I've been trying to alter the value of a form field the Django REST Framework's admin panel and for some reason the change never takes place. I have the serializer below
class SomeView(ModelViewSet):
queryset = MyModel.objects.all()
serializer_class = MyModelSerializer
# I Want to override this and change the POST data
def perform_create(self, serializer):
user = self.request.user.id
# this was a form field where I manually entered the user ID
# I now want it to default to the logged in user
serializer.data['user'] = user
# This returns the original user that was entered into the form field
print serializer.data
I checked out serializer.data with dir() and it's just a python dictionary so I can't figure out why I can't modify the value. As a test, I tried to add extra values but that doesn't work either
# this doesnt work
serializer.data['some_new_field'] = 'test'
EDIT
On another note, I can copy the data and edit it
fake_data = serializer.data.copy()
fake_data['old_value'] = 'new value'
However, it always fails to validate
serializer = MyModelSerializer(data=fake_data)
serializer.is_valid() # returns false
EDIT EDIT:
Ok, so the validation error was caused by Django returning a SimpleLazyObject. Everything works now when I perform a copy of the data, but I'm really curious as to why I can't edit serializer.data directly without copying it. The problem is solved now, but if anyone can provide insight on the issue just for curiosity, that would be awesome.
I checked out serializer.data with dir() and it's just a python dictionary so I can't figure out why I can't modify the value.
While the value returned from Serializer.data is indeed a dictionary, Serializer.data is not a simple instance variable.
If you look at rest_framework/serializers.py:
class Serializer(BaseSerializer, metaclass=SerializerMetaclass):
# [...]
#property
def data(self):
ret = super().data
return ReturnDict(ret, serializer=self)
ReturnDict inherits from OrderedDict, but you still get a new dictionary every time you access Serializer.data.
The real data is in _data, however as noted by the underscore you might not want to modify that either as it is not intended to be public. The values are filled by Serializer.to_representation() which you could override on the viewset.
As for the second part: ModelViewSet defines get_serializer() that is called with the request POST data to create the serializer you want to modify. I'd suggest try to change the input data before the serializer is created, instead.
Related
After updating queryset,
queryset.update(quantity=F('quantity') + increment_amount)
serializer = MySerializer(queryset, many=True, context={'product': product})
serializer.data # here sometimes wrong old data
queryset.values() # shows correct data.
If I evaluate queryset (in DRF), would it contain the update I made above?
My teammate found sometimes DRF serialized data doesn't have the updated info.
But queryset.values() shows the correct updated info.
I can't reproduce this and we are having hard time to google.
If I evaluate queryset (in DRF), would it contain the update I made above?
It should.
My teammate found sometimes DRF serialized data doesn't have the updated info.
Because you don't reevaluate the queryset after performing the update.
But queryset.values() shows the correct updated info.
Because that reevaluates the queryset.
Try with:
queryset.update(quantity=F('quantity') + increment_amount)
serializer = MySerializer(queryset.all(), many=True, context={'product': product})
serializer.data # Should always contain up to date data.
I need to perform some special serialization / deserialization with DRF, but one of the required parameters for those is the user that generated the query.
AFAIK a serializers.Field instance does not have access to ViewSet and thus no access to ViewSet.request.user.
So I thought I'd just add the user as a static field into the queryset so that each record would have access to it.
qry = qry.annotate(user=Value(user, models.ForeignKey(settings.AUTH_USER_MODEL)))
However, this gives me
ValueError: Related model 'auth.User' cannot be resolved
I also tried
q.annotate(user=Value(user, models.ForeignKey(user.__class__)))
but that also excepts.
Exactly what do I have to include so that this will resolve as needed?
