I want to write a function that gets a string of hex numbers (two hex numbers represent a u8 value) then returns a vector of u8 values, e.g. the string 1f 0a d1 should be converted into [31, 10, 209]. The input string may contains non hex characters, then the function must be able to filter these characters, such as:

1f \x0a\xD1
\x1F \x0a \xd1
\x1f    \x0A\xd1

All of them lead to the output [31, 10, 209]. My solution is that following:

fn parse_hex(hex_asm: &str) -> Vec<u8> {
    let hex_chars: Vec<char> = hex_asm.as_bytes().iter().filter_map(|b| {
        let ch = char::from(*b);
        if ('0' <= ch && ch <= '9') || ('a' <= ch && ch <= 'f') || ('A' <= ch && ch <= 'F') {
        } else {

    let mut index = 0usize;
    let (odd_chars, even_chars): (Vec<char>, Vec<char>) = hex_chars.into_iter().partition(|_| { 
        index = index + 1;
        index % 2 == 1

    odd_chars.into_iter().zip(even_chars.into_iter()).map(|(c0, c1)| {
        fn hexchar2int(ch: char) -> u8 {
            if '0' <= ch && ch <= '9' {
                ch as u8 - '0' as u8
            } else {
                0xa + 
                if 'a' <= ch && ch <= 'f' {
                    ch as u8 - 'a' as u8
                } else if 'A' <= ch && ch <= 'F' {
                    ch as u8 - 'A' as u8
                } else {
        hexchar2int(c0) * 0x10 + hexchar2int(c1)            

But it does make me happy. There are several problems that I may think of:

  1. It needs to create two sub-vectors, one for odd-index element, one for even-index
  2. Because of performance, I wrote the function hexchar2int (which is ugly) to convert two hex number into a u8 value instead of using u8::from_str_radix

I wonder if there is a better method to do that.

  • 1
    \$\begingroup\$ because of performance — you benchmarked to see that this was a bottleneck? \$\endgroup\$ – Shepmaster Aug 14 '18 at 23:15
  • \$\begingroup\$ @Shepmaster: I didn't, I just thought that u8::from_str_radix is overkill because in this case I need to create a string of two characters. \$\endgroup\$ – Ta Thanh Dinh Aug 14 '18 at 23:25
  1. Note that this code will not work on any non-ASCII strings. These are more and more common, especially considering the global community we are a part of (and don't forget emoji 🙃).

  2. Use Vec<_> to avoid redundantly specifying the inner type when collecting.

  3. Use byte literals b'x' instead of casting characters to bytes.

  4. You can match on ranges of characters. I find this aesthetically pleasing.

  5. Don't collect an iterator into a Vec just to call into_iter on the Vec. Instead, use the original iterator with partition.

  6. Don't convert from u8 to char and back. You only need to make a single transformation.

  7. There's no need to use the turbofish with the final collect since it's being returned and the type can be inferred.

  8. There's no need to call into_iter for zip's argument, it's implied because it takes an IntoIterator.

  9. Instead of partitioning into vectors, just grab two values out of the iterator at a time. Using fuse allows calling next after it's already returned None.

  10. Your code handles the check of a hex digit twice, leading to the unreachable. Instead, perform the conversion when performing the check.

fn parse_hex(hex_asm: &str) -> Vec<u8> {
    let mut hex_bytes = hex_asm.as_bytes().iter().filter_map(|b| {
        match b {
            b'0'...b'9' => Some(b - b'0'),
            b'a'...b'f' => Some(b - b'a' + 10),
            b'A'...b'F' => Some(b - b'A' + 10),
            _ => None,

    let mut bytes = Vec::new();
    while let (Some(h), Some(l)) = (hex_bytes.next(), hex_bytes.next()) {
        bytes.push(h << 4 | l)
| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.