ndonnx.additional package

Module contents

ndonnx.additional.datetime_to_year_month_day(arr: Array) → tuple[Array, Array, Array][source]: Split date time elements of the provided array into year, month, and day components.

ndonnx.additional.fill_null(x: Array, /, value: Array | SCALAR) → Array[source]

Returns a new Array with the null values filled with the given value.

Parameters:

x: Array: The Array to fill the null values of.
value: Array | Scalar: The value to fill the null values with.

Returns:

out: Array: A new Array with the null values filled with the given value.

ndonnx.additional.get_data(x: Array, /) → Array[source]

Get data part of a masked, datetime or timedelta array.

If the x is not masked, return x.

ndonnx.additional.get_mask(x: Array, /) → Array | None[source]: Get null-mask if there is any.

ndonnx.additional.is_float_dtype(dtype: DType, /) → TypeIs[Float16 | Float32 | Float64][source]

Return True if dtype is of a floating point data type.

This does not include masked data types.

Return True if dtype is of an integer data type.

Returns False for boolean data types.

Return True if dtype is a nullable (i.e. “masked”) data type.

Floating point and datetime data types are not considered as “nullable” by this function.

ndonnx.additional.is_nullable_float_dtype(dtype: DType, /) → TypeIs[NFloat16 | NFloat32 | NFloat64][source]: Return True if dtype is a nullable integer (i.e. “masked”) data type.

ndonnx.additional.is_nullable_integer_dtype(dtype: DType, /) → TypeIs[NInt8 | NInt16 | NInt32 | NInt64 | NUInt8 | NUInt16 | NUInt32 | NUInt64][source]: Return True if dtype is a nullable integer (i.e. “masked”) data type.

ndonnx.additional.is_nullable_numeric_dtype(dtype: DType, /) → TypeIs[NFloat16 | NFloat32 | NFloat64 | NInt8 | NInt16 | NInt32 | NInt64 | NUInt8 | NUInt16 | NUInt32 | NUInt64][source]

Return True if dtype is of a numeric data type.

This does not include masked data types.

ndonnx.additional.is_onnx_dtype(dtype: DType, /) → TypeIs[Float16 | Float32 | Float64 | Int8 | Int16 | Int32 | Int64 | UInt8 | UInt16 | UInt32 | UInt64 | Utf8 | Bool][source]: Return True if dtype is of a data type found in the ONNX standard.

ndonnx.additional.is_signed_integer_dtype(dtype: DType, /) → TypeIs[Int8 | Int16 | Int32 | Int64][source]

Return True if dtype is of a signed integer data type.

Returns False for boolean data types.

ndonnx.additional.is_time_unit(s: str, /) → TypeIs[Literal['ns', 'us', 'ms', 's']][source]

ndonnx.additional.is_unsigned_integer_dtype(dtype: DType, /) → TypeIs[UInt8 | UInt16 | UInt32 | UInt64][source]

Return True if dtype is of an unsigned integer data type.

Returns False for boolean data types.

ndonnx.additional.isin(x: Array, /, items: Sequence[SCALAR]) → Array[source]

Return true where the input Array contains an element in items.

NaN values do not compare equal.

Parameters:

x: Array: The input Array to check for the presence of items.
items: Sequence[Scalar]: Scalar items to check for in the input Array.

Returns:

out: Array: Array of booleans indicating whether each element of x is in items.

ndonnx.additional.make_nullable(x: Array, null: Array | None, /, *, merge_strategy: Literal['raise', 'merge'] = 'raise') → Array[source]

Given an array x of values and a null mask null, construct a new Array with a nullable data type.

Parameters:

x: Array: Array of values
null: Array: Array of booleans indicating whether each element of x is null.
merge_strategy: Literal[“raise”, “merge”]: If “raise”, a TypeError is raised if x is already of a nullable data type. If “merge” is provided, any mask existing on x is merged with null.

Returns:

out: Array: A new Array with a nullable data type.

Raises:

TypeError: If the data type of x does not have a nullable counterpart.

ndonnx.additional.put(a: Array, indices: Array, updates: Array, /) → None[source]

Replaces specified elements of an array with given values.

This function follows the semantics of numpy.put with `mode=”raises”. The data types of the update array and the updates must match. The indices must be provided as a 1D int64 array.

ndonnx.additional.shape(x: Array, /) → Array[source]

Returns shape of an array.

Parameters:

x: Array: Array to get shape of

Returns:

out: Array: Array of shape

ndonnx.additional.static_map(x: Array, /, mapping: Mapping[SCALAR, VALUE], default: VALUE | None = None) → Array[source]

Map values in x based on the static mapping.

Parameters:

x: Array

The Array whose values will be mapped.

mapping: Mapping[Key, Value]

A mapping from keys to values. The keys must be of the same type as the values in x.

default: Value, optional

The default value to use when a key is not found in the mapping. If None the value depends on the type of Value in the mapping:

float: 0.0

int: 0

bool: False

str: “MISSING”

Returns:

out: Array: A new Array with the values mapped according to the mapping.

Raises:

ValueError: If mapping is empty and default is None.