Have a look at the CurrentUserDefault:
A default class that can be used to represent the current user. In order to use this, the 'request' must have been provided as part of the context dictionary when instantiating the serializer.
owner = serializers.HiddenField(
default=serializers.CurrentUserDefault()
)
To use this, you need to pass the request in the context as following:
serializer = AccountSerializer(account, context={'request': request})
I have subclassed a text-field form field in Django to create my own custom widget for a field. I was wondering if it's possible to check if all other fields of the form are valid (I want its server side behavior to vary based on the validation of other fields)
See comment
Something like:
class CustomField(TextInput):
def __init__(self, *args, **kwargs):
...
super(CustomField, self).__init__(*args, **kwargs)
input_type = 'hidden'
def value_from_datadict(self, data, files, name):
aws_file_key = data.get(name, None)
_media_bucket = boto.connect_s3(settings.AWS_ACCESS_KEY_ID,
settings.AWS_SECRET_ACCESS_KEY)\
.lookup(settings.AWS_MEDIA_STORAGE_BUCKET_NAME)
try:
key = _media_bucket.get_key(aws_file_key)
except:
print 'Failed to get key.'
key = None
if key and aws_file_key:
fh = tempfile.TemporaryFile()
key.get_contents_to_file(fh)
fh.seek(0)
files = SimpleUploadedFile(key.name, fh.read())
### IF FORM IS VALID DELETE KEY, OTHERWISE, KEEP IT.
if code_to_check_if_valid:
_media_bucket.delete_key(key)
fh.close()
return files
...... etc.
If you want to validate a certain field depending on the values of other fields, you need to to it at the form level and overwrite the field's clean method. Here's the docs on the subject - they are very good.
class CustomForm(forms.Form):
custom_field = CustomField()
def clean(self):
cleaned_data = super(CustomForm, self).clean()
custom_field = cleaned_data.get("custom_field")
...
If you look at the flow of how forms are validated, you will see that the clean method is run if all the other fields validate independently, so at this stage, the form can be considered valid:
These methods are run in the order given above, one field at a time. That is, for each field in the form (in the order they are declared in the form definition), the Field.clean() method (or its override) is run, then clean_<fieldname>(). Finally, once those two methods are run for every field, the Form.clean() method, or its override, is executed.
The final clean method is actually run regardless of if there's an error so you have to iterate through the cleaned_data to make sure there are no errors
The clean() method for the Form class or subclass is always run. If that method raises a ValidationError, cleaned_data will be an empty dictionary.
The previous paragraph means that if you are overriding Form.clean(), you should iterate through self.cleaned_data.items(), possibly considering the _errors dictionary attribute on the form as well. In this way, you will already know which fields have passed their individual validation requirements.
The clean methods for individual fields are called in the same order as the form declaration order or explicitly specified order. [django source code]
Although I wouldn't recommend this approach over using the clean method for multi-field validation, if your custom field is the last field in the order, you can expect self._errors to indicate whether all other fields have passed validation or not. However at this stage, non-field errors won't be available.
Basically I have a model with a ManyToMany field, and then a modelform derived from that model where that field is rendered as a "multiple choice" selectbox. In my template I'm having that field omitted, electing instead to prepare the values for that field in the view, then pass those prepared values into request.POST (actually a copy of request.POST because request.POST is immutable), then feeding request.POST to the form and then carry on as normal. I can't figure out how to do this, because request.POST isn't just a simple python dictionary, but instead a QueryDict, which behaves a little differently.
The field I need to populate is called "not_bases". When I create the widget using the form, it works perfectly well internally, but just not to my liking UI-wise. When I inspect the django-form submitted POST value via django's handy debug error window, the working QueryDict looks like this:
<QueryDict: {u'not_bases': [u'005', u'00AR', u'00F', u'00FD'], [...] }>
It appears the value for "not_bases" is a list, but it's not simply a list. I can't just .append() to it because it won't work. I dug around the documentation and found .update(), which appears to work, but doesn't. Here is my code:
newPOST = request.POST.copy()
for base in bases:
newPOST.update({"not_bases": base.identifier})
and here is the output:
<QueryDict: {u'not_bases': [u'KMER', u'KYIP'], u'reference': [u''], [...] }>
But when I feed that QueryDict to the form, I get an form validation error that says "not_bases: Enter a list of values.". Its obvious that the list-looking things coming from the str() representation of the QueryDict are not the same in the two cases above, even though they look exactly the same
So how do I do this?
