source: request.py
If you're doing REST-based web service stuff ... you should ignore request.POST.
— Malcom Tredinnick, Django developers group
REST framework's Request class extends the standard HttpRequest, adding support for REST framework's flexible request parsing and request authentication.
REST framework's Request objects provide flexible request parsing that allows you to treat requests with JSON data or other media types in the same way that you would normally deal with form data.
request.data returns the parsed content of the request body. This is similar to the standard request.POST and request.FILES attributes except that:
- It includes all parsed content, including file and non-file inputs.
- It supports parsing the content of HTTP methods other than
POST, meaning that you can access the content ofPUTandPATCHrequests. - It supports REST framework's flexible request parsing, rather than just supporting form data. For example you can handle incoming JSON data in the same way that you handle incoming form data.
For more details see the parsers documentation.
request.query_params is a more correctly named synonym for request.GET.
For clarity inside your code, we recommend using request.query_params instead of the Django's standard request.GET. Doing so will help keep your codebase more correct and obvious - any HTTP method type may include query parameters, not just GET requests.
The APIView class or @api_view decorator will ensure that this property is automatically set to a list of Parser instances, based on the parser_classes set on the view or based on the DEFAULT_PARSER_CLASSES setting.
You won't typically need to access this property.
Note: If a client sends malformed content, then accessing request.data may raise a ParseError. By default REST framework's APIView class or @api_view decorator will catch the error and return a 400 Bad Request response.
If a client sends a request with a content-type that cannot be parsed then a UnsupportedMediaType exception will be raised, which by default will be caught and return a 415 Unsupported Media Type response.
The request exposes some properties that allow you to determine the result of the content negotiation stage. This allows you to implement behaviour such as selecting a different serialisation schemes for different media types.
The renderer instance what was selected by the content negotiation stage.
A string representing the media type that was accepted by the content negotiation stage.
REST framework provides flexible, per-request authentication, that gives you the ability to:
- Use different authentication policies for different parts of your API.
- Support the use of multiple authentication policies.
- Provide both user and token information associated with the incoming request.
request.user typically returns an instance of django.contrib.auth.models.User, although the behavior depends on the authentication policy being used.
If the request is unauthenticated the default value of request.user is an instance of django.contrib.auth.models.AnonymousUser.
For more details see the authentication documentation.
request.auth returns any additional authentication context. The exact behavior of request.auth depends on the authentication policy being used, but it may typically be an instance of the token that the request was authenticated against.
If the request is unauthenticated, or if no additional context is present, the default value of request.auth is None.
For more details see the authentication documentation.
The APIView class or @api_view decorator will ensure that this property is automatically set to a list of Authentication instances, based on the authentication_classes set on the view or based on the DEFAULT_AUTHENTICATORS setting.
You won't typically need to access this property.
Note: You may see a WrappedAttributeError raised when calling the .user or .auth properties. These errors originate from an authenticator as a standard AttributeError, however it's necessary that they be re-raised as a different exception type in order to prevent them from being suppressed by the outer property access. Python will not recognize that the AttributeError orginates from the authenticator and will instaed assume that the request object does not have a .user or .auth property. The authenticator will need to be fixed.
REST framework supports a few browser enhancements such as browser-based PUT, PATCH and DELETE forms.
request.method returns the uppercased string representation of the request's HTTP method.
Browser-based PUT, PATCH and DELETE forms are transparently supported.
For more information see the browser enhancements documentation.
request.content_type, returns a string object representing the media type of the HTTP request's body, or an empty string if no media type was provided.
You won't typically need to directly access the request's content type, as you'll normally rely on REST framework's default request parsing behavior.
If you do need to access the content type of the request you should use the .content_type property in preference to using request.META.get('HTTP_CONTENT_TYPE'), as it provides transparent support for browser-based non-form content.
For more information see the browser enhancements documentation.
request.stream returns a stream representing the content of the request body.
You won't typically need to directly access the request's content, as you'll normally rely on REST framework's default request parsing behavior.
As REST framework's Request extends Django's HttpRequest, all the other standard attributes and methods are also available. For example the request.META and request.session dictionaries are available as normal.
Note that due to implementation reasons the Request class does not inherit from HttpRequest class, but instead extends the class using composition.
source: response.py
Unlike basic HttpResponse objects, TemplateResponse objects retain the details of the context that was provided by the view to compute the response. The final output of the response is not computed until it is needed, later in the response process.
REST framework supports HTTP content negotiation by providing a Response class which allows you to return content that can be rendered into multiple content types, depending on the client request.
The Response class subclasses Django's SimpleTemplateResponse. Response objects are initialised with data, which should consist of native Python primitives. REST framework then uses standard HTTP content negotiation to determine how it should render the final response content.
There's no requirement for you to use the Response class, you can also return regular HttpResponse or StreamingHttpResponse objects from your views if required. Using the Response class simply provides a nicer interface for returning content-negotiated Web API responses, that can be rendered to multiple formats.
Unless you want to heavily customize REST framework for some reason, you should always use an APIView class or @api_view function for views that return Response objects. Doing so ensures that the view can perform content negotiation and select the appropriate renderer for the response, before it is returned from the view.
Signature: Response(data, status=None, template_name=None, headers=None, content_type=None)
Unlike regular HttpResponse objects, you do not instantiate Response objects with rendered content. Instead you pass in unrendered data, which may consist of any Python primitives.
The renderers used by the Response class cannot natively handle complex datatypes such as Django model instances, so you need to serialize the data into primitive datatypes before creating the Response object.
You can use REST framework's Serializer classes to perform this data serialization, or use your own custom serialization.
Arguments:
data: The serialized data for the response.status: A status code for the response. Defaults to 200. See also status codes.template_name: A template name to use ifHTMLRendereris selected.headers: A dictionary of HTTP headers to use in the response.content_type: The content type of the response. Typically, this will be set automatically by the renderer as determined by content negotiation, but there may be some cases where you need to specify the content type explicitly.
The unrendered, serialized data of the response.
The numeric status code of the HTTP response.
The rendered content of the response. The .render() method must have been called before .content can be accessed.
The template_name, if supplied. Only required if HTMLRenderer or some other custom template renderer is the accepted renderer for the response.
The renderer instance that will be used to render the response.
Set automatically by the APIView or @api_view immediately before the response is returned from the view.
The media type that was selected by the content negotiation stage.
Set automatically by the APIView or @api_view immediately before the response is returned from the view.
A dictionary of additional context information that will be passed to the renderer's .render() method.
Set automatically by the APIView or @api_view immediately before the response is returned from the view.
The Response class extends SimpleTemplateResponse, and all the usual attributes and methods are also available on the response. For example you can set headers on the response in the standard way:
response = Response()
response['Cache-Control'] = 'no-cache'
Signature: .render()
As with any other TemplateResponse, this method is called to render the serialized data of the response into the final response content. When .render() is called, the response content will be set to the result of calling the .render(data, accepted_media_type, renderer_context) method on the accepted_renderer instance.
You won't typically need to call .render() yourself, as it's handled by Django's standard response cycle.
source: decorators.py views.py
Django's class-based views are a welcome departure from the old-style views.
REST framework provides an APIView class, which subclasses Django's View class.
APIView classes are different from regular View classes in the following ways:
- Requests passed to the handler methods will be REST framework's
Requestinstances, not Django'sHttpRequestinstances. - Handler methods may return REST framework's
Response, instead of Django'sHttpResponse. The view will manage content negotiation and setting the correct renderer on the response. - Any
APIExceptionexceptions will be caught and mediated into appropriate responses. - Incoming requests will be authenticated and appropriate permission and/or throttle checks will be run before dispatching the request to the handler method.
Using the APIView class is pretty much the same as using a regular View class, as usual, the incoming request is dispatched to an appropriate handler method such as .get() or .post(). Additionally, a number of attributes may be set on the class that control various aspects of the API policy.
For example:
from rest_framework.views import APIView
from rest_framework.response import Response
from rest_framework import authentication, permissions
from django.contrib.auth.models import User
class ListUsers(APIView):
"""
View to list all users in the system.
* Requires token authentication.
* Only admin users are able to access this view.
"""
authentication_classes = (authentication.TokenAuthentication,)
permission_classes = (permissions.IsAdminUser,)
def get(self, request, format=None):
"""
Return a list of all users.
"""
usernames = [user.username for user in User.objects.all()]
return Response(usernames)
Note: The full methods, attributes on, and relations between Django REST Framework's APIView, GenericAPIView, various Mixins, and Viewsets can be initially complex. In addition to the documentation here, the Classy Django REST Framework resource provides a browsable reference, with full methods and attributes, for each of Django REST Framework's class-based views.
The following attributes control the pluggable aspects of API views.
The following methods are used by REST framework to instantiate the various pluggable API policies. You won't typically need to override these methods.
The following methods are called before dispatching to the handler method.
The following methods are called directly by the view's .dispatch() method.
These perform any actions that need to occur before or after calling the handler methods such as .get(), .post(), put(), patch() and .delete().
Performs any actions that need to occur before the handler method gets called. This method is used to enforce permissions and throttling, and perform content negotiation.
You won't typically need to override this method.
Any exception thrown by the handler method will be passed to this method, which either returns a Response instance, or re-raises the exception.
The default implementation handles any subclass of rest_framework.exceptions.APIException, as well as Django's Http404 and PermissionDenied exceptions, and returns an appropriate error response.
If you need to customize the error responses your API returns you should subclass this method.
Ensures that the request object that is passed to the handler method is an instance of Request, rather than the usual Django HttpRequest.
You won't typically need to override this method.
Ensures that any Response object returned from the handler method will be rendered into the correct content type, as determined by the content negotiation.
You won't typically need to override this method.
Saying [that class-based views] is always the superior solution is a mistake.
REST framework also allows you to work with regular function based views. It provides a set of simple decorators that wrap your function based views to ensure they receive an instance of Request (rather than the usual Django HttpRequest) and allows them to return a Response (instead of a Django HttpResponse), and allow you to configure how the request is processed.
Signature: @api_view(http_method_names=['GET'])
The core of this functionality is the api_view decorator, which takes a list of HTTP methods that your view should respond to. For example, this is how you would write a very simple view that just manually returns some data:
from rest_framework.decorators import api_view
@api_view()
def hello_world(request):
return Response({"message": "Hello, world!"})
This view will use the default renderers, parsers, authentication classes etc specified in the settings.
By default only GET methods will be accepted. Other methods will respond with "405 Method Not Allowed". To alter this behaviour, specify which methods the view allows, like so:
@api_view(['GET', 'POST'])
def hello_world(request):
if request.method == 'POST':
return Response({"message": "Got some data!", "data": request.data})
return Response({"message": "Hello, world!"})
To override the default settings, REST framework provides a set of additional decorators which can be added to your views. These must come after (below) the @api_view decorator. For example, to create a view that uses a throttle to ensure it can only be called once per day by a particular user, use the @throttle_classes decorator, passing a list of throttle classes:
from rest_framework.decorators import api_view, throttle_classes
from rest_framework.throttling import UserRateThrottle
class OncePerDayUserThrottle(UserRateThrottle):
rate = '1/day'
@api_view(['GET'])
@throttle_classes([OncePerDayUserThrottle])
def view(request):
return Response({"message": "Hello for today! See you tomorrow!"})
These decorators correspond to the attributes set on APIView subclasses, described above.
The available decorators are:
@renderer_classes(...)@parser_classes(...)@authentication_classes(...)@throttle_classes(...)@permission_classes(...)
Each of these decorators takes a single argument which must be a list or tuple of classes.
To override the default schema generation for function based views you may use
the @schema decorator. This must come after (below) the @api_view
decorator. For example:
from rest_framework.decorators import api_view, schema
from rest_framework.schemas import AutoSchema
class CustomAutoSchema(AutoSchema):
def get_link(self, path, method, base_url):
# override view introspection here...
@api_view(['GET'])
@schema(CustomAutoSchema())
def view(request):
return Response({"message": "Hello for today! See you tomorrow!"})
This decorator takes a single AutoSchema instance, an AutoSchema subclass
instance or ManualSchema instance as described in the Schemas documentation.
You may pass None in order to exclude the view from schema generation.
@api_view(['GET'])
@schema(None)
def view(request):
return Response({"message": "Will not appear in schema!"})
source: mixins.py generics.py
Django’s generic views... were developed as a shortcut for common usage patterns... They take certain common idioms and patterns found in view development and abstract them so that you can quickly write common views of data without having to repeat yourself.
One of the key benefits of class-based views is the way they allow you to compose bits of reusable behavior. REST framework takes advantage of this by providing a number of pre-built views that provide for commonly used patterns.
The generic views provided by REST framework allow you to quickly build API views that map closely to your database models.
If the generic views don't suit the needs of your API, you can drop down to using the regular APIView class, or reuse the mixins and base classes used by the generic views to compose your own set of reusable generic views.
Typically when using the generic views, you'll override the view, and set several class attributes.
from django.contrib.auth.models import User
from myapp.serializers import UserSerializer
from rest_framework import generics
from rest_framework.permissions import IsAdminUser
class UserList(generics.ListCreateAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
permission_classes = (IsAdminUser,)
For more complex cases you might also want to override various methods on the view class. For example.
class UserList(generics.ListCreateAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
permission_classes = (IsAdminUser,)
def list(self, request):
# Note the use of `get_queryset()` instead of `self.queryset`
queryset = self.get_queryset()
serializer = UserSerializer(queryset, many=True)
return Response(serializer.data)
For very simple cases you might want to pass through any class attributes using the .as_view() method. For example, your URLconf might include something like the following entry:
url(r'^/users/', ListCreateAPIView.as_view(queryset=User.objects.all(), serializer_class=UserSerializer), name='user-list')
This class extends REST framework's APIView class, adding commonly required behavior for standard list and detail views.
Each of the concrete generic views provided is built by combining GenericAPIView, with one or more mixin classes.
Basic settings:
The following attributes control the basic view behavior.
queryset- The queryset that should be used for returning objects from this view. Typically, you must either set this attribute, or override theget_queryset()method. If you are overriding a view method, it is important that you callget_queryset()instead of accessing this property directly, asquerysetwill get evaluated once, and those results will be cached for all subsequent requests.serializer_class- The serializer class that should be used for validating and deserializing input, and for serializing output. Typically, you must either set this attribute, or override theget_serializer_class()method.lookup_field- The model field that should be used to for performing object lookup of individual model instances. Defaults to'pk'. Note that when using hyperlinked APIs you'll need to ensure that both the API views and the serializer classes set the lookup fields if you need to use a custom value.lookup_url_kwarg- The URL keyword argument that should be used for object lookup. The URL conf should include a keyword argument corresponding to this value. If unset this defaults to using the same value aslookup_field.
Pagination:
The following attributes are used to control pagination when used with list views.
pagination_class- The pagination class that should be used when paginating list results. Defaults to the same value as theDEFAULT_PAGINATION_CLASSsetting, which is'rest_framework.pagination.PageNumberPagination'. Settingpagination_class=Nonewill disable pagination on this view.
Filtering:
filter_backends- A list of filter backend classes that should be used for filtering the queryset. Defaults to the same value as theDEFAULT_FILTER_BACKENDSsetting.
Base methods:
Returns the queryset that should be used for list views, and that should be used as the base for lookups in detail views. Defaults to returning the queryset specified by the queryset attribute.
This method should always be used rather than accessing self.queryset directly, as self.queryset gets evaluated only once, and those results are cached for all subsequent requests.
May be overridden to provide dynamic behavior, such as returning a queryset, that is specific to the user making the request.
For example:
def get_queryset(self):
user = self.request.user
return user.accounts.all()
Returns an object instance that should be used for detail views. Defaults to using the lookup_field parameter to filter the base queryset.
May be overridden to provide more complex behavior, such as object lookups based on more than one URL kwarg.
For example:
def get_object(self):
queryset = self.get_queryset()
filter = {}
for field in self.multiple_lookup_fields:
filter[field] = self.kwargs[field]
obj = get_object_or_404(queryset, **filter)
self.check_object_permissions(self.request, obj)
return obj
Note that if your API doesn't include any object level permissions, you may optionally exclude the self.check_object_permissions, and simply return the object from the get_object_or_404 lookup.
Given a queryset, filter it with whichever filter backends are in use, returning a new queryset.
For example:
def filter_queryset(self, queryset):
filter_backends = (CategoryFilter,)
if 'geo_route' in self.request.query_params:
filter_backends = (GeoRouteFilter, CategoryFilter)
elif 'geo_point' in self.request.query_params:
filter_backends = (GeoPointFilter, CategoryFilter)
for backend in list(filter_backends):
queryset = backend().filter_queryset(self.request, queryset, view=self)
return queryset
Returns the class that should be used for the serializer. Defaults to returning the serializer_class attribute.
May be overridden to provide dynamic behavior, such as using different serializers for read and write operations, or providing different serializers to different types of users.
For example:
def get_serializer_class(self):
if self.request.user.is_staff:
return FullAccountSerializer
return BasicAccountSerializer
Save and deletion hooks:
The following methods are provided by the mixin classes, and provide easy overriding of the object save or deletion behavior.
perform_create(self, serializer)- Called byCreateModelMixinwhen saving a new object instance.perform_update(self, serializer)- Called byUpdateModelMixinwhen saving an existing object instance.perform_destroy(self, instance)- Called byDestroyModelMixinwhen deleting an object instance.
These hooks are particularly useful for setting attributes that are implicit in the request, but are not part of the request data. For instance, you might set an attribute on the object based on the request user, or based on a URL keyword argument.
def perform_create(self, serializer):
serializer.save(user=self.request.user)
These override points are also particularly useful for adding behavior that occurs before or after saving an object, such as emailing a confirmation, or logging the update.
def perform_update(self, serializer):
instance = serializer.save()
send_email_confirmation(user=self.request.user, modified=instance)
You can also use these hooks to provide additional validation, by raising a ValidationError(). This can be useful if you need some validation logic to apply at the point of database save. For example:
def perform_create(self, serializer):
queryset = SignupRequest.objects.filter(user=self.request.user)
if queryset.exists():
raise ValidationError('You have already signed up')
serializer.save(user=self.request.user)
Note: These methods replace the old-style version 2.x pre_save, post_save, pre_delete and post_delete methods, which are no longer available.
Other methods:
You won't typically need to override the following methods, although you might need to call into them if you're writing custom views using GenericAPIView.
get_serializer_context(self)- Returns a dictionary containing any extra context that should be supplied to the serializer. Defaults to including'request','view'and'format'keys.get_serializer(self, instance=None, data=None, many=False, partial=False)- Returns a serializer instance.get_paginated_response(self, data)- Returns a paginated styleResponseobject.paginate_queryset(self, queryset)- Paginate a queryset if required, either returning a page object, orNoneif pagination is not configured for this view.filter_queryset(self, queryset)- Given a queryset, filter it with whichever filter backends are in use, returning a new queryset.
The mixin classes provide the actions that are used to provide the basic view behavior. Note that the mixin classes provide action methods rather than defining the handler methods, such as .get() and .post(), directly. This allows for more flexible composition of behavior.
The mixin classes can be imported from rest_framework.mixins.
Provides a .list(request, *args, **kwargs) method, that implements listing a queryset.
If the queryset is populated, this returns a 200 OK response, with a serialized representation of the queryset as the body of the response. The response data may optionally be paginated.
Provides a .create(request, *args, **kwargs) method, that implements creating and saving a new model instance.
If an object is created this returns a 201 Created response, with a serialized representation of the object as the body of the response. If the representation contains a key named url, then the Location header of the response will be populated with that value.
If the request data provided for creating the object was invalid, a 400 Bad Request response will be returned, with the error details as the body of the response.
Provides a .retrieve(request, *args, **kwargs) method, that implements returning an existing model instance in a response.
If an object can be retrieved this returns a 200 OK response, with a serialized representation of the object as the body of the response. Otherwise it will return a 404 Not Found.
Provides a .update(request, *args, **kwargs) method, that implements updating and saving an existing model instance.
Also provides a .partial_update(request, *args, **kwargs) method, which is similar to the update method, except that all fields for the update will be optional. This allows support for HTTP PATCH requests.
If an object is updated this returns a 200 OK response, with a serialized representation of the object as the body of the response.
If the request data provided for updating the object was invalid, a 400 Bad Request response will be returned, with the error details as the body of the response.
Provides a .destroy(request, *args, **kwargs) method, that implements deletion of an existing model instance.
If an object is deleted this returns a 204 No Content response, otherwise it will return a 404 Not Found.
The following classes are the concrete generic views. If you're using generic views this is normally the level you'll be working at unless you need heavily customized behavior.
The view classes can be imported from rest_framework.generics.
Used for create-only endpoints.
Provides a post method handler.
Extends: GenericAPIView, CreateModelMixin
Used for read-only endpoints to represent a collection of model instances.
Provides a get method handler.
Extends: GenericAPIView, ListModelMixin
Used for read-only endpoints to represent a single model instance.
Provides a get method handler.
Extends: GenericAPIView, RetrieveModelMixin
Used for delete-only endpoints for a single model instance.
Provides a delete method handler.
Extends: GenericAPIView, DestroyModelMixin
Used for update-only endpoints for a single model instance.
Provides put and patch method handlers.
Extends: GenericAPIView, UpdateModelMixin
Used for read-write endpoints to represent a collection of model instances.
Provides get and post method handlers.
Extends: GenericAPIView, ListModelMixin, CreateModelMixin
Used for read or update endpoints to represent a single model instance.
Provides get, put and patch method handlers.
Extends: GenericAPIView, RetrieveModelMixin, UpdateModelMixin
Used for read or delete endpoints to represent a single model instance.
Provides get and delete method handlers.
Extends: GenericAPIView, RetrieveModelMixin, DestroyModelMixin
Used for read-write-delete endpoints to represent a single model instance.
Provides get, put, patch and delete method handlers.
Extends: GenericAPIView, RetrieveModelMixin, UpdateModelMixin, DestroyModelMixin
Often you'll want to use the existing generic views, but use some slightly customized behavior. If you find yourself reusing some bit of customized behavior in multiple places, you might want to refactor the behavior into a common class that you can then just apply to any view or viewset as needed.
For example, if you need to lookup objects based on multiple fields in the URL conf, you could create a mixin class like the following:
class MultipleFieldLookupMixin(object):
"""
Apply this mixin to any view or viewset to get multiple field filtering
based on a `lookup_fields` attribute, instead of the default single field filtering.
"""
def get_object(self):
queryset = self.get_queryset() # Get the base queryset
queryset = self.filter_queryset(queryset) # Apply any filter backends
filter = {}
for field in self.lookup_fields:
if self.kwargs[field]: # Ignore empty fields.
filter[field] = self.kwargs[field]
obj = get_object_or_404(queryset, **filter) # Lookup the object
self.check_object_permissions(self.request, obj)
return obj
You can then simply apply this mixin to a view or viewset anytime you need to apply the custom behavior.
class RetrieveUserView(MultipleFieldLookupMixin, generics.RetrieveAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
lookup_fields = ('account', 'username')
Using custom mixins is a good option if you have custom behavior that needs to be used.
If you are using a mixin across multiple views, you can take this a step further and create your own set of base views that can then be used throughout your project. For example:
class BaseRetrieveView(MultipleFieldLookupMixin,
generics.RetrieveAPIView):
pass
class BaseRetrieveUpdateDestroyView(MultipleFieldLookupMixin,
generics.RetrieveUpdateDestroyAPIView):
pass
Using custom base classes is a good option if you have custom behavior that consistently needs to be repeated across a large number of views throughout your project.
Prior to version 3.0 the REST framework mixins treated PUT as either an update or a create operation, depending on if the object already existed or not.
Allowing PUT as create operations is problematic, as it necessarily exposes information about the existence or non-existence of objects. It's also not obvious that transparently allowing re-creating of previously deleted instances is necessarily a better default behavior than simply returning 404 responses.
Both styles "PUT as 404" and "PUT as create" can be valid in different circumstances, but from version 3.0 onwards we now use 404 behavior as the default, due to it being simpler and more obvious.
If you need to generic PUT-as-create behavior you may want to include something like this AllowPUTAsCreateMixin class as a mixin to your views.
The following third party packages provide additional generic view implementations.
The django-rest-framework-bulk package implements generic view mixins as well as some common concrete generic views to allow to apply bulk operations via API requests.
Django Rest Multiple Models provides a generic view (and mixin) for sending multiple serialized models and/or querysets via a single API request.
source: viewsets.py
After routing has determined which controller to use for a request, your controller is responsible for making sense of the request and producing the appropriate output.
Django REST framework allows you to combine the logic for a set of related views in a single class, called a ViewSet. In other frameworks you may also find conceptually similar implementations named something like 'Resources' or 'Controllers'.
A ViewSet class is simply a type of class-based View, that does not provide any method handlers such as .get() or .post(), and instead provides actions such as .list() and .create().
The method handlers for a ViewSet are only bound to the corresponding actions at the point of finalizing the view, using the .as_view() method.
Typically, rather than explicitly registering the views in a viewset in the urlconf, you'll register the viewset with a router class, that automatically determines the urlconf for you.
Let's define a simple viewset that can be used to list or retrieve all the users in the system.
from django.contrib.auth.models import User
from django.shortcuts import get_object_or_404
from myapps.serializers import UserSerializer
from rest_framework import viewsets
from rest_framework.response import Response
class UserViewSet(viewsets.ViewSet):
"""
A simple ViewSet for listing or retrieving users.
"""
def list(self, request):
queryset = User.objects.all()
serializer = UserSerializer(queryset, many=True)
return Response(serializer.data)
def retrieve(self, request, pk=None):
queryset = User.objects.all()
user = get_object_or_404(queryset, pk=pk)
serializer = UserSerializer(user)
return Response(serializer.data)
If we need to, we can bind this viewset into two separate views, like so:
user_list = UserViewSet.as_view({'get': 'list'})
user_detail = UserViewSet.as_view({'get': 'retrieve'})
Typically we wouldn't do this, but would instead register the viewset with a router, and allow the urlconf to be automatically generated.
from myapp.views import UserViewSet
from rest_framework.routers import DefaultRouter
router = DefaultRouter()
router.register(r'users', UserViewSet, base_name='user')
urlpatterns = router.urls
Rather than writing your own viewsets, you'll often want to use the existing base classes that provide a default set of behavior. For example:
class UserViewSet(viewsets.ModelViewSet):
"""
A viewset for viewing and editing user instances.
"""
serializer_class = UserSerializer
queryset = User.objects.all()
There are two main advantages of using a ViewSet class over using a View class.
- Repeated logic can be combined into a single class. In the above example, we only need to specify the
querysetonce, and it'll be used across multiple views. - By using routers, we no longer need to deal with wiring up the URL conf ourselves.
Both of these come with a trade-off. Using regular views and URL confs is more explicit and gives you more control. ViewSets are helpful if you want to get up and running quickly, or when you have a large API and you want to enforce a consistent URL configuration throughout.
The default routers included with REST framework will provide routes for a standard set of create/retrieve/update/destroy style actions, as shown below:
class UserViewSet(viewsets.ViewSet):
"""
Example empty viewset demonstrating the standard
actions that will be handled by a router class.
If you're using format suffixes, make sure to also include
the `format=None` keyword argument for each action.
"""
def list(self, request):
pass
def create(self, request):
pass
def retrieve(self, request, pk=None):
pass
def update(self, request, pk=None):
pass
def partial_update(self, request, pk=None):
pass
def destroy(self, request, pk=None):
pass
During dispatch, the following attributes are available on the ViewSet.
basename- the base to use for the URL names that are created.action- the name of the current action (e.g.,list,create).detail- boolean indicating if the current action is configured for a list or detail view.suffix- the display suffix for the viewset type - mirrors thedetailattribute.
You may inspect these attributes to adjust behaviour based on the current action. For example, you could restrict permissions to everything except the list action similar to this:
def get_permissions(self):
"""
Instantiates and returns the list of permissions that this view requires.
"""
if self.action == 'list':
permission_classes = [IsAuthenticated]
else:
permission_classes = [IsAdmin]
return [permission() for permission in permission_classes]
If you have ad-hoc methods that should be routable, you can mark them as such with the @action decorator. Like regular actions, extra actions may be intended for either a list of objects, or a single instance. To indicate this, set the detail argument to True or False. The router will configure its URL patterns accordingly. e.g., the DefaultRouter will configure detail actions to contain pk in their URL patterns.
A more complete example of extra actions:
from django.contrib.auth.models import User
from rest_framework import status, viewsets
from rest_framework.decorators import action
from rest_framework.response import Response
from myapp.serializers import UserSerializer, PasswordSerializer
class UserViewSet(viewsets.ModelViewSet):
"""
A viewset that provides the standard actions
"""
queryset = User.objects.all()
serializer_class = UserSerializer
@action(methods=['post'], detail=True)
def set_password(self, request, pk=None):
user = self.get_object()
serializer = PasswordSerializer(data=request.data)
if serializer.is_valid():
user.set_password(serializer.data['password'])
user.save()
return Response({'status': 'password set'})
else:
return Response(serializer.errors,
status=status.HTTP_400_BAD_REQUEST)
@action(detail=False)
def recent_users(self, request):
recent_users = User.objects.all().order('-last_login')
page = self.paginate_queryset(recent_users)
if page is not None:
serializer = self.get_serializer(page, many=True)
return self.get_paginated_response(serializer.data)
serializer = self.get_serializer(recent_users, many=True)
return Response(serializer.data)
The decorator can additionally take extra arguments that will be set for the routed view only. For example:
@action(methods=['post'], detail=True, permission_classes=[IsAdminOrIsSelf])
def set_password(self, request, pk=None):
...
