owl22python

Bidirectional OWL2 ⇄ Python over a renderable block/class system. Make blocks, make classes from data, and keep an OWL ontology for everything you make. Dependency-light (pydantic + jinja2), no network, no external services.

What's in it

• pydantic_stack — RenderablePiece (a block: a renderable typed pydantic model) + MetaStack (a stack of blocks that renders to one output).
• template_mixins — TemplateAttributeMixin / TemplateMethodMixin: class-level templating — add_attribute, add_method, execute_template_sequence — grow a class's attributes and methods from data at runtime.
• templated_class — TemplatedClass: a BaseModel + the mixins, configurable from a dict, that can emit real .py source of itself (generate_class_template / generate_code_files).
• owl22python(xml) / python22owl(cls) — OWL2/XML ⇄ a generated RenderablePiece class (one typed field per OWL DataProperty). Bidirectional: every class knows its own OWL, so you can define a thing in OWL or in Python and round-trip it.

Example

from owl22python import owl22python, python22owl, RenderablePiece, MetaStack, TemplatedClass

# OWL -> a live Python class
M = owl22python(EXAMPLE_OWL2_XML)["ModelClass"]
owl = M(species="Great Horned Owl", wingspan_cm=120.0, nocturnal=True)
print(owl.render())               # -> OWL individual (RDF/XML) + typed triples

# Python -> OWL (the inverse): the class knows its own ontology
print(python22owl(M))             # -> OWL2/XML schema for M

MIT.