Move jobparser from act repository to Gitea

[lunny]

The jobparser sub package in act is only used by Gitea. Move it to Gitea to make it more easier to maintain.

[silverwind]

Good call moving this here imho.

[silverwind]

CI failures are related.

[ChristopherHX]

Looks like a regression in this code snipped, is this due to me switching to yaml.v4...

gitea/services/actions/workflow.go

Lines 136 to 149 in 3716ed8

	singleWorkflow := &jobparser.SingleWorkflow{}
	if err := yaml.Unmarshal(content, singleWorkflow); err != nil {
	return fmt.Errorf("failed to unmarshal workflow content: %w", err)
	}
	// get inputs from post
	workflow := &model.Workflow{
	RawOn: singleWorkflow.RawOn,
	}
	inputsWithDefaults := make(map[string]any)
	if workflowDispatch := workflow.WorkflowDispatchConfig(); workflowDispatch != nil {
	if err = processInputs(workflowDispatch, inputsWithDefaults); err != nil {
	return err
	}
	}

Needs investigation, in worst case revert to yaml v3.

[wxiaoguang]

The jobparser sub package in act is only used by Gitea. Move it to Gitea to make it more easier to maintain.

If "jobparser" is only used in Gitea, should upstream delete this package first?

[TheFox0x7]

If "jobparser" is only used in Gitea, should upstream delete this package first?

https://gitea.com/gitea/act/pulls/155

[wxiaoguang]

And one more question: should we remove this "replace"? Just use "gitea.com/gitea/act" directly

replace github.com/nektos/act => gitea.com/gitea/act v0.261.8

[silverwind]

Yes please remove that replace directive. It just results in readers being confused which module is actually used. replace directives should all be eliminated imho.

[ChristopherHX]

And one more question: should we remove this "replace"? Just use "gitea.com/gitea/act" directly

replace github.com/nektos/act => gitea.com/gitea/act v0.261.8

Then we would need a PR to gitea/act first to replace this in all modules, yes this makes sense for me as well.

This does not 100% look to be in scope of this PR.

(At least I think that we are forced to (one exception is the fork is of an executable, then go does not care), please correct me if I am wrong)

[ChristopherHX]

I would have preserved the original authors as coauthors, other than that this looks good now.

The ci failure

• was unrelated to the yaml v4 version in go.mod
• the problem was the model used yaml.Node v4 and the yaml.Unmarshal v3 (that left the yaml.Node v4 fields blank)

[lunny]

I would have preserved the original authors as coauthors, other than that this looks good now.

The ci failure

• was unrelated to the yaml v4 version in go.mod
• the problem was the model used yaml.Node v4 and the yaml.Unmarshal v3 (that left the yaml.Node v4 fields blank)

This package was newly introduced by the Gitea maintainers specifically for Gitea. Therefore, the copyright holder “Gitea Authors” is appropriate.

Since the original files did not include a copyright year, I have added “2026” consistently to all of these files.

[ChristopherHX]

Yes copyright is fine, I know