dap-rs, a Rust implementation of the Debug Adapter Protocol

Introduction

This crate is a Rust implementation of the Debug Adapter Protocol (or DAP for short).

The best way to think of DAP is to compare it to LSP (Language Server Protocol) but for debuggers. The core idea is the same: a protocol that serves as lingua franca for editors and debuggers to talk to each other. This means that an editor that implements DAP can use a debugger that also implements DAP.

In practice, the adapter might be separate from the actual debugger. For example, one could implement an adapter that writes commands to the stdin of a gdb subprocess, then parses the output it receives (this is why it's called an "adapter" - it adapts the debugger to editors that know DAP).

Stability

This crate is in a fairly early stage and breakages will be frequent. Any version before 1.0 might be a breaking version.

Tutorial

For illustration purposes, we are going to recreate the dummy-server example, step by step.

To get started, create a binary project and add dap to your Cargo.toml:

[package]
name = "dummy-server"
version = "*"
edition = "2021"

[dependencies]
dap = "*"

Our dummy server is going to read its input from a text file and write the output to stdout.

To facilitate that, we import the necessary standard types.

Also, we are kinda lazy (err, smart) and we'll use thiserror to create a full-fledge error type.

use std::fs::File;
use std::io::{BufReader, BufWriter};

use thiserror::Error;

dap ships a prelude module and for most applications it's the easiest way to import the necessary types:

use dap::prelude::*;

Let's get the error type out of the way first. We don't plan on handling all commands in our dummy server, so we'll have an error variant that means that.

#[derive(Error, Debug)]
enum MyAdapterError {
  #[error("Unhandled command")]
  UnhandledCommandError,
}

Now we create our Adapter which is going to be the heart of the implementation.

Its accept function will be called for each incoming request, and each return type will be returned to the client in its serialized form.

struct MyAdapter;

impl Adapter for MyAdapter {
  type Error = MyAdapterError;

  fn accept(&mut self, request: Request, _ctx: &mut dyn Context) -> Result<Response, Self::Error> {
    todo!()
  }

...whew. I probably could not explain that to my grandma. So what's with that return type?

Let's see:

• The Result can be used to indicate a success or error in the Adapter itself. Since the Server does not know how to handle your custom errors, returning an error here will mean bubbling that error up through Server::run. In essence, this is an error for your application to handle.
• If you want a user-visible indication of an error, you should return an error response. Users interact with their editor and it's the job of the editor to display such errors.
• If, for any reason you don't want to send a response but you also don't want to return an error, you can use Response::empty(). But do note that clients will normally expect you to send a response, so use this sparingly.

The request argument is the deserialized request and its command field will be one of the requests variants. In practice, this function will likely contain a large match expression or some other means of dispatching the requests to code that can handle them.

The currently unused _ctx parameter could be used to send events and reverse requests to the client. We are not going to utilize that in this tutorial (check out the send_event example).

We'll come back to implementing the accept function after we set up the infrastructure for our server.

First, create an instance of your adapter in main:

let adapter = MyAdapter{};

Then, a client. In this crate, the Client is responsible for sending the responses, events and reverse requests to the actual client that is connected.

let client = BasicClient::new(BufWriter::new(std::io::stdout()));

BasicClient is a builtin implementation that takes a BufWriter where the serialized responses, event and reverse requests are written. It is easy and typical to write to the standard output, but some implementations may want to write to a socket instead.

The Client and Context traits can be implemented to provide different behavior.

Next, we create the Server. The Server ties together the Adapter and the Client. Most importantly, it is the server's responsibility to deserialize the incoming JSON requests, pass them to the Adapter, then take the return value and pass it to the Client (which in turn will serialize it and write it to the client's buffer - in this case, to stdout).

let mut server = Server::new(adapter, client);

Finally, we create a BufReader for the server which serves as the input mechanism and run the server. In this example, we are reading the requests from a file, but in a real life implementation this would be either stdin or a socket.

let f = File::open("testinput.txt")?;
let mut reader = BufReader::new(f);

And finally we run the server. It will run until EOF or an error from the adapter is encountered.

server.run(&mut reader)?;

The AdapterError variant of the error type that run may return will be the error type that you defined above for your Adapter.

Let's take a look at the accept implementation now. To allow separating adapter output from status messages, we will write the latter to stderr.

We will handle two commands in this example.

    match &request.command {
      Command::Initialize(args) => todo!(),
      Command::Next(_) => todo!(),
      _ => Err(MyAdapterError::UnhandledCommandError),
    }
  }

The Next command is one that only requires an ACK response. Keep in mind that it doesn't really make sense to get a Next command from a client out of the blue, so this is purely for examples's sake.

    match &request.command {
      Command::Initialize(args) => todo!(),
      Command::Next(_) => Ok(Response::make_ack(&request).unwrap()),
      _ => Err(MyAdapterError::UnhandledCommandError),
    }
  }

Response, Event and ReverseRequest have helper funtions to create them with certain defaults.

In this case, make_ack borrows the request to be able to copy the seq number from it. This function returns a Result because it checks the request type and only creates an ACK for requests that support it. We will ignore that potential error in this example and just unwrap the result.

Let's implement the Initialize request now. We will make up an error where our adapter absolutely needs the client_name (otherwise optional) field to be set. This is not really a sensible error, we are doing it for demonstration purposes.

We also handle the happy path by returning a Capabilites response with some fields set. A real-life application would like set many more fields. Overall, it looks like this:

    match &request.command {
      Command::Initialize(args) => {
        if let Some(client_name) = args.client_name.as_ref() {
          eprintln!("> Client '{client_name}' requested initialization.");
          Ok(Response::make_success(
            &request,
            ResponseBody::Initialize(Some(types::Capabilities {
              supports_configuration_done_request: Some(true),
              supports_evaluate_for_hovers: Some(true),
              ..Default::default()
            }),
          )))
        } else {
          Ok(Response::make_error(&request, "Missing client name"))
        }
      }
      Command::Next(_) => Ok(Response::make_ack(&request).unwrap()),
      _ => Err(MyAdapterError::UnhandledCommandError),
    }
  }

And that is it. The dummy server is ready to run now.

License

This library is dual-licensed as MIT and Apache 2.0. That means users may choose either of these licenses. In general, these are non-restrictive, non-viral licenses, a.k.a. "do what you want but no guarantees from me".

Commercial support is available on a contract basis (contact me: [email protected]).