[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin closed the pull request at: https://github.com/apache/spark/pull/14298 --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r76548403 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Double
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r75830247 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Double
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user clockfly commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r75709632 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Doub
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73995159 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Double
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user cloud-fan commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73993906 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Dou
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73993189 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Double
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user cloud-fan commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73991551 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Dou
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73990892 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Double
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user cloud-fan commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73832267 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,462 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either a single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Dou
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r73101674 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B --- End diff -- I don't have strong preference here -- let's see what reviewers say. --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user dongjoon-hyun commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r72295166 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B --- End diff -- Hi, @lw-lin . I know the reason why you define this as a capatal 'B', but I'm just wondering it's consistent with Spark naming rule. --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user dongjoon-hyun commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r72294685 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an --- End diff -- minor: `an single` -> `a single` --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user hvanhovell commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r72227222 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(D
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user thunterdb commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r72137984 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Do
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user lw-lin commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r71815241 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Doubl
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
Github user thunterdb commented on a diff in the pull request: https://github.com/apache/spark/pull/14298#discussion_r71740068 --- Diff: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/PercentileApprox.scala --- @@ -0,0 +1,456 @@ +/* + * 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 scala.collection.mutable.ArrayBuffer + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.analysis.TypeCheckResult +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.expressions.aggregate.QuantileSummaries.Stats +import org.apache.spark.sql.catalyst.util._ +import org.apache.spark.sql.types._ + +/** + * Computes an approximate percentile (quantile) using the G-K algorithm (see below), for very + * large numbers of rows where the regular percentile() UDAF might run out of memory. + * + * The input is a single double value or an array of double values representing the percentiles + * requested. The output, corresponding to the input, is either an single double value or an + * array of doubles that are the percentile values. + */ +@ExpressionDescription( + usage = """_FUNC_(col, p [, B]) - Returns an approximate pth percentile of a numeric column in the + group. The B parameter, which defaults to 1000, controls approximation accuracy at the cost of + memory; higher values yield better approximations. +_FUNC_(col, array(p1 [, p2]...) [, B]) - Same as above, but accepts and returns an array of + percentile values instead of a single one. +""") +case class PercentileApprox( +child: Expression, +percentilesExpr: Expression, +bExpr: Option[Expression], +percentiles: Seq[Double], // the extracted percentiles +B: Int,// the extracted B +resultAsArray: Boolean,// whether to return the result as an array +mutableAggBufferOffset: Int = 0, +inputAggBufferOffset: Int = 0) extends ImperativeAggregate { + + private def this(child: Expression, percentilesExpr: Expression, bExpr: Option[Expression]) = { +this( + child = child, + percentilesExpr = percentilesExpr, + bExpr = bExpr, + // validate and extract percentiles + percentiles = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._1, + // validate and extract B + B = bExpr.map(PercentileApprox.validateBLiteral(_)).getOrElse(PercentileApprox.B_DEFAULT), + // validate and mark whether we should return results as array of double or not + resultAsArray = PercentileApprox.validatePercentilesLiteral(percentilesExpr)._2) + } + + // Constructor for the "_FUNC_(col, p) / _FUNC_(col, array(p1, ...))" form + def this(child: Expression, percentilesExpr: Expression) = { +this(child, percentilesExpr, None) + } + + // Constructor for the "_FUNC_(col, p, B) / _FUNC_(col, array(p1, ...), B)" form + def this(child: Expression, percentilesExpr: Expression, bExpr: Expression) = { +this(child, percentilesExpr, Some(bExpr)) + } + + override def prettyName: String = "percentile_approx" + + override def withNewMutableAggBufferOffset(newMutableAggBufferOffset: Int): ImperativeAggregate = +copy(mutableAggBufferOffset = newMutableAggBufferOffset) + + override def withNewInputAggBufferOffset(newInputAggBufferOffset: Int): ImperativeAggregate = +copy(inputAggBufferOffset = newInputAggBufferOffset) + + override def children: Seq[Expression] = +bExpr.map(child :: percentilesExpr :: _ :: Nil).getOrElse(child :: percentilesExpr :: Nil) + + // we would return null for empty inputs + override def nullable: Boolean = true + + override def dataType: DataType = if (resultAsArray) ArrayType(Do
[GitHub] spark pull request #14298: [SPARK-16283][SQL] Implement `percentile_approx` ...
GitHub user lw-lin opened a pull request: https://github.com/apache/spark/pull/14298 [SPARK-16283][SQL] Implement `percentile_approx` SQL function ## What changes were proposed in this pull request? This patch Implements `percentile_approx` SQL function using Spark's implementation of G-K algorithm. - commit 1: moves the G-K algorithm implementation(`QuantileSummaries` and related tests) from `sql/core` to `sql/catalyst` - commit 2: implements `percentile_approx` using G-K algorithm ## How was this patch tested? - Jenkins - added new tests You can merge this pull request into a Git repository by running: $ git pull https://github.com/lw-lin/spark impl_percentile_approx Alternatively you can review and apply these changes as the patch at: https://github.com/apache/spark/pull/14298.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #14298 commit d3a6dc825577a4a5e44e8eb0f8e61ef2053e127d Author: Liwei Lin Date: 2016-07-21T08:29:00Z Move G-K all from `sql/core` to `sql/catalyst` commit 110158062cb1f6a571ad8e0bab9bc5962107b59a Author: Liwei Lin Date: 2016-07-21T08:38:06Z Implement percentile_approx --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org