Skip to main content

Add validator, serializer and deserializer to AnyBlok

Project description

Python versions Version status Build status Coverage Version status Documentation Status

Front Matter

Information about the AnyBlok / Marshmallow project.

Project Homepage

AnyBlok is hosted on github - the main project page is at Source code is tracked here using GIT.

Releases and project status are available on Pypi at

The most recent published version of this documentation should be at

Project Status

AnyBlok with Marshmallow is currently in beta status and is expected to be fairly stable. Users should take care to report bugs and missing features on an as-needed basis. It should be expected that the development version may be required for proper implementation of recently repaired issues in between releases;


Install released versions of AnyBlok from the Python package index with pip or a similar tool:

pip install anyblok_marshmallow

Installation via source distribution is via the script:

python install

Installation will add the anyblok commands to the environment.

Unit Test

Run the test with pytest:

pip install pytest pytest-cov
ANYBLOK_DRIVER_NAME=postgres ANYBLOK_DATABASE_NAME=mybase pytest anyblok_marshmallow/tests


AnyBlok works with Python 3.6 and later. The install process will ensure that AnyBlok, marshmallow >= 3.2.0 and marshmallow-sqlalchemy are installed, in addition to other dependencies. The latest version of them is strongly recommended.

Contributing (hackers needed!)

Anyblok / Marshmallow is at a very early stage, feel free to fork, talk with core dev, and spread the word!


Jean-Sébastien Suzanne


Anybox team:

  • Jean-Sébastien Suzanne

Sensee team:

  • Franck Bret


Declare your AnyBlok model

from anyblok.column import Integer, String
from anyblok.relationship import Many2One, Many2Many
from anyblok import Declarations

class City:

    id = Integer(primary_key=True)
    name = String(nullable=False)
    zipcode = String(nullable=False)

    def __repr__(self):
        return '<City(name={!r})>'.format(self=self)

class Tag:

    id = Integer(primary_key=True)
    name = String(nullable=False)

    def __repr__(self):
        return '<Tag(name={!r})>'.format(self=self)

class Customer:
    id = Integer(primary_key=True)
    name = String(nullable=False)
    tags = Many2Many(model=Declarations.Model.Tag)

    def __repr__(self):
        return '<Customer(name={!r}, '

class Address:

    id = Integer(primary_key=True)
    street = String(nullable=False)
    city = Many2One(model=Declarations.Model.City, nullable=False)
    customer = Many2One(
        model=Declarations.Model.Customer, nullable=False,


The AnyBlok model must be declared in a blok

Declare your schema

from anyblok_marshmallow import SchemaWrapper, PostLoadSchema, Nested

class CitySchema(SchemaWrapper):
    model = 'Model.City'

class TagSchema(SchemaWrapper):
    model = 'Model.Tag'

class AddressSchema(SchemaWrapper):
    model = 'Model.Address'

    class Schema:
        # Add some marshmallow fields or behaviours

        # follow the relationship Many2One and One2One
        city = Nested(CitySchema)

class CustomerSchema(SchemaWrapper):
    model = 'Model.Customer'
    # optionally attach an AnyBlok registry
    # to use for serialization, desarialization and validation
    registry = registry

    class Schema(PostLoadSchema):
        # follow the relationship One2Many and Many2Many
        # - the many=True is required because it is *2Many
        # - exclude is used to forbid the recurse loop
        addresses = Nested(AddressSchema, many=True, exclude=('customer', ))
        tags = Nested(TagSchema, many=True)

customer_schema = CustomerSchema()


New in version 1.1.0 the Nested field must come from anyblok_marshmallow, because marshmallow cache the Nested field with the context. And the context is not propagated again if it changed


Ref in version 1.4.0, post_load_return_instance was replaced by the mixin class PostLoadSchema


Ref in version 2.1.0, ModelSchema was replaced by SchemaWrapper. This action break the compatibility with the previous version, but allow to follow the upgrade of marshmallow

