Unable to register custom command

[gsabran]

I'm trying to register a custom command, following the documentation as well as I can. But when the server receives this command it fails. What should I do differently?

import logging
from pygls.server import LanguageServer as ls

logging.basicConfig(filename='pygls.log', filemode='w', level=logging.DEBUG)

server = ls('example-server', 'v0.1')

@server.command('saySomething')
def saySomething(ls, *args):
    """Returns completion items."""
    return "hello"

server.start_io()

The error log is:

ERROR:pygls.protocol.json_rpc:Failed to handle request 3 saySomething Object(lastToken=None, params=None)
Traceback (most recent call last):
  File ".../pygls/protocol/json_rpc.py", line 218, in _get_handler
    return self.fm.builtin_features[feature_name]
           ~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^
KeyError: 'saySomething'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File ".../pygls/protocol/json_rpc.py", line 221, in _get_handler
    return self.fm.features[feature_name]
           ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^
KeyError: 'saySomething'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File ".../pygls/protocol/json_rpc.py", line 260, in _handle_request
    handler = self._get_handler(method_name)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../pygls/protocol/json_rpc.py", line 223, in _get_handler
    raise JsonRpcMethodNotFound.of(feature_name)
pygls.exceptions.JsonRpcMethodNotFound: Method Not Found: saySomething
INFO:pygls.protocol.json_rpc:Sending data: {"error": {"code": -32601, "message": "Method Not Found: saySomething"}, "jsonrpc": "2.0", "id": 3}

Thanks a lot for the support.

[alcarney]

Your server code looks fine, the issue is probably with how the client is interacting with it.

Commands in LSP cannot be called directly, instead the client needs to send a workspace/executeCommand request, specifying the name of the command it wishes to call.

For example

import asyncio
import sys

from lsprotocol import types
from pygls.lsp.client import BaseLanguageClient

async def main():
    client = BaseLanguageClient("client", "v1")
    await client.start_io(sys.executable, "test.py")

    response = await client.initialize_async(
        types.InitializeParams(capabilities=types.ClientCapabilities())
    )
    available_commands = response.capabilities.execute_command_provider.commands
    print(f"Available commands: {available_commands}")

    command_name = available_commands[0]
    print(f"Calling command: {command_name!r}")

    response = await client.workspace_execute_command_async(
        types.ExecuteCommandParams(
            command=command_name, arguments=[]
        )
    )
    print(f"Result: {response!r}")

    await client.shutdown_async(None)
    client.exit(None)

    await client.stop()

asyncio.run(main())

Saving your code to a file called test.py I was able to use the code above to get the following result

$ python client.py
Available commands: ['saySomething']
Calling command: 'saySomething'
Result: 'hello'

Hope that helps! :)

[gsabran]

You are right, thanks a lot!