Examples of Tika

com.dotcms.repackage.org.apache.tika.Tika
org.apache.tika.Tika
Facade class for accessing Tika functionality. This class hides much of the underlying complexity of the lower level Tika classes and provides simple methods for many common parsing and type detection operations. @since Apache Tika 0.5 @see Parser @see Detector

Examples of com.dotcms.repackage.org.apache.tika.Tika

    Map<String, String> metaMap = new HashMap<String, String>();


    // store content metadata on disk
        File contentM=APILocator.getFileAssetAPI().getContentMetadataFile(inode);


    Tika t = new Tika();
    Metadata met = new Metadata();
    t.setMaxStringLength(-1);
    Reader fulltext = null;
    InputStream is = null;
    // if the limit is not "unlimited"
    // I can use the faster parseToString
    try {


      if(forceMemory){
        // no worry about the limit and less time to process.
        String content = t.parseToString(new FileInputStream(binFile), met);
        metaMap = new HashMap<String, String>();
        for (int i = 0; i < met.names().length; i++) {
          String name = met.names()[i];
          if (UtilMethods.isSet(name) && met.get(name) != null) {
            // we will want to normalize our metadata for searching
            String[] x = translateKey(name);
            for (String y : x)
              metaMap.put(y, met.get(name));
          }
        }
        metaMap.put(FileAssetAPI.CONTENT_FIELD, content);
      }
      else {




        is = TikaInputStream.get(binFile);
        fulltext = t.parse(is, met);
        metaMap = new HashMap<String, String>();
        for (int i = 0; i < met.names().length; i++) {
          String name = met.names()[i];
          if (UtilMethods.isSet(name) && met.get(name) != null) {
            // we will want to normalize our metadata for searching

View Full Code Here

Examples of org.apache.tika.Tika

  }


  public String getText() {
    if (text == null || lastTextUpdate < getFile().lastModified()) {
      try {
        Tika tika = new Tika();
        tika.setMaxStringLength(-1);
        text = tika.parseToString(getFile()).replaceAll("\n\\s*\n+","\n\n");
      } catch (TikaException e) {
        text = "";
      } catch (IOException e) {
        text = "";
      } finally {

View Full Code Here

Examples of org.apache.tika.Tika

    }


    public static String getMimeTypeWithByteBuffer(java.nio.ByteBuffer buffer) throws IOException {
        byte[] b = buffer.array();


        Tika tika = new Tika();
        return tika.detect(b);
    }

View Full Code Here

Examples of org.apache.tika.Tika

    }


    public static String getMimeTypeWithByteBuffer(java.nio.ByteBuffer buffer) throws IOException {
        byte[] b = buffer.array();


        Tika tika = new Tika();
        return tika.detect(b);
    }

View Full Code Here

Examples of org.apache.tika.Tika

    }


    public void XtestParseUTF8() throws IOException, SAXException, TikaException {
        String path = "/test-documents/testXHTML_utf8.html";
        Metadata metadata = new Metadata();
        String content = new Tika().parseToString(
                HtmlParserTest.class.getResourceAsStream(path), metadata);


        assertTrue("Did not contain expected text:"
                + "Title : Tilte with UTF-8 chars öäå", content
                .contains("Title : Tilte with UTF-8 chars öäå"));

View Full Code Here

Examples of org.apache.tika.Tika

    }


    public void testXhtmlParsing() throws Exception {
        String path = "/test-documents/testXHTML.html";
        Metadata metadata = new Metadata();
        String content = new Tika().parseToString(
                HtmlParserTest.class.getResourceAsStream(path), metadata);


        assertEquals("application/xhtml+xml", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("XHTML test document", metadata.get(Metadata.TITLE));

View Full Code Here

Examples of org.apache.tika.Tika

     * Test case for TIKA-210
     * @see <a href="https://issues.apache.org/jira/browse/TIKA-210">TIKA-210</a>
     */
    public void testCharactersDirectlyUnderBodyElement() throws Exception {
        String test = "<html><body>test</body></html>";
        String content = new Tika().parseToString(
                new ByteArrayInputStream(test.getBytes("UTF-8")));
        assertEquals("test", content);
    }

View Full Code Here

Examples of org.apache.tika.Tika

     * @see <a href="https://issues.apache.org/jira/browse/TIKA-268">TIKA-268</a>
     */
    public void testWhitespaceBetweenTableCells() throws Exception {
        String test =
            "<html><body><table><tr><td>a</td><td>b</td></table></body></html>";
        String content = new Tika().parseToString(
                new ByteArrayInputStream(test.getBytes("UTF-8")));
        assertTrue(content.contains("a"));
        assertTrue(content.contains("b"));
        assertFalse(content.contains("ab"));
    }

View Full Code Here

Examples of org.apache.tika.Tika

     *
     * @see <a href="https://issues.apache.org/jira/browse/TIKA-343">TIKA-343</a>
     */
    public void testLineBreak() throws Exception {
        String test = "<html><body><div>foo<br>bar</div>baz</body></html>";
        String text = new Tika().parseToString(
                new ByteArrayInputStream(test.getBytes("US-ASCII")));
        String[] parts = text.trim().split("\\s+");
        assertEquals(3, parts.length);
        assertEquals("foo", parts[0]);
        assertEquals("bar", parts[1]);

View Full Code Here

Examples of org.apache.tika.Tika

   * If you need better performance just use {@link com.gentics.cr.file.ResolvableFileBean#getMimeType()}
   * @throws IOException in case Tika detection fails.
   */
  public String getDetectedMimetype() throws IOException {
    if (file != null) {
      return new Tika().detect(file);
    } else {
      return UNKNOWN_MIMETYPE;
    }
  }

View Full Code Here

0 1 2 3 4 5

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.