Valedictory validates the data in dicts. It is designed for use in API validation, and other situations where you are receiving structured JSON data as opposed to key-value POST form data. It takes in a dict of data (probably obtained from a JSON POST request), and validates that data against some fields.

Validators are defined as classes. Declare fields on a Validator class. Once constructed, Validator instances are immutable.

from valedictory import Validator, fields, InvalidDataException

class PersonValidator(Validator):
    name = fields.CharField()
    height = fields.IntegerField()
    date_of_birth = fields.DateField()

person_validator = PersonValidator()

A Python dict can be checked to see if it conforms to this validator. The dict can come from a JSON POST request, or a configuration file, or any other external data source that needs validation and cleaning. The cleaned data will be returned. Validator classes will return a dict of cleaned data. Each field type may transform the data as part of cleaning it. For example, the DateField will transform the data into a instance.

input_data = json.loads(request.body)

    # cleaned_data will be a dict of cleaned, validated data
    cleaned_data = person_validator.clean(input_data)

    # Do something with the returned data

except InvalidDataException as errors:
    # The data did not pass validation
    for path, message in errors.flatten():
        # This will print something like 'name: This field is required'
        print("{0}: {1}".format('.'.join(path), message))

Validators can be nested, allowing dicts of arbitrary complexity:

class ArticleValidator(Validator):
    content = fields.CharField()
    published = fields.DateTimeField()
    author = fields.NestedValidator(PersonValidator())
    tags = fields.ListField(fields.CharField())

# Some example data that would pass validation:
data = {
    "content": "An interesting article",
    "published": "2018-06-13T1:44:00+10:00",
    "author": {
        "name": "Alex Smith",
        "height": 175,
        "date_of_birth": "1990-03-26",
    "tags": ["humour", "interesting", "clickbait"],

Read the documentation for more information.