Examples with TikaConfig org.apache.tika.config.TikaConfig used on opensource projects

Example 1 with TikaConfig

use of org.apache.tika.config.TikaConfig in project che by eclipse.

the class MediaTypeFilter method accept.

@Override
public boolean accept(VirtualFile file) {
    try (InputStream content = file.getContent()) {
        TikaConfig tikaConfig = new TikaConfig();
        MediaType mimeType = tikaConfig.getDetector().detect(content, new Metadata());
        if (excludedMediaTypes.contains(mimeType) || excludedTypes.contains(mimeType.getType())) {
            return true;
        }
        return false;
    } catch (TikaException | ForbiddenException | ServerException | IOException e) {
        return true;
    }
}

Example 2 with TikaConfig

use of org.apache.tika.config.TikaConfig in project lucene-solr by apache.

the class TikaEntityProcessor method firstInit.

@Override
protected void firstInit(Context context) {
    super.firstInit(context);
    try {
        String tikaConfigFile = context.getResolvedEntityAttribute("tikaConfig");
        if (tikaConfigFile == null) {
            ClassLoader classLoader = context.getSolrCore().getResourceLoader().getClassLoader();
            tikaConfig = new TikaConfig(classLoader);
        } else {
            File configFile = new File(tikaConfigFile);
            if (!configFile.isAbsolute()) {
                configFile = new File(context.getSolrCore().getResourceLoader().getConfigDir(), tikaConfigFile);
            }
            tikaConfig = new TikaConfig(configFile);
        }
    } catch (Exception e) {
        wrapAndThrow(SEVERE, e, "Unable to load Tika Config");
    }
    String extractEmbeddedString = context.getResolvedEntityAttribute("extractEmbedded");
    if ("true".equals(extractEmbeddedString)) {
        extractEmbedded = true;
    }
    format = context.getResolvedEntityAttribute("format");
    if (format == null)
        format = "text";
    if (!"html".equals(format) && !"xml".equals(format) && !"text".equals(format) && !"none".equals(format))
        throw new DataImportHandlerException(SEVERE, "'format' can be one of text|html|xml|none");
    htmlMapper = context.getResolvedEntityAttribute("htmlMapper");
    if (htmlMapper == null)
        htmlMapper = "default";
    if (!"default".equals(htmlMapper) && !"identity".equals(htmlMapper))
        throw new DataImportHandlerException(SEVERE, "'htmlMapper', if present, must be 'default' or 'identity'");
    parser = context.getResolvedEntityAttribute("parser");
    if (parser == null) {
        parser = AUTO_PARSER;
    }
    spatialMetadataField = context.getResolvedEntityAttribute("spatialMetadataField");
}

Example 3 with TikaConfig

use of org.apache.tika.config.TikaConfig in project lucene-solr by apache.

the class ExtractingRequestHandler method inform.

@Override
public void inform(SolrCore core) {
    if (initArgs != null) {
        //if relative,then relative to config dir, otherwise, absolute path
        String tikaConfigLoc = (String) initArgs.get(CONFIG_LOCATION);
        if (tikaConfigLoc != null) {
            File configFile = new File(tikaConfigLoc);
            if (configFile.isAbsolute() == false) {
                configFile = new File(core.getResourceLoader().getConfigDir(), configFile.getPath());
            }
            try {
                config = new TikaConfig(configFile);
            } catch (Exception e) {
                throw new SolrException(ErrorCode.SERVER_ERROR, e);
            }
        }
        String parseContextConfigLoc = (String) initArgs.get(PARSE_CONTEXT_CONFIG);
        if (parseContextConfigLoc != null) {
            try {
                parseContextConfig = new ParseContextConfig(core.getResourceLoader(), parseContextConfigLoc);
            } catch (Exception e) {
                throw new SolrException(ErrorCode.SERVER_ERROR, e);
            }
        }
        NamedList configDateFormats = (NamedList) initArgs.get(DATE_FORMATS);
        if (configDateFormats != null && configDateFormats.size() > 0) {
            dateFormats = new HashSet<>();
            Iterator<Map.Entry> it = configDateFormats.iterator();
            while (it.hasNext()) {
                String format = (String) it.next().getValue();
                log.info("Adding Date Format: " + format);
                dateFormats.add(format);
            }
        }
    }
    if (config == null) {
        try {
            config = getDefaultConfig(core.getResourceLoader().getClassLoader());
        } catch (MimeTypeException | IOException e) {
            throw new SolrException(ErrorCode.SERVER_ERROR, e);
        }
    }
    if (parseContextConfig == null) {
        parseContextConfig = new ParseContextConfig();
    }
    factory = createFactory();
}

Example 4 with TikaConfig

use of org.apache.tika.config.TikaConfig 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 5 with TikaConfig

use of org.apache.tika.config.TikaConfig 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);
    }
}