riptable.numba.indexing

Indexing-related helper functions for use when implementing other numba-based functions.

Functions

deref_idx(deref, idx)	A numba function for turning an 'indirect' index into a true index if a deference idx exists.
scalar_or_lookup(val, idx)	Allows a numba-based function to accept either a scalar value or an array for some parameter.

riptable.numba.indexing.deref_idx(deref, idx)

A numba function for turning an ‘indirect’ index into a true index if a deference idx exists.

This main use of this function is to seamlessly deal with the existence/non-existence of iGroup (from a Grouping object).

• When writing a function to operate on a single array, there are no groups. The data is contiguous, so there is no iGroup and an index doesn’t need to be ‘dereferenced’ to get a ‘real’ index.

• When writing a function to operate over Grouping data, we need to use Grouping.igroup to turn a 0-based index (scalar) within a particular group’s data into an 0-based index within the larger array whose shape matches the Grouping object itself.

Parameters:

• deref (None or Array) – an array to use for dereferencing or None.

• idx (Integer) – the index to grab (or to be returned if deref is None).

Returns:

idx if deref is None; otherwise deref[idx].

Return type:

int

riptable.numba.indexing.scalar_or_lookup(val, idx)

Allows a numba-based function to accept either a scalar value or an array for some parameter. Scalars are passed through, but if an array is provided, the array element at the specified index is returned.

For example, in a numba-based function operating over grouped data (with a Grouping), the function could accept a scalar parameter; by using scalar_or_lookup, the code can easily accept either a scalar (to be applied to all groups) or an array containing a specific value for each group.

Parameters:

• val –

• idx –

Return type:

retval