[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1140198056
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFEvaluators.scala:
##
@@ -0,0 +1,148 @@
+/*
+ * 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.hive
+
+import scala.collection.JavaConverters._
+
+import org.apache.hadoop.hive.ql.exec.{FunctionRegistry, UDF}
+import org.apache.hadoop.hive.ql.udf.{UDFType => HiveUDFType}
+import org.apache.hadoop.hive.ql.udf.generic.GenericUDF
+import org.apache.hadoop.hive.ql.udf.generic.GenericUDF._
+import org.apache.hadoop.hive.ql.udf.generic.GenericUDFUtils.ConversionHelper
+import org.apache.hadoop.hive.serde2.objectinspector.{ObjectInspector,
ObjectInspectorFactory}
+import
org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.ObjectInspectorOptions
+
+import org.apache.spark.sql.catalyst.expressions.Expression
+import org.apache.spark.sql.errors.QueryExecutionErrors
+import org.apache.spark.sql.hive.HiveShim.HiveFunctionWrapper
+import org.apache.spark.sql.types.DataType
+
+abstract class HiveUDFEvaluatorBase[UDFType <: AnyRef](
+funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
+ extends HiveInspectors with Serializable {
+
+ @transient
+ lazy val function = funcWrapper.createFunction[UDFType]()
+
+ @transient
+ val isUDFDeterministic = {
Review Comment:
It should be `lazy val`, as it accesses a lazy val.
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1139899830
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,137 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
@transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
+ }
+
+ override def foldable: Boolean = isUDFDeterministic &&
children.forall(_.foldable)
+
+ override lazy val dataType: DataType =
javaTypeToDataType(evaluator.method.getGenericReturnType)
+
+ // TODO: Finish input output types.
+ override def eval(input: InternalRow): Any = {
+children.zipWithIndex.map {
+ case (child, idx) =>
+evaluator.setArg(idx, child.eval(input))
+}
+evaluator.evaluate()
+ }
+
+ override def toString: String = {
+s"$nodeName#${funcWrapper.functionClassName}(${children.mkString(",")})"
+ }
+
+ override def prettyName: String = name
+
+ override def sql: String = s"$name(${children.map(_.sql).mkString(", ")})"
+
+ override protected def withNewChildrenInternal(newChildren:
IndexedSeq[Expression]): Expression =
+copy(children = newChildren)
+
+ protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+val refEvaluator = ctx.addReferenceObj("evaluator", evaluator)
+val evals = children.map(_.genCode(ctx))
+
+val setValues = evals.zipWithIndex.map {
+ case (eval, i) =>
+s"""
+ |if (${eval.isNull}) {
+ | $refEvaluator.setArg($i, null);
+ |} else {
+ | $refEvaluator.setArg($i, ${eval.value});
+ |}
+ |""".stripMargin
+}
+
+val resultType = CodeGenerator.boxedType(dataType)
+val resultTerm = ctx.freshName("result")
+ev.copy(code =
+ code"""
+ |${evals.map(_.code).mkString("\n")}
+ |${setValues.mkString("\n")}
+ |$resultType $resultTerm = null;
+ |boolean ${ev.isNull} = false;
+ |try {
+ | $resultTerm = ($resultType) $refEvaluator.evaluate();
+ | ${ev.isNull} = $resultTerm == null;
+ |} catch (Throwable e) {
+ | throw QueryExecutionErrors.failedExecuteUserDefinedFunctionError(
Review Comment:
BTW this seems like an unrelated change. The previous code does not rethrow
the exception.
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1139898978
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,137 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
@transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
+ }
+
+ override def foldable: Boolean = isUDFDeterministic &&
children.forall(_.foldable)
+
+ override lazy val dataType: DataType =
javaTypeToDataType(evaluator.method.getGenericReturnType)
+
+ // TODO: Finish input output types.
