Function bytes_to_int

crates/common/src/int.rs:41–145

https://github.com/python/cpython/blob/4e665351082c50018fb31d80db25b4693057393e/Objects/longobject.c#L2977 https://github.com/python/cpython/blob/4e665351082c50018fb31d80db25b4693057393e/Objects/longobject.c#L2884

(
    buf: &[u8],
    mut base: u32,
    digit_limit: usize,
)

// https://github.com/python/cpython/blob/4e665351082c50018fb31d80db25b4693057393e/Objects/longobject.c#L2977
// https://github.com/python/cpython/blob/4e665351082c50018fb31d80db25b4693057393e/Objects/longobject.c#L2884
pub fn bytes_to_int(
    buf: &[u8],
    mut base: u32,
    digit_limit: usize,
) -> Result<BigInt, BytesToIntError> {
    if base != 0 && !(2..=36).contains(&base) {
        return Err(BytesToIntError::InvalidBase);
    }

    let mut buf = buf.trim_ascii();

    // split sign
    let sign = match buf.first() {
        Some(b'+') => Some(Sign::Plus),
        Some(b'-') => Some(Sign::Minus),
        None => return Err(BytesToIntError::InvalidLiteral { base }),
        _ => None,
    };

    if sign.is_some() {
        buf = &buf[1..];
    }

    let mut error_if_nonzero = false;
    if base == 0 {
        match (buf.first(), buf.get(1)) {
            (Some(v), _) if *v != b'0' => base = 10,
            (_, Some(b'x' | b'X')) => base = 16,
            (_, Some(b'o' | b'O')) => base = 8,
            (_, Some(b'b' | b'B')) => base = 2,
            (_, _) => {
                // "old" (C-style) octal literal, now invalid. it might still be zero though
                base = 10;
                error_if_nonzero = true;
            }
        }
    }

    if error_if_nonzero {
        if let [_first, others @ .., last] = buf {
            let is_zero = *last == b'0' && others.iter().all(|&c| c == b'0' || c == b'_');
            if !is_zero {
                return Err(BytesToIntError::InvalidLiteral { base });
            }
        }
        return Ok(BigInt::zero());
    }

    if buf.first().is_some_and(|&v| v == b'0')
        && buf.get(1).is_some_and(|&v| {
            (base == 16 && (v == b'x' || v == b'X'))
                || (base == 8 && (v == b'o' || v == b'O'))
                || (base == 2 && (v == b'b' || v == b'B'))
        })
    {
        buf = &buf[2..];

        // One underscore allowed here