Validation

didactic checks three things at construction time:

1. Each value satisfies its annotation (handled by the type translation layer).
2. Each @validates-decorated method returns True.
3. Each entry in __axioms__ evaluates to True against the field environment.

A failure in any of the three raises didactic.api.ValidationError with one ValidationErrorEntry per failure.

Per-field validators

Decorate a method with @dx.validates to attach a check to a named field:

import didactic.api as dx


class User(dx.Model):
    id: str
    email: str

    @dx.validates("email")
    def _email_must_have_at(self, value: str) -> bool:
        return "@" in value

The validator runs after type translation. It receives the decoded value and returns a bool. A return of False raises ValidationError; raising directly from inside the validator works too, in which case the raised exception is wrapped.

Class-level axioms

Axioms are conditions on the field environment as a whole. Use them when the rule depends on more than one field:

class Range(dx.Model):
    low: int
    high: int

    __axioms__ = [
        dx.axiom("low >= 0", message="low must be non-negative"),
        dx.axiom("low <= high", message="low must not exceed high"),
    ]

Each axiom is parsed via panproto.parse_expr and evaluated against {field_name: decoded_value} at construction time. Axioms inherited from a base class are checked too.

The expression syntax is the panproto-Expr surface syntax: lambda expressions, comparisons, boolean connectives, arithmetic, and a small set of builtins. The most common forms (comparisons, ==, /=, &&, ||, not, arithmetic) are listed in Guides > Axioms.

Inspecting failures

A ValidationError carries a tuple of entries plus a reference to the Model class:

try:
    User(id="u1", email="not-an-email")
except dx.ValidationError as exc:
    for entry in exc.entries:
        print(entry.loc, entry.type, entry.msg)
    # ('email',) validates_failed `email` validator returned False

Per-entry types in v0.0.1: missing_required, extra_field, type_error, validates_failed, axiom_failed. New types may be added in subsequent releases; treat the set as open.

Next: serialisation.