Skip to content

Model Config

Behaviour of pydantic can be controlled via the Config class on a model or a pydantic dataclass.

from pydantic import BaseModel, ValidationError


class Model(BaseModel):
    v: str

    class Config:
        max_anystr_length = 10
        error_msg_templates = {
            'value_error.any_str.max_length': 'max_length:{limit_value}',
        }


try:
    Model(v='x' * 20)
except ValidationError as e:
    print(e)
    """
    1 validation error for Model
    v
      max_length:10 (type=value_error.any_str.max_length; limit_value=10)
    """

(This script is complete, it should run "as is")

Also, you can specify config options as model class kwargs:

from pydantic import BaseModel, ValidationError, Extra


class Model(BaseModel, extra=Extra.forbid):
    a: str


try:
    Model(a='spam', b='oh no')
except ValidationError as e:
    print(e)
    """
    1 validation error for Model
    b
      extra fields not permitted (type=value_error.extra)
    """

(This script is complete, it should run "as is")

Similarly, if using the @dataclass decorator:

from datetime import datetime

from pydantic import ValidationError
from pydantic.dataclasses import dataclass


class MyConfig:
    max_anystr_length = 10
    validate_assignment = True
    error_msg_templates = {
        'value_error.any_str.max_length': 'max_length:{limit_value}',
    }


@dataclass(config=MyConfig)
class User:
    id: int
    name: str = 'John Doe'
    signup_ts: datetime = None


user = User(id='42', signup_ts='2032-06-21T12:00')
try:
    user.name = 'x' * 20
except ValidationError as e:
    print(e)
    """
    1 validation error for User
    name
      max_length:10 (type=value_error.any_str.max_length; limit_value=10)
    """

(This script is complete, it should run "as is")

Options

title
the title for the generated JSON Schema
anystr_strip_whitespace
whether to strip leading and trailing whitespace for str & byte types (default: False)
anystr_upper
whether to make all characters uppercase for str & byte types (default: False)
anystr_lower
whether to make all characters lowercase for str & byte types (default: False)
min_anystr_length
the min length for str & byte types (default: 0)
max_anystr_length
the max length for str & byte types (default: None)
validate_all
whether to validate field defaults (default: False)
extra
whether to ignore, allow, or forbid extra attributes during model initialization. Accepts the string values of 'ignore', 'allow', or 'forbid', or values of the Extra enum (default: Extra.ignore). 'forbid' will cause validation to fail if extra attributes are included, 'ignore' will silently ignore any extra attributes, and 'allow' will assign the attributes to the model.
allow_mutation
whether or not models are faux-immutable, i.e. whether __setattr__ is allowed (default: True)

frozen

Warning

This parameter is in beta

setting frozen=True does everything that allow_mutation=False does, and also generates a __hash__() method for the model. This makes instances of the model potentially hashable if all the attributes are hashable. (default: False)
use_enum_values
whether to populate models with the value property of enums, rather than the raw enum. This may be useful if you want to serialise model.dict() later (default: False)
fields
a dict containing schema information for each field; this is equivalent to using the Field class, except when a field is already defined through annotation or the Field class, in which case only alias, include, exclude, min_length, max_length, regex, gt, lt, gt, le, multiple_of, max_digits, decimal_places, min_items, max_items, unique_items and allow_mutation can be set (for example you cannot set default of default_factory) (default: None)
validate_assignment
whether to perform validation on assignment to attributes (default: False)
allow_population_by_field_name
whether an aliased field may be populated by its name as given by the model attribute, as well as the alias (default: False)

Note

The name of this configuration setting was changed in v1.0 from allow_population_by_alias to allow_population_by_field_name.

error_msg_templates
a dict used to override the default error message templates. Pass in a dictionary with keys matching the error messages you want to override (default: {})
arbitrary_types_allowed
whether to allow arbitrary user types for fields (they are validated simply by checking if the value is an instance of the type). If False, RuntimeError will be raised on model declaration (default: False). See an example in Field Types.
orm_mode
whether to allow usage of ORM mode
getter_dict
a custom class (which should inherit from GetterDict) to use when decomposing arbitrary classes for validation, for use with orm_mode; see Data binding.
alias_generator
a callable that takes a field name and returns an alias for it; see the dedicated section
keep_untouched
a tuple of types (e.g. descriptors) for a model's default values that should not be changed during model creation and will not be included in the model schemas. Note: this means that attributes on the model with defaults of this type, not annotations of this type, will be left alone.
schema_extra
a dict used to extend/update the generated JSON Schema, or a callable to post-process it; see schema customization
json_loads
a custom function for decoding JSON; see custom JSON (de)serialisation
json_dumps
a custom function for encoding JSON; see custom JSON (de)serialisation
json_encoders
a dict used to customise the way types are encoded to JSON; see JSON Serialisation
underscore_attrs_are_private
whether to treat any underscore non-class var attrs as private, or leave them as is; see Private model attributes
copy_on_model_validation
string literal to control how models instances are processed during validation, with the following means (see #4093 for a full discussion of the changes to this field):
  • 'none' - models are not copied on validation, they're simply kept "untouched"
  • 'shallow' - models are shallow copied, this is the default
  • 'deep' - models are deep copied
smart_union
whether pydantic should try to check all types inside Union to prevent undesired coercion; see the dedicated section
post_init_call
whether stdlib dataclasses __post_init__ should be run before (default behaviour with value 'before_validation') or after (value 'after_validation') parsing and validation when they are converted.
allow_inf_nan
whether to allows infinity (+inf an -inf) and NaN values to float fields, defaults to True, set to False for compatibility with JSON, see #3994 for more details, added in V1.10

