icexelloss commented on code in PR #35514:
URL: https://github.com/apache/arrow/pull/35514#discussion_r1211899044
##########
python/pyarrow/_compute.pyx:
##########
@@ -2738,9 +2744,83 @@ def register_scalar_function(func, function_name,
function_doc, in_types, out_ty
21
]
"""
- return _register_scalar_like_function(get_register_scalar_function(),
- func, function_name, function_doc,
in_types,
- out_type, func_registry)
+ return _register_user_defined_function(get_register_scalar_function(),
+ func, function_name, function_doc,
in_types,
+ out_type, func_registry)
+
+
+def register_aggregate_function(func, function_name, function_doc, in_types,
out_type,
+ func_registry=None):
+ """
+ Register a user-defined non-decomposable aggregate function.
+
+ A non-decomposable aggregation function is a function that executes
+ aggregate operations on the whole data that it is aggregating.
+ In other words, non-decomposable aggregate function cannot be
+ split into consume/merge/finalize steps.
+
+ This is mostly useful with segemented aggregation, where the data
+ to be aggregated is continuous.
+
+ Parameters
+ ----------
+ func : callable
+ A callable implementing the user-defined function.
+ The first argument is the context argument of type
+ UdfContext.
+ Then, it must take arguments equal to the number of
+ in_types defined. It must return Scalar matching the
+ out_type.
+ To define a varargs function, pass a callable that takes
+ varargs. The in_type needs to match in type of inputs when
+ the function gets called.
+
+ function_name : str
+ Name of the function. This name must be globally unique.
Review Comment:
This is copied from the other UDF docs but I can expand on that here in the
doc
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
