spywhere/detect-language.nvim

neovim plugin for better language auto-detection, powered by tree-sitter

Hi 👋

My daily drivers

• neovim
• tmux
• zsh with powerlevel10k
• kitty
• And more...

This week I spent my time on

Swift          11 hrs 59 mins  █████████████████████▓░░░   87.11 %
Other          1 hr 10 mins    ██░░░░░░░░░░░░░░░░░░░░░░░   08.52 %
C#             10 mins         ▒░░░░░░░░░░░░░░░░░░░░░░░░   01.24 %
Bash           10 mins         ▒░░░░░░░░░░░░░░░░░░░░░░░░   01.21 %
Markdown       5 mins          ▒░░░░░░░░░░░░░░░░░░░░░░░░   00.71 %

Some statistics

Latest blog posts

• Why you should do an open-source project?
• Wireless Ad-hoc iOS Application Deployment on Dropbox
• Evolution of Atom language
• Keywords: Why we need it? Can’t we live without it?

Organization-based accounts

• Pomelo Fashion: @pmlo-sirisak-l
• MobiLife: @sirisak-lu

detect-language.nvim's Issues

Parsing timeout

Allow each of the language to timed out for better performance and not keep hanging the process.

Small code optimization

https://github.com/spywhere/detect-language.nvim/blob/main/lua/detect-language/init.lua#L73-L79

Instead of reading the whole document as lines and check if it's exceed the threshold, try check it from the last line number.

local last_line = fn.line('$')

if last_line > threshold then
  -- exceed limit
  return
end

local lines = api.nvim_buf_get_lines(0, 0, last_line, false)

doesnt work with lua files

Should instead make an exclusion list of filetypes you want to disable and also make it available in opts={} as well

disabled_filetypes = { "help", "text", "markdown" },

it's mainly because i sometimes have .bak backup files with source code and want to have treesitter enabled in them as well.

and also you'll probably never see a [No Name] file if you use a startpanel plugin like dashboard.nvim

related: #1

Add eager picker support

A picker that evaluate as each language get analyzed. Once the score stopped getting higher, it stop the analysis and return the last language immediately.

Try modular design

So behavior customization can be achieve easier.

For example, picker behavior can be import and customized separately to core detection logic.

Do not trigger for unsupported file types

When file type is set and it is not supported by tree-sitter, do not trigger auto detection. This should also make it ignore startify and nvimtree without any hard coded value as well.

For example, markdown is not supported by tree-sitter. Auto-detect would mess up the file type settings here.

Also only limited to only buftype= (empty)

Use coroutine for async parsing

If possible, try using coroutine instead for parsing.

Allow manually trigger detection

Through a custom command name if possible, so no need to come up with one 😉

Minimum score for sensible provider

Allow a customization of minimum score for sensible provider