Fix incorrect error message on missing comma

[bentsherman]

Close #3185

This PR fixes a particularly nasty error message that occurs with missing commas in process inputs/outputs.

Consider the following example:

process foo {
  debug true

  input:
  tuple val(x) val(y) // missing comma

  script:
  """
  echo "foo ${task.index} ${task.workDir}"
  """
}

workflow {
  foo(['x', 'y'])
}

Currently, the script compiles despite the missing comma but then Nextflow reports that the process has been used twice:

 nextflow run gh-3185/bad-error-message-4.nf
N E X T F L O W  ~  version 23.06.0-edge
Launching `gh-3185/bad-error-message-4.nf` [loving_keller] DSL2 - revision: 63eb6918e1
Process 'foo' has been already used -- If you need to reuse the same component, include it with a different name or include it in a different workflow context

 -- Check script 'gh-3185/bad-error-message-4.nf' at line: 14 or see '.nextflow.log' file for more details

The transpiled Groovy code looks like this:

this._in_tuple(
    new nextflow.script.TokenValCall(new nextflow.script.TokenVar('x'))
).val(new nextflow.script.TokenVar('y'))

So I just add a methodMissing() implementation to BaseParam to catch the incorrect .val() call. With that change, the error now looks like this:

$ ../launch.sh run gh-3185/bad-error-message-4.nf
N E X T F L O W  ~  version 23.06.0-edge
Launching `gh-3185/bad-error-message-4.nf` [nasty_bohr] DSL2 - revision: 63eb6918e1
Invalid method call `val([nextflow.script.TokenVar(y)])` -- possible syntax error

 -- Check script 'gh-3185/bad-error-message-4.nf' at line: 5 or see '.nextflow.log' file for more details

The error message is correctly called a syntax error and it gives the exact line where it happened.

Note however that it doesn't catch all missing comma errors. For example if you forget a comma before emit: ..., the script fails to compile so we can't catch the error.

[netlify]

✅ Deploy Preview for nextflow-docs-staging canceled.

Name	Link
🔨 Latest commit	ad5eda0
🔍 Latest deploy log	https://app.netlify.com/sites/nextflow-docs-staging/deploys/64c53756dcf0d30008be0d56

[bentsherman]

If we remove old DSL1 syntax (#3664 and #3671), this change should also catch things like from and into.

[pditommaso]

What about adding a unit test for capture this behavior

[bentsherman]

I added a unit test for process inputs.

As for outputs, I modified the example to have a tuple output with missing comma:

process foo {
  debug true

  output:
  tuple val(x) val(y) // missing comma

  script:
  """
  echo "foo ${task.index} ${task.workDir}"
  """
}

workflow {
  foo()
}

Which produces this syntax:

this._out_tuple(new nextflow.script.TokenValCall(new nextflow.script.TokenVar('x'))).val(y)

And finally this error:

N E X T F L O W  ~  version 23.07.0-edge
Launching `gh-3185/main.nf` [elegant_stone] DSL2 - revision: 6f14ffa021
ERROR ~ No such variable: y

 -- Check script 'gh-3185/main.nf' at line: 8 or see '.nextflow.log' file for more details

Since outputs are not wrapped in a TokenVar, the Groovy shell encounters a missing variable before it reaches the methodMissing(). This error is already specific enough so I left it alone.