+ override def eval(input: InternalRow): Any = {
+children.zipWithIndex.map {
+ case (child, idx) =>
+evaluator.setArg(idx, child.eval(input))
+}
+evaluator.evaluate()
+ }
+
+ override def toString: String = {
+s"$nodeName#${funcWrapper.functionClassName}(${children.mkString(",")})"
+ }
+
+ override def prettyName: String = name
+
+ override def sql: String = s"$name(${children.map(_.sql).mkString(", ")})"
+
+ override protected def withNewChildrenInternal(newChildren:
IndexedSeq[Expression]): Expression =
+copy(children = newChildren)
+
+ protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+val refEvaluator = ctx.addReferenceObj("evaluator", evaluator)
+val evals = children.map(_.genCode(ctx))
+
+val setValues = evals.zipWithIndex.map {
+ case (eval, i) =>
+s"""
+ |if (${eval.isNull}) {
+ | $refEvaluator.setArg($i, null);
+ |} else {
+ | $refEvaluator.setArg($i, ${eval.value});
+ |}
+ |""".stripMargin
+}
+
+val resultType = CodeGenerator.boxedType(dataType)
+val resultTerm = ctx.freshName("result")
+ev.copy(code =
+ code"""
+ |${evals.map(_.code).mkString("\n")}
+ |${setValues.mkString("\n")}
+ |$resultType $resultTerm = null;
+ |boolean ${ev.isNull} = false;
+ |try {
+ | $resultTerm = ($resultType) $refEvaluator.evaluate();
+ | ${ev.isNull} = $resultTerm == null;
+ |} catch (Throwable e) {
+ | throw QueryExecutionErrors.failedExecuteUserDefinedFunctionError(
Review Comment:
shall we move the try-catch to `evaluator.evaluate()`?
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1139897525
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,137 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
@transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
+ }
+
+ override def foldable: Boolean = isUDFDeterministic &&
children.forall(_.foldable)
+
+ override lazy val dataType: DataType =
javaTypeToDataType(evaluator.method.getGenericReturnType)
+
+ // TODO: Finish input output types.
+ override def eval(input: InternalRow): Any = {
+children.zipWithIndex.map {
+ case (child, idx) =>
+evaluator.setArg(idx, child.eval(input))
Review Comment:
```suggestion
case (child, idx) => evaluator.setArg(idx, child.eval(input))
```
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1139897200
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,137 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
@transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
Review Comment:
the code seems to be the same with generic UDF. maybe we can move it to
`HiveUDFEvaluatorBase`
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1139896355
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,137 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
@transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
+ }
+
+ override def foldable: Boolean = isUDFDeterministic &&
children.forall(_.foldable)
+
+ override lazy val dataType: DataType =
javaTypeToDataType(evaluator.method.getGenericReturnType)
+
+ // TODO: Finish input output types.
+ override def eval(input: InternalRow): Any = {
+children.zipWithIndex.map {
+ case (child, idx) =>
+evaluator.setArg(idx, child.eval(input))
+}
+evaluator.evaluate()
+ }
+
+ override def toString: String = {
+s"$nodeName#${funcWrapper.functionClassName}(${children.mkString(",")})"
+ }
+
+ override def prettyName: String = name
+
+ override def sql: String = s"$name(${children.map(_.sql).mkString(", ")})"
+
+ override protected def withNewChildrenInternal(newChildren:
IndexedSeq[Expression]): Expression =
+copy(children = newChildren)
+
+ protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+val refEvaluator = ctx.addReferenceObj("evaluator", evaluator)
+val evals = children.map(_.genCode(ctx))
+
+val setValues = evals.zipWithIndex.map {
+ case (eval, i) =>
+s"""
+ |if (${eval.isNull}) {
+ | $refEvaluator.setArg($i, null);
+ |} else {
+ | $refEvaluator.setArg($i, ${eval.value});
+ |}
+ |""".stripMargin
+}
+
+val resultType = CodeGenerator.boxedType(dataType)
+val resultTerm = ctx.freshName("result")
+ev.copy(code =
+ code"""
+ |${evals.map(_.code).mkString("\n")}
+ |${setValues.mkString("\n")}
+ |$resultType $resultTerm = null;
+ |boolean ${ev.isNull} = false;
+ |try {
+ | $resultTerm = ($resultType) $refEvaluator.evaluate();
+ | ${ev.isNull} = $resultTerm == null;
+ |} catch (Throwable e) {
+ | throw QueryExecutionErrors.failedExecuteUserDefinedFunctionError(
+ |"${funcWrapper.functionClassName}",
+ |"${children.map(_.dataType.catalogString).mkString(", ")}",
+ |"${dataType.catalogString}",
+ |e);
+ |}
+ |${CodeGenerator.javaType(dataType)} ${ev.value} =
${CodeGenerator.defaultValue(dataType)};
+ |if (!${ev.isNull}) {
+ | ${ev.value} = $resultTerm;
+ |}
+ |""".stripMargin
+)
+ }
+}
+
+abstract class HiveUDFEvaluatorBase[UDFType <: AnyRef](
Review Comment:
can we move evaluators to a separated file?
