RaggedArray

A RaggedArray is a 2d array, where the lengths of the rows can differ. These arrays behaves as much as possible like numpy array, meaning that you can index it like a numpy array, call unfuncs on them and also call a limited set of array functions on them.

API documentation

class npstructures.RaggedArray(data: _SupportsArray[dtype[Any]] | _NestedSequence[_SupportsArray[dtype[Any]]] | bool | int | float | complex | str | bytes | _NestedSequence[bool | int | float | complex | str | bytes], shape: List[int] | RaggedShape | RaggedView | Shape | None = None, dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any] = None, safe_mode: bool = True)[source]

Class to represent 2d arrays with differing row lengths

Provides objects that behaves similar to numpy ndarrays, but can represent arrays with differing row lengths. Numpy-like functionality is provided in three ways.

  1. ufuncs are supported, so addition, multiplication, sin etc. works just like numpy arrays

  2. Indexing works similar to numpy arrays. Slice objects, index lists and boolean arrays etc.

  3. Some numpy functions like np.concatenate, np.sum, np.nonzero are implemented.

See the examples for simple demonstrations.

Parameters:
datanested list or array_like

the nested list to be converted, or (if shape is provided) a continous array of values

shapelist or Shape, optional

the shape of the ragged array, or list of row lenghts

dtypeoptional

the data type to use for the array

Examples

>>> ra = RaggedArray([[2, 4, 8], [3, 2], [5, 7, 3, 1]])
>>> ra+1
RaggedArray([[3, 5, 9], [4, 3], [6, 8, 4, 2]])
>>> ra*2
RaggedArray([[4, 8, 16], [6, 4], [10, 14, 6, 2]])
>>> ra[1]
array([3, 2])
>>> ra[0:2]
RaggedArray([[2, 4, 8], [3, 2]])
>>> np.nonzero(ra>3)
(array([0, 0, 2, 2]), array([1, 2, 0, 1]))
Attributes:
shapeRaggedShape

the shape (row-lengths) of the array

sizeint

the total size of the array

dtype

the data-type of the array

astype(dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any]) → RaggedArray[source]

Return ragged array with underlying data changed to dtype

Parameters:
dtypenpt.DTypeLike
Returns:
‘RaggedArray’
cumsum(axis: int | None = None, dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any] = None) → RaggedArray[source]

Return an array with cumulative sums along the given axis

Parameters:
axisint
dtypenpt.DTypeLike
Returns:
RaggedArray
equals(other)[source]

Checks for euqality with other

fill(value: Number)[source]

Fill the whole array with value

classmethod load(filename: str) → RaggedArray[source]

Loads a ragged array from file

Parameters:
filenamestr

name of the file

Returns:
RaggedArray

The ragged array loaded from file

nonzero() → Tuple[_SupportsArray[dtype[Any]] | _NestedSequence[_SupportsArray[dtype[Any]]] | bool | int | float | complex | str | bytes | _NestedSequence[bool | int | float | complex | str | bytes]][source]

Return the indices of all nonzero entries in the array

Returns:
Tuple[npt.ArrayLike]
save(filename)[source]

Saves the ragged array to file using np.savez

Parameters:
filenamestr

name of the file

to_numpy_array() → ndarray[source]

Return a normal numpy array with the same shape and data

Requires that all the rows are of the same lengths

Returns:
np.ndarray
tolist() → List[List][source]

Return a list of list representing rows

Returns:
List[List]
npstructures.arrayfunctions.concatenate(ragged_arrays: List[RaggedArray], axis: int = 0) → RaggedArray[source]

Concatenate a set of raggedarrays along the given axis

Parameters:
ragged_arraysList[RaggedArray]
axisint
Returns:
RaggedArray
npstructures.arrayfunctions.diff(ragged_array: RaggedArray, n: int = 1, axis: int = -1) → RaggedArray[source]

Return diffs for each row in a raggedarray

Parameters:
ragged_arrayRaggedArray
nint
axisint
npstructures.arrayfunctions.zeros_like(ragged_array: RaggedArray, dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any] = None, shape: Tuple[int] | None = None) → RaggedArray[source]

Return a raggedarray with the same shape and dtype as raggedarray filled with zeros

Parameters:
ragged_arrayRaggedArray
dtypenpt.DTypeLike
shapeTuple[int]
Returns:
RaggedArray
npstructures.arrayfunctions.ones_like(ragged_array: RaggedArray, dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any] = None, shape: Tuple[int] | None = None) → RaggedArray[source]

Return a raggedarray with the same shape and dtype as raggedarray filled with ones

Parameters:
ragged_arrayRaggedArray
dtypenpt.DTypeLike
shapeTuple[int]
Returns:
RaggedArray
npstructures.arrayfunctions.empty_like(ragged_array: RaggedArray, dtype: dtype[Any] | None | type[Any] | _SupportsDType[dtype[Any]] | str | tuple[Any, int] | tuple[Any, SupportsIndex | Sequence[SupportsIndex]] | list[Any] | _DTypeDict | tuple[Any, Any] = None, shape: Tuple[int] | None = None) → RaggedArray[source]

Return an empty raggedarray with the same shape and dtype as raggedarray

Parameters:
ragged_arrayRaggedArray
dtypenpt.DTypeLike
shapeTuple[int]
Returns:
RaggedArray
npstructures.arrayfunctions.where(ragged_mask: RaggedArray, x: RaggedArray | None = None, y: RaggedArray | None = None) → RaggedArray[source]

Perform an ifelse (tertiary operator) on a raggedarray

Inicies where ragged_mask is True gets the corresponding value from x. Where False it gets from y

Parameters:
ragged_maskRaggedArray
xRaggedArray
yRaggedArray
Returns:
RaggedArray
npstructures.arrayfunctions.unique(ragged_array: RaggedArray, axis: int | None = None, return_counts: bool = False) → RaggedArray | Tuple[RaggedArray][source]

Get the unqiue values from ragged_array. If return_counts then also return the number of elemtents with each value

Parameters:
ragged_arrayRaggedArray
axisint
return_countsbool
Returns:
Union[RaggedArray, Tuple[RaggedArray]]