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

Example 6 with CompositeParser

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

the class TikaCLI method displayParser.

private void displayParser(Parser p, boolean includeMimeTypes, boolean apt, int i) {
    String decorated = null;
    if (p instanceof ParserDecorator) {
        ParserDecorator pd = (ParserDecorator) p;
        decorated = " (Wrapped by " + pd.getDecorationName() + ")";
        p = pd.getWrappedParser();
    }
    boolean isComposite = (p instanceof CompositeParser);
    String name = p.getClass().getName();
    if (apt) {
        name = name.substring(0, name.lastIndexOf(".") + 1) + "{{{./api/" + name.replace(".", "/") + "}" + name.substring(name.lastIndexOf(".") + 1) + "}}";
    } else if (decorated != null) {
        name += decorated;
    }
    if ((apt && !isComposite) || !apt) {
        // Don't display Composite parsers in the apt output.
        System.out.println(indent(i) + ((apt) ? "* " : "") + name + (isComposite ? " (Composite Parser):" : ""));
        if (apt)
            System.out.println();
        if (includeMimeTypes && !isComposite) {
            for (MediaType mt : p.getSupportedTypes(context)) {
                System.out.println(indent(i + 3) + ((apt) ? "* " : "") + mt);
                if (apt)
                    System.out.println();
            }
        }
    }
    if (isComposite) {
        Parser[] subParsers = sortParsers(invertMediaTypeMap(((CompositeParser) p).getParsers()));
        for (Parser sp : subParsers) {
            // Don't indent for Composites in apt.
            displayParser(sp, includeMimeTypes, apt, i + ((apt) ? 0 : 3));
        }
    }
}

Example 7 with CompositeParser

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

the class BundleIT method testBundleParsers.

@Test
public void testBundleParsers() throws Exception {
    // Get the classes found within OSGi
    ServiceReference<Parser> parserRef = bc.getServiceReference(Parser.class);
    DefaultParser parserService = (DefaultParser) bc.getService(parserRef);
    Set<String> osgiParsers = new HashSet<>();
    for (Parser p : parserService.getAllComponentParsers()) {
        osgiParsers.add(p.getClass().getName());
    }
    // Check we did get a few, just in case...
    assertTrue("Should have lots Parser names, found " + osgiParsers.size(), osgiParsers.size() > 15);
    // Get the raw parsers list from the traditional service loading mechanism
    CompositeParser parser = (CompositeParser) defaultParser;
    Set<String> rawParsers = new HashSet<>();
    for (Parser p : parser.getAllComponentParsers()) {
        if (p instanceof DefaultParser) {
            for (Parser pChild : ((DefaultParser) p).getAllComponentParsers()) {
                rawParsers.add(pChild.getClass().getName());
            }
        } else {
            rawParsers.add(p.getClass().getName());
        }
    }
    assertEquals(rawParsers, osgiParsers);
}

Example 8 with CompositeParser

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

the class EmbeddedDocumentUtil method tryToFindExistingLeafParser.

/**
     * Tries to find an existing parser within the ParseContext.
     * It looks inside of CompositeParsers and ParserDecorators.
     * The use case is when a parser needs to parse an internal stream
     * that is _part_ of the document, e.g. rtf body inside an msg.
     * <p/>
     * Can return <code>null</code> if the context contains no parser or
     * the correct parser can't be found.
     *
     * @param clazz parser class to search for
     * @param context
     * @return
     */
public static Parser tryToFindExistingLeafParser(Class clazz, ParseContext context) {
    Parser p = context.get(Parser.class);
    if (equals(p, clazz)) {
        return p;
    }
    Parser returnParser = null;
    if (p != null) {
        if (p instanceof ParserDecorator) {
            p = ((ParserDecorator) p).getWrappedParser();
        }
        if (equals(p, clazz)) {
            return p;
        }
        if (p instanceof CompositeParser) {
            returnParser = findInComposite((CompositeParser) p, clazz, context);
        }
    }
    if (returnParser != null && equals(returnParser, clazz)) {
        return returnParser;
    }
    return null;
}

Example 9 with CompositeParser

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

the class DumpTikaConfigExampleTest method testDump.

@Test
public void testDump() throws Exception {
    DumpTikaConfigExample ex = new DumpTikaConfigExample();
    for (Charset charset : new Charset[] { UTF_8, UTF_16LE }) {
        for (TikaConfigSerializer.Mode mode : TikaConfigSerializer.Mode.values()) {
            Writer writer = new OutputStreamWriter(new FileOutputStream(configFile), charset);
            TikaConfigSerializer.serialize(TikaConfig.getDefaultConfig(), mode, writer, charset);
            writer.flush();
            writer.close();
            TikaConfig c = new TikaConfig(configFile);
            assertTrue(c.getParser().toString(), c.getParser() instanceof CompositeParser);
            assertTrue(c.getDetector().toString(), c.getDetector() instanceof CompositeDetector);
            CompositeParser p = (CompositeParser) c.getParser();
            assertTrue("enough parsers?", p.getParsers().size() > 130);
            CompositeDetector d = (CompositeDetector) c.getDetector();
            assertTrue("enough detectors?", d.getDetectors().size() > 3);
            //just try to load it into autodetect to make sure no errors are thrown
            Parser auto = new AutoDetectParser(c);
            assertNotNull(auto);
        }
    }
}

Example 10 with CompositeParser

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

the class TikaConfigSerializer method addParser.

private static void addParser(Mode mode, Element rootElement, Document doc, Parser parser) throws Exception {
    // If the parser is decorated, is it a kind where we output the parser inside?
    ParserDecorator decoration = null;
    if (parser instanceof ParserDecorator) {
        if (parser.getClass().getName().startsWith(ParserDecorator.class.getName() + "$")) {
            decoration = ((ParserDecorator) parser);
            parser = decoration.getWrappedParser();
        }
    }
    boolean outputParser = true;
    List<Parser> children = Collections.emptyList();
    if (mode == Mode.CURRENT && parser instanceof DefaultParser) {
    // Only output the parser, not the children
    } else if (parser instanceof CompositeParser) {
        children = ((CompositeParser) parser).getAllComponentParsers();
        // Special case for a naked composite
        if (parser.getClass().equals(CompositeParser.class)) {
            outputParser = false;
        }
        // Special case for making Default to static
        if (parser instanceof DefaultParser && (mode == Mode.STATIC || mode == Mode.STATIC_FULL)) {
            outputParser = false;
        }
    }
    if (outputParser) {
        rootElement = addParser(mode, rootElement, doc, parser, decoration);
    }
    for (Parser childParser : children) {
        addParser(mode, rootElement, doc, childParser);
    }
// TODO Parser Exclusions
}