Hello, I am posting here because I have the feeling I might be doing something wrong.
I want to track computer files that can be copied to several data storage devices.
I would like to enforce, on the model level, that you can’t add a file which would exceed the device’s capacity.
As I understand it, you can’t use CheckConstraint across models. So I am using a through model for the ManyToManyField and validate this constraint manually in its save()/clean() method.
Here’s the gist of what I came up with:
class StorageDevice(models.Model):
name = models.CharField(max_length=256)
capacity = models.PositiveBigIntegerField()
def __str__(self):
return self.name
def used_capacity(self):
return self.file_set.aggregate(Sum("size"))["size__sum"] or 0
class File(models.Model):
name = models.CharField(max_length=256)
size = models.PositiveBigIntegerField()
storage_devices = models.ManyToManyField(StorageDevice,
through="FileStorage",
blank=True)
def __str__(self):
return self.name
class FileStorage(models.Model):
file = models.ForeignKey(File, on_delete=models.CASCADE)
storage_device = models.ForeignKey(StorageDevice, on_delete=models.CASCADE)
class Meta:
unique_together = ("file", "storage_device")
def __str__(self):
return f"'{self.file.name}' stored in '{self.storage_device.name}'"
def save(self, *args, **kwargs):
self.full_clean() # Also calls clean().
super().save(*args, **kwargs)
def clean(self):
"""Check if adding this file would exceed the device's capacity."""
current_total = self.storage_device.used_capacity()
if self.file.size + current_total > self.storage_device.capacity:
err = (f"Cannot store file '{self.file.name}' "
"on device '{self.storage_device.name}': capacity exceeded.")
raise ValidationError(err)
This seems to work. Doing this results in the expected ValidationError:
storage = StorageDevice.objects.create(name="device 1", capacity=100)
file = File.objects.create(name="file 1", size=200)
FileStorage.objects.create(file=file, storage_device=storage1)
But doing this, does not:
file.storage_devices.add(storage)
As far as I know this happens because Django directly modifies the data in the database in this case. I have to admit that I don’t understand why Django uses two different approaches here.
My questions:
How would you model such a relationship?
Would it make sense to use a custom ManyToManyDescriptor and overwrite its add() method? I guess I would also have to deal with set() and remove() in that case.
Where does Django use the add() method? Will I get away with simply not calling it myself? How would you prevent code from calling it accidentally?
I think the default admin forms use add(). But due to using a through model, I will have to write a custom admin form anyway.
Additional thoughts:
My approach isn’t reliable in case of changes to the file size or the device’s capacity after establishing their connection. I will probably have to overwrite their save() methods too, right?
Looking forward to your advice.
Thanks.