Opened 22 months ago
Last modified 21 months ago
#34388 closed New feature
Added support for direct usage of Choices classes on model fields — at Initial Version
Reported by: | T. Franzel | Owned by: | nobody |
---|---|---|---|
Component: | Database layer (models, ORM) | Version: | dev |
Severity: | Normal | Keywords: | |
Cc: | David Wobrock, Adam Johnson | Triage Stage: | Ready for checkin |
Has patch: | yes | Needs documentation: | no |
Needs tests: | no | Patch needs improvement: | no |
Easy pickings: | no | UI/UX: | no |
Description
Hi,
I would like to propose a feature addition on how Choices
are handled when used on model fields. Currently, Field.choices
only accepts iterables. This has 2 shortcommings imho:
- - Rejecting a
Choices
class as argument toField(choices=...)
seems counter-intuitive. Providing the class directly currently results in afields.E005
error.- To make this more pythonic, the field should also accept the Choice class directly and deal with the variation internally.
- I really can't come up with a scenario where a user would want a different behavior or rather that a user would be surprised by the implicit resolution.
- By forcing the user to expand the
Choices
class manually, essentially all meta information is lost. Downstream packages may benefit from this lost information.- Specifically, as maintainer of drf-spectcular (OpenAPI generator for DRF), I am interested in the name of the choice set (e.g. Suit, Vehicle, Gender) and potentially also the docstring. This would greatly improve OpenAPI generation of choice sets and take out the unnecessary guesswork to find a proper name. (And if anyone wonders, the model field name is not a good candidate for a choice set name.)
This PR allows to use Choices
classes directly as argument, while being transparent. No behavioral changes otherwise.
I marked this as dev
, but it would be awesome if it could still slip into 4.2
. Not sure if the feature window is still open, but this is more or less a trivial and backwards-compatible change with little risk. PR is still missing some docs, which I will write if this is considered.
class Suit(models.IntegerChoices): """ All possible card categories in a deck """ DIAMOND = 1, _("Diamond") SPADE = 2, _("Spade") HEART = 3, _("Heart") CLUB = 4, _("Club") class Choiceful(models.Model): # CURRENTLY: from_enum_old = models.IntegerField(choices=Suit.choices) # NEW: raised an fields.E005 prior to proposed PR. Now, retains reference # to class and transparently resolves via implicit `.choices` call from_new = models.IntegerField(choices=Suit)