Skip to content
Use type annotations to validate/deserialize request parameters in Dango REST Framework.
Python
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
rest_typed_views Fixes _get_raw_value for body param Oct 17, 2019
test_project Fixes _get_raw_value for body param Oct 17, 2019
.gitignore Adds pypy files Oct 3, 2019
LICENSE.md Initial commit Sep 15, 2019
README.md Cleans up some readme typos Oct 17, 2019
manage.py Initial commit Sep 15, 2019
pypi_submit.py Adds pypy files Oct 3, 2019
requirements.txt Updates requirements Sep 28, 2019
setup.py Bumps pypi version Oct 17, 2019

README.md

Django REST Framework - Typed Views

Package version Python versions

This project extends Django Rest Framework to allow use of Python's type annotations for automatically validating and casting view parameters. This pattern makes for code that is easier to read and write. View inputs are individually declared, not buried inside all-encompassing request objects. Meanwhile, you get even more out of type annotations: they can replace repetitive validation/sanitization code.

More features:

  • Pydantic models and Marshmallow schemas are compatible types for view parameters. Annotate your POST/PUT functions with them to automatically validate incoming request bodies.
  • Advanced validators for more than just the type: min_value/max_value for numbers
  • Validate string formats: email, uuid and ipv4/6; use Python's native Enum for "choices" validation

Quick example:

from rest_typed_views import typed_api_view

@typed_api_view(["GET"])
def get_users(registered_on: date = None, groups: List[int] = None, is_staff: bool = None):
    print(registered_on, groups, is_staff)

GET /users/registered/?registered_on=2019-03-03&groups=4,5&is_staff=yes
Status Code: 200

    date(2019, 3, 3)   [4, 5]  True

GET /users/?registered_on=9999&groups=admin&is_staff=maybe
🚫 Status Code: 400 ValidationError raised

    {
        "registered_on": "'9999' is not a valid date",
        "groups": "'admin' is not a valid integer",
        "is_staff": "'maybe' is not a valid boolean"
    }

Table of Contents

Install & Decorators

pip install rest_typed_views

You can add type annotation-enabled features to either ViewSet methods or function-based views using the typed_action and typed_api_view decorators. They take the exact same arguments as Django REST's api_view and action decorators.

How It Works: Simple Usage

For many cases, you can rely on implicit behavior for how different parts of the request (URL path variables, query parameters, body) map to the parameters of a view function/method.

The value of a view parameter will come from...

  • the URL path if the path variable and the view argument have the same name, or:
  • the request body if the view argument is annotated using a class from a supported library for complex object validation (Pydantic, MarshMallow), or:
  • a query parameter with the same name

Unless a default value is given, the parameter is required and a ValidationError will be raised if not set.

Basic GET Request

urlpatterns = [
    url(r"^(?P<city>[\w+])/restaurants/", search_restaurants)
]

from rest_typed_views import typed_api_view

# Example request: /chicago/restaurants?delivery=yes
@typed_api_view(["GET"])
def search_restaurants(city: str, rating: float = None, offers_delivery: bool = None):
    restaurants = Restaurant.objects.filter(city=city)

    if rating is not None:
        restaurants = restaurants.filter(rating__gte=rating)

    if offers_delivery is not None:
        restaurants = restaurants.filter(delivery=offers_delivery)

In this example, city is required and must be its string. Its value comes from the URL path variable with the same name. The other parameters, rating and offers_delivery, are not part of the path parameters and are assumed to be query parameters. They both have a default value, so they are optional.

Basic POST Request

# urls.py
urlpatterns = [url(r"^(?P<city>[\w+])/bookings/", create_booking)]

# settings.py
DRF_TYPED_VIEWS = {"schema_packages": ["pydantic"]}

# views.py
from pydantic import BaseModel
from rest_typed_views import typed_api_view


class RoomEnum(str, Enum):
    double = 'double'
    twin = 'twin'
    single = 'single'


class BookingSchema(BaseModel):
    start_date: date
    end_date: date
    room: RoomEnum = RoomEnum.double
    include_breakfast: bool = False

# Example request: /chicago/bookings/
@typed_api_view(["POST"])
def create_booking(city: str, booking: BookingSchema):
    # do something with the validated booking...