(De)serialize your data and validate it

customer = registry.Customer.insert(name="JS Suzanne")
tag1 = registry.Tag.insert(name="tag 1")
tag2 = registry.Tag.insert(name="tag 2")
rouen = registry.City.insert(name="Rouen", zipcode="76000")
paris = registry.City.insert(name="Paris", zipcode="75000")
registry.Address.insert(customer=customer, street="Somewhere", city=rouen)
registry.Address.insert(customer=customer, street="Another place", city=paris)

dump_data = customer_schema.dump(customer).data
# {
#     'id': 1,
#     'name': 'JS Suzanne',
#     'tags': [
#         {
#             'id': 1,
#             'name': 'tag 1',
#         },
#         {
#             'id': 2,
#             'name': 'tag 2',
#         },
#     ],
#     'addresses': [
#         {
#             'id': 1
#             'street': 'Somewhere'
#             'city': {
#                 'id': 1,
#                 'name': 'Rouen',
#                 'zipcode': '76000',
#             },
#         },
#         {
#             'id': 2
#             'street': 'Another place'
#             'city': {
#                 'id': 2,
#                 'name': 'Paris',
#                 'zipcode': '75000',
#             },
#         },
#     ],
# }

# <Customer(name='JS Suzanne' tags=[<Tag(name='tag 1')>, <Tag (name='tag 2')>])>

errors = customer_schema.validate(dump_data)
# dict with all the validating errors


We have an instance of the model cause of the mixin PostLoadSchema

Give the registry

The schema need to have the registry.

If no registry found when the de(serialization) or validation then the RegistryNotFound exception will be raised.

Add the registry by the class attribute

This is the solution given in the main exemple:

class CustomerSchema(SchemaWrapper):
    model = 'Model.Customer'
    registry = registry

Add the registry during init

This solution is use during the instanciation

customer_schema = CustomerSchema(registry=registry)

Add the registry by the context

This solution is use during the instanciation or after

customer_schema = CustomerSchema(context={'registry': registry})


customer_schema = CustomerSchema()
customer_schema.context['registry'] = registry

Add the registry when the de(serialization or validator is called

customer_schema.dumps(customer, registry=registry)
customer_schema.dump(customer, registry=registry)
customer_schema.loads(dump_data, registry=registry)
customer_schema.load(dump_data, registry=registry)
customer_schema.validate(dump_data, registry=registry)

model option

This option add in the model name. As the registry, this option can be passed by definition, initialization, context or during the call of the (de)serialization / validation

class AnySchema(SchemaWrapper):
    model = "Model.Customer"


any_schema = AnySchema(model="Model.customer")


any_schema.context['model'] = "Model.Customer"


any_schema.dumps(instance, model="Model.Customer")
any_schema.dump(instance, model="Model.Customer")
any_schema.loads(dump_data, model="Model.Customer")
any_schema.load(dump_data, model="Model.Customer")
any_schema.validate(dump_data, model="Model.Customer")

only_primary_key option

This option add in the only argument the primary keys of the model. As the registry, this option can be passed by definition, initialization, context or during the call of the (de)serialization / validation

class CustomerSchema(SchemaWrapper):
    model = "Model.Customer"
    only_primary_key = True


customer_schema = CustomerSchema(only_primary_key=True)


customer_schema.context['only_primary_key'] = True


customer_schema.dumps(instance, only_primary_key=True)
customer_schema.dump(instance, only_primary_key=True)
customer_schema.loads(dump_data, only_primary_key=True)
customer_schema.load(dump_data, only_primary_key=True)
customer_schema.validate(dump_data, only_primary_key=True)

required_fields option

This option force the generated fields, and only them to be requried.

class CustomerSchema(SchemaWrapper):
    model = "Model.Customer"
    required_fields = True
    # or required_fields = [ list of fieldname ]


customer_schema = CustomerSchema(required_fields=True)


customer_schema.context['required_fields'] = True


customer_schema.loads(dump_data, required_fields=True)
customer_schema.load(dump_data, required_fields=True)
customer_schema.validate(dump_data, required_fields=True)


All the attributes can take True or the list of the fieldname to be required

Use the field JsonCollection

This field allow the schema to inspect an AnyBlok.fields.Json in an any specific instance to validate the value.

