How to edit MS Word documents using Java?

Question

I do have few Word templates, and my requirement is to replace some of the words/place holders in the document based on the user input, using Java. I tried lot of libraries

Answer 1

One may use for docx (a zip with XML and other files) a java zip file system and XML or text processing.

URI docxUri = ,,, // "jar:file:/C:/... .docx"
Map<String, String> zipProperties = new HashMap<>();
zipProperties.put("encoding", "UTF-8");
try (FileSystem zipFS = FileSystems.newFileSystem(docxUri, zipProperties)) {
    Path documentXmlPath = zipFS.getPath("/word/document.xml");

When using XML:

    DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();

    factory.setNamespaceAware(true);
    DocumentBuilder builder = factory.newDocumentBuilder();

    Document doc = builder.parse(Files.newInputStream(documentXmlPath));
    //Element root = doc.getDocumentElement();

You can then use XPath to find the places, and write the XML back again.

It even might be that you do not need XML but could replace place holders:

    byte[] content = Files.readAllBytes(documentXmlPath);
    String xml = new String(content, StandardCharsets.UTF_8);
    xml = xml.replace("#DATE#", "2014-09-24");
    xml = xml.replace("#NAME#", StringEscapeUtils.escapeXml("Sniper")));
    ...
    content = xml.getBytes(StandardCharsets.UTF_8);
    Files.delete(documentXmlPath);
    Files.write(documentXmlPath, content);

For a fast development, rename a copy of the .docx to a name with the .zip file extension, and inspect the files.

File.write should already apply StandardOpenOption.TRUNCATE_EXISTING, but I have added Files.delete as some error occured. See comments.

Answer 2

Try Apache POI. POI can work with doc and docx, but docx is more documented therefore support of it better.

UPD: You can use XDocReport, which use POI. Also I recomend to use xlsx for templates because it more suitable and more documented

Answer 3

I have spent a few days on this issue, until I found that what makes the difference is the try-with-resources on the FileSystem instance, appearing in Joop Eggen's snippet but not in question snippet:
try (FileSystem zipFS = FileSystems.newFileSystem(docxUri, zipProperties))
Without such try-with-resources block, the FileSystem resource will not be closed (as explained in Java tutorial), and the word document not modified.

Answer 4

Stepping back a bit, there are about 4 different approaches for editing words/placeholders:

• MERGEFIELD or DOCPROPERTY fields (if you are having problems with this in docx4j, then you have probably not set up your input docx correctly)
• content control databinding
• variable replacement on the document surface (either at the DOM/SAX level, or using a library)
• do stuff as XHTML, then import that

Before choosing one, you should decide whether you also need to be able to handle:

• repeating data (eg adding table rows)
• conditional content (eg entire paragraphs which will either be present or absent)
• adding images

If you need these, then MERGEFIELD or DOCPROPERTY fields are probably out (though you can also use IF fields, if you can find a library which supports them). And adding images makes DOM/SAX manipulation as advocated in one of the other answers, messier and error prone.

The other things to consider are:

• your authors: how technical are they? What does that imply for the authoring UI?
• the "user input" you mention for variable replacement, is this given, or is obtaining it part of the problem you are solving?