cboumalh commented on code in PR #52883: URL: https://github.com/apache/spark/pull/52883#discussion_r2551011273
########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/tuplesketchesAggregates.scala: ########## @@ -0,0 +1,843 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.catalyst.expressions.aggregate + +import org.apache.datasketches.tuple.{Intersection, Sketch, Summary, Union, UpdatableSketch, UpdatableSketchBuilder, UpdatableSummary} + +import org.apache.spark.SparkUnsupportedOperationException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.{ExpectsInputTypes, Expression, ExpressionDescription, Literal} +import org.apache.spark.sql.catalyst.expressions.aggregate.TypedImperativeAggregate +import org.apache.spark.sql.catalyst.trees.{QuaternaryLike, TernaryLike} +import org.apache.spark.sql.catalyst.util.{ArrayData, CollationFactory, ThetaSketchUtils} +import org.apache.spark.sql.errors.QueryExecutionErrors +import org.apache.spark.sql.internal.types.StringTypeWithCollation +import org.apache.spark.sql.types.{AbstractDataType, ArrayType, BinaryType, DataType, DoubleType, FloatType, IntegerType, LongType, StringType, StructType} +import org.apache.spark.unsafe.types.UTF8String + +sealed trait TupleSketchState { + def serialize(): Array[Byte] + def eval(): Array[Byte] +} +case class UpdatableTupleSketchBuffer[U, S <: UpdatableSummary[U]](sketch: UpdatableSketch[U, S]) + extends TupleSketchState { + override def serialize(): Array[Byte] = sketch.compact.toByteArray + override def eval(): Array[Byte] = sketch.compact.toByteArray +} +case class UnionTupleAggregationBuffer[S <: Summary](union: Union[S]) extends TupleSketchState { + override def serialize(): Array[Byte] = union.getResult.toByteArray + override def eval(): Array[Byte] = union.getResult.toByteArray +} +case class IntersectionTupleAggregationBuffer[S <: Summary](intersection: Intersection[S]) + extends TupleSketchState { + override def serialize(): Array[Byte] = intersection.getResult.toByteArray + override def eval(): Array[Byte] = intersection.getResult.toByteArray +} +case class FinalizedTupleSketch[S <: Summary](sketch: Sketch[S]) extends TupleSketchState { + override def serialize(): Array[Byte] = sketch.toByteArray + override def eval(): Array[Byte] = sketch.toByteArray +} + +/** + * The TupleSketchAgg function utilizes a Datasketches TupleSketch instance to count a + * probabilistic approximation of the number of unique values in a given column with associated + * summary values, and outputs the binary representation of the TupleSketch. + * + * See [[https://datasketches.apache.org/docs/Tuple/TupleSketches.html]] for more information. + * + * @param child + * child expression (struct with key and summary value) against which unique counting will occur + * @param lgNomEntriesExpr + * the log-base-2 of nomEntries decides the number of buckets for the sketch + * @param summaryType + * the type of summary (double, integer, string) Review Comment: @dtenedor The highest priority would be double. Most of the integer cases can also use double. String summary types are a niche. The `lgNomEntries` and `mode` parameters are much more likely to be configured compared to the `summaryType`. So maybe I can start with: ``` tuple_sketch_agg_double( key => col, summary => 1.0D, lgNomEntries => 12, (optional, default 12) mode => 'sum') (optional, default sum) ) ``` I agree regarding creating SQL named arguments for the strings, I saw the examples you mentioned. I can add that. I also see the case for going away from struct usage. My initial intuition was have an input that looks like a tuple for a tuple sketch, but named arguments with explicit key and summary parameters are more readable and consistent with Spark. If things look good for `tuple_sketch_agg_double`, we can extend for other summary types. I see in your `tuple_sketch_agg_bigint`, the `summaryType` parameter was still configurable which I believe will make things a bit confusing since the function name already specifies the `summaryType` in the suffix. Also, the supported summary types are string, integer, and double so we would probably have `_string`, `_integer`, and `_double` as suffixes respectively. Thanks for the help with this! 🤝 Let me know what you think -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
