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

Example 11 with DefaultParser

use of org.apache.tika.parser.DefaultParser 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 DefaultParser

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

the class TesseractOCRParserTest method offersNoTypesIfNotFound.

/*
    Check that if Tesseract is not found, the TesseractOCRParser claims to not support
    any file types. So, the standard image parser is called instead.
     */
@Test
public void offersNoTypesIfNotFound() throws Exception {
    TesseractOCRParser parser = new TesseractOCRParser();
    DefaultParser defaultParser = new DefaultParser();
    MediaType png = MediaType.image("png");
    // With an invalid path, will offer no types
    TesseractOCRConfig invalidConfig = new TesseractOCRConfig();
    invalidConfig.setTesseractPath("/made/up/path");
    ParseContext parseContext = new ParseContext();
    parseContext.set(TesseractOCRConfig.class, invalidConfig);
    // No types offered
    assertEquals(0, parser.getSupportedTypes(parseContext).size());
    // And DefaultParser won't use us
    assertEquals(ImageParser.class, defaultParser.getParsers(parseContext).get(png).getClass());
}