The plugin does not always interpret correctly )} in Lua

[ChrisGVE]

Most of the time, the plugin works great, but I have encountered what is probably a border case where it does not show the context correctly:

This comes from a wezterm configuration file (which is partially supported by the LSP, especially when using plugins).

I'm happy to help you debug it, but I would need some directions.

[andersevenrud]

You can use the command :NvimContextVtDebug to print out some information you can paste here. That might help find a solution :)

https://github.com/andersevenrud/nvim_context_vt?tab=readme-ov-file#debug

[ChrisGVE]

Cool thanks for the quick reply, here is the output

   907:chunk --> ---------
509:table_constructor --> {
509:arguments --> ({
509:function_call --> resurrect.periodic_save({

[andersevenrud]

I was a little bit unsure about what was missing here, but did a quick test and found the following:

In JavaScript

foo.bar({
    a: b,
    c: d,
}) --> foo.bar({

In Lua

foo.bar({
    a = b,
    c = d,
}) --> ({

local foo = foo.bar({
    a = b,
    c = d,
}) --> local foo = foo.bar

---

I created a potential fix for this here: #79

Let me know if it works out :)

[ChrisGVE]

I've applied the fix manually, but unfortunately I still have the same result.

However, I am getting something different now with the debug, to recap:

(showing again as I made some changes, so line # are not the same as above)

and here is the debug output

16:45:02 msg_show   NvimContextVtDebug    517:table_constructor --> {
16:45:02 msg_show   NvimContextVtDebug    517:arguments --> ({
16:45:02 msg_show   NvimContextVtDebug    517:function_call --> resurrect.set_encryption({
16:45:02 msg_show   NvimContextVtDebug 907:chunk --> ---------

Well, maybe not so much change after all... just line #

[ChrisGVE]

Ok I went into Treesitter and here is the tree for the function above:

  (function_call ; [504, 0] - [509, 2]
    name: (dot_index_expression ; [504, 0] - [504, 23]
      table: (identifier) ; [504, 0] - [504, 9]
      field: (identifier)) ; [504, 10] - [504, 23]
    arguments: (arguments ; [504, 23] - [509, 2]
      (table_constructor ; [504, 24] - [509, 1]
        (field ; [505, 1] - [505, 23]
          name: (identifier) ; [505, 1] - [505, 17]
          value: (number)) ; [505, 20] - [505, 23]
        (comment ; [505, 25] - [505, 29]
          content: (comment_content)) ; [505, 27] - [505, 29]
        (field ; [506, 1] - [506, 22]
          name: (identifier) ; [506, 1] - [506, 15]
          value: (true)) ; [506, 18] - [506, 22]
        (field ; [507, 1] - [507, 19]
          name: (identifier) ; [507, 1] - [507, 12]
          value: (true)) ; [507, 15] - [507, 19]
        (field ; [508, 1] - [508, 17]
          name: (identifier) ; [508, 1] - [508, 10]
          value: (true))))) ; [508, 13] - [508, 17]

I hope it helps

[andersevenrud]

Thanks!

Sorry for premature closing of issue. I've just pushed another commit that might fix the issue.

---

My initial PR seemed to work fine on my end on initial testing, but it turned out that the ts structure changed when there's a root return statement (module pattern).

I believe I have added the correct context keyword now.

[ChrisGVE]

Cool, indeed it is fixed with the last PR. Thanks a lot for the swift action!

[ChrisGVE]

Fixed

[andersevenrud]

Awesome :)