[FLINK-26477][python] Support WindowedStream.aggregate in Python DataStream API #19176
Conversation
| .aggregate(new AverageAggregate) | ||
| ``` | ||
| {{< /tab >}} | ||
| {{< tab "Python" >}} |
There was a problem hiding this comment.
should also update the chinese doc: docs.zh/content/docs/dev/datastream/operators/windows.md
| ```python | ||
| class ProcessWindowFunction(Function, Generic[IN, OUT, KEY, W]): | ||
|
|
||
| @abstractmethod |
There was a problem hiding this comment.
The method clear is missing. Should also update the Java/Scala example.
| input \ | ||
| .key_by(<key selector>) \ | ||
| .window(<window assigner>) \ | ||
| .reduce(lambda v1, v2: (v1[0], v1[1] + v2[1]), |
There was a problem hiding this comment.
The implementation is incorrect. It doesn't match the Java/Scala example, also doesn't match the description of this section: "The following example shows how an incremental ReduceFunction can be combined with a ProcessWindowFunction to return the smallest event in a window along with the start time of the window."
| input \ | ||
| .key_by(<key selector>) \ | ||
| .window(<window assigner>) \ | ||
| .apply(new MyWindowFunction()) |
There was a problem hiding this comment.
| .apply(new MyWindowFunction()) | |
| .apply(MyWindowFunction()) |
| aggregation function. | ||
| :param result_type: Type information for the result type of the window function. | ||
| :return: The data stream that is the result of applying the window function to the window. | ||
| """ |
| aggregate_function: AggregateFunction, | ||
| window_function: Union[WindowFunction, ProcessWindowFunction] = None, | ||
| accumulator_type: TypeInformation = None, | ||
| result_type: TypeInformation = None) -> DataStream: |
There was a problem hiding this comment.
| result_type: TypeInformation = None) -> DataStream: | |
| output_type: TypeInformation = None) -> DataStream: |
Keep it consistent with the other methods.
| Arriving data is incrementally aggregated using the given aggregate function. This means | ||
| that the window function typically has only a single value to process when called. | ||
There was a problem hiding this comment.
What about showing a simple example about this API?
Vancior
force-pushed
the
feat/window_aggregate
branch
from
68fab99 to
17733c3
Compare
|
@flinkbot run azure |
4 participants
What is the purpose of the change
Support
WindowedStream.aggregatein Python DataStream API, which is an alignment to the Java API, with doc updated.Verifying this change
This change added tests and can be verified as follows:
WindowedStream.aggregateAPIDoes this pull request potentially affect one of the following parts:
@Public(Evolving): (yes)Documentation