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

Example 71 with ParseContext

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

the class ExcelParserTest method testExcelParser.

@Test
// Checks legacy Tika-1.0 style metadata keys
@SuppressWarnings("deprecation")
public void testExcelParser() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Simple Excel document", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Keith Bennett", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Keith Bennett", metadata.get(Metadata.AUTHOR));
        // Mon Oct 01 17:13:56 BST 2007
        assertEquals("2007-10-01T16:13:56Z", metadata.get(TikaCoreProperties.CREATED));
        assertEquals("2007-10-01T16:13:56Z", metadata.get(Metadata.CREATION_DATE));
        // Mon Oct 01 17:31:43 BST 2007
        assertEquals("2007-10-01T16:31:43Z", metadata.get(TikaCoreProperties.MODIFIED));
        assertEquals("2007-10-01T16:31:43Z", metadata.get(Metadata.DATE));
        String content = handler.toString();
        assertContains("Sample Excel Worksheet", content);
        assertContains("Numbers and their Squares", content);
        assertContains("\t\tNumber\tSquare", content);
        assertContains("9", content);
        assertNotContained("9.0", content);
        assertContains("196", content);
        assertNotContained("196.0", content);
    }
}

Example 72 with ParseContext

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

the class ExcelParserTest method testHeaderAndFooterExtraction.

@Test
public void testHeaderAndFooterExtraction() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_headers_footers.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.UK);
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals("Internal spreadsheet", metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Aeham Abushwashi", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("Aeham Abushwashi", metadata.get(Metadata.AUTHOR));
        String content = handler.toString();
        assertContains("John Smith1", content);
        assertContains("John Smith50", content);
        assertContains("1 Corporate HQ", content);
        assertContains("Header - Corporate Spreadsheet", content);
        assertContains("Header - For Internal Use Only", content);
        assertContains("Header - Author: John Smith", content);
        assertContains("Footer - Corporate Spreadsheet", content);
        assertContains("Footer - For Internal Use Only", content);
        assertContains("Footer - Author: John Smith", content);
    }
}

Example 73 with ParseContext

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

the class ExcelParserTest method testWorksSpreadsheet70.

@Test
public void testWorksSpreadsheet70() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testWORKSSpreadsheet7.0.xlr")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler(-1);
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
        String content = handler.toString();
        assertContains("Microsoft Works", content);
    }
}

Example 74 with ParseContext

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

the class ExcelParserTest method testExcelParserPassword.

@Test
public void testExcelParserPassword() throws Exception {
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_protected_passtika.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        new OfficeParser().parse(input, handler, metadata, context);
        fail("Document is encrypted, shouldn't parse");
    } catch (EncryptedDocumentException e) {
    // Good
    }
    // Try again, this time with the password
    try (InputStream input = ExcelParserTest.class.getResourceAsStream("/test-documents/testEXCEL_protected_passtika.xls")) {
        Metadata metadata = new Metadata();
        ContentHandler handler = new BodyContentHandler();
        ParseContext context = new ParseContext();
        context.set(Locale.class, Locale.US);
        context.set(PasswordProvider.class, new PasswordProvider() {

            @Override
            public String getPassword(Metadata metadata) {
                return "tika";
            }
        });
        new OfficeParser().parse(input, handler, metadata, context);
        assertEquals("application/vnd.ms-excel", metadata.get(Metadata.CONTENT_TYPE));
        assertEquals(null, metadata.get(TikaCoreProperties.TITLE));
        assertEquals("Antoni", metadata.get(TikaCoreProperties.CREATOR));
        assertEquals("2011-11-25T09:52:48Z", metadata.get(TikaCoreProperties.CREATED));
        String content = handler.toString();
        assertContains("This is an Encrypted Excel spreadsheet", content);
        assertNotContained("9.0", content);
    }
}

Example 75 with ParseContext

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

the class ExcelParserTest method testTurningOffTextBoxExtractionExcel.

//TIKA-2346
@Test
public void testTurningOffTextBoxExtractionExcel() throws Exception {
    ParseContext pc = new ParseContext();
    OfficeParserConfig officeParserConfig = new OfficeParserConfig();
    officeParserConfig.setIncludeShapeBasedContent(false);
    pc.set(OfficeParserConfig.class, officeParserConfig);
    String xml = getXML("testEXCEL_textbox.xls", pc).xml;
    assertNotContained("autoshape", xml);
}