MicroPython fork

[Josverl]

Just a small thank you, and to let you know that I've been able to re-use this work for MicroPython in a reduced form.

https://github.com/Josverl/mp_wcwidth

[jquast]

Very nice work, @Josverl thanks for sharing. Looks like you're behind a few versions :) I've been busy in the past few weeks, do you think you will also implement width(), center(), rjust(), wrap(), iter_graphemes(), etc. ?

I added these new API functions in 0.3.0 because of thorough analysis of downstream dependencies and found wc/swidth() is not used correctly in almost every case, so I am offering these more correct and battle-tested solutions (mostly ported from blessed library)

I'm surprised by your decision to use a small self-clearing dict_cache of size 128, this would give sawtooth performance variation, but I see lru_cache is not available in micropython which is surprising, it is very efficient last I looked, it I see that lru_cache is not in std micropython so that explains it -- I've looked at its source in python and it doesn't appear to use anything unavailable to micropython but this is eyesight only I may be mistaken. In any case, we use "UDHR Datasets" and have benchmarks now in this repository you might borrow from if performance is important to you.

Version 0.2.14 release accidentally included a table_vs15.py file, leftover from code generation of a WIP branch, it was erroneously included in the sdist file on pypi. it is interesting that it looks like your code agent picked it up, it was right to remove it (#120)

[Josverl]

Looks like you're behind a few versions

I noticed the updates.

on the small self-clearing dict_cache
MCUs only have so much memory, and I could not find a decent implementation , so built a mediocre one that can be improved/replaced later.
Indeed, it might be good to bring lru_cache to MicroPython.

My current goal is to see if I can improve the MicroPython REPL and input / readline processing wrt to unicode.
For that it needs to have some awareness of mapping code-points to cursor movements , across 4 different host platforms ( Linux(es), MacOS, Windows, Web/WASM ) , so that will be fun .
Depending on what I need for that I may add additional functionality, and then likely in a lazy-import fashion.

If I may ask: what would you suggest as the best/proper API to use for that ?

[jquast]

I designed wcwidth.width() to process most terminal output sequences and discover the maximum "real estate" taken up, so "AA\b\b" is 2, but, I also briefly entertained an option to measure the "final cursor position", so that "AA\b\b" is 0, which might be what you're looking for, so I regret I pulled it out.

In any case the algorithm for a single-line REPL would be some variation of the width() function.

For a multi-line REPL, you're basically writing an editor, and something like pyte might help. Or at least to take a look.

For the ability to move your cursor around complex ZWJ emoji, Flags, or even just simple combining takes some care, while editing, like a REPL would want, you'll need to use iter_graphemes() like what is done in pyte updated by my pr

And anytime you get lost, you can always ask the terminal "where is the cursor?" but there is a round-trip delay over network/serial/etc