Java TikaCoreProperties类代码示例

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
public void handle(Directory directory, Metadata metadata)
        throws MetadataException {
    if (directory.containsTag(IptcDirectory.TAG_KEYWORDS)) {
        String[] keywords = directory.getStringArray(IptcDirectory.TAG_KEYWORDS);
        for (String k : keywords) {
            metadata.add(TikaCoreProperties.KEYWORDS, k);
        }
    }
    if (directory.containsTag(IptcDirectory.TAG_HEADLINE)) {
        metadata.set(TikaCoreProperties.TITLE, directory.getString(IptcDirectory.TAG_HEADLINE));
    } else if (directory.containsTag(IptcDirectory.TAG_OBJECT_NAME)) {
        metadata.set(TikaCoreProperties.TITLE, directory.getString(IptcDirectory.TAG_OBJECT_NAME));
    }
    if (directory.containsTag(IptcDirectory.TAG_BY_LINE)) {
        metadata.set(TikaCoreProperties.CREATOR, directory.getString(IptcDirectory.TAG_BY_LINE));
        metadata.set(IPTC.CREATOR, directory.getString(IptcDirectory.TAG_BY_LINE));
    }
    if (directory.containsTag(IptcDirectory.TAG_CAPTION)) {
        metadata.set(TikaCoreProperties.DESCRIPTION,
                // Looks like metadata extractor returns IPTC newlines as a single carriage return,
                // but the exiv2 command does not so we change to line feed here because that is less surprising to users                        
                directory.getString(IptcDirectory.TAG_CAPTION).replaceAll("\r\n?", "\n"));
    }
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
@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);
		}
	}
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
private void parse(DocumentSummaryInformation summary) {
  set(OfficeOpenXMLExtended.COMPANY, summary.getCompany());
  set(OfficeOpenXMLExtended.MANAGER, summary.getManager());
  set(TikaCoreProperties.LANGUAGE, getLanguage(summary));
  set(OfficeOpenXMLCore.CATEGORY, summary.getCategory());

  // New style counts
  set(Office.SLIDE_COUNT, summary.getSlideCount());
  if (summary.getSlideCount() > 0) {
    metadata.set(PagedText.N_PAGES, summary.getSlideCount());
  }
  // Old style, Tika 1.0 counts
  // TODO Remove these in Tika 2.0
  set(Metadata.COMPANY, summary.getCompany());
  set(Metadata.MANAGER, summary.getManager());
  set(MSOffice.SLIDE_COUNT, summary.getSlideCount());
  set(Metadata.CATEGORY, summary.getCategory());