These decorator will route GET requests by default, but may also accept other HTTP methods by setting the methods argument. For example:
@action(methods=['post', 'delete'], detail=True)
def unset_password(self, request, pk=None):
...
The two new actions will then be available at the urls ^users/{pk}/set_password/$ and ^users/{pk}/unset_password/$
To view all extra actions, call the .get_extra_actions() method.
If you need to get the URL of an action, use the .reverse_action() method. This is a convenience wrapper for reverse(), automatically passing the view's request object and prepending the url_name with the .basename attribute.
Note that the basename is provided by the router during ViewSet registration. If you are not using a router, then you must provide the basename argument to the .as_view() method.
Using the example from the previous section:
>>> view.reverse_action('set-password', args=['1'])
'http://localhost:8000/api/users/1/set_password'Alternatively, you can use the url_name attribute set by the @action decorator.
>>> view.reverse_action(view.set_password.url_name, args=['1'])
'http://localhost:8000/api/users/1/set_password'The url_name argument for .reverse_action() should match the same argument to the @action decorator. Additionally, this method can be used to reverse the default actions, such as list and create.
The ViewSet class inherits from APIView. You can use any of the standard attributes such as permission_classes, authentication_classes in order to control the API policy on the viewset.
The ViewSet class does not provide any implementations of actions. In order to use a ViewSet class you'll override the class and define the action implementations explicitly.
The GenericViewSet class inherits from GenericAPIView, and provides the default set of get_object, get_queryset methods and other generic view base behavior, but does not include any actions by default.
In order to use a GenericViewSet class you'll override the class and either mixin the required mixin classes, or define the action implementations explicitly.
The ModelViewSet class inherits from GenericAPIView and includes implementations for various actions, by mixing in the behavior of the various mixin classes.
The actions provided by the ModelViewSet class are .list(), .retrieve(), .create(), .update(), .partial_update(), and .destroy().
Because ModelViewSet extends GenericAPIView, you'll normally need to provide at least the queryset and serializer_class attributes. For example:
class AccountViewSet(viewsets.ModelViewSet):
"""
A simple ViewSet for viewing and editing accounts.
"""
queryset = Account.objects.all()
serializer_class = AccountSerializer
permission_classes = [IsAccountAdminOrReadOnly]
Note that you can use any of the standard attributes or method overrides provided by GenericAPIView. For example, to use a ViewSet that dynamically determines the queryset it should operate on, you might do something like this:
class AccountViewSet(viewsets.ModelViewSet):
"""
A simple ViewSet for viewing and editing the accounts
associated with the user.
"""
serializer_class = AccountSerializer
permission_classes = [IsAccountAdminOrReadOnly]
def get_queryset(self):
return self.request.user.accounts.all()
Note however that upon removal of the queryset property from your ViewSet, any associated router will be unable to derive the base_name of your Model automatically, and so you will have to specify the base_name kwarg as part of your router registration.
Also note that although this class provides the complete set of create/list/retrieve/update/destroy actions by default, you can restrict the available operations by using the standard permission classes.
The ReadOnlyModelViewSet class also inherits from GenericAPIView. As with ModelViewSet it also includes implementations for various actions, but unlike ModelViewSet only provides the 'read-only' actions, .list() and .retrieve().
As with ModelViewSet, you'll normally need to provide at least the queryset and serializer_class attributes. For example:
class AccountViewSet(viewsets.ReadOnlyModelViewSet):
"""
A simple ViewSet for viewing accounts.
"""
queryset = Account.objects.all()
serializer_class = AccountSerializer
Again, as with ModelViewSet, you can use any of the standard attributes and method overrides available to GenericAPIView.
You may need to provide custom ViewSet classes that do not have the full set of ModelViewSet actions, or that customize the behavior in some other way.
To create a base viewset class that provides create, list and retrieve operations, inherit from GenericViewSet, and mixin the required actions:
from rest_framework import mixins
class CreateListRetrieveViewSet(mixins.CreateModelMixin,
mixins.ListModelMixin,
mixins.RetrieveModelMixin,
viewsets.GenericViewSet):
"""
A viewset that provides `retrieve`, `create`, and `list` actions.
To use it, override the class and set the `.queryset` and
`.serializer_class` attributes.
"""
pass
By creating your own base ViewSet classes, you can provide common behavior that can be reused in multiple viewsets across your API.
source: routers.py
Resource routing allows you to quickly declare all of the common routes for a given resourceful controller. Instead of declaring separate routes for your index... a resourceful route declares them in a single line of code.
Some Web frameworks such as Rails provide functionality for automatically determining how the URLs for an application should be mapped to the logic that deals with handling incoming requests.
REST framework adds support for automatic URL routing to Django, and provides you with a simple, quick and consistent way of wiring your view logic to a set of URLs.
Here's an example of a simple URL conf, that uses SimpleRouter.
from rest_framework import routers
router = routers.SimpleRouter()
router.register(r'users', UserViewSet)
router.register(r'accounts', AccountViewSet)
urlpatterns = router.urls
There are two mandatory arguments to the register() method:
prefix- The URL prefix to use for this set of routes.viewset- The viewset class.
Optionally, you may also specify an additional argument:
base_name- The base to use for the URL names that are created. If unset the basename will be automatically generated based on thequerysetattribute of the viewset, if it has one. Note that if the viewset does not include aquerysetattribute then you must setbase_namewhen registering the viewset.
The example above would generate the following URL patterns:
- URL pattern:
^users/$Name:'user-list' - URL pattern:
^users/{pk}/$Name:'user-detail' - URL pattern:
^accounts/$Name:'account-list' - URL pattern:
^accounts/{pk}/$Name:'account-detail'
Note: The base_name argument is used to specify the initial part of the view name pattern. In the example above, that's the user or account part.
Typically you won't need to specify the base_name argument, but if you have a viewset where you've defined a custom get_queryset method, then the viewset may not have a .queryset attribute set. If you try to register that viewset you'll see an error like this:
'base_name' argument not specified, and could not automatically determine the name from the viewset, as it does not have a '.queryset' attribute.
This means you'll need to explicitly set the base_name argument when registering the viewset, as it could not be automatically determined from the model name.
The .urls attribute on a router instance is simply a standard list of URL patterns. There are a number of different styles for how you can include these URLs.
For example, you can append router.urls to a list of existing views…
router = routers.SimpleRouter()
router.register(r'users', UserViewSet)
router.register(r'accounts', AccountViewSet)
urlpatterns = [
url(r'^forgot-password/$', ForgotPasswordFormView.as_view()),
]
urlpatterns += router.urls
Alternatively you can use Django's include function, like so…
urlpatterns = [
url(r'^forgot-password/$', ForgotPasswordFormView.as_view()),
url(r'^', include(router.urls)),
]
You may use include with an application namespace:
urlpatterns = [
url(r'^forgot-password/$', ForgotPasswordFormView.as_view()),
url(r'^api/', include((router.urls, 'app_name'))),
]
Or both an application and instance namespace:
urlpatterns = [
url(r'^forgot-password/$', ForgotPasswordFormView.as_view()),
url(r'^api/', include((router.urls, 'app_name'), namespace='instance_name')),
]
See Django's URL namespaces docs and the include API reference for more details.
Note: If using namespacing with hyperlinked serializers you'll also need to ensure that any view_name parameters
on the serializers correctly reflect the namespace. In the examples above you'd need to include a parameter such as
view_name='app_name:user-detail' for serializer fields hyperlinked to the user detail view.
The automatic view_name generation uses a pattern like %(model_name)-detail. Unless your models names actually clash
you may be better off not namespacing your Django REST Framework views when using hyperlinked serializers.
A viewset may mark extra actions for routing by decorating a method with the @action decorator. These extra actions will be included in the generated routes. For example, given the set_password method on the UserViewSet class:
from myapp.permissions import IsAdminOrIsSelf
from rest_framework.decorators import action
class UserViewSet(ModelViewSet):
...
@action(methods=['post'], detail=True, permission_classes=[IsAdminOrIsSelf])
def set_password(self, request, pk=None):
...
The following route would be generated:
- URL pattern:
^users/{pk}/set_password/$ - URL name:
'user-set-password'
By default, the URL pattern is based on the method name, and the URL name is the combination of the ViewSet.basename and the hyphenated method name.
If you don't want to use the defaults for either of these values, you can instead provide the url_path and url_name arguments to the @action decorator.
For example, if you want to change the URL for our custom action to ^users/{pk}/change-password/$, you could write:
from myapp.permissions import IsAdminOrIsSelf
from rest_framework.decorators import action
class UserViewSet(ModelViewSet):
...
@action(methods=['post'], detail=True, permission_classes=[IsAdminOrIsSelf],
url_path='change-password', url_name='change_password')
def set_password(self, request, pk=None):
...
The above example would now generate the following URL pattern:
- URL path:
^users/{pk}/change-password/$ - URL name:
'user-change_password'
This router includes routes for the standard set of list, create, retrieve, update, partial_update and destroy actions. The viewset can also mark additional methods to be routed, using the @action decorator.
| URL Style | HTTP Method | Action | URL Name |
|---|---|---|---|
| {prefix}/ | GET | list | {basename}-list |
| POST | create | ||
| {prefix}/{url_path}/ | GET, or as specified by `methods` argument | `@action(detail=False)` decorated method | {basename}-{url_name} |
| {prefix}/{lookup}/ | GET | retrieve | {basename}-detail |
| PUT | update | ||
| PATCH | partial_update | ||
| DELETE | destroy | ||
| {prefix}/{lookup}/{url_path}/ | GET, or as specified by `methods` argument | `@action(detail=True)` decorated method | {basename}-{url_name} |
By default the URLs created by SimpleRouter are appended with a trailing slash.
This behavior can be modified by setting the trailing_slash argument to False when instantiating the router. For example:
router = SimpleRouter(trailing_slash=False)
Trailing slashes are conventional in Django, but are not used by default in some other frameworks such as Rails. Which style you choose to use is largely a matter of preference, although some javascript frameworks may expect a particular routing style.
The router will match lookup values containing any characters except slashes and period characters. For a more restrictive (or lenient) lookup pattern, set the lookup_value_regex attribute on the viewset. For example, you can limit the lookup to valid UUIDs:
class MyModelViewSet(mixins.RetrieveModelMixin, viewsets.GenericViewSet):
lookup_field = 'my_model_id'
lookup_value_regex = '[0-9a-f]{32}'
This router is similar to SimpleRouter as above, but additionally includes a default API root view, that returns a response containing hyperlinks to all the list views. It also generates routes for optional .json style format suffixes.
| URL Style | HTTP Method | Action | URL Name |
|---|---|---|---|
| [.format] | GET | automatically generated root view | api-root |
| {prefix}/[.format] | GET | list | {basename}-list |
| POST | create | ||
| {prefix}/{url_path}/[.format] | GET, or as specified by `methods` argument | `@action(detail=False)` decorated method | {basename}-{url_name} |
| {prefix}/{lookup}/[.format] | GET | retrieve | {basename}-detail |
| PUT | update | ||
| PATCH | partial_update | ||
| DELETE | destroy | ||
| {prefix}/{lookup}/{url_path}/[.format] | GET, or as specified by `methods` argument | `@action(detail=True)` decorated method | {basename}-{url_name} |
As with SimpleRouter the trailing slashes on the URL routes can be removed by setting the trailing_slash argument to False when instantiating the router.
router = DefaultRouter(trailing_slash=False)
Implementing a custom router isn't something you'd need to do very often, but it can be useful if you have specific requirements about how the URLs for your API are structured. Doing so allows you to encapsulate the URL structure in a reusable way that ensures you don't have to write your URL patterns explicitly for each new view.
The simplest way to implement a custom router is to subclass one of the existing router classes. The .routes attribute is used to template the URL patterns that will be mapped to each viewset. The .routes attribute is a list of Route named tuples.
The arguments to the Route named tuple are:
url: A string representing the URL to be routed. May include the following format strings:
{prefix}- The URL prefix to use for this set of routes.{lookup}- The lookup field used to match against a single instance.{trailing_slash}- Either a '/' or an empty string, depending on thetrailing_slashargument.
mapping: A mapping of HTTP method names to the view methods
name: The name of the URL as used in reverse calls. May include the following format string:
{basename}- The base to use for the URL names that are created.
initkwargs: A dictionary of any additional arguments that should be passed when instantiating the view. Note that the detail, basename, and suffix arguments are reserved for viewset introspection and are also used by the browsable API to generate the view name and breadcrumb links.
You can also customize how the @action decorator is routed. Include the DynamicRoute named tuple in the .routes list, setting the detail argument as appropriate for the list-based and detail-based routes. In addition to detail, the arguments to DynamicRoute are:
url: A string representing the URL to be routed. May include the same format strings as Route, and additionally accepts the {url_path} format string.
name: The name of the URL as used in reverse calls. May include the following format strings:
{basename}- The base to use for the URL names that are created.{url_name}- Theurl_nameprovided to the@action.
initkwargs: A dictionary of any additional arguments that should be passed when instantiating the view.
The following example will only route to the list and retrieve actions, and does not use the trailing slash convention.
from rest_framework.routers import Route, DynamicRoute, SimpleRouter
class CustomReadOnlyRouter(SimpleRouter):
"""
A router for read-only APIs, which doesn't use trailing slashes.
"""
routes = [
Route(
url=r'^{prefix}$',
mapping={'get': 'list'},
name='{basename}-list',
detail=False,
initkwargs={'suffix': 'List'}
),
Route(
url=r'^{prefix}/{lookup}$',
mapping={'get': 'retrieve'},
name='{basename}-detail',
detail=True,
initkwargs={'suffix': 'Detail'}
),
DynamicRoute(
url=r'^{prefix}/{lookup}/{url_path}$',
name='{basename}-{url_name}',
detail=True,
initkwargs={}
)
]
Let's take a look at the routes our CustomReadOnlyRouter would generate for a simple viewset.
views.py:
class UserViewSet(viewsets.ReadOnlyModelViewSet):
"""
A viewset that provides the standard actions
"""
queryset = User.objects.all()
serializer_class = UserSerializer
lookup_field = 'username'
@action(detail=True)
def group_names(self, request, pk=None):
"""
Returns a list of all the group names that the given
user belongs to.
"""
user = self.get_object()
groups = user.groups.all()
return Response([group.name for group in groups])
urls.py:
router = CustomReadOnlyRouter()
router.register('users', UserViewSet)
urlpatterns = router.urls
The following mappings would be generated...
| URL | HTTP Method | Action | URL Name |
|---|---|---|---|
| /users | GET | list | user-list |
| /users/{username} | GET | retrieve | user-detail |
| /users/{username}/group-names | GET | group_names | user-group-names |
For another example of setting the .routes attribute, see the source code for the SimpleRouter class.
If you want to provide totally custom behavior, you can override BaseRouter and override the get_urls(self) method. The method should inspect the registered viewsets and return a list of URL patterns. The registered prefix, viewset and basename tuples may be inspected by accessing the self.registry attribute.
You may also want to override the get_default_base_name(self, viewset) method, or else always explicitly set the base_name argument when registering your viewsets with the router.
The following third party packages are also available.
The drf-nested-routers package provides routers and relationship fields for working with nested resources.
The wq.db package provides an advanced ModelRouter class (and singleton instance) that extends DefaultRouter with a register_model() API. Much like Django's admin.site.register, the only required argument to rest.router.register_model is a model class. Reasonable defaults for a url prefix, serializer, and viewset will be inferred from the model and global configuration.
from wq.db import rest
from myapp.models import MyModel
rest.router.register_model(MyModel)
The DRF-extensions package provides routers for creating nested viewsets, collection level controllers with customizable endpoint names.
source: parsers.py
Machine interacting web services tend to use more structured formats for sending data than form-encoded, since they're sending more complex data than simple forms
— Malcom Tredinnick, Django developers group
REST framework includes a number of built in Parser classes, that allow you to accept requests with various media types. There is also support for defining your own custom parsers, which gives you the flexibility to design the media types that your API accepts.
The set of valid parsers for a view is always defined as a list of classes. When request.data is accessed, REST framework will examine the Content-Type header on the incoming request, and determine which parser to use to parse the request content.
Note: When developing client applications always remember to make sure you're setting the Content-Type header when sending data in an HTTP request.
If you don't set the content type, most clients will default to using 'application/x-www-form-urlencoded', which may not be what you wanted.
As an example, if you are sending json encoded data using jQuery with the .ajax() method, you should make sure to include the contentType: 'application/json' setting.
The default set of parsers may be set globally, using the DEFAULT_PARSER_CLASSES setting. For example, the following settings would allow only requests with JSON content, instead of the default of JSON or form data.
REST_FRAMEWORK = {
'DEFAULT_PARSER_CLASSES': (
'rest_framework.parsers.JSONParser',
)
}
You can also set the parsers used for an individual view, or viewset,
using the APIView class-based views.
from rest_framework.parsers import JSONParser
from rest_framework.response import Response
from rest_framework.views import APIView
class ExampleView(APIView):
"""
A view that can accept POST requests with JSON content.
"""
parser_classes = (JSONParser,)
def post(self, request, format=None):
return Response({'received data': request.data})
Or, if you're using the @api_view decorator with function based views.
from rest_framework.decorators import api_view
from rest_framework.decorators import parser_classes
from rest_framework.parsers import JSONParser
@api_view(['POST'])
@parser_classes((JSONParser,))
def example_view(request, format=None):
"""
A view that can accept POST requests with JSON content.
"""
return Response({'received data': request.data})
Parses JSON request content.
.media_type: application/json
Parses HTML form content. request.data will be populated with a QueryDict of data.
You will typically want to use both FormParser and MultiPartParser together in order to fully support HTML form data.
.media_type: application/x-www-form-urlencoded
Parses multipart HTML form content, which supports file uploads. Both request.data will be populated with a QueryDict.
You will typically want to use both FormParser and MultiPartParser together in order to fully support HTML form data.
.media_type: multipart/form-data
Parses raw file upload content. The request.data property will be a dictionary with a single key 'file' containing the uploaded file.
If the view used with FileUploadParser is called with a filename URL keyword argument, then that argument will be used as the filename.
If it is called without a filename URL keyword argument, then the client must set the filename in the Content-Disposition HTTP header. For example Content-Disposition: attachment; filename=upload.jpg.
.media_type: */*
- The
FileUploadParseris for usage with native clients that can upload the file as a raw data request. For web-based uploads, or for native clients with multipart upload support, you should use theMultiPartParserparser instead. - Since this parser's
media_typematches any content type,FileUploadParsershould generally be the only parser set on an API view. FileUploadParserrespects Django's standardFILE_UPLOAD_HANDLERSsetting, and therequest.upload_handlersattribute. See the Django documentation for more details.
# views.py
class FileUploadView(views.APIView):
parser_classes = (FileUploadParser,)
def put(self, request, filename, format=None):
file_obj = request.data['file']
# ...
# do some stuff with uploaded file
# ...
return Response(status=204)
# urls.py
urlpatterns = [
# ...
url(r'^upload/(?P<filename>[^/]+)$', FileUploadView.as_view())
]
To implement a custom parser, you should override BaseParser, set the .media_type property, and implement the .parse(self, stream, media_type, parser_context) method.
The method should return the data that will be used to populate the request.data property.
The arguments passed to .parse() are:
A stream-like object representing the body of the request.
Optional. If provided, this is the media type of the incoming request content.
Depending on the request's Content-Type: header, this may be more specific than the renderer's media_type attribute, and may include media type parameters. For example "text/plain; charset=utf-8".
Optional. If supplied, this argument will be a dictionary containing any additional context that may be required to parse the request content.
By default this will include the following keys: view, request, args, kwargs.
The following is an example plaintext parser that will populate the request.data property with a string representing the body of the request.
class PlainTextParser(BaseParser):
"""
Plain text parser.
"""
media_type = 'text/plain'
def parse(self, stream, media_type=None, parser_context=None):
"""
Simply return a string representing the body of the request.
"""
return stream.read()
The following third party packages are also available.
REST framework YAML provides YAML parsing and rendering support. It was previously included directly in the REST framework package, and is now instead supported as a third-party package.
Install using pip.
$ pip install djangorestframework-yaml
Modify your REST framework settings.
REST_FRAMEWORK = {
'DEFAULT_PARSER_CLASSES': (
'rest_framework_yaml.parsers.YAMLParser',
),
'DEFAULT_RENDERER_CLASSES': (
'rest_framework_yaml.renderers.YAMLRenderer',
),
}
REST Framework XML provides a simple informal XML format. It was previously included directly in the REST framework package, and is now instead supported as a third-party package.
Install using pip.
$ pip install djangorestframework-xml
Modify your REST framework settings.
REST_FRAMEWORK = {
'DEFAULT_PARSER_CLASSES': (
'rest_framework_xml.parsers.XMLParser',
),
'DEFAULT_RENDERER_CLASSES': (
'rest_framework_xml.renderers.XMLRenderer',
),
}
MessagePack is a fast, efficient binary serialization format. Juan Riaza maintains the djangorestframework-msgpack package which provides MessagePack renderer and parser support for REST framework.
djangorestframework-camel-case provides camel case JSON renderers and parsers for REST framework. This allows serializers to use Python-style underscored field names, but be exposed in the API as Javascript-style camel case field names. It is maintained by Vitaly Babiy.
source: renderers.py
Before a TemplateResponse instance can be returned to the client, it must be rendered. The rendering process takes the intermediate representation of template and context, and turns it into the final byte stream that can be served to the client.
REST framework includes a number of built in Renderer classes, that allow you to return responses with various media types. There is also support for defining your own custom renderers, which gives you the flexibility to design your own media types.
The set of valid renderers for a view is always defined as a list of classes. When a view is entered REST framework will perform content negotiation on the incoming request, and determine the most appropriate renderer to satisfy the request.
The basic process of content negotiation involves examining the request's Accept header, to determine which media types it expects in the response. Optionally, format suffixes on the URL may be used to explicitly request a particular representation. For example the URL http://example.com/api/users_count.json might be an endpoint that always returns JSON data.
For more information see the documentation on content negotiation.
The default set of renderers may be set globally, using the DEFAULT_RENDERER_CLASSES setting. For example, the following settings would use JSON as the main media type and also include the self describing API.
REST_FRAMEWORK = {
'DEFAULT_RENDERER_CLASSES': (
'rest_framework.renderers.JSONRenderer',
'rest_framework.renderers.BrowsableAPIRenderer',
)
}
You can also set the renderers used for an individual view, or viewset,
using the APIView class-based views.
from django.contrib.auth.models import User
from rest_framework.renderers import JSONRenderer
from rest_framework.response import Response
from rest_framework.views import APIView
class UserCountView(APIView):
"""
A view that returns the count of active users in JSON.
"""
renderer_classes = (JSONRenderer, )
def get(self, request, format=None):
user_count = User.objects.filter(active=True).count()
content = {'user_count': user_count}
return Response(content)
Or, if you're using the @api_view decorator with function based views.
@api_view(['GET'])
@renderer_classes((JSONRenderer,))
def user_count_view(request, format=None):
"""
A view that returns the count of active users in JSON.
"""
user_count = User.objects.filter(active=True).count()
content = {'user_count': user_count}
return Response(content)
It's important when specifying the renderer classes for your API to think about what priority you want to assign to each media type. If a client underspecifies the representations it can accept, such as sending an Accept: */* header, or not including an Accept header at all, then REST framework will select the first renderer in the list to use for the response.
For example if your API serves JSON responses and the HTML browsable API, you might want to make JSONRenderer your default renderer, in order to send JSON responses to clients that do not specify an Accept header.
If your API includes views that can serve both regular webpages and API responses depending on the request, then you might consider making TemplateHTMLRenderer your default renderer, in order to play nicely with older browsers that send broken accept headers.
Renders the request data into JSON, using utf-8 encoding.
Note that the default style is to include unicode characters, and render the response using a compact style with no unnecessary whitespace:
{"unicode black star":"★","value":999}
The client may additionally include an 'indent' media type parameter, in which case the returned JSON will be indented. For example Accept: application/json; indent=4.
{
"unicode black star": "★",
"value": 999
}
The default JSON encoding style can be altered using the UNICODE_JSON and COMPACT_JSON settings keys.
.media_type: application/json
.format: '.json'
.charset: None
Renders data to HTML, using Django's standard template rendering.
Unlike other renderers, the data passed to the Response does not need to be serialized. Also, unlike other renderers, you may want to include a template_name argument when creating the Response.
The TemplateHTMLRenderer will create a RequestContext, using the response.data as the context dict, and determine a template name to use to render the context.
The template name is determined by (in order of preference):
- An explicit
template_nameargument passed to the response. - An explicit
.template_nameattribute set on this class. - The return result of calling
view.get_template_names().
An example of a view that uses TemplateHTMLRenderer:
class UserDetail(generics.RetrieveAPIView):
"""
A view that returns a templated HTML representation of a given user.
"""
queryset = User.objects.all()
renderer_classes = (TemplateHTMLRenderer,)
def get(self, request, *args, **kwargs):
self.object = self.get_object()
return Response({'user': self.object}, template_name='user_detail.html')
You can use TemplateHTMLRenderer either to return regular HTML pages using REST framework, or to return both HTML and API responses from a single endpoint.
If you're building websites that use TemplateHTMLRenderer along with other renderer classes, you should consider listing TemplateHTMLRenderer as the first class in the renderer_classes list, so that it will be prioritised first even for browsers that send poorly formed ACCEPT: headers.
See the HTML & Forms Topic Page for further examples of TemplateHTMLRenderer usage.
.media_type: text/html
.format: '.html'
.charset: utf-8
See also: StaticHTMLRenderer
A simple renderer that simply returns pre-rendered HTML. Unlike other renderers, the data passed to the response object should be a string representing the content to be returned.
An example of a view that uses StaticHTMLRenderer:
@api_view(('GET',))
@renderer_classes((StaticHTMLRenderer,))
def simple_html_view(request):
data = '<html><body><h1>Hello, world</h1></body></html>'
return Response(data)
You can use StaticHTMLRenderer either to return regular HTML pages using REST framework, or to return both HTML and API responses from a single endpoint.
.media_type: text/html
.format: '.html'
.charset: utf-8
See also: TemplateHTMLRenderer
Renders data into HTML for the Browsable API:
This renderer will determine which other renderer would have been given highest priority, and use that to display an API style response within the HTML page.
.media_type: text/html
.format: '.api'
.charset: utf-8
.template: 'rest_framework/api.html'
By default the response content will be rendered with the highest priority renderer apart from BrowsableAPIRenderer. If you need to customize this behavior, for example to use HTML as the default return format, but use JSON in the browsable API, you can do so by overriding the get_default_renderer() method. For example:
class CustomBrowsableAPIRenderer(BrowsableAPIRenderer):
def get_default_renderer(self, view):
return JSONRenderer()
Renders data into HTML for an admin-like display:
This renderer is suitable for CRUD-style web APIs that should also present a user-friendly interface for managing the data.
Note that views that have nested or list serializers for their input won't work well with the AdminRenderer, as the HTML forms are unable to properly support them.
Note: The AdminRenderer is only able to include links to detail pages when a properly configured URL_FIELD_NAME (url by default) attribute is present in the data. For HyperlinkedModelSerializer this will be the case, but for ModelSerializer or plain Serializer classes you'll need to make sure to include the field explicitly. For example here we use models get_absolute_url method:
class AccountSerializer(serializers.ModelSerializer):
url = serializers.CharField(source='get_absolute_url', read_only=True)
class Meta:
model = Account
.media_type: text/html
.format: '.admin'
.charset: utf-8
.template: 'rest_framework/admin.html'
Renders data returned by a serializer into an HTML form. The output of this renderer does not include the enclosing <form> tags, a hidden CSRF input or any submit buttons.
This renderer is not intended to be used directly, but can instead be used in templates by passing a serializer instance to the render_form template tag.
{% load rest_framework %}
<form action="/submit-report/" method="post">
{% csrf_token %}
{% render_form serializer %}
<input type="submit" value="Save" />
</form>
For more information see the HTML & Forms documentation.
.media_type: text/html
.format: '.form'
.charset: utf-8
.template: 'rest_framework/horizontal/form.html'
This renderer is used for rendering HTML multipart form data. It is not suitable as a response renderer, but is instead used for creating test requests, using REST framework's test client and test request factory.
.media_type: multipart/form-data; boundary=BoUnDaRyStRiNg
.format: '.multipart'
.charset: utf-8
To implement a custom renderer, you should override BaseRenderer, set the .media_type and .format properties, and implement the .render(self, data, media_type=None, renderer_context=None) method.
The method should return a bytestring, which will be used as the body of the HTTP response.
The arguments passed to the .render() method are:
The request data, as set by the Response() instantiation.
Optional. If provided, this is the accepted media type, as determined by the content negotiation stage.
Depending on the client's Accept: header, this may be more specific than the renderer's media_type attribute, and may include media type parameters. For example "application/json; nested=true".
Optional. If provided, this is a dictionary of contextual information provided by the view.
By default this will include the following keys: view, request, response, args, kwargs.
