[TIKA-690] WordExtractor doesn't extract text from HWPFDocument

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Not A Problem
• Affects Version/s: 0.9, 0.10
• Component/s: parser
• Labels: parsing
• Assignee: Unassigned
• Reporter: Joseph Vychtrle
• Created: 14/Aug/11 13:17
• Updated: 14/Aug/11 18:30
• Resolved: 14/Aug/11 18:29

Description

If I use apache poi's HWPF component to create MS doc, and pass it to tika.parseToString(is); it returns just carriage return "\n". I tested that with tons of different input text. Adding paragraphs doesn't help.

private void createDOCDocument(String from, File file) throws Exception {

    POIFSFileSystem fs = new POIFSFileSystem(DOCGenerator.class.getClass().getResourceAsStream("/poi/template.doc"));
    HWPFDocument doc = new HWPFDocument(fs);

    Range range = doc.getRange();
    CharacterRun run1 = range.insertBefore(from);
    run1.setFontSize(11);

    DocumentSummaryInformation dsi = doc.getDocumentSummaryInformation();
    CustomProperties cp = dsi.getCustomProperties();
    if (cp == null)
        cp = new CustomProperties();
    cp.put("myProperty", "foo bar baz");
    dsi.setCustomProperties(cp);

    doc.write(new FileOutputStream(file));
}

protected String extractText(InputStream is) throws SystemException {
	Tika tika = new Tika();
	tika.setMaxStringLength(new Long(maxCharCount).intValue());
	String text;
	try {
		text = tika.parseToString(is);
	} catch (IOException ioe) {
		throw new SystemException(ioe.getMessage(), ioe);
	} catch (TikaException te) {
		throw new SystemException(te.getMessage(), te);
	}
	return text;

}