  parse(summary.getCustomProperties());
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
/**
 * Returns a resolved key that is common in other document types or returns
 * the specified metaDataLocalName if no common key could be found. The key
 * could be a simple String key, or could be a {@link Property}
 *
 * @param metaDataLocalName
 *          The localname of the element containing metadata
 * @return a resolved key that is common in other document types
 */
private Object resolveMetaDataKey(String metaDataLocalName) {
  Object metaDataKey = metaDataLocalName;
  if ("sf:authors".equals(metaDataQName)) {
    metaDataKey = TikaCoreProperties.CREATOR;
  } else if ("sf:title".equals(metaDataQName)) {
    metaDataKey = TikaCoreProperties.TITLE;
  } else if ("sl:SLCreationDateProperty".equals(metaDataQName)) {
    metaDataKey = TikaCoreProperties.CREATED;
  } else if ("sl:SLLastModifiedDateProperty".equals(metaDataQName)) {
    metaDataKey = Metadata.LAST_MODIFIED;
  } else if ("sl:language".equals(metaDataQName)) {
    metaDataKey = TikaCoreProperties.LANGUAGE;
  }
  return metaDataKey;
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
protected ContentHandler getContentHandler(
        ContentHandler handler, Metadata metadata, ParseContext context) {
    return new TeeContentHandler(
            super.getContentHandler(handler, metadata, context),
            getDublinCoreHandler(metadata, TikaCoreProperties.TITLE, "title"),
            getDublinCoreHandler(metadata, TikaCoreProperties.KEYWORDS, "subject"),
            getDublinCoreHandler(metadata, TikaCoreProperties.CREATOR, "creator"),
            getDublinCoreHandler(metadata, TikaCoreProperties.DESCRIPTION, "description"),
            getDublinCoreHandler(metadata, TikaCoreProperties.PUBLISHER, "publisher"),
            getDublinCoreHandler(metadata, TikaCoreProperties.CONTRIBUTOR, "contributor"),
            getDublinCoreHandler(metadata, TikaCoreProperties.CREATED, "date"),
            getDublinCoreHandler(metadata, TikaCoreProperties.TYPE, "type"),
            getDublinCoreHandler(metadata, TikaCoreProperties.FORMAT, "format"),
            getDublinCoreHandler(metadata, TikaCoreProperties.IDENTIFIER, "identifier"),
            getDublinCoreHandler(metadata, TikaCoreProperties.LANGUAGE, "language"),
            getDublinCoreHandler(metadata, TikaCoreProperties.RIGHTS, "rights"));
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
/**
 * Test the plain text output of the Word converter
 * @throws Exception
 */
@Test
public void testWord() throws Exception {
    Metadata metadata = new Metadata();
    ContentHandler handler = new BodyContentHandler();
    ParseContext context = new ParseContext();

    InputStream input = getTestDocument("testWORD.docx");
    try {
        parser.parse(input, handler, metadata, context);
        
      
        assertEquals(
                "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
                metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Sample Word Document", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Keith Bennett", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
        assertTrue(handler.toString().contains("Sample Word Document"));
    } finally {
        input.close();
    }
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
@Test
public void testWordCustomProperties() throws Exception {
   InputStream input = OOXMLParserTest.class.getResourceAsStream(
         "/test-documents/testWORD_custom_props.docx");
   Metadata metadata = new Metadata();

   try {
      ContentHandler handler = new BodyContentHandler(-1);
      ParseContext context = new ParseContext();
      context.set(Locale.class, Locale.US);
      new OOXMLParser().parse(input, handler, metadata, context);
   } finally {
      input.close();
   }

   assertEquals(
         "application/vnd.openxmlformats-officedocument.wordprocessingml.document", 
         metadata.get(Metadata.CONTENT_TYPE));
   assertEquals("EJ04325S",             metadata.get(TikaCoreProperties.CREATOR));
   assertEquals("Etienne Jouvin",       metadata.get(TikaCoreProperties.MODIFIER));
   assertEquals("Etienne Jouvin",       metadata.get(Metadata.LAST_AUTHOR));
   assertEquals("2011-07-29T16:52:00Z", metadata.get(TikaCoreProperties.CREATED));
   assertEquals("2011-07-29T16:52:00Z", metadata.get(Metadata.CREATION_DATE));
   assertEquals("2012-01-03T22:14:00Z", metadata.get(TikaCoreProperties.MODIFIED));
   assertEquals("2012-01-03T22:14:00Z", metadata.get(Metadata.DATE));
   assertEquals("Microsoft Office Word",metadata.get(Metadata.APPLICATION_NAME));
   assertEquals("Microsoft Office Word",metadata.get(OfficeOpenXMLExtended.APPLICATION));
   assertEquals("1",                    metadata.get(Office.PAGE_COUNT));
   assertEquals("2",                    metadata.get(Office.WORD_COUNT));
   assertEquals("My Title",             metadata.get(TikaCoreProperties.TITLE));
   assertEquals("My Keyword",           metadata.get(TikaCoreProperties.KEYWORDS));
   assertEquals("Normal.dotm",          metadata.get(Metadata.TEMPLATE));
   assertEquals("Normal.dotm",          metadata.get(OfficeOpenXMLExtended.TEMPLATE));
   // TODO: Remove subject in Tika 2.0
   assertEquals("My subject",           metadata.get(Metadata.SUBJECT));
   assertEquals("My subject",           metadata.get(OfficeOpenXMLCore.SUBJECT));
   assertEquals("EDF-DIT",              metadata.get(TikaCoreProperties.PUBLISHER));
   assertEquals("true",                 metadata.get("custom:myCustomBoolean"));
   assertEquals("3",                    metadata.get("custom:myCustomNumber"));
   assertEquals("MyStringValue",        metadata.get("custom:MyCustomString"));
   assertEquals("2010-12-30T23:00:00Z", metadata.get("custom:MyCustomDate"));
   assertEquals("2010-12-29T22:00:00Z", metadata.get("custom:myCustomSecondDate"));
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
@Test
public void testCustomMetadata() throws Exception {
    Parser parser = new AutoDetectParser(); // Should auto-detect!
    Metadata metadata = new Metadata();

