Examples with Metadata org.apache.tika.metadata.Metadata used on opensource projects

Example 81 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class ForkParserIntegrationTest method testParsingErrorInForkedParserShouldBeReported.

/**
     * TIKA-831 Parsers throwing errors should be caught and
     *  properly reported
     */
@Test
public void testParsingErrorInForkedParserShouldBeReported() throws Exception {
    BrokenParser brokenParser = new BrokenParser();
    ForkParser parser = new ForkParser(ForkParser.class.getClassLoader(), brokenParser);
    InputStream stream = getClass().getResourceAsStream("/test-documents/testTXT.txt");
    // With a serializable error, we'll get that back
    try {
        ContentHandler output = new BodyContentHandler();
        ParseContext context = new ParseContext();
        parser.parse(stream, output, new Metadata(), context);
        fail("Expected TikaException caused by Error");
    } catch (TikaException e) {
        assertEquals(brokenParser.err, e.getCause());
    } finally {
        parser.close();
    }
    // With a non serializable one, we'll get something else
    // TODO Fix this test
    brokenParser = new BrokenParser();
    brokenParser.re = new WontBeSerializedError("Can't Serialize");
    parser = new ForkParser(ForkParser.class.getClassLoader(), brokenParser);
//        try {
//           ContentHandler output = new BodyContentHandler();
//           ParseContext context = new ParseContext();
//           parser.parse(stream, output, new Metadata(), context);
//           fail("Expected TikaException caused by Error");
//       } catch (TikaException e) {
//           assertEquals(TikaException.class, e.getCause().getClass());
//           assertEquals("Bang!", e.getCause().getMessage());
//       }
}

Example 82 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class ForkParserIntegrationTest method testParserHandlingOfNonSerializable.

/**
     * If we supply a non serializable object on the ParseContext,
     *  check we get a helpful exception back
     */
@Test
public void testParserHandlingOfNonSerializable() throws Exception {
    ForkParser parser = new ForkParser(ForkParserIntegrationTest.class.getClassLoader(), tika.getParser());
    ParseContext context = new ParseContext();
    context.set(Detector.class, new Detector() {

        public MediaType detect(InputStream input, Metadata metadata) {
            return MediaType.OCTET_STREAM;
        }
    });
    try {
        ContentHandler output = new BodyContentHandler();
        InputStream stream = ForkParserIntegrationTest.class.getResourceAsStream("/test-documents/testTXT.txt");
        parser.parse(stream, output, new Metadata(), context);
        fail("Should have blown up with a non serializable ParseContext");
    } catch (TikaException e) {
        // Check the right details
        assertNotNull(e.getCause());
        assertEquals(NotSerializableException.class, e.getCause().getClass());
        assertEquals("Unable to serialize ParseContext to pass to the Forked Parser", e.getMessage());
    } finally {
        parser.close();
    }
}

Example 83 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class ForkParserIntegrationTest method testAttachingADebuggerOnTheForkedParserShouldWork.

/**
     * TIKA-832
     */
@Test
public void testAttachingADebuggerOnTheForkedParserShouldWork() throws Exception {
    ParseContext context = new ParseContext();
    context.set(Parser.class, tika.getParser());
    ForkParser parser = new ForkParser(ForkParserIntegrationTest.class.getClassLoader(), tika.getParser());
    parser.setJavaCommand(Arrays.asList("java", "-Xmx32m", "-Xdebug", "-Xrunjdwp:transport=dt_socket,address=54321,server=y,suspend=n"));
    try {
        ContentHandler body = new BodyContentHandler();
        InputStream stream = ForkParserIntegrationTest.class.getResourceAsStream("/test-documents/testTXT.txt");
        parser.parse(stream, body, new Metadata(), context);
        String content = body.toString();
        assertContains("Test d'indexation", content);
        assertContains("http://www.apache.org", content);
    } finally {
        parser.close();
    }
}

Example 84 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class RecursiveParserWrapperTest method testMaxEmbedded.

@Test
public void testMaxEmbedded() throws Exception {
    int maxEmbedded = 4;
    //including outer container file
    int totalNoLimit = 12;
    ParseContext context = new ParseContext();
    Metadata metadata = new Metadata();
    String limitReached = null;
    Parser wrapped = new AutoDetectParser();
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, -1));
    InputStream stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    List<Metadata> list = wrapper.getMetadata();
    //test default
    assertEquals(totalNoLimit, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertNull(limitReached);
    wrapper.reset();
    stream.close();
    //test setting value
    metadata = new Metadata();
    stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.setMaxEmbeddedResources(maxEmbedded);
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    list = wrapper.getMetadata();
    //add 1 for outer container file
    assertEquals(maxEmbedded + 1, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertEquals("true", limitReached);
    wrapper.reset();
    stream.close();
    //test setting value < 0
    metadata = new Metadata();
    stream = RecursiveParserWrapperTest.class.getResourceAsStream("/test-documents/test_recursive_embedded.docx");
    wrapper.setMaxEmbeddedResources(-2);
    wrapper.parse(stream, new DefaultHandler(), metadata, context);
    assertEquals(totalNoLimit, list.size());
    limitReached = list.get(0).get(RecursiveParserWrapper.EMBEDDED_RESOURCE_LIMIT_REACHED);
    assertNull(limitReached);
}

Example 85 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class RecursiveParserWrapperTest method testPrimaryExcWEmbedded.

@Test
public void testPrimaryExcWEmbedded() throws Exception {
    //if embedded content is handled and then
    //the parser hits an exception in the container document,
    //that the first element of the returned list is the container document
    //and the second is the embedded content
    Metadata metadata = new Metadata();
    metadata.set(Metadata.RESOURCE_NAME_KEY, "embedded_then_npe.xml");
    ParseContext context = new ParseContext();
    Parser wrapped = new AutoDetectParser();
    RecursiveParserWrapper wrapper = new RecursiveParserWrapper(wrapped, new BasicContentHandlerFactory(BasicContentHandlerFactory.HANDLER_TYPE.TEXT, -1), true);
    String path = "/test-documents/mock/embedded_then_npe.xml";
    InputStream stream = null;
    boolean npe = false;
    try {
        stream = RecursiveParserWrapperTest.class.getResourceAsStream(path);
        wrapper.parse(stream, new DefaultHandler(), metadata, context);
    } catch (TikaException e) {
        if (e.getCause().getClass().equals(NullPointerException.class)) {
            npe = true;
        }
    } finally {
        IOUtils.closeQuietly(stream);
    }
    assertTrue("npe", npe);
    List<Metadata> metadataList = wrapper.getMetadata();
    assertEquals(2, metadataList.size());
    Metadata outerMetadata = metadataList.get(0);
    Metadata embeddedMetadata = metadataList.get(1);
    assertContains("main_content", outerMetadata.get(RecursiveParserWrapper.TIKA_CONTENT));
    assertEquals("embedded_then_npe.xml", outerMetadata.get(TikaMetadataKeys.RESOURCE_NAME_KEY));
    assertEquals("Nikolai Lobachevsky", outerMetadata.get("author"));
    assertContains("some_embedded_content", embeddedMetadata.get(RecursiveParserWrapper.TIKA_CONTENT));
    assertEquals("embed1.xml", embeddedMetadata.get(TikaMetadataKeys.RESOURCE_NAME_KEY));
    assertEquals("embeddedAuthor", embeddedMetadata.get("author"));
}