From the Django docs:
How are the backward relationships possible?
Other object-relational mappers require you to define relationships on
both sides. The Django developers believe this is a violation of the
DRY (Don’t Repeat Yourself) principle, so Django only requires you to
define the relationship on one end.
But how is this possible, given that a model class doesn’t know which
other model classes are related to it until those other model classes
are loaded?
The answer lies in the INSTALLED_APPS setting. The first time any
model is loaded, Django iterates over every model in INSTALLED_APPS
and creates the backward relationships in memory as needed.
Essentially, one of the functions of INSTALLED_APPS is to tell Django
the entire model domain.
Is there a way to get this ORM model? I am trying to debug some reverse relations that are not automagically created and it would really help to see the whole ORM Django has created.
There is no specific ORM "Model", however there are a few things that may help you.
from django.db.models.loading import get_models
get_models() will return you a list of every registered model, this list is what the mechanism that you are describing loops over.
YourModel._meta.get_all_related_objects_with_model()
This function loops over every field in every registered model and finds and returns any reverse relations to your YourModel.
The Options class from django.db.models.options (YourModel._meta is an Options object) is a good place to look around for this stuff.
Django doesn't "create" an ORM so the question makes no sense. If you want to know what properties the ORM adds to your model classes to support backward relationships then you can
open a django shell, import your models and inspect them
read the source code (hey, it's open source isn't it ?)
if that's not enough, add breakpoints at appropriate places and run the whole thing thru the debugger
Related
Back in previous versions of Django, we all used South to do migrations, and because it wasn't as smart as we might have liked, we sometimes needed to tell it to explicitly ignore some fields because such fields were too complicated for it to handle. We did this with add_ignored_fields and everything pretty much worked.
In our case, we have a "field" on our model that's a subclass of CharField that actually attaches two additional fields to the model via the contribute_to_class method. It's kind of a weird process, but it makes sense for our use case and worked just fine until we upgraded to Django 1.7.
However, with the new migrations system, our use of add_ignored_fields() no longer works and our unit tests are exploding as it tries to create a test database table with these fields. Specifically it says:
django.db.utils.OperationalError: duplicate column name: our_column_name
In this case our_column_name is one of the fields this special CharField subclass creates.
In a perfect world, we could have the migration code handle these fields properly, but I'd settle for the 1.7 version of add_ignored_fields() if it exists.
I should also point out that we've found the explanation for deconstruct in the Django documentation, but I'm afraid that it doesn't make any sense to me. I can't figure out if it applies to this situation or not.
We've discovered this Django ticket that pretty much states that this is a design pattern that the devs aren't supporting, so we're just going to have to rewrite the model to explicitly create the other fields on the model and then reference said fields from the "parent" field a la ImageField.
I have defined some models within an application, call it "blog".
djangoproject
/blog
models.py
I now want to change the models location, for example put them here:
djangoproject
/blog
xxx
/all_models
models.py
From the code point of view, this is pretty trivial, but the I guess there will be problems on the database since all the tables Django and South created are now called blog_posts blog_comments, Django relies on this naming convention and I don't want to lose the data already present in the database.
How to do this?
The easiest thing is not to bother changing the tables at all, but the code. Inside the Meta class of each of your models, put the declaration db_table = "blog_tablename", and Django will find them without problems.
You can solve this in two ways.
The first and easier one is to provide a db_table in Meta class of each of your models. The other is create a migration to apply the change.
As far as I know south doesn't support table rename, so you should do it as a three way migration:
Move de models, and create migration (now you have both tables old and new)
Create a data migration and iterate over the former table, copying objects to later
Remove the former model, and create a migration for it.
You can read a little bit more about the second way in south docs
How does Django handle changes to my Model? Or, what help does it offer me to do this?
I am thinking of a situation where I have already have published data to my DB which I don't want to lose, but I need to make changes to my data model - for example, adding extra fields to a particular class, changing the types of fields, etc. My understanding is that syncdb won't ever alter tables that already exist in the DB.
For example, let's say I have the following model:
class Person(models.Model):
name = models.CharField(max_length=200)
phone_number=models.CharField(max_length=200)
hair_colour=CharField(max_length=50)
Things I might want to do to Person off the top of my head:
I wish to add an 'age' field.
I realise I want to use IntegerField instead of CharField for phone_number (whether this is a good idea or not, is out of scope...) - assuming it's possible.
I realise that I no longer wish to define hair_colour 'inline' within Person, because several people share the same hair colour - I wish instead to change this to be a foreign key to some other model.
Whilst I can imagine some of these are tough/impossible for the framework to 'guess' exactly what needs to be done to my data if all I do is update the models.py, I can imagine that there might still be some tooling to help enable it - does it exist?
