In the docs for Django's System check framework it says:
Checks should be registered in a file that’s loaded when your application is loaded; for example, in the AppConfig.ready() method.
None of the examples on that page, or around the AppConfig.ready() method, show how to do this. Given a checking method like:
from django.core.checks import register, Tags
#register(Tags.compatibility)
def my_check(app_configs, **kwargs):
# ... perform compatibility checks and collect errors
return errors
How would you do this in/from the AppConfig.ready() method? Is one called from the other? Which file should the above method go in? Do you add #register(...) to the ready() method?
From reading the examples on this page about the Apps Registry and System Checks Framework, it seems there are (at least) two ways to add your own System Checks. To adapt that page's examples (assuming you're creating an app called myapp):
1) Create a myapp/checks.py file like this:
from django.apps import apps as camelot_apps
from django.core.checks import register, Warning
from django.core.checks import Tags as DjangoTags
class Tags(DjangoTags):
"""Do this if none of the existing tags work for you:
https://docs.djangoproject.com/en/1.8/ref/checks/#builtin-tags
"""
my_new_tag = 'my_new_tag'
#register(Tags.my_new_tag)
def check_taggit_is_installed(app_configs=None, **kwargs):
"Check that django-taggit is installed when usying myapp."
errors = []
try:
from taggit import models
except ImportError:
errors.append(
Warning(
"The django-taggit app is required to use myapp.",
hint=("Install django-taggit"),
# A unique ID so it's easier to find this warning:
id='myapp.W001',
)
)
return errors
And then in myapp/__init__.py (create it if it doesn't exist):
from . import checks
Running this should run that check above:
$ ./manage.py check myapp
2) Or, as I thought in my initial question, you can register that check in your AppConfig. So, keep the above code in myapp/check.py, but remove the #register(Tags.my_new_tag) line.
Then create myapp/apps.py containing this:
from django.core.checks import register
from .checks import Tags, check_taggit_is_installed
class MyappConfig(AppConfig):
name = 'myapp'
def ready(self):
super(MyappConfig, self).ready()
register(Tags.my_new_tag)(check_taggit_is_installed)
And alter myapps/__init__.py so it contains this:
from . import checks
default_app_config = 'myapp.apps.MyappConfig'
The first example seems simpler, with no need for a custom AppConfig.
Django recommends:
Checks should be registered in a file that’s loaded when your application is loaded; for example, in the AppConfig.ready() method.
Therefore, place the checks code in a checks.py file. Then simply in apps.py, as with signals:
from django.apps import AppConfig
class MyAppConfig(AppConfig):
name = 'MyApp'
verbose_name = "My App"
def ready(self):
# import myapp.signals
import myapp.checks
My guess is that you can just place this code in ready() method. For example of usage you can take a look at django.contrib.auth (or other contrib app). This app has check in checks.py file, that imported in apps.py and registered in ready() method: checks.py, apps.py.
Related
I have functions in my admin Class that I'd like to perform on a regular basis (every 10 minutes). I am trying to use the "custom django-admin commands" solution ( https://docs.djangoproject.com/en/3.1/howto/custom-management-commands/#howto-custom-management-commands ) and will use a cron job to run it.
I do have the custom command working with test code, however I am running into a few issues:
I'm not sure how to initialize the class (I get errors such as "missing 1 required positional argument"). If I skip instantiation I can call the functions explicitly, but then I run into these next two problems.
I'm used to calling internal Admin class functions using "self.functionname()". But 'self' doesn't work when called from my custom management command.
Many of my admin functions use the standard 'request' argument; however I don't have a 'request' from my custom management command.
