Skip to content

response-model.md

Latest commit

 

History

History
291 lines (203 loc) · 9.08 KB

response-model.md

File metadata and controls

291 lines (203 loc) · 9.08 KB

Response Models

Each route handler has key-value pair of status codes and a response model. This response model holds information on the type of response to be returned.

# project_name/apps/items/controllers.py

from ellar.common import Controller, get, Serializer, ControllerBase

class UserSchema(Serializer):
    username: str
    email: str = None
    first_name: str = None
    last_name: str = None


@Controller
class ItemsController(ControllerBase):
    @get("/me", response=UserSchema)
    def me(self):
        return dict(username='Ellar', email='ellar@example.com')

During route response computation, the me route handler response will evaluate to a JSONResponseModel with UserSchema as content validation schema.

The resulting route responses will be:

from ellar.common import Serializer
from ellar.common.responses.models import JSONResponseModel


class UserSchema(Serializer):
    username: str
    email: str = None
    first_name: str = None
    last_name: str = None

    
response = {200: JSONResponseModel(model_field_or_schema=UserSchema)}

For documentation purposes, we can apply some description to the returned response

@get("/me", response=(UserSchema, 'User Schema Response'))
def me(self):
    return dict(username='Ellar', email='ellar@example.com')

This will be translated to:

response = {200: JSONResponseModel(model_field_or_schema=UserSchema, description='User Schema Response')}

response description

!!! info Each route handler has its own ResponseModel computation and validation. If there is no response definition, Ellar default the route handler model to EmptyAPIResponseModel.

Override Response Type

When you use a Response class as response, a ResponseModel is used and the response_type is replaced with applied response class.

For example:

# project_name/apps/items/controllers.py

from ellar.common import Controller, get, ControllerBase,PlainTextResponse, Serializer


class UserSchema(Serializer):
    username: str
    email: str = None
    first_name: str = None
    last_name: str = None

    
@Controller
class ItemsController(ControllerBase):
    @get("/me", response={200: PlainTextResponse, 201: UserSchema})
    def me(self):
        return "some text response."

This will be translated to:

from ellar.common.responses.models import ResponseModel, JSONResponseModel
from ellar.common import PlainTextResponse

response = {200: ResponseModel(response_type=PlainTextResponse), 201: JSONResponseModel(model_field_or_schema=UserSchema)}

Response Model Properties

All response model follows IResponseModel contract.

import typing as t

from ellar.pydantic.fields import ModelField
from ellar.common import IExecutionContext, Response


class IResponseModel:
    media_type: str
    description: str
    get_model_field: t.Callable[..., t.Optional[t.Union[ModelField, t.Any]]]
    create_response: t.Callable[[IExecutionContext, t.Any], Response]

Properties Overview:

  • media_type: Read from response media type. Required
  • description: For documentation purpose. Default: Success Response. Optional
  • get_model_field: returns response schema if any. Optional
  • create_response: returns a response for the client. Optional

There is also a BaseResponseModel concrete class for more generic implementation. And its adds extra properties for configuration purposes.

They include:

  • response_type: Response classes eg. JSONResponse, PlainResponse, HTMLResponse. etc. Default: Response. Required
  • model_field_or_schema: Optional property. For return data validation. Default: None Optional

Response Resolution Process

When a route handler returns a response, Ellar goes through a resolution process to determine which response model to use. This process is handled by the response_resolver method and follows these steps:

1. Determine the Status Code

The status code is determined in the following priority order:

  1. Single Model Case: If only one response model is defined, its status code is used as the default.

    @get("/item", response=UserSchema)  # Defaults to status code 200
def get_item(self):
    return dict(username='Ellar')

  2. Response Object Status Code: If a Response object was created and has a status code set, that takes precedence.

    @get("/item", response={200: UserSchema, 201: UserSchema})
def get_item(self, res: Response):
    res.status_code = 201  # This status code will be used
    return dict(username='Ellar')

  3. Tuple Return Value: If the handler returns a tuple of (response_obj, status_code), the status code from the tuple is used.