The following is an example plaintext renderer that will return a response with the data parameter as the content of the response.
from django.utils.encoding import smart_unicode
from rest_framework import renderers
class PlainTextRenderer(renderers.BaseRenderer):
media_type = 'text/plain'
format = 'txt'
def render(self, data, media_type=None, renderer_context=None):
return data.encode(self.charset)
By default renderer classes are assumed to be using the UTF-8 encoding. To use a different encoding, set the charset attribute on the renderer.
class PlainTextRenderer(renderers.BaseRenderer):
media_type = 'text/plain'
format = 'txt'
charset = 'iso-8859-1'
def render(self, data, media_type=None, renderer_context=None):
return data.encode(self.charset)
Note that if a renderer class returns a unicode string, then the response content will be coerced into a bytestring by the Response class, with the charset attribute set on the renderer used to determine the encoding.
If the renderer returns a bytestring representing raw binary content, you should set a charset value of None, which will ensure the Content-Type header of the response will not have a charset value set.
In some cases you may also want to set the render_style attribute to 'binary'. Doing so will also ensure that the browsable API will not attempt to display the binary content as a string.
class JPEGRenderer(renderers.BaseRenderer):
media_type = 'image/jpeg'
format = 'jpg'
charset = None
render_style = 'binary'
def render(self, data, media_type=None, renderer_context=None):
return data
You can do some pretty flexible things using REST framework's renderers. Some examples...
- Provide either flat or nested representations from the same endpoint, depending on the requested media type.
- Serve both regular HTML webpages, and JSON based API responses from the same endpoints.
- Specify multiple types of HTML representation for API clients to use.
- Underspecify a renderer's media type, such as using
media_type = 'image/*', and use theAcceptheader to vary the encoding of the response.
In some cases you might want your view to use different serialization styles depending on the accepted media type. If you need to do this you can access request.accepted_renderer to determine the negotiated renderer that will be used for the response.
For example:
@api_view(('GET',))
@renderer_classes((TemplateHTMLRenderer, JSONRenderer))
def list_users(request):
"""
A view that can return JSON or HTML representations
of the users in the system.
"""
queryset = Users.objects.filter(active=True)
if request.accepted_renderer.format == 'html':
# TemplateHTMLRenderer takes a context dict,
# and additionally requires a 'template_name'.
# It does not require serialization.
data = {'users': queryset}
return Response(data, template_name='list_users.html')
# JSONRenderer requires serialized data as normal.
serializer = UserSerializer(instance=queryset)
data = serializer.data
return Response(data)
In some cases you might want a renderer to serve a range of media types.
In this case you can underspecify the media types it should respond to, by using a media_type value such as image/*, or */*.
If you underspecify the renderer's media type, you should make sure to specify the media type explicitly when you return the response, using the content_type attribute. For example:
return Response(data, content_type='image/png')
For the purposes of many Web APIs, simple JSON responses with hyperlinked relations may be sufficient. If you want to fully embrace RESTful design and HATEOAS you'll need to consider the design and usage of your media types in more detail.
In the words of Roy Fielding, "A REST API should spend almost all of its descriptive effort in defining the media type(s) used for representing resources and driving application state, or in defining extended relation names and/or hypertext-enabled mark-up for existing standard media types.".
For good examples of custom media types, see GitHub's use of a custom application/vnd.github+json media type, and Mike Amundsen's IANA approved application/vnd.collection+json JSON-based hypermedia.
Typically a renderer will behave the same regardless of if it's dealing with a regular response, or with a response caused by an exception being raised, such as an Http404 or PermissionDenied exception, or a subclass of APIException.
If you're using either the TemplateHTMLRenderer or the StaticHTMLRenderer and an exception is raised, the behavior is slightly different, and mirrors Django's default handling of error views.
Exceptions raised and handled by an HTML renderer will attempt to render using one of the following methods, by order of precedence.
- Load and render a template named
{status_code}.html. - Load and render a template named
api_exception.html. - Render the HTTP status code and text, for example "404 Not Found".
Templates will render with a RequestContext which includes the status_code and details keys.
Note: If DEBUG=True, Django's standard traceback error page will be displayed instead of rendering the HTTP status code and text.
The following third party packages are also available.
REST framework YAML provides YAML parsing and rendering support. It was previously included directly in the REST framework package, and is now instead supported as a third-party package.
Install using pip.
$ pip install djangorestframework-yaml
Modify your REST framework settings.
REST_FRAMEWORK = {
'DEFAULT_PARSER_CLASSES': (
'rest_framework_yaml.parsers.YAMLParser',
),
'DEFAULT_RENDERER_CLASSES': (
'rest_framework_yaml.renderers.YAMLRenderer',
),
}
REST Framework XML provides a simple informal XML format. It was previously included directly in the REST framework package, and is now instead supported as a third-party package.
Install using pip.
$ pip install djangorestframework-xml
Modify your REST framework settings.
REST_FRAMEWORK = {
'DEFAULT_PARSER_CLASSES': (
'rest_framework_xml.parsers.XMLParser',
),
'DEFAULT_RENDERER_CLASSES': (
'rest_framework_xml.renderers.XMLRenderer',
),
}
REST framework JSONP provides JSONP rendering support. It was previously included directly in the REST framework package, and is now instead supported as a third-party package.
Warning: If you require cross-domain AJAX requests, you should generally be using the more modern approach of CORS as an alternative to JSONP. See the CORS documentation for more details.
The jsonp approach is essentially a browser hack, and is only appropriate for globally readable API endpoints, where GET requests are unauthenticated and do not require any user permissions.
Install using pip.
$ pip install djangorestframework-jsonp
Modify your REST framework settings.
REST_FRAMEWORK = {
'DEFAULT_RENDERER_CLASSES': (
'rest_framework_jsonp.renderers.JSONPRenderer',
),
}
MessagePack is a fast, efficient binary serialization format. Juan Riaza maintains the djangorestframework-msgpack package which provides MessagePack renderer and parser support for REST framework.
Comma-separated values are a plain-text tabular data format, that can be easily imported into spreadsheet applications. Mjumbe Poe maintains the djangorestframework-csv package which provides CSV renderer support for REST framework.
UltraJSON is an optimized C JSON encoder which can give significantly faster JSON rendering. Jacob Haslehurst maintains the drf-ujson-renderer package which implements JSON rendering using the UJSON package.
djangorestframework-camel-case provides camel case JSON renderers and parsers for REST framework. This allows serializers to use Python-style underscored field names, but be exposed in the API as Javascript-style camel case field names. It is maintained by Vitaly Babiy.
Django REST Pandas provides a serializer and renderers that support additional data processing and output via the Pandas DataFrame API. Django REST Pandas includes renderers for Pandas-style CSV files, Excel workbooks (both .xls and .xlsx), and a number of other formats. It is maintained by S. Andrew Sheppard as part of the wq Project.
Rest Framework Latex provides a renderer that outputs PDFs using Laulatex. It is maintained by Pebble (S/F Software).
source: serializers.py
Expanding the usefulness of the serializers is something that we would like to address. However, it's not a trivial problem, and it will take some serious design work.
— Russell Keith-Magee, Django users group
Serializers allow complex data such as querysets and model instances to be converted to native Python datatypes that can then be easily rendered into JSON, XML or other content types. Serializers also provide deserialization, allowing parsed data to be converted back into complex types, after first validating the incoming data.
The serializers in REST framework work very similarly to Django's Form and ModelForm classes. We provide a Serializer class which gives you a powerful, generic way to control the output of your responses, as well as a ModelSerializer class which provides a useful shortcut for creating serializers that deal with model instances and querysets.
Let's start by creating a simple object we can use for example purposes:
from datetime import datetime
class Comment(object):
def __init__(self, email, content, created=None):
self.email = email
self.content = content
self.created = created or datetime.now()
comment = Comment(email='leila@example.com', content='foo bar')
We'll declare a serializer that we can use to serialize and deserialize data that corresponds to Comment objects.
Declaring a serializer looks very similar to declaring a form:
from rest_framework import serializers
class CommentSerializer(serializers.Serializer):
email = serializers.EmailField()
content = serializers.CharField(max_length=200)
created = serializers.DateTimeField()
We can now use CommentSerializer to serialize a comment, or list of comments. Again, using the Serializer class looks a lot like using a Form class.
serializer = CommentSerializer(comment)
serializer.data
# {'email': 'leila@example.com', 'content': 'foo bar', 'created': '2016-01-27T15:17:10.375877'}
At this point we've translated the model instance into Python native datatypes. To finalise the serialization process we render the data into json.
from rest_framework.renderers import JSONRenderer
json = JSONRenderer().render(serializer.data)
json
# b'{"email":"leila@example.com","content":"foo bar","created":"2016-01-27T15:17:10.375877"}'
Deserialization is similar. First we parse a stream into Python native datatypes...
from django.utils.six import BytesIO
from rest_framework.parsers import JSONParser
stream = BytesIO(json)
data = JSONParser().parse(stream)
...then we restore those native datatypes into a dictionary of validated data.
serializer = CommentSerializer(data=data)
serializer.is_valid()
# True
serializer.validated_data
# {'content': 'foo bar', 'email': 'leila@example.com', 'created': datetime.datetime(2012, 08, 22, 16, 20, 09, 822243)}
If we want to be able to return complete object instances based on the validated data we need to implement one or both of the .create() and .update() methods. For example:
class CommentSerializer(serializers.Serializer):
email = serializers.EmailField()
content = serializers.CharField(max_length=200)
created = serializers.DateTimeField()
def create(self, validated_data):
return Comment(**validated_data)
def update(self, instance, validated_data):
instance.email = validated_data.get('email', instance.email)
instance.content = validated_data.get('content', instance.content)
instance.created = validated_data.get('created', instance.created)
return instance
If your object instances correspond to Django models you'll also want to ensure that these methods save the object to the database. For example, if Comment was a Django model, the methods might look like this:
def create(self, validated_data):
return Comment.objects.create(**validated_data)
def update(self, instance, validated_data):
instance.email = validated_data.get('email', instance.email)
instance.content = validated_data.get('content', instance.content)
instance.created = validated_data.get('created', instance.created)
instance.save()
return instance
Now when deserializing data, we can call .save() to return an object instance, based on the validated data.
comment = serializer.save()
Calling .save() will either create a new instance, or update an existing instance, depending on if an existing instance was passed when instantiating the serializer class:
# .save() will create a new instance.
serializer = CommentSerializer(data=data)
# .save() will update the existing `comment` instance.
serializer = CommentSerializer(comment, data=data)
Both the .create() and .update() methods are optional. You can implement either neither, one, or both of them, depending on the use-case for your serializer class.
Sometimes you'll want your view code to be able to inject additional data at the point of saving the instance. This additional data might include information like the current user, the current time, or anything else that is not part of the request data.
You can do so by including additional keyword arguments when calling .save(). For example:
serializer.save(owner=request.user)
Any additional keyword arguments will be included in the validated_data argument when .create() or .update() are called.
In some cases the .create() and .update() method names may not be meaningful. For example, in a contact form we may not be creating new instances, but instead sending an email or other message.
In these cases you might instead choose to override .save() directly, as being more readable and meaningful.
For example:
class ContactForm(serializers.Serializer):
email = serializers.EmailField()
message = serializers.CharField()
def save(self):
email = self.validated_data['email']
message = self.validated_data['message']
send_email(from=email, message=message)
Note that in the case above we're now having to access the serializer .validated_data property directly.
When deserializing data, you always need to call is_valid() before attempting to access the validated data, or save an object instance. If any validation errors occur, the .errors property will contain a dictionary representing the resulting error messages. For example:
serializer = CommentSerializer(data={'email': 'foobar', 'content': 'baz'})
serializer.is_valid()
# False
serializer.errors
# {'email': [u'Enter a valid e-mail address.'], 'created': [u'This field is required.']}
Each key in the dictionary will be the field name, and the values will be lists of strings of any error messages corresponding to that field. The non_field_errors key may also be present, and will list any general validation errors. The name of the non_field_errors key may be customized using the NON_FIELD_ERRORS_KEY REST framework setting.
When deserializing a list of items, errors will be returned as a list of dictionaries representing each of the deserialized items.
#### Raising an exception on invalid data
The .is_valid() method takes an optional raise_exception flag that will cause it to raise a serializers.ValidationError exception if there are validation errors.
These exceptions are automatically dealt with by the default exception handler that REST framework provides, and will return HTTP 400 Bad Request responses by default.
# Return a 400 response if the data was invalid.
serializer.is_valid(raise_exception=True)
You can specify custom field-level validation by adding .validate_<field_name> methods to your Serializer subclass. These are similar to the .clean_<field_name> methods on Django forms.
These methods take a single argument, which is the field value that requires validation.
Your validate_<field_name> methods should return the validated value or raise a serializers.ValidationError. For example:
from rest_framework import serializers
class BlogPostSerializer(serializers.Serializer):
title = serializers.CharField(max_length=100)
content = serializers.CharField()
def validate_title(self, value):
"""
Check that the blog post is about Django.
"""
if 'django' not in value.lower():
raise serializers.ValidationError("Blog post is not about Django")
return value
Note: If your <field_name> is declared on your serializer with the parameter required=False then this validation step will not take place if the field is not included.
To do any other validation that requires access to multiple fields, add a method called .validate() to your Serializer subclass. This method takes a single argument, which is a dictionary of field values. It should raise a serializers.ValidationError if necessary, or just return the validated values. For example:
from rest_framework import serializers
class EventSerializer(serializers.Serializer):
description = serializers.CharField(max_length=100)
start = serializers.DateTimeField()
finish = serializers.DateTimeField()
def validate(self, data):
"""
Check that the start is before the stop.
"""
if data['start'] > data['finish']:
raise serializers.ValidationError("finish must occur after start")
return data
Individual fields on a serializer can include validators, by declaring them on the field instance, for example:
def multiple_of_ten(value):
if value % 10 != 0:
raise serializers.ValidationError('Not a multiple of ten')
class GameRecord(serializers.Serializer):
score = IntegerField(validators=[multiple_of_ten])
...
Serializer classes can also include reusable validators that are applied to the complete set of field data. These validators are included by declaring them on an inner Meta class, like so:
class EventSerializer(serializers.Serializer):
name = serializers.CharField()
room_number = serializers.IntegerField(choices=[101, 102, 103, 201])
date = serializers.DateField()
class Meta:
# Each room only has one event per day.
validators = UniqueTogetherValidator(
queryset=Event.objects.all(),
fields=['room_number', 'date']
)
For more information see the validators documentation.
When passing an initial object or queryset to a serializer instance, the object will be made available as .instance. If no initial object is passed then the .instance attribute will be None.
When passing data to a serializer instance, the unmodified data will be made available as .initial_data. If the data keyword argument is not passed then the .initial_data attribute will not exist.
By default, serializers must be passed values for all required fields or they will raise validation errors. You can use the partial argument in order to allow partial updates.
# Update `comment` with partial data
serializer = CommentSerializer(comment, data={'content': u'foo bar'}, partial=True)
The previous examples are fine for dealing with objects that only have simple datatypes, but sometimes we also need to be able to represent more complex objects, where some of the attributes of an object might not be simple datatypes such as strings, dates or integers.
The Serializer class is itself a type of Field, and can be used to represent relationships where one object type is nested inside another.
class UserSerializer(serializers.Serializer):
email = serializers.EmailField()
username = serializers.CharField(max_length=100)
class CommentSerializer(serializers.Serializer):
user = UserSerializer()
content = serializers.CharField(max_length=200)
created = serializers.DateTimeField()
If a nested representation may optionally accept the None value you should pass the required=False flag to the nested serializer.
class CommentSerializer(serializers.Serializer):
user = UserSerializer(required=False) # May be an anonymous user.
content = serializers.CharField(max_length=200)
created = serializers.DateTimeField()
Similarly if a nested representation should be a list of items, you should pass the many=True flag to the nested serialized.
class CommentSerializer(serializers.Serializer):
user = UserSerializer(required=False)
edits = EditItemSerializer(many=True) # A nested list of 'edit' items.
content = serializers.CharField(max_length=200)
created = serializers.DateTimeField()
When dealing with nested representations that support deserializing the data, any errors with nested objects will be nested under the field name of the nested object.
serializer = CommentSerializer(data={'user': {'email': 'foobar', 'username': 'doe'}, 'content': 'baz'})
serializer.is_valid()
# False
serializer.errors
# {'user': {'email': [u'Enter a valid e-mail address.']}, 'created': [u'This field is required.']}
Similarly, the .validated_data property will include nested data structures.
If you're supporting writable nested representations you'll need to write .create() or .update() methods that handle saving multiple objects.
The following example demonstrates how you might handle creating a user with a nested profile object.
class UserSerializer(serializers.ModelSerializer):
profile = ProfileSerializer()
class Meta:
model = User
fields = ('username', 'email', 'profile')
def create(self, validated_data):
profile_data = validated_data.pop('profile')
user = User.objects.create(**validated_data)
Profile.objects.create(user=user, **profile_data)
return user
For updates you'll want to think carefully about how to handle updates to relationships. For example if the data for the relationship is None, or not provided, which of the following should occur?
- Set the relationship to
NULLin the database. - Delete the associated instance.
- Ignore the data and leave the instance as it is.
- Raise a validation error.
Here's an example for an .update() method on our previous UserSerializer class.
def update(self, instance, validated_data):
profile_data = validated_data.pop('profile')
# Unless the application properly enforces that this field is
# always set, the follow could raise a `DoesNotExist`, which
# would need to be handled.
profile = instance.profile
instance.username = validated_data.get('username', instance.username)
instance.email = validated_data.get('email', instance.email)
instance.save()
profile.is_premium_member = profile_data.get(
'is_premium_member',
profile.is_premium_member
)
profile.has_support_contract = profile_data.get(
'has_support_contract',
profile.has_support_contract
)
profile.save()
return instance
Because the behavior of nested creates and updates can be ambiguous, and may require complex dependencies between related models, REST framework 3 requires you to always write these methods explicitly. The default ModelSerializer .create() and .update() methods do not include support for writable nested representations.
There are however, third-party packages available such as DRF Writable Nested that support automatic writable nested representations.
An alternative to saving multiple related instances in the serializer is to write custom model manager classes that handle creating the correct instances.
For example, suppose we wanted to ensure that User instances and Profile instances are always created together as a pair. We might write a custom manager class that looks something like this:
class UserManager(models.Manager):
...
def create(self, username, email, is_premium_member=False, has_support_contract=False):
user = User(username=username, email=email)
user.save()
profile = Profile(
user=user,
is_premium_member=is_premium_member,
has_support_contract=has_support_contract
)
profile.save()
return user
This manager class now more nicely encapsulates that user instances and profile instances are always created at the same time. Our .create() method on the serializer class can now be re-written to use the new manager method.
def create(self, validated_data):
return User.objects.create(
username=validated_data['username'],
email=validated_data['email']
is_premium_member=validated_data['profile']['is_premium_member']
has_support_contract=validated_data['profile']['has_support_contract']
)
For more details on this approach see the Django documentation on model managers, and this blogpost on using model and manager classes.
The Serializer class can also handle serializing or deserializing lists of objects.
To serialize a queryset or list of objects instead of a single object instance, you should pass the many=True flag when instantiating the serializer. You can then pass a queryset or list of objects to be serialized.
queryset = Book.objects.all()
serializer = BookSerializer(queryset, many=True)
serializer.data
# [
# {'id': 0, 'title': 'The electric kool-aid acid test', 'author': 'Tom Wolfe'},
# {'id': 1, 'title': 'If this is a man', 'author': 'Primo Levi'},
# {'id': 2, 'title': 'The wind-up bird chronicle', 'author': 'Haruki Murakami'}
# ]
The default behavior for deserializing multiple objects is to support multiple object creation, but not support multiple object updates. For more information on how to support or customize either of these cases, see the ListSerializer documentation below.
There are some cases where you need to provide extra context to the serializer in addition to the object being serialized. One common case is if you're using a serializer that includes hyperlinked relations, which requires the serializer to have access to the current request so that it can properly generate fully qualified URLs.
You can provide arbitrary additional context by passing a context argument when instantiating the serializer. For example:
serializer = AccountSerializer(account, context={'request': request})
serializer.data
# {'id': 6, 'owner': u'denvercoder9', 'created': datetime.datetime(2013, 2, 12, 09, 44, 56, 678870), 'details': 'http://example.com/accounts/6/details'}
The context dictionary can be used within any serializer field logic, such as a custom .to_representation() method, by accessing the self.context attribute.
Often you'll want serializer classes that map closely to Django model definitions.
The ModelSerializer class provides a shortcut that lets you automatically create a Serializer class with fields that correspond to the Model fields.
The ModelSerializer class is the same as a regular Serializer class, except that:
- It will automatically generate a set of fields for you, based on the model.
- It will automatically generate validators for the serializer, such as unique_together validators.
- It includes simple default implementations of
.create()and.update().
Declaring a ModelSerializer looks like this:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = ('id', 'account_name', 'users', 'created')
By default, all the model fields on the class will be mapped to a corresponding serializer fields.
Any relationships such as foreign keys on the model will be mapped to PrimaryKeyRelatedField. Reverse relationships are not included by default unless explicitly included as specified in the serializer relations documentation.
Serializer classes generate helpful verbose representation strings, that allow you to fully inspect the state of their fields. This is particularly useful when working with ModelSerializers where you want to determine what set of fields and validators are being automatically created for you.
To do so, open the Django shell, using python manage.py shell, then import the serializer class, instantiate it, and print the object representation…
>>> from myapp.serializers import AccountSerializer
>>> serializer = AccountSerializer()
>>> print(repr(serializer))
AccountSerializer():
id = IntegerField(label='ID', read_only=True)
name = CharField(allow_blank=True, max_length=100, required=False)
owner = PrimaryKeyRelatedField(queryset=User.objects.all())
If you only want a subset of the default fields to be used in a model serializer, you can do so using fields or exclude options, just as you would with a ModelForm. It is strongly recommended that you explicitly set all fields that should be serialized using the fields attribute. This will make it less likely to result in unintentionally exposing data when your models change.
For example:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = ('id', 'account_name', 'users', 'created')
You can also set the fields attribute to the special value '__all__' to indicate that all fields in the model should be used.
For example:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = '__all__'
You can set the exclude attribute to a list of fields to be excluded from the serializer.
For example:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
exclude = ('users',)
In the example above, if the Account model had 3 fields account_name, users, and created, this will result in the fields account_name and created to be serialized.
The names in the fields and exclude attributes will normally map to model fields on the model class.
Alternatively names in the fields options can map to properties or methods which take no arguments that exist on the model class.
Since version 3.3.0, it is mandatory to provide one of the attributes fields or exclude.
The default ModelSerializer uses primary keys for relationships, but you can also easily generate nested representations using the depth option:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = ('id', 'account_name', 'users', 'created')
depth = 1
The depth option should be set to an integer value that indicates the depth of relationships that should be traversed before reverting to a flat representation.
If you want to customize the way the serialization is done you'll need to define the field yourself.
You can add extra fields to a ModelSerializer or override the default fields by declaring fields on the class, just as you would for a Serializer class.
class AccountSerializer(serializers.ModelSerializer):
url = serializers.CharField(source='get_absolute_url', read_only=True)
groups = serializers.PrimaryKeyRelatedField(many=True)
class Meta:
model = Account
Extra fields can correspond to any property or callable on the model.
You may wish to specify multiple fields as read-only. Instead of adding each field explicitly with the read_only=True attribute, you may use the shortcut Meta option, read_only_fields.
This option should be a list or tuple of field names, and is declared as follows:
class AccountSerializer(serializers.ModelSerializer):
class Meta:
model = Account
fields = ('id', 'account_name', 'users', 'created')
read_only_fields = ('account_name',)
Model fields which have editable=False set, and AutoField fields will be set to read-only by default, and do not need to be added to the read_only_fields option.
Note: There is a special-case where a read-only field is part of a unique_together constraint at the model level. In this case the field is required by the serializer class in order to validate the constraint, but should also not be editable by the user.
The right way to deal with this is to specify the field explicitly on the serializer, providing both the read_only=True and default=… keyword arguments.
One example of this is a read-only relation to the currently authenticated User which is unique_together with another identifier. In this case you would declare the user field like so:
user = serializers.PrimaryKeyRelatedField(read_only=True, default=serializers.CurrentUserDefault())
Please review the Validators Documentation for details on the UniqueTogetherValidator and CurrentUserDefault classes.
There is also a shortcut allowing you to specify arbitrary additional keyword arguments on fields, using the extra_kwargs option. As in the case of read_only_fields, this means you do not need to explicitly declare the field on the serializer.
This option is a dictionary, mapping field names to a dictionary of keyword arguments. For example:
class CreateUserSerializer(serializers.ModelSerializer):
class Meta:
model = User
fields = ('email', 'username', 'password')
extra_kwargs = {'password': {'write_only': True}}
def create(self, validated_data):
user = User(
email=validated_data['email'],
username=validated_data['username']
)
user.set_password(validated_data['password'])
user.save()
return user
When serializing model instances, there are a number of different ways you might choose to represent relationships. The default representation for ModelSerializer is to use the primary keys of the related instances.
Alternative representations include serializing using hyperlinks, serializing complete nested representations, or serializing with a custom representation.
For full details see the serializer relations documentation.
The ModelSerializer class also exposes an API that you can override in order to alter how serializer fields are automatically determined when instantiating the serializer.
Normally if a ModelSerializer does not generate the fields you need by default then you should either add them to the class explicitly, or simply use a regular Serializer class instead. However in some cases you may want to create a new base class that defines how the serializer fields are created for any given model.
A mapping of Django model classes to REST framework serializer classes. You can override this mapping to alter the default serializer classes that should be used for each model class.
This property should be the serializer field class, that is used for relational fields by default.
For ModelSerializer this defaults to PrimaryKeyRelatedField.
For HyperlinkedModelSerializer this defaults to serializers.HyperlinkedRelatedField.
The serializer field class that should be used for any url field on the serializer.
Defaults to serializers.HyperlinkedIdentityField
The serializer field class that should be used for any choice fields on the serializer.
Defaults to serializers.ChoiceField
The following methods are called to determine the class and keyword arguments for each field that should be automatically included on the serializer. Each of these methods should return a two tuple of (field_class, field_kwargs).
Called to generate a serializer field that maps to a standard model field.
The default implementation returns a serializer class based on the serializer_field_mapping attribute.
Called to generate a serializer field that maps to a relational model field.
The default implementation returns a serializer class based on the serializer_relational_field attribute.
The relation_info argument is a named tuple, that contains model_field, related_model, to_many and has_through_model properties.
Called to generate a serializer field that maps to a relational model field, when the depth option has been set.
The default implementation dynamically creates a nested serializer class based on either ModelSerializer or HyperlinkedModelSerializer.
The nested_depth will be the value of the depth option, minus one.
The relation_info argument is a named tuple, that contains model_field, related_model, to_many and has_through_model properties.
Called to generate a serializer field that maps to a property or zero-argument method on the model class.
The default implementation returns a ReadOnlyField class.
Called to generate a serializer field for the serializer's own url field. The default implementation returns a HyperlinkedIdentityField class.
Called when the field name did not map to any model field or model property. The default implementation raises an error, although subclasses may customize this behavior.
The HyperlinkedModelSerializer class is similar to the ModelSerializer class except that it uses hyperlinks to represent relationships, rather than primary keys.
By default the serializer will include a url field instead of a primary key field.
The url field will be represented using a HyperlinkedIdentityField serializer field, and any relationships on the model will be represented using a HyperlinkedRelatedField serializer field.
You can explicitly include the primary key by adding it to the fields option, for example:
class AccountSerializer(serializers.HyperlinkedModelSerializer):
class Meta:
model = Account
fields = ('url', 'id', 'account_name', 'users', 'created')
When instantiating a HyperlinkedModelSerializer you must include the current
request in the serializer context, for example:
serializer = AccountSerializer(queryset, context={'request': request})
Doing so will ensure that the hyperlinks can include an appropriate hostname, so that the resulting representation uses fully qualified URLs, such as:
http://api.example.com/accounts/1/
Rather than relative URLs, such as:
/accounts/1/
If you do want to use relative URLs, you should explicitly pass {'request': None}
in the serializer context.
There needs to be a way of determining which views should be used for hyperlinking to model instances.
By default hyperlinks are expected to correspond to a view name that matches the style '{model_name}-detail', and looks up the instance by a pk keyword argument.
You can override a URL field view name and lookup field by using either, or both of, the view_name and lookup_field options in the extra_kwargs setting, like so:
class AccountSerializer(serializers.HyperlinkedModelSerializer):
class Meta:
model = Account
fields = ('account_url', 'account_name', 'users', 'created')
extra_kwargs = {
'url': {'view_name': 'accounts', 'lookup_field': 'account_name'},
'users': {'lookup_field': 'username'}
}
Alternatively you can set the fields on the serializer explicitly. For example:
class AccountSerializer(serializers.HyperlinkedModelSerializer):
url = serializers.HyperlinkedIdentityField(
view_name='accounts',
lookup_field='slug'
)
users = serializers.HyperlinkedRelatedField(
view_name='user-detail',
lookup_field='username',
many=True,
read_only=True
)
class Meta:
model = Account
fields = ('url', 'account_name', 'users', 'created')
Tip: Properly matching together hyperlinked representations and your URL conf can sometimes be a bit fiddly. Printing the repr of a HyperlinkedModelSerializer instance is a particularly useful way to inspect exactly which view names and lookup fields the relationships are expected to map too.
