Package org.apache.any23.rdf

Examples of org.apache.any23.rdf.Any23ValueFactoryWrapper


    public void testGZippedContent() throws IOException, URISyntaxException, ExtractionException {
        assumeOnlineAllowed();

        Any23 runner = new Any23();
        runner.setHTTPUserAgent("test-user-agent");
        HTTPClient httpClient = runner.getHTTPClient();
        DocumentSource source = new HTTPDocumentSource(
                httpClient,
                "http://products.semweb.bestbuy.com/y/products/7590289/"
        );
        ByteArrayOutputStream out = new ByteArrayOutputStream();
View Full Code Here


                configuration,
                new HTMLFixture(copyResourceToTempFile(file)).getOpener("http://nested.test.com"),
                extractorGroup,
                cth
        );
        instance.setMIMETypeDetector( new TikaMIMETypeDetector(new WhiteSpacesPurifier()) );
        return instance;
    }
View Full Code Here

                configuration,
                new HTMLFixture(copyResourceToTempFile(file)).getOpener("http://nested.test.com"),
                extractorGroup,
                cth
        );
        instance.setMIMETypeDetector( new TikaMIMETypeDetector(new WhiteSpacesPurifier()) );
        return instance;
    }
View Full Code Here

    private TikaMIMETypeDetector detector;

    @Before
    public void setUp() throws Exception {
        detector = new TikaMIMETypeDetector(new WhiteSpacesPurifier());
    }
View Full Code Here

            verifyDataType ? RDFParser.DatatypeHandling.VERIFY : RDFParser.DatatypeHandling.IGNORE
        );
        parser.setStopAtFirstError(stopAtFirstError);
        parser.setParseErrorListener( new InternalParseErrorListener(extractionResult) );
        parser.setValueFactory(
                new Any23ValueFactoryWrapper(
                        ValueFactoryImpl.getInstance(),
                        extractionResult,
                        extractionContext.getDefaultLanguage()
                )
        );
View Full Code Here

            lon = Float.parseFloat(coords[1]);
        } catch (NumberFormatException nfe) {
            return;
        }

        final ValueFactory factory = new Any23ValueFactoryWrapper(ValueFactoryImpl.getInstance(), out);
        BNode point = factory.createBNode();
        out.writeTriple(extractionContext.getDocumentURI(), expand("dcterms:related"), point);
        out.writeTriple(point, expand("rdf:type"), expand("geo:Point"));
        out.writeTriple(point, expand("geo:lat"), factory.createLiteral(Float.toString(lat)));
        out.writeTriple(point, expand("geo:long"), factory.createLiteral(Float.toString(lon)));
    }
View Full Code Here

        }

        final String contextURI = extractionParameters.getProperty(EXTRACTION_CONTEXT_URI_PROPERTY);
        ensureHasLocalCopy();
        try {
            this.documentURI = new Any23ValueFactoryWrapper(
                    ValueFactoryImpl.getInstance()
            ).createURI( "?".equals(contextURI) ? in.getDocumentURI() : contextURI);
        } catch (Exception ex) {
            throw new IllegalArgumentException("Invalid URI: " + in.getDocumentURI(), ex);
        }
View Full Code Here

            ExtractionParameters extractionParameters,
            ExtractionContext extractionContext,
            Document in,
            ExtractionResult out
    ) throws IOException, ExtractionException {
        final Any23ValueFactoryWrapper valueFactory = new Any23ValueFactoryWrapper(
            ValueFactoryImpl.getInstance(), out, extractionContext.getDefaultLanguage()
        );
       
        try {
            String title = DomUtils.find(in, "/HTML/HEAD/TITLE/text()").trim();
            if (title != null && (title.length() != 0)) {
                out.writeTriple(extractionContext.getDocumentURI(), vDCTERMS.title, valueFactory.createLiteral(title));
            }
        } finally {
            valueFactory.setIssueReport(null);
        }
    }
View Full Code Here

            verifyDataType ? RDFParser.DatatypeHandling.VERIFY : RDFParser.DatatypeHandling.IGNORE
        );
        parser.setStopAtFirstError(stopAtFirstError);
        parser.setParseErrorListener( new InternalParseErrorListener(extractionResult) );
        parser.setValueFactory(
                new Any23ValueFactoryWrapper(
                        ValueFactoryImpl.getInstance(),
                        extractionResult,
                        extractionContext.getDefaultLanguage()
                )
        );
View Full Code Here

            ExtractionParameters extractionParameters,
            ExtractionContext extractionContext,
            Document in,
            ExtractionResult out
    ) throws IOException, ExtractionException {
        final Any23ValueFactoryWrapper valueFactory = new Any23ValueFactoryWrapper(
            ValueFactoryImpl.getInstance(), out, extractionContext.getDefaultLanguage()
        );
       
        try {
            String title = DomUtils.find(in, "/HTML/HEAD/TITLE/text()").trim();
            if (title != null && (title.length() != 0)) {
                out.writeTriple(extractionContext.getDocumentURI(), vDCTERMS.title, valueFactory.createLiteral(title));
            }
        } finally {
            valueFactory.setIssueReport(null);
        }
    }
View Full Code Here

TOP

Related Classes of org.apache.any23.rdf.Any23ValueFactoryWrapper

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.