dracor-org/dracor-api

Harmonize handling of missing information in responses

Closed this issue · 3 comments

ingoboerner commented

The information that a value is not available is not consistent across the API JSON responses and even not with a single endpoint, e.g.

  • If a digital source URL "sourceUrl" is not available, the endpoint /corpora/{corpusname} returns an empty string.
  • If there is no information on a year (written,printed,premiered,normalized), null is returned.

I am working on marshmallow schemas to validate the response against. In case of the field "sourceUrl" I expected the return value to be a valid URL. This I can check within the schema:

sourceUrl = fields.Url(allow_none=True)

I set allow_none to True because I figured, that if there are no dates, the value of the respective field is null, so this works:

yearWritten = fields.Str(allow_none=True)

I expected, that if no URL is available, it would also return Null/None, but

sourceUrl = fields.Url(allow_none=True)

throws an error:

ValidationError: {'dramas': {1: {'sourceUrl': ['Not a valid URL.']}

In CalDraCor, for example, you find:

"wikidataId": null,
"yearPremiered": null,
 "source": "",
 "yearWritten": "1681",
 "id": "cal000152",
 "title": "La divina Filotea",
 "sourceUrl": "",
cmil commented

In 2cb885e the string properties source and sourceUrl have been replaced by and object source with the properties name and url. If a play has only partial or no source information at all the respective properties are currently omitted.

@ingoboerner Does this solution work with the marshmallow schemas? If so, I'd suggest to close this issue. If there are other properties showing the same problem, let's open a new issue explicitly naming the properties in question.

cmil commented

Closing this. Please open a new issue if there is a need for or plan of how to further harmonise this.