Examples with OutlookTextExtactor org.apache.poi.hsmf.extractor.OutlookTextExtactor used on opensource projects

Example 6 with OutlookTextExtactor

use of org.apache.poi.hsmf.extractor.OutlookTextExtactor in project poi by apache.

the class OLE2ScratchpadExtractorFactory method identifyEmbeddedResources.

/**
     * Returns an array of text extractors, one for each of
     *  the embedded documents in the file (if there are any).
     * If there are no embedded documents, you'll get back an
     *  empty array. Otherwise, you'll get one open
     *  {@link POITextExtractor} for each embedded file.
     */
public static void identifyEmbeddedResources(POIOLE2TextExtractor ext, List<Entry> dirs, List<InputStream> nonPOIFS) throws IOException {
    // Find all the embedded directories
    DirectoryEntry root = ext.getRoot();
    if (root == null) {
        throw new IllegalStateException("The extractor didn't know which POIFS it came from!");
    }
    if (ext instanceof WordExtractor) {
        // These are in ObjectPool -> _... under the root
        try {
            DirectoryEntry op = (DirectoryEntry) root.getEntry("ObjectPool");
            Iterator<Entry> it = op.getEntries();
            while (it.hasNext()) {
                Entry entry = it.next();
                if (entry.getName().startsWith("_")) {
                    dirs.add(entry);
                }
            }
        } catch (FileNotFoundException e) {
        // ignored here
        }
    //} else if(ext instanceof PowerPointExtractor) {
    // Tricky, not stored directly in poifs
    // TODO
    } else if (ext instanceof OutlookTextExtactor) {
        // Stored in the Attachment blocks
        MAPIMessage msg = ((OutlookTextExtactor) ext).getMAPIMessage();
        for (AttachmentChunks attachment : msg.getAttachmentFiles()) {
            if (attachment.getAttachData() != null) {
                byte[] data = attachment.getAttachData().getValue();
                nonPOIFS.add(new ByteArrayInputStream(data));
            } else if (attachment.getAttachmentDirectory() != null) {
                dirs.add(attachment.getAttachmentDirectory().getDirectory());
            }
        }
    }
}

Example 7 with OutlookTextExtactor

use of org.apache.poi.hsmf.extractor.OutlookTextExtactor in project poi by apache.

the class TestExtractorFactory method testPOIFS.

@Test
public void testPOIFS() throws Exception {
    // Excel
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(xls))) instanceof ExcelExtractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(xls))).getText().length() > 200);
    // Word
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc))) instanceof WordExtractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc))).getText().length() > 120);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc6))) instanceof Word6Extractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc6))).getText().length() > 20);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc95))) instanceof Word6Extractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(doc95))).getText().length() > 120);
    // PowerPoint
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(ppt))) instanceof PowerPointExtractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(ppt))).getText().length() > 120);
    // Visio
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(vsd))) instanceof VisioTextExtractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(vsd))).getText().length() > 50);
    // Publisher
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(pub))) instanceof PublisherTextExtractor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(pub))).getText().length() > 50);
    // Outlook msg
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg))) instanceof OutlookTextExtactor);
    assertTrue(ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(msg))).getText().length() > 50);
    // Text
    try {
        ExtractorFactory.createExtractor(new POIFSFileSystem(new FileInputStream(txt)));
        fail();
    } catch (IOException e) {
    // Good
    }
}

Example 8 with OutlookTextExtactor

use of org.apache.poi.hsmf.extractor.OutlookTextExtactor in project poi by apache.

the class TestExtractorFactory method testFile.