The name of the URL field defaults to 'url'. You can override this globally, by using the URL_FIELD_NAME setting.
The ListSerializer class provides the behavior for serializing and validating multiple objects at once. You won't typically need to use ListSerializer directly, but should instead simply pass many=True when instantiating a serializer.
When a serializer is instantiated and many=True is passed, a ListSerializer instance will be created. The serializer class then becomes a child of the parent ListSerializer
The following argument can also be passed to a ListSerializer field or a serializer that is passed many=True:
This is True by default, but can be set to False if you want to disallow empty lists as valid input.
There are a few use cases when you might want to customize the ListSerializer behavior. For example:
- You want to provide particular validation of the lists, such as checking that one element does not conflict with another element in a list.
- You want to customize the create or update behavior of multiple objects.
For these cases you can modify the class that is used when many=True is passed, by using the list_serializer_class option on the serializer Meta class.
For example:
class CustomListSerializer(serializers.ListSerializer):
...
class CustomSerializer(serializers.Serializer):
...
class Meta:
list_serializer_class = CustomListSerializer
The default implementation for multiple object creation is to simply call .create() for each item in the list. If you want to customize this behavior, you'll need to customize the .create() method on ListSerializer class that is used when many=True is passed.
For example:
class BookListSerializer(serializers.ListSerializer):
def create(self, validated_data):
books = [Book(**item) for item in validated_data]
return Book.objects.bulk_create(books)
class BookSerializer(serializers.Serializer):
...
class Meta:
list_serializer_class = BookListSerializer
By default the ListSerializer class does not support multiple updates. This is because the behavior that should be expected for insertions and deletions is ambiguous.
To support multiple updates you'll need to do so explicitly. When writing your multiple update code make sure to keep the following in mind:
- How do you determine which instance should be updated for each item in the list of data?
- How should insertions be handled? Are they invalid, or do they create new objects?
- How should removals be handled? Do they imply object deletion, or removing a relationship? Should they be silently ignored, or are they invalid?
- How should ordering be handled? Does changing the position of two items imply any state change or is it ignored?
You will need to add an explicit id field to the instance serializer. The default implicitly-generated id field is marked as read_only. This causes it to be removed on updates. Once you declare it explicitly, it will be available in the list serializer's update method.
Here's an example of how you might choose to implement multiple updates:
class BookListSerializer(serializers.ListSerializer):
def update(self, instance, validated_data):
# Maps for id->instance and id->data item.
book_mapping = {book.id: book for book in instance}
data_mapping = {item['id']: item for item in validated_data}
# Perform creations and updates.
ret = []
for book_id, data in data_mapping.items():
book = book_mapping.get(book_id, None)
if book is None:
ret.append(self.child.create(data))
else:
ret.append(self.child.update(book, data))
# Perform deletions.
for book_id, book in book_mapping.items():
if book_id not in data_mapping:
book.delete()
return ret
class BookSerializer(serializers.Serializer):
# We need to identify elements in the list using their primary key,
# so use a writable field here, rather than the default which would be read-only.
id = serializers.IntegerField()
...
class Meta:
list_serializer_class = BookListSerializer
It is possible that a third party package may be included alongside the 3.1 release that provides some automatic support for multiple update operations, similar to the allow_add_remove behavior that was present in REST framework 2.
When a serializer with many=True is instantiated, we need to determine which arguments and keyword arguments should be passed to the .__init__() method for both the child Serializer class, and for the parent ListSerializer class.
The default implementation is to pass all arguments to both classes, except for validators, and any custom keyword arguments, both of which are assumed to be intended for the child serializer class.
Occasionally you might need to explicitly specify how the child and parent classes should be instantiated when many=True is passed. You can do so by using the many_init class method.
@classmethod
def many_init(cls, *args, **kwargs):
# Instantiate the child serializer.
kwargs['child'] = cls()
# Instantiate the parent list serializer.
return CustomListSerializer(*args, **kwargs)
BaseSerializer class that can be used to easily support alternative serialization and deserialization styles.
This class implements the same basic API as the Serializer class:
.data- Returns the outgoing primitive representation..is_valid()- Deserializes and validates incoming data..validated_data- Returns the validated incoming data..errors- Returns any errors during validation..save()- Persists the validated data into an object instance.
There are four methods that can be overridden, depending on what functionality you want the serializer class to support:
.to_representation()- Override this to support serialization, for read operations..to_internal_value()- Override this to support deserialization, for write operations..create()and.update()- Override either or both of these to support saving instances.
Because this class provides the same interface as the Serializer class, you can use it with the existing generic class-based views exactly as you would for a regular Serializer or ModelSerializer.
The only difference you'll notice when doing so is the BaseSerializer classes will not generate HTML forms in the browsable API. This is because the data they return does not include all the field information that would allow each field to be rendered into a suitable HTML input.
To implement a read-only serializer using the BaseSerializer class, we just need to override the .to_representation() method. Let's take a look at an example using a simple Django model:
class HighScore(models.Model):
created = models.DateTimeField(auto_now_add=True)
player_name = models.CharField(max_length=10)
score = models.IntegerField()
It's simple to create a read-only serializer for converting HighScore instances into primitive data types.
class HighScoreSerializer(serializers.BaseSerializer):
def to_representation(self, obj):
return {
'score': obj.score,
'player_name': obj.player_name
}
We can now use this class to serialize single HighScore instances:
@api_view(['GET'])
def high_score(request, pk):
instance = HighScore.objects.get(pk=pk)
serializer = HighScoreSerializer(instance)
return Response(serializer.data)
Or use it to serialize multiple instances:
@api_view(['GET'])
def all_high_scores(request):
queryset = HighScore.objects.order_by('-score')
serializer = HighScoreSerializer(queryset, many=True)
return Response(serializer.data)
To create a read-write serializer we first need to implement a .to_internal_value() method. This method returns the validated values that will be used to construct the object instance, and may raise a serializers.ValidationError if the supplied data is in an incorrect format.
Once you've implemented .to_internal_value(), the basic validation API will be available on the serializer, and you will be able to use .is_valid(), .validated_data and .errors.
If you want to also support .save() you'll need to also implement either or both of the .create() and .update() methods.
Here's a complete example of our previous HighScoreSerializer, that's been updated to support both read and write operations.
class HighScoreSerializer(serializers.BaseSerializer):
def to_internal_value(self, data):
score = data.get('score')
player_name = data.get('player_name')
# Perform the data validation.
if not score:
raise serializers.ValidationError({
'score': 'This field is required.'
})
if not player_name:
raise serializers.ValidationError({
'player_name': 'This field is required.'
})
if len(player_name) > 10:
raise serializers.ValidationError({
'player_name': 'May not be more than 10 characters.'
})
# Return the validated values. This will be available as
# the `.validated_data` property.
return {
'score': int(score),
'player_name': player_name
}
def to_representation(self, obj):
return {
'score': obj.score,
'player_name': obj.player_name
}
def create(self, validated_data):
return HighScore.objects.create(**validated_data)
The BaseSerializer class is also useful if you want to implement new generic serializer classes for dealing with particular serialization styles, or for integrating with alternative storage backends.
The following class is an example of a generic serializer that can handle coercing arbitrary objects into primitive representations.
class ObjectSerializer(serializers.BaseSerializer):
"""
A read-only serializer that coerces arbitrary complex objects
into primitive representations.
"""
def to_representation(self, obj):
for attribute_name in dir(obj):
attribute = getattr(obj, attribute_name)
if attribute_name('_'):
# Ignore private attributes.
pass
elif hasattr(attribute, '__call__'):
# Ignore methods and other callables.
pass
elif isinstance(attribute, (str, int, bool, float, type(None))):
# Primitive types can be passed through unmodified.
output[attribute_name] = attribute
elif isinstance(attribute, list):
# Recursively deal with items in lists.
output[attribute_name] = [
self.to_representation(item) for item in attribute
]
elif isinstance(attribute, dict):
# Recursively deal with items in dictionaries.
output[attribute_name] = {
str(key): self.to_representation(value)
for key, value in attribute.items()
}
else:
# Force anything else to its string representation.
output[attribute_name] = str(attribute)
If you need to alter the serialization or deserialization behavior of a serializer class, you can do so by overriding the .to_representation() or .to_internal_value() methods.
Some reasons this might be useful include...
- Adding new behavior for new serializer base classes.
- Modifying the behavior slightly for an existing class.
- Improving serialization performance for a frequently accessed API endpoint that returns lots of data.
The signatures for these methods are as follows:
Takes the object instance that requires serialization, and should return a primitive representation. Typically this means returning a structure of built-in Python datatypes. The exact types that can be handled will depend on the render classes you have configured for your API.
May be overridden in order modify the representation style. For example:
def to_representation(self, instance):
"""Convert `username` to lowercase."""
ret = super().to_representation(instance)
ret['username'] = ret['username'].lower()
return ret
Takes the unvalidated incoming data as input and should return the validated data that will be made available as serializer.validated_data. The return value will also be passed to the .create() or .update() methods if .save() is called on the serializer class.
If any of the validation fails, then the method should raise a serializers.ValidationError(errors). The errors argument should be a dictionary mapping field names (or settings.NON_FIELD_ERRORS_KEY) to a list of error messages. If you don't need to alter deserialization behavior and instead want to provide object-level validation, it's recommended that you instead override the .validate() method.
The data argument passed to this method will normally be the value of request.data, so the datatype it provides will depend on the parser classes you have configured for your API.
Similar to Django forms, you can extend and reuse serializers through inheritance. This allows you to declare a common set of fields or methods on a parent class that can then be used in a number of serializers. For example,
class MyBaseSerializer(Serializer):
my_field = serializers.CharField()
def validate_my_field(self):
...
class MySerializer(MyBaseSerializer):
...
Like Django's Model and ModelForm classes, the inner Meta class on serializers does not implicitly inherit from it's parents' inner Meta classes. If you want the Meta class to inherit from a parent class you must do so explicitly. For example:
class AccountSerializer(MyBaseSerializer):
class Meta(MyBaseSerializer.Meta):
model = Account
Typically we would recommend not using inheritance on inner Meta classes, but instead declaring all options explicitly.
Additionally, the following caveats apply to serializer inheritance:
-
Normal Python name resolution rules apply. If you have multiple base classes that declare a
Metainner class, only the first one will be used. This means the child’sMeta, if it exists, otherwise theMetaof the first parent, etc. -
It’s possible to declaratively remove a
Fieldinherited from a parent class by setting the name to beNoneon the subclass.class MyBaseSerializer(ModelSerializer): my_field = serializers.CharField() class MySerializer(MyBaseSerializer): my_field = NoneHowever, you can only use this technique to opt out from a field defined declaratively by a parent class; it won’t prevent the
ModelSerializerfrom generating a default field. To opt-out from default fields, see Specifying which fields to include.
Once a serializer has been initialized, the dictionary of fields that are set on the serializer may be accessed using the .fields attribute. Accessing and modifying this attribute allows you to dynamically modify the serializer.
Modifying the fields argument directly allows you to do interesting things such as changing the arguments on serializer fields at runtime, rather than at the point of declaring the serializer.
For example, if you wanted to be able to set which fields should be used by a serializer at the point of initializing it, you could create a serializer class like so:
class DynamicFieldsModelSerializer(serializers.ModelSerializer):
"""
A ModelSerializer that takes an additional `fields` argument that
controls which fields should be displayed.
"""
def __init__(self, *args, **kwargs):
# Don't pass the 'fields' arg up to the superclass
fields = kwargs.pop('fields', None)
# Instantiate the superclass normally
super(DynamicFieldsModelSerializer, self).__init__(*args, **kwargs)
if fields is not None:
# Drop any fields that are not specified in the `fields` argument.
allowed = set(fields)
existing = set(self.fields)
for field_name in existing - allowed:
self.fields.pop(field_name)
This would then allow you to do the following:
>>> class UserSerializer(DynamicFieldsModelSerializer):
>>> class Meta:
>>> model = User
>>> fields = ('id', 'username', 'email')
>>>
>>> print UserSerializer(user)
{'id': 2, 'username': 'jonwatts', 'email': 'jon@example.com'}
>>>
>>> print UserSerializer(user, fields=('id', 'email'))
{'id': 2, 'email': 'jon@example.com'}
REST framework 2 provided an API to allow developers to override how a ModelSerializer class would automatically generate the default set of fields.
This API included the .get_field(), .get_pk_field() and other methods.
Because the serializers have been fundamentally redesigned with 3.0 this API no longer exists. You can still modify the fields that get created but you'll need to refer to the source code, and be aware that if the changes you make are against private bits of API then they may be subject to change.
The following third party packages are also available.
The django-rest-marshmallow package provides an alternative implementation for serializers, using the python marshmallow library. It exposes the same API as the REST framework serializers, and can be used as a drop-in replacement in some use-cases.
The serpy package is an alternative implementation for serializers that is built for speed. Serpy serializes complex datatypes to simple native types. The native types can be easily converted to JSON or any other format needed.
The django-rest-framework-mongoengine package provides a MongoEngineModelSerializer serializer class that supports using MongoDB as the storage layer for Django REST framework.
The django-rest-framework-gis package provides a GeoFeatureModelSerializer serializer class that supports GeoJSON both for read and write operations.
The django-rest-framework-hstore package provides an HStoreSerializer to support django-hstore DictionaryField model field and its schema-mode feature.
The dynamic-rest package extends the ModelSerializer and ModelViewSet interfaces, adding API query parameters for filtering, sorting, and including / excluding all fields and relationships defined by your serializers.
The drf-dynamic-fields package provides a mixin to dynamically limit the fields per serializer to a subset specified by an URL parameter.
The drf-flex-fields package extends the ModelSerializer and ModelViewSet to provide commonly used functionality for dynamically setting fields and expanding primitive fields to nested models, both from URL parameters and your serializer class definitions.
The django-rest-framework-serializer-extensions package provides a collection of tools to DRY up your serializers, by allowing fields to be defined on a per-view/request basis. Fields can be whitelisted, blacklisted and child serializers can be optionally expanded.
The html-json-forms package provides an algorithm and serializer for processing <form> submissions per the (inactive) HTML JSON Form specification. The serializer facilitates processing of arbitrarily nested JSON structures within HTML. For example, <input name="items[0][id]" value="5"> will be interpreted as {"items": [{"id": "5"}]}.
DRF-Base64 provides a set of field and model serializers that handles the upload of base64-encoded files.
djangorestframework-queryfields allows API clients to specify which fields will be sent in the response via inclusion/exclusion query parameters.
The drf-writable-nested package provides writable nested model serializer which allows to create/update models with nested related data.
404: Not Found 404: Not Found source: validators.py
Validators can be useful for re-using validation logic between different types of fields.
Most of the time you're dealing with validation in REST framework you'll simply be relying on the default field validation, or writing explicit validation methods on serializer or field classes.
However, sometimes you'll want to place your validation logic into reusable components, so that it can easily be reused throughout your codebase. This can be achieved by using validator functions and validator classes.
## Validation in REST framework
Validation in Django REST framework serializers is handled a little differently to how validation works in Django's ModelForm class.
With ModelForm the validation is performed partially on the form, and partially on the model instance. With REST framework the validation is performed entirely on the serializer class. This is advantageous for the following reasons:
- It introduces a proper separation of concerns, making your code behavior more obvious.
- It is easy to switch between using shortcut
ModelSerializerclasses and using explicitSerializerclasses. Any validation behavior being used forModelSerializeris simple to replicate. - Printing the
reprof a serializer instance will show you exactly what validation rules it applies. There's no extra hidden validation behavior being called on the model instance.
When you're using ModelSerializer all of this is handled automatically for you. If you want to drop down to using Serializer classes instead, then you need to define the validation rules explicitly.
As an example of how REST framework uses explicit validation, we'll take a simple model class that has a field with a uniqueness constraint.
class CustomerReportRecord(models.Model):
time_raised = models.DateTimeField(default=timezone.now, editable=False)
reference = models.CharField(unique=True, max_length=20)
description = models.TextField()
Here's a basic ModelSerializer that we can use for creating or updating instances of CustomerReportRecord:
class CustomerReportSerializer(serializers.ModelSerializer):
class Meta:
model = CustomerReportRecord
If we open up the Django shell using manage.py shell we can now
>>> from project.example.serializers import CustomerReportSerializer
>>> serializer = CustomerReportSerializer()
>>> print(repr(serializer))
CustomerReportSerializer():
id = IntegerField(label='ID', read_only=True)
time_raised = DateTimeField(read_only=True)
reference = CharField(max_length=20, validators=[<UniqueValidator(queryset=CustomerReportRecord.objects.all())>])
description = CharField(style={'type': 'textarea'})
The interesting bit here is the reference field. We can see that the uniqueness constraint is being explicitly enforced by a validator on the serializer field.
Because of this more explicit style REST framework includes a few validator classes that are not available in core Django. These classes are detailed below.
This validator can be used to enforce the unique=True constraint on model fields.
It takes a single required argument, and an optional messages argument:
querysetrequired - This is the queryset against which uniqueness should be enforced.message- The error message that should be used when validation fails.lookup- The lookup used to find an existing instance with the value being validated. Defaults to'exact'.
This validator should be applied to serializer fields, like so:
from rest_framework.validators import UniqueValidator
slug = SlugField(
max_length=100,
validators=[UniqueValidator(queryset=BlogPost.objects.all())]
)
## UniqueTogetherValidator
This validator can be used to enforce unique_together constraints on model instances.
It has two required arguments, and a single optional messages argument:
querysetrequired - This is the queryset against which uniqueness should be enforced.fieldsrequired - A list or tuple of field names which should make a unique set. These must exist as fields on the serializer class.message- The error message that should be used when validation fails.
The validator should be applied to serializer classes, like so:
from rest_framework.validators import UniqueTogetherValidator
class ExampleSerializer(serializers.Serializer):
# ...
class Meta:
# ToDo items belong to a parent list, and have an ordering defined
# by the 'position' field. No two items in a given list may share
# the same position.
validators = [
UniqueTogetherValidator(
queryset=ToDoItem.objects.all(),
fields=('list', 'position')
)
]
Note: The UniqueTogetherValidation class always imposes an implicit constraint that all the fields it applies to are always treated as required. Fields with default values are an exception to this as they always supply a value even when omitted from user input.
These validators can be used to enforce the unique_for_date, unique_for_month and unique_for_year constraints on model instances. They take the following arguments:
querysetrequired - This is the queryset against which uniqueness should be enforced.fieldrequired - A field name against which uniqueness in the given date range will be validated. This must exist as a field on the serializer class.date_fieldrequired - A field name which will be used to determine date range for the uniqueness constrain. This must exist as a field on the serializer class.message- The error message that should be used when validation fails.
The validator should be applied to serializer classes, like so:
from rest_framework.validators import UniqueForYearValidator
class ExampleSerializer(serializers.Serializer):
# ...
class Meta:
# Blog posts should have a slug that is unique for the current year.
validators = [
UniqueForYearValidator(
queryset=BlogPostItem.objects.all(),
field='slug',
date_field='published'
)
]
The date field that is used for the validation is always required to be present on the serializer class. You can't simply rely on a model class default=..., because the value being used for the default wouldn't be generated until after the validation has run.
There are a couple of styles you may want to use for this depending on how you want your API to behave. If you're using ModelSerializer you'll probably simply rely on the defaults that REST framework generates for you, but if you are using Serializer or simply want more explicit control, use on of the styles demonstrated below.
If you want the date field to be writable the only thing worth noting is that you should ensure that it is always available in the input data, either by setting a default argument, or by setting required=True.
published = serializers.DateTimeField(required=True)
If you want the date field to be visible, but not editable by the user, then set read_only=True and additionally set a default=... argument.
published = serializers.DateTimeField(read_only=True, default=timezone.now)
The field will not be writable to the user, but the default value will still be passed through to the validated_data.
Using with a hidden date field.
If you want the date field to be entirely hidden from the user, then use HiddenField. This field type does not accept user input, but instead always returns its default value to the validated_data in the serializer.
published = serializers.HiddenField(default=timezone.now)
Note: The UniqueFor<Range>Validation classes impose an implicit constraint that the fields they are applied to are always treated as required. Fields with default values are an exception to this as they always supply a value even when omitted from user input.
Validators that are applied across multiple fields in the serializer can sometimes require a field input that should not be provided by the API client, but that is available as input to the validator.
Two patterns that you may want to use for this sort of validation include:
- Using
HiddenField. This field will be present invalidated_databut will not be used in the serializer output representation. - Using a standard field with
read_only=True, but that also includes adefault=…argument. This field will be used in the serializer output representation, but cannot be set directly by the user.
REST framework includes a couple of defaults that may be useful in this context.
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()
)
#### CreateOnlyDefault
A default class that can be used to only set a default argument during create operations. During updates the field is omitted.
It takes a single argument, which is the default value or callable that should be used during create operations.
created_at = serializers.DateTimeField(
default=serializers.CreateOnlyDefault(timezone.now)
)
There are some ambiguous cases where you'll need to instead handle validation
explicitly, rather than relying on the default serializer classes that
ModelSerializer generates.
In these cases you may want to disable the automatically generated validators,
by specifying an empty list for the serializer Meta.validators attribute.
By default "unique together" validation enforces that all fields be
required=True. In some cases, you might want to explicit apply
required=False to one of the fields, in which case the desired behaviour
of the validation is ambiguous.
In this case you will typically need to exclude the validator from the
serializer class, and instead write any validation logic explicitly, either
in the .validate() method, or else in the view.
For example:
class BillingRecordSerializer(serializers.ModelSerializer):
def validate(self, data):
# Apply custom validation either here, or in the view.
class Meta:
fields = ('client', 'date', 'amount')
extra_kwargs = {'client': {'required': False}}
validators = [] # Remove a default "unique together" constraint.
When applying an update to an existing instance, uniqueness validators will
exclude the current instance from the uniqueness check. The current instance
is available in the context of the uniqueness check, because it exists as
an attribute on the serializer, having initially been passed using
instance=... when instantiating the serializer.
In the case of update operations on nested serializers there's no way of applying this exclusion, because the instance is not available.
Again, you'll probably want to explicitly remove the validator from the
serializer class, and write the code the for the validation constraint
explicitly, in a .validate() method, or in the view.
If you're not sure exactly what behavior a ModelSerializer class will
generate it is usually a good idea to run manage.py shell, and print
an instance of the serializer, so that you can inspect the fields and
validators that it automatically generates for you.
>>> serializer = MyComplexModelSerializer()
>>> print(serializer)
class MyComplexModelSerializer:
my_fields = ...
Also keep in mind that with complex cases it can often be better to explicitly
define your serializer classes, rather than relying on the default
ModelSerializer behavior. This involves a little more code, but ensures
that the resulting behavior is more transparent.
You can use any of Django's existing validators, or write your own custom validators.
A validator may be any callable that raises a serializers.ValidationError on failure.
def even_number(value):
if value % 2 != 0:
raise serializers.ValidationError('This field must be an even number.')
You can specify custom field-level validation by adding .validate_<field_name> methods
to your Serializer subclass. This is documented in the
Serializer docs
To write a class-based validator, use the __call__ method. Class-based validators are useful as they allow you to parameterize and reuse behavior.
class MultipleOf(object):
def __init__(self, base):
self.base = base
def __call__(self, value):
if value % self.base != 0:
message = 'This field must be a multiple of %d.' % self.base
raise serializers.ValidationError(message)
In some advanced cases you might want a validator to be passed the serializer field it is being used with as additional context. You can do so by declaring a set_context method on a class-based validator.
def set_context(self, serializer_field):
# Determine if this is an update or a create operation.
# In `__call__` we can then use that information to modify the validation behavior.
self.is_update = serializer_field.parent.instance is not None
source: authentication.py
Auth needs to be pluggable.
— Jacob Kaplan-Moss, "REST worst practices"
Authentication is the mechanism of associating an incoming request with a set of identifying credentials, such as the user the request came from, or the token that it was signed with. The permission and throttling policies can then use those credentials to determine if the request should be permitted.
REST framework provides a number of authentication schemes out of the box, and also allows you to implement custom schemes.
Authentication is always run at the very start of the view, before the permission and throttling checks occur, and before any other code is allowed to proceed.
The request.user property will typically be set to an instance of the contrib.auth package's User class.
The request.auth property is used for any additional authentication information, for example, it may be used to represent an authentication token that the request was signed with.
Note: Don't forget that authentication by itself won't allow or disallow an incoming request, it simply identifies the credentials that the request was made with.
For information on how to setup the permission polices for your API please see the permissions documentation.
The authentication schemes are always defined as a list of classes. REST framework will attempt to authenticate with each class in the list, and will set request.user and request.auth using the return value of the first class that successfully authenticates.
If no class authenticates, request.user will be set to an instance of django.contrib.auth.models.AnonymousUser, and request.auth will be set to None.
The value of request.user and request.auth for unauthenticated requests can be modified using the UNAUTHENTICATED_USER and UNAUTHENTICATED_TOKEN settings.
The default authentication schemes may be set globally, using the DEFAULT_AUTHENTICATION_CLASSES setting. For example.
REST_FRAMEWORK = {
'DEFAULT_AUTHENTICATION_CLASSES': (
'rest_framework.authentication.BasicAuthentication',
'rest_framework.authentication.SessionAuthentication',
)
}
You can also set the authentication scheme on a per-view or per-viewset basis,
using the APIView class-based views.
from rest_framework.authentication import SessionAuthentication, BasicAuthentication
from rest_framework.permissions import IsAuthenticated
from rest_framework.response import Response
from rest_framework.views import APIView
class ExampleView(APIView):
authentication_classes = (SessionAuthentication, BasicAuthentication)
permission_classes = (IsAuthenticated,)
def get(self, request, format=None):
content = {
'user': unicode(request.user), # `django.contrib.auth.User` instance.
'auth': unicode(request.auth), # None
}
return Response(content)
Or, if you're using the @api_view decorator with function based views.
@api_view(['GET'])
@authentication_classes((SessionAuthentication, BasicAuthentication))
@permission_classes((IsAuthenticated,))
def example_view(request, format=None):
content = {
'user': unicode(request.user), # `django.contrib.auth.User` instance.
'auth': unicode(request.auth), # None
}
return Response(content)
When an unauthenticated request is denied permission there are two different error codes that may be appropriate.
HTTP 401 responses must always include a WWW-Authenticate header, that instructs the client how to authenticate. HTTP 403 responses do not include the WWW-Authenticate header.
The kind of response that will be used depends on the authentication scheme. Although multiple authentication schemes may be in use, only one scheme may be used to determine the type of response. The first authentication class set on the view is used when determining the type of response.
Note that when a request may successfully authenticate, but still be denied permission to perform the request, in which case a 403 Permission Denied response will always be used, regardless of the authentication scheme.
Note that if deploying to Apache using mod_wsgi, the authorization header is not passed through to a WSGI application by default, as it is assumed that authentication will be handled by Apache, rather than at an application level.
If you are deploying to Apache, and using any non-session based authentication, you will need to explicitly configure mod_wsgi to pass the required headers through to the application. This can be done by specifying the WSGIPassAuthorization directive in the appropriate context and setting it to 'On'.
# this can go in either server config, virtual host, directory or .htaccess
WSGIPassAuthorization On
This authentication scheme uses HTTP Basic Authentication, signed against a user's username and password. Basic authentication is generally only appropriate for testing.
If successfully authenticated, BasicAuthentication provides the following credentials.
request.userwill be a DjangoUserinstance.request.authwill beNone.
Unauthenticated responses that are denied permission will result in an HTTP 401 Unauthorized response with an appropriate WWW-Authenticate header. For example:
WWW-Authenticate: Basic realm="api"
Note: If you use BasicAuthentication in production you must ensure that your API is only available over https. You should also ensure that your API clients will always re-request the username and password at login, and will never store those details to persistent storage.
This authentication scheme uses a simple token-based HTTP Authentication scheme. Token authentication is appropriate for client-server setups, such as native desktop and mobile clients.
To use the TokenAuthentication scheme you'll need to configure the authentication classes to include TokenAuthentication, and additionally include rest_framework.authtoken in your INSTALLED_APPS setting:
INSTALLED_APPS = (
...
'rest_framework.authtoken'
)
Note: Make sure to run manage.py migrate after changing your settings. The rest_framework.authtoken app provides Django database migrations.
You'll also need to create tokens for your users.
from rest_framework.authtoken.models import Token
token = Token.objects.create(user=...)
print token.key
For clients to authenticate, the token key should be included in the Authorization HTTP header. The key should be prefixed by the string literal "Token", with whitespace separating the two strings. For example:
Authorization: Token 9944b09199c62bcf9418ad846dd0e4bbdfc6ee4b
Note: If you want to use a different keyword in the header, such as Bearer, simply subclass TokenAuthentication and set the keyword class variable.
If successfully authenticated, TokenAuthentication provides the following credentials.
request.userwill be a DjangoUserinstance.request.authwill be arest_framework.authtoken.models.Tokeninstance.
Unauthenticated responses that are denied permission will result in an HTTP 401 Unauthorized response with an appropriate WWW-Authenticate header. For example:
WWW-Authenticate: Token
The curl command line tool may be useful for testing token authenticated APIs. For example:
curl -X GET http://127.0.0.1:8000/api/example/ -H 'Authorization: Token 9944b09199c62bcf9418ad846dd0e4bbdfc6ee4b'
Note: If you use TokenAuthentication in production you must ensure that your API is only available over https.
