Examples with CompositeParser org.apache.tika.parser.CompositeParser used on opensource projects

Example 11 with CompositeParser

use of org.apache.tika.parser.CompositeParser in project tika by apache.

the class TikaParserConfigTest method testParserExcludeFromDefault.

@Test
public void testParserExcludeFromDefault() throws Exception {
    TikaConfig config = getConfig("TIKA-1558-blacklist.xml");
    assertNotNull(config.getParser());
    assertNotNull(config.getDetector());
    CompositeParser parser = (CompositeParser) config.getParser();
    MediaType PE_EXE = MediaType.application("x-msdownload");
    MediaType ELF = MediaType.application("x-elf");
    // Get the DefaultParser from the config
    ParserDecorator confWrappedParser = (ParserDecorator) parser.getParsers().get(MediaType.APPLICATION_XML);
    assertNotNull(confWrappedParser);
    DefaultParser confParser = (DefaultParser) confWrappedParser.getWrappedParser();
    // Get a fresh "default" DefaultParser
    DefaultParser normParser = new DefaultParser(config.getMediaTypeRegistry());
    // The default one will offer the Executable Parser
    assertContains(PE_EXE, normParser.getSupportedTypes(context));
    assertContains(ELF, normParser.getSupportedTypes(context));
    boolean hasExec = false;
    for (Parser p : normParser.getParsers().values()) {
        if (p instanceof ExecutableParser) {
            hasExec = true;
            break;
        }
    }
    assertTrue(hasExec);
    // The one from the config won't
    assertNotContained(PE_EXE, confParser.getSupportedTypes(context));
    assertNotContained(ELF, confParser.getSupportedTypes(context));
    for (Parser p : confParser.getParsers().values()) {
        if (p instanceof ExecutableParser)
            fail("Shouldn't have the Executable Parser from config");
    }
}

Example 12 with CompositeParser

use of org.apache.tika.parser.CompositeParser in project tika by apache.

the class TikaCLI method displaySupportedTypes.

/**
     * Prints all the known media types, aliases and matching parser classes.
     */
private void displaySupportedTypes() {
    AutoDetectParser parser = new AutoDetectParser();
    MediaTypeRegistry registry = parser.getMediaTypeRegistry();
    Map<MediaType, Parser> parsers = parser.getParsers();
    for (MediaType type : registry.getTypes()) {
        System.out.println(type);
        for (MediaType alias : registry.getAliases(type)) {
            System.out.println("  alias:     " + alias);
        }
        MediaType supertype = registry.getSupertype(type);
        if (supertype != null) {
            System.out.println("  supertype: " + supertype);
        }
        Parser p = parsers.get(type);
        if (p != null) {
            if (p instanceof CompositeParser) {
                p = ((CompositeParser) p).getParsers().get(type);
            }
            System.out.println("  parser:    " + p.getClass().getName());
        }
    }
}

Example 13 with CompositeParser

use of org.apache.tika.parser.CompositeParser in project tika by apache.

the class TikaMimeTypes method getMediaTypes.

protected List<MediaTypeDetails> getMediaTypes() {
    MediaTypeRegistry registry = TikaResource.getConfig().getMediaTypeRegistry();
    Map<MediaType, Parser> parsers = ((CompositeParser) TikaResource.getConfig().getParser()).getParsers();
    List<MediaTypeDetails> types = new ArrayList<TikaMimeTypes.MediaTypeDetails>(registry.getTypes().size());
    for (MediaType type : registry.getTypes()) {
        MediaTypeDetails details = new MediaTypeDetails();
        details.type = type;
        details.aliases = registry.getAliases(type).toArray(new MediaType[0]);
        MediaType supertype = registry.getSupertype(type);
        if (supertype != null && !MediaType.OCTET_STREAM.equals(supertype)) {
            details.supertype = supertype;
        }
        Parser p = parsers.get(type);
        if (p != null) {
            if (p instanceof CompositeParser) {
                p = ((CompositeParser) p).getParsers().get(type);
            }
            details.parser = p.getClass().getName();
        }
        types.add(details);
    }
    return types;
}

Example 14 with CompositeParser

use of org.apache.tika.parser.CompositeParser in project tika by apache.

the class TikaConfigTest method parserWithChildParsers.

/**
     * TIKA-1653 If one parser has child parsers, those child parsers shouldn't
     *  show up at the top level as well
     */
@Test
public void parserWithChildParsers() throws Exception {
    try {
        TikaConfig config = getConfig("TIKA-1653-norepeat.xml");
        CompositeParser cp = (CompositeParser) config.getParser();
        List<Parser> parsers = cp.getAllComponentParsers();
        Parser p;
        // Just 2 top level parsers
        assertEquals(2, parsers.size());
        // Should have a CompositeParser with 2 child ones, and
        //  and a wrapped empty parser
        p = parsers.get(0);
        assertTrue(p.toString(), p instanceof CompositeParser);
        assertEquals(2, ((CompositeParser) p).getAllComponentParsers().size());
        p = parsers.get(1);
        assertTrue(p.toString(), p instanceof ParserDecorator);
        assertEquals(EmptyParser.class, ((ParserDecorator) p).getWrappedParser().getClass());
        assertEquals("hello/world", p.getSupportedTypes(null).iterator().next().toString());
    } catch (TikaException e) {
        fail("Unexpected TikaException: " + e);
    }
}

Example 15 with CompositeParser

use of org.apache.tika.parser.CompositeParser in project tika by apache.

the class TIAParsingExample method useCompositeParser.

public static void useCompositeParser() throws Exception {
    InputStream stream = new ByteArrayInputStream(new byte[0]);
    ContentHandler handler = new DefaultHandler();
    ParseContext context = new ParseContext();
    Map<MediaType, Parser> parsersByType = new HashMap<MediaType, Parser>();
    parsersByType.put(MediaType.parse("text/html"), new HtmlParser());
    parsersByType.put(MediaType.parse("application/xml"), new XMLParser());
    CompositeParser parser = new CompositeParser();
    parser.setParsers(parsersByType);
    parser.setFallback(new TXTParser());
    Metadata metadata = new Metadata();
    metadata.set(Metadata.CONTENT_TYPE, "text/html");
    parser.parse(stream, handler, metadata, context);
}