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

Example 26 with TikaConfig

use of org.apache.tika.config.TikaConfig in project jackrabbit-oak by apache.

the class BinaryTextExtractor method initializeTikaConfig.

private static TikaConfigHolder initializeTikaConfig(@Nullable IndexDefinition definition) {
    ClassLoader current = Thread.currentThread().getContextClassLoader();
    InputStream configStream = null;
    String configSource = null;
    try {
        Thread.currentThread().setContextClassLoader(LuceneIndexEditorContext.class.getClassLoader());
        if (definition != null && definition.hasCustomTikaConfig()) {
            log.debug("[{}] Using custom tika config", definition.getIndexName());
            configSource = "Custom config at " + definition.getIndexPath();
            configStream = definition.getTikaConfig();
        } else {
            URL configUrl = LuceneIndexEditorContext.class.getResource("tika-config.xml");
            if (configUrl != null) {
                configSource = configUrl.toString();
                configStream = configUrl.openStream();
            }
        }
        if (configStream != null) {
            return new TikaConfigHolder(new TikaConfig(configStream), configSource);
        }
    } catch (TikaException | IOException | SAXException e) {
        log.warn("Tika configuration not available : " + configSource, e);
    } finally {
        IOUtils.closeQuietly(configStream);
        Thread.currentThread().setContextClassLoader(current);
    }
    return new TikaConfigHolder(TikaConfig.getDefaultConfig(), "Default Config");
}

Example 27 with TikaConfig

use of org.apache.tika.config.TikaConfig in project gate-core by GateNLP.

the class TikaFormat method unpackMarkup.

@Override
public void unpackMarkup(Document doc, RepositioningInfo repInfo, RepositioningInfo ampCodingInfo) throws DocumentFormatException {
    if (doc == null || doc.getSourceUrl() == null) {
        throw new DocumentFormatException("GATE document is null or no content found. Nothing to parse!");
    }
    // End if
    // Create a status listener
    StatusListener statusListener = new StatusListener() {

        @Override
        public void statusChanged(String text) {
            // This is implemented in DocumentFormat.java and inherited here
            fireStatusChanged(text);
        }
    };
    XmlDocumentHandler ch = new XmlDocumentHandler(doc, this.markupElementsMap, this.element2StringMap);
    Metadata metadata = extractParserTips(doc);
    ch.addStatusListener(statusListener);
    ch.setRepositioningInfo(repInfo);
    // set the object with ampersand coding positions
    ch.setAmpCodingInfo(ampCodingInfo);
    InputStream input = null;
    try {
        Parser tikaParser = new TikaConfig().getParser();
        input = doc.getSourceUrl().openStream();
        tikaParser.parse(input, ch, metadata, new ParseContext());
        setDocumentFeatures(metadata, doc);
    } catch (IOException e) {
        throw new DocumentFormatException(e);
    } catch (SAXException e) {
        throw new DocumentFormatException(e);
    } catch (TikaException e) {
        throw new DocumentFormatException(e);
    } finally {
        // null safe
        IOUtils.closeQuietly(input);
        ch.removeStatusListener(statusListener);
    }
    if (doc instanceof DocumentImpl) {
        ((DocumentImpl) doc).setNextAnnotationId(ch.getCustomObjectsId());
    }
}

Example 28 with TikaConfig

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

the class TikaComponent method createEndpoint.

@Override
protected Endpoint createEndpoint(String uri, String remaining, Map<String, Object> parameters) throws Exception {
    TikaConfiguration tikaConfiguration = new TikaConfiguration();
    setProperties(tikaConfiguration, parameters);
    TikaConfig config = resolveAndRemoveReferenceParameter(parameters, TIKA_CONFIG, TikaConfig.class);
    if (config != null) {
        tikaConfiguration.setTikaConfig(config);
    }
    tikaConfiguration.setOperation(new URI(uri).getHost());
    return new TikaEndpoint(uri, this, tikaConfiguration);
}

Example 29 with TikaConfig

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

the class TikaConfiguration method setTikaConfigUri.

/**
     * 
     * Tika Config Uri: The URI of  tika-config.xml
     * 
     */
public void setTikaConfigUri(String tikaConfigUri) throws TikaException, IOException, SAXException {
    this.tikaConfigUri = tikaConfigUri;
    this.tikaConfig = new TikaConfig(tikaConfigUri);
}

Example 30 with TikaConfig

use of org.apache.tika.config.TikaConfig in project jackrabbit-oak by apache.

the class TikaHelper method getTikaConfig.

private static TikaConfig getTikaConfig(File tikaConfig) throws TikaException, IOException, SAXException {
    TikaConfig config;
    if (tikaConfig == null) {
        URL configUrl = TextExtractor.class.getResource(DEFAULT_TIKA_CONFIG);
        if (configUrl != null) {
            log.info("Loading default Tika config from {}", configUrl);
            config = new TikaConfig(configUrl);
        } else {
            log.info("Using default Tika config");
            config = TikaConfig.getDefaultConfig();
        }
    } else {
        log.info("Loading external Tika config from {}", tikaConfig);
        config = new TikaConfig(tikaConfig);
    }
    return config;
}