If you want every user to have an automatically generated Token, you can simply catch the User's post_save signal.
from django.conf import settings
from django.db.models.signals import post_save
from django.dispatch import receiver
from rest_framework.authtoken.models import Token
@receiver(post_save, sender=settings.AUTH_USER_MODEL)
def create_auth_token(sender, instance=None, created=False, **kwargs):
if created:
Token.objects.create(user=instance)
Note that you'll want to ensure you place this code snippet in an installed models.py module, or some other location that will be imported by Django on startup.
If you've already created some users, you can generate tokens for all existing users like this:
from django.contrib.auth.models import User
from rest_framework.authtoken.models import Token
for user in User.objects.all():
Token.objects.get_or_create(user=user)
When using TokenAuthentication, you may want to provide a mechanism for clients to obtain a token given the username and password. REST framework provides a built-in view to provide this behavior. To use it, add the obtain_auth_token view to your URLconf:
from rest_framework.authtoken import views
urlpatterns += [
url(r'^api-token-auth/', views.obtain_auth_token)
]
Note that the URL part of the pattern can be whatever you want to use.
The obtain_auth_token view will return a JSON response when valid username and password fields are POSTed to the view using form data or JSON:
{ 'token' : '9944b09199c62bcf9418ad846dd0e4bbdfc6ee4b' }
Note that the default obtain_auth_token view explicitly uses JSON requests and responses, rather than using default renderer and parser classes in your settings.
By default there are no permissions or throttling applied to the obtain_auth_token view. If you do wish to apply throttling you'll need to override the view class,
and include them using the throttle_classes attribute.
If you need a customized version of the obtain_auth_token view, you can do so by subclassing the ObtainAuthToken view class, and using that in your url conf instead.
For example, you may return additional user information beyond the token value:
from rest_framework.authtoken.views import ObtainAuthToken
from rest_framework.authtoken.models import Token
from rest_framework.response import Response
class CustomAuthToken(ObtainAuthToken):
def post(self, request, *args, **kwargs):
serializer = self.serializer_class(data=request.data,
context={'request': request})
serializer.is_valid(raise_exception=True)
user = serializer.validated_data['user']
token, created = Token.objects.get_or_create(user=user)
return Response({
'token': token.key,
'user_id': user.pk,
'email': user.email
})
And in your urls.py:
urlpatterns += [
url(r'^api-token-auth/', CustomAuthToken.as_view())
]
It is also possible to create Tokens manually through admin interface. In case you are using a large user base, we recommend that you monkey patch the TokenAdmin class to customize it to your needs, more specifically by declaring the user field as raw_field.
your_app/admin.py:
from rest_framework.authtoken.admin import TokenAdmin
TokenAdmin.raw_id_fields = ('user',)
Since version 3.6.4 it's possible to generate a user token using the following command:
./manage.py drf_create_token <username>
this command will return the API token for the given user, creating it if it doesn't exist:
Generated token 9944b09199c62bcf9418ad846dd0e4bbdfc6ee4b for user user1
In case you want to regenerate the token (for example if it has been compromised or leaked) you can pass an additional parameter:
./manage.py drf_create_token -r <username>
This authentication scheme uses Django's default session backend for authentication. Session authentication is appropriate for AJAX clients that are running in the same session context as your website.
If successfully authenticated, SessionAuthentication provides the following credentials.
request.userwill be a DjangoUserinstance.request.authwill beNone.
Unauthenticated responses that are denied permission will result in an HTTP 403 Forbidden response.
If you're using an AJAX style API with SessionAuthentication, you'll need to make sure you include a valid CSRF token for any "unsafe" HTTP method calls, such as PUT, PATCH, POST or DELETE requests. See the Django CSRF documentation for more details.
Warning: Always use Django's standard login view when creating login pages. This will ensure your login views are properly protected.
CSRF validation in REST framework works slightly differently to standard Django due to the need to support both session and non-session based authentication to the same views. This means that only authenticated requests require CSRF tokens, and anonymous requests may be sent without CSRF tokens. This behaviour is not suitable for login views, which should always have CSRF validation applied.
This authentication scheme allows you to delegate authentication to your web server, which sets the REMOTE_USER
environment variable.
To use it, you must have django.contrib.auth.backends.RemoteUserBackend (or a subclass) in your
AUTHENTICATION_BACKENDS setting. By default, RemoteUserBackend creates User objects for usernames that don't
already exist. To change this and other behaviour, consult the
Django documentation.
If successfully authenticated, RemoteUserAuthentication provides the following credentials:
request.userwill be a DjangoUserinstance.request.authwill beNone.
Consult your web server's documentation for information about configuring an authentication method, e.g.:
To implement a custom authentication scheme, subclass BaseAuthentication and override the .authenticate(self, request) method. The method should return a two-tuple of (user, auth) if authentication succeeds, or None otherwise.
In some circumstances instead of returning None, you may want to raise an AuthenticationFailed exception from the .authenticate() method.
Typically the approach you should take is:
- If authentication is not attempted, return
None. Any other authentication schemes also in use will still be checked. - If authentication is attempted but fails, raise a
AuthenticationFailedexception. An error response will be returned immediately, regardless of any permissions checks, and without checking any other authentication schemes.
You may also override the .authenticate_header(self, request) method. If implemented, it should return a string that will be used as the value of the WWW-Authenticate header in a HTTP 401 Unauthorized response.
If the .authenticate_header() method is not overridden, the authentication scheme will return HTTP 403 Forbidden responses when an unauthenticated request is denied access.
Note: When your custom authenticator is invoked by the request object's .user or .auth properties, you may see an AttributeError re-raised as a WrappedAttributeError. This is necessary to prevent the original exception from being suppressed by the outer property access. Python will not recognize that the AttributeError orginates from your custom authenticator and will instead assume that the request object does not have a .user or .auth property. These errors should be fixed or otherwise handled by your authenticator.
The following example will authenticate any incoming request as the user given by the username in a custom request header named 'X_USERNAME'.
from django.contrib.auth.models import User
from rest_framework import authentication
from rest_framework import exceptions
class ExampleAuthentication(authentication.BaseAuthentication):
def authenticate(self, request):
username = request.META.get('X_USERNAME')
if not username:
return None
try:
user = User.objects.get(username=username)
except User.DoesNotExist:
raise exceptions.AuthenticationFailed('No such user')
return (user, None)
The following third party packages are also available.
The Django OAuth Toolkit package provides OAuth 2.0 support, and works with Python 2.7 and Python 3.3+. The package is maintained by Evonove and uses the excellent OAuthLib. The package is well documented, and well supported and is currently our recommended package for OAuth 2.0 support.
Install using pip.
pip install django-oauth-toolkit
Add the package to your INSTALLED_APPS and modify your REST framework settings.
INSTALLED_APPS = (
...
'oauth2_provider',
)
REST_FRAMEWORK = {
'DEFAULT_AUTHENTICATION_CLASSES': (
'oauth2_provider.contrib.rest_framework.OAuth2Authentication',
)
}
For more details see the Django REST framework - Getting started documentation.
The Django REST framework OAuth package provides both OAuth1 and OAuth2 support for REST framework.
This package was previously included directly in REST framework but is now supported and maintained as a third party package.
Install the package using pip.
pip install djangorestframework-oauth
For details on configuration and usage see the Django REST framework OAuth documentation for authentication and permissions.
HTTP digest authentication is a widely implemented scheme that was intended to replace HTTP basic authentication, and which provides a simple encrypted authentication mechanism. Juan Riaza maintains the djangorestframework-digestauth package which provides HTTP digest authentication support for REST framework.
The Django OAuth2 Consumer library from Rediker Software is another package that provides OAuth 2.0 support for REST framework. The package includes token scoping permissions on tokens, which allows finer-grained access to your API.
JSON Web Token is a fairly new standard which can be used for token-based authentication. Unlike the built-in TokenAuthentication scheme, JWT Authentication doesn't need to use a database to validate a token. Blimp maintains the djangorestframework-jwt package which provides a JWT Authentication class as well as a mechanism for clients to obtain a JWT given the username and password. An alternative package for JWT authentication is djangorestframework-simplejwt which provides different features as well as a pluggable token blacklist app.
The HawkREST library builds on the Mohawk library to let you work with Hawk signed requests and responses in your API. Hawk lets two parties securely communicate with each other using messages signed by a shared key. It is based on HTTP MAC access authentication (which was based on parts of OAuth 1.0).
HTTP Signature (currently a IETF draft) provides a way to achieve origin authentication and message integrity for HTTP messages. Similar to Amazon's HTTP Signature scheme, used by many of its services, it permits stateless, per-request authentication. Elvio Toccalino maintains the djangorestframework-httpsignature package which provides an easy to use HTTP Signature Authentication mechanism.
Djoser library provides a set of views to handle basic actions such as registration, login, logout, password reset and account activation. The package works with a custom user model and it uses token based authentication. This is a ready to use REST implementation of Django authentication system.
Django-rest-auth library provides a set of REST API endpoints for registration, authentication (including social media authentication), password reset, retrieve and update user details, etc. By having these API endpoints, your client apps such as AngularJS, iOS, Android, and others can communicate to your Django backend site independently via REST APIs for user management.
Django-rest-framework-social-oauth2 library provides an easy way to integrate social plugins (facebook, twitter, google, etc.) to your authentication system and an easy oauth2 setup. With this library, you will be able to authenticate users based on external tokens (e.g. facebook access token), convert these tokens to "in-house" oauth2 tokens and use and generate oauth2 tokens to authenticate your users.
Django-rest-knox library provides models and views to handle token based authentication in a more secure and extensible way than the built-in TokenAuthentication scheme - with Single Page Applications and Mobile clients in mind. It provides per-client tokens, and views to generate them when provided some other authentication (usually basic authentication), to delete the token (providing a server enforced logout) and to delete all tokens (logs out all clients that a user is logged into).
drfpasswordless adds (Medium, Square Cash inspired) passwordless support to Django REST Framework's own TokenAuthentication scheme. Users log in and sign up with a token sent to a contact point like an email address or a mobile number.
source: permissions.py
Authentication or identification by itself is not usually sufficient to gain access to information or code. For that, the entity requesting access must have authorization.
Together with authentication and throttling, permissions determine whether a request should be granted or denied access.
Permission checks are always run at the very start of the view, before any other code is allowed to proceed. Permission checks will typically use the authentication information in the request.user and request.auth properties to determine if the incoming request should be permitted.
Permissions are used to grant or deny access different classes of users to different parts of the API.
The simplest style of permission would be to allow access to any authenticated user, and deny access to any unauthenticated user. This corresponds the IsAuthenticated class in REST framework.
A slightly less strict style of permission would be to allow full access to authenticated users, but allow read-only access to unauthenticated users. This corresponds to the IsAuthenticatedOrReadOnly class in REST framework.
Permissions in REST framework are always defined as a list of permission classes.
Before running the main body of the view each permission in the list is checked.
If any permission check fails an exceptions.PermissionDenied or exceptions.NotAuthenticated exception will be raised, and the main body of the view will not run.
When the permissions checks fail either a "403 Forbidden" or a "401 Unauthorized" response will be returned, according to the following rules:
- The request was successfully authenticated, but permission was denied. — An HTTP 403 Forbidden response will be returned.
- The request was not successfully authenticated, and the highest priority authentication class does not use
WWW-Authenticateheaders. — An HTTP 403 Forbidden response will be returned. - The request was not successfully authenticated, and the highest priority authentication class does use
WWW-Authenticateheaders. — An HTTP 401 Unauthorized response, with an appropriateWWW-Authenticateheader will be returned.
REST framework permissions also support object-level permissioning. Object level permissions are used to determine if a user should be allowed to act on a particular object, which will typically be a model instance.
Object level permissions are run by REST framework's generic views when .get_object() is called.
As with view level permissions, an exceptions.PermissionDenied exception will be raised if the user is not allowed to act on the given object.
If you're writing your own views and want to enforce object level permissions,
or if you override the get_object method on a generic view, then you'll need to explicitly call the .check_object_permissions(request, obj) method on the view at the point at which you've retrieved the object.
This will either raise a PermissionDenied or NotAuthenticated exception, or simply return if the view has the appropriate permissions.
For example:
def get_object(self):
obj = get_object_or_404(self.get_queryset(), pk=self.kwargs["pk"])
self.check_object_permissions(self.request, obj)
return obj
For performance reasons the generic views will not automatically apply object level permissions to each instance in a queryset when returning a list of objects.
Often when you're using object level permissions you'll also want to filter the queryset appropriately, to ensure that users only have visibility onto instances that they are permitted to view.
The default permission policy may be set globally, using the DEFAULT_PERMISSION_CLASSES setting. For example.
REST_FRAMEWORK = {
'DEFAULT_PERMISSION_CLASSES': (
'rest_framework.permissions.IsAuthenticated',
)
}
If not specified, this setting defaults to allowing unrestricted access:
'DEFAULT_PERMISSION_CLASSES': (
'rest_framework.permissions.AllowAny',
)
You can also set the authentication policy on a per-view, or per-viewset basis,
using the APIView class-based views.
from rest_framework.permissions import IsAuthenticated
from rest_framework.response import Response
from rest_framework.views import APIView
class ExampleView(APIView):
permission_classes = (IsAuthenticated,)
def get(self, request, format=None):
content = {
'status': 'request was permitted'
}
return Response(content)
Or, if you're using the @api_view decorator with function based views.
from rest_framework.decorators import api_view, permission_classes
from rest_framework.permissions import IsAuthenticated
from rest_framework.response import Response
@api_view(['GET'])
@permission_classes((IsAuthenticated, ))
def example_view(request, format=None):
content = {
'status': 'request was permitted'
}
return Response(content)
Note: when you set new permission classes through class attribute or decorators you're telling the view to ignore the default list set over the settings.py file.
The AllowAny permission class will allow unrestricted access, regardless of if the request was authenticated or unauthenticated.
This permission is not strictly required, since you can achieve the same result by using an empty list or tuple for the permissions setting, but you may find it useful to specify this class because it makes the intention explicit.
The IsAuthenticated permission class will deny permission to any unauthenticated user, and allow permission otherwise.
This permission is suitable if you want your API to only be accessible to registered users.
The IsAdminUser permission class will deny permission to any user, unless user.is_staff is True in which case permission will be allowed.
This permission is suitable if you want your API to only be accessible to a subset of trusted administrators.
The IsAuthenticatedOrReadOnly will allow authenticated users to perform any request. Requests for unauthorised users will only be permitted if the request method is one of the "safe" methods; GET, HEAD or OPTIONS.
This permission is suitable if you want to your API to allow read permissions to anonymous users, and only allow write permissions to authenticated users.
This permission class ties into Django's standard django.contrib.auth model permissions. This permission must only be applied to views that have a .queryset property set. Authorization will only be granted if the user is authenticated and has the relevant model permissions assigned.
POSTrequests require the user to have theaddpermission on the model.PUTandPATCHrequests require the user to have thechangepermission on the model.DELETErequests require the user to have thedeletepermission on the model.
The default behaviour can also be overridden to support custom model permissions. For example, you might want to include a view model permission for GET requests.
To use custom model permissions, override DjangoModelPermissions and set the .perms_map property. Refer to the source code for details.
If you're using this permission with a view that uses an overridden get_queryset() method there may not be a queryset attribute on the view. In this case we suggest also marking the view with a sentinel queryset, so that this class can determine the required permissions. For example:
queryset = User.objects.none() # Required for DjangoModelPermissions
Similar to DjangoModelPermissions, but also allows unauthenticated users to have read-only access to the API.
## DjangoObjectPermissions
This permission class ties into Django's standard object permissions framework that allows per-object permissions on models. In order to use this permission class, you'll also need to add a permission backend that supports object-level permissions, such as django-guardian.
As with DjangoModelPermissions, this permission must only be applied to views that have a .queryset property or .get_queryset() method. Authorization will only be granted if the user is authenticated and has the relevant per-object permissions and relevant model permissions assigned.
POSTrequests require the user to have theaddpermission on the model instance.PUTandPATCHrequests require the user to have thechangepermission on the model instance.DELETErequests require the user to have thedeletepermission on the model instance.
Note that DjangoObjectPermissions does not require the django-guardian package, and should support other object-level backends equally well.
As with DjangoModelPermissions you can use custom model permissions by overriding DjangoObjectPermissions and setting the .perms_map property. Refer to the source code for details.
Note: If you need object level view permissions for GET, HEAD and OPTIONS requests, you'll want to consider also adding the DjangoObjectPermissionsFilter class to ensure that list endpoints only return results including objects for which the user has appropriate view permissions.
To implement a custom permission, override BasePermission and implement either, or both, of the following methods:
.has_permission(self, request, view).has_object_permission(self, request, view, obj)
The methods should return True if the request should be granted access, and False otherwise.
If you need to test if a request is a read operation or a write operation, you should check the request method against the constant SAFE_METHODS, which is a tuple containing 'GET', 'OPTIONS' and 'HEAD'. For example:
if request.method in permissions.SAFE_METHODS:
# Check permissions for read-only request
else:
# Check permissions for write request
Note: The instance-level has_object_permission method will only be called if the view-level has_permission checks have already passed. Also note that in order for the instance-level checks to run, the view code should explicitly call .check_object_permissions(request, obj). If you are using the generic views then this will be handled for you by default. (Function-based views will need to check object permissions explicitly, raising PermissionDenied on failure.)
Custom permissions will raise a PermissionDenied exception if the test fails. To change the error message associated with the exception, implement a message attribute directly on your custom permission. Otherwise the default_detail attribute from PermissionDenied will be used.
from rest_framework import permissions
class CustomerAccessPermission(permissions.BasePermission):
message = 'Adding customers not allowed.'
def has_permission(self, request, view):
...
The following is an example of a permission class that checks the incoming request's IP address against a blacklist, and denies the request if the IP has been blacklisted.
from rest_framework import permissions
class BlacklistPermission(permissions.BasePermission):
"""
Global permission check for blacklisted IPs.
"""
def has_permission(self, request, view):
ip_addr = request.META['REMOTE_ADDR']
blacklisted = Blacklist.objects.filter(ip_addr=ip_addr).exists()
return not blacklisted
As well as global permissions, that are run against all incoming requests, you can also create object-level permissions, that are only run against operations that affect a particular object instance. For example:
class IsOwnerOrReadOnly(permissions.BasePermission):
"""
Object-level permission to only allow owners of an object to edit it.
Assumes the model instance has an `owner` attribute.
"""
def has_object_permission(self, request, view, obj):
# Read permissions are allowed to any request,
# so we'll always allow GET, HEAD or OPTIONS requests.
if request.method in permissions.SAFE_METHODS:
return True
# Instance must have an attribute named `owner`.
return obj.owner == request.user
Note that the generic views will check the appropriate object level permissions, but if you're writing your own custom views, you'll need to make sure you check the object level permission checks yourself. You can do so by calling self.check_object_permissions(request, obj) from the view once you have the object instance. This call will raise an appropriate APIException if any object-level permission checks fail, and will otherwise simply return.
Also note that the generic views will only check the object-level permissions for views that retrieve a single model instance. If you require object-level filtering of list views, you'll need to filter the queryset separately. See the filtering documentation for more details.
The following third party packages are also available.
The Composed Permissions package provides a simple way to define complex and multi-depth (with logic operators) permission objects, using small and reusable components.
The REST Condition package is another extension for building complex permissions in a simple and convenient way. The extension allows you to combine permissions with logical operators.
The DRY Rest Permissions package provides the ability to define different permissions for individual default and custom actions. This package is made for apps with permissions that are derived from relationships defined in the app's data model. It also supports permission checks being returned to a client app through the API's serializer. Additionally it supports adding permissions to the default and custom list actions to restrict the data they retrieve per user.
The Django Rest Framework Roles package makes it easier to parameterize your API over multiple types of users.
The Django Rest Framework API Key package allows you to ensure that every request made to the server requires an API key header. You can generate one from the django admin interface.
The Django Rest Framework Role Filters package provides simple filtering over multiple types of roles.
source: throttling.py
HTTP/1.1 420 Enhance Your Calm
Throttling is similar to permissions, in that it determines if a request should be authorized. Throttles indicate a temporary state, and are used to control the rate of requests that clients can make to an API.
As with permissions, multiple throttles may be used. Your API might have a restrictive throttle for unauthenticated requests, and a less restrictive throttle for authenticated requests.
Another scenario where you might want to use multiple throttles would be if you need to impose different constraints on different parts of the API, due to some services being particularly resource-intensive.
Multiple throttles can also be used if you want to impose both burst throttling rates, and sustained throttling rates. For example, you might want to limit a user to a maximum of 60 requests per minute, and 1000 requests per day.
Throttles do not necessarily only refer to rate-limiting requests. For example a storage service might also need to throttle against bandwidth, and a paid data service might want to throttle against a certain number of a records being accessed.
As with permissions and authentication, throttling in REST framework is always defined as a list of classes.
Before running the main body of the view each throttle in the list is checked.
If any throttle check fails an exceptions.Throttled exception will be raised, and the main body of the view will not run.
The default throttling policy may be set globally, using the DEFAULT_THROTTLE_CLASSES and DEFAULT_THROTTLE_RATES settings. For example.
REST_FRAMEWORK = {
'DEFAULT_THROTTLE_CLASSES': (
'rest_framework.throttling.AnonRateThrottle',
'rest_framework.throttling.UserRateThrottle'
),
'DEFAULT_THROTTLE_RATES': {
'anon': '100/day',
'user': '1000/day'
}
}
The rate descriptions used in DEFAULT_THROTTLE_RATES may include second, minute, hour or day as the throttle period.
You can also set the throttling policy on a per-view or per-viewset basis,
using the APIView class-based views.
from rest_framework.response import Response
from rest_framework.throttling import UserRateThrottle
from rest_framework.views import APIView
class ExampleView(APIView):
throttle_classes = (UserRateThrottle,)
def get(self, request, format=None):
content = {
'status': 'request was permitted'
}
return Response(content)
Or, if you're using the @api_view decorator with function based views.
@api_view(['GET'])
@throttle_classes([UserRateThrottle])
def example_view(request, format=None):
content = {
'status': 'request was permitted'
}
return Response(content)
## How clients are identified
The X-Forwarded-For HTTP header and REMOTE_ADDR WSGI variable are used to uniquely identify client IP addresses for throttling. If the X-Forwarded-For header is present then it will be used, otherwise the value of the REMOTE_ADDR variable from the WSGI environment will be used.
If you need to strictly identify unique client IP addresses, you'll need to first configure the number of application proxies that the API runs behind by setting the NUM_PROXIES setting. This setting should be an integer of zero or more. If set to non-zero then the client IP will be identified as being the last IP address in the X-Forwarded-For header, once any application proxy IP addresses have first been excluded. If set to zero, then the REMOTE_ADDR value will always be used as the identifying IP address.
It is important to understand that if you configure the NUM_PROXIES setting, then all clients behind a unique NAT'd gateway will be treated as a single client.
Further context on how the X-Forwarded-For header works, and identifying a remote client IP can be found here.
The throttle classes provided by REST framework use Django's cache backend. You should make sure that you've set appropriate cache settings. The default value of LocMemCache backend should be okay for simple setups. See Django's cache documentation for more details.
If you need to use a cache other than 'default', you can do so by creating a custom throttle class and setting the cache attribute. For example:
class CustomAnonRateThrottle(AnonRateThrottle):
cache = get_cache('alternate')
You'll need to remember to also set your custom throttle class in the 'DEFAULT_THROTTLE_CLASSES' settings key, or using the throttle_classes view attribute.
The AnonRateThrottle will only ever throttle unauthenticated users. The IP address of the incoming request is used to generate a unique key to throttle against.
The allowed request rate is determined from one of the following (in order of preference).
- The
rateproperty on the class, which may be provided by overridingAnonRateThrottleand setting the property. - The
DEFAULT_THROTTLE_RATES['anon']setting.
AnonRateThrottle is suitable if you want to restrict the rate of requests from unknown sources.
The UserRateThrottle will throttle users to a given rate of requests across the API. The user id is used to generate a unique key to throttle against. Unauthenticated requests will fall back to using the IP address of the incoming request to generate a unique key to throttle against.
The allowed request rate is determined from one of the following (in order of preference).
- The
rateproperty on the class, which may be provided by overridingUserRateThrottleand setting the property. - The
DEFAULT_THROTTLE_RATES['user']setting.
An API may have multiple UserRateThrottles in place at the same time. To do so, override UserRateThrottle and set a unique "scope" for each class.
For example, multiple user throttle rates could be implemented by using the following classes...
class BurstRateThrottle(UserRateThrottle):
scope = 'burst'
class SustainedRateThrottle(UserRateThrottle):
scope = 'sustained'
...and the following settings.
REST_FRAMEWORK = {
'DEFAULT_THROTTLE_CLASSES': (
'example.throttles.BurstRateThrottle',
'example.throttles.SustainedRateThrottle'
),
'DEFAULT_THROTTLE_RATES': {
'burst': '60/min',
'sustained': '1000/day'
}
}
UserRateThrottle is suitable if you want simple global rate restrictions per-user.
The ScopedRateThrottle class can be used to restrict access to specific parts of the API. This throttle will only be applied if the view that is being accessed includes a .throttle_scope property. The unique throttle key will then be formed by concatenating the "scope" of the request with the unique user id or IP address.
The allowed request rate is determined by the DEFAULT_THROTTLE_RATES setting using a key from the request "scope".
For example, given the following views...
class ContactListView(APIView):
throttle_scope = 'contacts'
...
class ContactDetailView(APIView):
throttle_scope = 'contacts'
...
class UploadView(APIView):
throttle_scope = 'uploads'
...
...and the following settings.
REST_FRAMEWORK = {
'DEFAULT_THROTTLE_CLASSES': (
'rest_framework.throttling.ScopedRateThrottle',
),
'DEFAULT_THROTTLE_RATES': {
'contacts': '1000/day',
'uploads': '20/day'
}
}
User requests to either ContactListView or ContactDetailView would be restricted to a total of 1000 requests per-day. User requests to UploadView would be restricted to 20 requests per day.
To create a custom throttle, override BaseThrottle and implement .allow_request(self, request, view). The method should return True if the request should be allowed, and False otherwise.
Optionally you may also override the .wait() method. If implemented, .wait() should return a recommended number of seconds to wait before attempting the next request, or None. The .wait() method will only be called if .allow_request() has previously returned False.
If the .wait() method is implemented and the request is throttled, then a Retry-After header will be included in the response.
The following is an example of a rate throttle, that will randomly throttle 1 in every 10 requests.
import random
class RandomRateThrottle(throttling.BaseThrottle):
def allow_request(self, request, view):
return random.randint(1, 10) != 1
source: filters.py
The root QuerySet provided by the Manager describes all objects in the database table. Usually, though, you'll need to select only a subset of the complete set of objects.
The default behavior of REST framework's generic list views is to return the entire queryset for a model manager. Often you will want your API to restrict the items that are returned by the queryset.
The simplest way to filter the queryset of any view that subclasses GenericAPIView is to override the .get_queryset() method.
Overriding this method allows you to customize the queryset returned by the view in a number of different ways.
You might want to filter the queryset to ensure that only results relevant to the currently authenticated user making the request are returned.
You can do so by filtering based on the value of request.user.
For example:
from myapp.models import Purchase
from myapp.serializers import PurchaseSerializer
from rest_framework import generics
class PurchaseList(generics.ListAPIView):
serializer_class = PurchaseSerializer
def get_queryset(self):
"""
This view should return a list of all the purchases
for the currently authenticated user.
"""
user = self.request.user
return Purchase.objects.filter(purchaser=user)
Another style of filtering might involve restricting the queryset based on some part of the URL.
For example if your URL config contained an entry like this:
url('^purchases/(?P<username>.+)/$', PurchaseList.as_view()),
You could then write a view that returned a purchase queryset filtered by the username portion of the URL:
class PurchaseList(generics.ListAPIView):
serializer_class = PurchaseSerializer
def get_queryset(self):
"""
This view should return a list of all the purchases for
the user as determined by the username portion of the URL.
"""
username = self.kwargs['username']
return Purchase.objects.filter(purchaser__username=username)
A final example of filtering the initial queryset would be to determine the initial queryset based on query parameters in the url.
We can override .get_queryset() to deal with URLs such as http://example.com/api/purchases?username=denvercoder9, and filter the queryset only if the username parameter is included in the URL:
class PurchaseList(generics.ListAPIView):
serializer_class = PurchaseSerializer
def get_queryset(self):
"""
Optionally restricts the returned purchases to a given user,
by filtering against a `username` query parameter in the URL.
"""
queryset = Purchase.objects.all()
username = self.request.query_params.get('username', None)
if username is not None:
queryset = queryset.filter(purchaser__username=username)
return queryset
As well as being able to override the default queryset, REST framework also includes support for generic filtering backends that allow you to easily construct complex searches and filters.
Generic filters can also present themselves as HTML controls in the browsable API and admin API.
The default filter backends may be set globally, using the DEFAULT_FILTER_BACKENDS setting. For example.
