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

Example 21 with TikaConfig

use of org.apache.tika.config.TikaConfig 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 22 with TikaConfig

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

the class TikaServerCli method main.

public static void main(String[] args) {
    LOG.info("Starting {} server", new Tika());
    try {
        Options options = getOptions();
        CommandLineParser cliParser = new GnuParser();
        CommandLine line = cliParser.parse(options, args);
        if (line.hasOption("help")) {
            HelpFormatter helpFormatter = new HelpFormatter();
            helpFormatter.printHelp("tikaserver", options);
            System.exit(-1);
        }
        String host = DEFAULT_HOST;
        if (line.hasOption("host")) {
            host = line.getOptionValue("host");
            if ("*".equals(host)) {
                host = "0.0.0.0";
            }
        }
        int port = DEFAULT_PORT;
        if (line.hasOption("port")) {
            port = Integer.valueOf(line.getOptionValue("port"));
        }
        boolean returnStackTrace = false;
        if (line.hasOption("includeStack")) {
            returnStackTrace = true;
        }
        TikaLoggingFilter logFilter = null;
        if (line.hasOption("log")) {
            String logLevel = line.getOptionValue("log");
            if (LOG_LEVELS.contains(logLevel)) {
                boolean isInfoLevel = "info".equals(logLevel);
                logFilter = new TikaLoggingFilter(isInfoLevel);
            } else {
                LOG.info("Unsupported request URI log level: {}", logLevel);
            }
        }
        CrossOriginResourceSharingFilter corsFilter = null;
        if (line.hasOption("cors")) {
            corsFilter = new CrossOriginResourceSharingFilter();
            String url = line.getOptionValue("cors");
            List<String> origins = new ArrayList<String>();
            // Empty list allows all origins.
            if (!url.equals("*"))
                origins.add(url);
            corsFilter.setAllowOrigins(origins);
        }
        // The Tika Configuration to use throughout            
        TikaConfig tika;
        if (line.hasOption("config")) {
            String configFilePath = line.getOptionValue("config");
            LOG.info("Using custom config: {}", configFilePath);
            tika = new TikaConfig(configFilePath);
        } else {
            tika = TikaConfig.getDefaultConfig();
        }
        DigestingParser.Digester digester = null;
        if (line.hasOption("digest")) {
            int digestMarkLimit = DEFAULT_DIGEST_MARK_LIMIT;
            if (line.hasOption("dml")) {
                String dmlS = line.getOptionValue("dml");
                try {
                    digestMarkLimit = Integer.parseInt(dmlS);
                } catch (NumberFormatException e) {
                    throw new RuntimeException("Must have parseable int after digestMarkLimit(dml): " + dmlS);
                }
            }
            digester = new CommonsDigester(digestMarkLimit, CommonsDigester.parse(line.getOptionValue("digest")));
        }
        if (line.hasOption("enableFileUrl") && !line.hasOption("enableUnsecureFeatures")) {
            System.err.println("If you want to enable fileUrl, you must also acknowledge the security risks\n" + "by including --enableUnsecureFeatures.  See CVE-2015-3271.");
            System.exit(-1);
        }
        InputStreamFactory inputStreamFactory = null;
        if (line.hasOption("enableFileUrl") && line.hasOption("enableUnsecureFeatures")) {
            inputStreamFactory = new URLEnabledInputStreamFactory();
            System.out.println(FILE_URL_WARNING);
        } else {
            inputStreamFactory = new DefaultInputStreamFactory();
        }
        TikaResource.init(tika, digester, inputStreamFactory);
        JAXRSServerFactoryBean sf = new JAXRSServerFactoryBean();
        List<ResourceProvider> rCoreProviders = new ArrayList<>();
        rCoreProviders.add(new SingletonResourceProvider(new MetadataResource()));
        rCoreProviders.add(new SingletonResourceProvider(new RecursiveMetadataResource()));
        rCoreProviders.add(new SingletonResourceProvider(new DetectorResource()));
        rCoreProviders.add(new SingletonResourceProvider(new LanguageResource()));
        rCoreProviders.add(new SingletonResourceProvider(new TranslateResource()));
        rCoreProviders.add(new SingletonResourceProvider(new TikaResource()));
        rCoreProviders.add(new SingletonResourceProvider(new UnpackerResource()));
        rCoreProviders.add(new SingletonResourceProvider(new TikaMimeTypes()));
        rCoreProviders.add(new SingletonResourceProvider(new TikaDetectors()));
        rCoreProviders.add(new SingletonResourceProvider(new TikaParsers()));
        rCoreProviders.add(new SingletonResourceProvider(new TikaVersion()));
        List<ResourceProvider> rAllProviders = new ArrayList<>(rCoreProviders);
        rAllProviders.add(new SingletonResourceProvider(new TikaWelcome(rCoreProviders)));
        sf.setResourceProviders(rAllProviders);
        List<Object> providers = new ArrayList<>();
        providers.add(new TarWriter());
        providers.add(new ZipWriter());
        providers.add(new CSVMessageBodyWriter());
        providers.add(new MetadataListMessageBodyWriter());
        providers.add(new JSONMessageBodyWriter());
        providers.add(new XMPMessageBodyWriter());
        providers.add(new TextMessageBodyWriter());
        providers.add(new TikaServerParseExceptionMapper(returnStackTrace));
        if (logFilter != null) {
            providers.add(logFilter);
        }
        if (corsFilter != null) {
            providers.add(corsFilter);
        }
        sf.setProviders(providers);
        String url = "http://" + host + ":" + port + "/";
        sf.setAddress(url);
        BindingFactoryManager manager = sf.getBus().getExtension(BindingFactoryManager.class);
        JAXRSBindingFactory factory = new JAXRSBindingFactory();
        factory.setBus(sf.getBus());
        manager.registerBindingFactory(JAXRSBindingFactory.JAXRS_BINDING_ID, factory);
        sf.create();
        LOG.info("Started Apache Tika server at {}", url);
    } catch (Exception ex) {
        LOG.error("Can't start", ex);
        System.exit(-1);
    }
}

