Karana.KUtils.kdpy_model_io

Load and identify serialized kdFlex models for the kdpy launcher.

The launcher accepts one --from-model path without requiring the user to name its serialized root type. This module owns that probing policy and normalizes every successful load into LoadedModel, allowing the rest of kdpy to work with one small interface.

Attributes

MODEL_LOADERS

Classes

LoadedModel

Normalize a successfully loaded root model for launcher orchestration.

Functions

loadModel(→ LoadedModel)

Load the first supported kdFlex model type that accepts the file.

Module Contents

class Karana.KUtils.kdpy_model_io.LoadedModel

Normalize a successfully loaded root model for launcher orchestration.

state_target is the object reset from --from-state. It may be a subtree even when sim.mb is the complete multibody. model_target is retained until output serialization so kdpy writes the same kind of model that it read. The caller owns sim and therefore controls the lifetime of every native object reachable through the other fields.

sim: Karana.KUtils.Sim.Sim

model_type: str

state_target: Karana.Dynamics.SubTree

model_target: Any

Karana.KUtils.kdpy_model_io.MODEL_LOADERS: tuple[tuple[str, Callable[[pathlib.Path], tuple[Karana.KUtils.Sim.Sim, Karana.Dynamics.SubTree, Any]]], Ellipsis]

Karana.KUtils.kdpy_model_io.loadModel(model_file: pathlib.Path) → LoadedModel

Load the first supported kdFlex model type that accepts the file.

Each kdFlex fromFile call performs schema validation. A schema mismatch is therefore the dispatch signal rather than a fatal error. Failed probes can allocate native objects and register cleanup callbacks, so each attempt is treated as a small transaction: remember the callback boundary, try the loader, and unwind only callbacks introduced by that attempt on failure.