These code snippets are paraphrased from the originals.
polls/management/commands/performTasks.py
from django.core.management.base import BaseCommand, CommandError
from polls.models import Question
from polls.admin import QuestionAdmin
class Command(BaseCommand):
help = 'Starts the process.'
def handle(self, *args, **options):
QuestionAdmin.tasks(self, 'YES')
# The following 2 lines I simply use to prove a successful query:
myqueryset = Question.objects.filter(author="Tolkien")
print(myqueryset)
admin.py
from django.contrib import admin
from .models import Question
from django.urls import path
from django.http import HttpResponse, HttpResponseRedirect
#admin.register(Question)
class QuestionAdmin(admin.ModelAdmin):
fields = ['pub_date', 'question_text']
def tasks(self, arg1):
# This does print "YES" so I know the custom management command works.
print(arg1)
# again, the following 2 lines I simply use to prove a successful query:
myqueryset = Question.objects.filter(author="Tolkien")
print(myqueryset)
# This next line fails from the Custom management command with:
# AttributeError: 'Command' object has no attribute 'stepOne'
self.stepOne(self, request, myqueryset)
def stepOne(self, request, queryset):
result = self.stepTwo(request, queryset)
def stepTwo(self, request, queryset):
''' The actual function performs a logic test and returns True or False. '''
return True
models.py
from django.db import models
class Question(models.Model):
question_text = models.CharField(max_length=200)
pub_date = models.DateTimeField('date published')
author = models.CharField(max_length=200, unique=True)
So 3 questions... heavily related:
What's the correct way to initialize my admin Class?
How do I propagate 'self' through my admin Classes functions?
How could I handle the missing 'request' arguments?
Also any comments about the feasibility of using the Custom Management command would be welcome. My goal is to use a complex set of function (that already work from the Admin class) on a regular (every 10 minutes) basis.
I'm working on a simple Django reputation app, based on Actstream justquick/django-activity-stream/ that creates a signal to be used as:
rep.send(request.user, action='increase', target=obj, val=10)
However, even with dispatch UID:
# apps.py
from django.apps import AppConfig
from . signals import rep
from . receivers import rep_handler
class DjangoReputationConfig(AppConfig):
name = 'django_rep'
def ready(self):
rep.connect(rep_handler, dispatch_uid='django_rep.models')`
Which I copied from Actstream/apps.py, I cannot get my signals to send only once. I've tried other unique strings and nothing seems to make a difference.
In my __ init__.py have:
default_app_config = 'django_rep.apps.DjangoReputationConfig'
Thanks! The repo is here if you want to see the code!
I had exactly the same problem. I tried many things, but in the end I had to resort to using an environment variable, like this:
import os
from django.apps import AppConfig
class MyConfig(AppConfig):
name = "xyz"
def ready(self):
if not os.environ.get("ready_called"):
# The development server instantiates MyConfig twice; running via gunicorn only once.
os.environ["ready_called"] = "1" # Use this environment semaphore to prevent starting backgroundtaks more than once.
import backgroundtasks
backgroundtasks.install()
I implemented authentication management using Django auth with the default admin site but then I wanted to use my own AdminSite to rewrite some behaviors:
class OptiAdmin(admin.AdminSite):
site_title = "Optimizer site's admin"
#...Other stuff here
Then registered my own models:
admin_site = OptiAdmin(name='opti_admin')
admin.site.register(MyModel, MyModelAdmin)
#Other stuff here
But when I go to the admin site I am only able to see the models I just registered, which sounds fair to me but I would like to see all the other apps models in this new custom site including the auth's users and groups and I don't know how to do this automatically like the default admin does, pls help :).
Create your own AdminSite with a simple __init__() override.
Import your admin in urls.py.
Replacing the Django Admin and getting the autodiscover() behavior is possible with minimal effort. Here's a project structure generated in the typical django-admin startproject project fashion:
project/
manage.py
project/
__init__.py
settings.py
urls.py
wsgi.py
admin.py # CREATE THIS FILE
project/admin.py: (I think it makes the most sense to do this at the project level.)
from django.contrib.admin import * # PART 1
class MyAdminSite(AdminSite):
site_header = "My Site"
def __init__(self, *args, **kwargs):
super(MyAdminSite, self).__init__(*args, **kwargs)
self._registry.update(site._registry) # PART 2
site = MyAdminSite()
project/urls.py (snippet):
from . import admin # PART 3
urlpatterns = [
url(r'^admin/', admin.site.urls),
]
Part 1 is simple Python. By importing everything from django.contrib.admin into your namespace, it acts as a drop-in replacement. I suppose you don't have to do this, but it helps preserve expectations. Part 3, simply connect up your admin. Part 2 is the real trick. As the documentation says, autodiscover() is called to do the work. All autodiscover does is go through INSTALLED_APPS attempting to import a file called admin.py. Importing runs the code of course and that code is doing the same thing you do to register models (example by decorator and example by method). No magic. You don't have to register your models with your customized admin (as the documentation says).