In particular I imagine there must be some good patterns for option 1.
I'm very new to Django and have no experience with any other ORM-type stuff, which I think this is - I've always been a bit suspicious of ORMs, mainly for the reasons above :)
Django itself will not attempt to modify an already created database table. What you are trying to do is typically called "Migration" and there are a couple of different Database Migration tools available for Django.
South
Schema Migrations
Data Migrations
Backwards Migrations
Nash Vegas
Schema Migrations
Data Migrations
Django Evolution
Schema Migrations
Data Migrations (Unknown)
Backwards Migrations (Unknown)
Of the three South is probably the most widely used but they each have different ways of dealing with migrations. You can see more details on the comparison on Django Packages.
Much of what you're asking about can be done with the django project South. You add it as an INSTALLED_APP. Create a baseline, then as your model changes it creates SQL statements to convert your tables and the rows with-in the tables to the new model format.
I'm developing web page with many type of users, each with different profiles properties. I would like to use built in auth system but:
a) I want to use django-registration.
b) I can point User.get_profile to only one profile model.
How to accomplish that in nice fashion?
I haven't used django-registration so I don't know what it entails. For the second part of your question one way would be to
Associate an UserProfile with every User
Add different kinds of ProfileProperty classes and link to them from UseProfile using a generic relationship.
I know, it is a bit of a stretch.
I'm not an expert nor in python, django or database but I encountered a somewhat similar issue few weeks ago and on #django someone advised me to use Generic relation to achieve that.
I created a UserProfile model that is liked (through a OneToOnefield) to the "true" profile, and using contrib.content-type and generic relation I'm able to use several distinct porfile types.
Note that should work for you if you don't fear to hit you database one more time on each get_profile().
An alternative would be to create a big table that contain all the field for all the profile type and using some kind of hook (or reimplementing a custom save()) to check the retired fields according the profile type. But it look complicated to me, especially if you want to have a lot of different profile type.
Another alternative could be to create a TextField derivated custom field and use it as a storage for a dictionary that you pickle to it on save and unpickle from it on load. With some hacking you could certainly map some of the model attribute to the dictionary key. That would allow a lot of flexibility.
Free hint for ya : I also use fixtures to test my application and forgot to check if the raw parameter of the post_save signal was True to prevent executing the UserProfile creation when using manage.py loaddata. As I had coded the creation of the "true" profile in my post_save callback the exception what kind of weird until I find out what was happening.
Usefull ressource :
defining a profile
generic relation
Scenario: large project with many third party apps. Want to add tagging to those apps without having to modify the apps' source.
My first thought was to first specify a list of models in settings.py (like ['appname.modelname',], and call django-tagging's register function on each of them. The register function adds a TagField and a custom manager to the specified model. The problem with that approach is that the function needs to run BEFORE the DB schema is generated.
I tried running the register function directly in settings.py, but I need django.db.models.get_model to get the actual model reference from only a string, and I can't seem to import that from settings.py - no matter what I try I get an ImportError. The tagging.register function imports OK however.
So I changed tactics and wrote a custom management command in an otherwise empty app. The problem there is that the only signal which hooks into syncdb is post_syncdb which is useless to me since it fires after the DB schema has been generated.
The only other approach I can think of at the moment is to generate and run a 'south' like database schema migration. This seems more like a hack than a solution.
This seems like it should be a pretty common need, but I haven't been able to find a clean solution.
So my question is: Is it possible to dynamically add fields to a model BEFORE the schema is generated, but more specifically, is it possible to add tagging to a third party model without editing it's source.
To clarify, I know it is possible to create and store Tags without having a TagField on the model, but there is a major flaw in that approach in that it is difficult to simultaneously create and tag a new model.
From the docs:
You don't have to register your models
in order to use them with the tagging
application - many of the features
added by registration are just
convenience wrappers around the
tagging API provided by the Tag and
TaggedItem models and their managers,
as documented further below.
Take a look at the API documentation and the examples that follow for how you can add tags to any arbitrary object in the system.
http://api.rst2a.com/1.0/rst2/html?uri=http://django-tagging.googlecode.com/svn/trunk/docs/overview.txt#tags
Updated
#views.py
def tag_model_view(request, model_id):
instance_to_tag = SomeModel.objects.get(pk=model_id)
setattr(instance_to_tag, 'tags_for_instance', request.POST['tags'])
...
instance_to_tag.save()
...returns response
#models.py
#this is the post_save signal receiver
def tagging_post_save_handler(sender, instance, created):
if hasattr(instance, 'tags_for_instance'):
Tag.objects.update_tags(instance, instance.tags_for_instance)