How to read Unicode codepoints from an unbuffered file in a mio event loop?

295 views Asked by At

I'd like to use the mio crate to read keypresses as they arrive in an unbuffered fashion. I already have the code to unbuffer stdin, and I have scaffolded the event loop:

extern crate mio;
extern crate termios;

use termios::{Termios, TCSANOW, ICANON, ECHO, tcsetattr};
use mio::*;
use mio::unix::EventedFd;

fn unbuffer_stdin() {
    let termios = Termios::from_fd(0).unwrap();
    let mut new_termios = termios.clone();
    new_termios.c_lflag &= !(ICANON | ECHO);
    tcsetattr(0, TCSANOW, &mut new_termios).unwrap();
}

fn main() {
    let stdin = 0;
    unbuffer_stdin();

    let poll = Poll::new().unwrap();

    const STDIN: Token = Token(0);
    let ev_fd = EventedFd(&stdin);
    poll.register(&ev_fd, STDIN, Ready::readable(), PollOpt::edge()).unwrap();

    let mut events = Events::with_capacity(1024);
    loop {
        poll.poll(&mut events, None).unwrap();

        for event in events.iter() {
            match event.token() {
                STDIN => {
                    println!("keypress");
                    // XXX read in ready codepoints to a buffer
                }
                _ => unreachable!(),
            }
        }
    }
}

How do I implement the part marked XXX? There are a couple of challenges:

  • How do I know how many bytes to read? I'm not sure mio tells me this.
  • How do I deal with partial codepoints, where the read spans a multi-byte character.

One solution that might work is to use oneshot events instead of edge events, then read one byte per event into a temporary buffer. Each time the buffer makes sense as a complete codepoint, I could then convert it to a char and store it away, and clear the scratch buffer.

This seems a little inefficient though. What is the best way?

0

There are 0 answers