AnyBlok models:

class Template:
    name = anyblok.column.String(primary_key=True)
    properties = anyblok.column.Json(defaumt={})

class SaleOrder:
    id = anyblok.column.Integer(primary_key=True)
    number = anyblok.column.Integer(nullable=False)
    discount = anyblok.column.Integer()

AnyBlok / Marchmallow schema:

class SaleOrderSchema(SchemaWrapper):
    model = 'Model.SaleOrder'

    class Schema:
        discount = JsonCollection(


goodcustomer = registry.Template.insert(
    name='Good customer',
    properties={'allowed_discount': [10, 15, 30]
customer = registry.Template.insert(
    properties={'allowed_discount': [0, 5, 10]
badcustomer = registry.Template.insert(
    name='Bad customer',
    properties={'allowed_discount': [0]

schema = SaleOrderSchema(registry=registry)


data = schema.load(
    instances={'default': goodcustomer}


data = schema.load(
    instances={'default': customer}
==> error = {}


data = schema.load(
    instances={'default': badcustomer}
==> error = {'discount': ['Not a valid choice']}


2.3.0 (2019-10-31)

  • Fixed du of Marshmallow 3.2.1 release

2.2.3 (2019-05-06)

  • Fixed du of Marshmallow 3.0.0RC5 release
  • Refactored unittest from nosetest to pytest

2.2.2 (2018-12-06)

  • Fixed du of Marshmallow 3.0.0RC1 release

2.2.1 (2018-10-26)

  • Fixed du of Marshmallow 3.0.0b19 release

2.2.0 (2018-10-17)

  • Fixed the convertion of type between AnyBlok.Column and marshmallow.Field

2.1.0 (2018-09-26)

  • Fixed the compatibility with Marshmallow > 3.0.0b8
  • Removed ModelSchema class
  • Added SchemaWrapper, this is the best way to defined a generated schema with the marshmallow_sqlalchemy library


This version break the compatibility with previous version, in the only goal to be adapted with the latest version of marshmallow

2.0.1 (2018-06-07)

  • Fix required_field put allow_none to False

2.0.0 (2018-05-30)

  • Add JsonCollection field, Allow to add a check in function of an collection stored in a AnyBlok.fields.Json
  • Add Text field, to represent an anyblok.column.Text
  • Migration of the code and unit test to marshmallow 3.0.0
  • Add Email matching for anyblok.column.Email
  • Add URL matching for anyblok.column.URL
  • Add PhoneNumber matching for anyblok.column.PhoneNumber
  • Add Country matching for anyblok.column.Country
  • Add required_fields option
  • Add InstanceField

1.4.0 (2018-04-07)

  • Replace post_load_return_instance method by PostLoadSchema class
  • In the case of the field Selection, the validator OneOf is applied with the available values come from the AnyBlok columns
  • Replace marshmallow_sqlalchemy.fields.Related by anyblok_marshmallow.fields.Nested. The goal is to improve the consistent between all field in the schema

1.3.0 (2017-12-23)

  • [ADD] unittest on some case
  • [FIX] AnyBlok field.Function is return as MarshMallow fields.Raw
  • [ADD] fields.File, type to encode and decode to/from base 64

1.2.0 (2017-11-30)

  • [REF] decrease complexity
  • [IMP] Add validates_schema on ModelSchema to automaticly check if the field exist on the model

1.1.0 (2017-11-02)

  • Add option put only the primary keys
  • Fix the Front page
  • REF model option, can be given by another way than Meta
  • Put RegistryNotFound in exceptions
  • Add Nested field, this field is not and have not to be cached

1.0.2 (2017-10-25)

  • Fix pypi documentation

1.0.0 (2017-10-24)

  • Add marshmallow schema for AnyBlok for:
    • Serialization
    • Deserialization
    • Validation

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for anyblok_marshmallow, version 2.3.0
Filename, size File type Python version Upload date Hashes
Filename, size anyblok_marshmallow-2.3.0-py3-none-any.whl (34.5 kB) File type Wheel Python version 3.6 Upload date Hashes View
Filename, size anyblok_marshmallow-2.3.0.tar.gz (27.9 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page