In this example, city will again be populated using the URL path variable. The booking parameter is annotated using a supported complex schema class (Pydantic), so it's assumed to come from the request body, which will be read in as JSON, used to hydrate the Pydantic BookingSchema and then validated. If validation fails a ValidationError will be raised.

How It Works: Advanced Usage

For more advanced use cases, you can explicitly declare how each parameter's value is sourced from the request -- from the query parameters, path, body or headers -- as well as define additional validation rules. You import a class named after the request element that is expected to hold the value and assign it to the parameter's default.

from rest_typed_views import typed_api_view, Query, Path

@typed_api_view(["GET"])
def list_documents(year: date = Path(), title: str = Query(default=None)):
    # ORM logic here...

In this example, year is required and must come from the URL path and title is an optional query parameter because the default is set. This is similar to Django REST's serializer fields: passing a default implies that the filed is not required.

Additional Validation Rules

You can use the request element class (Query, Path, Body) to set additional validation constraints. You'll find that these keywords are consistent with Django REST's serializer fields.

from rest_typed_views import typed_api_view, Query, Path

@typed_api_view(["GET"])
def search_restaurants(
    year: date = Path(), 
    rating: int = Query(default=None, min_value=1, max_value=5)
):
    # ORM logic here...


@typed_api_view(["GET"])
def get_document(id: str = Path(format="uuid")):
    # ORM logic here...


@typed_api_view(["GET"])
def search_users(
    email: str = Query(default=None, format="email"), 
    ip_address: str = Query(default=None, format="ip"), 
):
    # ORM logic here...

View a full list of supported types and additional validation rules.

Nested Body Fields

Similar to how source is used in Django REST to control field mappings during serialization, you can use it to specify the exact path to the request data.

from pydantic import BaseModel
from rest_typed_views import typed_api_view, Query, Path

class Document(BaseModel):
    title: str
    body: str

"""
    POST
    {
        "strict": false,
        "data": {
            "title": "A Dark and Stormy Night",
            "body": "Once upon a time"
        }
    }
"""
@typed_api_view(["POST"])
def create_document(
    strict_mode: bool = Body(source="strict"), 
    item: Document = Body(source="data")
):
    # ORM logic here...

You can also use dot-notation to source data multiple levels deep in the JSON payload.

List Validation

For the basic case of list validation - validating types within a comma-delimited string - declare the type to get automatic validation/coercion:

from rest_typed_views import typed_api_view, Query

@typed_api_view(["GET"])
def search_movies(item_ids: List[int] = [])):
    print(item_ids)

# GET /movies?items_ids=41,64,3
# [41, 64, 3]

But you can also specify min_length and max_length, as well as the delimiter and specify additional rules for the child items -- think Django REST's ListField.

Import the generic Param class and use it to set the rules for the child elements:

from rest_typed_views import typed_api_view, Query, Param

@typed_api_view(["GET"])
def search_outcomes(
    scores: List[int] = Query(delimiter="|", child=Param(min_value=0, max_value=100))
):
    # ORM logic ...

@typed_api_view(["GET"])
def search_message(
    recipients: List[str] = Query(min_length=1, max_length=10, child=Param(format="email"))
):
    # ORM logic ...

Accessing the Request Object

You probably won't need to access the request object directly, as this package will provide its relevant properties as view arguments. However, you can include it as a parameter annotated with its type and it will be injected:

from rest_framework.request import Request
from rest_typed_views import typed_api_view

@typed_api_view(["GET"])
def search_documens(request: Request, q: str = None):
    # ORM logic ...