    InputStream stream = PDFParserTest.class.getResourceAsStream(
            "/test-documents/testPDF-custommetadata.pdf");

    String content = getText(stream, parser, metadata);

    assertEquals("application/pdf", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("Document author", metadata.get(TikaCoreProperties.CREATOR));
    assertEquals("Document author", metadata.get(Metadata.AUTHOR));
    assertEquals("Document title", metadata.get(TikaCoreProperties.TITLE));
    
    assertEquals("Custom Value", metadata.get("Custom Property"));
    
    assertEquals("Array Entry 1", metadata.get("Custom Array"));
    assertEquals(2, metadata.getValues("Custom Array").length);
    assertEquals("Array Entry 1", metadata.getValues("Custom Array")[0]);
    assertEquals("Array Entry 2", metadata.getValues("Custom Array")[1]);
    
    assertTrue(content.contains("Hello World!"));
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
/**
 * Ensures that custom OLE2 (HPSF) properties are extracted
 */
@Test
public void testCustomProperties() throws Exception {
   InputStream input = ExcelParserTest.class.getResourceAsStream(
         "/test-documents/testEXCEL_custom_props.xls");
   Metadata metadata = new Metadata();
   
   try {
      ContentHandler handler = new BodyContentHandler(-1);
      ParseContext context = new ParseContext();
      context.set(Locale.class, Locale.US);
      new OfficeParser().parse(input, handler, metadata, context);
   } finally {
      input.close();
   }
   
   assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
   assertEquals("",                     metadata.get(TikaCoreProperties.CREATOR));
   assertEquals("",                     metadata.get(TikaCoreProperties.MODIFIER));
   assertEquals("2011-08-22T13:45:54Z", metadata.get(TikaCoreProperties.MODIFIED));
   assertEquals("2006-09-12T15:06:44Z", metadata.get(TikaCoreProperties.CREATED));
   assertEquals("Microsoft Excel",      metadata.get(OfficeOpenXMLExtended.APPLICATION));
   assertEquals("true",                 metadata.get("custom:myCustomBoolean"));
   assertEquals("3",                    metadata.get("custom:myCustomNumber"));
   assertEquals("MyStringValue",        metadata.get("custom:MyCustomString"));
   assertEquals("2010-12-30T22:00:00Z", metadata.get("custom:MyCustomDate"));
   assertEquals("2010-12-29T22:00:00Z", metadata.get("custom:myCustomSecondDate"));
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
FileTikaMediaImpl(File file, Metadata md)
        throws IOException {
    this.file = file;
    this.id = file.getCanonicalPath();
    this.lastModified = new Date(file.lastModified());

