Package org.apache.any23.writer

Examples of org.apache.any23.writer.CountingTripleHandler


                "/microformats/nested-microformats-a1.html",
                "http://www.test.com");

        ByteArrayOutputStream baos = new ByteArrayOutputStream();

        CountingTripleHandler cth1 = new CountingTripleHandler();
        RDFXMLWriter ctw1 = new RDFXMLWriter(baos);
        CompositeTripleHandler compositeTH1 = new CompositeTripleHandler();
        compositeTH1.addChild(cth1);
        compositeTH1.addChild(ctw1);
        runner.extract(
                new ExtractionParameters(DefaultConfiguration.singleton(),
                        ValidationMode.None, true), source, compositeTH1);
        compositeTH1.close();
        logger.debug("Out1: " + baos.toString());
        Assert.assertEquals("Unexpected number of triples.",
                EXPECTED_TRIPLES + 3, cth1.getCount());

        baos.reset();
        CountingTripleHandler cth2 = new CountingTripleHandler();
        NTriplesWriter ctw2 = new NTriplesWriter(baos);
        CompositeTripleHandler compositeTH2 = new CompositeTripleHandler();
        compositeTH2.addChild(cth2);
        compositeTH2.addChild(ctw2);
        runner.extract(
                new ExtractionParameters(DefaultConfiguration.singleton(),
                        ValidationMode.ValidateAndFix, false), source,
                compositeTH2);
        compositeTH2.close();
        logger.debug("Out2: " + baos.toString());
        Assert.assertEquals("Unexpected number of triples.", EXPECTED_TRIPLES,
                cth2.getCount());
    }
View Full Code Here


    @Test
    public void testExceptionPropagation() throws IOException {
        Any23 any23 = new Any23();
        DocumentSource source = getDocumentSourceFromResource(
                "/application/turtle/geolinkeddata.ttl", "http://www.test.com");
        CountingTripleHandler cth1 = new CountingTripleHandler();
        try {
            any23.extract(source, cth1);
        } catch (ExtractionException e) {
            Assert.assertTrue(e.getCause() instanceof RDFParseException);
        }
View Full Code Here

        final String in = StreamUtils.asString(this.getClass()
                .getResourceAsStream("any23-xml-mimetype.xml"));
        final DocumentSource doc = new StringDocumentSource(in, documentURI,
                contentType);
        final Any23 any23 = new Any23();
        final CountingTripleHandler cth = new CountingTripleHandler(false);
        final ReportingTripleHandler rth = new ReportingTripleHandler(cth);
        final ExtractionReport report = any23.extract(doc, rth);
        Assert.assertFalse(report.hasMatchingExtractors());
        Assert.assertEquals(0, cth.getCount());
    }
View Full Code Here

    public void testXMLMimeTypeManagementViaURL() throws IOException,
            ExtractionException {
        assumeOnlineAllowed();
        final Any23 any23 = new Any23();
        any23.setHTTPUserAgent("test-user-agent");
        final CountingTripleHandler cth = new CountingTripleHandler(false);
        final ReportingTripleHandler rth = new ReportingTripleHandler(cth);
        final ExtractionReport report = any23.extract(
                "http://www.nativeremedies.com/XML/combos.xml", rth);
        Assert.assertFalse(report.hasMatchingExtractors());
        Assert.assertEquals(0, cth.getCount());
    }
View Full Code Here

    @Test
    public void testBlankNodesViaURL() throws IOException, ExtractionException {
        assumeOnlineAllowed();
        final Any23 any23 = new Any23();
        any23.setHTTPUserAgent("test-user-agent");
        final CountingTripleHandler cth = new CountingTripleHandler(false);
        final ReportingTripleHandler rth = new ReportingTripleHandler(cth);
        final ExtractionReport report = any23.extract(
                "http://www.usarab.org/news/?tag=england", rth);
        Assert.assertTrue(report.hasMatchingExtractors());
    }
View Full Code Here

        this.in = in;
        this.extractors = extractors;

        List<TripleHandler> tripleHandlers = new ArrayList<TripleHandler>();
        tripleHandlers.add(output);
        tripleHandlers.add(new CountingTripleHandler());
        this.output = new CompositeTripleHandler(tripleHandlers);
        this.encoderDetector = new TikaEncodingDetector();
    }
View Full Code Here

                    null,
                    report
            );
            return false;
        }
        FormatWriter fw = factory.getRdfWriter(byteOutStream);
        fw.setAnnotated(annotate);
        outputMediaType = factory.getMimeType();
        List<TripleHandler> tripleHandlers = new ArrayList<TripleHandler>();
        tripleHandlers.add(new IgnoreAccidentalRDFa(fw));
        tripleHandlers.add(new CountingTripleHandler());
        rdfWriter = new CompositeTripleHandler(tripleHandlers);
View Full Code Here

        final TripleHandler verifierTripleHandler = Mockito.mock(TripleHandler.class);
        compositeTripleHandler.addChild(verifierTripleHandler);
        final CountingTripleHandler countingTripleHandler = new CountingTripleHandler();
        compositeTripleHandler.addChild(countingTripleHandler);
        final ByteArrayOutputStream out = new ByteArrayOutputStream();
        compositeTripleHandler.addChild( new NTriplesWriter(out) );
        final ExtractionResult extractionResult = new ExtractionResultImpl(
                extractionContext, extractor, compositeTripleHandler
        );
        extractor.run(extractionParameters, extractionContext, is, extractionResult);
        compositeTripleHandler.close();
View Full Code Here

    @Test
    public void testTypedLiteralIncompatibleValueSupport()
    throws IOException, ExtractionException, TripleHandlerException {
        final URI uri = RDFUtils.uri("http://host.com/test-malformed-literal.turtle");
        ByteArrayOutputStream baos = new ByteArrayOutputStream();
        final TripleHandler th = new RDFXMLWriter(baos);
        final ExtractionContext extractionContext = new ExtractionContext("turtle-extractor", uri);
        final ExtractionResult result = new ExtractionResultImpl(extractionContext, extractor, th);
        extractor.setStopAtFirstError(false);
        try {
            extractor.run(
                    ExtractionParameters.newDefault(),
                    extractionContext,
                    this.getClass().getResourceAsStream("/org/apache/any23/extractor/rdf/testMalformedLiteral"),
                    result
            );
        } finally {
            logger.debug(baos.toString());
            th.close();
            result.close();
        }
    }
View Full Code Here

        assertTripleCount(vSINDICE.getProperty(SINDICE.NESTING_ORIGINAL, vREVIEW.hasReview, 1);
    }

    private SingleDocumentExtraction getInstance(String file) throws FileNotFoundException, IOException {
        baos = new ByteArrayOutputStream();
        rdfxmlWriter = new RDFXMLWriter(baos);
        repositoryWriter = new RepositoryWriter(conn);

        final CompositeTripleHandler cth = new CompositeTripleHandler();
        cth.addChild(rdfxmlWriter);
        cth.addChild(repositoryWriter);
View Full Code Here

TOP

Related Classes of org.apache.any23.writer.CountingTripleHandler

Copyright © 2018 www.massapicom. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.