Autodiscover looks smarter than it is with its register_to kwarg. That indicates you could call autodiscover() yourself passing your own admin. Nope; there's no wiring connected there (future feature?). The assignment happens here and is fixed to the native AdminSite instance here (or here using the decorator). Django contrib models register to that instance and so will any third-party libraries. It's not something you can hook into.
Here's the trick though, _registry is just a dictionary mapping. Let Django autodiscover all the things and then just copy the mapping. That's why self._registry.update(site._registry) works. "self" is your customized AdminSite instance, "site" is Django's instance and you can register your models with either.
(Final note: If models are missing, it's because of import order. All the registration to Django's AdminSite needs to happen before you copy _registry. Registering directly to your customized admin is probably the easiest thing.)
The Django docs suggest using SimpleAdminConfig with a custom admin site.
INSTALLED_APPS = (
...
'django.contrib.admin.apps.SimpleAdminConfig',
...
)
That prevents the models being registered with the default AdminSite.
The docs seem to assume that you will import the models individually and add them to your custom admin site:
from django.contrib.auth.models import Group, User
from django.contrib.auth.admin import GroupAdmin, UserAdmin
admin_site.register(Group, GroupAdmin)
admin_site.register(User, UserAdmin)
This would be very repetitive if you have models in many apps. It doesn't offer any advice how to automatically register models from all your apps with your custom site.
You could try monkey patching admin, and replacing admin.site with your own.
from django.contrib import admin
admin.site = OptiAdmin(name='opti_admin')
Then, when code called admin.site.register(), it would register the model with your admin site. This code would have to run before any models were registered. You could try putting it in the AppConfig for your app, and make sure that your app is above django.contrib.admin.
Adding to JCotton's great answer:
Using django 2.0, overriding site_header and site_title in the custom admin site only works for the index page.
To get it to work with all admin views, extend JCotton's code with the following:
def __init__(self, *args, **kwargs):
super(MyAdminSite, self).__init__(*args, **kwargs)
self._registry.update(site._registry) # PART 2
for model, model_admin in self._registry.items():
model_admin.admin_site = self
Just include init method in your CustomAdminSite class like this.
class CustomAdminSite(admin.AdminSite):
def __init__(self, *args, **kwargs):
super(CustomAdminSite, self).__init__(*args, **kwargs)
self._registry.update(admin.site._registry)
I realize there are many other questions related to custom django signals that don't work, and believe me, I have read all of them several times with no luck for getting my personal situation to work.
Here's the deal: I'm using django-rq to manage a lengthy background process that is set off by a particular http request. When that background process is done, I want it to fire off a custom Django signal so that the django-rq can be checked for any job failure/exceptions.
Two applications, both on the INSTALLED_APPS list, are at the same level. Inside of app1 there is a file:
signals.py
import django.dispatch
file_added = django.dispatch.Signal(providing_args=["issueKey", "file"])
fm_job_done = django.dispatch.Signal(providing_args=["jobId"])
and also a file jobs.py
from app1 import signals
from django.conf import settings
jobId = 23
issueKey = "fake"
fileObj = "alsoFake"
try:
pass
finally:
signals.file_added.send(sender=settings.SIGNAL_SENDER,issueKey=issueKey,fileName=fileObj)
signals.fm_job_done.send(sender=settings.SIGNAL_SENDER,jobId=jobId)
then inside of app2, in views.py
from app1.signals import file_added, fm_job_done
from django.conf import settings
#Setup signal handlers
def fm_job_done_callback(sender, **kwargs):
print "hellooooooooooooooooooooooooooooooooooo"
logging.info("file manager job done signal fired")
def file_added_callback(sender, **kwargs):
print "hellooooooooooooooooooooooooooooooooooo"
logging.info("file added signal fired")
file_added.connect(file_added_callback,sender=settings.SIGNAL_SENDER,weak=False)
fm_job_done.connect(fm_job_done_callback,sender=settings.SIGNAL_SENDER,weak=False)
I don't get any feedback whatsoever though and am at a total loss. I know for fact that jobs.py is executing, and therefore also that the block of code that should be firing the signals is executing as well since it is in a finally block (no the try is not actually empty - I just put pass there for simplicity) Please feel free to ask for more information - I'll respond asap.
