Turtle Language Server

This is a LSP server implementation for RDF graphs serialized as Turtle.

Install with: pip install turtle_language_server.

If you wish to modify the code instead of just running it, see the "Development" section below.

Commands and Features

• Use the loadGraphs command to pull in the graph definitions listed as prefixes in your Turtle file:

  # caches graphs in-memory for NeoVim session
  :CocCommand loadGraphs
  

  # overrides cache and pulls graphs anyway
  :CocCommand loadGraphs force
  

• syntax checking your Turtle file (including undefined namespaces)

• auto-complete when adding statements to the file (relies on loadGraphs)

TODOs:

• verify shapes and add validation information to UI

Setup With NeoVim and CoC

If you are using coc.nvim, you can configure the language server as follows:

1. Make sure Vim correctly detects turtle files and sets the filetype. One way to achieve this is by adding the following line to your .vimrc or init.nvim:

   au BufRead,BufNewFile *.ttl set filetype=turtle

2. Modify your CoC settings to use the turtle_language_server when you open a Turtle file.

   1. First, run :CocConfig or edit coc-settings.json
   2. Add the following (merge with existing keys in "languageserver" if needed):

      {
        "languageserver": {
          "turtle": {
            "command": "turtle_langserver",
            "filetypes": ["ttl", "turtle"]
          }
        }
      }

Development

First, clone the git repository:

git clone https://github.com/Spyderisk/langserver
cd langserver

You need the poetry python package management tool, which you need to install using your operating systenm's native package manager. For example, apt install python3-poetry or pacman -S poetry.

Then, in the new langserver directory:

poetry install

This should install all of the python dependencies listed in the pyproject.toml file. Poetry installs to a standard Python venv virtual environment.

Next decide on a port you wish the LSP server to listen on. Since this is an experimental service it should never be exposed on the internet, so the only interface to use with it is localhost. You will need to choose an unnused port, which you can verify using nmap localhost. In the following we have chosen port 7213 for no particular reason. You should now be able to start the LSP server like this:

$ python turtle_language_server/entry.py --tcp localhost:7213
DEBUG:asyncio:Using selector: EpollSelector
INFO:pygls.feature_manager:Registered builtin feature exit
INFO:pygls.feature_manager:Registered builtin feature initialize
INFO:pygls.feature_manager:Registered builtin feature initialized
INFO:pygls.feature_manager:Registered builtin feature shutdown
INFO:pygls.feature_manager:Registered builtin feature textDocument/didChange
INFO:pygls.feature_manager:Registered builtin feature textDocument/didClose
INFO:pygls.feature_manager:Registered builtin feature textDocument/didOpen
INFO:pygls.feature_manager:Registered builtin feature workspace/didChangeWorkspaceFolders
INFO:pygls.feature_manager:Registered builtin feature workspace/executeCommand
INFO:pygls.feature_manager:Command loadGraphs is successfully registered.
INFO:pygls.feature_manager:Registered textDocument/completion with options {'trigger_characters': [':']}
INFO:pygls.feature_manager:Command countDownBlocking is successfully registered.
INFO:pygls.feature_manager:Command countDownNonBlocking is successfully registered.
INFO:pygls.feature_manager:Registered textDocument/didChange with options {}
INFO:pygls.feature_manager:Registered textDocument/didClose with options {}
INFO:pygls.feature_manager:Registered textDocument/didOpen with options {}
INFO:pygls.feature_manager:Registered textDocument/didSave with options {}
INFO:pygls.feature_manager:Command registerCompletions is successfully registered.
INFO:pygls.server:Starting server on localhost:1113

Now you can start testing, and modifying server.py.