[API Integration] Sketchfab

[annatuma]

Sketchfab has over 250,000 CC-licensed images, many from GLAM institutions. We'd like to prioritize integration with them at this time.

API info:
https://docs.sketchfab.com/data-api/v3/index.html#/

Per our contact, once you have created a user account you're able to apply for an API key. If that isn't the case, we have a contact to quickly connect with for access.

[akshgpt7]

Are only CC-licensed images to be fetched through this API integration?

[mathemancer]

@akshgpt7 That is correct. I should work on this ticket to reflect that. I'll also mark this as in progress, since it's on my plate for this sprint (i.e., I'll be the one working on it).

[mathemancer]

Marking this as 'blocked' until we figure out how to proceed regarding the fact that it doesn't seem possible to systematically pull their entire collection via some sequence of API calls.

[akshgpt7]

@mathemancer I think we can pull the required data using this way:

The https://api.sketchfab.com/v3/search?type=models&license=cc0 endpoint returns the various models. For each model, on calling the endpoint in the uri field, we get the metadata for that particular model, which we can easily extract.
(In the initial endpoint, we can replace cc0 by by, by-sa, by-nd, by-nc, by-nc-sa, by-nc-nd to get different licensed models)
So, we can either take the license URL from the license field in the response, otherwise, we already know it by the endpoint we're calling. For example https://api.sketchfab.com/v3/search?type=models&license=by-sa responses will only contain BY-SA licenses.

[mathemancer]

@akshgpt7 Some context: There's a technical limitation in their API that means it's impossible to obtain more than 10,000 images per license that way. (That's actually the approach we tried initially). To see the script, check out the sketchfab_integration branch in this repo.