int numHits = TestUtil.NextInt32(Random, 1, numDocs + 5); //final int numHits = 5; if (VERBOSE) { Console.WriteLine("TEST: search query=" + query + " sort=" + sort + " numHits=" + numHits); } int from = -1; int size = -1; // First search on whole index: TopDocs topHits; if (sort == null) { if (useFrom) { TopScoreDocCollector c = TopScoreDocCollector.Create(numHits, Random.NextBoolean()); searcher.Search(query, c); from = TestUtil.NextInt32(Random, 0, numHits - 1); size = numHits - from; TopDocs tempTopHits = c.GetTopDocs(); if (from < tempTopHits.ScoreDocs.Length) { // Can't use TopDocs#topDocs(start, howMany), since it has different behaviour when start >= hitCount // than TopDocs#merge currently has ScoreDoc[] newScoreDocs = new ScoreDoc[Math.Min(size, tempTopHits.ScoreDocs.Length - from)]; Array.Copy(tempTopHits.ScoreDocs, from, newScoreDocs, 0, newScoreDocs.Length); tempTopHits.ScoreDocs = newScoreDocs; topHits = tempTopHits; } else { topHits = new TopDocs(tempTopHits.TotalHits, new ScoreDoc[0], tempTopHits.MaxScore); } } else { topHits = searcher.Search(query, numHits); } } else { TopFieldCollector c = TopFieldCollector.Create(sort, numHits, true, true, true, Random.NextBoolean()); searcher.Search(query, c); if (useFrom) { from = TestUtil.NextInt32(Random, 0, numHits - 1); size = numHits - from; TopDocs tempTopHits = c.GetTopDocs(); if (from < tempTopHits.ScoreDocs.Length) { // Can't use TopDocs#topDocs(start, howMany), since it has different behaviour when start >= hitCount // than TopDocs#merge currently has ScoreDoc[] newScoreDocs = new ScoreDoc[Math.Min(size, tempTopHits.ScoreDocs.Length - from)]; Array.Copy(tempTopHits.ScoreDocs, from, newScoreDocs, 0, newScoreDocs.Length); tempTopHits.ScoreDocs = newScoreDocs; topHits = tempTopHits; } else { topHits = new TopDocs(tempTopHits.TotalHits, new ScoreDoc[0], tempTopHits.MaxScore); } } else { topHits = c.GetTopDocs(0, numHits); } } if (VERBOSE) { if (useFrom) { Console.WriteLine("from=" + from + " size=" + size); } Console.WriteLine(" top search: " + topHits.TotalHits + " totalHits; hits=" + (topHits.ScoreDocs == null ? "null" : topHits.ScoreDocs.Length + " maxScore=" + topHits.MaxScore)); if (topHits.ScoreDocs != null) { for (int hitIDX = 0; hitIDX < topHits.ScoreDocs.Length; hitIDX++) { ScoreDoc sd = topHits.ScoreDocs[hitIDX]; Console.WriteLine(" doc=" + sd.Doc + " score=" + sd.Score); } } } // ... then all shards: Weight w = searcher.CreateNormalizedWeight(query); TopDocs[] shardHits = new TopDocs[subSearchers.Length]; for (int shardIDX = 0; shardIDX < subSearchers.Length; shardIDX++) { TopDocs subHits; ShardSearcher subSearcher = subSearchers[shardIDX]; if (sort == null) { subHits = subSearcher.Search(w, numHits); } else { TopFieldCollector c = TopFieldCollector.Create(sort, numHits, true, true, true, Random.NextBoolean()); subSearcher.Search(w, c); subHits = c.GetTopDocs(0, numHits); } shardHits[shardIDX] = subHits; if (VERBOSE) { Console.WriteLine(" shard=" + shardIDX + " " + subHits.TotalHits + " totalHits hits=" + (subHits.ScoreDocs == null ? "null" : subHits.ScoreDocs.Length.ToString())); if (subHits.ScoreDocs != null) { foreach (ScoreDoc sd in subHits.ScoreDocs) { Console.WriteLine(" doc=" + sd.Doc + " score=" + sd.Score); } } } } // Merge: TopDocs mergedHits; if (useFrom) { mergedHits = TopDocs.Merge(sort, from, size, shardHits); } else { mergedHits = TopDocs.Merge(sort, numHits, shardHits); } if (mergedHits.ScoreDocs != null) { // Make sure the returned shards are correct: for (int hitIDX = 0; hitIDX < mergedHits.ScoreDocs.Length; hitIDX++) { ScoreDoc sd = mergedHits.ScoreDocs[hitIDX]; Assert.AreEqual(ReaderUtil.SubIndex(sd.Doc, docStarts), sd.ShardIndex, "doc=" + sd.Doc + " wrong shard"); } } TestUtil.AssertEquals(topHits, mergedHits); } reader.Dispose(); dir.Dispose(); }