Initial Notes on Making the Neopoligen Template LSP

The Notes

• This is for .neoj template files that Neopoligen uses.

• They're the same thing as Jinja files but flip the tokens to something I find more readable. Specifically:

  {{ }} becomes [@ @]

  {% %} becomes [! !]

  {# #} becomes [# #]

• I started by watching a video where someone was using the tower_lsp^tower rust crate. They were basically just typing it in. When I realized that I just went to the page myself.

• First code was just a copy paste from the example on the tower_lsp home page

  I dropped the code in a new Rust project named neoj_lsp and ran cargo install --path . in it to install it on my system.

• I updated my Neovim config (which for me is at ~/.config/nvim/init.lua to add this at the bottom:

  vim.lsp.config['neoj_lsp'] = {
  cmd = { 'neoj_lsp' },
  filetypes = {'html', 'neoj'},
  -- root_markers = { '.git' },
  --
  }
  
  vim.lsp.enable('neoj_lsp')
  
  vim.filetype.add({
    extension = {
      neoj = 'neoj',
    }
  })

  This is straight from the Neovim LSP Quick-Start^qs and filetype^ft docs.

• The quick start also had lines for root_makers and settings like:

  vim.lsp.config['luals'] = {
    -- previous stuff
    root_markers = { { '.luarc.json', '.luarc.jsonc' }, '.git' },
    settings = {
      Lua = {
        runtime = {
          version = 'LuaJIT',
        }
      }
    }
  }

  I didn't need those to get started though. (That uses the luals from the demo instead of my neoj_lsp. Left that as is so the lua stuff in the body of the code makes a little more sense)

• I'm sure there's a more standard place to put that, This got me started until I figure that out.
• The vim.filetype.add() also supports a pattern key that can be used to match the content. Will look at that later so that .html files can use it too.

• That's all it took to get the basic text completion for the word "Hello" to show up in Neovim.

  
• Took a few more hours after that to figure out how to return a formatted version of a document. I've got the pipe working. Now it's on to writing the parser. I already know how to do that, though. It'll take a while, but it's not looking for new knowledge. Less burning of the mental energy required.