Highlight update

[hugopl]

This updates the highlight queries with new recent features.

It also add a injections.scm, currently just for heredoc and comments, because yes... there's a comments tree-sitter parser, hehehe. I knew about this when I tried to submit the queries to nvim-tree-sitter.

heredoc injections works nicely if you do SQL queries in heredoc strings, as you can see on github itself:

sql = <<-SQL
select foo from bar;
SQL

sql = <<-NOTSQL
select foo from bar;
NOTSQL

The same can be done for regexes when #49 gets fixed.

[hugopl]

my nvim is now crystal colorful ❤️ , despite of the errors in macros 😅

If my time permits I'll try to write other queries for nvim like indent, fold, etc... and submit the patch to nvim-tree-sitter.

[devnote-dev]

Do these changes also highlight calls without parentheses? It has been a frequent request recently.

[keidax]

Why remove this one? Is it conflicting with something?

[nobodywasishere]

For helix at least, that would match every identifier as a variable, regardless of the rest of the file due to how it parses in comparison to neovim

[keidax]

I guess we'd want more sophisticated matching of local variables anyways: https://tree-sitter.github.io/tree-sitter/syntax-highlighting#local-variables

[nobodywasishere]

I have the start of a locals.scm above that uses those actually

[hugopl]

Yep, I noticed it was matching too much due to being too generic.

So I removed in a attempt to avoid multiple matches for the same node.

[hugopl]

Helix uses @keyword.control.if while nvim uses @keyword.conditional... I should had put the list of captures of each editor on txt files and diff them instead of being lazy and try to find the differences by eye 😅

Ok to have the directory structure?

└── queries
    ├── helix
    ├── nvim
    └── zed

[nobodywasishere]

@hugopl that works for me!

[hugopl]

PR Updated, I didn't added the indent queries because I wasn't able to make it work on nvim yet.

But folding is working, still need some adjustments left for future PRs. e.g. some nodes must be created on if (and rescue IIRC), so folding an if wont fold entire if/else/end block, but just the if.

[nobodywasishere]

some nodes must be created on if (and rescue IIRC)

Can you create an issue for what's needed to accomplish this?

[hugopl]

some nodes must be created on if (and rescue IIRC)

Can you create an issue for what's needed to accomplish this?

Nevermind... the behavior I was expecting seems not to be the behavior I see on other grammars.

On:

if foo
  hey
else
  ho
end

when folding the if, I was expecting to have:

if foo...................
else
  ho
end

but what I get is what other languages seem to also do, fold the entire if/else/end.

if foo...................

So not an issue.

since it's approved, could someone merge it, plz? 😁

[nobodywasishere]

See comments

[hugopl]

Ooops, sorry for not seeing the comments. I always got confused by the github interface and keep missing things in the communication.