FileType

markdown

Initial buffer

[|]

Input key-sequence

]

Expected buffer

[](|)

Actual buffer

[]|

Information

I'm trying to create a mapping for ] in markdown that inserts a () after the square brackets (for links). This is the configuration I'm using to test this (it also supports Pandoc markdown citations, which are of the form [@citation_key] and do not need parentheses afterwards):

		local insx = require("insx")
		local esc = require("insx.helper.regex").esc

		insx.add(
			"[",
			require("insx.recipe.auto_pair").strings({
				open = "[",
				close = "]",
			})
		)
		insx.add(
			"<BS>",
			require("insx.recipe.delete_pair").strings({
				open_pat = esc("["),
				close_pat = esc("]"),
			})
		)
		insx.add(
			"]",
			require("insx.recipe.jump_next")({
				jump_pat = {
					[[\\\@<!\%#]] .. esc("]") .. [[\zs]],
				},
			})
		)
		insx.add("]", {
			action = function(ctx)
				ctx.move(ctx.row(), ctx.col() + 1)
				if not ctx.data.citation then
					ctx.send("()")
					ctx.move(ctx.row(), ctx.col() - 1)
				end
			end,
			enabled = function(ctx)
				if not vim.tbl_contains({ "markdown", "text" }, ctx.filetype, {}) then
					return false
				end
				if ctx.match([==[\[@[^]]*\%#[^]]*\]]==]) then
					ctx.data.citation = true
					return true
				end
				return ctx.match([==[\%#[^]]*\]]==])
			end,
		})

The issue is that there's no way to prioritize one rule over another, as far as I can tell. If I remove the general jump_next rule, the one for Markdown works (but only in Markdown).

I'm coming from lexima.vim which prioritizes rules (either explicitly by weighting them or implicitly by doing things like applying filetype matches over general rules), but this doesn't seem to be possible in insx.

I don't think this is a bug, but it seems like there could be a use for having multiple mappings for the same key.

nvim-autopairs can integrate with nvim-cmp by automatically adding parantheses when you confirm a completion on a function entry:
https://github.com/windwp/nvim-autopairs#mapping-cr
It would be nice to have something similar for this plugin.

only this recipe should consider treesitter nodes.

FileType

markdown

Initial buffer

``|

Input key-sequence

`

Expected buffer

```|

Actual buffer

```|`

Information

When entering three consecutive backquotes in markdown, the following behavior is observed:

```|`

It would be appreciated if changes could be made to edit this as below, for writing the markdown code block:

```|

Description

Hi!
Thank you for creating and maintaining this amazing plugin!

I found that when we set up insx for <space> key, the abbrev feature does not work.

Here is a minimal configuration.

local tmp_dir = "/tmp"
local commit_hash = "c1d9efb592787f3f8c6d99db6ed5f51dd9475d4f"

-- fectch insx if not exists {{
local insx_path = tmp_dir .. "/insx"
if not (vim.uv or vim.loop).fs_stat(insx_path) then
	vim.fn.system({
		"git",
		"clone",
		"--filter=blob:none",
		"--single-branch",
		"https://github.com/hrsh7th/nvim-insx.git",
		insx_path,
	})
end

-- checkout commit_hash
vim.fn.system({
	"git",
	"-C",
	insx_path,
	"checkout",
	commit_hash,
})

vim.opt.runtimepath:prepend(insx_path)
-- }}

-- setup insx
require("insx").add("<Space>", {
	enabled = function()
		return false
	end,
	action = function(ctx) end,
})

-- define abbreviations
if (vim.fn.has("nvim-0.10")) ~= 0 then
	vim.keymap.set("ia", "fu", "fuhahahaha")
	vim.keymap.set("ca", "sa", "sasumatu")
else
	vim.cmd([[iabbrev <buffer> fu fuhahahaha]])
	vim.cmd([[cabbrev <buffer> sa sasumatu]])
end

How to reproduce

1. Create some lua file(like init.lua) and yank-paste the above code to the file.
2. Open Neovim with the file described earlier (ex. NVIM_APPNAME=insx_abbr nvim -u init.lua)
3. Go to insert mode, then type fu<space> to the buffer (and you can see that abbr does not work)
4. Go to command mode, then type sa<space> (and you can see that ca works)

I tested on [email protected] and [email protected]+ga41546d4a, and the OS is macOS 14.

Thanks

This feature is very similar to express/koa's middleware.

local delete_pair = {
  action = function(ctx)
    if not ctx.match('(|)') then
      return ctx.next()
    end
    ctx.send('<BS><Del>')
  end
}

local remove_cursor_around_spaces = {
  action = function(ctx)
    local before_space = ctx.before():match('%s+$') or ''
    ctx.send(('<BS>'):rep(#before_space)
    local after_space = ctx.after():match('^%s+') or ''
    ctx.send(('<Del>'):rep(#after_space)
    ctx.next()
  end
}

local delete_pair_with_spaces = insx.compose(remove_cursor_around_spaces, delete_pair)

Description

The following bug could be avoided using insx.recipe.substitute.

