Schemas and predicates

Describe a shape with Python values

A dict maps field names to schemas. A one-element list describes each item in a list. A Python type requires that exact type. PEP 604 unions combine types. Optional wraps an optional field.

Predicates receive the original value and accept it when they return a truthy result. They do not transform values. A predicate should check its own input type before comparing or indexing it.

from zodify import validate

schema = {"users": [{"name": str}], "enabled": bool}
assert validate(schema, {"users": [{"name": "Ada"}], "enabled": True})["users"][0]["name"] == "Ada"
# A predicate accepts the original value when it returns a truthy result.
assert validate({"port": lambda x: type(x) is int and 1 <= x <= 65535}, {"port": 8080}) == {"port": 8080}
print("schemas: passed")

Keep schema boundaries explicit

Top-level data must be a dict (or a supported Schema declaration at the schema boundary). Lists with zero or multiple schema items are invalid. Malformed schemas may raise TypeError. Validation has a max_depth option, defaulting to 32; cyclic models and arbitrary typing expressions are not promised.

Dict-shaped and list-shaped validation creates result containers, but leaf objects and defaults can remain shared. Validation is not a universal deep copy.

• Ownership and defaults →