ak.with_cache

Defined in awkward.operations.structure on line 4203.

ak.with_cache(array, cache, highlevel=True, behavior=None)
Parameters

  • array – Data to search for nested virtual arrays.

  • cache (None or MutableMapping) – If None, arrays are generated every time they are needed; otherwise, generated arrays are stored in the mapping with __setitem__, retrieved with __getitem__, and only re-generated if __getitem__ raises a KeyError. This mapping may evict elements according to any caching algorithm (LRU, LFR, RR, TTL, etc.). If “new”, a new dict (keep-forever cache) is created.

  • highlevel (bool) – If True, return an ak.Array; otherwise, return a low-level ak.layout.Content subclass.

  • behavior (None or dict) – Custom ak.behavior for the output array, if high-level.

Remove caches from all virtual arrays nested within array if cache is None; adds a cache otherwise.

For example:

>>> cache1 = {}
>>> one = ak.virtual(lambda: [[1.1, 2.2, 3.3], [], [4.4, 5.5]], cache=cache1, length=3)
>>> two = ak.virtual(lambda: [100, 200, 300], cache=cache1, length=3)
>>> array1 = ak.zip({"x": one, "y": two}, depth_limit=1)
>>> len(cache1)
0

creates an array of records with virtual fields that would fill cache1.

We can then switch every instance of cache1 in array with cache2:

>>> cache2 = {}
>>> array2 = ak.with_cache(array1, cache2)
>>> array2["x"]
<Array [[1.1, 2.2, 3.3], [], [4.4, 5.5]] type='3 * var * float64'>
>>>
>>> len(cache1), len(cache2)
(0, 1)

Viewing the array2["x"] filled cache2 and not cache1.

See ak.virtual.