Code

local insx = require('insx')
local pair = require('insx.recipe.auto_pair')
insx.add('(', pair({
  open = [[\\(]],
  close = [[\)]]
}))

Expected behaviour

| is a cursor. When I type ( before the backslash, I expect to complete the close bracket \).

\| ==> \(| ==> \(|\)

Actual behaviour

| is a cursor. When I type ( before the backslash, I expect to complete the close bracket \) and \.

\| ==> \(| ==> \\(|\)

')' and '}' are set to fast_wrap, but if I enter {} with console.log(<cursor>foo), the order of the parentheses is reversed.
If you put a space inside the parentheses before executing, the problem does not occur.

local minx = require('minx')
minx.add(')', require('minx.recipe.fast_wrap')({ close = ')' }))
minx.add('}', require('minx.recipe.fast_wrap')({ close = '}' }))

FileType

n/a

Initial buffer

n/a

Input key-sequence

n/a

Expected buffer

n/a

Actual buffer

n/a

Information

I haven't been able to get the .mov currently in the README.md (https://user-images.githubusercontent.com/629908/212733495-f8e5486c-215c-4c01-b53c-b720b9779c3f.mov) to work as of this writing.

I've tried it on:

• Firefox Linux (type error -- by the way, why isn't this a .webm?)
• Chrome Linux (starts to load, then stalls)
• Chrome Android (starts to load, then stalls)
• Firefox Android (starts to load, then stalls)

My guess is there are two snags, one of which is a problem with the data hosted by github for the .mov, and the other is the choice to host the data as a .mov in the first place, when .webm (with an open codec) would be the obvious cross-platform choice in 2023.

Description

Since i updated my plugin, i got :

E5113: Error while calling lua chunk: .../pack/packer/start/nvim-insx/lua/insx/kit/Vim/Keymap.lua:29: attempt to call field 'keycode' (a nil value)
stack traceback:
        .../pack/packer/start/nvim-insx/lua/insx/kit/Vim/Keymap.lua:29: in function 'normalize'
        .../nvim/site/pack/packer/start/nvim-insx/lua/insx/init.lua:297: in function 'add'
        ...pack/packer/start/nvim-insx/lua/insx/preset/standard.lua:16: in function 'setup_insert_mode'
        ...pack/packer/start/nvim-insx/lua/insx/preset/standard.lua:10: in function 'setup'
        /home/dk/.config/nvim//lua/plugins_config.lua:439: in main chunk
        [C]: in function 'require'
        /home/dk/.config/nvim/init.lua:21: in main chunk

Maybe from latest updates on the plugin ?

I've been using this plugin for a long time and it is always been great. But there was one annoying problem that came with the standard preset: the ' single quote got auto-pairing when I was coding in Rust or OCaml. In Rust programming language, 'a is used to denote the lifetime of a reference, and in OCaml, 'a is used to represent a type variable. These two usages are very common, and getting automatic single quote auto pairing means that every time when I was working around with the type system, I have to press single quotes twice and delete the pair, then I can finally got a type, not an 'a' character.

Actually it can be fixed by manually adding filetype to the single quote rule myself, but I don't want to copy and paste all the other default presets just to fix this issue. So if there is a way to inject a condition into the default preset and override a part of it, it can save me from getting a huge but redundant neovim dotfiles.

Using the presets, when typing a ' anywhere automatically adds another '. This is usually great, but inside comments, this is rather annoying, since in comments, I often want to write something like "isn't" or "it's", and always get an unnecessary extra ' there.

I'd suggest either not auto-enclosing ' in comments, or adding some conditionals like for example only adding ' if the previous character is a non-letter.

The < operator is often use individually as a comparison operator. But the nvim-insx add <> as default preset which will force the user to delete extra operator. This also let writing code like std::cout << "Hello World" << std::endl harder when coding in C++. I will prefer letting user to manually add < by their own instead of passing this preset as default.

it would be useful to have auto-pairing in vim's cmdline as well

Description

When inserting empty brackets followed by brackets, dot-repeats will nest the latter by the prior.

config

vim.opt.runtimepath:prepend("path/to/nvim-insx")
require("insx.preset.standard").setup()

steps to reproduce

o[]()<Esc>.

becomes

[]()
[()]

o[](foo.html)<Esc>.

becomes

[](foo.html)
[(foo.html)]

Currently the endwise recipe is very basic. I think it should be smarter.

However, I don't use endwise, so I don't know what kind of functionality is needed.