Search in sources :

Example 1 with Parser

use of org.apache.clerezza.rdf.core.serializedform.Parser in project stanbol by apache.

the class ContentItemReaderWriterTest method createTestContentItem.

/**
     * @return
     */
@BeforeClass
public static void createTestContentItem() throws IOException {
    contentItem = ciFactory.createContentItem(new IRI("urn:test"), new StringSource("<html>\n" + "  <body>\n" + "    This is a <b>ContentItem</b> to <i>Mime Multipart</i> test!\n" + "  </body>\n" + "</html>", "text/html"));
    RuntimeDelegate.setInstance(new RuntimeDelegateImpl());
    contentItem.addPart(new IRI("run:text:text"), ciFactory.createBlob(new StringSource("This is a ContentItem to Mime Multipart test!")));
    contentItem.getMetadata().add(new TripleImpl(new IRI("urn:test"), RDF.type, new IRI("urn:types:Document")));
    //mark the main content as parsed and also that all 
    //contents and contentparts should be included
    Map<String, Object> properties = initRequestPropertiesContentPart(contentItem);
    properties.put(PARSED_CONTENT_URIS, Collections.singleton(contentItem.getPartUri(0).getUnicodeString()));
    properties.put(OUTPUT_CONTENT, Collections.singleton("*/*"));
    properties.put(OUTPUT_CONTENT_PART, Collections.singleton("*"));
    properties.put(RDF_FORMAT, "application/rdf+xml");
    Graph em = initExecutionMetadataContentPart(contentItem);
    BlankNodeOrIRI ep = createExecutionPlan(em, "testChain", null);
    writeExecutionNode(em, ep, "testEngine", true, null, null);
    initExecutionMetadata(em, em, contentItem.getUri(), "testChain", false);
    ciWriter = new ContentItemWriter(Serializer.getInstance());
    ciReader = new ContentItemReader() {

        @Override
        protected Parser getParser() {
            return Parser.getInstance();
        }

        @Override
        protected ContentItemFactory getContentItemFactory() {
            return ciFactory;
        }
    };
}
Also used : IRI(org.apache.clerezza.commons.rdf.IRI) BlankNodeOrIRI(org.apache.clerezza.commons.rdf.BlankNodeOrIRI) ContentItemFactory(org.apache.stanbol.enhancer.servicesapi.ContentItemFactory) InMemoryContentItemFactory(org.apache.stanbol.enhancer.contentitem.inmemory.InMemoryContentItemFactory) RuntimeDelegateImpl(org.glassfish.jersey.internal.RuntimeDelegateImpl) BlankNodeOrIRI(org.apache.clerezza.commons.rdf.BlankNodeOrIRI) Parser(org.apache.clerezza.rdf.core.serializedform.Parser) ContentItemReader(org.apache.stanbol.enhancer.jersey.reader.ContentItemReader) SimpleGraph(org.apache.clerezza.commons.rdf.impl.utils.simple.SimpleGraph) Graph(org.apache.clerezza.commons.rdf.Graph) StringSource(org.apache.stanbol.enhancer.servicesapi.impl.StringSource) TripleImpl(org.apache.clerezza.commons.rdf.impl.utils.TripleImpl) ContentItemWriter(org.apache.stanbol.enhancer.jersey.writers.ContentItemWriter) BeforeClass(org.junit.BeforeClass)

Example 2 with Parser

use of org.apache.clerezza.rdf.core.serializedform.Parser in project stanbol by apache.

the class TopicEngineTest method testImportModelFromSKOS.

@Test
public void testImportModelFromSKOS() throws Exception {
    log.info(" --- testImportModelFromSKOS --- ");
    Parser parser = Parser.getInstance();
    parser.bindParsingProvider(new JenaParserProvider());
    ImmutableGraph graph = parser.parse(getClass().getResourceAsStream("/sample-scheme.skos.rdf.xml"), SupportedFormat.RDF_XML);
    int imported = classifier.importConceptsFromGraph(graph, OntologicalClasses.SKOS_CONCEPT, Properties.SKOS_BROADER);
    assertEquals(imported, 4);
    assertEquals(0, classifier.getBroaderConcepts("http://example.com/ns#someconceptscheme/100").size());
    assertEquals(0, classifier.getBroaderConcepts("http://example.com/ns#someconceptscheme/200").size());
    assertEquals(1, classifier.getBroaderConcepts("http://example.com/ns#someconceptscheme/010").size());
    assertEquals(1, classifier.getBroaderConcepts("http://example.com/ns#someconceptscheme/020").size());
    assertEquals(2, classifier.getRootConcepts().size());
}
Also used : JenaParserProvider(org.apache.clerezza.rdf.jena.parser.JenaParserProvider) ImmutableGraph(org.apache.clerezza.commons.rdf.ImmutableGraph) Parser(org.apache.clerezza.rdf.core.serializedform.Parser) Test(org.junit.Test)

Example 3 with Parser

use of org.apache.clerezza.rdf.core.serializedform.Parser in project stanbol by apache.

the class OpenCalaisEngine method readModel.

