Optimize Semantic Token Language Server Request

[SeanLedford]

These changes are meant to optimize the request for semantic tokens between the client and the language server. Instead of preparing the module required to parse the textDocument prior to every request, the module is now loaded from the regal bundle.

[anderseknert]

Looks great! Some minor things to fix, then good to merge 👍

[anderseknert]

It's an advanced topic sure, but args aren't necessarily vars! If they aren't they'll be pattern matched for equality in a call. So for example this definition:

f("foo", y) if {
    y > 100
}

Would be the same as:

f(x, y) if {
    x == "foo"
    y > 100
}

Here's one example of where we use this in Regal to describe an "or switch" pretty much.

So while it's not too common, you'll want to ensure the type is var here:

arg_tokens.declaration contains arg if {
    some rule in module.rules
    some arg in rule.head.args
    arg.type == "var"
}

[SeanLedford]

Ah gotcha, addressed! Went ahead and changed the arg_token naming to var_token as well to be more exact. On second thought, we're still checking for function args only here, just checking that they're used as variables. So I'll leave it as is

[anderseknert]

It's just complaining about not having a last newline in the file.

[anderseknert]

Add count(rule.head.args) > 0 below or else you'll be running this on all rules (not just functions)

[SeanLedford]

Ah good idea, added.

[anderseknert]

You'll want to do the same check for type == "var" here.

Also, this will be the same for all vars in the rule, so this should be moved up above the walk

[SeanLedford]

Addressed the location of the function arg name check. I did already have a type check for the arg I was pulling from the array of terms, so just want to verify that's what you meant.

[anderseknert]

Same as above applies here.

[anderseknert]

Thanks!