Add actionable item to PatternMatchExhaustivity diagnostic

[ghostbuster91]

The purpose of this PR is to add an actionable item for non-exhaustive pattern match diagnostic, so that people can auto insert missing cases.

Relates to scalameta/metals-feature-requests#350

[ghostbuster91]

@ckipp01 could you give it a look?

[ckipp01]

Just a note about something we need to figure out. Testing this I now see our e2e with Dotty/sbt/Metals isn't working as expected. I know there was some recent changes in sbt, but the behavior I'm seeing now is that the diagnostic is published with the actions, but then published again without them with reset to true. For example:

[Trace - 00:47:25 pm] Received notification 'build/publishDiagnostics'
Params: {
  "textDocument": {
    "uri": "file:///Users/ckipp/Documents/scala-workspace/actionable-diagnostic/src/main/scala/example/Hello.scala"
  },
  "buildTarget": {
    "uri": "file:/Users/ckipp/Documents/scala-workspace/actionable-diagnostic/#root/Compile"
  },
  "diagnostics": [
    {
      "range": {
        "start": {
          "line": 17,
          "character": 26
        },
        "end": {
          "line": 17,
          "character": 30
        }
      },
      "severity": 2,
      "code": "29",
      "source": "sbt",
      "message": "\u001b[33mmatch\u001b[0m may not be exhaustive.\n\nIt would fail on pattern case: Tree.Leaf(_)\n",
      "relatedInformation": [],
      "dataKind": "scala",
      "data": {
        "actions": [
          {
            "title": "Insert missing cases (1)",
            "edit": {
              "changes": [
                {
                  "range": {
                    "start": {
                      "line": 18,
                      "character": 33
                    },
                    "end": {
                      "line": 18,
                      "character": 33
                    }
                  },
                  "newText": "\n      case Tree.Leaf(_) \u003d\u003e ???"
                }
              ]
            }
          }
        ]
      }
    }
  ],
  "reset": false
}


[Trace - 00:47:25 pm] Received notification 'build/logMessage'
Params: {
  "type": 2,
  "message": "\u001b[33m\u001b[33m-- [E029] Pattern Match Exhaustivity Warning: /Users/ckipp/Documents/scala-workspace/actionable-diagnostic/src/main/scala/example/Hello.scala:18:26 \u001b[0m\u001b[0m\n\u001b[33m18 |\u001b[0m    \u001b[33mdef\u001b[0m \u001b[36mfoo\u001b[0m(\u001b[36mtree\u001b[0m: \u001b[35mTree\u001b[0m) \u003d tree \u001b[33mmatch\u001b[0m {\n\u001b[33m\u001b[33m   |\u001b[0m                          ^^^^\u001b[0m\n\u001b[33m   |\u001b[0m                          \u001b[33mmatch\u001b[0m may not be exhaustive.\n\u001b[33m   |\u001b[0m\n\u001b[33m   |\u001b[0m                          It would fail on pattern case: Tree.Leaf(_)\n\u001b[33m   |\u001b[0m\n\u001b[33m   |\u001b[0m longer explanation available when compiling with `-explain`"
}


[Trace - 00:47:25 pm] Received notification 'build/logMessage'
Params: {
  "type": 2,
  "message": "one warning found"
}


[Trace - 00:47:25 pm] Received notification 'build/logMessage'
Params: {
  "type": 3,
  "message": "done compiling"
}


[Trace - 00:47:25 pm] Received notification 'build/publishDiagnostics'
Params: {
  "textDocument": {
    "uri": "file:///Users/ckipp/Documents/scala-workspace/actionable-diagnostic/src/main/scala/example/Hello.scala"
  },
  "buildTarget": {
    "uri": "file:/Users/ckipp/Documents/scala-workspace/actionable-diagnostic/#root/Compile"
  },
  "diagnostics": [
    {
      "range": {
        "start": {
          "line": 17,
          "character": 26
        },
        "end": {
          "line": 17,
          "character": 30
        }
      },
      "severity": 2,
      "source": "sbt",
      "message": "\u001b[33mmatch\u001b[0m may not be exhaustive.\n\nIt would fail on pattern case: Tree.Leaf(_)\n",
      "relatedInformation": []
    }
  ],
  "reset": true
}

I'm not really sure why this is happening now.

[tgodzik]

I think this was already fixed in sbt/zinc#1225 ?

We had same issues when discussing sbt/zinc#1214

[ckipp01]

I think this was already fixed in sbt/zinc#1225 ?

We had same issues when discussing sbt/zinc#1214

Yea, it's just odd that we didn't hit on this before with the other examples for Scala 3. I'm trying to update https://github.com/lampepfl/dotty/blob/10503b0345550259e9381f4fe59da80fb35b32b9/sbt-bridge/src/dotty/tools/xsbt/CompilerBridgeDriver.java#L139-L140 right now which I also believe might be problematic here because it's not using the new problem2.

[ckipp01]

Alright I hope it's ok, but I just rebase this @ghostbuster91 because I want to test it now that #18322 is merged in.

[ghostbuster91]

Alright I hope it's ok, but I just rebase this @ghostbuster91 because I want to test it now that #18322 is merged in.

Nice, thanks:) Could you also share how you are testing it together with these e2e tests with metals?

[ckipp01]

Nice, thanks:) Could you also share how you are testing it together with these e2e tests with metals?

Sure, I know you're using nvim-metals as well, so the steps should be the same:

1. Do a sbt publishLocal on this repo
2. Create a minimal sbt project using 1.9.3
3. Set the scalaVersion to the version you just published locally
4. Switch your build server to use sbt instead of Bloop
5. Then just go into your main file and drop the code from your test in there. You should see a diagnostic and be able to trigger a code action on it.
6. If you don't you can start debugging by creating a .metals/bsp.trace.json file and look in the publishDiagnostics and see what they look like, or if they are there at all.

O and make sure you're on the latest release of Metals.

Very cool 😄

[ckipp01]

This looks great @ghostbuster91! Thanks for doing this! Just one small nit on the test, let's make sure we also have one with significant whitespace being used just to be safe.

[ghostbuster91]

This looks great @ghostbuster91! Thanks for doing this! Just one small nit on the test, let's make sure we also have one with significant whitespace being used just to be safe.

Thanks for making it possible :)

[ckipp01]

This looks good to me. Thanks a lot @ghostbuster91! LGTM

[ckipp01]

It's not actually letting me merge. @dwijnand I think since you requested changes it's waiting for your approval.