Tag registry¶

Each YAML() starts with its own registrations. Use yaml.register_class() for most tasks, either as the decorator @yaml.register_class or by calling it with the class; create a TagRegistry when several instances should share them.

from dataclasses import dataclass
from yamluna import YAML, TagRegistry


@dataclass
class Server:
    port: int


registry = TagRegistry()
writer = YAML(registry=registry)
reader = YAML(registry=registry)
writer.register_class(Server, source='myapp')

text = writer.dump(Server(8080))
print(reader.load(text))

Output:

Server(port=8080)

Registration options¶

yaml.register_class(cls, *, tag=None, source=None, to_yaml=None, from_yaml=None)

Option	Default	Use
`tag`	`cls.yaml_tag` or the class name	Choose the tag name
`source`	`cls.yaml_source` or the root package	Choose its namespace
`to_yaml`	The class hook, or its state	Write a custom representation
`from_yaml`	The class hook, or restore its state	Construct a custom value

Registering the same class again replaces its registration, including hooks. Two classes with the same explicit source and tag are ambiguous; give one a different source or tag.

For application-wide registration, yamluna.register_class() writes to yamluna.default_registry. Opt into it with YAML(registry=yamluna.default_registry); a plain YAML() stays independent.

See custom classes for complete save-and-load examples, including types you do not own such as numpy's ndarray.

Tag registry¶

Share classes between a writer and a reader¶

Registration options¶