    artist = md.get(XMPDM.ARTIST);
    album = md.get(XMPDM.ALBUM);
    genre = md.get(XMPDM.GENRE);
    try {
        year = Integer.parseInt(md.get(XMPDM.RELEASE_DATE));
    } catch (NumberFormatException nfEx) {
        year = -1;
    }
    requestedBy = md.get(XMPDM.LOG_COMMENT);
    title = md.get(TikaCoreProperties.TITLE);
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
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));
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
private Property resolveMetadataKey(String localName) {
    if ("authors".equals(localName)) {
        return TikaCoreProperties.CREATOR;
    }
    if ("title".equals(localName)) {
        return TikaCoreProperties.TITLE;
    }
    if ("comment".equals(localName)) {
        return TikaCoreProperties.COMMENTS;
    }
    return Property.internalText(localName);
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
@Override
public void parse(
        InputStream stream, ContentHandler handler,
        Metadata metadata, ParseContext context)
        throws IOException, SAXException, TikaException {
    super.parse(stream, handler, metadata, context);
    // Copy subject to description for OO2
    String odfSubject = metadata.get(OfficeOpenXMLCore.SUBJECT);
    if (odfSubject != null && !odfSubject.equals("") && 
            (metadata.get(TikaCoreProperties.DESCRIPTION) == null || metadata.get(TikaCoreProperties.DESCRIPTION).equals(""))) {
        metadata.set(TikaCoreProperties.DESCRIPTION, odfSubject);
    }
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
@Override
public void endElement(
        String uri, String local, String name) throws SAXException {
    if (bodyLevel > 0 && discardLevel == 0) {
        String safe = mapper.mapSafeElement(name);
        if (safe != null) {
            xhtml.endElement(safe);
        } else if (XHTMLContentHandler.ENDLINE.contains(
                name.toLowerCase(Locale.ENGLISH))) {
            // TIKA-343: Replace closing block tags (and <br/>) with a
            // newline unless the HtmlMapper above has already mapped
            // them to something else
            xhtml.newline();
        }
    }

    if (titleLevel > 0) {
        titleLevel--;
        if (titleLevel == 0) {
            metadata.set(TikaCoreProperties.TITLE, title.toString().trim());
        }
    }
    if (bodyLevel > 0) {
        bodyLevel--;
    }
    if (discardLevel > 0) {
        discardLevel--;
    }
}
 

import org.apache.tika.metadata.TikaCoreProperties; //导入依赖的package包/类
public void parse(InputStream file) throws IOException, TikaException {
    ByteArrayOutputStream xmpraw = new ByteArrayOutputStream();
    if (!scanner.parse(file, xmpraw)) {
        return;
    }

    Reader decoded = new InputStreamReader(
            new ByteArrayInputStream(xmpraw.toByteArray()),
            DEFAULT_XMP_CHARSET);
    try {
        XMPMetadata xmp = XMPMetadata.load(new InputSource(decoded));
        XMPSchemaDublinCore dc = xmp.getDublinCoreSchema();
        if (dc != null) {
            if (dc.getTitle() != null) {
                metadata.set(TikaCoreProperties.TITLE, dc.getTitle());
            }
            if (dc.getDescription() != null) {
                metadata.set(TikaCoreProperties.DESCRIPTION, dc.getDescription());
            }
            if (dc.getCreators() != null && dc.getCreators().size() > 0) {
                metadata.set(TikaCoreProperties.CREATOR, joinCreators(dc.getCreators()));
            }
            if (dc.getSubjects() != null && dc.getSubjects().size() > 0) {
                Iterator<String> keywords = dc.getSubjects().iterator();
                while (keywords.hasNext()) {
                    metadata.add(TikaCoreProperties.KEYWORDS, keywords.next());
                }
                // TODO should we set KEYWORDS too?
                // All tested photo managers set the same in Iptc.Application2.Keywords and Xmp.dc.subject
            }
        }
    } catch (IOException e) {
        // Could not parse embedded XMP metadata. That's not a serious
        // problem, so we'll just ignore the issue for now.
        // TODO: Make error handling like this configurable.
    }
}