Examples with Metadata org.apache.tika.metadata.Metadata used on opensource projects

Example 41 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class RollbackSoftware method rollback.

public void rollback(File deployArea) throws IOException, SAXException, TikaException {
    LinkContentHandler handler = new LinkContentHandler();
    Metadata met = new Metadata();
    DeploymentAreaParser parser = new DeploymentAreaParser();
    parser.parse(IOUtils.toInputStream(deployArea.getAbsolutePath(), UTF_8), handler, met);
    List<Link> links = handler.getLinks();
    if (links.size() < 2)
        throw new IOException("Must have installed at least 2 versions!");
    Collections.sort(links, new Comparator<Link>() {

        public int compare(Link o1, Link o2) {
            return o1.getText().compareTo(o2.getText());
        }
    });
    this.updateVersion(links.get(links.size() - 2).getText());
}

Example 42 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class SimpleComparerTest method testGetContent.

@Test
public void testGetContent() throws Exception {
    Metadata m = new Metadata();
    m.add(RecursiveParserWrapper.TIKA_CONTENT, "0123456789");
    Map<Cols, String> data = new HashMap<>();
    String content = getContent(m, 10, data);
    assertEquals(10, content.length());
    assertEquals("FALSE", data.get(Cols.CONTENT_TRUNCATED_AT_MAX_LEN));
    content = getContent(m, 4, data);
    assertEquals(4, content.length());
    assertEquals("TRUE", data.get(Cols.CONTENT_TRUNCATED_AT_MAX_LEN));
    //test Metadata with no content
    content = getContent(new Metadata(), 10, data);
    assertEquals(0, content.length());
    assertEquals("FALSE", data.get(Cols.CONTENT_TRUNCATED_AT_MAX_LEN));
    //test null Metadata
    content = getContent(null, 10, data);
    assertEquals(0, content.length());
    assertEquals("FALSE", data.get(Cols.CONTENT_TRUNCATED_AT_MAX_LEN));
}

Example 43 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class SimpleComparerTest method testAttachmentCounts.

@Test
public void testAttachmentCounts() {
    List<Metadata> list = new ArrayList<>();
    Metadata m0 = new Metadata();
    //bad data should be ignored
    m0.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "dir1/dir2/file.zip");
    //in the first metadata object
    list.add(m0);
    Metadata m1 = new Metadata();
    m1.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "/f1.docx/f2.zip/text1.txt");
    list.add(m1);
    Metadata m2 = new Metadata();
    m2.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "/f1.docx/f2.zip/text2.txt");
    list.add(m2);
    Metadata m3 = new Metadata();
    m3.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "/f1.docx/f2.zip");
    list.add(m3);
    Metadata m4 = new Metadata();
    m4.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "/f1.docx");
    list.add(m4);
    Metadata m5 = new Metadata();
    m5.set(RecursiveParserWrapper.EMBEDDED_RESOURCE_PATH, "/f1.docx/text3.txt");
    list.add(m5);
    List<Integer> counts = AbstractProfiler.countAttachments(list);
    List<Integer> expected = new ArrayList<>();
    expected.add(5);
    expected.add(0);
    expected.add(0);
    expected.add(2);
    expected.add(4);
    expected.add(0);
    assertEquals(expected, counts);
}

Example 44 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class ExtractEmbeddedFiles method extract.

public void extract(InputStream is, Path outputDir) throws SAXException, TikaException, IOException {
    Metadata m = new Metadata();
    ParseContext c = new ParseContext();
    ContentHandler h = new BodyContentHandler(-1);
    c.set(Parser.class, parser);
    EmbeddedDocumentExtractor ex = new MyEmbeddedDocumentExtractor(outputDir, c);
    c.set(EmbeddedDocumentExtractor.class, ex);
    parser.parse(is, h, m, c);
}

Example 45 with Metadata

use of org.apache.tika.metadata.Metadata in project tika by apache.

the class Language method languageDetectionWithHandler.

public static void languageDetectionWithHandler() throws Exception {
    LanguageHandler handler = new LanguageHandler();
    new AutoDetectParser().parse(System.in, handler, new Metadata(), new ParseContext());
    LanguageResult result = handler.getLanguage();
    System.out.println(result.getLanguage());
}