Django Rest Framework React Template Framework is a Python package that plugs into Django Rest Framework to generate json schemas for react-jsonschema-form. The concept is similar to the DRF HTMLFormRenderer, but specifically designed for React frontends.
pip install drf-react-template-framework
See /example for a complete simple example.
serializers.py:
from rest_framework import serializers
class ChoiceSerializer(serializers.ModelSerializer):
choice_text = serializers.CharField()
votes = serializers.IntegerField()
class Meta:
fields = ('choice_text', 'votes')viewsets.py:
from rest_framework.mixins import RetrieveModelMixin
from drf_react_template.mixins import FormSchemaViewSetMixin
class ChoiceViewSet(
RetrieveModelMixin,
FormSchemaViewSetMixin,
):
queryset = models.Choice.objects.all()
serializer_class = serializers.ChoiceSerializerNow when the retrieve endpoint is called, FormSchemaViewSetMixin will update the response
to include a serializer object in the json, while the original payload is sorted in formData.
Inside the serializer object, two fields schema and uiSchema contain the objects which
can be used inside react-jsonschema-form.
While the above quick start is useful for super simple cases, the framework has a large number of additional parameters which can be used to customize behaviour.
All viewsets must inherit from drf_react_template.mixins.FormSchemaViewSetMixin.
This class inherits from the
DRF GenericViewSet,
while additionally providing custom finialize response functionality and the additional endpoint:
GET */create/
>> {'serializer': ..., 'formData': {}}
Which can be used to generate an empty form for create.
The only other specific customization that can be applied in the viewset is different
serializers for different endpoints. For example, update actions often show a subset of fields;
as such it is possible to override get_serializer_class to return the specific form required.
For example:
# serializers.py
class ChoiceSerializer(serializers.ModelSerializer):
choice_text = serializers.CharField()
votes = serializers.IntegerField()
class Meta:
fields = ('choice_text', 'votes')
class ChoiceUpdateSerializer(serializers.ModelSerializer):
choice_text = serializers.CharField()
class Meta:
fields = ('choice_text',)
# viewsets.py
def get_serializer_class(self):
if self.action == 'update':
return ChoiceUpdateSerializer
return ChoiceSerializerSince this having a separate list serializer is so common, the above can be avoided by using
the serializer_list_class class attribute provided by FormSchemaViewSetMixin.
The majority of the customization will occur inside serializer classes;
a real world example will often require custom create, update, to_representation, etc methods.
At the class level the serializer defines the form name:
ChoiceSerializer -> Choice
QuestionAndChoiceSerializer -> Question And Choice
Choice -> Choice
If a different title is required (or no title for nested forms), then the __init__ method can be updated:
def __init__(self, *args, **kwargs):
kwargs['label'] = 'New Form Title'
super().__init__(*args, **kwargs)The following is a list of parameters that can be added to individual fields which modifies
react-jsonschema-form functionality on the front-end.
Updates the form input title text. Can also be used to provide translations.
choice_text = serializers.CharField(label='What is the choice label?')The label can also be disabled completely via style:
choice_text = serializers.CharField(style={'ui:options': {'label': False}})Note: This does not work for list actions.
Setting read_only=True forces the encoder to skip it when building the form,
but still means the value gets sent in the formData:
choice_text = serializers.CharField(read_only=True)This is useful for custom widgets which rely on data not related to the form (e.g. id).
Setting required=False removes frontend validation of the field:
choice_text = serializers.CharField(required=False)Note: This has no effect on list actions.
Setting allow_null=True allows null values to be sent back in formData:
choice_text = serializers.CharField(allow_null=True)Note: This has no effect on list actions.
By setting a default value, a field is automatically set to required=False.
The formData object will also contain the default value if it is not provided one.
choice_text = serializers.CharField(default='example text')Note: This has no effect on list actions.
Built-in validators are automatically added to the schema. For instance:
char_text = serializers.CharField(min_length=5, max_length=10)Will result in the following schema portion:
{
"type": "string",
"title": "Char text",
"maxLength": 10,
"minLength": 5
}The following core validators have mapping to json schema (https://ajv.js.org/json-schema.html):
- MaxLengthValidator: maxLength
- MinLengthValidator: minLength
- MaxValueValidator: maximum
- MinValueValidator: minimum
- RegexValidator: pattern
Custom validators can not be mapped directly to Ajv schema properies so they are rather provided in the json ui schema as hints.
class MinSizeImageValidator:
message = _('Image is too small, must be 1KB minimum.')
code = 'image_min_1KB'
def __call__(self, value):
min_size = 1024 # 1KB
if value.size < min_size:
raise serializers.ValidationError(self.message, code=self.code)
image_field = serializers.ImageField(
required=True, validators=[MinSizeImageValidator]
)Will result in the following ui schema portion:
{
"ui:widget": "file",
"ui:custom-validators": [
{"code": "image_min_1KB", "message": "Image is too small, must be 1KB minimum."}
]
}The DRF style parameter is a dict and is therefore used for a number of different parameters.