@Test
public void testFile() throws Exception {
    // Excel
    POITextExtractor xlsExtractor = ExtractorFactory.createExtractor(xls);
    assertNotNull("Had empty extractor for " + xls, xlsExtractor);
    assertTrue("Expected instanceof ExcelExtractor, but had: " + xlsExtractor.getClass(), xlsExtractor instanceof ExcelExtractor);
    assertTrue(xlsExtractor.getText().length() > 200);
    xlsExtractor.close();
    POITextExtractor extractor = ExtractorFactory.createExtractor(xlsx);
    assertTrue(extractor.getClass().getName(), extractor instanceof XSSFExcelExtractor);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(xlsx);
    assertTrue(extractor.getText().length() > 200);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(xltx);
    assertTrue(extractor.getClass().getName(), extractor instanceof XSSFExcelExtractor);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(xlsb);
    assertContains(extractor.getText(), "test");
    extractor.close();
    extractor = ExtractorFactory.createExtractor(xltx);
    assertContains(extractor.getText(), "test");
    extractor.close();
    // TODO Support OOXML-Strict, see bug #57699
    try {
        /*extractor =*/
        ExtractorFactory.createExtractor(xlsxStrict);
        fail("OOXML-Strict isn't yet supported");
    } catch (POIXMLException e) {
    // Expected, for now
    }
    //        extractor = ExtractorFactory.createExtractor(xlsxStrict);
    //        assertTrue(
    //                extractor
    //                instanceof XSSFExcelExtractor
    //        );
    //        extractor.close();
    //
    //        extractor = ExtractorFactory.createExtractor(xlsxStrict);
    //        assertTrue(
    //                extractor.getText().contains("test")
    //        );
    //        extractor.close();
    // Word
    extractor = ExtractorFactory.createExtractor(doc);
    assertTrue(extractor instanceof WordExtractor);
    assertTrue(extractor.getText().length() > 120);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(doc6);
    assertTrue(extractor instanceof Word6Extractor);
    assertTrue(extractor.getText().length() > 20);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(doc95);
    assertTrue(extractor instanceof Word6Extractor);
    assertTrue(extractor.getText().length() > 120);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(docx);
    assertTrue(extractor instanceof XWPFWordExtractor);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(docx);
    assertTrue(extractor.getText().length() > 120);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(dotx);
    assertTrue(extractor instanceof XWPFWordExtractor);
    extractor.close();
    extractor = ExtractorFactory.createExtractor(dotx);
    assertContains(extractor.getText(), "Test");
    extractor.close();
    // PowerPoint (PPT)
    extractor = ExtractorFactory.createExtractor(ppt);
    assertTrue(extractor instanceof PowerPointExtractor);
    assertTrue(extractor.getText().length() > 120);
    extractor.close();
    // PowerPoint (PPTX)
    extractor = ExtractorFactory.createExtractor(pptx);
    assertTrue(extractor instanceof XSLFPowerPointExtractor);
    assertTrue(extractor.getText().length() > 120);
    extractor.close();
    // Visio - binary
    extractor = ExtractorFactory.createExtractor(vsd);
    assertTrue(extractor instanceof VisioTextExtractor);
    assertTrue(extractor.getText().length() > 50);
    extractor.close();
    // Visio - vsdx
    extractor = ExtractorFactory.createExtractor(vsdx);
    assertTrue(extractor instanceof XDGFVisioExtractor);
    assertTrue(extractor.getText().length() > 20);
    extractor.close();
    // Publisher
    extractor = ExtractorFactory.createExtractor(pub);
    assertTrue(extractor instanceof PublisherTextExtractor);
    assertTrue(extractor.getText().length() > 50);
    extractor.close();
    // Outlook msg
    extractor = ExtractorFactory.createExtractor(msg);
    assertTrue(extractor instanceof OutlookTextExtactor);
    assertTrue(extractor.getText().length() > 50);
    extractor.close();
    // Text
    try {
        ExtractorFactory.createExtractor(txt);
        fail();
    } catch (IllegalArgumentException e) {
    // Good
    }
}

Example 9 with OutlookTextExtactor

use of org.apache.poi.hsmf.extractor.OutlookTextExtactor in project poi by apache.

the class TestFixedSizedProperties method testReadMessageDateFailsWithOutlookTextExtractor.

/**
    * Test to see if we can read the Date Chunk with OutlookTextExtractor.
    */
@Test
public // @Ignore("TODO Work out why the Thu 21st vs Monday 25th problem is occurring and fix")
void testReadMessageDateFailsWithOutlookTextExtractor() throws Exception {
    OutlookTextExtactor ext = new OutlookTextExtactor(mapiMessageFails);
    // Don't close re-used test resources here
    ext.setFilesystem(null);
    String text = ext.getText();
    assertContains(text, "Date: Thu, 21 Jun 2012 14:14:04 +0000\n");
    ext.close();
}