Merge pull request ClickHouse#83029 from ClickHouse/backport/25.6/82950

Backport ClickHouse#82950 to 25.6: Fix filter conjuncts removal in StorageMerge

Author: clickhouse-gh[bot]

src/Analyzer/Utils.cpp

@@ -1122,7 +1122,9 @@ bool hasUnknownColumn(const QueryTreeNodePtr & node, QueryTreeNodePtr table_expr
             {
                 auto * column_node = current->as<ColumnNode>();
                 auto source = column_node->getColumnSourceOrNull();
-                if (source && table_expression && !source->isEqual(*table_expression))
+                /// Column source can be nullptr if JOIN node was replaced with a table expression.
+                /// In that case column is not from the specified table expression.
+                if (source != table_expression)
                     return true;
                 break;
             }

src/Storages/transformQueryForExternalDatabaseAnalyzer.cpp

@@ -60,7 +60,8 @@ class PrepareForExternalDatabaseVisitor : public InDepthQueryTreeVisitor<Prepare
 
 ASTPtr getASTForExternalDatabaseFromQueryTree(ContextPtr context, const QueryTreeNodePtr & query_tree, const QueryTreeNodePtr & table_expression)
 {
-    auto new_tree = query_tree->clone();
+    auto replacement_table_expression = table_expression->clone();
+    auto new_tree = query_tree->cloneAndReplace(table_expression, replacement_table_expression);
 
     PrepareForExternalDatabaseVisitor visitor;
     visitor.visit(new_tree);
@@ -71,9 +72,9 @@ ASTPtr getASTForExternalDatabaseFromQueryTree(ContextPtr context, const QueryTre
     if (const auto * join_node = join_tree->as<JoinNode>())
     {
         if (join_node->getKind() == JoinKind::Left)
-            allow_where = join_node->getLeftTableExpression()->isEqual(*table_expression);
+            allow_where = join_node->getLeftTableExpression()->isEqual(*replacement_table_expression);
         else if (join_node->getKind() == JoinKind::Right)
-            allow_where = join_node->getRightTableExpression()->isEqual(*table_expression);
+            allow_where = join_node->getRightTableExpression()->isEqual(*replacement_table_expression);
         else
             allow_where = (join_node->getKind() == JoinKind::Inner);
     }
@@ -83,9 +84,9 @@ ASTPtr getASTForExternalDatabaseFromQueryTree(ContextPtr context, const QueryTre
     if (allow_where)
     {
         if (query_node->hasPrewhere())
-            removeExpressionsThatDoNotDependOnTableIdentifiers(query_node->getPrewhere(), table_expression, context);
+            removeExpressionsThatDoNotDependOnTableIdentifiers(query_node->getPrewhere(), replacement_table_expression, context);
         if (query_node->hasWhere())
-            removeExpressionsThatDoNotDependOnTableIdentifiers(query_node->getWhere(), table_expression, context);
+            removeExpressionsThatDoNotDependOnTableIdentifiers(query_node->getWhere(), replacement_table_expression, context);
     }
 
     auto query_node_ast = query_node->toAST({ .add_cast_for_constants = false, .fully_qualified_identifiers = false });

tests/queries/0_stateless/03549_analyzer_fix_filter_removal.reference

@@ -0,0 +1,8 @@
+SET allow_experimental_analyzer = 1;
+
+CREATE TABLE m (`key` UInt32) ENGINE = Merge(currentDatabase(), 'a');
+CREATE TABLE b (`key` UInt32, `ID` UInt32) ENGINE = MergeTree ORDER BY key;
+
+CREATE TABLE a1 (`day` Date, `id` UInt32) ENGINE = Distributed('test_cluster_two_shards', currentDatabase(), a1_replicated, id);
+
+SELECT * FROM m INNER JOIN b USING (key) WHERE ID = 1; -- { serverError UNKNOWN_TABLE, ALL_CONNECTION_TRIES_FAILED }