thoas / django-sluggable Goto Github PK

Hi,
I have a scenario where I do id-slug as the actual slug, therefore, I don't need the slug to be unique, but it seems that wasn't planned ahead. What would you suggest as a path to handle that?

My idea would be: simply make is_slug_available aware of the uniqueness, that would be set either in the SluggableField instance, or in the decider field?

SluggableField(decider=MySlug, unique=False, populate_from='title')

Or in the decider slug field:

class MySlug(Slug): slug = models.CharField(db_index=True, unique=False)

What do you think?

Hi,
This might not be directly caused by sluggable, but it occurs and I wanted your opinion. Thanks a lot btw for that simple yet very functional app.

sluggable version is 0.7.0, installed from pip, python is 3.8, and Django 2.2

So here goes, I do have a model called Workshop, that has a slug field, defined as:

SluggableField(decider=WorkshopSlug, null=False, populate_from='title', max_length=150, always_update=True)

And WorkshopSlug looks like:

class WorkshopSlug(Slug):
    slug = models.CharField(
        max_length=255, verbose_name=_("URL"), db_index=True
    )
    class Meta:
        abstract = False

This seems rather basic, my idea is to roll out a migration to populate that slug, on an existing db. The title is not allowed to be blank, and none are NULL.
I don't need uniqueness on the slug as I use pk-slug, but I tried without redefining WorkshopSlug.slug, I have the same issue.

I then tried a few things, and in the shell did something like:
for w in Workshop.objects.filter(slug__isnull=True): w.save()
And that worked nicely, the slugs were defined, I was happy. I rolled back the migrations, replayed the one creating the field and model.
Then, I wrote a migration, supposed to run after the one introducing the slug field, and the WorkshopSlug model in the db. Here's the migration:

from django.db import migrations
def set_initial_slugs(apps, schema_editor):
    Workshop = apps.get_model('workshops', 'Workshop')
    db_alias = schema_editor.connection.alias
    for w in Workshop.objects.using(db_alias).all():
        w.save(using=db_alias)
        print('Workshop', w.title, 'has now slug', w.slug)

class Migration(migrations.Migration):

    dependencies = [
        ('workshops', '0019_workshop_slugs'),
    ]

    operations = [
        migrations.RunPython(set_initial_slugs, migrations.RunPython.noop),
    ]

And that doesn't work for some reason, because the Slug is not instanciated (i.e. no signals, no self.decider and what not.). If I try something like w.slug = 'blah', it crashes as it can't get self.decider.

So easy workaround is to directly import the Workshop model, and that works for me. But you're not supposed to do that in a migration, and get them from get_model... However, is that a bug? Is it expected (took me a good hour to understand why it worked in the shell and not in my migration, because in the shell.... apps.get_model ... works!)