Skip to content

hishnash/hypermediachannels

Repository files navigation

Hyper Media Channels Rest Framework

A Hyper Media style serializer for DjangoChannelsRestFramework.

https://travis-ci.org/hishnash/hypermediachannels.svg?branch=master

API USAGE

This is a collection of serialisers and serialiser fields that creats a hypermidea style like PK

Setting up your consumers

All of your consumers should be mapped through a AsyncJsonWebsocketDemultiplexer.

from channelsmultiplexer import AsyncJsonWebsocketDemultiplexer
from djangochannelsrestframework.generics import GenericAsyncAPIConsumer
from hypermediachannels.serializers import HyperChannelsApiModelSerializer

class UserSerializer(HyperChannelsApiModelSerializer):
  class Meta:
      model = User
      fields = (
          '@id',
          'username'
      )

class UserConsumer(GenericAsyncAPIConsumer):
    queryset = User.objects.all()
    serializer_class = UserSerializer

class MainDemultiplexer(AsyncJsonWebsocketDemultiplexer):
    applications = {
        'users': UserConsumer.as_asgi(),
    }

Then when configuring your Channels Application add the MainDemultiplexer as your main consumer. This way all Websocket connections on that URL will run through the Demultiplexer. See DjangoChannelsRestFramework for more deaitls on how to write consumers.

HyperChannelsApiModelSerializer

This can be used inplace of the DRF ModelSerializer. It will (if you include @id in the fields list) add a self reference to the model being displayed)

eg User model Serialiser might respond like this if its fields are ('@id', 'username', 'profile')

{
   @id: {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 1023
      }
   },
   username: '[email protected]',
   profile: {
     stream: 'profile',
     payload: {
       action: 'retrieve',
       pk: 23
     }
   }
}

This will under the hood use HyperlinkedIdentityField to create the @id and profile fields and they will (by default) return these retrieve objects.

Why do we do this.

this means that if we then need to lookup that profile for this user we can just send the msg:

{
  stream: 'profile',
  payload: {
    action: 'retrieve',
    pk: 23
  }
}

Down the websocket and we will get that item, the frontend code does not need to track all of these lockup logic, (consider that some models might have lookup that is not based on pk for example).

If you need to define a different set of lookup params. You can use the kwarg_mappings, stream_name and action_name kwargs to override this.

eg:

class UserSerializer(HyperChannelsApiModelSerializer):
    class Meta:
        model = get_user_model()
        fields = (
            '@id', 'username', 'profile'
        )

        extra_kwargs = {
            'profile': {
                'action_name': 'user_profile',
                'kwarg_mappings': {
                    'user_pk': 'self.pk',
                    'team_pk': 'team.pk'
                }
            },
        }

the kwarg_mappings will set the value in the response user_pk by extracting the pk value on from the User instance.

(pre-appending self to the kwarg_mappings value means it will do the lookup based on the instance parsed to the parent Serializer rather than the instance for this field. In this case a user profile).

so the above would return:

{
   @id: {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 1023
      }
   },
   username: '[email protected]',
   profile: {
     stream: 'user_profile',
     payload: {
       action: 'retrieve',
       user_pk: 1023,
       team_pk: 234234
     }
   }
}

You can use . to access nested values eg. profile.team.name.

Alternatively you can create fields as you would in DRF.
class UserSerializer(HyperChannelsApiModelSerializer):
    team = HyperChannelsApiRelationField(
        source='profile.team',
        kwarg_mappings={
            'member_username': 'self.username'
        }
    )

    class Meta:
        model = get_user_model()
        fields = (
            '@id', 'username', 'team'
        )

this will return:

{
   @id: {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 1023
      }
   },
   username: '[email protected]',
   team: {
     stream: 'team',
     payload: {
       action: 'retrieve',
       member_username: '[email protected]'
     }
   }
}

If you reference a Many field the HyperChannelsApiModelSerializer will do some magic so that:

class UserSerializer(HyperChannelsApiModelSerializer):
    friends = HyperChannelsApiRelationField(
        source='profile.friends'
    )

    class Meta:
        model = get_user_model()
        fields = (
            '@id', 'username', 'friends'
        )



        extra_kwargs = {
            'friends': {
                'kwarg_mappings': {
                    'user_pk': 'self.user.pk',
                }
            },
        }

Adding extra_kwargs for any Many field can be important so that you can controle the lookup params used.

NOTE all Many fields (forwards and backwards) will extract values from the parent instance regardless of if you use self. in the kwarg_mappings value.)

this will return:

{
   @id: {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 1023
      }
   },
   username: '[email protected]',
   friends: {
        stream: 'user_profile', payload: {action: 'list', user_pk: 1023}
    }
}

Remember you can also override the @id lookup/action and stream if needed, eg:

extra_kwargs = {
    '@id': {
        'action_name': 'subscribe_status',
        'kwarg_mappings': {
            'username': 'username'
        }
    },
}

Returning Many items.

Expect to get:

[
    {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 1023
      }
    },
    {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 234
      }
   },
    {
     stream: 'user',
     payload: {
       action: 'retrieve',
       pk: 103223
      }
   },
]

Rather than getting a fully expanded value for each instance you will rather just get a list of hyper media paths you can use to lookup the instance you need.

If you need to override the stream action or lookup do this:

class UserSerializer(HyperChannelsApiModelSerializer):

    class Meta:
        model = User
        fields = (
            '@id',
            'username'
        )

        many_stream_name = 'active_users'

        many_kwarg_mappings = {
            'username': 'username'
        }

        many_action_name = 'subscribe'

Sponsor this project

Packages

No packages published

Languages