[MuhammadAashirAslam]

Add sync_tutorials script and workflow

Updates the tutorials in module.toml by using a temporary directory to shallow-clone precice/tutorials, discovers all tutorial cases and their image folders, and updates the Hugo mounts alphabetically. The workflow then opens an automated PR when changes are detected.

Testing:

• Removing tutorials: MuhammadAashirAslam@554699b , automatic PR Automated: Sync new tutorials from precice/tutorials MuhammadAashirAslam/precice.github.io#14
• Adding non-existent tutorials: MuhammadAashirAslam@c090c78 , automatic PR Automated: Sync new tutorials from precice/tutorials MuhammadAashirAslam/precice.github.io#15
• Without any new tutorials (nothing happens): https://github.com/MuhammadAashirAslam/precice.github.io/actions/runs/36220170196
• Also tested locally on if we add new tutorials in the tutorial repo (local clone) and the dummy mount was added in module.toml (alphabetically)

Reference #979 (comment)

[fsimonis]

For clarification regarding #979 (comment) I think this file is pretty much unmaintainable unless there is a way to generate it. And by this I mean generate entirely.
So you should be able to update the hugo modules, then delete the modules.toml and generate an updated one using a script.
The tutorials repo is only a part of it.

That said, the real issue is that this file is gigantic.
Is it possible to split it up into multiple files? Maybe even one for each external repository?
Then they could be updated individually, and only a script for the complicated tutorials repo would be necessary.

Regarding the technical side, I suggest generating a Python dict with all the info and then using toml.dumps to generate the corresponding toml content.

[MakisH]

I like the idea of one file per repository.

For the tutorials, I see that it is mostly duplication, so that should be easy to generate: each README.md corresponds to an _index.md, and we currently only have one page per tutorial. There is already a rather consistent structure there.

However, I don't see the reason for completely auto-generating the file. There are aspects here that need human input: for example, that tutorials/tests/README.md is mounted at content/docs/dev-docs/system-tests/_index.md. We could get more pages like these, and we should be free to define where they get mounted to.

There are also pages that are inconsistent across repositories, and I would expect that more could appear at times if we integrate more contributed repositories. For example:

• docs/adapter-calculix-overview.md
• docs/adapter-code_aster.md
• docs/overview.md
• docs/adapter-fenics.md

are all meant to be mounted to some _index.md.

To me, such a long file is not an issue, as long as it is clear how to update it. Things could be further automated and made consistent (and I am happy to come up with conventions and apply them), but I would not wait for that before we can merge the rest.

[MuhammadAashirAslam]

I checked the Hugo configuration behavior and documentation, but unfortunately splitting this across multiple files isn't supported natively.

In Hugo's configuration directory, filenames in config/_default/ map directly to root configuration keys (e.g. module.toml maps to [module]). Hugo doesn't append or merge slice/array structures like [[module.imports]] across multiple files—any custom file like tutorials.toml would just be treated as an unrecognized top-level key.

The only workaround would be keeping separate files per repository and writing a custom script to stitch/concatenate them into module.toml during CI/build, but this seems a little bit too hacky and it doesnot seem a right step to me.

[fsimonis]

unfortunately splitting this across multiple files isn't supported natively.

This is really unfortunate.

Then I suggest the following:

• please remove the workflow part of this PR, as it needs to be manual anyways.
• then merge this PR with the tool you propose, so that we have something to work with.