REST_FRAMEWORK = {
'DEFAULT_FILTER_BACKENDS': ('django_filters.rest_framework.DjangoFilterBackend',)
}
You can also set the filter backends on a per-view, or per-viewset basis,
using the GenericAPIView class-based views.
import django_filters.rest_framework
from django.contrib.auth.models import User
from myapp.serializers import UserSerializer
from rest_framework import generics
class UserListView(generics.ListAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
filter_backends = (django_filters.rest_framework.DjangoFilterBackend,)
Note that if a filter backend is configured for a view, then as well as being used to filter list views, it will also be used to filter the querysets used for returning a single object.
For instance, given the previous example, and a product with an id of 4675, the following URL would either return the corresponding object, or return a 404 response, depending on if the filtering conditions were met by the given product instance:
http://example.com/api/products/4675/?category=clothing&max_price=10.00
Note that you can use both an overridden .get_queryset() and generic filtering together, and everything will work as expected. For example, if Product had a many-to-many relationship with User, named purchase, you might want to write a view like this:
class PurchasedProductsList(generics.ListAPIView):
"""
Return a list of all the products that the authenticated
user has ever purchased, with optional filtering.
"""
model = Product
serializer_class = ProductSerializer
filter_class = ProductFilter
def get_queryset(self):
user = self.request.user
return user.purchase_set.all()
The django-filter library includes a DjangoFilterBackend class which
supports highly customizable field filtering for REST framework.
To use DjangoFilterBackend, first install django-filter. Then add django_filters to Django's INSTALLED_APPS
pip install django-filter
You should now either add the filter backend to your settings:
REST_FRAMEWORK = {
'DEFAULT_FILTER_BACKENDS': ('django_filters.rest_framework.DjangoFilterBackend',)
}
Or add the filter backend to an individual View or ViewSet.
from django_filters.rest_framework import DjangoFilterBackend
class UserListView(generics.ListAPIView):
...
filter_backends = (DjangoFilterBackend,)
If all you need is simple equality-based filtering, you can set a filter_fields attribute on the view, or viewset, listing the set of fields you wish to filter against.
class ProductList(generics.ListAPIView):
queryset = Product.objects.all()
serializer_class = ProductSerializer
filter_backends = (DjangoFilterBackend,)
filter_fields = ('category', 'in_stock')
This will automatically create a FilterSet class for the given fields, and will allow you to make requests such as:
http://example.com/api/products?category=clothing&in_stock=True
For more advanced filtering requirements you can specify a FilterSet class that should be used by the view.
You can read more about FilterSets in the django-filter documentation.
It's also recommended that you read the section on DRF integration.
The SearchFilter class supports simple single query parameter based searching, and is based on the Django admin's search functionality.
When in use, the browsable API will include a SearchFilter control:
The SearchFilter class will only be applied if the view has a search_fields attribute set. The search_fields attribute should be a list of names of text type fields on the model, such as CharField or TextField.
class UserListView(generics.ListAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
filter_backends = (filters.SearchFilter,)
search_fields = ('username', 'email')
This will allow the client to filter the items in the list by making queries such as:
http://example.com/api/users?search=russell
You can also perform a related lookup on a ForeignKey or ManyToManyField with the lookup API double-underscore notation:
search_fields = ('username', 'email', 'profile__profession')
By default, searches will use case-insensitive partial matches. The search parameter may contain multiple search terms, which should be whitespace and/or comma separated. If multiple search terms are used then objects will be returned in the list only if all the provided terms are matched.
The search behavior may be restricted by prepending various characters to the search_fields.
- '^' Starts-with search.
- '=' Exact matches.
- '@' Full-text search. (Currently only supported Django's MySQL backend.)
- '$' Regex search.
For example:
search_fields = ('=username', '=email')
By default, the search parameter is named 'search', but this may be overridden with the SEARCH_PARAM setting.
For more details, see the Django documentation.
The OrderingFilter class supports simple query parameter controlled ordering of results.
By default, the query parameter is named 'ordering', but this may by overridden with the ORDERING_PARAM setting.
For example, to order users by username:
http://example.com/api/users?ordering=username
The client may also specify reverse orderings by prefixing the field name with '-', like so:
http://example.com/api/users?ordering=-username
Multiple orderings may also be specified:
http://example.com/api/users?ordering=account,username
It's recommended that you explicitly specify which fields the API should allowing in the ordering filter. You can do this by setting an ordering_fields attribute on the view, like so:
class UserListView(generics.ListAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
filter_backends = (filters.OrderingFilter,)
ordering_fields = ('username', 'email')
This helps prevent unexpected data leakage, such as allowing users to order against a password hash field or other sensitive data.
If you don't specify an ordering_fields attribute on the view, the filter class will default to allowing the user to filter on any readable fields on the serializer specified by the serializer_class attribute.
If you are confident that the queryset being used by the view doesn't contain any sensitive data, you can also explicitly specify that a view should allow ordering on any model field or queryset aggregate, by using the special value '__all__'.
class BookingsListView(generics.ListAPIView):
queryset = Booking.objects.all()
serializer_class = BookingSerializer
filter_backends = (filters.OrderingFilter,)
ordering_fields = '__all__'
If an ordering attribute is set on the view, this will be used as the default ordering.
Typically you'd instead control this by setting order_by on the initial queryset, but using the ordering parameter on the view allows you to specify the ordering in a way that it can then be passed automatically as context to a rendered template. This makes it possible to automatically render column headers differently if they are being used to order the results.
class UserListView(generics.ListAPIView):
queryset = User.objects.all()
serializer_class = UserSerializer
filter_backends = (filters.OrderingFilter,)
ordering_fields = ('username', 'email')
ordering = ('username',)
The ordering attribute may be either a string or a list/tuple of strings.
The DjangoObjectPermissionsFilter is intended to be used together with the django-guardian package, with custom 'view' permissions added. The filter will ensure that querysets only returns objects for which the user has the appropriate view permission.
If you're using DjangoObjectPermissionsFilter, you'll probably also want to add an appropriate object permissions class, to ensure that users can only operate on instances if they have the appropriate object permissions. The easiest way to do this is to subclass DjangoObjectPermissions and add 'view' permissions to the perms_map attribute.
A complete example using both DjangoObjectPermissionsFilter and DjangoObjectPermissions might look something like this.
permissions.py:
class CustomObjectPermissions(permissions.DjangoObjectPermissions):
"""
Similar to `DjangoObjectPermissions`, but adding 'view' permissions.
"""
perms_map = {
'GET': ['%(app_label)s.view_%(model_name)s'],
'OPTIONS': ['%(app_label)s.view_%(model_name)s'],
'HEAD': ['%(app_label)s.view_%(model_name)s'],
'POST': ['%(app_label)s.add_%(model_name)s'],
'PUT': ['%(app_label)s.change_%(model_name)s'],
'PATCH': ['%(app_label)s.change_%(model_name)s'],
'DELETE': ['%(app_label)s.delete_%(model_name)s'],
}
views.py:
class EventViewSet(viewsets.ModelViewSet):
"""
Viewset that only lists events if user has 'view' permissions, and only
allows operations on individual events if user has appropriate 'view', 'add',
'change' or 'delete' permissions.
"""
queryset = Event.objects.all()
serializer_class = EventSerializer
filter_backends = (filters.DjangoObjectPermissionsFilter,)
permission_classes = (myapp.permissions.CustomObjectPermissions,)
For more information on adding 'view' permissions for models, see the relevant section of the django-guardian documentation, and this blogpost.
You can also provide your own generic filtering backend, or write an installable app for other developers to use.
To do so override BaseFilterBackend, and override the .filter_queryset(self, request, queryset, view) method. The method should return a new, filtered queryset.
As well as allowing clients to perform searches and filtering, generic filter backends can be useful for restricting which objects should be visible to any given request or user.
For example, you might need to restrict users to only being able to see objects they created.
class IsOwnerFilterBackend(filters.BaseFilterBackend):
"""
Filter that only allows users to see their own objects.
"""
def filter_queryset(self, request, queryset, view):
return queryset.filter(owner=request.user)
We could achieve the same behavior by overriding get_queryset() on the views, but using a filter backend allows you to more easily add this restriction to multiple views, or to apply it across the entire API.
Generic filters may also present an interface in the browsable API. To do so you should implement a to_html() method which returns a rendered HTML representation of the filter. This method should have the following signature:
to_html(self, request, queryset, view)
The method should return a rendered HTML string.
You can also make the filter controls available to the schema autogeneration
that REST framework provides, by implementing a get_schema_fields() method. This method should have the following signature:
get_schema_fields(self, view)
The method should return a list of coreapi.Field instances.
The following third party packages provide additional filter implementations.
The django-rest-framework-filters package works together with the DjangoFilterBackend class, and allows you to easily create filters across relationships, or create multiple filter lookup types for a given field.
The djangorestframework-word-filter developed as alternative to filters.SearchFilter which will search full word in text, or exact match.
django-url-filter provides a safe way to filter data via human-friendly URLs. It works very similar to DRF serializers and fields in a sense that they can be nested except they are called filtersets and filters. That provides easy way to filter related data. Also this library is generic-purpose so it can be used to filter other sources of data and not only Django QuerySets.
drf-url-filter is a simple Django app to apply filters on drf ModelViewSet's Queryset in a clean, simple and configurable way. It also supports validations on incoming query params and their values. A beautiful python package Voluptuous is being used for validations on the incoming query parameters. The best part about voluptuous is you can define your own validations as per your query params requirements.
source: pagination.py
Django provides a few classes that help you manage paginated data – that is, data that’s split across several pages, with “Previous/Next” links.
REST framework includes support for customizable pagination styles. This allows you to modify how large result sets are split into individual pages of data.
The pagination API can support either:
- Pagination links that are provided as part of the content of the response.
- Pagination links that are included in response headers, such as
Content-RangeorLink.
The built-in styles currently all use links included as part of the content of the response. This style is more accessible when using the browsable API.
Pagination is only performed automatically if you're using the generic views or viewsets. If you're using a regular APIView, you'll need to call into the pagination API yourself to ensure you return a paginated response. See the source code for the mixins.ListModelMixin and generics.GenericAPIView classes for an example.
Pagination can be turned off by setting the pagination class to None.
The pagination style may be set globally, using the DEFAULT_PAGINATION_CLASS and PAGE_SIZE setting keys. For example, to use the built-in limit/offset pagination, you would do something like this:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'rest_framework.pagination.LimitOffsetPagination',
'PAGE_SIZE': 100
}
Note that you need to set both the pagination class, and the page size that should be used. Both DEFAULT_PAGINATION_CLASS and PAGE_SIZE are None by default.
You can also set the pagination class on an individual view by using the pagination_class attribute. Typically you'll want to use the same pagination style throughout your API, although you might want to vary individual aspects of the pagination, such as default or maximum page size, on a per-view basis.
If you want to modify particular aspects of the pagination style, you'll want to override one of the pagination classes, and set the attributes that you want to change.
class LargeResultsSetPagination(PageNumberPagination):
page_size = 1000
page_size_query_param = 'page_size'
max_page_size = 10000
class StandardResultsSetPagination(PageNumberPagination):
page_size = 100
page_size_query_param = 'page_size'
max_page_size = 1000
You can then apply your new style to a view using the .pagination_class attribute:
class BillingRecordsView(generics.ListAPIView):
queryset = Billing.objects.all()
serializer_class = BillingRecordsSerializer
pagination_class = LargeResultsSetPagination
Or apply the style globally, using the DEFAULT_PAGINATION_CLASS settings key. For example:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'apps.core.pagination.StandardResultsSetPagination'
}
This pagination style accepts a single number page number in the request query parameters.
Request:
GET https://api.example.org/accounts/?page=4
Response:
HTTP 200 OK
{
"count": 1023
"next": "https://api.example.org/accounts/?page=5",
"previous": "https://api.example.org/accounts/?page=3",
"results": [
…
]
}
To enable the PageNumberPagination style globally, use the following configuration, and set the PAGE_SIZE as desired:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'rest_framework.pagination.PageNumberPagination',
'PAGE_SIZE': 100
}
On GenericAPIView subclasses you may also set the pagination_class attribute to select PageNumberPagination on a per-view basis.
The PageNumberPagination class includes a number of attributes that may be overridden to modify the pagination style.
To set these attributes you should override the PageNumberPagination class, and then enable your custom pagination class as above.
django_paginator_class- The Django Paginator class to use. Default isdjango.core.paginator.Paginator, which should be fine for most use cases.page_size- A numeric value indicating the page size. If set, this overrides thePAGE_SIZEsetting. Defaults to the same value as thePAGE_SIZEsettings key.page_query_param- A string value indicating the name of the query parameter to use for the pagination control.page_size_query_param- If set, this is a string value indicating the name of a query parameter that allows the client to set the page size on a per-request basis. Defaults toNone, indicating that the client may not control the requested page size.max_page_size- If set, this is a numeric value indicating the maximum allowable requested page size. This attribute is only valid ifpage_size_query_paramis also set.last_page_strings- A list or tuple of string values indicating values that may be used with thepage_query_paramto request the final page in the set. Defaults to('last',)template- The name of a template to use when rendering pagination controls in the browsable API. May be overridden to modify the rendering style, or set toNoneto disable HTML pagination controls completely. Defaults to"rest_framework/pagination/numbers.html".
This pagination style mirrors the syntax used when looking up multiple database records. The client includes both a "limit" and an
"offset" query parameter. The limit indicates the maximum number of items to return, and is equivalent to the page_size in other styles. The offset indicates the starting position of the query in relation to the complete set of unpaginated items.
Request:
GET https://api.example.org/accounts/?limit=100&offset=400
Response:
HTTP 200 OK
{
"count": 1023
"next": "https://api.example.org/accounts/?limit=100&offset=500",
"previous": "https://api.example.org/accounts/?limit=100&offset=300",
"results": [
…
]
}
To enable the LimitOffsetPagination style globally, use the following configuration:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'rest_framework.pagination.LimitOffsetPagination'
}
Optionally, you may also set a PAGE_SIZE key. If the PAGE_SIZE parameter is also used then the limit query parameter will be optional, and may be omitted by the client.
On GenericAPIView subclasses you may also set the pagination_class attribute to select LimitOffsetPagination on a per-view basis.
The LimitOffsetPagination class includes a number of attributes that may be overridden to modify the pagination style.
To set these attributes you should override the LimitOffsetPagination class, and then enable your custom pagination class as above.
default_limit- A numeric value indicating the limit to use if one is not provided by the client in a query parameter. Defaults to the same value as thePAGE_SIZEsettings key.limit_query_param- A string value indicating the name of the "limit" query parameter. Defaults to'limit'.offset_query_param- A string value indicating the name of the "offset" query parameter. Defaults to'offset'.max_limit- If set this is a numeric value indicating the maximum allowable limit that may be requested by the client. Defaults toNone.template- The name of a template to use when rendering pagination controls in the browsable API. May be overridden to modify the rendering style, or set toNoneto disable HTML pagination controls completely. Defaults to"rest_framework/pagination/numbers.html".
The cursor-based pagination presents an opaque "cursor" indicator that the client may use to page through the result set. This pagination style only presents forward and reverse controls, and does not allow the client to navigate to arbitrary positions.
Cursor based pagination requires that there is a unique, unchanging ordering of items in the result set. This ordering might typically be a creation timestamp on the records, as this presents a consistent ordering to paginate against.
Cursor based pagination is more complex than other schemes. It also requires that the result set presents a fixed ordering, and does not allow the client to arbitrarily index into the result set. However it does provide the following benefits:
- Provides a consistent pagination view. When used properly
CursorPaginationensures that the client will never see the same item twice when paging through records, even when new items are being inserted by other clients during the pagination process. - Supports usage with very large datasets. With extremely large datasets pagination using offset-based pagination styles may become inefficient or unusable. Cursor based pagination schemes instead have fixed-time properties, and do not slow down as the dataset size increases.
Proper use of cursor based pagination requires a little attention to detail. You'll need to think about what ordering you want the scheme to be applied against. The default is to order by "-created". This assumes that there must be a 'created' timestamp field on the model instances, and will present a "timeline" style paginated view, with the most recently added items first.
You can modify the ordering by overriding the 'ordering' attribute on the pagination class, or by using the OrderingFilter filter class together with CursorPagination. When used with OrderingFilter you should strongly consider restricting the fields that the user may order by.
Proper usage of cursor pagination should have an ordering field that satisfies the following:
- Should be an unchanging value, such as a timestamp, slug, or other field that is only set once, on creation.
- Should be unique, or nearly unique. Millisecond precision timestamps are a good example. This implementation of cursor pagination uses a smart "position plus offset" style that allows it to properly support not-strictly-unique values as the ordering.
- Should be a non-nullable value that can be coerced to a string.
- Should not be a float. Precision errors easily lead to incorrect results.
Hint: use decimals instead.
(If you already have a float field and must paginate on that, an
example
CursorPaginationsubclass that uses decimals to limit precision is available here.) - The field should have a database index.
Using an ordering field that does not satisfy these constraints will generally still work, but you'll be losing some of the benefits of cursor pagination.
For more technical details on the implementation we use for cursor pagination, the "Building cursors for the Disqus API" blog post gives a good overview of the basic approach.
To enable the CursorPagination style globally, use the following configuration, modifying the PAGE_SIZE as desired:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'rest_framework.pagination.CursorPagination',
'PAGE_SIZE': 100
}
On GenericAPIView subclasses you may also set the pagination_class attribute to select CursorPagination on a per-view basis.
The CursorPagination class includes a number of attributes that may be overridden to modify the pagination style.
To set these attributes you should override the CursorPagination class, and then enable your custom pagination class as above.
page_size= A numeric value indicating the page size. If set, this overrides thePAGE_SIZEsetting. Defaults to the same value as thePAGE_SIZEsettings key.cursor_query_param= A string value indicating the name of the "cursor" query parameter. Defaults to'cursor'.ordering= This should be a string, or list of strings, indicating the field against which the cursor based pagination will be applied. For example:ordering = 'slug'. Defaults to-created. This value may also be overridden by usingOrderingFilteron the view.template= The name of a template to use when rendering pagination controls in the browsable API. May be overridden to modify the rendering style, or set toNoneto disable HTML pagination controls completely. Defaults to"rest_framework/pagination/previous_and_next.html".
To create a custom pagination serializer class you should subclass pagination.BasePagination and override the paginate_queryset(self, queryset, request, view=None) and get_paginated_response(self, data) methods:
- The
paginate_querysetmethod is passed the initial queryset and should return an iterable object that contains only the data in the requested page. - The
get_paginated_responsemethod is passed the serialized page data and should return aResponseinstance.
Note that the paginate_queryset method may set state on the pagination instance, that may later be used by the get_paginated_response method.
Suppose we want to replace the default pagination output style with a modified format that includes the next and previous links under in a nested 'links' key. We could specify a custom pagination class like so:
class CustomPagination(pagination.PageNumberPagination):
def get_paginated_response(self, data):
return Response({
'links': {
'next': self.get_next_link(),
'previous': self.get_previous_link()
},
'count': self.page.paginator.count,
'results': data
})
We'd then need to setup the custom class in our configuration:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'my_project.apps.core.pagination.CustomPagination',
'PAGE_SIZE': 100
}
Note that if you care about how the ordering of keys is displayed in responses in the browsable API you might choose to use an OrderedDict when constructing the body of paginated responses, but this is optional.
To have your custom pagination class be used by default, use the DEFAULT_PAGINATION_CLASS setting:
REST_FRAMEWORK = {
'DEFAULT_PAGINATION_CLASS': 'my_project.apps.core.pagination.LinkHeaderPagination',
'PAGE_SIZE': 100
}
API responses for list endpoints will now include a Link header, instead of including the pagination links as part of the body of the response, for example:
You can also make the pagination controls available to the schema autogeneration
that REST framework provides, by implementing a get_schema_fields() method. This method should have the following signature:
get_schema_fields(self, view)
The method should return a list of coreapi.Field instances.
A custom pagination style, using the 'Link' header'
By default using the pagination classes will cause HTML pagination controls to be displayed in the browsable API. There are two built-in display styles. The PageNumberPagination and LimitOffsetPagination classes display a list of page numbers with previous and next controls. The CursorPagination class displays a simpler style that only displays a previous and next control.
You can override the templates that render the HTML pagination controls. The two built-in styles are:
rest_framework/pagination/numbers.htmlrest_framework/pagination/previous_and_next.html
Providing a template with either of these paths in a global template directory will override the default rendering for the relevant pagination classes.
Alternatively you can disable HTML pagination controls completely by subclassing on of the existing classes, setting template = None as an attribute on the class. You'll then need to configure your DEFAULT_PAGINATION_CLASS settings key to use your custom class as the default pagination style.
The low-level API for determining if a pagination class should display the controls or not is exposed as a display_page_controls attribute on the pagination instance. Custom pagination classes should be set to True in the paginate_queryset method if they require the HTML pagination controls to be displayed.
The .to_html() and .get_html_context() methods may also be overridden in a custom pagination class in order to further customize how the controls are rendered.
The following third party packages are also available.
The DRF-extensions package includes a PaginateByMaxMixin mixin class that allows your API clients to specify ?page_size=max to obtain the maximum allowed page size.
The drf-proxy-pagination package includes a ProxyPagination class which allows to choose pagination class with a query parameter.
The django-rest-framework-link-header-pagination package includes a LinkHeaderPagination class which provides pagination via an HTTP Link header as desribed in Github's developer documentation.
source: versioning.py
Versioning an interface is just a "polite" way to kill deployed clients.
— Roy Fielding.
API versioning allows you to alter behavior between different clients. REST framework provides for a number of different versioning schemes.
Versioning is determined by the incoming client request, and may either be based on the request URL, or based on the request headers.
There are a number of valid approaches to approaching versioning. Non-versioned systems can also be appropriate, particularly if you're engineering for very long-term systems with multiple clients outside of your control.
When API versioning is enabled, the request.version attribute will contain a string that corresponds to the version requested in the incoming client request.
By default, versioning is not enabled, and request.version will always return None.
How you vary the API behavior is up to you, but one example you might typically want is to switch to a different serialization style in a newer version. For example:
def get_serializer_class(self):
if self.request.version == 'v1':
return AccountSerializerVersion1
return AccountSerializer
The reverse function included by REST framework ties in with the versioning scheme. You need to make sure to include the current request as a keyword argument, like so.
from rest_framework.reverse import reverse
reverse('bookings-list', request=request)
The above function will apply any URL transformations appropriate to the request version. For example:
- If
NamespaceVersioningwas being used, and the API version was 'v1', then the URL lookup used would be'v1:bookings-list', which might resolve to a URL likehttp://example.org/v1/bookings/. - If
QueryParameterVersioningwas being used, and the API version was1.0, then the returned URL might be something likehttp://example.org/bookings/?version=1.0
When using hyperlinked serialization styles together with a URL based versioning scheme make sure to include the request as context to the serializer.
def get(self, request):
queryset = Booking.objects.all()
serializer = BookingsSerializer(queryset, many=True, context={'request': request})
return Response({'all_bookings': serializer.data})
Doing so will allow any returned URLs to include the appropriate versioning.
The versioning scheme is defined by the DEFAULT_VERSIONING_CLASS settings key.
REST_FRAMEWORK = {
'DEFAULT_VERSIONING_CLASS': 'rest_framework.versioning.NamespaceVersioning'
}
Unless it is explicitly set, the value for DEFAULT_VERSIONING_CLASS will be None. In this case the request.version attribute will always return None.
You can also set the versioning scheme on an individual view. Typically you won't need to do this, as it makes more sense to have a single versioning scheme used globally. If you do need to do so, use the versioning_class attribute.
class ProfileList(APIView):
versioning_class = versioning.QueryParameterVersioning
The following settings keys are also used to control versioning:
DEFAULT_VERSION. The value that should be used forrequest.versionwhen no versioning information is present. Defaults toNone.ALLOWED_VERSIONS. If set, this value will restrict the set of versions that may be returned by the versioning scheme, and will raise an error if the provided version is not in this set. Note that the value used for theDEFAULT_VERSIONsetting is always considered to be part of theALLOWED_VERSIONSset (unless it isNone). Defaults toNone.VERSION_PARAM. The string that should be used for any versioning parameters, such as in the media type or URL query parameters. Defaults to'version'.
You can also set your versioning class plus those three values on a per-view or a per-viewset basis by defining your own versioning scheme and using the default_version, allowed_versions and version_param class variables. For example, if you want to use URLPathVersioning:
from rest_framework.versioning import URLPathVersioning
from rest_framework.views import APIView
class ExampleVersioning(URLPathVersioning):
default_version = ...
allowed_versions = ...
version_param = ...
class ExampleView(APIVIew):
versioning_class = ExampleVersioning
## AcceptHeaderVersioning
This scheme requires the client to specify the version as part of the media type in the Accept header. The version is included as a media type parameter, that supplements the main media type.
Here's an example HTTP request using the accept header versioning style.
GET /bookings/ HTTP/1.1
Host: example.com
Accept: application/json; version=1.0
In the example request above request.version attribute would return the string '1.0'.
Versioning based on accept headers is generally considered as best practice, although other styles may be suitable depending on your client requirements.
Strictly speaking the json media type is not specified as including additional parameters. If you are building a well-specified public API you might consider using a vendor media type. To do so, configure your renderers to use a JSON based renderer with a custom media type:
class BookingsAPIRenderer(JSONRenderer):
media_type = 'application/vnd.megacorp.bookings+json'
Your client requests would now look like this:
GET /bookings/ HTTP/1.1
Host: example.com
Accept: application/vnd.megacorp.bookings+json; version=1.0
This scheme requires the client to specify the version as part of the URL path.
GET /v1/bookings/ HTTP/1.1
Host: example.com
Accept: application/json
Your URL conf must include a pattern that matches the version with a 'version' keyword argument, so that this information is available to the versioning scheme.
urlpatterns = [
url(
r'^(?P<version>(v1|v2))/bookings/$',
bookings_list,
name='bookings-list'
),
url(
r'^(?P<version>(v1|v2))/bookings/(?P<pk>[0-9]+)/$',
bookings_detail,
name='bookings-detail'
)
]
To the client, this scheme is the same as URLPathVersioning. The only difference is how it is configured in your Django application, as it uses URL namespacing, instead of URL keyword arguments.
GET /v1/something/ HTTP/1.1
Host: example.com
Accept: application/json
With this scheme the request.version attribute is determined based on the namespace that matches the incoming request path.
In the following example we're giving a set of views two different possible URL prefixes, each under a different namespace:
# bookings/urls.py
urlpatterns = [
url(r'^$', bookings_list, name='bookings-list'),
url(r'^(?P<pk>[0-9]+)/$', bookings_detail, name='bookings-detail')
]
# urls.py
urlpatterns = [
url(r'^v1/bookings/', include('bookings.urls', namespace='v1')),
url(r'^v2/bookings/', include('bookings.urls', namespace='v2'))
]
Both URLPathVersioning and NamespaceVersioning are reasonable if you just need a simple versioning scheme. The URLPathVersioning approach might be better suitable for small ad-hoc projects, and the NamespaceVersioning is probably easier to manage for larger projects.
The hostname versioning scheme requires the client to specify the requested version as part of the hostname in the URL.
For example the following is an HTTP request to the http://v1.example.com/bookings/ URL:
GET /bookings/ HTTP/1.1
Host: v1.example.com
Accept: application/json
By default this implementation expects the hostname to match this simple regular expression:
^([a-zA-Z0-9]+)\.[a-zA-Z0-9]+\.[a-zA-Z0-9]+$
Note that the first group is enclosed in brackets, indicating that this is the matched portion of the hostname.
The HostNameVersioning scheme can be awkward to use in debug mode as you will typically be accessing a raw IP address such as 127.0.0.1. There are various online tutorials on how to access localhost with a custom subdomain which you may find helpful in this case.
Hostname based versioning can be particularly useful if you have requirements to route incoming requests to different servers based on the version, as you can configure different DNS records for different API versions.
This scheme is a simple style that includes the version as a query parameter in the URL. For example:
GET /something/?version=0.1 HTTP/1.1
Host: example.com
Accept: application/json
To implement a custom versioning scheme, subclass BaseVersioning and override the .determine_version method.
The following example uses a custom X-API-Version header to determine the requested version.
class XAPIVersionScheme(versioning.BaseVersioning):
def determine_version(self, request, *args, **kwargs):
return request.META.get('HTTP_X_API_VERSION', None)
If your versioning scheme is based on the request URL, you will also want to alter how versioned URLs are determined. In order to do so you should override the .reverse() method on the class. See the source code for examples.
source: negotiation.py
HTTP has provisions for several mechanisms for "content negotiation" - the process of selecting the best representation for a given response when there are multiple representations available.
— RFC 2616, Fielding et al.
Content negotiation is the process of selecting one of multiple possible representations to return to a client, based on client or server preferences.
REST framework uses a simple style of content negotiation to determine which media type should be returned to a client, based on the available renderers, the priorities of each of those renderers, and the client's Accept: header. The style used is partly client-driven, and partly server-driven.
- More specific media types are given preference to less specific media types.
- If multiple media types have the same specificity, then preference is given to based on the ordering of the renderers configured for the given view.
For example, given the following Accept header:
application/json; indent=4, application/json, application/yaml, text/html, */*
The priorities for each of the given media types would be:
application/json; indent=4application/json,application/yamlandtext/html*/*
If the requested view was only configured with renderers for YAML and HTML, then REST framework would select whichever renderer was listed first in the renderer_classes list or DEFAULT_RENDERER_CLASSES setting.
