We should get a boolean from the imageinfo API for this. "Includable" or "licensed" or something like that. Certainly OCG shouldn't be trying to parse license fields itself.

Certainly OCG shouldn't be trying to parse license fields itself.

What is generating the "?" then?

There is an [[ https://www.mediawiki.org/wiki/Extension:CommonsMetadata#Returned_data | AttributionRequired ]] field already.

@Nemo_bis: we're just repeating the data provided by the extmetadata property of the action API imageinfo query: https://gerrit.wikimedia.org/r/107562

We can grab AttributionRequired as well, but it would be nicer if extmetadata would also give us a boolean for AttributeValid rather than making us try to match against ?. After all -- is that ? localized? What character is returned in arabic locales?

? does not come from any internal logic in CommonsMetadata - if the author or source is not specified, the relevant key will simply be missing from the response:

• https://commons.wikimedia.org/w/api.php?action=query&titles=File%3ABathsOfCaracalla.jpg&prop=imageinfo&iiprop=extmetadata&format=jsonfm
• https://commons.wikimedia.org/w/api.php?action=query&titles=File%3ADomusAurea.jpg&prop=imageinfo&iiprop=extmetadata&format=jsonfm

The required logic is: AttributionRequired is true(-ish - T86726) but Artist and Attribution are both missing. (This assumes you are using the Attribution field, which did not exist at the time the above patch was made against OCG.) I can put this logic into CMD if you prefer that.

Change 218950 had a related patch set uploaded (by Cscott):
Add "AttributionRequired" metadata to OCG bundle.

https://gerrit.wikimedia.org/r/218950

Change 218950 merged by jenkins-bot:
Add "AttributionRequired" metadata to OCG bundle.

https://gerrit.wikimedia.org/r/218950

Though this task might be do-able once SDC is in production, it's not an SDC task itself.