Unable to round-trip NumericString with APER

[FrancisRussell]

So I originally was attempting to decode a PER NumericString and got weird results, so I wanted to compare with the rasn output for the same value, and encountered an encoding error.

Here is a example which demonstrates the issue:

use rasn::AsnType;
use rasn::prelude::*;

#[derive(Debug, AsnType, rasn::Encode, rasn::Decode)]
#[rasn(delegate, from("0..=9"), size("1..=255"))]
struct SimpleNumericString(pub NumericString);

fn main() {
    let string = SimpleNumericString(NumericString::from_bytes(b"123451234512345").expect("Failed to construct NumericString"));
    println!("Original: {:?}", string);
    let encoded = rasn::aper::encode(&string).expect("Failed to encode");
    println!("Encoded: {:?}", encoded);
    let decoded: SimpleNumericString = rasn::aper::decode(&encoded).expect("Failed to decode");
    println!("Decoded: {:?}", decoded);
}

which crashes with

thread 'main' panicked at /home/fpr/projects/rasn/src/per/enc.rs:280:79:
range end index 15 out of range for slice of length 8

I did some initial investigation at the panic site here. At this point, range is 0..15, which appears to be the related to the length of the original string, but value is Some([18, 52, 81, 35, 69, 18, 52, 5]) which appears to be the packed string data, so the range is invalid to index the slice.

I was using rasn 0.12.1 and rustc 1.76.0-nightly (2b603f95a 2023-11-12) from nightly-x86_64-unknown-linux-gnu.

[Nicceboy]

Values are calculated based on indexes of the valid characters on numeric string, so there is something wrong when calculating the range. I have time later this week to attempt to fix it.