Instant on DocumentReference not properly output on json

[richard-keebler]

• fhir.resources version: v5
• Python version: 3.11
• Operating System: macos

Description

When loading a document reference from a json representation that originally came from a DocumentReference, it fails on validating the date. Below is a contrived example, but the json we're getting (and loading) is from a fhir repository that delivers and accepts json representations.

What I Did

d = DocumentReference(
  id="test",
  status=Code('unknown'),
  content=[DocumentReferenceContent(id="testc", attachment=Attachment(contentType=Code("text/plain"), data="hi"))],
  date=Instant(year=2023, day=8, month=8, hour=8, minute=8, second=8)
)

j = d.json()

item = json.loads(j)

dd = DocumentReference(**item) <-- this throws error invalid datetime format (type=value_error.datetime)

There is a hacky fix here. Just add a Z to the end of the date string representation and the problem is solved, i.e. item['date'] + "Z". I would expect though, that the string representation would already have that.

[janwijbrand]

Ah, you ran into this issue more or less simultaneously with me. Let say +1 to see if we can find a solution to this issue of "instants" that require timezone information in their string representation. What I did to circumvent for the time being is to add timezone info to the dt object that gets serialized later: dt.replace(tzinfo=datetime.timezone.utc)