[Plugin] Whitaker's WORDS running natively on Kindle (KOReader)

[ddd]

Hey!

I've been working on a library that reimplements Whitaker's WORDS in a form that can run efficiently in modern web browsers, but also on low-memory devices like an old Kindle.

The reason I started was pretty simple. I spend a lot of my day at a computer, and I wanted to be able to learn to read Latin away from it, using my Kindle.

The problem I was running into was the delay involved in looking up words. Having to wait for a translation or for a slower application to open takes away some of the fun and flow of reading for me.

So I built WhitakerLib, a native C++ reimplementation of Whitaker's WORDS, which now serves as the backend for a KOReader plugin.

One thing worth mentioning is that this isn't a dictionary containing precomputed results from WORDS. It's the analyzer itself, reimplemented to run directly on the Kindle. The whole thing, including everything needed for the Latin analysis, fits in a roughly 3.5 MB binary. No internet connection or external dictionary needed, and it does it fast, very fast.

Even on my old Paperwhite 3, the Kindle can analyze a word much faster than it can update the screen to show the result.

I probably spent more time making the analysis faster than I needed to, mostly because I enjoyed it. The original goal was just to make looking up words feel instant, but once I got there, I wanted to see how much further I could take it. So now the Latin analysis is pretty much free for what I'm trying to do on the Kindle.

Currently, the plugin is still a fairly thin wrapper around the library. I've tried to make the results from WORDS a bit easier to read and navigate, but I don't think the reading experience is that good yet.

This is probably where I'll be spending most of my time next. The analysis is already fast enough for what I wanted, although there's still plenty to improve in the library itself, like accuracy across more books, macron support, and a few other things. For the Kindle plugin, I'd rather focus on making the results more useful when reading.

I'd be interested to hear how people might use it, what could be improved, and whether it works on other devices. So far, I've only tested it on the one I own!

I've packaged the plugin as a ZIP for my Kindle, so there's no need to compile anything yourself if you're using a compatible device. Installation instructions are on GitHub.

The library uses C++23, which not everyone has a suitable toolchain for. So if anyone wants to try it on another e-reader but can't compile it themselves, just let me know! I'd be happy to try making a ZIP for their device as well.

https://github.com/ddd2951/whitakerLib-kindle

I noticed other developers have shared their KOReader plugins here, so I hope this is the right place. If I've misunderstood the forum rules, apologies!