Contextful validity errors, null-byte specs fix

[diamondburned]

This pull request adds meaningful errors while decoding as well as proper null-byte encoding.

Benchmarks

New is the pull request and old is upstream/main. Test performed on a laptop with an Intel Core i5-8250U.

name          old time/op    new time/op    delta
Encode-8         269ns ±14%     269ns ±21%     ~     (p=1.000 n=8+8)
EncodeTo-8      1.78µs ±13%    1.70µs ± 9%     ~     (p=0.266 n=8+8)
Decode-8        2.74µs ±12%    3.40µs ± 6%  +24.26%  (p=0.000 n=8+8) <-- more is slower
DecodeFrom-8    3.38µs ±16%    4.69µs ± 7%  +38.76%  (p=0.000 n=8+8) <-- more is slower

name          old speed      new speed      delta
Encode-8      44.8MB/s ±15%  45.2MB/s ±19%     ~     (p=1.000 n=8+8)
EncodeTo-8     185MB/s ±14%   193MB/s ± 9%     ~     (p=0.279 n=8+8)
Decode-8       120MB/s ±11%    96MB/s ± 6%  -19.68%  (p=0.000 n=8+8) <-- less is slower
DecodeFrom-8  97.5MB/s ±14%  69.9MB/s ± 7%  -28.33%  (p=0.000 n=8+8) <-- less is slower

The performance loss in DecodeFrom is probably scanner.Text() having to allocate a new string for each chunk. The speed is still fairly decent, though.

[codecov-io]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[nihaals]

Based on GitHub Actions benchmarks, BenchmarkDecodeFrom is the main benchmark affected, the rest looks like random variance.

It would be nice to have some tests for invalid inputs, maybe even checking the position (although that might be overkill).

[diamondburned]

maybe even checking the position

type InvalidRuneError rune can be feasibly changed to something else to add this, for example

type InvalidRuneError struct {
    Rune     rune
    Position int
}

[nihaals]

type InvalidRuneError rune can be feasibly changed to something else to add this, for example

type InvalidRuneError struct {
    Rune     rune
    Position int
}

I think this change would almost exclusively be used for tests, so I'm not too sure. I was originally thinking parsing the error string for the reported position and checking if it matches what is expected (hence me calling it overkill).

[nihaals]

I think we should add the position to InvalidRuneError and add tests for invalid inputs, also checking the position to ensure it matches.

[diamondburned]

add tests for invalid inputs

Do you have a set of reference invalid inputs?

[nihaals]

add tests for invalid inputs

Do you have a set of reference invalid inputs?

A few examples:

1. Decode: 💖✨✨✨,,,,👉👈test💖💖,👉👈❤️👉👈💖💖✨🥺👉👈💖💖✨🥺,👉👈 (non-bottom characters)
2. Decode: 💖✨✨✨,,,,👉test👈💖💖,👉👈❤️👉👈💖💖✨🥺👉👈💖💖✨🥺,👉👈 (non-bottom characters)
3. Decode: 💖✨✨✨,,,,👉💖👈💖💖,👉👈❤️👉👈💖💖✨🥺👉👈💖💖✨🥺,👉👈 (character in byte separator)
4. Decode: 💖✨✨✨,,,,👉test👈💖💖,👉👈❤️👉👈💖💖✨🥺👉👈💖💖✨🥺,👉👈 (character in byte separator)
5. Decode: 💖✨✨✨,,,,👉👈,💖💖👉👈❤️👉👈💖💖✨🥺👉👈💖💖✨🥺,👉👈 (characters not in order)
6. Decode: 💖✨✨✨,,,,👉👈💖💖,👉👈❤️💖💖✨🥺👉👈💖💖✨🥺,👉👈 (missing byte separator after null byte)