At the moment relationships within our API are represented by using primary keys. In this part of the tutorial we'll improve the cohesion and discoverability of our API, by instead using hyperlinking for relationships.
Right now we have endpoints for 'snippets' and 'users', but we don't have a single entry point to our API. To create one, we'll use a regular function-based view and the `@api_view` decorator we introduced earlier. In your `snippets/views.py` add:
Unlike all our other API endpoints, we don't want to use JSON, but instead just present an HTML representation. There are two styles of HTML renderer provided by REST framework, one for dealing with HTML rendered using templates, the other for dealing with pre-rendered HTML. The second renderer is the one we'd like to use for this endpoint.
The other thing we need to consider when creating the code highlight view is that there's no existing concrete generic view that we can use. We're not returning an object instance, but instead a property of an object instance.
Instead of using a concrete generic view, we'll use the base class for representing instances, and create our own `.get()` method. In your `snippets/views.py` add:
Dealing with relationships between entities is one of the more challenging aspects of Web API design. There are a number of different ways that we might choose to represent a relationship:
* Using primary keys.
* Using hyperlinking between entities.
* Using a unique identifying slug field on the related entity.
* Using the default string representation of the related entity.
* Nesting the related entity inside the parent representation.
* Some other custom representation.
REST framework supports all of these styles, and can apply them across forward or reverse relationships, or apply them across custom managers such as generic foreign keys.
In this case we'd like to use a hyperlinked style between entities. In order to do so, we'll modify our serializers to extend `HyperlinkedModelSerializer` instead of the existing `ModelSerializer`.
The `HyperlinkedModelSerializer` has the following differences from `ModelSerializer`:
* It does not include the `pk` field by default.
* It includes a `url` field, using `HyperlinkedIdentityField`.
Notice that we've also added a new `'highlight'` field. This field is of the same type as the `url` field, except that it points to the `'snippet-highlight'` url pattern, instead of the `'snippet-detail'` url pattern.
Because we've included format suffixed URLs such as `'.json'`, we also need to indicate on the `highlight` field that any format suffixed hyperlinks it returns should use the `'.html'` suffix.
If we're going to have a hyperlinked API, we need to make sure we name our URL patterns. Let's take a look at which URL patterns we need to name.
* The root of our API refers to `'user-list'` and `'snippet-list'`.
* Our snippet serializer includes a field that refers to `'snippet-highlight'`.
* Our user serializer includes a field that refers to `'snippet-detail'`.
* Our snippet and user serializers include `'url'` fields that by default will refer to `'{model_name}-detail'`, which in this case will be `'snippet-detail'` and `'user-detail'`.
The list views for users and code snippets could end up returning quite a lot of instances, so really we'd like to make sure we paginate the results, and allow the API client to step through each of the individual pages.
Note that settings in REST framework are all namespaced into a single dictionary setting, named 'REST_FRAMEWORK', which helps keep them well separated from your other project settings.