Interdependent Query Parameter Validation

Often, it's useful to validate a combination of query parameters - for instance, a start_date shouldn't come after an end_date. You can use complex schema object (Pydantic or Marshmallow) for this scenario. In the example below, Query(source="*") is instructing an instance of SearchParamsSchema to be populated/validated using all of the query parameters together: request.query_params.dict().

from marshmallow import Schema, fields, validates_schema, ValidationError
from rest_typed_views import typed_api_view

class SearchParamsSchema(Schema):
    start_date = fields.Date()
    end_date = fields.Date()

    @validates_schema
    def validate_numbers(self, data, **kwargs):
        if data["start_date"] >= data["end_date"]:
            raise ValidationError("end_date must come after start_date")

@typed_api_view(["GET"])
def search_documens(search_params: SearchParamsSchema = Query(source="*")):
    # ORM logic ...

(Simple) Access Control

You can apply some very basic access control by applying some validation rules to a view parameter sourced from the CurrentUser request element class. In the example below, a ValidationError will be raised if the request.user is not a member of either super_users or admins.

    from my_pydantic_schemas import BookingSchema
    from rest_typed_views import typed_api_view, CurrentUser

    @typed_api_view(["POST"])
    def create_booking(
        booking: BookingSchema, 
        user: User = CurrentUser(member_of_any=["super_users", "admins"])
    ):
        # Do something with the request.user

Read more about the Current User request element class.

Enabling Marshmallow, Pydantic Schemas

As an alternative to Django REST's serializers, you can annotate views with Pydantic models or Marshmallow schemas to have their parameters automatically validated and pass an instance of the Pydantic/Marshmallow class to your method/function.

To enable support for third-party libraries for complex object validation, modify your settings:

DRF_TYPED_VIEWS = {
    "schema_packages": ["pydantic", "marshmallow"]
}

These third-party packages must be installed in your virtual environment/runtime.

Request Element Classes

You can specify the part of the request that holds each view parameter by using default function arguments, for example:

    from rest_typed_views import Body, Query

    @typed_api_view(["PUT"])
    def update_user(
        user: UserSchema = Body(), 
        optimistic_update: bool = Query(default=False)
    ):

The user parameter will come from the request body and is required because no default is provided. Meanwhile, optimistic_update is not required and will be populated from a query parameter with the same name.

The core keyword arguments to these classes are:

  • default the default value for the parameter, which is required unless set
  • source if the view parameter has a different name than its key embedded in the request

Passing keywords for additional validation constraints is a powerful capability that gets you almost the same feature set as Django REST's flexible serializer fields. See a complete list of validation keywords.

Query

Use the source argument to alias the parameter value and pass keywords to set additional constraints. For example, your query parameters can have dashes, but be mapped to a parameter that have underscores:

    from rest_typed_views import typed_api_view, Query

    @typed_api_view(["GET"])
    def search_events(
        starting_after: date = Query(source="starting-after"),
        available_tickets: int = Query(default=0, min_value=0)
    ):
        # ORM logic here...

Body

By default, the entire request body is used to populate parameters marked with this class (source="*"):

    from rest_typed_views import typed_api_view, Body
    from my_pydantic_schemas import ResidenceListing

    @typed_api_view(["POST"])
    def create_listing(residence: ResidenceListing = Body()):
        # ORM logic ...

However, you can also specify nested fields in the request body, with support for dot notation.

    """
        POST  /users/
        {
            "first_name": "Homer",
            "last_name": "Simpson",
            "contact": {
                "phone" : "800-123-456",
                "fax": "13235551234"
            }
        }
    """
    from rest_typed_views import typed_api_view, Body

    @typed_api_view(["POST"])
    def create_user(
        first_name: str = Body(source="first_name"),
        last_name: str = Body(source="last_name"),
        phone: str = Body(source="contact.phone", min_length=10, max_length=20)
    ):
        # ORM logic ...

