Exemple #1
0
        public virtual void TestPayloadSpanUtil()
        {
            Directory         directory = NewDirectory();
            RandomIndexWriter writer    = new RandomIndexWriter(Random(), directory, NewIndexWriterConfig(TEST_VERSION_CURRENT, new PayloadAnalyzer(this)).SetSimilarity(Similarity));

            Document doc = new Document();

            doc.Add(NewTextField(PayloadHelper.FIELD, "xx rr yy mm  pp", Field.Store.YES));
            writer.AddDocument(doc);

            IndexReader reader = writer.Reader;

            writer.Dispose();
            IndexSearcher searcher = NewSearcher(reader);

            PayloadSpanUtil psu = new PayloadSpanUtil(searcher.TopReaderContext);

            var payloads = psu.GetPayloadsForQuery(new TermQuery(new Term(PayloadHelper.FIELD, "rr")));

            if (VERBOSE)
            {
                Console.WriteLine("Num payloads:" + payloads.Count);
                foreach (var bytes in payloads)
                {
                    Console.WriteLine(Encoding.UTF8.GetString((byte[])(Array)bytes));
                }
            }
            reader.Dispose();
            directory.Dispose();
        }
        public virtual void TestPayloadsPos0()
        {
            Directory dir = NewDirectory();
            RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, new MockPayloadAnalyzer());
            Document doc = new Document();
            doc.Add(new TextField("content", new StringReader("a a b c d e a f g h i j a b k k")));
            writer.AddDocument(doc);

            IndexReader readerFromWriter = writer.Reader;
            AtomicReader r = SlowCompositeReaderWrapper.Wrap(readerFromWriter);

            DocsAndPositionsEnum tp = r.TermPositionsEnum(new Term("content", "a"));

            int count = 0;
            Assert.IsTrue(tp.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
            // "a" occurs 4 times
            Assert.AreEqual(4, tp.Freq());
            Assert.AreEqual(0, tp.NextPosition());
            Assert.AreEqual(1, tp.NextPosition());
            Assert.AreEqual(3, tp.NextPosition());
            Assert.AreEqual(6, tp.NextPosition());

            // only one doc has "a"
            Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, tp.NextDoc());

            IndexSearcher @is = NewSearcher(readerFromWriter);

            SpanTermQuery stq1 = new SpanTermQuery(new Term("content", "a"));
            SpanTermQuery stq2 = new SpanTermQuery(new Term("content", "k"));
            SpanQuery[] sqs = new SpanQuery[] { stq1, stq2 };
            SpanNearQuery snq = new SpanNearQuery(sqs, 30, false);

            count = 0;
            bool sawZero = false;
            if (VERBOSE)
            {
                Console.WriteLine("\ngetPayloadSpans test");
            }
            Search.Spans.Spans pspans = MultiSpansWrapper.Wrap(@is.TopReaderContext, snq);
            while (pspans.Next())
            {
                if (VERBOSE)
                {
                    Console.WriteLine("doc " + pspans.Doc() + ": span " + pspans.Start() + " to " + pspans.End());
                }
                var payloads = pspans.Payload;
                sawZero |= pspans.Start() == 0;
                foreach (var bytes in payloads)
                {
                    count++;
                    if (VERBOSE)
                    {
                        Console.WriteLine("  payload: " + Encoding.UTF8.GetString((byte[])(Array)bytes));
                    }
                }
            }
            Assert.IsTrue(sawZero);
            Assert.AreEqual(5, count);

            // System.out.println("\ngetSpans test");
            Search.Spans.Spans spans = MultiSpansWrapper.Wrap(@is.TopReaderContext, snq);
            count = 0;
            sawZero = false;
            while (spans.Next())
            {
                count++;
                sawZero |= spans.Start() == 0;
                // System.out.println(spans.Doc() + " - " + spans.Start() + " - " +
                // spans.End());
            }
            Assert.AreEqual(4, count);
            Assert.IsTrue(sawZero);

            // System.out.println("\nPayloadSpanUtil test");

            sawZero = false;
            PayloadSpanUtil psu = new PayloadSpanUtil(@is.TopReaderContext);
            var pls = psu.GetPayloadsForQuery(snq);
            count = pls.Count;
            foreach (var bytes in pls)
            {
                string s = Encoding.UTF8.GetString(bytes);
                //System.out.println(s);
                sawZero |= s.Equals("pos: 0");
            }
            Assert.AreEqual(5, count);
            Assert.IsTrue(sawZero);
            writer.Dispose();
            @is.IndexReader.Dispose();
            dir.Dispose();
        }
Exemple #3
0
        public virtual void TestPayloadsPos0()
        {
            Directory         dir    = NewDirectory();
            RandomIndexWriter writer = new RandomIndexWriter(Random, dir, new MockPayloadAnalyzer());
            Document          doc    = new Document();

            doc.Add(new TextField("content", new StringReader("a a b c d e a f g h i j a b k k")));
            writer.AddDocument(doc);

