django-choicer

Simplify dealing with (large) choices.

Getting started

Imagine we got a model called Subscription providing a field state.

models:

from django.db import models
import choicer

subscription_choicer = choicer.Choicer([
    dict(
        name='started',
        value=0,
        verbose_name='Approved'
    ),
    dict(
        name='inprogress',
        value=1,
        verbose_name='In progress'
    ),
    dict(
        name='completed',
        value=2,
        verbose_name='Completed'
    ),
    dict(
        name='approved',
        value=3,
        verbose_name='Approved'
    ),
    dict(
        name='dismissed',
        value=4,
        verbose_name='Dismissed'
    ),
    dict(
        name='wartinglist',
        value=5,
        verbose_name='On waiting list'
    ),
])

@subscription_choicer.apply(field_name='state')
class Subscription(models.Model):
    # some fields
    state = models.IntegerField(choices=sub_choicer.get_choices())

After we defined our model and applied our choicer, we now want to play with the api:

s = Subscription()
s.set_state_started()
s.save()

print(s.is_state_started())
# True

s.set_state_inprogress()
print(s.is_state_approved())
# False
print (s.is_state_started())
# True

choice = s.STATE_CHOICER.get_by_name('dismissed')
print(choice)
# {'name': 'dismissed', 'value': 4, verbose_name='Dismissed'}

choice = s.STATE_CHOICER.get_by_value(5)
print(choice)
# {'name': 'waitinglist', 'value': 5, verbose_name='On waiting list'}

s.set_state_waitinglist()
print(s.state)
# 5
print(s.get_state())
# {'name': 'waitinglist', 'value': 5, verbose_name='On waiting list'}


s.STATE_CHOICER.get_choices()
# returns the choices

s.STATE_CHOICER.get_dict()
s.STATE_CHOICER.get_list()

Installation

code:

pip install git+https://github.com/tonimichel/django-choicer.git

As django-choicer neither provides models, nor templates, nor static files we dont neet to add it to our project's INSTALLED_APPS.

Motivation

Sometimes models need to provide a large set of choices. This may bring the following issues to deal with:

• Querying

Actually we dont want to query our model by integer or string values which are not human-readable. So, instead of doing

MyModel.objects.filter(type=0)

its better to do something like this

MyModel.objects.filter(state=STATES.approved)

Furthermore it is sometimes necessary to provide instance methods that check for a certain model state. So instead of doing

if obj.state == 0:
    # do something

or a little better

if obj.state == STATES.approved

we actually want to do

if obj.is_state_approved():
    pass

which is the way to go as our model provide an explicit api check for a given state. Considering assignment of a choice, we got similar issues:

obj.state = 0

is worse than

obj.state = STATES.approved

But what we actually want to do is

obj.set_state_approved()

So, now imagine the following scenario. We got 8 different choices, so the naive way of checkin against the integers or strings defined in our choices is inacceptable. But also the way of constructing a "STATE" class for doing enum-like checks (obj.state == STATES.approved) is no that cool, as we actually want to do obj.is_state_approved. But, it would also be totally unacceptable to write 8 getters and 8 setters each providing the same code, especially when we need to change the naming of a choice during early development, which actually happend quite often after Phil Karlton's "There are only two hard things in Computer Science: cache invalidation and naming things".