Github user viirya commented on a diff in the pull request:
https://github.com/apache/spark/pull/15544#discussion_r139877802
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/ApproxCountDistinctForIntervals.scala
---
@@ -0,0 +1,235 @@
+/*
+ * 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 java.util
+
+import org.apache.spark.sql.catalyst.InternalRow
+import org.apache.spark.sql.catalyst.analysis.TypeCheckResult
+import
org.apache.spark.sql.catalyst.analysis.TypeCheckResult.{TypeCheckFailure,
TypeCheckSuccess}
+import org.apache.spark.sql.catalyst.expressions.{AttributeReference,
ExpectsInputTypes, Expression}
+import org.apache.spark.sql.catalyst.util.{ArrayData, GenericArrayData,
HyperLogLogPlusPlusHelper}
+import org.apache.spark.sql.types._
+
+/**
+ * This function counts the approximate number of distinct values (ndv) in
+ * intervals constructed from endpoints specified in
`endpointsExpression`. The endpoints should be
+ * sorted into ascending order. E.g., given an array of endpoints
+ * (endpoint_1, endpoint_2, ... endpoint_N), returns the approximate ndv's
for intervals
+ * [endpoint_1, endpoint_2], (endpoint_2, endpoint_3], ... (endpoint_N-1,
endpoint_N].
+ * To count ndv's in these intervals, apply the HyperLogLogPlusPlus
algorithm in each of them.
+ * @param child to estimate the ndv's of.
+ * @param endpointsExpression to construct the intervals, should be sorted
into ascending order.
+ * @param relativeSD The maximum estimation error allowed in the
HyperLogLogPlusPlus algorithm.
+ */
+case class ApproxCountDistinctForIntervals(
+ child: Expression,
+ endpointsExpression: Expression,
+ relativeSD: Double = 0.05,
+ mutableAggBufferOffset: Int = 0,
+ inputAggBufferOffset: Int = 0)
+ extends ImperativeAggregate with ExpectsInputTypes {
+
+ def this(child: Expression, endpointsExpression: Expression) = {
+ this(
+ child = child,
+ endpointsExpression = endpointsExpression,
+ relativeSD = 0.05,
+ mutableAggBufferOffset = 0,
+ inputAggBufferOffset = 0)
+ }
+
+ def this(child: Expression, endpointsExpression: Expression, relativeSD:
Expression) = {
+ this(
+ child = child,
+ endpointsExpression = endpointsExpression,
+ relativeSD = HyperLogLogPlusPlus.validateDoubleLiteral(relativeSD),
+ mutableAggBufferOffset = 0,
+ inputAggBufferOffset = 0)
+ }
+
+ override def inputTypes: Seq[AbstractDataType] = {
+ Seq(TypeCollection(NumericType, TimestampType, DateType), ArrayType)
+ }
+
+ // Mark as lazy so that endpointsExpression is not evaluated during tree
transformation.
+ lazy val endpoints: Array[Double] =
+ (endpointsExpression.dataType, endpointsExpression.eval()) match {
+ case (ArrayType(baseType: NumericType, _), arrayData: ArrayData) =>
--- End diff --
The type of `child` can be `TimestampType` and `DateType`, but endpoints
can only be `ArrayType` of `NumericType`. It may not be convenient to set up
numeric endpoints for a timestamp or date child column.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