here is the solution for django > 2.0
settings.py:
change name of your INSTALLED_APPS from 'app2' to
'app2.apps.App2Config'
app2 -> apps.py:
from app1.signals import file_added, fm_job_done
Class App2Config(AppConfig):
name = 'app2'
def ready(self):
from .views import fm_job_done_callback, file_added_callback
file_added.connect(file_added_callback)
fm_job_done.connect(fm_job_done_callback)
use django receiver decorator
from django.dispatch import receiver
from app1.signals import file_added, fm_job_done
#receiver(fm_job_done)
def fm_job_done_callback(sender, **kwargs):
print "helloooooooooooooo"
#receiver(file_added)
def file_added_callback(sender, **kwargs):
print "helloooooooooooooo"
Also, I prefer to handle signals in models.py
I've been struggling with this problem for 5 hours and I have a feeling it's a simple solution that I'm just overlooking.
I'm trying to tie in a third party module (Django Activity Stream) that uses a series of senders and receivers to post data about user activity to a database table. Everything is set up and installed correctly, but I get a 'Signal' Object has No Attribute 'Save' error when I try to run it.
I suspect the problem is in my syntax somewhere. I'm just getting started with Signals, so am probably overlooking something a veteran will spot immediately.
In views.py I have:
from django.db.models.signals import pre_save
from actstream import action ##This is the third-party app
from models import Bird
def my_handler(sender, **kwargs):
action.save(sender, verb='was saved')
#return HttpResponse("Working Great")
pre_save.connect(my_handler, sender=Bird)
def animal(request):
animal = Bird()
animal.name = "Douglas"
animal.save()
The Django Activity Stream app has this signals.py file:
from django.dispatch import Signal
action = Signal(providing_args=['actor','verb','target','description','timestamp'])
And then this models.py file:
from datetime import datetime
from operator import or_
from django.db import models
from django.db.models.query import QuerySet
from django.core.urlresolvers import reverse
from django.utils.translation import ugettext_lazy as _
from django.utils.timesince import timesince as timesince_
from django.contrib.contenttypes import generic
from django.contrib.contenttypes.models import ContentType
from django.contrib.auth.models import User
from actstream import action
...
def action_handler(verb, target=None, **kwargs):
actor = kwargs.pop('sender')
kwargs.pop('signal', None)
action = Action(actor_content_type=ContentType.objects.get_for_model(actor),
actor_object_id=actor.pk,
verb=unicode(verb),
public=bool(kwargs.pop('public', True)),
description=kwargs.pop('description', None),
timestamp=kwargs.pop('timestamp', datetime.now()))
if target:
action.target_object_id=target.pk
action.target_content_type=ContentType.objects.get_for_model(target)
action.save()
action.connect(action_handler, dispatch_uid="actstream.models")
Your main problem is in the discipline in maintaining coding style, or rather in this case, lack of. You will find that it is easier to identify problems in your code if you do not use the same name to refer to multiple things within the same module; give each object a unique, meaningful name, and refer to it using only that name.
The bottom line here is that the docs for that project contain bad code. This line:
action.save(sender, verb='was saved')
isn't ever going to work. The from actstream import action ultimately imports a signal from actstream.signals, and signals do not and never have had a save method. Especially not with such an odd signature of sender, verb.
At first I thought maybe the author had done something odd with subclassing Signal, but after looking at the rest of the codebase, that's just not the case. I'm not entirely sure what the intention of those docs was supposed to be, but the right thing to do in your handler will either be to save a new Action (imported from actstream.models) instance, or to do something with your model.
Sadly, the project's repository has a pretty sorry set of tests/examples, so without downloading and trying the app myself, I can't tell you what needs to happen there. You might try contacting the author or simply try finding a better-documented/better-maintained Activity Streams app.