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] cloud-fan commented on a diff in pull request #40397: [SPARK-42052][SQL] Codegen Support for HiveSimpleUDF
cloud-fan commented on code in PR #40397:
URL: https://github.com/apache/spark/pull/40397#discussion_r1137003368
##
sql/hive/src/main/scala/org/apache/spark/sql/hive/hiveUDFs.scala:
##
@@ -49,68 +48,140 @@ private[hive] case class HiveSimpleUDF(
name: String, funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
extends Expression
with HiveInspectors
- with CodegenFallback
- with Logging
with UserDefinedExpression {
+ @transient
+ private lazy val evaluator = new HiveSimpleUDFEvaluator(funcWrapper,
children)
+
+ @transient
+ private val isUDFDeterministic = {
+val udfType =
evaluator.function.getClass.getAnnotation(classOf[HiveUDFType])
+udfType != null && udfType.deterministic() && !udfType.stateful()
+ }
+
override lazy val deterministic: Boolean = isUDFDeterministic &&
children.forall(_.deterministic)
override def nullable: Boolean = true
- @transient
- lazy val function = funcWrapper.createFunction[UDF]()
+ override def foldable: Boolean = isUDFDeterministic &&
children.forall(_.foldable)
+
+ override lazy val dataType: DataType =
javaTypeToDataType(evaluator.method.getGenericReturnType)
+
+ // TODO: Finish input output types.
+ override def eval(input: InternalRow): Any = {
+children.zipWithIndex.map {
+ case (child, idx) =>
+evaluator.setArg(idx, child.eval(input))
+}
+evaluator.evaluate()
+ }
+
+ override def toString: String = {
+s"$nodeName#${funcWrapper.functionClassName}(${children.mkString(",")})"
+ }
+
+ override def prettyName: String = name
+
+ override def sql: String = s"$name(${children.map(_.sql).mkString(", ")})"
+
+ override protected def withNewChildrenInternal(newChildren:
IndexedSeq[Expression]): Expression =
+copy(children = newChildren)
+
+ protected def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+evaluator.doGenCode(ctx, ev, dataType)
+ }
+}
+
+abstract class HiveUDFEvaluatorBase[UDFType <: AnyRef](
+funcWrapper: HiveFunctionWrapper, children: Seq[Expression])
+ extends HiveInspectors with Serializable {
@transient
- private lazy val method =
-
function.getResolver.getEvalMethod(children.map(_.dataType.toTypeInfo).asJava)
+ lazy val function = funcWrapper.createFunction[UDFType]()
+
+ def setArg(index: Int, arg: Any): Unit
+
+ def evaluate(): Any
+
+ final def doGenCode(ctx: CodegenContext, ev: ExprCode, dataType: DataType):
ExprCode = {
Review Comment:
It's weird to implement codegen in the evaluator. If we really want to
deduplicate the code, let's add `HiveUDFExpressionBase` later.
--
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: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