For more information on the HTTP Accept header, see RFC 2616
Note: "q" values are not taken into account by REST framework when determining preference. The use of "q" values negatively impacts caching, and in the author's opinion they are an unnecessary and overcomplicated approach to content negotiation.
This is a valid approach as the HTTP spec deliberately underspecifies how a server should weight server-based preferences against client-based preferences.
It's unlikely that you'll want to provide a custom content negotiation scheme for REST framework, but you can do so if needed. To implement a custom content negotiation scheme override BaseContentNegotiation.
REST framework's content negotiation classes handle selection of both the appropriate parser for the request, and the appropriate renderer for the response, so you should implement both the .select_parser(request, parsers) and .select_renderer(request, renderers, format_suffix) methods.
The select_parser() method should return one of the parser instances from the list of available parsers, or None if none of the parsers can handle the incoming request.
The select_renderer() method should return a two-tuple of (renderer instance, media type), or raise a NotAcceptable exception.
The following is a custom content negotiation class which ignores the client request when selecting the appropriate parser or renderer.
from rest_framework.negotiation import BaseContentNegotiation
class IgnoreClientContentNegotiation(BaseContentNegotiation):
def select_parser(self, request, parsers):
"""
Select the first parser in the `.parser_classes` list.
"""
return parsers[0]
def select_renderer(self, request, renderers, format_suffix):
"""
Select the first renderer in the `.renderer_classes` list.
"""
return (renderers[0], renderers[0].media_type)
The default content negotiation class may be set globally, using the DEFAULT_CONTENT_NEGOTIATION_CLASS setting. For example, the following settings would use our example IgnoreClientContentNegotiation class.
REST_FRAMEWORK = {
'DEFAULT_CONTENT_NEGOTIATION_CLASS': 'myapp.negotiation.IgnoreClientContentNegotiation',
}
You can also set the content negotiation used for an individual view, or viewset, using the APIView class-based views.
from myapp.negotiation import IgnoreClientContentNegotiation
from rest_framework.response import Response
from rest_framework.views import APIView
class NoNegotiationView(APIView):
"""
An example view that does not perform content negotiation.
"""
content_negotiation_class = IgnoreClientContentNegotiation
def get(self, request, format=None):
return Response({
'accepted media type': request.accepted_renderer.media_type
})
source: metadata.py
[The
OPTIONS] method allows a client to determine the options and/or requirements associated with a resource, or the capabilities of a server, without implying a resource action or initiating a resource retrieval.
REST framework includes a configurable mechanism for determining how your API should respond to OPTIONS requests. This allows you to return API schema or other resource information.
There are not currently any widely adopted conventions for exactly what style of response should be returned for HTTP OPTIONS requests, so we provide an ad-hoc style that returns some useful information.
Here's an example response that demonstrates the information that is returned by default.
HTTP 200 OK
Allow: GET, POST, HEAD, OPTIONS
Content-Type: application/json
{
"name": "To Do List",
"description": "List existing 'To Do' items, or create a new item.",
"renders": [
"application/json",
"text/html"
],
"parses": [
"application/json",
"application/x-www-form-urlencoded",
"multipart/form-data"
],
"actions": {
"POST": {
"note": {
"type": "string",
"required": false,
"read_only": false,
"label": "title",
"max_length": 100
}
}
}
}
You can set the metadata class globally using the 'DEFAULT_METADATA_CLASS' settings key:
REST_FRAMEWORK = {
'DEFAULT_METADATA_CLASS': 'rest_framework.metadata.SimpleMetadata'
}
Or you can set the metadata class individually for a view:
class APIRoot(APIView):
metadata_class = APIRootMetadata
def get(self, request, format=None):
return Response({
...
})
The REST framework package only includes a single metadata class implementation, named SimpleMetadata. If you want to use an alternative style you'll need to implement a custom metadata class.
If you have specific requirements for creating schema endpoints that are accessed with regular GET requests, you might consider re-using the metadata API for doing so.
For example, the following additional route could be used on a viewset to provide a linkable schema endpoint.
@action(methods=['GET'], detail=False)
def schema(self, request):
meta = self.metadata_class()
data = meta.determine_metadata(request, self)
return Response(data)
There are a couple of reasons that you might choose to take this approach, including that OPTIONS responses are not cacheable.
If you want to provide a custom metadata class you should override BaseMetadata and implement the determine_metadata(self, request, view) method.
Useful things that you might want to do could include returning schema information, using a format such as JSON schema, or returning debug information to admin users.
The following class could be used to limit the information that is returned to OPTIONS requests.
class MinimalMetadata(BaseMetadata):
"""
Don't include field and other information for `OPTIONS` requests.
Just return the name and description.
"""
def determine_metadata(self, request, view):
return {
'name': view.get_view_name(),
'description': view.get_view_description()
}
Then configure your settings to use this custom class:
REST_FRAMEWORK = {
'DEFAULT_METADATA_CLASS': 'myproject.apps.core.MinimalMetadata'
}
The following third party packages provide additional metadata implementations.
drf-schema-adapter is a set of tools that makes it easier to provide schema information to frontend frameworks and libraries. It provides a metadata mixin as well as 2 metadata classes and several adapters suitable to generate json-schema as well as schema information readable by various libraries.
You can also write your own adapter to work with your specific frontend. If you wish to do so, it also provides an exporter that can export those schema information to json files.
source: schemas.py
A machine-readable [schema] describes what resources are available via the API, what their URLs are, how they are represented and what operations they support.
— Heroku, JSON Schema for the Heroku Platform API
API schemas are a useful tool that allow for a range of use cases, including generating reference documentation, or driving dynamic client libraries that can interact with your API.
You'll need to install the coreapi package in order to add schema support
for REST framework.
pip install coreapi
REST framework uses Core API in order to model schema information in a format-independent representation. This information can then be rendered into various different schema formats, or used to generate API documentation.
When using Core API, a schema is represented as a Document which is the
top-level container object for information about the API. Available API
interactions are represented using Link objects. Each link includes a URL,
HTTP method, and may include a list of Field instances, which describe any
parameters that may be accepted by the API endpoint. The Link and Field
instances may also include descriptions, that allow an API schema to be
rendered into user documentation.
Here's an example of an API description that includes a single search
endpoint:
coreapi.Document(
title='Flight Search API',
url='https://api.example.org/',
content={
'search': coreapi.Link(
url='/search/',
action='get',
fields=[
coreapi.Field(
name='from',
required=True,
location='query',
description='City name or airport code.'
),
coreapi.Field(
name='to',
required=True,
location='query',
description='City name or airport code.'
),
coreapi.Field(
name='date',
required=True,
location='query',
description='Flight date in "YYYY-MM-DD" format.'
)
],
description='Return flight availability and prices.'
)
}
)
In order to be presented in an HTTP response, the internal representation has to be rendered into the actual bytes that are used in the response.
Core JSON is designed as a canonical format for use with Core API.
REST framework includes a renderer class for handling this media type, which
is available as renderers.CoreJSONRenderer.
Other schema formats such as Open API ("Swagger"),
JSON HyperSchema, or API Blueprint can also
be supported by implementing a custom renderer class that handles converting a
Document instance into a bytestring representation.
If there is a Core API codec package that supports encoding into the format you want to use then implementing the renderer class can be done by using the codec.
For example, the openapi_codec package provides support for encoding or decoding
to the Open API ("Swagger") format:
from rest_framework import renderers
from openapi_codec import OpenAPICodec
class SwaggerRenderer(renderers.BaseRenderer):
media_type = 'application/openapi+json'
format = 'swagger'
def render(self, data, media_type=None, renderer_context=None):
codec = OpenAPICodec()
return codec.dump(data)
It's worth pointing out here that Core API can also be used to model hypermedia responses, which present an alternative interaction style to API schemas.
With an API schema, the entire available interface is presented up-front as a single endpoint. Responses to individual API endpoints are then typically presented as plain data, without any further interactions contained in each response.
With Hypermedia, the client is instead presented with a document containing both data and available interactions. Each interaction results in a new document, detailing both the current state and the available interactions.
Further information and support on building Hypermedia APIs with REST framework is planned for a future version.
REST framework includes functionality for auto-generating a schema, or allows you to specify one explicitly.
To manually specify a schema you create a Core API Document, similar to the
example above.
schema = coreapi.Document(
title='Flight Search API',
content={
...
}
)
Automatic schema generation is provided by the SchemaGenerator class.
SchemaGenerator processes a list of routed URL pattterns and compiles the
appropriately structured Core API Document.
Basic usage is just to provide the title for your schema and call
get_schema():
generator = schemas.SchemaGenerator(title='Flight Search API')
schema = generator.get_schema()
By default, view introspection is performed by an AutoSchema instance
accessible via the schema attribute on APIView. This provides the
appropriate Core API Link object for the view, request method and path:
auto_schema = view.schema
coreapi_link = auto_schema.get_link(...)
(In compiling the schema, SchemaGenerator calls view.schema.get_link() for
each view, allowed method and path.)
Note: For basic APIView subclasses, default introspection is essentially
limited to the URL kwarg path parameters. For GenericAPIView
subclasses, which includes all the provided class based views, AutoSchema will
attempt to introspect serialiser, pagination and filter fields, as well as
provide richer path field descriptions. (The key hooks here are the relevant
GenericAPIView attributes and methods: get_serializer, pagination_class,
filter_backends and so on.)
To customise the Link generation you may:
-
Instantiate
AutoSchemaon your view with themanual_fieldskwarg:from rest_framework.views import APIView from rest_framework.schemas import AutoSchema class CustomView(APIView): ... schema = AutoSchema( manual_fields=[ coreapi.Field("extra_field", ...), ] )This allows extension for the most common case without subclassing.
-
Provide an
AutoSchemasubclass with more complex customisation:from rest_framework.views import APIView from rest_framework.schemas import AutoSchema class CustomSchema(AutoSchema): def get_link(...): # Implement custom introspection here (or in other sub-methods) class CustomView(APIView): ... schema = CustomSchema()This provides complete control over view introspection.
-
Instantiate
ManualSchemaon your view, providing the Core APIFieldsfor the view explicitly:from rest_framework.views import APIView from rest_framework.schemas import ManualSchema class CustomView(APIView): ... schema = ManualSchema(fields=[ coreapi.Field( "first_field", required=True, location="path", schema=coreschema.String() ), coreapi.Field( "second_field", required=True, location="path", schema=coreschema.String() ), ])This allows manually specifying the schema for some views whilst maintaining automatic generation elsewhere.
You may disable schema generation for a view by setting schema to None:
class CustomView(APIView):
...
schema = None # Will not appear in schema
Note: For full details on SchemaGenerator plus the AutoSchema and
ManualSchema descriptors see the API Reference below.
There are a few different ways to add a schema view to your API, depending on exactly what you need.
The simplest way to include a schema in your project is to use the
get_schema_view() function.
from rest_framework.schemas import get_schema_view
schema_view = get_schema_view(title="Server Monitoring API")
urlpatterns = [
url('^$', schema_view),
...
]
Once the view has been added, you'll be able to make API requests to retrieve the auto-generated schema definition.
$ http http://127.0.0.1:8000/ Accept:application/coreapi+json
HTTP/1.0 200 OK
Allow: GET, HEAD, OPTIONS
Content-Type: application/vnd.coreapi+json
{
"_meta": {
"title": "Server Monitoring API"
},
"_type": "document",
...
}
The arguments to get_schema_view() are:
May be used to provide a descriptive title for the schema definition.
May be used to pass a canonical URL for the schema.
schema_view = get_schema_view(
title='Server Monitoring API',
url='https://www.example.org/api/'
)
A string representing the import path to the URL conf that you want to generate an API schema for. This defaults to the value of Django's ROOT_URLCONF setting.
schema_view = get_schema_view(
title='Server Monitoring API',
url='https://www.example.org/api/',
urlconf='myproject.urls'
)
May be used to pass the set of renderer classes that can be used to render the API root endpoint.
from rest_framework.schemas import get_schema_view
from rest_framework.renderers import CoreJSONRenderer
from my_custom_package import APIBlueprintRenderer
schema_view = get_schema_view(
title='Server Monitoring API',
url='https://www.example.org/api/',
renderer_classes=[CoreJSONRenderer, APIBlueprintRenderer]
)
List of url patterns to limit the schema introspection to. If you only want the myproject.api urls
to be exposed in the schema:
schema_url_patterns = [
url(r'^api/', include('myproject.api.urls')),
]
schema_view = get_schema_view(
title='Server Monitoring API',
url='https://www.example.org/api/',
patterns=schema_url_patterns,
)
May be used to specify a SchemaGenerator subclass to be passed to the
SchemaView.
May be used to specify the list of authentication classes that will apply to the schema endpoint.
Defaults to settings.DEFAULT_AUTHENTICATION_CLASSES
May be used to specify the list of permission classes that will apply to the schema endpoint.
Defaults to settings.DEFAULT_PERMISSION_CLASSES
If you need a little more control than the get_schema_view() shortcut gives you,
then you can use the SchemaGenerator class directly to auto-generate the
Document instance, and to return that from a view.
This option gives you the flexibility of setting up the schema endpoint with whatever behaviour you want. For example, you can apply different permission, throttling, or authentication policies to the schema endpoint.
Here's an example of using SchemaGenerator together with a view to
return the schema.
views.py:
from rest_framework.decorators import api_view, renderer_classes
from rest_framework import renderers, response, schemas
generator = schemas.SchemaGenerator(title='Bookings API')
@api_view()
@renderer_classes([renderers.CoreJSONRenderer])
def schema_view(request):
schema = generator.get_schema(request)
return response.Response(schema)
urls.py:
urlpatterns = [
url('/', schema_view),
...
]
You can also serve different schemas to different users, depending on the permissions they have available. This approach can be used to ensure that unauthenticated requests are presented with a different schema to authenticated requests, or to ensure that different parts of the API are made visible to different users depending on their role.
In order to present a schema with endpoints filtered by user permissions,
you need to pass the request argument to the get_schema() method, like so:
@api_view()
@renderer_classes([renderers.CoreJSONRenderer])
def schema_view(request):
generator = schemas.SchemaGenerator(title='Bookings API')
return response.Response(generator.get_schema(request=request))
An alternative to the auto-generated approach is to specify the API schema
explicitly, by declaring a Document object in your codebase. Doing so is a
little more work, but ensures that you have full control over the schema
representation.
import coreapi
from rest_framework.decorators import api_view, renderer_classes
from rest_framework import renderers, response
schema = coreapi.Document(
title='Bookings API',
content={
...
}
)
@api_view()
@renderer_classes([renderers.CoreJSONRenderer])
def schema_view(request):
return response.Response(schema)
A final option is to write your API schema as a static file, using one of the available formats, such as Core JSON or Open API.
You could then either:
- Write a schema definition as a static file, and serve the static file directly.
- Write a schema definition that is loaded using
Core API, and then rendered to one of many available formats, depending on the client request.
One common usage of API schemas is to use them to build documentation pages.
The schema generation in REST framework uses docstrings to automatically populate descriptions in the schema document.
These descriptions will be based on:
- The corresponding method docstring if one exists.
- A named section within the class docstring, which can be either single line or multi-line.
- The class docstring.
An APIView, with an explicit method docstring.
class ListUsernames(APIView):
def get(self, request):
"""
Return a list of all user names in the system.
"""
usernames = [user.username for user in User.objects.all()]
return Response(usernames)
A ViewSet, with an explict action docstring.
class ListUsernames(ViewSet):
def list(self, request):
"""
Return a list of all user names in the system.
"""
usernames = [user.username for user in User.objects.all()]
return Response(usernames)
A generic view with sections in the class docstring, using single-line style.
class UserList(generics.ListCreateAPIView):
"""
get: List all the users.
post: Create a new user.
"""
queryset = User.objects.all()
serializer_class = UserSerializer
permission_classes = (IsAdminUser,)
A generic viewset with sections in the class docstring, using multi-line style.
class UserViewSet(viewsets.ModelViewSet):
"""
API endpoint that allows users to be viewed or edited.
retrieve:
Return a user instance.
list:
Return all users, ordered by most recently joined.
"""
queryset = User.objects.all().order_by('-date_joined')
serializer_class = UserSerializer
A class that walks a list of routed URL patterns, requests the schema for each view, and collates the resulting CoreAPI Document.
Typically you'll instantiate SchemaGenerator with a single argument, like so:
generator = SchemaGenerator(title='Stock Prices API')
Arguments:
titlerequired - The name of the API.url- The root URL of the API schema. This option is not required unless the schema is included under path prefix.patterns- A list of URLs to inspect when generating the schema. Defaults to the project's URL conf.urlconf- A URL conf module name to use when generating the schema. Defaults tosettings.ROOT_URLCONF.
Returns a coreapi.Document instance that represents the API schema.
@api_view
@renderer_classes([renderers.CoreJSONRenderer])
def schema_view(request):
generator = schemas.SchemaGenerator(title='Bookings API')
return Response(generator.get_schema())
The request argument is optional, and may be used if you want to apply per-user
permissions to the resulting schema generation.
Return a nested dictionary containing all the links that should be included in the API schema.
This is a good point to override if you want to modify the resulting structure of the generated schema, as you can build a new dictionary with a different layout.
A class that deals with introspection of individual views for schema generation.
AutoSchema is attached to APIView via the schema attribute.
The AutoSchema constructor takes a single keyword argument manual_fields.
manual_fields: a list of coreapi.Field instances that will be added to
the generated fields. Generated fields with a matching name will be overwritten.
class CustomView(APIView):
schema = AutoSchema(manual_fields=[
coreapi.Field(
"my_extra_field",
required=True,
location="path",
schema=coreschema.String()
),
])
For more advanced customisation subclass AutoSchema to customise schema generation.
class CustomViewSchema(AutoSchema):
"""
Overrides `get_link()` to provide Custom Behavior X
"""
def get_link(self, path, method, base_url):
link = super().get_link(path, method, base_url)
# Do something to customize link here...
return link
class MyView(APIView):
schema = CustomViewSchema()
The following methods are available to override.
Returns a coreapi.Link instance corresponding to the given view.
This is the main entry point. You can override this if you need to provide custom behaviors for particular views.
Returns a string to use as the link description. By default this is based on the view docstring as described in the "Schemas as Documentation" section above.
Returns a string to indicate the encoding for any request body, when interacting
with the given view. Eg. 'application/json'. May return a blank string for views
that do not expect a request body.
Return a list of coreapi.Link() instances. One for each path parameter in the URL.
Return a list of coreapi.Link() instances. One for each field in the serializer class used by the view.
Return a list of coreapi.Link() instances, as returned by the get_schema_fields() method on any pagination class used by the view.
Return a list of coreapi.Link() instances, as returned by the get_schema_fields() method of any filter classes used by the view.
Return a list of coreapi.Field() instances to be added to or replace generated fields. Defaults to (optional) manual_fields passed to AutoSchema constructor.
May be overridden to customise manual fields by path or method. For example, a per-method adjustment may look like this:
def get_manual_fields(self, path, method):
"""Example adding per-method fields."""
extra_fields = []
if method=='GET':
extra_fields = # ... list of extra fields for GET ...
if method=='POST':
extra_fields = # ... list of extra fields for POST ...
manual_fields = super().get_manual_fields(path, method)
return manual_fields + extra_fieldsUtility staticmethod. Encapsulates logic to add or replace fields from a list
by Field.name. May be overridden to adjust replacement criteria.
Allows manually providing a list of coreapi.Field instances for the schema,
plus an optional description.
class MyView(APIView):
schema = ManualSchema(fields=[
coreapi.Field(
"first_field",
required=True,
location="path",
schema=coreschema.String()
),
coreapi.Field(
"second_field",
required=True,
location="path",
schema=coreschema.String()
),
]
)
The ManualSchema constructor takes two arguments:
fields: A list of coreapi.Field instances. Required.
description: A string description. Optional.
encoding: Default None. A string encoding, e.g application/json. Optional.
This documentation gives a brief overview of the components within the coreapi
package that are used to represent an API schema.
Note that these classes are imported from the coreapi package, rather than
from the rest_framework package.
Represents a container for the API schema.
A name for the API.
A canonical URL for the API.
A dictionary, containing the Link objects that the schema contains.
In order to provide more structure to the schema, the content dictionary
may be nested, typically to a second level. For example:
content={
"bookings": {
"list": Link(...),
"create": Link(...),
...
},
"venues": {
"list": Link(...),
...
},
...
}
Represents an individual API endpoint.
The URL of the endpoint. May be a URI template, such as /users/{username}/.
The HTTP method associated with the endpoint. Note that URLs that support
more than one HTTP method, should correspond to a single Link for each.
A list of Field instances, describing the available parameters on the input.
A short description of the meaning and intended usage of the endpoint.
Represents a single input parameter on a given API endpoint.
A descriptive name for the input.
A boolean, indicated if the client is required to included a value, or if the parameter can be omitted.
Determines how the information is encoded into the request. Should be one of the following strings:
"path"
Included in a templated URI. For example a url value of /products/{product_code}/ could be used together with a "path" field, to handle API inputs in a URL path such as /products/slim-fit-jeans/.
These fields will normally correspond with named arguments in the project URL conf.
"query"
Included as a URL query parameter. For example ?search=sale. Typically for GET requests.
These fields will normally correspond with pagination and filtering controls on a view.
"form"
Included in the request body, as a single item of a JSON object or HTML form. For example {"colour": "blue", ...}. Typically for POST, PUT and PATCH requests. Multiple "form" fields may be included on a single link.
These fields will normally correspond with serializer fields on a view.
"body"
Included as the complete request body. Typically for POST, PUT and PATCH requests. No more than one "body" field may exist on a link. May not be used together with "form" fields.
These fields will normally correspond with views that use ListSerializer to validate the request input, or with file upload views.
"application/json"
JSON encoded request content. Corresponds to views using JSONParser.
Valid only if either one or more location="form" fields, or a single
location="body" field is included on the Link.
"multipart/form-data"
Multipart encoded request content. Corresponds to views using MultiPartParser.
Valid only if one or more location="form" fields is included on the Link.
"application/x-www-form-urlencoded"
URL encoded request content. Corresponds to views using FormParser. Valid
only if one or more location="form" fields is included on the Link.
"application/octet-stream"
Binary upload request content. Corresponds to views using FileUploadParser.
Valid only if a location="body" field is included on the Link.
A short description of the meaning and intended usage of the input field.
drf-yasg generates OpenAPI documents suitable for code generation - nested schemas, named models, response bodies, enum/pattern/min/max validators, form parameters, etc.
DRF OpenAPI renders the schema generated by Django Rest Framework in OpenAPI format.
source: urlpatterns.py
Section 6.2.1 does not say that content negotiation should be used all the time.
— Roy Fielding, REST discuss mailing list
A common pattern for Web APIs is to use filename extensions on URLs to provide an endpoint for a given media type. For example, 'http://example.com/api/users.json' to serve a JSON representation.
Adding format-suffix patterns to each individual entry in the URLconf for your API is error-prone and non-DRY, so REST framework provides a shortcut to adding these patterns to your URLConf.
Signature: format_suffix_patterns(urlpatterns, suffix_required=False, allowed=None)
Returns a URL pattern list which includes format suffix patterns appended to each of the URL patterns provided.
Arguments:
- urlpatterns: Required. A URL pattern list.
- suffix_required: Optional. A boolean indicating if suffixes in the URLs should be optional or mandatory. Defaults to
False, meaning that suffixes are optional by default. - allowed: Optional. A list or tuple of valid format suffixes. If not provided, a wildcard format suffix pattern will be used.
Example:
from rest_framework.urlpatterns import format_suffix_patterns
from blog import views
urlpatterns = [
url(r'^/$', views.apt_root),
url(r'^comments/$', views.comment_list),
url(r'^comments/(?P<pk>[0-9]+)/$', views.comment_detail)
]
urlpatterns = format_suffix_patterns(urlpatterns, allowed=['json', 'html'])
When using format_suffix_patterns, you must make sure to add the 'format' keyword argument to the corresponding views. For example:
@api_view(('GET', 'POST'))
def comment_list(request, format=None):
# do stuff...
Or with class-based views:
class CommentList(APIView):
def get(self, request, format=None):
# do stuff...
def post(self, request, format=None):
# do stuff...
The name of the kwarg used may be modified by using the FORMAT_SUFFIX_KWARG setting.
Also note that format_suffix_patterns does not support descending into include URL patterns.
If using the i18n_patterns function provided by Django, as well as format_suffix_patterns you should make sure that the i18n_patterns function is applied as the final, or outermost function. For example:
url patterns = [
…
]
urlpatterns = i18n_patterns(
format_suffix_patterns(urlpatterns, allowed=['json', 'html'])
)
An alternative to the format suffixes is to include the requested format in a query parameter. REST framework provides this option by default, and it is used in the browsable API to switch between differing available representations.
To select a representation using its short format, use the format query parameter. For example: http://example.com/organizations/?format=csv.
The name of this query parameter can be modified using the URL_FORMAT_OVERRIDE setting. Set the value to None to disable this behavior.
There seems to be a view among some of the Web community that filename extensions are not a RESTful pattern, and that HTTP Accept headers should always be used instead.
It is actually a misconception. For example, take the following quote from Roy Fielding discussing the relative merits of query parameter media-type indicators vs. file extension media-type indicators:
“That's why I always prefer extensions. Neither choice has anything to do with REST.” — Roy Fielding, REST discuss mailing list
The quote does not mention Accept headers, but it does make it clear that format suffixes should be considered an acceptable pattern.
404: Not Found source: exceptions.py
Exceptions… allow error handling to be organized cleanly in a central or high-level place within the program structure.
— Doug Hellmann, Python Exception Handling Techniques
REST framework's views handle various exceptions, and deal with returning appropriate error responses.
The handled exceptions are:
- Subclasses of
APIExceptionraised inside REST framework. - Django's
Http404exception. - Django's
PermissionDeniedexception.
In each case, REST framework will return a response with an appropriate status code and content-type. The body of the response will include any additional details regarding the nature of the error.
Most error responses will include a key detail in the body of the response.
For example, the following request:
DELETE http://api.example.com/foo/bar HTTP/1.1
Accept: application/json
Might receive an error response indicating that the DELETE method is not allowed on that resource:
HTTP/1.1 405 Method Not Allowed
Content-Type: application/json
Content-Length: 42
{"detail": "Method 'DELETE' not allowed."}
Validation errors are handled slightly differently, and will include the field names as the keys in the response. If the validation error was not specific to a particular field then it will use the "non_field_errors" key, or whatever string value has been set for the NON_FIELD_ERRORS_KEY setting.
Any example validation error might look like this:
HTTP/1.1 400 Bad Request
Content-Type: application/json
Content-Length: 94
{"amount": ["A valid integer is required."], "description": ["This field may not be blank."]}
You can implement custom exception handling by creating a handler function that converts exceptions raised in your API views into response objects. This allows you to control the style of error responses used by your API.
The function must take a pair of arguments, the first is the exception to be handled, and the second is a dictionary containing any extra context such as the view currently being handled. The exception handler function should either return a Response object, or return None if the exception cannot be handled. If the handler returns None then the exception will be re-raised and Django will return a standard HTTP 500 'server error' response.
For example, you might want to ensure that all error responses include the HTTP status code in the body of the response, like so:
HTTP/1.1 405 Method Not Allowed
Content-Type: application/json
Content-Length: 62
{"status_code": 405, "detail": "Method 'DELETE' not allowed."}
In order to alter the style of the response, you could write the following custom exception handler:
from rest_framework.views import exception_handler
def custom_exception_handler(exc, context):
# Call REST framework's default exception handler first,
# to get the standard error response.
response = exception_handler(exc, context)
# Now add the HTTP status code to the response.
if response is not None:
response.data['status_code'] = response.status_code
return response
The context argument is not used by the default handler, but can be useful if the exception handler needs further information such as the view currently being handled, which can be accessed as context['view'].
The exception handler must also be configured in your settings, using the EXCEPTION_HANDLER setting key. For example:
REST_FRAMEWORK = {
'EXCEPTION_HANDLER': 'my_project.my_app.utils.custom_exception_handler'
}
If not specified, the 'EXCEPTION_HANDLER' setting defaults to the standard exception handler provided by REST framework:
REST_FRAMEWORK = {
'EXCEPTION_HANDLER': 'rest_framework.views.exception_handler'
}
Note that the exception handler will only be called for responses generated by raised exceptions. It will not be used for any responses returned directly by the view, such as the HTTP_400_BAD_REQUEST responses that are returned by the generic views when serializer validation fails.
Signature: APIException()
The base class for all exceptions raised inside an APIView class or @api_view.
To provide a custom exception, subclass APIException and set the .status_code, .default_detail, and default_code attributes on the class.
For example, if your API relies on a third party service that may sometimes be unreachable, you might want to implement an exception for the "503 Service Unavailable" HTTP response code. You could do this like so:
from rest_framework.exceptions import APIException
class ServiceUnavailable(APIException):
status_code = 503
default_detail = 'Service temporarily unavailable, try again later.'
default_code = 'service_unavailable'
There are a number of different properties available for inspecting the status of an API exception. You can use these to build custom exception handling for your project.
The available attributes and methods are:
.detail- Return the textual description of the error..get_codes()- Return the code identifier of the error..get_full_details()- Return both the textual description and the code identifier.
