#utf8 #conversion #stream #iterator

yanked chisel-charstream

Simple and relatively fast u8 -> char stream implementation

0.1.3 Mar 17, 2023
0.1.2 Mar 17, 2023
0.1.1 Mar 17, 2023
0.1.0 Mar 17, 2023

#41 in #utf8

28 downloads per month


261 lines

Chisel - CharStream



This repository contains a very simple, lean implementation of a transcoder that will consume u8 bytes from a given Read implementation, and convert into the Rust internal char type. This is an offshoot lib from an ongoing toy parser project, and is used as the first stage of the scanning/lexing phase of the parser in order avoid unnecessary allocations during the u8 sequence -> char conversion.

Note that the implementation is pretty fast and loose, and under the covers utilises some bit-twiddlin' in conjunction with the unsafe transmute function to do the conversions. No string allocations are used during conversion. There is minimal checking (other than bit-masking) of the inbound bytes - it is not intended to be a UTF8 validation library.


Usage is very simple, provided you have something that implements Read in order to source some bytes:

Create from a slice

Just wrap your array in a reader, and then plug it into a new instance of CharStream:

    let buffer: &[u8] = &[0x10, 0x12, 0x23, 0x12];
    let mut reader = BufReader::new(buffer);
    let _stream = CharStream::new(&mut reader);

Create from a file

Just crack open your file, wrap in a Read instance and then plug into a new instance of CharStream:

    let path = PathBuf::from("somefile.txt");
    let f = File::open(path);
    let mut reader = BufReader::new(f.unwrap());
    let _stream = CharStream::new(&mut reader);

Consuming chars

You can either pull out new chars from the reader wrapped inside a Result type:

    loop {
        let result = stream.next_char();
        if result.is_err() {

Alternatively, you can just use the CharStream as an Iterator:

    let stream = CharStream::new(&mut reader);
    for c in stream {
        match c {
            Some(c) => ...
            None => ...

No runtime deps