Merge pull request ClickHouse#44716 from ClickHouse/vdimir/merge_engi…

…ne_nullable

src/Storages/StorageMerge.cpp

@@ -145,8 +145,43 @@ bool StorageMerge::canMoveConditionsToPrewhere() const
     /// If new table that matches regexp for current storage and doesn't support PREWHERE
     /// will appear after this check and before calling "read" method, the optimized query may fail.
     /// Since it's quite rare case, we just ignore this possibility.
+    const auto & table_doesnt_support_prewhere = getFirstTable([](const auto & table) { return !table->canMoveConditionsToPrewhere(); });
+    bool can_move = (table_doesnt_support_prewhere == nullptr);
 
-    return getFirstTable([](const auto & table) { return !table->canMoveConditionsToPrewhere(); }) == nullptr;
+    if (!can_move)
+        return false;
+
+    if (!getInMemoryMetadataPtr())
+        return false;
+
+    std::unordered_map<std::string, const IDataType *> column_types;
+    for (const auto & name_type : getInMemoryMetadataPtr()->getColumns().getAll())
+    {
+        column_types.emplace(name_type.name, name_type.type.get());
+    }
+
+    /// Check that all tables have the same column types, otherwise prewhere will fail
+    forEachTable([&](const StoragePtr & table)
+    {
+        const auto & metadata_ptr = table->getInMemoryMetadataPtr();
+        if (!metadata_ptr)
+            can_move = false;
+
+        if (!can_move)
+            return;
+
+        for (const auto & column : metadata_ptr->getColumns().getAll())
+        {
+            const auto * src_type = column_types[column.name];
+            if (src_type && !src_type->equals(*column.type))
+            {
+                can_move = false;
+                return;
+            }
+        }
+    });
+
+    return can_move;
 }
 
 bool StorageMerge::mayBenefitFromIndexForIn(const ASTPtr & left_in_operand, ContextPtr query_context, const StorageMetadataPtr & /*metadata_snapshot*/) const

tests/queries/0_stateless/02518_merge_engine_nullable_43324.sql

@@ -0,0 +1,18 @@
+
+DROP TABLE IF EXISTS foo;
+DROP TABLE IF EXISTS foo__fuzz_0;
+DROP TABLE IF EXISTS foo_merge;
+
+CREATE TABLE foo (`Id` Int32, `Val` Int32) ENGINE = MergeTree ORDER BY Id;
+CREATE TABLE foo__fuzz_0 (`Id` Int64, `Val` Nullable(Int32)) ENGINE = MergeTree ORDER BY Id;
+
+INSERT INTO foo SELECT number, number % 5 FROM numbers(10);
+INSERT INTO foo__fuzz_0 SELECT number, number % 5 FROM numbers(10);
+
+CREATE TABLE merge1 AS foo ENGINE = Merge(currentDatabase(), '^foo');
+CREATE TABLE merge2 (`Id` Int32, `Val` Int32) ENGINE = Merge(currentDatabase(), '^foo');
+CREATE TABLE merge3 (`Id` Int32, `Val` Int32) ENGINE = Merge(currentDatabase(), '^foo__fuzz_0');
+
+SELECT * FROM merge1 WHERE Val = 3 AND Val = 1;
+SELECT * FROM merge2 WHERE Val = 3 AND Val = 1;
+SELECT * FROM merge3 WHERE Val = 3 AND Val = 1;