ak.type — Awkward Array 2.8.2 documentation (original) (raw)

Defined in awkward.operations.ak_type on line 21.

ak.type(array, *, behavior=None)#

Parameters:

• array – Array-like data (anything ak.to_layout recognizes).
• behavior (None or dict) – Custom ak.behavior for the output type, if high-level.

The high-level type of an array (many types supported, including all Awkward Arrays and Records) as ak.types.Type objects.

The high-level type ignores layout differences likeak.contents.ListArray versus ak.contents.ListOffsetArray, but not differences like “regular-sized lists” (i.e.ak.contents.RegularArray) versus “variable-sized lists” (i.e.ak.contents.ListArray and similar).

Types are rendered as Datashapestrings, which makes the same distinctions.

For example,

array = ak.Array([[{"x": 1.1, "y": [1]}, {"x": 2.2, "y": [2, 2]}], ... [], ... [{"x": 3.3, "y": [3, 3, 3]}]])

has type

ak.type(array).show() 3 * var * { x: float64, y: var * int64 }

but

array = ak.Array(np.arange(235).reshape(2, 3, 5))

has type

ak.type(array).show() 2 * 3 * 5 * int64

Some cases, like heterogeneous data, require extensions beyond the Datashape specification. For example,

array = ak.Array([1, "two", [3, 3, 3]])

has type

ak.type(array).show() 3 * union[ int64, string, var * int64 ]

but “union” is not a Datashape type-constructor. (Its syntax is similar to existing type-constructors, so it’s a plausible addition to the language.)