calculate isRemote() from underlying tables of StorageMerge [#CLICKHOUSE-2911]

dbms/src/Interpreters/InJoinSubqueriesPreprocessor.cpp

@@ -209,9 +209,6 @@ void InJoinSubqueriesPreprocessor::process(ASTSelectQuery * query) const
 
 bool InJoinSubqueriesPreprocessor::hasAtLeastTwoShards(const IStorage & table) const
 {
-    if (!table.isRemote())
-        return false;
-
     const StorageDistributed * distributed = typeid_cast<const StorageDistributed *>(&table);
     if (!distributed)
         return false;

dbms/src/Interpreters/InterpreterSelectQuery.cpp

@@ -838,24 +838,13 @@ QueryProcessingStage::Enum InterpreterSelectQuery::executeFetchColumns()
         if (max_streams > 1 && !is_remote)
             max_streams *= settings.max_streams_to_max_threads_ratio;
 
-        ASTPtr actual_query_ptr;
-        if (storage->isRemote())
-        {
-            /// In case of a remote query, we send only SELECT, which will be executed.
-            actual_query_ptr = query.cloneFirstSelect();
-        }
-        else
-            actual_query_ptr = query_ptr;
-
         /// PREWHERE optimization
         {
             auto optimize_prewhere = [&](auto & merge_tree)
             {
-                const ASTSelectQuery & actual_select = typeid_cast<const ASTSelectQuery &>(*actual_query_ptr);
-
                 /// Try transferring some condition from WHERE to PREWHERE if enabled and viable
-                if (settings.optimize_move_to_prewhere && actual_select.where_expression && !actual_select.prewhere_expression && !actual_select.final())
-                    MergeTreeWhereOptimizer{actual_query_ptr, context, merge_tree.getData(), required_columns, log};
+                if (settings.optimize_move_to_prewhere && query.where_expression && !query.prewhere_expression && !query.final())
+                    MergeTreeWhereOptimizer{query_ptr, context, merge_tree.getData(), required_columns, log};
             };
 
             if (const StorageMergeTree * merge_tree = typeid_cast<const StorageMergeTree *>(storage.get()))
@@ -864,8 +853,7 @@ QueryProcessingStage::Enum InterpreterSelectQuery::executeFetchColumns()
                 optimize_prewhere(*merge_tree);
         }
 
-        streams = storage->read(required_columns, actual_query_ptr,
-            context, from_stage, max_block_size, max_streams);
+        streams = storage->read(required_columns, query_ptr, context, from_stage, max_block_size, max_streams);
 
         if (alias_actions)
         {
@@ -1316,11 +1304,6 @@ void InterpreterSelectQuery::executeLimit()
 
 void InterpreterSelectQuery::executeSubqueriesInSetsAndJoins(SubqueriesForSets & subqueries_for_sets)
 {
-    /// If the query is not distributed, then remove the creation of temporary tables from subqueries (intended for sending to remote servers).
-    if (!(storage && storage->isRemote()))
-        for (auto & elem : subqueries_for_sets)
-            elem.second.table.reset();
-
     const Settings & settings = context.getSettingsRef();
 
     executeUnion();

dbms/src/Parsers/ASTSelectQuery.cpp

@@ -173,7 +173,7 @@ ASTPtr ASTSelectQuery::clone() const
     return ptr;
 }
 
-ASTPtr ASTSelectQuery::cloneFirstSelect() const
+std::shared_ptr<ASTSelectQuery> ASTSelectQuery::cloneFirstSelect() const
 {
     auto res = cloneImpl(false);
     res->prev_union_all = nullptr;

dbms/src/Parsers/ASTSelectQuery.h

@@ -39,7 +39,7 @@ public:
     ASTPtr clone() const override;
 
     /// Get a deep copy of the first SELECT query tree.
-    ASTPtr cloneFirstSelect() const;
+    std::shared_ptr<ASTSelectQuery> cloneFirstSelect() const;
 
 private:
     std::shared_ptr<ASTSelectQuery> cloneImpl(bool traverse_union_all) const;

dbms/src/Storages/StorageDistributed.cpp

@@ -60,8 +60,8 @@ namespace
 /// Creates a copy of query, changes database and table names.
 ASTPtr rewriteSelectQuery(const ASTPtr & query, const std::string & database, const std::string & table)
 {
-    auto modified_query_ast = query->clone();
-    typeid_cast<ASTSelectQuery &>(*modified_query_ast).replaceDatabaseAndTable(database, table);
+    auto modified_query_ast = typeid_cast<const ASTSelectQuery &>(*query).cloneFirstSelect();
+    modified_query_ast->replaceDatabaseAndTable(database, table);
     return modified_query_ast;
 }
 

dbms/src/Storages/StorageMerge.cpp

@@ -50,6 +50,19 @@ StorageMerge::StorageMerge(
 {
 }
 
+bool StorageMerge::isRemote() const
+{
+    auto database = context.getDatabase(source_database);
+    auto iterator = database->getIterator();
+
+    while (iterator->isValid())
+    {
+        if (table_name_regexp.match(iterator->name()) && iterator->table()->isRemote())
+            return true;
+    }
+    return false;
+}
+
 NameAndTypePair StorageMerge::getColumn(const String & column_name) const
 {
     auto type = VirtualColumnFactory::tryGetType(column_name);

dbms/src/Storages/StorageMerge.h

@@ -20,6 +20,8 @@ public:
     std::string getName() const override { return "Merge"; }
     std::string getTableName() const override { return name; }
 
+    bool isRemote() const override;
+
     /// The check is delayed to the read method. It checks the support of the tables used.
     bool supportsSampling() const override { return true; }
     bool supportsPrewhere() const override { return true; }