Merge pull request #19991 from ClickHouse/constant-propagation-fix

Do not use inputs which values are known constants in ActionsDAG.

src/Interpreters/ActionsDAG.cpp

@@ -39,7 +39,17 @@ ActionsDAG::ActionsDAG(const ColumnsWithTypeAndName & inputs_)
     for (const auto & input : inputs_)
     {
         if (input.column && isColumnConst(*input.column))
+        {
             addInput(input, true);
+
+            /// Here we also add column.
+            /// It will allow to remove input which is actually constant (after projection).
+            /// Also, some transforms from query pipeline may randomly materialize constants,
+            ///   without any respect to header structure. So, it is a way to drop materialized column and use
+            ///   constant value from header.
+            /// We cannot remove such input right now cause inputs positions are important in some cases.
+            addColumn(input, true);
+        }
         else
             addInput(input.name, input.type, true);
     }
@@ -116,7 +126,6 @@ ActionsDAG::Node & ActionsDAG::addAlias(Node & child, std::string alias, bool ca
     node.result_type = child.result_type;
     node.result_name = std::move(alias);
     node.column = child.column;
-    node.allow_constant_folding = child.allow_constant_folding;
     node.children.emplace_back(&child);
 
     return addNode(std::move(node), can_replace);
@@ -184,7 +193,6 @@ ActionsDAG::Node & ActionsDAG::addFunction(
     for (size_t i = 0; i < num_arguments; ++i)
     {
         auto & child = *node.children[i];
-        node.allow_constant_folding = node.allow_constant_folding && child.allow_constant_folding;
 
         ColumnWithTypeAndName argument;
         argument.column = child.column;
@@ -349,10 +357,15 @@ void ActionsDAG::removeUnusedActions()
         stack.push(node);
     }
 
-    /// We cannot remove arrayJoin because it changes the number of rows.
     for (auto & node : nodes)
     {
-        if (node.type == ActionType::ARRAY_JOIN && visited_nodes.count(&node) == 0)
+        /// We cannot remove function with side effects even if it returns constant (e.g. ignore(...)).
+        bool prevent_constant_folding = node.column && isColumnConst(*node.column) && !node.allow_constant_folding;
+        /// We cannot remove arrayJoin because it changes the number of rows.
+        bool is_array_join = node.type == ActionType::ARRAY_JOIN;
+
+        bool must_keep_node = is_array_join || prevent_constant_folding;
+        if (must_keep_node && visited_nodes.count(&node) == 0)
         {
             visited_nodes.insert(&node);
             stack.push(&node);
@@ -410,7 +423,6 @@ void ActionsDAG::addAliases(const NamesWithAliases & aliases, std::vector<Node *
             node.result_type = child->result_type;
             node.result_name = std::move(item.second);
             node.column = child->column;
-            node.allow_constant_folding = child->allow_constant_folding;
             node.children.emplace_back(child);
 
             auto & alias = addNode(std::move(node), true);

src/Storages/MergeTree/MergeTreeRangeReader.cpp

@@ -867,7 +867,7 @@ void MergeTreeRangeReader::executePrewhereActionsAndFilterColumns(ReadResult & r
     if (result.totalRowsPerGranule() == 0)
         result.setFilterConstFalse();
     /// If we need to filter in PREWHERE
-    else if (prewhere->need_filter || result.need_filter)
+    else if (prewhere->need_filter || result.need_filter || prewhere->remove_prewhere_column)
     {
         /// If there is a filter and without optimized
         if (result.getFilter() && last_reader_in_chain)

tests/queries/0_stateless/01671_merge_join_and_constants.reference

@@ -3,3 +3,4 @@
 │ b │ 2018-01-01 │ B │ 2018-01-01 │ 0.10      │
 │ c │ 2018-01-01 │ C │ 2018-01-01 │ 0.10      │
 └───┴────────────┴───┴────────────┴───────────┘
+\N	\N	\N	\N	0	0

tests/queries/0_stateless/01671_merge_join_and_constants.sql

@@ -11,5 +11,7 @@ set join_algorithm = 'partial_merge';
 
 SELECT * FROM table1 AS t1 ALL LEFT JOIN (SELECT *, '0.10', c, d AS b FROM table2) AS t2 USING (a, b) ORDER BY d ASC FORMAT PrettyCompact settings max_rows_in_join = 1;
 
+SELECT pow('0.0000000257', NULL), pow(pow(NULL, NULL), NULL) - NULL, (val + NULL) = (rval * 0), * FROM (SELECT (val + 256) = (NULL * NULL), toLowCardinality(toNullable(dummy)) AS val FROM system.one) AS s1 ANY LEFT JOIN (SELECT toLowCardinality(dummy) AS rval FROM system.one) AS s2 ON (val + 0) = (rval * 255) settings max_rows_in_join = 1;
+
 DROP TABLE IF EXISTS table1;
 DROP TABLE IF EXISTS table2;