예제 #1
0
        public virtual void TestMultipleSources()
        {
            TeeSinkTokenFilter tee1 = new TeeSinkTokenFilter(new MockTokenizer(new StringReader(buffer1.ToString()), MockTokenizer.WHITESPACE, false));

            TeeSinkTokenFilter.SinkTokenStream dogDetector = tee1.NewSinkTokenStream(dogFilter);
            TeeSinkTokenFilter.SinkTokenStream theDetector = tee1.NewSinkTokenStream(theFilter);
            tee1.Reset();
            TokenStream source1 = new CachingTokenFilter(tee1);

            tee1.AddAttribute <ICheckClearAttributesAttribute>();
            dogDetector.AddAttribute <ICheckClearAttributesAttribute>();
            theDetector.AddAttribute <ICheckClearAttributesAttribute>();

            TeeSinkTokenFilter tee2 = new TeeSinkTokenFilter(new MockTokenizer(new StringReader(buffer2.ToString()), MockTokenizer.WHITESPACE, false));

            tee2.AddSinkTokenStream(dogDetector);
            tee2.AddSinkTokenStream(theDetector);
            TokenStream source2 = tee2;

            AssertTokenStreamContents(source1, tokens1);
            AssertTokenStreamContents(source2, tokens2);

            AssertTokenStreamContents(theDetector, new string[] { "The", "the", "The", "the" });
            AssertTokenStreamContents(dogDetector, new string[] { "Dogs", "Dogs" });

            source1.Reset();
            TokenStream lowerCasing = new LowerCaseFilter(TEST_VERSION_CURRENT, source1);

            string[] lowerCaseTokens = new string[tokens1.Length];
            for (int i = 0; i < tokens1.Length; i++)
            {
                lowerCaseTokens[i] = CultureInfo.InvariantCulture.TextInfo.ToLower(tokens1[i]);
            }
            AssertTokenStreamContents(lowerCasing, lowerCaseTokens);
        }
예제 #2
0
        public virtual void Test()
        {
            TokenTypeSinkFilter sinkFilter = new TokenTypeSinkFilter("D");
            string test = "The quick red fox jumped over the lazy brown dogs";

            TeeSinkTokenFilter ttf = new TeeSinkTokenFilter(new WordTokenFilter(new MockTokenizer(new StringReader(test), MockTokenizer.WHITESPACE, false)));

            TeeSinkTokenFilter.SinkTokenStream sink = ttf.NewSinkTokenStream(sinkFilter);

            bool seenDogs = false;

            ICharTermAttribute termAtt = ttf.AddAttribute <ICharTermAttribute>();
            ITypeAttribute     typeAtt = ttf.AddAttribute <ITypeAttribute>();

            ttf.Reset();
            while (ttf.IncrementToken())
            {
                if (termAtt.ToString().Equals("dogs", StringComparison.Ordinal))
                {
                    seenDogs = true;
                    assertTrue(typeAtt.Type + " is not equal to " + "D", typeAtt.Type.Equals("D", StringComparison.Ordinal) == true);
                }
                else
                {
                    assertTrue(typeAtt.Type + " is not null and it should be", typeAtt.Type.Equals("word", StringComparison.Ordinal));
                }
            }
            assertTrue(seenDogs + " does not equal: " + true, seenDogs == true);

            int sinkCount = 0;

            sink.Reset();
            while (sink.IncrementToken())
            {
                sinkCount++;
            }

            assertTrue("sink Size: " + sinkCount + " is not: " + 1, sinkCount == 1);
        }
예제 #3
0
        public virtual void TestGeneral()
        {
            TeeSinkTokenFilter source = new TeeSinkTokenFilter(new MockTokenizer(new StringReader(buffer1.ToString()), MockTokenizer.WHITESPACE, false));
            TokenStream        sink1  = source.NewSinkTokenStream();
            TokenStream        sink2  = source.NewSinkTokenStream(theFilter);

            source.AddAttribute <ICheckClearAttributesAttribute>();
            sink1.AddAttribute <ICheckClearAttributesAttribute>();
            sink2.AddAttribute <ICheckClearAttributesAttribute>();

            AssertTokenStreamContents(source, tokens1);
            AssertTokenStreamContents(sink1, tokens1);
            AssertTokenStreamContents(sink2, new string[] { "The", "the" });
        }
        public virtual void Test()
        {
            TokenTypeSinkFilter sinkFilter = new TokenTypeSinkFilter("D");
            string test = "The quick red fox jumped over the lazy brown dogs";

            TeeSinkTokenFilter ttf = new TeeSinkTokenFilter(new WordTokenFilter(this, new MockTokenizer(new StringReader(test), MockTokenizer.WHITESPACE, false)));
            TeeSinkTokenFilter.SinkTokenStream sink = ttf.NewSinkTokenStream(sinkFilter);

            bool seenDogs = false;

            ICharTermAttribute termAtt = ttf.AddAttribute<ICharTermAttribute>();
            ITypeAttribute typeAtt = ttf.AddAttribute<ITypeAttribute>();
            ttf.Reset();
            while (ttf.IncrementToken())
            {
                if (termAtt.ToString().Equals("dogs"))
                {
                    seenDogs = true;
                    assertTrue(typeAtt.Type + " is not equal to " + "D", typeAtt.Type.Equals("D") == true);
                }
                else
                {
                    assertTrue(typeAtt.Type + " is not null and it should be", typeAtt.Type.Equals("word"));
                }
            }
            assertTrue(seenDogs + " does not equal: " + true, seenDogs == true);

            int sinkCount = 0;
            sink.Reset();
            while (sink.IncrementToken())
            {
                sinkCount++;
            }

            assertTrue("sink Size: " + sinkCount + " is not: " + 1, sinkCount == 1);
        }