There are a number of options
that react-jsonschema-form provides, many of which should work out-of-the-box,
although not all options will have been tested.
Since style params are applied last, they can overwrite other keys. Additionally, any key not
starting with schema: will be sent in the json payload, so this can be a good place to send
additional attributes for custom widgets.
The following are a list of valid (tested) keys and their uses.
Note: This can also be done at the serializer level:
def __init__(self, *args, **kwargs):
kwargs['style'] = {'ui:template': 'Card'}
super().__init__(*args, **kwargs)react-jsonschema-form allows for the use of dependencies
between multiple fields. This library has analogues for the four types (bidirectional, unidirectional, conditional, and
dynamic).
# Unidirectional
choice_text = serializers.CharField(
style={'schema:dependencies:simple': ['votes']}
)
# Bidirectional - note the different value styles, either is fine.
choice_text = serializers.CharField(
style={'schema:dependencies:simple': ['votes']}
)
votes = serializers.IntegerField(
default=0, style={'schema:dependencies:simple': 'choice_text'}
)
# Conditional
choice_text = serializers.CharField(
style={'schema:dependencies:conditional': ['votes']}
)
# Dynamic - must be an enum field, and only one choice per dependency,
# multiple choices can have the same dependencies (side-stepping this issue).
choice_text = serializers.ChoiceField(
choices=(('yes', 'Yes'), ('no', 'No')),
style={'schema:dependencies:dynamic': {'yes': ['votes'], 'no': None}}
)Since this framework doesn't cover 100% of the features of the react-jsonschema-form, it is possible to provide
dict objects which override the individual field properties:
choice_text = serializers.CharField(
style={
'schema:override': {'type': 'string', 'title': 'Overridden title'},
'uiSchema:override': {'ui:widget': 'updown'},
'column:override': {
'title': 'Overridden title',
'dataIndex': 'question_text',
'key': 'question_text',
},
'schema:dependencies:override': ['votes']
}
)Note: There is no validation around these overrides, so it is left up to the developer to ensure the resulting
schema is valid. For example, 'schema:dependencies:override' will not remove fields from the main 'properties' or
'required' objects. (This can be done by omitting the dependant field from the serializer or using read_only=True).
Sends the defaultSortOrder key with the list action serializer:
choice_text = serializers.CharField(, style={'schema:sort': 'ascend'})While the framework tries to provide sensible defaults for DRF fields, sometimes custom frontend widgets need to provide custom behaviour.
choice_text = serializers.CharField(style={
'ui:widget': 'textarea',
'schema:type': 'string',
'schema:enum': 'choices'
})More information can be found on
types
and enum.
enum can also be set to the string choices (as in the example),
to try and use the field.choices attribute.
The HTML placeholder text can be updated in the following way:
choice_text = serializers.CharField(style={'ui:placeholder': 'This a placeholder value'})When the ui:widget is set to textarea (see above), the rows can be set via:
choice_text = serializers.CharField(style={
'ui:widget': 'textarea',
'ui:options': {'rows': 8},
})An input field can be disabled in the following way:
choice_text = serializers.CharField(style={'ui:disabled': 'true'})Allows for custom fields to be added to ProcessingMixin.TYPE_MAP in this way:
DRF_REACT_TEMPLATE_TYPE_MAP = {
'ImageField': {'type': 'image', 'widget': 'file'},
}The default supported field types are:
{
'CharField': {'type': 'string'},
'IntegerField': {'type': 'integer', 'widget': 'updown'},
'FloatField': {'type': 'number', 'widget': 'updown'},
'DecimalField': {'type': 'number', 'widget': 'updown'},
'BooleanField': {'type': 'boolean'},
'DateTimeField': {'type': 'string', 'widget': 'date-time'},
'DateField': {'type': 'string', 'widget': 'date'},
'URLField': {'type': 'string', 'widget': 'uri'},
'ChoiceField': {'type': 'string', 'enum': 'choices'},
'EmailField': {'type': 'string', 'widget': 'email'},
'ListField': {'type': 'array'},
}This Repo uses Poetry, and setup for development is as simple as follows:
poetry init
pre-commit install