org.apache.tika.metadata.TikaCoreProperties Java Examples

@Override
public void parseEmbedded(final InputStream input, final ContentHandler handler, final Metadata metadata,
                          final boolean outputHtml) throws SAXException, IOException {

	// There's no need to spawn inline embeds, like images in PDFs. These should be concatenated to the main
	// document as usual.
	if (TikaCoreProperties.EmbeddedResourceType.INLINE.toString().equals(metadata
			.get(TikaCoreProperties.EMBEDDED_RESOURCE_TYPE))) {
		final ContentHandler embedHandler = new EmbeddedContentHandler(new BodyContentHandler(handler));

		if (outputHtml) {
			writeStart(handler, metadata);
		}

		delegateParsing(input, embedHandler, metadata);

		if (outputHtml) {
			writeEnd(handler);
		}
	} else {
		try (final TikaInputStream tis = TikaInputStream.get(input)) {
			spawnEmbedded(tis, metadata);
		}
	}
}
 

private void setDocumentFeatures(Metadata metadata, Document doc) {
  FeatureMap fmap = doc.getFeatures();
  setTikaFeature(metadata, TikaCoreProperties.TITLE, fmap);
  setTikaFeature(metadata, Office.AUTHOR, fmap);
  setTikaFeature(metadata, TikaCoreProperties.COMMENTS, fmap);
  setTikaFeature(metadata, TikaCoreProperties.CREATOR, fmap);
  if (fmap.get("AUTHORS") == null && fmap.get("AUTHOR") != null)
    fmap.put("AUTHORS", fmap.get(Office.AUTHOR));
  fmap.put("MimeType", metadata.get(Metadata.CONTENT_TYPE));
}
 

/**
 * See if a language was set by the parser, from meta tags. As a last resort falls back to the result from the
 * ProfilingHandler.
 * 
 * @param metadata
 * @param profilingHandler
 * @return The first language found (two char lang code) or empty string if no language was detected.
 */
private static String detectLanguage(Metadata metadata, ProfilingHandler profilingHandler) {
    String result = null;

    String dubCoreLang = metadata.get(TikaCoreProperties.LANGUAGE);
    String httpEquivLang = metadata.get(Metadata.CONTENT_LANGUAGE);

    if (dubCoreLang != null) {
        result = dubCoreLang;
    } else if (httpEquivLang != null) {
        result = httpEquivLang;
    }

    result = getFirstLanguage(result);

    if (result == null) {
        // Language is still unspecified, so use ProfileHandler's result
        LanguageIdentifier langIdentifier = profilingHandler.getLanguage();
        // FUTURE KKr - provide config for specifying required certainty level.
        if (langIdentifier.isReasonablyCertain()) {
            result = langIdentifier.getLanguage();
            LOGGER.trace("Using language specified by profiling handler: " + result);
        } else {
            result = "";
        }

    }

    return result;
}
 

public static byte[] generateMetadataXML(byte[] file_bytes, String file_name) throws IOException {
	
	// Input & Output Variables
	ByteArrayInputStream base_input = new ByteArrayInputStream(file_bytes);
	byte[]               xml_bytes  = null;
	
	// Tika Parser Objects
    Parser               parser     = new AutoDetectParser();
    BodyContentHandler   handler    = new BodyContentHandler();
    Metadata             metadata   = new Metadata();
    ParseContext         context    = new ParseContext();
	  
    try {
    	// Populate Metadata Object with Tika Parser
    	parser.parse(base_input, handler, metadata, context);
    	
    	// Container & Writer for Metadata
    	Properties   meta_props = new Properties();
    	StringWriter sw         = new StringWriter();
    	
    	// Put Tika Metadata in Properties
    	for(String name : metadata.names()) {
    		if (!metadata.get(name).isEmpty()) {
    			meta_props.put(name, metadata.get(name));
    		}
    	}
    	meta_props.store(sw, "Tika Values");

    	// Expected Harvester Properties
    	String     meta_descr  = metadata.get(TikaCoreProperties.DESCRIPTION);
    	String     meta_modif  = metadata.get(TikaCoreProperties.MODIFIED);
    	String     meta_title  = metadata.get(TikaCoreProperties.TITLE);
    	
    	// Default Label for Undefined Tika Properties
    	DateFormat date_format = new SimpleDateFormat("yyyy/MM/dd");
    	Date       date        = new Date();
    	String     date_today  = date_format.format(date);
    	String     tika_label  = String.format("TIKA_%s", date_today);
    	
    	// Check For Null Values & Set Defaults
    	if (meta_descr == null) {
    		meta_props.put(WKAConstants.WKA_DESCRIPTION, "" + sw.toString());
    	} else {
    		meta_props.put(WKAConstants.WKA_DESCRIPTION, meta_descr);
    	}
    	
    	if (meta_modif == null) {
    		meta_props.put(WKAConstants.WKA_MODIFIED, tika_label);
    	} else {
    		meta_props.put(WKAConstants.WKA_MODIFIED, meta_modif);
    	}
    	
    	if (meta_title == null) {
    		meta_props.put(WKAConstants.WKA_TITLE, file_name);
    	} else {
    		meta_props.put(WKAConstants.WKA_TITLE, meta_title);
    	}
 	
    	// Build XML as Bytes
    	MapAttribute attr = AttributeUtils.fromProperties(meta_props);
		Document document = new SimpleDcMetaBuilder().create(attr);
		xml_bytes = XmlUtils.toString(document).getBytes("UTF-8");
    		
    } catch (Exception ex) {
      LOG.error(String.format("Error reading data."), ex);
    } finally {
    	base_input.close();
    }
	
	return xml_bytes;
	
}
 

private String getMetadataKey() {
    return TikaCoreProperties.TIKA_META_PREFIX +
            "digest" + Metadata.NAMESPACE_PREFIX_DELIMITER +
            algorithmKeyName;
}
 

static String getKey(String algorithm) {
	return TikaCoreProperties.TIKA_META_PREFIX + "digest" + Metadata.NAMESPACE_PREFIX_DELIMITER + algorithm
			.replace("-", "");
}