Print workflow outputs on run completion

[bentsherman]

Spun off from #6574

This PR updates OutputDsl to print the workflow outputs at the end of a run.

By default it prints a human-readable summary. You can set -output-format json to output JSON instead.

It has no effect when there is no output block

When combined with -q, it is useful for printing a comprehensive pipeline output that can be passed to downstream operations

[netlify]

✅ Deploy Preview for nextflow-docs-staging canceled.

Name	Link
🔨 Latest commit	8fc849a
🔍 Latest deploy log	https://app.netlify.com/projects/nextflow-docs-staging/deploys/69c415da423db000085ba920

[pditommaso]

This could be useful in several scenarios. Could not make sense to print it by default in human readable manner and use an output format flag to switch to json, similarly to module run?

[bentsherman]

Yes I would be happy to always print something for the output.

The -o option is already used by the run command to specify the output directory, so we would probably call it -output-format

[bentsherman]

@jorgee if I add an -output-format option to the run command then you could reuse it in module run to control the output format like we were saying

[jorgee]

@jorgee if I add an -output-format option to the run command then you could reuse it in module run to control the output format like we were saying

This will fit perfect for module run.

[bentsherman]

Updated to print human-readable output by default.

This is what it looks like for the output-dsl.nf test:

[bentsherman]

Refining the output:

Outputs:

  /home/bent/projects/nextflow-io/nextflow/tests/checks/output-dsl.nf/results

  samples:
    - {id: sample9, fastqc: log/sample9.fastqc.log, bam: null, bai: null, quant: quant/sample9}
    - {id: sample4, fastqc: log/sample4.fastqc.log, bam: null, bai: null, quant: quant/sample4}
    - {id: sample5, fastqc: log/sample5.fastqc.log, bam: null, bai: null, quant: quant/sample5}
    - {id: sample15, fastqc: log/sample15.fastqc.log, bam: null, bai: null, quant: quant/sample15}
    - {id: sample8, fastqc: log/sample8.fastqc.log, bam: null, bai: null, quant: quant/sample8}
    - {id: sample11, fastqc: log/sample11.fastqc.log, bam: null, bai: null, quant: quant/sample11}
    - {id: sample1, fastqc: log/sample1.fastqc.log, bam: null, bai: null, quant: quant/sample1}
    - {id: sample14, fastqc: log/sample14.fastqc.log, bam: null, bai: null, quant: quant/sample14}
    - {id: sample6, fastqc: log/sample6.fastqc.log, bam: null, bai: null, quant: quant/sample6}
    - {id: sample3, fastqc: log/sample3.fastqc.log, bam: null, bai: null, quant: quant/sample3}
    - ... (90 more items)

  summary:
    - summary_report.html
    - summary_data/data.json
    - summary_data/fastqc.txt

[pditommaso]

it looks nice! is it aligned with modules output? or the other way around?

[bentsherman]

It is now 😄

$ nextflow -q module run test/my-test-module --meta.id 42 --id 42 --text input.txt --reads sample.fastq 
Outputs:

  /home/bent/projects/sketches/module-system-example/results

  txt:
    - {id: '42'}
    - out-42.txt

  id: '42'

$ nextflow -q module run test/my-test-module --meta.id 42 --id 42 --text input.txt --reads sample.fastq -output-format json
{
    "txt": [
        {
            "id": "42"
        },
        "/home/bent/projects/sketches/module-system-example/results/out-42.txt"
    ],
    "id": "42"
}

[pditommaso]

Cool, i'll play hard with modules this w-e

[pditommaso]

Code Review

A few issues worth addressing:

1. replaceAll → replace in OutputDsl.normalizeOutput
replaceAll(outputDir + '/', '') treats outputDir as a regex. Cloud URIs like s3://bucket.name/path will misbehave since . is a regex wildcard. Use replace() (literal string replacement) instead.

2. Default outputFormat='text' always overrides config
   In CmdRun, outputFormat defaults to 'text' (truthy). ConfigBuilder checks if( cmdRun.outputFormat ) — which is always true, so the CLI default always wins even when the user didn't pass -output-format. Other CLI options use null as default to avoid this. Consider defaulting to null in CmdRun and falling back to 'text' at the usage site in OutputDsl.apply().

3. No validation for -output-format values
   Invalid values like -output-format xml silently fall through to text format. Consider validating early with a helpful error message.

4. No unit tests for DumpHelper new methods
   The new normalize(), prettyPrintYaml(), prettyPrintJson() public methods have no unit tests. Would improve confidence in edge cases (null values, nested structures, Path serialization).

5. Typo in Session.groovy
   Line ~153: "Output format for worklfow outputs" → "workflow".