aca / neuron-language-server

Language server implementation for neuron

neuron-language-server

Language server for neuron.

Neuron will embed language server in neuron. Check sric/neuron#213 for updates.
This is just an personal experiment.

Supports

• textDocument/completion(search by title/id)
  
• textDocument/definition
• textDocument/hover
  
• textDocument/publishDiagnostics
  
  Virtual text is available with nvim-lsp

TODO

• Renaming tag

• improve completion

• LSP snippets

  • header snippets?

• textDocument/codeAction
  https://github.com/felko/neuron-mode has some awesome features that might be converted to code actions.

Prerequisites

• neuron

Installation

go get -u github.com/aca/neuron-language-server

LSP client settings

• vim/neovim, coc.nvim

  "languageserver": {
    "neuron": {
      "command": "neuron-language-server",
      "filetypes": ["markdown"]
    },
  

• neovim, nvim-lspconfig

  local nvim_lsp = require'lspconfig'
  local configs = require'lspconfig/configs'
  
  configs.neuron_ls = {
    default_config = {
      cmd = {'neuron-language-server'};
      filetypes = {'markdown'};
      root_dir = function()
        return vim.loop.cwd()
      end;
      settings = {};
    };
  }
  
  nvim_lsp.neuron_ls.setup{}

neuron-language-server's Issues

IDs containing "-" don‘t get parsed as Zettel IDs

The majority of my zettel IDs contain -, could you make the parser a bit less restrictive?

Doesn't work with latest neuron.

Query result seems to have changed.

Completion by title missing?

README says

textDocument/completion(search by title/id)

However in practice it only completes by ID. When I type in the title, it doesn't do any completion.

Title as virtual text for links not working for new zettel

When I open a fresh neovim (with coc setup) I see the titles of links as virtual text when I open my zettel. For new zettels created, inserted link don't show the title as virtual text.

Zettel titles are displayed as warnings in coc.nvim

I am not sure if this is a bug.

I use nvim with coc.nvim and the number of annotated links counts against the number of warnings in my document.
I think there could be another way seeing that they get displayed as blue annotations, where other tools use a white/grey annotation that does not contribute to warning count. Probably the colors don‘t mean much to you, I just want to say, that I guess that lsp provides different types of annotations and maybe the one you are using might not be the optimal choice. (But this might also be a ui issue of coc.nvim.)

nix-build install error on macOS

macos version 10.15.6
neovim 0.5 nightly

implement completionItem/resolve.

need atomic update of documents

VS Code?

Do you know how this can be used in VS code?

Port to typescript. sourcegraph/go-lsp is no longer maintained.

support wiki-style links

srid/neuron#351

go-to-definition does not work on ?cf links.

Hey and thank you so much for your previous bug fix (the hyphens in ids).
neuron-language-server is now already a great help in my day-to-day activities.

I noticed a small other issue though: When I have a link of the form <id?cf> it correctly annotates the link with the title of zettel id, but go-to-definition does only work, when I remove the ?cf. Otherwise I get a "definition not found" error.

Autocomplete tags

Neuron now supports inline tags, like #foo.

Typing # should autocomplete tags, like [[ already does for zettel titles.