Path

Use the source argument to alias a view parameter name. More commonly, though, you can set additional validation rules for parameters coming from the URL path.

    from rest_typed_views import typed_api_view, Query

    @typed_api_view(["GET"])
    def retrieve_event(id: int = Path(min_value=0, max_value=1000)):
        # ORM logic here...

CurrentUser

Use this class to have a view parameter populated with the current user of the request. You can even extract fields from the current user using the source option.

    from my_pydantic_schemas import BookingSchema
    from rest_typed_views import typed_api_view, CurrentUser

    @typed_api_view(["POST"])
    def create_booking(booking: BookingSchema, user: User = CurrentUser()):
        # Do something with the request.user

    @typed_api_view(["GET"])
    def retrieve_something(first_name: str = CurrentUser(source="first_name")):
        # Do something with the request.user's first name

You can also pass some additional parameters to the CurrentUser request element class to implement simple access control:

  • member_of (str) Validates that the current request.user is a member of a group with this name
  • member_of_any (List[str]) Validates that the current request.user is a member of one of these groups

Using these keyword validators assumes that your User model has a many-to-many relationship with django.contrib.auth.models.Group via user.groups.

An example:

from django.contrib.auth.models import User
from rest_typed_views import typed_api_view, CurrentUser

@typed_api_view(["GET"])
def do_something(user: User = CurrentUser(member_of="admin")):
    # now have a user instance (assuming ValidationError wasn't raised)

Supported Types and Validator Rules

The following native Python types are supported. Depending on the type, you can pass additional validation rules to the request element class (Query, Path, Body). You can think of the type combining with the validation rules to create a Django REST serializer field on the fly -- in fact, that's what happens behind the scenes.

str

Additional arguments:

  • max_length Validates that the input contains no more than this number of characters.
  • min_length Validates that the input contains no fewer than this number of characters.
  • trim_whitespace (bool; default True) Whether to trim leading and trailing white space.
  • format Validates that the string matches a common format; supported values:
    • email validates the text to be a valid e-mail address.
    • slug validates the input against the pattern [a-zA-Z0-9_-]+.
    • uuid validates the input is a valid UUID string
    • url validates fully qualified URLs of the form http://<host>/<path>
    • ip validates input is a valid IPv4 or IPv6 string
    • ipv4 validates input is a valid IPv4 string
    • ipv6 validates input is a valid IPv6 string
    • file_path validates that the input corresponds to filenames in a certain directory on the filesystem; allows all the same keyword arguments as Django REST's FilePathField

Some examples:

from rest_typed_views import typed_api_view, Query

@typed_api_view(["GET"])
def search_users(email: str = Query(format='email')):
    # ORM logic here...
    return Response(data)

@typed_api_view(["GET"])
def search_shared_links(url: str = Query(default=None, format='url')):
    # ORM logic here...
    return Response(data)

@typed_api_view(["GET"])
def search_request_logs(ip_address: str = Query(default=None, format='ip')):
    # ORM logic here...
    return Response(data)

int

Additional arguments:

  • max_value Validate that the number provided is no greater than this value.
  • min_value Validate that the number provided is no less than this value.

An example:

from rest_typed_views import typed_api_view, Query

@typed_api_view(["GET"])
def search_products(inventory: int = Query(min_value=0)):
    # ORM logic here...

float

Additional arguments:

  • max_value Validate that the number provided is no greater than this value.
  • min_value Validate that the number provided is no less than this value.

An example:

from rest_typed_views import typed_api_view, Query

@typed_api_view(["GET"])
def search_products(price: float = Query(min_value=0)):
    # ORM logic here...

