bytes()

The .bytes() method on str returns a std::str::Bytes<'_> iterator that yields each raw UTF-8 byte (u8) of the string by value, while .as_bytes() returns the underlying byte slice &[u8] directly.

Signatures and Traits

pub fn bytes(&self) -> std::str::Bytes<'_>
// where Bytes<'_>: Iterator<Item = u8> + DoubleEndedIterator + ExactSizeIterator + FusedIterator

pub const fn as_bytes(&self) -> &[u8]

Unlike .chars(), Bytes<'_> implements ExactSizeIterator because every item is a single u8 byte and the total byte length s.len() is known in $O(1)$ time.

fn main() {
    let text = "Aé";

    let raw_bytes: Vec<u8> = text.bytes().collect();
    println!("Bytes of \"{text}\" (len={}): {raw_bytes:#04X?}", text.bytes().len());
    println!("as_bytes() slice:          {:#04X?}", text.as_bytes());
}

Bytes of "Aé" (len=3): [
    0x41,
    0xC3,
    0xA9,
]
as_bytes() slice:          [0x41, 0xC3, 0xA9]

Fast ASCII Scanning and Checksums with .bytes() and .as_bytes()

Because UTF-8 guarantees that an ASCII byte (0x00..=0x7F) can never appear as a leading or continuation byte of a multi-byte Unicode character (0x80..=0xFF), searching for ASCII delimiters (b':', b',', b'\n') or computing byte-level checksums via .bytes() or .as_bytes() is 100% UTF-8-safe and avoids UTF-8 code-point decoding overhead:

fn count_ascii_delimiters(csv_line: &str) -> (usize, u32) {
    let commas = csv_line.bytes().filter(|&b| b == b',').count();
    let checksum: u32 = csv_line.bytes().map(u32::from).sum();
    (commas, checksum)
}

fn main() {
    let row = "id,নাম,score";
    let (commas, checksum) = count_ascii_delimiters(row);
    println!("Commas: {commas}, byte checksum: {checksum}");
}

Commas: 2, byte checksum: 2348

.bytes() vs. .as_bytes()