Apache POI replaceText() side effect, changes line space

Question

I am using POI 3.15 in Java to replace some text in my .doc template.

private HWPFDocument replaceText(HWPFDocument doc, String findText, String replaceText) {
    Range r = doc.getRange();
    for (int i = 0; i < r.numSections(); ++i) {
        Section s = r.getSection(i);
        for (int j = 0; j < s.numParagraphs(); j++) {
            Paragraph p = s.getParagraph(j);
            for (int k = 0; k < p.numCharacterRuns(); k++) {
                CharacterRun run = p.getCharacterRun(k);
                String text = run.text();
                if (text.contains(findText)) {
                    run.replaceText(findText, replaceText);
                }
            }
        }
    }
    return doc;
}

After I save the document. All content inside are correct. But the style of the document is not. The space between lines is changed. The original gap between lines is missing. All line are closely packed together.

Why? How do I keep the style of my template?

Answer 1

The HWPF library may not support all features, which exist in your doc file and this may result in changed formats. It may also result in unreadable files.

Some years ago I created a customized HWPF library, which could properly modify and write a wide variety of doc files for one of my clients and I gained a lot of experience about the doc file format and the HWPF library.

The problem is, that one has to properly support all features in HWPF, which may be present in the doc file. For instance, if clipart is included in the file, there will be separate tables, which maintain position and properties of the cliparts. If the content (text) is changed without adjusting the addresses in the other internal tables, formats etc. can be shifted, ignored or lost. (or in worst case, the document is unreadable)

I am not sure about the status of HWPF these days, but I expect, that it does not fully support the main relevant doc file features.

If you want to use HWPF for modifying / writing doc files, you may succeed with files, which have a reduced "feature set". For instance no tables, no cliparts, no text boxes - things like that. If you need to support almost any document, which a user may provide, I'd recommend to find a different solution.

One option could be to use rtf files, which are named .doc. Or use the XWPF library, which works for .docx files.