From what I remember RTF is the easiest to implement from common filetypes, and openable in office suites like MS or LibreOffice. Then, in theory (unless it breaks a bunch of stuff inside), you can save it as DOCX or something. Anyone has experience with this?
The early WordPerfect file format is pretty straightforward. It pretty much mirrors what you see in Reveal Codes, which is a single stream of text interspersed with special inline codes that switch things on/off (eg bold) or do something special (eg table of contents).
As it's linear you can just whizz along taking plain text as it is, converting features as you reach them, and discarding the ones you don't care about. I did it myself in C# for WP4 (DOS) files to Markdown, but RTF is also trivial to write to. Search for the filename "WP42FF.TXT" and you'll find the format online.
Incidentally, DOCX is just a ZIP file with special contents. Make a single sentence document in Word and save it out, rename the file to .ZIP and extract it. The contents are obvious when you look through them, and you can use that as a 'skeleton' and switch the text with your own. For RTF look for a copy of "RTF Pocket Guide" by Sean Burke.
Here's how the free law project parses it: wpd2html
https://github.com/freelawproject/doctor/blob/main/doctor/ta...