how create a Word document from a template or existing document with Java?

Question

I have a document template where some fields are static and others are dynamic. I need to replace some data (name, last name, salary) and generate the new file. What library do you recommend to do this? Is POI appropriate? I am working with Spring, Java EE6 and Oracle.

Answer 1

You can give Apache POI a try but the HWPF and XWPF part of POI which are required to manipulate word files are really complicated to use - you need to have at least a good understanding how a word file is structured!

Solution using iText and PDF

I did something similar with PDF (this might be an option for you)

1) You can use LibreOffice to create fields in the document (like in Acrobat Pro)

• Create an .odt file and style it
• or convert your template to it using MS Word or LibreOffice Writer
• Then go to View -> Toolsbars -> Form Design and set "Design Mode on/off"
• Now you can add fields to your file (double clicking on it will open properties of the fields)
• When you are finished: "File -> Export as PDF"

2) Now you can use iText to fill in the created fields

The following is just example code:

    public byte[] getDocumentAsByteArray(Object dataBean, String pdfTemplateName) throws KkmsException {

    ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
    PdfStamper stamp = null;
    InputStream templateInputStream = null;

    Locale local = new Locale(language);

    try {
        templateInputStream = // get the file input stream of the pdf
        PdfReader reader = new PdfReader(templateInputStream);

        // Create a stamper that will copy the document to a new file
        stamp = new PdfStamper(reader, outputStream);

        AcroFields form = stamp.getAcroFields();

        // form fields are normal text in the end
        stamp.setFormFlattening(true);
        Map<String, AcroFields.Item> map = (Map<String, AcroFields.Item>)form.getFields();
        if (map != null) {
            if (map.size() == 0) {
                logger.debug("There are no fields in this PDF layout");
            }
            for (Entry<String, AcroFields.Item> e : map.entrySet()) {
                logger.debug("PDF fieldname = " + e.getKey());

                // at the moment we only handle text fields
                if (AcroFields.FIELD_TYPE_TEXT == form.getFieldType(e.getKey())) {
                    fillForm(dataBean, form, e.getKey(), local);
                } else {
                    logger.warn("Field type is not supported: "+form.getFieldType(e.getKey()));
                }
            }
        }

        stamp.close();
    } catch (Exception e) {
        logger.warn("Failed to create PDF document", e);
        throw new KkmsException("Failed to create PDF document: "+e.getMessage());
    } finally {
        if (templateInputStream != null) {
            try {
                templateInputStream.close();
            } catch (IOException e) {
                throw new KkmsException("Failed to close InputStream of PDF document", e);
            }
        }
    }
    return outputStream.toByteArray();
}

In the end you get a PDF -> hope this helps you at least a little bit!

Another quick and dirty solution

Could be to use the power of odt or docx -> convert your doc to docx or odt -> its just a zip file -> so unzip it -> you will see a content.xml file in the root of the zip -> there is all the document content in there Now you could add some magic tags (e.g. $$$) here which can later be replaced by your program

<text:p text:style-name="P3">SAP Customer Number:</text:p>

<text:p text:style-name="P3">SAP Customer Number: $$$sapCustomerNumber$$$</text:p>

Now create a program which unzips the odt/docx file -> replaces the tags -> zips the file again

Answer 2

These slides, from a presentation I gave at OSDC 2012, outline some major approaches.

These days I'd likely add "generate what you want as XHTML, then export that to docx". Since we introduced docx4j-ImportXHTML with support for converting CSS @class values to Word styles, we've seen this approach more and more.