This repo contains my custom treesitter queries for helix :3
Helix will first read its “preset” runtime, which is stored relatively to where your helix binary is. Here's a visualizer:
hx
runtime/
grammars/
themes/
queries/
ada/
folds.scm
highlights.scm
injections.scm
locals.scm
textobjects.scm
... other languages ...
zig/
highlights.scm
indents.scm
injections.scm
locals.scm
rainbows.scm
tags.scm
textobjects.scm
tutor
If you compile helix from source, the best approach to get this going is to symlink the runtime/ directory from where your helix binary is stored, to the runtime/ directory in your locally cloned helix repo.
My helix binary is ~/fes/eva/hx, and I store my local clone in ~/fes/ork/hx, so I use the below ↓
rm -fr ~/fes/eva/runtime
ln -sf ~/fes/ork/hx/runtime ~/fes/evaNow that you have this set up, you can override particular files in your config. ~/.config/helix/ repeats the similar structure:
~/.config/helix/runtime/
grammars/
themes/
queries/
fish/
highlights.scm
tutor
So! This repo is the queries/ in the above visualization. I have this repo stored locally as ~/fes/ork/qiri, so:
mkdir -p ~/.config/helix/runtime
ln -sf ~/fes/ork/qiri/ ~/.config/helix/runtime/queriesRealistically you won't want to use all of my queries wholesale; you'll probably want to pick particular files and yank them.
Put them in ~/.config/helix/runtime/, not forgetting that each language needs to be stored in its own subdirectory.
Most of my queries can be used as is, but for some languages I use a custom parser. If something is not working, check what I'm using in my config.
Don't. This is mine. Feel totally free to take my work as a base and start your own repo though 😌