            IndexReader  readerFromWriter = writer.GetReader();
            AtomicReader r = SlowCompositeReaderWrapper.Wrap(readerFromWriter);

            DocsAndPositionsEnum tp = r.GetTermPositionsEnum(new Term("content", "a"));

            int count = 0;

            Assert.IsTrue(tp.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
            // "a" occurs 4 times
            Assert.AreEqual(4, tp.Freq);
            Assert.AreEqual(0, tp.NextPosition());
            Assert.AreEqual(1, tp.NextPosition());
            Assert.AreEqual(3, tp.NextPosition());
            Assert.AreEqual(6, tp.NextPosition());

            // only one doc has "a"
            Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, tp.NextDoc());

            IndexSearcher @is = NewSearcher(readerFromWriter);

            SpanTermQuery stq1 = new SpanTermQuery(new Term("content", "a"));
            SpanTermQuery stq2 = new SpanTermQuery(new Term("content", "k"));

            SpanQuery[]   sqs = new SpanQuery[] { stq1, stq2 };
            SpanNearQuery snq = new SpanNearQuery(sqs, 30, false);

            count = 0;
            bool sawZero = false;

            if (Verbose)
            {
                Console.WriteLine("\ngetPayloadSpans test");
            }
            Search.Spans.Spans pspans = MultiSpansWrapper.Wrap(@is.TopReaderContext, snq);
            while (pspans.MoveNext())
            {
                if (Verbose)
                {
                    Console.WriteLine("doc " + pspans.Doc + ": span " + pspans.Start + " to " + pspans.End);
                }
                var payloads = pspans.GetPayload();
                sawZero |= pspans.Start == 0;
                foreach (var bytes in payloads)
                {
                    count++;
                    if (Verbose)
                    {
                        Console.WriteLine("  payload: " + Encoding.UTF8.GetString(bytes));
                    }
                }
            }
            Assert.IsTrue(sawZero);
            Assert.AreEqual(5, count);

            // System.out.println("\ngetSpans test");
            Search.Spans.Spans spans = MultiSpansWrapper.Wrap(@is.TopReaderContext, snq);
            count   = 0;
            sawZero = false;
            while (spans.MoveNext())
            {
                count++;
                sawZero |= spans.Start == 0;
                // System.out.println(spans.Doc() + " - " + spans.Start() + " - " +
                // spans.End());
            }
            Assert.AreEqual(4, count);
            Assert.IsTrue(sawZero);

            // System.out.println("\nPayloadSpanUtil test");

            sawZero = false;
            PayloadSpanUtil psu = new PayloadSpanUtil(@is.TopReaderContext);
            var             pls = psu.GetPayloadsForQuery(snq);

            count = pls.Count;
            foreach (var bytes in pls)
            {
                string s = Encoding.UTF8.GetString(bytes);
                //System.out.println(s);
                sawZero |= s.Equals("pos: 0", StringComparison.Ordinal);
            }
            Assert.AreEqual(5, count);
            Assert.IsTrue(sawZero);
            writer.Dispose();
            @is.IndexReader.Dispose();
            dir.Dispose();
        }
        public virtual void TestPayloadSpanUtil()
        {
            Directory directory = NewDirectory();
            RandomIndexWriter writer = new RandomIndexWriter(Random(), directory, NewIndexWriterConfig(TEST_VERSION_CURRENT, new PayloadAnalyzer(this)).SetSimilarity(Similarity));

            Document doc = new Document();
            doc.Add(NewTextField(PayloadHelper.FIELD, "xx rr yy mm  pp", Field.Store.YES));
            writer.AddDocument(doc);

            IndexReader reader = writer.Reader;
            writer.Dispose();
            IndexSearcher searcher = NewSearcher(reader);

            PayloadSpanUtil psu = new PayloadSpanUtil(searcher.TopReaderContext);

            ICollection<sbyte[]> payloads = psu.GetPayloadsForQuery(new TermQuery(new Term(PayloadHelper.FIELD, "rr")));
            if (VERBOSE)
            {
                Console.WriteLine("Num payloads:" + payloads.Count);
                foreach (sbyte[] bytes in payloads)
                {
                    Console.WriteLine(Encoding.UTF8.GetString((byte[])(Array)bytes));
                }
            }
            reader.Dispose();
            directory.Dispose();
        }