Examples with ParseContext org.apache.tika.parser.ParseContext used on opensource projects

Example 11 with ParseContext

use of org.apache.tika.parser.ParseContext 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 12 with ParseContext

use of org.apache.tika.parser.ParseContext 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());
}

Example 13 with ParseContext

use of org.apache.tika.parser.ParseContext in project tika by apache.

the class DisplayMetInstance method getMet.

public static Metadata getMet(URL url) throws IOException, SAXException, TikaException {
    Metadata met = new Metadata();
    PDFParser parser = new PDFParser();
    parser.parse(url.openStream(), new BodyContentHandler(), met, new ParseContext());
    return met;
}

Example 14 with ParseContext

use of org.apache.tika.parser.ParseContext in project tika by apache.

the class FontParsersTest method testTTFParsing.

@Test
public void testTTFParsing() throws Exception {
    // Should auto-detect!
    Parser parser = new AutoDetectParser();
    ContentHandler handler = new BodyContentHandler();
    Metadata metadata = new Metadata();
    ParseContext context = new ParseContext();
    try (TikaInputStream stream = TikaInputStream.get(FontParsersTest.class.getResource("/test-documents/testTrueType3.ttf"))) {
        parser.parse(stream, handler, metadata, context);
    }
    assertEquals("application/x-font-ttf", metadata.get(Metadata.CONTENT_TYPE));
    assertEquals("Open Sans Bold", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("2010-12-30T11:04:00Z", metadata.get(Metadata.CREATION_DATE));
    assertEquals("2010-12-30T11:04:00Z", metadata.get(TikaCoreProperties.CREATED));
    assertEquals("2011-05-05T12:37:53Z", metadata.get(TikaCoreProperties.MODIFIED));
    assertEquals("Open Sans Bold", metadata.get(MET_FONT_NAME));
    assertEquals("Open Sans", metadata.get(MET_FONT_FAMILY_NAME));
    assertEquals("Bold", metadata.get(MET_FONT_SUB_FAMILY_NAME));
    assertEquals("OpenSans-Bold", metadata.get(MET_PS_NAME));
    assertEquals("Digitized", metadata.get("Copyright").substring(0, 9));
    assertEquals("Open Sans", metadata.get("Trademark").substring(0, 9));
    // Not extracted
    assertEquals(null, metadata.get(MET_FONT_FULL_NAME));
    assertEquals(null, metadata.get(MET_FONT_WEIGHT));
    assertEquals(null, metadata.get(MET_FONT_VERSION));
    // Currently, the parser doesn't extract any contents
    String content = handler.toString();
    assertEquals("", content);
}

Example 15 with ParseContext

use of org.apache.tika.parser.ParseContext in project tika by apache.

the class TestGDALParser method testParseMetadata.

@Test
public void testParseMetadata() {
    assumeTrue(canRun());
    final String expectedNcInst = "NCAR (National Center for Atmospheric Research, Boulder, CO, USA)";
    final String expectedModelNameEnglish = "NCAR CCSM";
    final String expectedProgramId = "Source file unknown Version unknown Date unknown";
    final String expectedProjectId = "IPCC Fourth Assessment";
    final String expectedRealization = "1";
    final String expectedTitle = "model output prepared for IPCC AR4";
    final String expectedSub8Name = "\":ua";
    final String expectedSub8Desc = "[1x17x128x256] eastward_wind (32-bit floating-point)";
    GDALParser parser = new GDALParser();
    InputStream stream = TestGDALParser.class.getResourceAsStream("/test-documents/sresa1b_ncar_ccsm3_0_run1_200001.nc");
    Metadata met = new Metadata();
    BodyContentHandler handler = new BodyContentHandler();
    try {
        parser.parse(stream, handler, met, new ParseContext());
        assertNotNull(met);
        assertNotNull(met.get("NC_GLOBAL#institution"));
        assertEquals(expectedNcInst, met.get("NC_GLOBAL#institution"));
        assertNotNull(met.get("NC_GLOBAL#model_name_english"));
        assertEquals(expectedModelNameEnglish, met.get("NC_GLOBAL#model_name_english"));
        assertNotNull(met.get("NC_GLOBAL#prg_ID"));
        assertEquals(expectedProgramId, met.get("NC_GLOBAL#prg_ID"));
        assertNotNull(met.get("NC_GLOBAL#prg_ID"));
        assertEquals(expectedProgramId, met.get("NC_GLOBAL#prg_ID"));
        assertNotNull(met.get("NC_GLOBAL#project_id"));
        assertEquals(expectedProjectId, met.get("NC_GLOBAL#project_id"));
        assertNotNull(met.get("NC_GLOBAL#realization"));
        assertEquals(expectedRealization, met.get("NC_GLOBAL#realization"));
        assertNotNull(met.get("NC_GLOBAL#title"));
        assertEquals(expectedTitle, met.get("NC_GLOBAL#title"));
        assertNotNull(met.get("SUBDATASET_8_NAME"));
        assertTrue(met.get("SUBDATASET_8_NAME").endsWith(expectedSub8Name));
        assertNotNull(met.get("SUBDATASET_8_DESC"));
        assertEquals(expectedSub8Desc, met.get("SUBDATASET_8_DESC"));
    } catch (Exception e) {
        e.printStackTrace();
        fail(e.getMessage());
    }
}