Change behaviour globally

If you wish to change the behaviour of pydantic globally, you can create your own custom BaseModel with custom Config since the config is inherited

from pydantic import BaseModel as PydanticBaseModel


class BaseModel(PydanticBaseModel):
    class Config:
        arbitrary_types_allowed = True


class MyClass:
    """A random class"""


class Model(BaseModel):
    x: MyClass

(This script is complete, it should run "as is")

Alias Generator

If data source field names do not match your code style (e. g. CamelCase fields), you can automatically generate aliases using alias_generator:

from pydantic import BaseModel


def to_camel(string: str) -> str:
    return ''.join(word.capitalize() for word in string.split('_'))


class Voice(BaseModel):
    name: str
    language_code: str

    class Config:
        alias_generator = to_camel


voice = Voice(Name='Filiz', LanguageCode='tr-TR')
print(voice.language_code)
#> tr-TR
print(voice.dict(by_alias=True))
#> {'Name': 'Filiz', 'LanguageCode': 'tr-TR'}

(This script is complete, it should run "as is")

Here camel case refers to "upper camel case" aka pascal case e.g. CamelCase. If you'd like instead to use lower camel case e.g. camelCase, instead use the to_lower_camel function.

Alias Precedence

Warning

Alias priority logic changed in v1.4 to resolve buggy and unexpected behaviour in previous versions. In some circumstances this may represent a breaking change, see #1178 and the precedence order below for details.

In the case where a field's alias may be defined in multiple places, the selected value is determined as follows (in descending order of priority):

  1. Set via Field(..., alias=<alias>), directly on the model
  2. Defined in Config.fields, directly on the model
  3. Set via Field(..., alias=<alias>), on a parent model
  4. Defined in Config.fields, on a parent model
  5. Generated by alias_generator, regardless of whether it's on the model or a parent

Note

This means an alias_generator defined on a child model does not take priority over an alias defined on a field in a parent model.

For example:

from pydantic import BaseModel, Field


class Voice(BaseModel):
    name: str = Field(None, alias='ActorName')
    language_code: str = None
    mood: str = None


class Character(Voice):
    act: int = 1

    class Config:
        fields = {'language_code': 'lang'}

        @classmethod
        def alias_generator(cls, string: str) -> str:
            # this is the same as `alias_generator = to_camel` above
            return ''.join(word.capitalize() for word in string.split('_'))


print(Character.schema(by_alias=True))
"""
{
    'title': 'Character',
    'type': 'object',
    'properties': {
        'ActorName': {'title': 'Actorname', 'type': 'string'},
        'lang': {'title': 'Lang', 'type': 'string'},
        'Mood': {'title': 'Mood', 'type': 'string'},
        'Act': {'title': 'Act', 'default': 1, 'type': 'integer'},
    },
}
"""

(This script is complete, it should run "as is")

Smart Union

By default, as explained here, pydantic tries to validate (and coerce if it can) in the order of the Union. So sometimes you may have unexpected coerced data.

from typing import Union

from pydantic import BaseModel


class Foo(BaseModel):
    pass


class Bar(BaseModel):
    pass


class Model(BaseModel):
    x: Union[str, int]
    y: Union[Foo, Bar]


print(Model(x=1, y=Bar()))
#> x='1' y=Foo()
from pydantic import BaseModel


class Foo(BaseModel):
    pass


class Bar(BaseModel):
    pass


class Model(BaseModel):
    x: str | int
    y: Foo | Bar


print(Model(x=1, y=Bar()))
#> x='1' y=Foo()

(This script is complete, it should run "as is")

To prevent this, you can enable Config.smart_union. Pydantic will then check all allowed types before even trying to coerce. Know that this is of course slower, especially if your Union is quite big.

from typing import Union

from pydantic import BaseModel


class Foo(BaseModel):
    pass


class Bar(BaseModel):
    pass


class Model(BaseModel):
    x: Union[str, int]
    y: Union[Foo, Bar]

    class Config:
        smart_union = True


print(Model(x=1, y=Bar()))
#> x=1 y=Bar()
from pydantic import BaseModel


class Foo(BaseModel):
    pass


class Bar(BaseModel):
    pass


class Model(BaseModel):
    x: str | int
    y: Foo | Bar

    class Config:
        smart_union = True


print(Model(x=1, y=Bar()))
#> x=1 y=Bar()

(This script is complete, it should run "as is")

Warning

Note that this option does not support compound types yet (e.g. differentiate List[int] and List[str]). This option will be improved further once a strict mode is added in pydantic and will probably be the default behaviour in v2!

from typing import List, Union

from pydantic import BaseModel


class Model(BaseModel, smart_union=True):
    x: Union[List[str], List[int]]


# Expected coercion
print(Model(x=[1, '2']))
#> x=['1', '2']

# Unexpected coercion
print(Model(x=[1, 2]))
#> x=['1', '2']
from typing import Union

from pydantic import BaseModel


class Model(BaseModel, smart_union=True):
    x: Union[list[str], list[int]]


# Expected coercion
print(Model(x=[1, '2']))
#> x=['1', '2']

# Unexpected coercion
print(Model(x=[1, 2]))
#> x=['1', '2']
from pydantic import BaseModel


class Model(BaseModel, smart_union=True):
    x: list[str] | list[int]


# Expected coercion
print(Model(x=[1, '2']))
#> x=['1', '2']

# Unexpected coercion
print(Model(x=[1, 2]))
#> x=['1', '2']

(This script is complete, it should run "as is")