高效截断字符串副本`str`到`[u8]`（utf8 aware strlcpy）？

Question

While Rust provides str.as_bytes , I'm looking to copy a string into a fixed sized buffer, where only full unicode-scalar-values are copied into the buffer, and are instead truncated with a null terminator written at the end, in C terms, I'd call this a utf8 aware strlcpy (that is - it copies into a fixed size buffer and ensures its null terminated) .

---

This is a function I came up with, but I expect there are better ways to do this in Rust:

// return the number of bytes written to
pub fn strlcpy_utf8(utf8_dst: &mut [u8], str_src: &str) -> usize {
    let utf8_dst_len = utf8_dst.len();
    if utf8_dst_len == 0 {
        return 0;
    }
    let mut index: usize = 0;
    if utf8_dst_len > 1 {
        let mut utf8_buf: [u8; 4] = [0; 4];
        for c in str_src.chars() {
            let len_utf8 = c.len_utf8();
            let index_next = index + len_utf8;
            c.encode_utf8(&mut utf8_buf);
            if index_next >= utf8_dst_len {
                break;
            }
            utf8_dst[index..index_next].clone_from_slice(&utf8_buf[0..len_utf8]);
            index = index_next;
        }
    }
    utf8_dst[index] = 0;
    return index + 1;
}

---

Note): I realize this isn't ideal since multiple UCS may make up a single glyph, however the result will at least be able to decoded back into a str .

Answer 1

Rust's str has a handy method char_indices for when you need to know the actual character boundaries. This would immediately simplify your function somewhat:

pub fn strlcpy_utf8(utf8_dst: &mut [u8], str_src: &str) -> usize {
    let utf8_dst_len = utf8_dst.len();
    if utf8_dst_len == 0 {
        return 0;
    }
    let mut last_index = 0;
    for (idx, _) in str_src.char_indices() {
        if (idx+1) > utf8_dst_len {
            break;
        }
        last_index = idx;
    }
    utf8_dst[0..last_index].copy_from_slice(&str_src.as_bytes()[0..last_index]);
    utf8_dst[last_index] = 0;
    return last_index + 1;
}

Playground

However you don't actually need to iterate through every character except when copying, as it turns out it's easy to find a boundary in UTF8; Rust has str::is_char_boundary() . This lets you instead look backwards from the end:

pub fn strlcpy_utf8(utf8_dst: &mut [u8], str_src: &str) -> usize {
    let utf8_dst_len = utf8_dst.len();
    if utf8_dst_len == 0 {
        return 0;
    }
    let mut last_index = min(utf8_dst_len-1, str_src.len());
    while last_index > 0 && !str_src.is_char_boundary(last_index) {
        last_index -= 1;
    }
    utf8_dst[0..last_index].copy_from_slice(&str_src.as_bytes()[0..last_index]);
    utf8_dst[last_index] = 0;
    return last_index + 1;
}

Playground

Answer 2

Based on Chris Emerson's answer and @Matthieu-m's suggestion to remove a redundant check.

// returns the number of bytes written to
pub fn strlcpy_utf8(utf8_dst: &mut [u8], str_src: &str) -> usize {
    let utf8_dst_len = utf8_dst.len();
    if utf8_dst_len == 0 {
        return 0;
    }
    // truncate if 'str_src' is too long
    let mut last_index = str_src.len();
    if last_index >= utf8_dst_len {
        last_index = utf8_dst_len - 1;
        // no need to check last_index > 0 here,
        // is_char_boundary covers that case
        while !str_src.is_char_boundary(last_index) {
            last_index -= 1;
        }
    }
    utf8_dst[0..last_index].clone_from_slice(&str_src.as_bytes()[0..last_index]);
    utf8_dst[last_index] = 0;
    return last_index + 1;
}

---

@ChrisEmerson: I'm posting this since it's the code I'm going to use for my project, feel free to update your answer with the changes if you like and I'll remove this answer.