Skip to content

window

function window(
timeColumn,
windowDuration,
slideDuration?,
startTime?): Column;

Defined in: functions/datetime.ts:164

Bucket rows into tumbling or sliding event-time windows.

Returns a struct { start, end } timestamp column. Combine with withWatermark and groupBy for stateful streaming aggregations.

Parameter Type Description
timeColumn ColOrName event-time column (timestamp)
windowDuration string Spark interval string, e.g. "5 minutes"
slideDuration? string optional slide; defaults to windowDuration (tumbling)
startTime? string optional offset from epoch, e.g. "15 seconds"

Column

df.groupBy(window(col("ts"), "5 minutes"))