take_while_m_n splits at wrong index

take_while_m_n splits a string slice with multibyte characters (e.g., arab letters) at the wrong position. input.position(condition) returns the byte to split at (6 in the following example case) while input.splice_index(idx) takes that byte position as char index and translates that index to byte index (9 in the example), effectively splitting at the wrong byte position (9 instead of 6).

Rust version: rustc 1.67.1 (d5a82bbd2 2023-02-07)
nom version: 7.1.3

Test case

Example test case:

use nom::IResult;
use nom::bytes::complete::take_while_m_n;

fn is_arab_letter(c: char) -> bool {
    match c {
        '\u{061c}'..='\u{064a}' => true,
        _ => false
    }
}

fn arab_string(s: &str) -> IResult<&str, &str> {
    take_while_m_n(0, 64, is_arab_letter)(s)
}

fn main() {
    assert_eq!(arab_string("باب latin"), Ok((&" latin"[..], &"باب"[..])));
}

Output:

thread 'main' panicked at 'assertion failed: `(left == right)`
  left: `Ok(("tin", "باب la"))`,
 right: `Ok((" latin", "باب"))`', src/main.rs:16:5

rust-bakery / nom

take_while_m_n splits at wrong index #1642

Test case