Markdown code block injection can leak to the surrounding context

[kirawi]

Summary

If the code block's closing backticks have trailing space, the injected syntax highlighting will apply to the text outside of the code block.

Reproduction Steps

Open a markdown buffer with the contents:

'''c
''' 
int i = 0;

where the quotes are replaced with backticks.

I expected this to happen:

int i = 0; should be highlighted as markdown

Instead, this happened:

It is highlighted as C

Helix log

No response

Platform

Linux/macOS

Terminal Emulator

Kitty 0.30.1

Installation Method

Source

Helix Version

990378a

[the-mikedavis]

It seems like this is because the markdown grammar only treats ``` followed by newlines as a terminator for a codefence. I'm not sure if that's intentional though

[pascalkuthe]

yeah I am pretty sure that is the reason. That may genuinely be the way markdown parsers behave. You can report that upstream to figure out if its intentional or not but we can't change this downstream

[kirawi]

According to the spec it references:

The closing code fence may be indented up to three spaces, and may be followed only by spaces, which are ignored.

I'll report it upstream as a bug.

[haitrungle]

I think this is fixed in the latest version?

[the-mikedavis]

Looking at tree-sitter-grammars/tree-sitter-markdown#135 it looks like this should be fixed upstream now. We'll need to update

helix/languages.toml

Lines 1787 to 1789 in 2ee11a0

	[[grammar]]
	name = "markdown"
	source = { git = "https://github.com/tree-sitter-grammars/tree-sitter-markdown", rev = "62516e8c78380e3b51d5b55727995d2c511436d8", subpath = "tree-sitter-markdown" }

and markdown_inline

[the-mikedavis]

Fixed in #14814