In most cases the error detail will be a simple item:
>>> print(exc.detail)
You do not have permission to perform this action.
>>> print(exc.get_codes())
permission_denied
>>> print(exc.get_full_details())
{'message':'You do not have permission to perform this action.','code':'permission_denied'}
In the case of validation errors the error detail will be either a list or dictionary of items:
>>> print(exc.detail)
{"name":"This field is required.","age":"A valid integer is required."}
>>> print(exc.get_codes())
{"name":"required","age":"invalid"}
>>> print(exc.get_full_details())
{"name":{"message":"This field is required.","code":"required"},"age":{"message":"A valid integer is required.","code":"invalid"}}
Signature: ParseError(detail=None, code=None)
Raised if the request contains malformed data when accessing request.data.
By default this exception results in a response with the HTTP status code "400 Bad Request".
Signature: AuthenticationFailed(detail=None, code=None)
Raised when an incoming request includes incorrect authentication.
By default this exception results in a response with the HTTP status code "401 Unauthenticated", but it may also result in a "403 Forbidden" response, depending on the authentication scheme in use. See the authentication documentation for more details.
Signature: NotAuthenticated(detail=None, code=None)
Raised when an unauthenticated request fails the permission checks.
By default this exception results in a response with the HTTP status code "401 Unauthenticated", but it may also result in a "403 Forbidden" response, depending on the authentication scheme in use. See the authentication documentation for more details.
Signature: PermissionDenied(detail=None, code=None)
Raised when an authenticated request fails the permission checks.
By default this exception results in a response with the HTTP status code "403 Forbidden".
Signature: NotFound(detail=None, code=None)
Raised when a resource does not exists at the given URL. This exception is equivalent to the standard Http404 Django exception.
By default this exception results in a response with the HTTP status code "404 Not Found".
Signature: MethodNotAllowed(method, detail=None, code=None)
Raised when an incoming request occurs that does not map to a handler method on the view.
By default this exception results in a response with the HTTP status code "405 Method Not Allowed".
Signature: NotAcceptable(detail=None, code=None)
Raised when an incoming request occurs with an Accept header that cannot be satisfied by any of the available renderers.
By default this exception results in a response with the HTTP status code "406 Not Acceptable".
Signature: UnsupportedMediaType(media_type, detail=None, code=None)
Raised if there are no parsers that can handle the content type of the request data when accessing request.data.
By default this exception results in a response with the HTTP status code "415 Unsupported Media Type".
Signature: Throttled(wait=None, detail=None, code=None)
Raised when an incoming request fails the throttling checks.
By default this exception results in a response with the HTTP status code "429 Too Many Requests".
Signature: ValidationError(detail, code=None)
The ValidationError exception is slightly different from the other APIException classes:
- The
detailargument is mandatory, not optional. - The
detailargument may be a list or dictionary of error details, and may also be a nested data structure. - By convention you should import the serializers module and use a fully qualified
ValidationErrorstyle, in order to differentiate it from Django's built-in validation error. For example.raise serializers.ValidationError('This field must be an integer value.')
The ValidationError class should be used for serializer and field validation, and by validator classes. It is also raised when calling serializer.is_valid with the raise_exception keyword argument:
serializer.is_valid(raise_exception=True)
The generic views use the raise_exception=True flag, which means that you can override the style of validation error responses globally in your API. To do so, use a custom exception handler, as described above.
By default this exception results in a response with the HTTP status code "400 Bad Request".
Django REST Framework provides two error views suitable for providing generic JSON 500 Server Error and
400 Bad Request responses. (Django's default error views provide HTML responses, which may not be appropriate for an
API-only application.)
Use these as per Django's Customizing error views documentation.
Returns a response with status code 500 and application/json content type.
Set as handler500:
handler500 = 'rest_framework.exceptions.server_error'
Returns a response with status code 400 and application/json content type.
Set as handler400:
handler400 = 'rest_framework.exceptions.bad_request'
source: status.py
418 I'm a teapot - Any attempt to brew coffee with a teapot should result in the error code "418 I'm a teapot". The resulting entity body MAY be short and stout.
— RFC 2324, Hyper Text Coffee Pot Control Protocol
Using bare status codes in your responses isn't recommended. REST framework includes a set of named constants that you can use to make your code more obvious and readable.
from rest_framework import status
from rest_framework.response import Response
def empty_view(self):
content = {'please move along': 'nothing to see here'}
return Response(content, status=status.HTTP_404_NOT_FOUND)
The full set of HTTP status codes included in the status module is listed below.
The module also includes a set of helper functions for testing if a status code is in a given range.
from rest_framework import status
from rest_framework.test import APITestCase
class ExampleTestCase(APITestCase):
def test_url_root(self):
url = reverse('index')
response = self.client.get(url)
self.assertTrue(status.is_success(response.status_code))
For more information on proper usage of HTTP status codes see RFC 2616 and RFC 6585.
This class of status code indicates a provisional response. There are no 1xx status codes used in REST framework by default.
HTTP_100_CONTINUE
HTTP_101_SWITCHING_PROTOCOLS
This class of status code indicates that the client's request was successfully received, understood, and accepted.
HTTP_200_OK
HTTP_201_CREATED
HTTP_202_ACCEPTED
HTTP_203_NON_AUTHORITATIVE_INFORMATION
HTTP_204_NO_CONTENT
HTTP_205_RESET_CONTENT
HTTP_206_PARTIAL_CONTENT
HTTP_207_MULTI_STATUS
This class of status code indicates that further action needs to be taken by the user agent in order to fulfill the request.
HTTP_300_MULTIPLE_CHOICES
HTTP_301_MOVED_PERMANENTLY
HTTP_302_FOUND
HTTP_303_SEE_OTHER
HTTP_304_NOT_MODIFIED
HTTP_305_USE_PROXY
HTTP_306_RESERVED
HTTP_307_TEMPORARY_REDIRECT
The 4xx class of status code is intended for cases in which the client seems to have erred. Except when responding to a HEAD request, the server SHOULD include an entity containing an explanation of the error situation, and whether it is a temporary or permanent condition.
HTTP_400_BAD_REQUEST
HTTP_401_UNAUTHORIZED
HTTP_402_PAYMENT_REQUIRED
HTTP_403_FORBIDDEN
HTTP_404_NOT_FOUND
HTTP_405_METHOD_NOT_ALLOWED
HTTP_406_NOT_ACCEPTABLE
HTTP_407_PROXY_AUTHENTICATION_REQUIRED
HTTP_408_REQUEST_TIMEOUT
HTTP_409_CONFLICT
HTTP_410_GONE
HTTP_411_LENGTH_REQUIRED
HTTP_412_PRECONDITION_FAILED
HTTP_413_REQUEST_ENTITY_TOO_LARGE
HTTP_414_REQUEST_URI_TOO_LONG
HTTP_415_UNSUPPORTED_MEDIA_TYPE
HTTP_416_REQUESTED_RANGE_NOT_SATISFIABLE
HTTP_417_EXPECTATION_FAILED
HTTP_422_UNPROCESSABLE_ENTITY
HTTP_423_LOCKED
HTTP_424_FAILED_DEPENDENCY
HTTP_428_PRECONDITION_REQUIRED
HTTP_429_TOO_MANY_REQUESTS
HTTP_431_REQUEST_HEADER_FIELDS_TOO_LARGE
HTTP_451_UNAVAILABLE_FOR_LEGAL_REASONS
Response status codes beginning with the digit "5" indicate cases in which the server is aware that it has erred or is incapable of performing the request. Except when responding to a HEAD request, the server SHOULD include an entity containing an explanation of the error situation, and whether it is a temporary or permanent condition.
HTTP_500_INTERNAL_SERVER_ERROR
HTTP_501_NOT_IMPLEMENTED
HTTP_502_BAD_GATEWAY
HTTP_503_SERVICE_UNAVAILABLE
HTTP_504_GATEWAY_TIMEOUT
HTTP_505_HTTP_VERSION_NOT_SUPPORTED
HTTP_507_INSUFFICIENT_STORAGE
HTTP_511_NETWORK_AUTHENTICATION_REQUIRED
The following helper functions are available for identifying the category of the response code.
is_informational() # 1xx
is_success() # 2xx
is_redirect() # 3xx
is_client_error() # 4xx
is_server_error() # 5xx
source: test.py
Code without tests is broken as designed.
REST framework includes a few helper classes that extend Django's existing test framework, and improve support for making API requests.
Extends Django's existing RequestFactory class.
The APIRequestFactory class supports an almost identical API to Django's standard RequestFactory class. This means that the standard .get(), .post(), .put(), .patch(), .delete(), .head() and .options() methods are all available.
from rest_framework.test import APIRequestFactory
# Using the standard RequestFactory API to create a form POST request
factory = APIRequestFactory()
request = factory.post('/notes/', {'title': 'new idea'})
Methods which create a request body, such as post, put and patch, include a format argument, which make it easy to generate requests using a content type other than multipart form data. For example:
# Create a JSON POST request
factory = APIRequestFactory()
request = factory.post('/notes/', {'title': 'new idea'}, format='json')
By default the available formats are 'multipart' and 'json'. For compatibility with Django's existing RequestFactory the default format is 'multipart'.
To support a wider set of request formats, or change the default format, see the configuration section.
If you need to explicitly encode the request body, you can do so by setting the content_type flag. For example:
request = factory.post('/notes/', json.dumps({'title': 'new idea'}), content_type='application/json')
One difference worth noting between Django's RequestFactory and REST framework's APIRequestFactory is that multipart form data will be encoded for methods other than just .post().
For example, using APIRequestFactory, you can make a form PUT request like so:
factory = APIRequestFactory()
request = factory.put('/notes/547/', {'title': 'remember to email dave'})
Using Django's RequestFactory, you'd need to explicitly encode the data yourself:
from django.test.client import encode_multipart, RequestFactory
factory = RequestFactory()
data = {'title': 'remember to email dave'}
content = encode_multipart('BoUnDaRyStRiNg', data)
content_type = 'multipart/form-data; boundary=BoUnDaRyStRiNg'
request = factory.put('/notes/547/', content, content_type=content_type)
When testing views directly using a request factory, it's often convenient to be able to directly authenticate the request, rather than having to construct the correct authentication credentials.
To forcibly authenticate a request, use the force_authenticate() method.
from rest_framework.test import force_authenticate
factory = APIRequestFactory()
user = User.objects.get(username='olivia')
view = AccountDetail.as_view()
# Make an authenticated request to the view...
request = factory.get('/accounts/django-superstars/')
force_authenticate(request, user=user)
response = view(request)
The signature for the method is force_authenticate(request, user=None, token=None). When making the call, either or both of the user and token may be set.
For example, when forcibly authenticating using a token, you might do something like the following:
user = User.objects.get(username='olivia')
request = factory.get('/accounts/django-superstars/')
force_authenticate(request, user=user, token=user.auth_token)
Note: force_authenticate directly sets request.user to the in-memory user instance. If you are re-using the same user instance across multiple tests that update the saved user state, you may need to call refresh_from_db() between tests.
Note: When using APIRequestFactory, the object that is returned is Django's standard HttpRequest, and not REST framework's Request object, which is only generated once the view is called.
This means that setting attributes directly on the request object may not always have the effect you expect. For example, setting .token directly will have no effect, and setting .user directly will only work if session authentication is being used.
# Request will only authenticate if `SessionAuthentication` is in use.
request = factory.get('/accounts/django-superstars/')
request.user = user
response = view(request)
By default, requests created with APIRequestFactory will not have CSRF validation applied when passed to a REST framework view. If you need to explicitly turn CSRF validation on, you can do so by setting the enforce_csrf_checks flag when instantiating the factory.
factory = APIRequestFactory(enforce_csrf_checks=True)
Note: It's worth noting that Django's standard RequestFactory doesn't need to include this option, because when using regular Django the CSRF validation takes place in middleware, which is not run when testing views directly. When using REST framework, CSRF validation takes place inside the view, so the request factory needs to disable view-level CSRF checks.
Extends Django's existing Client class.
The APIClient class supports the same request interface as Django's standard Client class. This means the that standard .get(), .post(), .put(), .patch(), .delete(), .head() and .options() methods are all available. For example:
from rest_framework.test import APIClient
client = APIClient()
client.post('/notes/', {'title': 'new idea'}, format='json')
To support a wider set of request formats, or change the default format, see the configuration section.
The login method functions exactly as it does with Django's regular Client class. This allows you to authenticate requests against any views which include SessionAuthentication.
# Make all requests in the context of a logged in session.
client = APIClient()
client.login(username='lauren', password='secret')
To logout, call the logout method as usual.
# Log out
client.logout()
The login method is appropriate for testing APIs that use session authentication, for example web sites which include AJAX interaction with the API.
The credentials method can be used to set headers that will then be included on all subsequent requests by the test client.
from rest_framework.authtoken.models import Token
from rest_framework.test import APIClient
# Include an appropriate `Authorization:` header on all requests.
token = Token.objects.get(user__username='lauren')
client = APIClient()
client.credentials(HTTP_AUTHORIZATION='Token ' + token.key)
Note that calling credentials a second time overwrites any existing credentials. You can unset any existing credentials by calling the method with no arguments.
# Stop including any credentials
client.credentials()
The credentials method is appropriate for testing APIs that require authentication headers, such as basic authentication, OAuth1a and OAuth2 authentication, and simple token authentication schemes.
Sometimes you may want to bypass authentication entirely and force all requests by the test client to be automatically treated as authenticated.
This can be a useful shortcut if you're testing the API but don't want to have to construct valid authentication credentials in order to make test requests.
user = User.objects.get(username='lauren')
client = APIClient()
client.force_authenticate(user=user)
To unauthenticate subsequent requests, call force_authenticate setting the user and/or token to None.
client.force_authenticate(user=None)
By default CSRF validation is not applied when using APIClient. If you need to explicitly enable CSRF validation, you can do so by setting the enforce_csrf_checks flag when instantiating the client.
client = APIClient(enforce_csrf_checks=True)
As usual CSRF validation will only apply to any session authenticated views. This means CSRF validation will only occur if the client has been logged in by calling login().
REST framework also includes a client for interacting with your application
using the popular Python library, requests. This may be useful if:
- You are expecting to interface with the API primarily from another Python service, and want to test the service at the same level as the client will see.
- You want to write tests in such a way that they can also be run against a staging or live environment. (See "Live tests" below.)
This exposes exactly the same interface as if you were using a requests session directly.
client = RequestsClient()
response = client.get('http://testserver/users/')
assert response.status_code == 200
Note that the requests client requires you to pass fully qualified URLs.
The RequestsClient class is useful if you want to write tests that solely interact with the service interface. This is a little stricter than using the standard Django test client, as it means that all interactions should be via the API.
If you're using RequestsClient you'll want to ensure that test setup, and results assertions are performed as regular API calls, rather than interacting with the database models directly. For example, rather than checking that Customer.objects.count() == 3 you would list the customers endpoint, and ensure that it contains three records.
Custom headers and authentication credentials can be provided in the same way
as when using a standard requests.Session instance.
from requests.auth import HTTPBasicAuth
client.auth = HTTPBasicAuth('user', 'pass')
client.headers.update({'x-test': 'true'})
If you're using SessionAuthentication then you'll need to include a CSRF token
for any POST, PUT, PATCH or DELETE requests.
You can do so by following the same flow that a JavaScript based client would use.
First make a GET request in order to obtain a CRSF token, then present that
token in the following request.
For example...
client = RequestsClient()
# Obtain a CSRF token.
response = client.get('/homepage/')
assert response.status_code == 200
csrftoken = response.cookies['csrftoken']
# Interact with the API.
response = client.post('/organisations/', json={
'name': 'MegaCorp',
'status': 'active'
}, headers={'X-CSRFToken': csrftoken})
assert response.status_code == 200
With careful usage both the RequestsClient and the CoreAPIClient provide
the ability to write test cases that can run either in development, or be run
directly against your staging server or production environment.
Using this style to create basic tests of a few core piece of functionality is a powerful way to validate your live service. Doing so may require some careful attention to setup and teardown to ensure that the tests run in a way that they do not directly affect customer data.
The CoreAPIClient allows you to interact with your API using the Python
coreapi client library.
# Fetch the API schema
client = CoreAPIClient()
schema = client.get('http://testserver/schema/')
# Create a new organisation
params = {'name': 'MegaCorp', 'status': 'active'}
client.action(schema, ['organisations', 'create'], params)
# Ensure that the organisation exists in the listing
data = client.action(schema, ['organisations', 'list'])
assert(len(data) == 1)
assert(data == [{'name': 'MegaCorp', 'status': 'active'}])
Custom headers and authentication may be used with CoreAPIClient in a
similar way as with RequestsClient.
from requests.auth import HTTPBasicAuth
client = CoreAPIClient()
client.session.auth = HTTPBasicAuth('user', 'pass')
client.session.headers.update({'x-test': 'true'})
REST framework includes the following test case classes, that mirror the existing Django test case classes, but use APIClient instead of Django's default Client.
APISimpleTestCaseAPITransactionTestCaseAPITestCaseAPILiveServerTestCase
You can use any of REST framework's test case classes as you would for the regular Django test case classes. The self.client attribute will be an APIClient instance.
from django.urls import reverse
from rest_framework import status
from rest_framework.test import APITestCase
from myproject.apps.core.models import Account
class AccountTests(APITestCase):
def test_create_account(self):
"""
Ensure we can create a new account object.
"""
url = reverse('account-list')
data = {'name': 'DabApps'}
response = self.client.post(url, data, format='json')
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
self.assertEqual(Account.objects.count(), 1)
self.assertEqual(Account.objects.get().name, 'DabApps')
REST framework also provides a test case class for isolating urlpatterns on a per-class basis. Note that this inherits from Django's SimpleTestCase, and will most likely need to be mixed with another test case class.
from django.urls import include, path, reverse
from rest_framework.test import APITestCase, URLPatternsTestCase
class AccountTests(APITestCase, URLPatternsTestCase):
urlpatterns = [
path('api/', include('api.urls')),
]
def test_create_account(self):
"""
Ensure we can create a new account object.
"""
url = reverse('account-list')
response = self.client.get(url, format='json')
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertEqual(len(response.data), 1)
When checking the validity of test responses it's often more convenient to inspect the data that the response was created with, rather than inspecting the fully rendered response.
For example, it's easier to inspect response.data:
response = self.client.get('/users/4/')
self.assertEqual(response.data, {'id': 4, 'username': 'lauren'})
Instead of inspecting the result of parsing response.content:
response = self.client.get('/users/4/')
self.assertEqual(json.loads(response.content), {'id': 4, 'username': 'lauren'})
If you're testing views directly using APIRequestFactory, the responses that are returned will not yet be rendered, as rendering of template responses is performed by Django's internal request-response cycle. In order to access response.content, you'll first need to render the response.
view = UserDetail.as_view()
request = factory.get('/users/4')
response = view(request, pk='4')
response.render() # Cannot access `response.content` without this.
self.assertEqual(response.content, '{"username": "lauren", "id": 4}')
The default format used to make test requests may be set using the TEST_REQUEST_DEFAULT_FORMAT setting key. For example, to always use JSON for test requests by default instead of standard multipart form requests, set the following in your settings.py file:
REST_FRAMEWORK = {
...
'TEST_REQUEST_DEFAULT_FORMAT': 'json'
}
If you need to test requests using something other than multipart or json requests, you can do so by setting the TEST_REQUEST_RENDERER_CLASSES setting.
For example, to add support for using format='html' in test requests, you might have something like this in your settings.py file.
REST_FRAMEWORK = {
...
'TEST_REQUEST_RENDERER_CLASSES': (
'rest_framework.renderers.MultiPartRenderer',
'rest_framework.renderers.JSONRenderer',
'rest_framework.renderers.TemplateHTMLRenderer'
)
}
source: settings.py
Namespaces are one honking great idea - let's do more of those!
Configuration for REST framework is all namespaced inside a single Django setting, named REST_FRAMEWORK.
For example your project's settings.py file might include something like this:
REST_FRAMEWORK = {
'DEFAULT_RENDERER_CLASSES': (
'rest_framework.renderers.JSONRenderer',
),
'DEFAULT_PARSER_CLASSES': (
'rest_framework.parsers.JSONParser',
)
}
If you need to access the values of REST framework's API settings in your project,
you should use the api_settings object. For example.
from rest_framework.settings import api_settings
print api_settings.DEFAULT_AUTHENTICATION_CLASSES
The api_settings object will check for any user-defined settings, and otherwise fall back to the default values. Any setting that uses string import paths to refer to a class will automatically import and return the referenced class, instead of the string literal.
The following settings control the basic API policies, and are applied to every APIView class-based view, or @api_view function based view.
A list or tuple of renderer classes, that determines the default set of renderers that may be used when returning a Response object.
Default:
(
'rest_framework.renderers.JSONRenderer',
'rest_framework.renderers.BrowsableAPIRenderer',
)
A list or tuple of parser classes, that determines the default set of parsers used when accessing the request.data property.
Default:
(
'rest_framework.parsers.JSONParser',
'rest_framework.parsers.FormParser',
'rest_framework.parsers.MultiPartParser'
)
A list or tuple of authentication classes, that determines the default set of authenticators used when accessing the request.user or request.auth properties.
Default:
(
'rest_framework.authentication.SessionAuthentication',
'rest_framework.authentication.BasicAuthentication'
)
A list or tuple of permission classes, that determines the default set of permissions checked at the start of a view. Permission must be granted by every class in the list.
Default:
(
'rest_framework.permissions.AllowAny',
)
A list or tuple of throttle classes, that determines the default set of throttles checked at the start of a view.
Default: ()
A content negotiation class, that determines how a renderer is selected for the response, given an incoming request.
Default: 'rest_framework.negotiation.DefaultContentNegotiation'
A view inspector class that will be used for schema generation.
Default: 'rest_framework.schemas.AutoSchema'
The following settings control the behavior of the generic class-based views.
This setting has been removed.
The pagination API does not use serializers to determine the output format, and you'll need to instead override the `get_paginated_response method on a pagination class in order to specify how the output format is controlled.
A list of filter backend classes that should be used for generic filtering.
If set to None then generic filtering is disabled.
This setting has been removed.
See the pagination documentation for further guidance on setting the pagination style.
The default page size to use for pagination. If set to None, pagination is disabled by default.
Default: None
This setting has been removed.
See the pagination documentation for further guidance on setting the pagination style.
This setting has been removed.
See the pagination documentation for further guidance on setting the pagination style.
The name of a query parameter, which can be used to specify the search term used by SearchFilter.
Default: search
The name of a query parameter, which can be used to specify the ordering of results returned by OrderingFilter.
Default: ordering
The value that should be used for request.version when no versioning information is present.
Default: None
If set, this value will restrict the set of versions that may be returned by the versioning scheme, and will raise an error if the provided version if not in this set.
Default: None
The string that should used for any versioning parameters, such as in the media type or URL query parameters.
Default: 'version'
The following settings control the behavior of unauthenticated requests.
The class that should be used to initialize request.user for unauthenticated requests.
(If removing authentication entirely, e.g. by removing django.contrib.auth from
INSTALLED_APPS, set UNAUTHENTICATED_USER to None.)
Default: django.contrib.auth.models.AnonymousUser
The class that should be used to initialize request.auth for unauthenticated requests.
Default: None
The following settings control the behavior of APIRequestFactory and APIClient
The default format that should be used when making test requests.
This should match up with the format of one of the renderer classes in the TEST_REQUEST_RENDERER_CLASSES setting.
Default: 'multipart'
The renderer classes that are supported when building test requests.
The format of any of these renderer classes may be used when constructing a test request, for example: client.post('/users', {'username': 'jamie'}, format='json')
Default:
(
'rest_framework.renderers.MultiPartRenderer',
'rest_framework.renderers.JSONRenderer'
)
If set, this maps the 'pk' identifier in the URL conf onto the actual field
name when generating a schema path parameter. Typically this will be 'id'.
This gives a more suitable representation as "primary key" is an implementation
detail, whereas "identifier" is a more general concept.
Default: True
If set, this is used to map internal viewset method names onto external action names used in the schema generation. This allows us to generate names that are more suitable for an external representation than those that are used internally in the codebase.
Default: {'retrieve': 'read', 'destroy': 'delete'}
The name of a URL parameter that may be used to override the default content negotiation Accept header behavior, by using a format=… query parameter in the request URL.
For example: http://example.com/organizations/?format=csv
If the value of this setting is None then URL format overrides will be disabled.
Default: 'format'
The name of a parameter in the URL conf that may be used to provide a format suffix. This setting is applied when using format_suffix_patterns to include suffixed URL patterns.
For example: http://example.com/organizations.csv/
Default: 'format'
The following settings are used to control how date and time representations may be parsed and rendered.
A format string that should be used by default for rendering the output of DateTimeField serializer fields. If None, then DateTimeField serializer fields will return Python datetime objects, and the datetime encoding will be determined by the renderer.
May be any of None, 'iso-8601' or a Python strftime format string.
Default: 'iso-8601'
A list of format strings that should be used by default for parsing inputs to DateTimeField serializer fields.
May be a list including the string 'iso-8601' or Python strftime format strings.
Default: ['iso-8601']
A format string that should be used by default for rendering the output of DateField serializer fields. If None, then DateField serializer fields will return Python date objects, and the date encoding will be determined by the renderer.
May be any of None, 'iso-8601' or a Python strftime format string.
Default: 'iso-8601'
A list of format strings that should be used by default for parsing inputs to DateField serializer fields.
May be a list including the string 'iso-8601' or Python strftime format strings.
Default: ['iso-8601']
A format string that should be used by default for rendering the output of TimeField serializer fields. If None, then TimeField serializer fields will return Python time objects, and the time encoding will be determined by the renderer.
May be any of None, 'iso-8601' or a Python strftime format string.
Default: 'iso-8601'
A list of format strings that should be used by default for parsing inputs to TimeField serializer fields.
May be a list including the string 'iso-8601' or Python strftime format strings.
Default: ['iso-8601']
When set to True, JSON responses will allow unicode characters in responses. For example:
{"unicode black star":"★"}
When set to False, JSON responses will escape non-ascii characters, like so:
{"unicode black star":"\u2605"}
Both styles conform to RFC 4627, and are syntactically valid JSON. The unicode style is preferred as being more user-friendly when inspecting API responses.
Default: True
When set to True, JSON responses will return compact representations, with no spacing after ':' and ',' characters. For example:
{"is_admin":false,"email":"jane@example"}
When set to False, JSON responses will return slightly more verbose representations, like so:
{"is_admin": false, "email": "jane@example"}
The default style is to return minified responses, in line with Heroku's API design guidelines.
Default: True
When set to True, JSON rendering and parsing will only observe syntactically valid JSON, raising an exception for the extended float values (nan, inf, -inf) accepted by Python's json module. This is the recommended setting, as these values are not generally supported. e.g., neither Javascript's JSON.Parse nor PostgreSQL's JSON data type accept these values.
When set to False, JSON rendering and parsing will be permissive. However, these values are still invalid and will need to be specially handled in your code.
Default: True
When returning decimal objects in API representations that do not support a native decimal type, it is normally best to return the value as a string. This avoids the loss of precision that occurs with binary floating point implementations.
When set to True, the serializer DecimalField class will return strings instead of Decimal objects. When set to False, serializers will return Decimal objects, which the default JSON encoder will return as floats.
Default: True
The following settings are used to generate the view names and descriptions, as used in responses to OPTIONS requests, and as used in the browsable API.
A string representing the function that should be used when generating view names.
This should be a function with the following signature:
view_name(cls, suffix=None)
cls: The view class. Typically the name function would inspect the name of the class when generating a descriptive name, by accessingcls.__name__.suffix: The optional suffix used when differentiating individual views in a viewset.
Default: 'rest_framework.views.get_view_name'
A string representing the function that should be used when generating view descriptions.
This setting can be changed to support markup styles other than the default markdown. For example, you can use it to support rst markup in your view docstrings being output in the browsable API.
This should be a function with the following signature:
view_description(cls, html=False)
cls: The view class. Typically the description function would inspect the docstring of the class when generating a description, by accessingcls.__doc__html: A boolean indicating if HTML output is required.Truewhen used in the browsable API, andFalsewhen used in generatingOPTIONSresponses.
Default: 'rest_framework.views.get_view_description'
Global settings for select field cutoffs for rendering relational fields in the browsable API.
Global setting for the html_cutoff value. Must be an integer.
Default: 1000
A string representing a global setting for html_cutoff_text.
Default: "More than {count} items..."
A string representing the function that should be used when returning a response for any given exception. If the function returns None, a 500 error will be raised.
This setting can be changed to support error responses other than the default {"detail": "Failure..."} responses. For example, you can use it to provide API responses like {"errors": [{"message": "Failure...", "code": ""} ...]}.
This should be a function with the following signature:
exception_handler(exc, context)
exc: The exception.
Default: 'rest_framework.views.exception_handler'
A string representing the key that should be used for serializer errors that do not refer to a specific field, but are instead general errors.
Default: 'non_field_errors'
A string representing the key that should be used for the URL fields generated by HyperlinkedModelSerializer.
Default: 'url'
An integer of 0 or more, that may be used to specify the number of application proxies that the API runs behind. This allows throttling to more accurately identify client IP addresses. If set to None then less strict IP matching will be used by the throttle classes.
Default: None