It's really not clear what you're trying to do here, but I doubt that hacking the QueryDict is the right way to achieve it.
If you are trying to customise the display of the not_bases field, you can simply override the definition in your modelform declaration:
class MyModelForm(forms.ModelForm):
not_bases = forms.ChoiceField(choices=[(base, base) for base in bases])
class Meta:
model = MyModel
Or, if you simply want to avoid showing it on the form, you can exclude it from the form and set the value after validation.
class MyModelForm(forms.ModelForm):
class Meta:
model = MyModel
exclude = ['not_bases']
....
if request.POST:
if form.is_valid():
instance = form.save(commit=False)
instance.not_bases = bases
instance.save()
Does either of these do what you want?
Thanks to Insin for answering a previous question related to this one.
His answer worked and works well, however, I'm perplexed at the provision of 'cleaned_data', or more precisely, how to use it?
class RegistrationFormPreview(FormPreview):
preview_template = 'workshops/workshop_register_preview.html'
form_template = 'workshops/workshop_register_form.html'
def done(self, request, cleaned_data):
# Do something with the cleaned_data, then redirect
# to a "success" page.
registration = Registration(cleaned_data)
registration.user = request.user
registration.save()
# an attempt to work with cleaned_data throws the error: TypeError
# int() argument must be a string or a number, not 'dict'
# obviously the fk are python objects(?) and not fk_id
# but how to proceed here in an easy way?
# the following works fine, however, it seems to be double handling the POST data
# which had already been processed in the django.formtools.preview.post_post
# method, and passed through to this 'done' method, which is designed to
# be overidden.
'''
form = self.form(request.POST) # instansiate the form with POST data
registration = form.save(commit=False) # save before adding the user
registration.user = request.user # add the user
registration.save() # and save.
'''
return HttpResponseRedirect('/register/success')
For quick reference, here's the contents of the post_post method:
def post_post(self, request):
"Validates the POST data. If valid, calls done(). Else, redisplays form."
f = self.form(request.POST, auto_id=AUTO_ID)
if f.is_valid():
if self.security_hash(request, f) != request.POST.get(self.unused_name('hash')):
return self.failed_hash(request) # Security hash failed.
return self.done(request, f.cleaned_data)
else:
return render_to_response(self.form_template,
{'form': f, 'stage_field': self.unused_name('stage'), 'state': self.state},
context_instance=RequestContext(request))
I've never tried what you're doing here with a ModelForm before, but you might be able to use the ** operator to expand your cleaned_data dictionary into the keyword arguments expected for your Registration constructor:
registration = Registration (**cleaned_data)
The constructor to your model classes take keyword arguments that Django's Model meta class converts to instance-level attributes on the resulting object. The ** operator is a calling convention that tells Python to expand your dictionary into those keyword arguments.
In other words...
What you're doing currently is tantamount to this:
registration = Registration ({'key':'value', ...})
Which is not what you want because the constructor expects keyword arguments as opposed to a dictionary that contains your keyword arguments.
What you want to be doing is this
registration = Registration (key='value', ...)
Which is analogous to this:
registration = Registration (**{'key':'value', ...})
Again, I've never tried it, but it seems like it would work as long as you aren't doing anything fancy with your form, such as adding new attributes to it that aren't expected by your Registration constructor. In that case you'd likely have to modify the items in the cleaned_data dictionary prior to doing this.
It does seem like you're losing out on some of the functionality inherent in ModelForms by going through the form preview utility, though. Perhaps you should take your use case to the Django mailing list and see if there's a potential enhancement to this API that could make it work better with ModelForms.
Edit
Short of what I've described above, you can always just extract the fields from your cleaned_data dictionary "by hand" and pass those into your Registration constructor too, but with the caveat that you have to remember to update this code as you add new fields to your model.
registration = Registration (
x=cleaned_data['x'],
y=cleaned_data['y'],
z=cleaned_data['z'],
...
)