nv_ingest_api.util.introspection package#

Submodules#

nv_ingest_api.util.introspection.class_inspect module#

nv_ingest_api.util.introspection.class_inspect.find_pydantic_config_schema(
actor_class: Type,
base_class_to_find: Type,
param_name: str = 'config',
) → Type[BaseModel] | None[source]#

Introspects a class’s MRO to find a Pydantic model in its __init__ signature.

This function is designed to find the specific Pydantic configuration model for a pipeline actor, which might be a direct class or a proxy object.

Parameters:

  • actor_class (Type) – The actor class or proxy object to inspect.

  • base_class_to_find (Type) – The specific base class (e.g., RaySource, RayStage) to look for when resolving the true actor class from a proxy.

  • param_name (str, optional) – The name of the __init__ parameter to inspect for the Pydantic schema, by default “config”.

Returns:

The Pydantic BaseModel class if found, otherwise None.

Return type:

Optional[Type[BaseModel]]

nv_ingest_api.util.introspection.class_inspect.find_pydantic_config_schema_for_callable(
callable_fn: Callable,
param_name: str = 'stage_config',
) → Type[BaseModel] | None[source]#

Introspects a callable’s signature to find a Pydantic model parameter.

This function is designed to find the specific Pydantic configuration model for a pipeline callable function.

Parameters:

  • callable_fn (Callable) – The callable function to inspect.

  • param_name (str, optional) – The name of the parameter to inspect for the Pydantic schema, by default “stage_config”.

Returns:

The Pydantic BaseModel class if found, otherwise None.

Return type:

Optional[Type[BaseModel]]

nv_ingest_api.util.introspection.class_inspect.find_pydantic_config_schema_unified(
target: Type | Callable,
base_class_to_find: Type | None = None,
param_name: str = 'config',
) → Type[BaseModel] | None[source]#

Unified function to find Pydantic schema for either classes or callables.

Parameters:

  • target (Union[Type, Callable]) – The class or callable to inspect.

  • base_class_to_find (Optional[Type], optional) – The specific base class to look for when resolving actor classes from proxies. Only used for class inspection.

  • param_name (str, optional) – The name of the parameter to inspect for the Pydantic schema. For classes: defaults to “config” For callables: should be “stage_config”

Returns:

The Pydantic BaseModel class if found, otherwise None.

Return type:

Optional[Type[BaseModel]]

nv_ingest_api.util.introspection.function_inspect module#

Utilities for introspecting and analyzing UDF function specifications.

nv_ingest_api.util.introspection.function_inspect.infer_udf_function_name(udf_function: str) → str | None[source]#

Attempts to infer the UDF function name from the provided function string.

Supports three formats: 1. Inline function: ‘def my_func(control_message): …’ -> ‘my_func’ 2. Import path: ‘my_module.my_function’ -> ‘my_function’ 3. File path: ‘/path/to/file.py:function_name’ -> ‘function_name’

Parameters:

udf_function (str) – The UDF function string.

Returns:

The inferred UDF function name, or None if inference is not possible.

Return type:

Optional[str]

Examples

>>> infer_udf_function_name("def my_custom_func(control_message): pass")
'my_custom_func'

>>> infer_udf_function_name("my_module.submodule.process_data")
'process_data'

>>> infer_udf_function_name("/path/to/script.py:custom_function")
'custom_function'

>>> infer_udf_function_name("/path/to/script.py")
None

Module contents#