use *file* identifier: produces a tree-sitter error

[CabalCrow]

Using the pattern use file identifier produces a TS error. This happens when the identifier is not put in a list ex: use file [identifier - this produces no error. Also happens when * is used instead of an identifier.

Picture (look at the highlighting of the use - on the line before the incorrectly highlighted use there is an TS error, which can be seen via the TSPlayground):

[fdncred]

Good catch. Thanks!

[CabalCrow]

After testing a bit more I think the issue is not actually in the final argument of the use. The actual issue compes from using a pipeline to get the module name.

For example use std * works properly. However use ('std') * produces a TS error. However it is valid nu syntax.

[fdncred]

This seems like invalid syntax use ('std') *. @kubouch Do you know if this syntax is valid?

[CabalCrow]

The syntax definitely works. The whole point of that syntaxs is to be able to use pipes to provide the module name.

[kubouch]

In theory it should be valid, I think we added subexpressions to const eval. Might be a bug.

[CabalCrow]

After looking through the grammar.js, I think this could be a potential fix. On line 121 it is originally:

          field("module", choice($.val_string, $.unquoted)),

instead this can be used:

          field("module", choice($.val_string, $.expr_parenthesized, $.unquoted)),

From what I saw looking through the grammar expr_parenthesized is used exactly for this type of expression. I tested this via the treesitter cli (both by running test & also by parsing individual files) and it seems to have fixed the issue.

If someone want to give this change a try with their collection of nushell files to see if this creates any ERRORs or incorrect highlighting, give it a go.

[kubouch]

The module field can be any expression that can be evaluated at parse time, I think the only relevant missing is string interpolation, e.g., use $"('s' + 't' + 'd')".

[CabalCrow]

I don't think treesitter can test for expressions that could be evaluated at parse time, but string interpolation can be added in. I will add this to the PR: #90