Validators¶

Validators are the same as any other subclass of input_algorithms.spec_base.Spec in that it has a normalise method that takes in meta and val and returns a new val.

It is convention that a validator subclasses input_algorithms.validators.Validator and implements a validate method. This means NotSpecified values are ignored and any specified value goes through the validate method.

It is the job of the validator to raise a sublcass of input_algorithms.errors.BadSpec if something is wrong, otherwise just return val.

Available Validators¶

has_either¶

Usage
has_either([key1, ..., keyn]).normalise(meta, val)
Will complain if the val.get(key, NotSpecified) returns NotSpecified for all the choices.

I.e. A valid dictionary has either one of the specified keys!

has_only_one_of¶

Usage
has_only_one_of([key1, ..., keyn]).normalise(meta, val)
Will complain if the val.get(key, NotSpecified) returns NotSpecified for all but one of the choices.

I.e. A valid dictionary must have exactly one of the specified keys!

either_keys¶

Usage
either_keys([k1, k2], [k3]).normalise(meta, val)
Will complain if the value is not a dictionary

Will complain if the keys from one group are mixed with keys from another group

Will complain if keys from one group appear without the rest of the keys in that group

Will not complain if unspecified keys are in the val

no_whitespace¶

Usage
no_whitespace().normalise(meta, val)
Raises an error if we can find the regex \s+ in the val.

no_dots¶

Usage
no_dots().normalise(meta, val)

# or

no_dots(reason="dots are evil!").normalise(meta, val)
This will complain if val contains any dots

It will use the reason to explain why it’s an error if it’s provided.

regexed¶

Usage
regexed(regex1, ..., regexn).normalise(meta, val)
This will match the val against all the regex``s and will complain if any of them fail, otherwise the ``val is returned.

deprecated_key¶

Usage
deprecated_key(key, reason).normalise(meta, val)
This will raise an error if val is nonempty and contains key. The error will use reason in it’s message.

choice¶

Usage

choice(choice1, ..., choicen).normalise(meta, val)