Force the use of @id in place of identifier for GeoNetwork SO

[iannesbitt]

We've identified some GeoNetwork repositories that incorrectly use @id instead of identifier to define the dataset id. While this is technically incorrect in SO parlance, it is part of some (older?) GeoNetwork versions and thus in use in production environments. To correct for this, the SO normalization process should be able to treat @id as if it were the identifier field.

This can be achieved by setting "use_at_id": true in the SO node settings.json file.

[iannesbitt]

Unfortunately, I had to commit a grave sin in order to make this work. The opersist Thing model wouldn't accept a json document that didn't have an identifier definition. So when the use_at_id settings flag is set, during SO normalization, I just copy the @id to identifier.

I know it's not in the spirit of the network's metadata harvesting practice, but I don't want to have to mess with how the model is structured. I suppose it's "in kind" with the sins that GeoNetwork is committing...

[mbjones]

@datadavev can you review this set of changes please?

[iannesbitt]

I made a mistake and committed directly to the develop branch, let me know if you'd like me to roll those back before you review @datadavev

[datadavev]

This all looks OK. Using @id isn't so terrible, but not preferred.

There is a question of what happens in the future should the geonetwork sources change, though I think these changes are good.

[iannesbitt]

Since this is a repo-specific setting, it can easily be turned off if Geonetwork restructures their schema.org definitions.

[iannesbitt]

• Make sure SHA-256 doesn't change when use_at_id flag is changed