Russian spelllang dictionary

[0rtz]

Hello, thank you for the plugin. Can you please hint me on how to set up suggestions from russian dictionary? I have the following init.vim:

let s:plug_dir = expand('/tmp/plugged/vim-plug')
if !filereadable(s:plug_dir . '/autoload/plug.vim')
  execute printf('!curl -fLo %s/autoload/plug.vim --create-dirs https://raw.githubusercontent.com/junegunn/vim-plug/master/plug.vim', s:plug_dir)
end

call plug#begin(s:plug_dir)
Plug 'hrsh7th/nvim-cmp'
Plug 'uga-rosa/cmp-dictionary'
call plug#end()

if len(filter(values(g:plugs), '!isdirectory(v:val.dir)'))
	echom 'Some modules are missing, running :PlugInstall'
	PlugInstall --sync
endif

lua <<EOF

local cmp = require('cmp')

require("cmp_dictionary").setup({
	dic = {
		spelllang = {
			en = "/path/to/en.dict",
			ru = "/path/to/ru.dict",
		},
	},
})

cmp.setup{
  mapping = {
		["<C-j>"] = cmp.mapping(function(fallback)
			if cmp.visible() then
				cmp.select_next_item()
			end
		end, { "i", "s" }),
		["<C-k>"] = cmp.mapping(function(fallback)
			if cmp.visible() then
				cmp.select_prev_item()
			end
		end, { "i", "s" }),
		['<C-l>'] = cmp.mapping({
			i = cmp.mapping.confirm({ select = false }),
		}),
	},

	sources = {
		{
			name = "dictionary",
			keyword_length = 2,
		},
	}
}

EOF

set keymap=russian-jcuken
set spelllang=ru

But I do not get any completions as I type in russian:

Executing CmpDictionaryUpdate after entering buffer also does not help.

While english seems to work fine:

I use aspell-en and aspell-ru packages on archlinux and the following command to get dictionaries:

aspell -d en dump master | aspell -l en expand > en.dict
aspell -d ru dump master | aspell -l ru expand > ru.dict 

[uga-rosa]

I think the problem is related to multibyte characters. Please wait while I investigate.

[uga-rosa]

Fixed in 94cadf9.
Please update the latest version and test it.

[0rtz]

Yeah, I see it works now, thank you.
Tho, completion is really slow, can I speed up it somehow? I see that russian dictionary in aspell after expansion is really huge, is it possible to truncate it?

[uga-rosa]

The candidate acquisition itself is done by bisection search, so I don't think it will be that slow. I think that a large number of candidates are passed to the main body of cmp, and the subsequent sorting and display process may be taking time. I will try to create a mechanism to limit the number of candidates returned to the main body.

[uga-rosa]

Can you try #32?

[0rtz]

Well, it doesn't lag, but results of completion seems odd

[uga-rosa]

That's right. If we return all the candidates, it would be too laggy, so to avoid that, we reduce the number of candidates to be returned.
You can increase the number of max_items in setup, and/or increase the number of exact, or set it to -1 (the candidates returned by this plugin to the main body are those looked for by prefix exact match for the number of exact only, and fuzzy match is the job of the main body).

require("cmp_dictionary").setup({
    --- other settings
    max_items = 1000, -- Default: 100
    exact = -1, -- Prefix exact match
})

[uga-rosa]

Please try and let me know how much these should be set to just the right numbers, as I would like to determine the appropriate default settings. Is it less at 100?

[uga-rosa]

PR for max_items has been merged. The default value is set to -1 (unlimited).
Please make good use of options max_items and exact.

[0rtz]

Thank you

max_items = 2000,
exact = -1,

works decent for Russian language