/**
     * Parses an InputStream of RDF data and produces an Graph from them
     *
     * @param in The InputStream of RDF data
     * @param format the format of the RDF data
     *
     * @return the resulting Graph or null if the RDF serialization format is not supported by the parser
     */
public Graph readModel(InputStream in, String format) {
    Parser parser = Parser.getInstance();
    if (parser.getSupportedFormats().contains(format)) {
        ImmutableGraph graph = parser.parse(in, format);
        Graph model = new SimpleGraph(graph);
        return model;
    } else {
        log.warn("Unsupported RDF format: {}\nSupported RDF formats: {}", format, parser.getSupportedFormats());
    }
    return null;
}
Also used : ImmutableGraph(org.apache.clerezza.commons.rdf.ImmutableGraph) SimpleGraph(org.apache.clerezza.commons.rdf.impl.utils.simple.SimpleGraph) Graph(org.apache.clerezza.commons.rdf.Graph) SimpleGraph(org.apache.clerezza.commons.rdf.impl.utils.simple.SimpleGraph) ImmutableGraph(org.apache.clerezza.commons.rdf.ImmutableGraph) QueryParser(org.apache.clerezza.rdf.core.sparql.QueryParser) Parser(org.apache.clerezza.rdf.core.serializedform.Parser)

Example 4 with Parser

use of org.apache.clerezza.rdf.core.serializedform.Parser in project stanbol by apache.

the class XsltExtractor method extract.

public synchronized void extract(String id, Document doc, Map<String, Object> params, Graph result) throws ExtractorException {
    if (params == null) {
        params = new HashMap<String, Object>();
    }
    params.put(this.uriParameter, id);
    initTransformerParameters(params);
    Source source = new DOMSource(doc);
    ByteArrayOutputStream writer = new ByteArrayOutputStream(8192);
    StreamResult output = new StreamResult(writer);
    try {
        this.transformer.transform(source, output);
        if (LOG.isDebugEnabled()) {
            String rdf = writer.toString("UTF-8");
            LOG.debug(rdf);
        }
        InputStream reader = new ByteArrayInputStream(writer.toByteArray());
        Parser rdfParser = Parser.getInstance();
        ImmutableGraph graph = rdfParser.parse(reader, this.syntax);
        result.addAll(graph);
    } catch (TransformerException e) {
        throw new ExtractorException(e.getMessage(), e);
    } catch (IOException e) {
        throw new ExtractorException(e.getMessage(), e);
    }
}
Also used : DOMSource(javax.xml.transform.dom.DOMSource) StreamResult(javax.xml.transform.stream.StreamResult) ByteArrayInputStream(java.io.ByteArrayInputStream) InputStream(java.io.InputStream) ByteArrayOutputStream(java.io.ByteArrayOutputStream) IOException(java.io.IOException) DOMSource(javax.xml.transform.dom.DOMSource) StreamSource(javax.xml.transform.stream.StreamSource) Source(javax.xml.transform.Source) Parser(org.apache.clerezza.rdf.core.serializedform.Parser) ByteArrayInputStream(java.io.ByteArrayInputStream) ImmutableGraph(org.apache.clerezza.commons.rdf.ImmutableGraph) TransformerException(javax.xml.transform.TransformerException)

Aggregations

Parser (org.apache.clerezza.rdf.core.serializedform.Parser)4 ImmutableGraph (org.apache.clerezza.commons.rdf.ImmutableGraph)3 Graph (org.apache.clerezza.commons.rdf.Graph)2 SimpleGraph (org.apache.clerezza.commons.rdf.impl.utils.simple.SimpleGraph)2 ByteArrayInputStream (java.io.ByteArrayInputStream)1 ByteArrayOutputStream (java.io.ByteArrayOutputStream)1 IOException (java.io.IOException)1 InputStream (java.io.InputStream)1 Source (javax.xml.transform.Source)1 TransformerException (javax.xml.transform.TransformerException)1 DOMSource (javax.xml.transform.dom.DOMSource)1 StreamResult (javax.xml.transform.stream.StreamResult)1 StreamSource (javax.xml.transform.stream.StreamSource)1 BlankNodeOrIRI (org.apache.clerezza.commons.rdf.BlankNodeOrIRI)1 IRI (org.apache.clerezza.commons.rdf.IRI)1 TripleImpl (org.apache.clerezza.commons.rdf.impl.utils.TripleImpl)1 QueryParser (org.apache.clerezza.rdf.core.sparql.QueryParser)1 JenaParserProvider (org.apache.clerezza.rdf.jena.parser.JenaParserProvider)1 InMemoryContentItemFactory (org.apache.stanbol.enhancer.contentitem.inmemory.InMemoryContentItemFactory)1 ContentItemReader (org.apache.stanbol.enhancer.jersey.reader.ContentItemReader)1