Using [url=...] breaks get_character_paragraph and get_character_line

[nerdbeard]

Godot version

4.0.beta2

System information

Linux

Issue description

RichTextLabel fails to account for the characters taken up by the URL (I think) when [url=...] tags are used. This causes get_character_paragraph and get_character_line to return incorrect values when several long URLs are used.

Steps to reproduce

Please see https://github.com/nerdbeard/RichTextLabel-offset/blob/main/anchor.gd for a minimal example.

Minimal reproduction project

https://github.com/nerdbeard/RichTextLabel-offset

[nerdbeard]

A comment in RichTextLabel::get_total_character_count() says:

// Note: Do not use line buffer "char_count", it includes only visible characters.

This appears to be exactly what get_character_line() and get_character_paragraph() do, they use char_count and char_offset, which only include visible characters. I think they need to be rewritten to use item->text.length() to calculate their own offsets.

I don't understand whether this would have other consequences but I'm going to try and code it and see what happens.

[nerdbeard]

Maybe I need to teach RichTextLabel::append_text() to push some kind of null to represent the invisible characters

[bruvzg]

This is how it is supposed to work. get_character_line() and get_character_paragraph() count only text characters, tags are not counted (this is true for all tags, not just url).

[bruvzg]

I guess getting line/paragraph for the position in the BBCode source is useful to have, so we should add an option to do so (an extra argument for the get_character_* or a new set or functions).

[nerdbeard]

Thanks, @bruvzg! Is how it's supposed to work documented anywhere?

My use case is that I want to scroll to the position of a string in the text. I only know the offset into the text I submitted to the control (which likely has BBCode in it) and I don't see that I have access to the offsets into the rendered text. I'm not sure how scroll_to_line and scroll_to_paragraph are meant to be used in conjunction with the source text, but I think I'm doing things the most obvious way.

The Line private struct of RichTextLabel has members char_offset and char_count. Could I add char_raw_offset and char_raw_count?

[nerdbeard]

Calculating the required address in GDScript was a lot easier than changing the engine for me obviously but if there's any desire for this feature I'd be interested in working on it.