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

Example 1 with CompositeParser

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

the class TikaConfigTest method testUnknownParser.

/**
     * Make sure that with a service loader given, we can
     * get different configurable behaviour on parser classes
     * which can't be found.
     */
@Test
public void testUnknownParser() throws Exception {
    ServiceLoader ignoreLoader = new ServiceLoader(getClass().getClassLoader(), LoadErrorHandler.IGNORE);
    ServiceLoader warnLoader = new ServiceLoader(getClass().getClassLoader(), LoadErrorHandler.WARN);
    ServiceLoader throwLoader = new ServiceLoader(getClass().getClassLoader(), LoadErrorHandler.THROW);
    Path configPath = Paths.get(new URI(getConfigPath("TIKA-1700-unknown-parser.xml")));
    TikaConfig ignore = new TikaConfig(configPath, ignoreLoader);
    assertNotNull(ignore);
    assertNotNull(ignore.getParser());
    assertEquals(1, ((CompositeParser) ignore.getParser()).getAllComponentParsers().size());
    TikaConfig warn = new TikaConfig(configPath, warnLoader);
    assertNotNull(warn);
    assertNotNull(warn.getParser());
    assertEquals(1, ((CompositeParser) warn.getParser()).getAllComponentParsers().size());
    try {
        new TikaConfig(configPath, throwLoader);
        fail("Shouldn't get here, invalid parser class");
    } catch (TikaException expected) {
    }
}

Example 2 with CompositeParser

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

the class TikaConfigTest method defaultParserWithExcludes.

/**
     * TIKA-1445 It should be possible to exclude DefaultParser from
     *  certain types, so another parser explicitly listed will take them
     */
@Test
public void defaultParserWithExcludes() throws Exception {
    try {
        TikaConfig config = getConfig("TIKA-1445-default-except.xml");
        CompositeParser cp = (CompositeParser) config.getParser();
        List<Parser> parsers = cp.getAllComponentParsers();
        Parser p;
        // Will be the three parsers defined in the xml
        assertEquals(3, parsers.size());
        // Should have a wrapped DefaultParser, not the main DefaultParser,
        //  as it is excluded from handling certain classes
        p = parsers.get(0);
        assertTrue(p.toString(), p instanceof ParserDecorator);
        assertEquals(DefaultParser.class, ((ParserDecorator) p).getWrappedParser().getClass());
        // Should have two others which claim things, which they wouldn't
        //  otherwise handle
        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());
        p = parsers.get(2);
        assertTrue(p.toString(), p instanceof ParserDecorator);
        assertEquals(ErrorParser.class, ((ParserDecorator) p).getWrappedParser().getClass());
        assertEquals("fail/world", p.getSupportedTypes(null).iterator().next().toString());
    } catch (TikaException e) {
        fail("Unexpected TikaException: " + e);
    }
}

Example 3 with CompositeParser

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

the class ExternalParsersFactory method attachExternalParsers.

public static void attachExternalParsers(List<ExternalParser> parsers, TikaConfig config) {
    Parser parser = config.getParser();
    if (parser instanceof CompositeParser) {
        CompositeParser cParser = (CompositeParser) parser;
        Map<MediaType, Parser> parserMap = cParser.getParsers();
    }
// TODO
}

Example 4 with CompositeParser

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

the class TikaParserConfigTest method testMimeExcludeInclude.

@Test
public void testMimeExcludeInclude() throws Exception {
    TikaConfig config = getConfig("TIKA-1558-blacklist.xml");
    assertNotNull(config.getParser());
    assertNotNull(config.getDetector());
    Parser parser = config.getParser();
    MediaType PDF = MediaType.application("pdf");
    MediaType JPEG = MediaType.image("jpeg");
    // Has two parsers
    assertEquals(CompositeParser.class, parser.getClass());
    CompositeParser cParser = (CompositeParser) parser;
    assertEquals(2, cParser.getAllComponentParsers().size());
    // Both are decorated
    assertTrue(cParser.getAllComponentParsers().get(0) instanceof ParserDecorator);
    assertTrue(cParser.getAllComponentParsers().get(1) instanceof ParserDecorator);
    ParserDecorator p0 = (ParserDecorator) cParser.getAllComponentParsers().get(0);
    ParserDecorator p1 = (ParserDecorator) cParser.getAllComponentParsers().get(1);
    // DefaultParser will be wrapped with excludes
    assertEquals(DefaultParser.class, p0.getWrappedParser().getClass());
    assertNotContained(PDF, p0.getSupportedTypes(context));
    assertContains(PDF, p0.getWrappedParser().getSupportedTypes(context));
    assertNotContained(JPEG, p0.getSupportedTypes(context));
    assertContains(JPEG, p0.getWrappedParser().getSupportedTypes(context));
    // Will have an empty parser for PDF
    assertEquals(EmptyParser.class, p1.getWrappedParser().getClass());
    assertEquals(1, p1.getSupportedTypes(context).size());
    assertContains(PDF, p1.getSupportedTypes(context));
    assertNotContained(PDF, p1.getWrappedParser().getSupportedTypes(context));
}

Example 5 with CompositeParser

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

the class TikaParserConfigTest method defaultParserBlacklist.

/**
     * TIKA-1558 It should be possible to exclude Parsers from being picked up by
     * DefaultParser.
     */
@Test
public void defaultParserBlacklist() throws Exception {
    TikaConfig config = new TikaConfig();
    assertNotNull(config.getParser());
    assertNotNull(config.getDetector());
    CompositeParser cp = (CompositeParser) config.getParser();
    List<Parser> parsers = cp.getAllComponentParsers();
    boolean hasXML = false;
    for (Parser p : parsers) {
        if (p instanceof XMLParser) {
            hasXML = true;
            break;
        }
    }
    assertTrue("Default config should include an XMLParser.", hasXML);
    // This custom TikaConfig should exclude XMLParser and all of its subclasses.
    config = getConfig("TIKA-1558-blacklistsub.xml");
    cp = (CompositeParser) config.getParser();
    parsers = cp.getAllComponentParsers();
    for (Parser p : parsers) {
        if (p instanceof XMLParser)
            fail("Custom config should not include an XMLParser (" + p.getClass() + ").");
    }
}