Backport ClickHouse#88350 to 25.8: count returns wrong answer

Author: robot-clickhouse

src/Analyzer/Passes/RemoveUnusedProjectionColumnsPass.cpp

@@ -183,40 +183,6 @@ void RemoveUnusedProjectionColumnsPass::run(QueryTreeNodePtr & query_tree_node,
                 if (query_node->isDistinct())
                     continue;
             }
-            else
-            {
-                auto * union_node = query_or_union_node->as<UnionNode>();
-                chassert(union_node != nullptr);
-
-                /// We can't remove unused projections in the case of EXCEPT and INTERSECT
-                /// because it can lead to incorrect query results. Example:
-                ///
-                /// SELECT count()
-                /// FROM
-                /// (
-                ///     SELECT
-                ///         1 AS a,
-                ///         2 AS b
-                ///     INTERSECT ALL
-                ///     SELECT
-                ///         1,
-                ///         1
-                /// )
-                ///
-                /// Will be transformed into the following query with output 1 instead of 0:
-                ///
-                /// SELECT count()
-                /// FROM
-                /// (
-                ///     SELECT
-                ///         1 AS a, -- we must keep at least 1 column
-                ///     INTERSECT ALL
-                ///     SELECT
-                ///         1
-                /// );
-                if (union_node->getUnionMode() > SelectUnionMode::UNION_DISTINCT)
-                    continue;
-            }
 
             auto used_projection_indexes = convertUsedColumnNamesToUsedProjectionIndexes(query_or_union_node, used_columns);
             updateUsedProjectionIndexes(query_or_union_node, used_projection_indexes);

src/Analyzer/UnionNode.cpp

@@ -122,6 +122,35 @@ void UnionNode::removeUnusedProjectionColumns(const std::unordered_set<size_t> &
     if (recursive_cte_table)
         return;
 
+    /// We can't remove unused projections in the case of EXCEPT and INTERSECT
+    /// because it can lead to incorrect query results. Example:
+    ///
+    /// SELECT count()
+    /// FROM
+    /// (
+    ///     SELECT
+    ///         1 AS a,
+    ///         2 AS b
+    ///     INTERSECT ALL
+    ///     SELECT
+    ///         1,
+    ///         1
+    /// )
+    ///
+    /// Will be transformed into the following query with output 1 instead of 0:
+    ///
+    /// SELECT count()
+    /// FROM
+    /// (
+    ///     SELECT
+    ///         1 AS a, -- we must keep at least 1 column
+    ///     INTERSECT ALL
+    ///     SELECT
+    ///         1
+    /// );
+    if (union_mode > SelectUnionMode::UNION_DISTINCT)
+        return;
+
     auto & query_nodes = getQueries().getNodes();
     for (auto & query_node : query_nodes)
     {

src/Parsers/SelectUnionMode.h

@@ -7,7 +7,7 @@ namespace DB
 {
 enum class SelectUnionMode : uint8_t
 {
-    UNION_DEFAULT,
+    UNION_DEFAULT = 0,
     UNION_ALL,
     UNION_DISTINCT,
     EXCEPT_DEFAULT,

tests/queries/0_stateless/03652_union_columns_2.reference

@@ -0,0 +1,8 @@
+run enable-analyzer=1
+0	5
+1	1
+2	1
+run enable-analyzer=1 ignore
+0	5
+1	1
+2	1

tests/queries/0_stateless/03652_union_columns_2.sql

@@ -0,0 +1,65 @@
+CREATE TABLE left (g UInt32, i UInt32)
+  ORDER BY (g, i);
+
+INSERT INTO left VALUES
+(0, 1), (0, 2), (0, 3), (0, 4), (0, 5), (2, 0);
+
+CREATE TABLE right (g UInt32, i UInt32)
+  ORDER BY (g, i);
+
+INSERT INTO right VALUES
+(0,0), (0, 3), (0, 4), (0, 6), (1, 0);
+
+SET enable_analyzer = 1;
+
+SELECT 'run enable-analyzer=1';
+with differences as
+    (
+        (
+            select g, i from left
+            where g BETWEEN 0 and 10
+            EXCEPT ALL
+            select g, i from right
+            where g BETWEEN 0 and 10
+        )
+        UNION ALL
+        (
+            select g, i from right
+            where g BETWEEN 0 and 10
+            EXCEPT ALL
+            select g, i from left
+            where g BETWEEN 0 and 10
+        )
+    ),
+diff_counts as
+    (
+        select g, count(*) from differences group by g
+    )
+select * from diff_counts
+ORDER BY g;
+
+SELECT 'run enable-analyzer=1 ignore';
+with differences as
+    (
+        (
+            select g, i from left
+            where g BETWEEN 0 and 10
+            EXCEPT ALL
+            select g, i from right
+            where g BETWEEN 0 and 10
+        )
+        UNION ALL
+        (
+            select g, i from right
+            where g BETWEEN 0 and 10
+            EXCEPT ALL
+            select g, i from left
+            where g BETWEEN 0 and 10
+        )
+    ),
+diff_counts as
+    (
+        select g, count(ignore(*)) from differences group by g
+    )
+select * from diff_counts
+ORDER BY g;