Decimal

Additional arguments:

  • max_value Validate that the number provided is no greater than this value.
  • min_value Validate that the number provided is no less than this value.
  • .. even more ... accepts the same arguments as Django REST's DecimalField

bool

View parameters annotated with this type will validate and coerce the same values as Django REST's BooleanField, including but not limited to the following:

    true_values = ["yes", 1, "on", "y", "true"]
    false_values = ["no", 0, "off", "n", "false"]

datetime

Additional arguments:

  • input_formats A list of input formats which may be used to parse the date-time, defaults to Django's DATETIME_INPUT_FORMATS settings, which defaults to ['iso-8601']
  • default_timezone A pytz.timezone of the timezone. If not specified, falls back to Django's USE_TZ setting.

date

Additional arguments:

  • input_formats A list of input formats which may be used to parse the date, defaults to Django's DATETIME_INPUT_FORMATS settings, which defaults to ['iso-8601']

time

Additional arguments:

  • input_formats A list of input formats which may be used to parse the time, defaults to Django's TIME_INPUT_FORMATS settings, which defaults to ['iso-8601']

timedelta

Validates strings of the format '[DD] [HH:[MM:]]ss[.uuuuuu]' and converts them to a datetime.timedelta instance.

Additional arguments:

  • max_value Validate that the input duration is no greater than this value.
  • min_value Validate that the input duration is no less than this value.

List

Validates strings of the format '[DD] [HH:[MM:]]ss[.uuuuuu]' and converts them to a datetime.timedelta instance.

Additional arguments:

  • min_length Validates that the list contains no fewer than this number of elements.
  • max_length Validates that the list contains no more than this number of elements.
  • child Pass keyword constraints via a Param instance to to validate the members of the list.

An example:

from rest_typed_views import typed_api_view, Param, Query

@typed_api_view(["GET"])
def search_contacts(emails: List[str] = Query(max_length=10, child=Param(format="email"))):
    # ORM logic here...

Enum

Validates that the value of the input is one of a limited set of choices. Think of this as mapping to a Django REST ChoiceField.

An example:

from rest_typed_views import typed_api_view, Query

class Straws(str, Enum):
    paper = "paper"
    plastic = "plastic"

@typed_api_view(["GET"])
def search_straws(type: Straws = None):
    # ORM logic here...

marshmallow.Schema

You can annotate view parameters with Marshmallow schemas to validate request data and pass an instance of the schema to the view.

from marshmallow import Schema, fields
from rest_typed_views import typed_api_view, Query

class ArtistSchema(Schema):
    name = fields.Str()

class AlbumSchema(Schema):
    title = fields.Str()
    release_date = fields.Date()
    artist = fields.Nested(ArtistSchema())

"""
    POST 
    {
        "title": "Michael Scott's Greatest Hits",
        "release_date": "2019-03-03",
        "artist": {
            "name": "Michael Scott"
        }
    }
"""
@typed_api_view(["POST"])
def create_album(album: AlbumSchema):
    # now have an album instance (assuming ValidationError wasn't raised)

pydantic.BaseModel

You can annotate view parameters with Pydantic models to validate request data and pass an instance of the model to the view.

from pydantic import BaseModel
from rest_typed_views import typed_api_view, Query

class User(BaseModel):
    id: int
    name: str
    signup_ts: datetime = None
    friends: List[int] = []

"""
    POST 
    {
        "id": 24529782,
        "name": "Michael Scott",
        "friends": [24529782]
    }
"""
@typed_api_view(["POST"])
def create_user(user: User):
    # now have a user instance (assuming ValidationError wasn't raised)

Motivation

While REST Framework's ModelViewSets and ModelSerializers are very productive when building out CRUD resources, I've felt less productive in the framework when developing other types of operations. Serializers are a powerful and flexible way to validate incoming request data, but are not as self-documenting as type annotations. Furthermore, the Django ecosystem is hugely productive and I see no reason why REST Framework cannot take advantage of more Python 3 features.

Inspiration

I first came across type annotations for validation in API Star, which has since evolved into an OpenAPI toolkit. This pattern has also been offered by Hug and Molten (I believe in that order). Furthermore, I've borrowed ideas from FastAPI, specifically its use of default values to declare additional validation rules. Finally, this blog post from Instagram's engineering team showed me how decorators can be used to implement these features on view functions.

You can’t perform that action at this time.