anchovYu commented on code in PR #39508: URL: https://github.com/apache/spark/pull/39508#discussion_r1091428757
########## sql/core/src/test/resources/sql-tests/inputs/column-resolution-aggregate.sql: ########## @@ -0,0 +1,30 @@ +-- Tests covering column resolution priority in Aggregate. + +CREATE TEMPORARY VIEW v1 AS VALUES (1, 1, 1), (2, 2, 1) AS t(a, b, k); +CREATE TEMPORARY VIEW v2 AS VALUES (1, 1, 1), (2, 2, 1) AS t(x, y, all); + +-- Relation output columns have higher priority than lateral column alias. This query +-- should fail as `b` is not in GROUP BY. +SELECT max(a) AS b, b FROM v1 GROUP BY k; + +-- Lateral column alias has higher priority than outer reference. +SELECT a FROM v1 WHERE (12, 13) IN (SELECT max(x + 10) AS a, a + 1 FROM v2); + +-- Relation output columns have higher priority than GROUP BY alias. This query should +-- fail as `a` is not in GROUP BY. +SELECT a AS k FROM v1 GROUP BY k; + +-- Relation output columns have higher priority than GROUP BY ALL. This query should +-- fail as `x` is not in GROUP BY. +SELECT x FROM v2 GROUP BY all; + +-- GROUP BY alias has higher priority than GROUP BY ALL, this query fails as `b` is not in GROUP BY. +SELECT a AS all, b FROM v1 GROUP BY all; + +-- GROUP BY alias/ALL does not support lateral column alias. +SELECT k AS lca, lca + 1 AS col FROM v1 GROUP BY k, col; +SELECT k AS lca, lca + 1 AS col FROM v1 GROUP BY all; Review Comment: Could you add one more test which is how I misunderstood the error message: ``` SELECT k AS lca, lca + 1 AS col FROM v1 GROUP BY lca; ``` Having LCA in SELECT list but group by alias not containing LCA should still resolve. ########## sql/core/src/test/resources/sql-tests/inputs/column-resolution-sort.sql: ########## @@ -0,0 +1,20 @@ +--SET spark.sql.leafNodeDefaultParallelism=1 +-- Tests covering column resolution priority in Sort. + +CREATE TEMPORARY VIEW v1 AS VALUES (1, 2, 2), (2, 1, 1) AS t(a, b, k); +CREATE TEMPORARY VIEW v2 AS VALUES (1, 2, 2), (2, 1, 1) AS t(a, b, all); + +-- Relation output columns have higher priority than missing reference. +-- Results will be [2, 1] if we order by the column `v1.b`. Review Comment: Test is good. just comment needs correction. ########## sql/core/src/test/resources/sql-tests/inputs/column-resolution-sort.sql: ########## @@ -0,0 +1,20 @@ +--SET spark.sql.leafNodeDefaultParallelism=1 +-- Tests covering column resolution priority in Sort. + +CREATE TEMPORARY VIEW v1 AS VALUES (1, 2, 2), (2, 1, 1) AS t(a, b, k); +CREATE TEMPORARY VIEW v2 AS VALUES (1, 2, 2), (2, 1, 1) AS t(a, b, all); + +-- Relation output columns have higher priority than missing reference. +-- Results will be [2, 1] if we order by the column `v1.b`. Review Comment: Missing reference won't be able to cover this case, it should only add grouping expressions to the Aggregate. v1.b is not a grouping expression, so if order by v1.b the query should throw exception. ########## sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ResolveReferencesInSort.scala: ########## @@ -0,0 +1,81 @@ +/* + * 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.analysis + +import org.apache.spark.sql.catalyst.SQLConfHelper +import org.apache.spark.sql.catalyst.expressions.SortOrder +import org.apache.spark.sql.catalyst.plans.logical.{LogicalPlan, Project, Sort} + +/** + * A virtual rule to resolve [[UnresolvedAttribute]] in [[Sort]]. It's only used by the real + * rule `ResolveReferences`. The column resolution order for [[Sort]] is: + * 1. Resolves the column to [[AttributeReference]] with the output of the child plan. This + * includes metadata columns as well. + * 2. Resolves the column to a literal function which is allowed to be invoked without braces, e.g. + * `SELECT col, current_date FROM t`. + * 3. If the child plan is Aggregate, resolves the column to [[TempResolvedColumn]] with the output + * of Aggregate's child plan. This is to allow Sort to host grouping expressions and aggregate + * functions, which can be pushed down to the Aggregate later. For example, + * `SELECT max(a) FROM t GROUP BY b ORDER BY min(a)`. + * 4. Resolves the column to [[AttributeReference]] with the output of a descendant plan node. + * Spark will propagate the missing attributes from the descendant plan node to the Sort node. + * This is to allow users to ORDER BY columns that are not in the SELECT clause, which is + * widely supported in other SQL dialects. For example, `SELECT a FROM t ORDER BY b`. + * 5. If the order by expressions only have one single unresolved column named ALL, expanded it to + * include all columns in the SELECT list. This is to support SQL pattern like + * `SELECT col1, col2 FROM t ORDER BY ALL`. This should also support specifying asc/desc, and + * nulls first/last. + * 6. Resolves the column to outer references with the outer plan if we are resolving subquery + * expressions. + */ +object ResolveReferencesInSort extends SQLConfHelper with ColumnResolutionHelper { + + def apply(s: Sort): LogicalPlan = { + val resolvedNoOuter = s.order.map(resolveExpressionByPlanOutput(_, s.child)) + val resolvedWithAgg = resolvedNoOuter.map(resolveColWithAgg(_, s.child)) + val (missingAttrResolved, newChild) = resolveExprsAndAddMissingAttrs(resolvedWithAgg, s.child) + val orderByAllResolved = resolveOrderByAll( + s.global, newChild, missingAttrResolved.map(_.asInstanceOf[SortOrder])) + val finalOrdering = orderByAllResolved.map(e => resolveOuterRef(e).asInstanceOf[SortOrder]) Review Comment: haha I forgot .. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