    @get("/item", response={200: UserSchema, 201: UserSchema})
def get_item(self):
    return dict(username='Ellar'), 201  # Returns with status code 201

2. Match to Response Model

After determining the status code, Ellar matches it to the appropriate response model:

  1. Exact Match: If a response model is defined for the specific status code, it's used.
  2. Ellipsis Fallback: If no exact match is found but an Ellipsis (...) key exists, that model is used as a catch-all.
  3. Default Fallback: If no match is found, EmptyAPIResponseModel is used with a warning logged.

Example with Ellipsis fallback:

from ellar.common import Controller, get, ControllerBase, Serializer

class UserSchema(Serializer):
    username: str
    email: str = None

class ErrorSchema(Serializer):
    message: str
    code: int

@Controller
class ItemsController(ControllerBase):
    @get("/item", response={200: UserSchema, ...: ErrorSchema})
    def get_item(self, status: int):
        if status == 200:
            return dict(username='Ellar', email='ellar@example.com')
        # Any other status code will use ErrorSchema
        return dict(message='Error occurred', code=status), status

In this example, returning with status code 200 uses UserSchema, but any other status code (404, 500, etc.) will use ErrorSchema as the fallback.

!!! tip Using the Ellipsis (...) key is useful when you want to define a catch-all response model for various status codes (like error responses) without defining each one explicitly.

Different Response Models

Let's see different ResponseModel available in Ellar and how you can create one too.

ResponseModel

Response model that manages rendering of other response types.

  • Location: ellar.common.responses.models.ResponseModel
  • response_type: Response
  • model_field_or_schema: None
  • media_type: text/plain

JSONResponseModel

Response model that manages JSON response.

  • Location: ellar.common.responses.models.json.JSONResponseModel
  • response_type: JSONResponse OR config.DEFAULT_JSON_CLASS
  • model_field_or_schema: Required
  • media_type: application/json

HTMLResponseModel

Response model that manages HTML templating response. see @render decorator.

  • Location: ellar.common.responses.models.html.HTMLResponseModel
  • response_type: TemplateResponse
  • model_field_or_schema: None
  • media_type: text/html

FileResponseModel

Response model that manages FILE response. see @file decorator.

  • Location: ellar.common.responses.models.file.FileResponseModel
  • response_type: FileResponse
  • model_field_or_schema: ellar.common.responses.models.file.FileResponseModelSchema
  • media_type: Required

StreamingResponseModel

Response model that manages STREAMING response. see @file decorator.

  • Location: ellar.common.responses.models.file.StreamingResponseModel
  • response_type: StreamingResponse
  • model_field_or_schema: ellar.common.responses.models.file.StreamResponseModelSchema
  • media_type: Required

EmptyAPIResponseModel

Default ResponseModel applied when no response is defined.

  • Location: ellar.common.responses.models.json.EmptyAPIResponseModel
  • response_type: JSONResponse OR config.DEFAULT_JSON_CLASS
  • model_field_or_schema: dict
  • media_type: application/json

Custom Response Model

Lets create a new JSON response model.

# project_name/apps/items/controllers.py

import typing as t
from ellar.common import Controller, get, ControllerBase, JSONResponse, Serializer
from ellar.common.responses.models import ResponseModel


class NoteSchema(Serializer):
    id: t.Union[int, None]
    text: str
    completed: bool


class JsonApiResponse(JSONResponse):
    media_type = "application/vnd.api+json"


class JsonApiResponseModel(ResponseModel):
    response_type = JsonApiResponse
    model_field_or_schema = t.List[NoteSchema]
    default_description = 'Successful JsonAPI Response'


@Controller
class ItemsController(ControllerBase):
    @get("/notes/", response=JsonApiResponseModel())
    def get_notes(self):
        return [
            dict(id=1, text='My Json Api Response 1', completed=True),
            dict(id=2, text='My Json Api Response 2', completed=True),
        ]

JsonApiResponseModel Image