[bug] Tokenizer line/char is not always accurate

[Vurv78]

Describe the bug
Sometimes the line / char is inaccurate when parsing something with the tokenizer. This is very annoying with error traces.

Note this is a shortcoming from using lua patterns as the tokenizer, which gives less control as to what happens for more performance. There is probably an elegant solution to this that maintains the pattern matching tokenizer, but it is definitely a lot more complex than if it looped over all chars.

To Reproduce
Nothing right now

Expected behaviour
Get proper line/char of the token.