Octal numbers are parsed according to 1.2 spec not 1.1

[firstdorsal]

In yaml 1.1 this is valid octal:

defaultMode: 0400

In yaml 1.2 only this is valid

defaultMode: 0o400

serde-yaml and serde-yaml-ng parse it according to 1.2 spec

This a problem when for example parsing kubernetes manifests because kubernetes uses 1.1

Adding this to parse_unsigned_int should implement the old behavior, allowing both:

fn parse_unsigned_int<T>(
    scalar: &str,
    from_str_radix: fn(&str, radix: u32) -> Result<T, ParseIntError>,
) -> Option<T> {
    let unpositive = scalar.strip_prefix('+').unwrap_or(scalar);
    if let Some(rest) = unpositive.strip_prefix("0x") {
        if rest.starts_with(['+', '-']) {
            return None;
        }
        if let Ok(int) = from_str_radix(rest, 16) {
            return Some(int);
        }
    }
    if let Some(rest) = unpositive.strip_prefix("0o") {
        if rest.starts_with(['+', '-']) {
            return None;
        }
        if let Ok(int) = from_str_radix(rest, 8) {
            return Some(int);
        }
    }
    if let Some(rest) = unpositive.strip_prefix("0b") {
        if rest.starts_with(['+', '-']) {
            return None;
        }
        if let Ok(int) = from_str_radix(rest, 2) {
            return Some(int);
        }
    }



    // THIS
    if unpositive.starts_with("0")
        && unpositive.len() > 1
        && unpositive.chars().all(|c| c.is_digit(8))
    {
        if let Ok(int) = from_str_radix(unpositive, 8) {
            return Some(int);
        }
    }




    if unpositive.starts_with(['+', '-']) {
        return None;
    }
    if digits_but_not_number(scalar) {
        return None;
    }
    from_str_radix(unpositive, 10).ok()
}

[acatton]

Yeah it's unclear to me if the documentation is wrong, or if this is a bug. I think the documentation is wrong.

Following #2, I specified in the README that this crate only supports YAML 1.1.

However, we have plan in #3 to support YAML 1.2.

This is difficult, because this bug comes from unsafe-libyaml which is an old transpiled version of the C library "libyaml" which has many bugs like this. We could upgrade, but I'm in favor of migrating to libyaml-safer (see #5)

In the mean time this issue is a real issue, that I will leave open. But it might get closed with "won't fix, we are now supporting 1.2.", since #5 is the task I want to tackle next, in order to fix #3.

I'm leaving this issue open for documentation purposes.