Examples with TikaConfig org.apache.tika.config.TikaConfig used on opensource projects

Example 16 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class CompositeParserTest method testDefaultParser.

@Test
public void testDefaultParser() throws Exception {
    TikaConfig config = TikaConfig.getDefaultConfig();
    CompositeParser parser = (CompositeParser) config.getParser();
    // Check it has the full registry
    assertEquals(config.getMediaTypeRegistry(), parser.getMediaTypeRegistry());
}

Example 17 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class InitializableParserTest method testInitializableParser.

@Test
public void testInitializableParser() throws Exception {
    URL configFileUrl = getClass().getClassLoader().getResource(TIKA_CFG_FILE);
    assert configFileUrl != null;
    TikaConfig config = new TikaConfig(configFileUrl);
    Tika tika = new Tika(config);
    Metadata md = new Metadata();
    tika.parse(TikaInputStream.get("someString".getBytes(StandardCharsets.ISO_8859_1)), md);
    assertEquals("5", md.get(DummyInitializableParser.SUM_FIELD));
}

Example 18 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class ParameterizedParserTest method getMetadata.

//TODO later -- add a test for a parser that isn't configurable
//but that has params in the config file
private Metadata getMetadata(String name) throws TikaException, IOException, SAXException {
    URL url = this.getClass().getResource("/org/apache/tika/config/" + name);
    assertNotNull("couldn't find: " + name, url);
    TikaConfig tikaConfig = new TikaConfig(url);
    Tika tika = new Tika(tikaConfig);
    Metadata metadata = new Metadata();
    tika.parse(url.openStream(), metadata);
    return metadata;
}

Example 19 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class RecursiveParserWrapperFSConsumerTest method testEmbeddedWithNPE.

@Test
public void testEmbeddedWithNPE() throws Exception {
    final String path = "/test-documents/embedded_with_npe.xml";
    final Metadata metadata = new Metadata();
    metadata.add(Metadata.RESOURCE_NAME_KEY, "embedded_with_npe.xml");
    ArrayBlockingQueue<FileResource> queue = new ArrayBlockingQueue<FileResource>(2);
    queue.add(new FileResource() {

        @Override
        public String getResourceId() {
            return "testFile";
        }

        @Override
        public Metadata getMetadata() {
            return metadata;
        }

        @Override
        public InputStream openInputStream() throws IOException {
            return this.getClass().getResourceAsStream(path);
        }
    });
    queue.add(new PoisonFileResource());
    MockOSFactory mockOSFactory = new MockOSFactory();
    RecursiveParserWrapperFSConsumer consumer = new RecursiveParserWrapperFSConsumer(queue, new AutoDetectParserFactory(), new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, -1), mockOSFactory, new TikaConfig());
    IFileProcessorFutureResult result = consumer.call();
    mockOSFactory.getStreams().get(0).flush();
    byte[] bytes = mockOSFactory.getStreams().get(0).toByteArray();
    List<Metadata> results = JsonMetadataList.fromJson(new InputStreamReader(new ByteArrayInputStream(bytes), UTF_8));
    assertEquals(4, results.size());
    assertContains("another null pointer", results.get(2).get(RecursiveParserWrapper.EMBEDDED_EXCEPTION));
    assertEquals("Nikolai Lobachevsky", results.get(0).get("author"));
    for (int i = 1; i < 4; i++) {
        assertEquals("embeddedAuthor" + i, results.get(i).get("author"));
        assertContains("some_embedded_content" + i, results.get(i).get(RecursiveParserWrapper.TIKA_CONTENT));
    }
}

Example 20 with TikaConfig

use of org.apache.tika.config.TikaConfig in project tika by apache.

the class NLTKNERecogniserTest method testGetEntityTypes.

@Test
public void testGetEntityTypes() throws Exception {
    String text = "America is a big country.";
    System.setProperty(NamedEntityParser.SYS_PROP_NER_IMPL, NLTKNERecogniser.class.getName());
    Tika tika = new Tika(new TikaConfig(NamedEntityParser.class.getResourceAsStream("tika-config.xml")));
    Metadata md = new Metadata();
    tika.parse(new ByteArrayInputStream(text.getBytes(StandardCharsets.UTF_8)), md);
    Set<String> names = new HashSet<>(Arrays.asList(md.getValues("NER_NAMES")));
    if (names.size() != 0) {
        assertTrue(names.contains("America"));
        assertTrue(names.size() == 1);
    }
}