Example 23 with TikaConfig

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

the class RTFParserTest method testConfig.

@Test
public void testConfig() throws Exception {
    //test that memory allocation of the bin element is limited
    //via the config file.  Unfortunately, this test file's bin embedding contains 10 bytes
    //so we had to set the config to 0.
    InputStream is = getClass().getResourceAsStream("/org/apache/tika/parser/rtf/tika-config.xml");
    assertNotNull(is);
    TikaConfig tikaConfig = new TikaConfig(is);
    Parser p = new AutoDetectParser(tikaConfig);
    List<Metadata> metadataList = getRecursiveMetadata("testBinControlWord.rtf", p);
    assertEquals(1, metadataList.size());
    assertContains("TikaMemoryLimitException", metadataList.get(0).get(TikaCoreProperties.TIKA_META_EXCEPTION_EMBEDDED_STREAM));
}

Example 24 with TikaConfig

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

the class ConcurrentUtilsTest method testExecuteExecutor.

@Test
public void testExecuteExecutor() throws Exception {
    TikaConfig config = TikaConfig.getDefaultConfig();
    ParseContext context = new ParseContext();
    context.set(ExecutorService.class, config.getExecutorService());
    Future result = ConcurrentUtils.execute(context, new Runnable() {

        @Override
        public void run() {
        //Do nothing
        }
    });
    assertNull(result.get());
}

Example 25 with TikaConfig

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

the class DL4JInceptionV3NetTest method recognise.

@Test
@Ignore("until we can make this more robust across platforms")
public void recognise() throws Exception {
    TikaConfig config;
    try {
        config = new TikaConfig(getClass().getResourceAsStream("dl4j-inception3-config.xml"));
    } catch (TikaConfigException e) {
        if (e.getMessage() != null && e.getMessage().contains("Connection refused")) {
            return;
        }
        throw e;
    }
    Tika tika = new Tika(config);
    Metadata md = new Metadata();
    tika.parse(getClass().getResourceAsStream("cat.jpg"), md);
    String[] objects = md.getValues("OBJECT");
    boolean found = false;
    for (String object : objects) {
        if (object.contains("_cat")) {
            found = true;
        }
    }
    assertTrue(found);
}