Repository: spark Updated Branches: refs/heads/master ac10171be -> 881c5c807
[SPARK-22548][SQL] Incorrect nested AND expression pushed down to JDBC data source ## What changes were proposed in this pull request? Letâs say I have a nested AND expression shown below and p2 can not be pushed down, (p1 AND p2) OR p3 In current Spark code, during data source filter translation, (p1 AND p2) is returned as p1 only and p2 is simply lost. This issue occurs with JDBC data source and is similar to [SPARK-12218](https://github.com/apache/spark/pull/10362) for Parquet. When we have AND nested below another expression, we should either push both legs or nothing. Note that: - The current Spark code will always split conjunctive predicate before it determines if a predicate can be pushed down or not - If I have (p1 AND p2) AND p3, it will be split into p1, p2, p3. There won't be nested AND expression. - The current Spark code logic for OR is OK. It either pushes both legs or nothing. The same translation method is also called by Data Source V2. ## How was this patch tested? Added new unit test cases to JDBCSuite gatorsmile Author: Jia Li <[email protected]> Closes #19776 from jliwork/spark-22548. Project: http://git-wip-us.apache.org/repos/asf/spark/repo Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/881c5c80 Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/881c5c80 Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/881c5c80 Branch: refs/heads/master Commit: 881c5c807304a305ef96e805d51afbde097f7f4f Parents: ac10171 Author: Jia Li <[email protected]> Authored: Tue Nov 21 17:30:02 2017 -0800 Committer: gatorsmile <[email protected]> Committed: Tue Nov 21 17:30:02 2017 -0800 ---------------------------------------------------------------------- .../datasources/DataSourceStrategy.scala | 14 +- .../datasources/DataSourceStrategySuite.scala | 231 +++++++++++++++++++ .../org/apache/spark/sql/jdbc/JDBCSuite.scala | 9 +- 3 files changed, 250 insertions(+), 4 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/spark/blob/881c5c80/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala ---------------------------------------------------------------------- diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala index 04d6f3f..400f2e0 100644 --- a/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala +++ b/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala @@ -497,7 +497,19 @@ object DataSourceStrategy { Some(sources.IsNotNull(a.name)) case expressions.And(left, right) => - (translateFilter(left) ++ translateFilter(right)).reduceOption(sources.And) + // See SPARK-12218 for detailed discussion + // It is not safe to just convert one side if we do not understand the + // other side. Here is an example used to explain the reason. + // Let's say we have (a = 2 AND trim(b) = 'blah') OR (c > 0) + // and we do not understand how to convert trim(b) = 'blah'. + // If we only convert a = 2, we will end up with + // (a = 2) OR (c > 0), which will generate wrong results. + // Pushing one leg of AND down is only safe to do at the top level. + // You can see ParquetFilters' createFilter for more details. + for { + leftFilter <- translateFilter(left) + rightFilter <- translateFilter(right) + } yield sources.And(leftFilter, rightFilter) case expressions.Or(left, right) => for { http://git-wip-us.apache.org/repos/asf/spark/blob/881c5c80/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategySuite.scala ---------------------------------------------------------------------- diff --git a/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategySuite.scala b/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategySuite.scala new file mode 100644 index 0000000..f20aded --- /dev/null +++ b/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategySuite.scala @@ -0,0 +1,231 @@ +/* + * 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.execution.datasources + +import org.apache.spark.sql.catalyst.dsl.expressions._ +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.catalyst.plans.PlanTest +import org.apache.spark.sql.sources +import org.apache.spark.sql.test.SharedSQLContext + +class DataSourceStrategySuite extends PlanTest with SharedSQLContext { + + test("translate simple expression") { + val attrInt = 'cint.int + val attrStr = 'cstr.string + + testTranslateFilter(EqualTo(attrInt, 1), Some(sources.EqualTo("cint", 1))) + testTranslateFilter(EqualTo(1, attrInt), Some(sources.EqualTo("cint", 1))) + + testTranslateFilter(EqualNullSafe(attrStr, Literal(null)), + Some(sources.EqualNullSafe("cstr", null))) + testTranslateFilter(EqualNullSafe(Literal(null), attrStr), + Some(sources.EqualNullSafe("cstr", null))) + + testTranslateFilter(GreaterThan(attrInt, 1), Some(sources.GreaterThan("cint", 1))) + testTranslateFilter(GreaterThan(1, attrInt), Some(sources.LessThan("cint", 1))) + + testTranslateFilter(LessThan(attrInt, 1), Some(sources.LessThan("cint", 1))) + testTranslateFilter(LessThan(1, attrInt), Some(sources.GreaterThan("cint", 1))) + + testTranslateFilter(GreaterThanOrEqual(attrInt, 1), Some(sources.GreaterThanOrEqual("cint", 1))) + testTranslateFilter(GreaterThanOrEqual(1, attrInt), Some(sources.LessThanOrEqual("cint", 1))) + + testTranslateFilter(LessThanOrEqual(attrInt, 1), Some(sources.LessThanOrEqual("cint", 1))) + testTranslateFilter(LessThanOrEqual(1, attrInt), Some(sources.GreaterThanOrEqual("cint", 1))) + + testTranslateFilter(InSet(attrInt, Set(1, 2, 3)), Some(sources.In("cint", Array(1, 2, 3)))) + + testTranslateFilter(In(attrInt, Seq(1, 2, 3)), Some(sources.In("cint", Array(1, 2, 3)))) + + testTranslateFilter(IsNull(attrInt), Some(sources.IsNull("cint"))) + testTranslateFilter(IsNotNull(attrInt), Some(sources.IsNotNull("cint"))) + + // cint > 1 AND cint < 10 + testTranslateFilter(And( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10)), + Some(sources.And( + sources.GreaterThan("cint", 1), + sources.LessThan("cint", 10)))) + + // cint >= 8 OR cint <= 2 + testTranslateFilter(Or( + GreaterThanOrEqual(attrInt, 8), + LessThanOrEqual(attrInt, 2)), + Some(sources.Or( + sources.GreaterThanOrEqual("cint", 8), + sources.LessThanOrEqual("cint", 2)))) + + testTranslateFilter(Not(GreaterThanOrEqual(attrInt, 8)), + Some(sources.Not(sources.GreaterThanOrEqual("cint", 8)))) + + testTranslateFilter(StartsWith(attrStr, "a"), Some(sources.StringStartsWith("cstr", "a"))) + + testTranslateFilter(EndsWith(attrStr, "a"), Some(sources.StringEndsWith("cstr", "a"))) + + testTranslateFilter(Contains(attrStr, "a"), Some(sources.StringContains("cstr", "a"))) + } + + test("translate complex expression") { + val attrInt = 'cint.int + + // ABS(cint) - 2 <= 1 + testTranslateFilter(LessThanOrEqual( + // Expressions are not supported + // Functions such as 'Abs' are not supported + Subtract(Abs(attrInt), 2), 1), None) + + // (cin1 > 1 AND cint < 10) OR (cint > 50 AND cint > 100) + testTranslateFilter(Or( + And( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10) + ), + And( + GreaterThan(attrInt, 50), + LessThan(attrInt, 100))), + Some(sources.Or( + sources.And( + sources.GreaterThan("cint", 1), + sources.LessThan("cint", 10)), + sources.And( + sources.GreaterThan("cint", 50), + sources.LessThan("cint", 100))))) + + // SPARK-22548 Incorrect nested AND expression pushed down to JDBC data source + // (cint > 1 AND ABS(cint) < 10) OR (cint < 50 AND cint > 100) + testTranslateFilter(Or( + And( + GreaterThan(attrInt, 1), + // Functions such as 'Abs' are not supported + LessThan(Abs(attrInt), 10) + ), + And( + GreaterThan(attrInt, 50), + LessThan(attrInt, 100))), None) + + // NOT ((cint <= 1 OR ABS(cint) >= 10) AND (cint <= 50 OR cint >= 100)) + testTranslateFilter(Not(And( + Or( + LessThanOrEqual(attrInt, 1), + // Functions such as 'Abs' are not supported + GreaterThanOrEqual(Abs(attrInt), 10) + ), + Or( + LessThanOrEqual(attrInt, 50), + GreaterThanOrEqual(attrInt, 100)))), None) + + // (cint = 1 OR cint = 10) OR (cint > 0 OR cint < -10) + testTranslateFilter(Or( + Or( + EqualTo(attrInt, 1), + EqualTo(attrInt, 10) + ), + Or( + GreaterThan(attrInt, 0), + LessThan(attrInt, -10))), + Some(sources.Or( + sources.Or( + sources.EqualTo("cint", 1), + sources.EqualTo("cint", 10)), + sources.Or( + sources.GreaterThan("cint", 0), + sources.LessThan("cint", -10))))) + + // (cint = 1 OR ABS(cint) = 10) OR (cint > 0 OR cint < -10) + testTranslateFilter(Or( + Or( + EqualTo(attrInt, 1), + // Functions such as 'Abs' are not supported + EqualTo(Abs(attrInt), 10) + ), + Or( + GreaterThan(attrInt, 0), + LessThan(attrInt, -10))), None) + + // In end-to-end testing, conjunctive predicate should has been split + // before reaching DataSourceStrategy.translateFilter. + // This is for UT purpose to test each [[case]]. + // (cint > 1 AND cint < 10) AND (cint = 6 AND cint IS NOT NULL) + testTranslateFilter(And( + And( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10) + ), + And( + EqualTo(attrInt, 6), + IsNotNull(attrInt))), + Some(sources.And( + sources.And( + sources.GreaterThan("cint", 1), + sources.LessThan("cint", 10)), + sources.And( + sources.EqualTo("cint", 6), + sources.IsNotNull("cint"))))) + + // (cint > 1 AND cint < 10) AND (ABS(cint) = 6 AND cint IS NOT NULL) + testTranslateFilter(And( + And( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10) + ), + And( + // Functions such as 'Abs' are not supported + EqualTo(Abs(attrInt), 6), + IsNotNull(attrInt))), None) + + // (cint > 1 OR cint < 10) AND (cint = 6 OR cint IS NOT NULL) + testTranslateFilter(And( + Or( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10) + ), + Or( + EqualTo(attrInt, 6), + IsNotNull(attrInt))), + Some(sources.And( + sources.Or( + sources.GreaterThan("cint", 1), + sources.LessThan("cint", 10)), + sources.Or( + sources.EqualTo("cint", 6), + sources.IsNotNull("cint"))))) + + // (cint > 1 OR cint < 10) AND (cint = 6 OR cint IS NOT NULL) + testTranslateFilter(And( + Or( + GreaterThan(attrInt, 1), + LessThan(attrInt, 10) + ), + Or( + // Functions such as 'Abs' are not supported + EqualTo(Abs(attrInt), 6), + IsNotNull(attrInt))), None) + } + + /** + * Translate the given Catalyst [[Expression]] into data source [[sources.Filter]] + * then verify against the given [[sources.Filter]]. + */ + def testTranslateFilter(catalystFilter: Expression, result: Option[sources.Filter]): Unit = { + assertResult(result) { + DataSourceStrategy.translateFilter(catalystFilter) + } + } +} http://git-wip-us.apache.org/repos/asf/spark/blob/881c5c80/sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala ---------------------------------------------------------------------- diff --git a/sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala b/sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala index 88a5f61..61571bc 100644 --- a/sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala +++ b/sql/core/src/test/scala/org/apache/spark/sql/jdbc/JDBCSuite.scala @@ -294,10 +294,13 @@ class JDBCSuite extends SparkFunSuite // This is a test to reflect discussion in SPARK-12218. // The older versions of spark have this kind of bugs in parquet data source. - val df1 = sql("SELECT * FROM foobar WHERE NOT (THEID != 2 AND NAME != 'mary')") - val df2 = sql("SELECT * FROM foobar WHERE NOT (THEID != 2) OR NOT (NAME != 'mary')") + val df1 = sql("SELECT * FROM foobar WHERE NOT (THEID != 2) OR NOT (NAME != 'mary')") assert(df1.collect.toSet === Set(Row("mary", 2))) - assert(df2.collect.toSet === Set(Row("mary", 2))) + + // SPARK-22548: Incorrect nested AND expression pushed down to JDBC data source + val df2 = sql("SELECT * FROM foobar " + + "WHERE (THEID > 0 AND TRIM(NAME) = 'mary') OR (NAME = 'fred')") + assert(df2.collect.toSet === Set(Row("fred", 1), Row("mary", 2))) def checkNotPushdown(df: DataFrame): DataFrame = { val parentPlan = df.queryExecution.executedPlan --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
