morpheus.utils.type_utils#

Utility functions for working with types.

Functions

`cpp_mode_to_exec_mode`()	Return the execution mode based on the configuration of the global `morpheus.config.CppConfig` singleton.
`df_type_str_to_exec_mode`(df_type_str)	Return the appropriate execution mode based on the DataFrame type string.
`df_type_str_to_pkg`(df_type_str)	Import and return the appropriate DataFrame package based on the DataFrame type string.
`exec_mode_to_df_type_str`(execution_mode)	Return the appropriate DataFrame type string based on the execution mode.
`flatten_types`(type_list)	Flattens a list of types, removing any union and `typing.Any` types.
`get_array_pkg`([execution_mode])	Return the appropriate array package (CuPy for GPU, NumPy for CPU) based on the execution mode.
`get_df_class`()	Return the appropriate DataFrame `selector` which can be either an `ExecutionMode` instance, a DataFrame type string, or `None`.
`get_df_pkg`()	Return the appropriate DataFrame package based on `selector` which can be either an `ExecutionMode` instance, a DataFrame type string, or `None`.
`get_df_pkg_from_obj`(obj)	Return the appropriate DataFrame package based on a given object (DataFrame, Series, RangeIndex etc...).
`get_full_qualname`(klass)	Returns the fully qualified name of a class.
`greatest_ancestor`(*cls_list)	Returns the greatest common ancestor of the classes in the class list
`is_cudf_type`(obj)	Check if a given object (DataFrame, Series, RangeIndex etc...) is a cuDF type.
`is_dataframe`(obj)	Check if a given object is a pandas or cudf DataFrame.
`is_union_type`(type_)	Returns True if the type is a `typing.Union` or a `types.UnionType`.
`pretty_print_type_name`(type_)	Determines a good label to use for a type.
`unpack_tuple`()
`unpack_union`()

cpp_mode_to_exec_mode()[source]#

Return the execution mode based on the configuration of the global morpheus.config.CppConfig singleton.

Returns:

morpheus.config.ExecutionMode: The execution mode.

df_type_str_to_exec_mode(df_type_str)[source]#

Return the appropriate execution mode based on the DataFrame type string.

Parameters:

df_type_strmorpheus.utils.type_aliases.DataFrameModule: The DataFrame type string.

Returns:

morpheus.config.ExecutionMode: The associated execution mode based on the DataFrame type string.

df_type_str_to_pkg(df_type_str)[source]#

Import and return the appropriate DataFrame package based on the DataFrame type string.

Parameters:

df_type_strmorpheus.utils.type_aliases.DataFrameModule: The DataFrame type string.

Returns:

types.ModuleType: The associated DataFrame package based on the DataFrame type string.

exec_mode_to_df_type_str(execution_mode)[source]#

Return the appropriate DataFrame type string based on the execution mode.

Parameters:

execution_modemorpheus.config.ExecutionMode: The execution mode.

Returns:

morpheus.utils.type_aliases.DataFrameModule: The associated DataFrame type string based on the execution mode.

flatten_types(type_list)[source]#

Flattens a list of types, removing any union and typing.Any types.

get_array_pkg(execution_mode=None)[source]#

Return the appropriate array package (CuPy for GPU, NumPy for CPU) based on the execution mode.

When None the execution mode is determined by the global morpheus.config.CppConfig singleton.

Parameters:

execution_modemorpheus.config.ExecutionMode, optional: The execution mode, by default None.

Returns:

types.ModuleType: The associated array package based on the execution mode.

get_df_class( selector: Literal['cudf', 'pandas'] = None, ) → type[DataFrameType][source]#

get_df_class( selector: ExecutionMode = None, ) → type[DataFrameType]

Return the appropriate DataFrame selector which can be either an ExecutionMode instance, a DataFrame type string, or None.

When None the execution mode is determined by the global morpheus.config.CppConfig singleton.

This method is best used within code which needs to construct a DataFrame in both CPU and GPU modes. Example usage:

from morpheus.utils.type_utils import get_df_class
df_class = get_df_class()
df = df_class({"a": [1, 2, 3], "b": [4, 5, 6], "c": [7, 8, 9]})

Parameters:

selectormorpheus.utils.type_aliases.DataFrameModule | morpheus.config.ExecutionMode | None, optional: The selector to determine the DataFrame class, by default None.

Returns:

type[DataFrameType]

get_df_pkg( selector: Literal['cudf', 'pandas'] = None, ) → ModuleType[source]#

get_df_pkg( selector: ExecutionMode = None, ) → ModuleType

Return the appropriate DataFrame package based on selector which can be either an ExecutionMode instance, a DataFrame type string, or None.

When None the execution mode is determined by the global morpheus.config.CppConfig singleton.

This method is best used within code which needs to operate in both CPU and GPU modes, where simply importing cudf would cause an import error if the user is not using a GPU. Example usage:

from morpheus.utils.type_utils import get_df_pkg
df_pkg = get_df_pkg()
ser = df_pkg.Series([1,2,3])

Parameters:

selectormorpheus.utils.type_aliases.DataFrameModule | morpheus.config.ExecutionMode | None, optional: The selector to determine the DataFrame package, by default None.

Returns:

types.ModuleType: The associated DataFrame package based on the selector.

get_df_pkg_from_obj(obj)[source]#

Return the appropriate DataFrame package based on a given object (DataFrame, Series, RangeIndex etc…).

Parameters:

objtyping.Any: The object to check.

Returns:

types.ModuleType: The associated DataFrame package based on the object.

get_full_qualname(klass)[source]#

Returns the fully qualified name of a class.

greatest_ancestor(*cls_list)[source]#: Returns the greatest common ancestor of the classes in the class list

is_cudf_type(obj)[source]#

Check if a given object (DataFrame, Series, RangeIndex etc…) is a cuDF type.

Parameters:

objtyping.Any: The object to check.

Returns:

bool: True if the object is a cuDF type, False otherwise.

is_dataframe(obj)[source]#

Check if a given object is a pandas or cudf DataFrame.

Parameters:

objtyping.Any: The object to check.

Returns:

bool: True if the object is a DataFrame, False otherwise.

is_union_type(type_)[source]#

Returns True if the type is a typing.Union or a types.UnionType.

pretty_print_type_name(type_)[source]#

Determines a good label to use for a type. Keeps the strings shorter.

unpack_tuple( cls_1: Type[T], ) → Tuple[Type[T]][source]#
unpack_tuple( cls_1: Type[T1], cls_2: Type[T2], ) → Tuple[Type[T1], Type[T2]]
unpack_tuple( cls_1: Type[T1], cls_2: Type[T2], cls_3: Type[T3], ) → Tuple[Type[T1], Type[T2], Type[T3]]

unpack_union( cls_1: Type[T], ) → Type[T][source]#
unpack_union( cls_1: Type[T1], cls_2: Type[T2], ) → Type[T1] | Type[T2]
unpack_union( cls_1: Type[T1], cls_2: Type[T2], cls_3: Type[T3], ) → Type[T1] | Type[T2] | Type[T3]