django many to many field - programmatically retrieve relations - django

Question

I have been working on a model for tags and am trying to avoid using contenttypes. I have couple questions related to ManyToManyField in django.
I have the following model
taggables/models.py
class Tag(models.Model):
tag_statuses = (
(u'P', _('Pending approval')),
(u'A', _('Approved')),
)
slug = models.SlugField()
created_at = models.DateTimeField(null=True, blank=True)
created_by = models.ForeignKey(User, related_name='tagged_item_created_by')
status = models.CharField(max_length=20, choices=tag_statuses)
site = models.ForeignKey(Site, default=settings.SITE_ID, related_name='tagged_item_site')
def __unicode__(self):
return self.slug
class TagI18n(models.Model):
tag = models.CharField(max_length=100)
descriptor = models.TextField(null=True, blank=True)
# i18n properties
item = models.ForeignKey(Tag)
language = models.CharField(max_length=6, choices=settings.LANGUAGES, default=settings.LANGUAGE_CODE)
class Meta:
unique_together = (("language", "item"))
def __unicode__(self):
return self.tag
I also have different apps around my project that uses tag model as many to many field. such as events for example
evetns/models.py
class Item(models.Model):
event_status_list = (
(u'P', _('Pending approval')),
(u'A', _('Approved')),
(u'R', _('Rejected')),
(u'S', _('Spam')),
)
published_at = models.DateTimeField(null=True, blank=True)
published_by = models.ForeignKey(User, null=True, blank=True, related_name='item_published_by')
updated_by = models.ForeignKey(User, null=True, blank=True, related_name='item_updated_by')
updated_at = models.DateTimeField(null=True, blank=True)
site = models.ForeignKey(Site, default=settings.SITE_ID, related_name='events_item_site')
event_slug = models.SlugField(null=True, blank=True)
# event timing
event_start_date = models.DateField()
event_start_time = models.TimeField(null=True, blank=True)
event_end_date = models.DateField()
event_end_time = models.TimeField(null=True, blank=True)
event_recurrent = models.BooleanField(default=False)
event_status = models.CharField(max_length=20, choices=event_status_list, default=u'P')
# relations
media = models.ManyToManyField(ImageFile, null=True, blank=True)
comments = models.ManyToManyField(Comment, null=True, blank=True)
votes = models.ManyToManyField(Vote, null=True, blank=True)
tags = models.ManyToManyField(Tag, null=True, blank=True)
audience = models.ManyToManyField(Audience, null=True, blank=True)
Now what am trying to do here is run a query to programmatically retrieve all the related models to Tag and then count how many a times a tag was used. Am sure I can do that with contenttypes (generic types) but I don't know how it will perform under heavy usage that's why I wanted to do the many to many fields.

Answer 1

If you are interested in the total number of usage ( aka reference count ) of a tag very often, I think you should store it in the database, example put one extra field to the Tag model, like
referencecount = models.IntegerField( default=0 )
Than in the appropriate places, ( example models .save() )you can increment or decrements it's value.

Answer 2

For your use case, the performance of generic wouldn't matter, because you need anyway to do N queries over 2N tables (one for each "taggable" model and one for each m2m join table, at least).
With the m2m approach, you should have the list of 'taggable' models stored somewhere, at least as a list of ('app_name', 'model') pairs. Then use ContentType (it's very performant) to get the actual model class or query directly from there:
counts = {}
for m in taggable_models:
ct = ContentType.get_by_natural_key(*m)
c = ct.model_class().objects.filter(tags=yourtag).distinct().count()
counts[ct.name] = c