Merge pull request ClickHouse#80158 from Avogar/merge-engine-subcolumns

Fix reading subcolumns from Merge engine

Author: Avogar

src/Storages/StorageMerge.cpp

@@ -529,14 +529,6 @@ void ReadFromMerge::initializePipeline(QueryPipelineBuilder & pipeline, const Bu
     {
         auto & child_plan = child_plans->at(i);
         const auto & table = *table_it;
-
-        const auto storage = std::get<1>(table);
-        const auto storage_metadata_snapshot = storage->getInMemoryMetadataPtr();
-        const auto nested_storage_snapshot = storage->getStorageSnapshot(storage_metadata_snapshot, context);
-
-        Names column_names_as_aliases;
-        Aliases aliases;
-
         auto source_pipeline = buildPipeline(child_plan, common_processed_stage);
 
         if (source_pipeline && source_pipeline->initialized())
@@ -580,7 +572,7 @@ void ReadFromMerge::filterTablesAndCreateChildrenPlans()
     has_database_virtual_column = false;
     has_table_virtual_column = false;
     column_names.clear();
-    column_names.reserve(column_names.size());
+    column_names.reserve(all_column_names.size());
 
     for (const auto & column_name : all_column_names)
     {
@@ -665,7 +657,7 @@ std::vector<ReadFromMerge::ChildPlan> ReadFromMerge::createChildrenPlans(SelectQ
 
             if (storage_metadata_snapshot->getColumns().empty())
             {
-                /// (Assuming that view has empty list of columns iff it's parameterized.)
+                /// (Assuming that view has empty list of columns if it's parameterized.)
                 if (storage->isView() && storage->as<StorageView>() && storage->as<StorageView>()->isParameterizedView())
                     throw Exception(ErrorCodes::STORAGE_REQUIRES_PARAMETER, "Parameterized view can't be queried through a Merge table.");
                 else
@@ -747,7 +739,7 @@ std::vector<ReadFromMerge::ChildPlan> ReadFromMerge::createChildrenPlans(SelectQ
 
             Names column_names_to_read = column_names_as_aliases.empty() ? std::move(real_column_names) : std::move(column_names_as_aliases);
 
-            std::erase_if(column_names_to_read, [existing_columns = nested_storage_snapshot->getAllColumnsDescription()](const auto & column_name){ return !existing_columns.has(column_name); });
+            std::erase_if(column_names_to_read, [existing_columns = nested_storage_snapshot->getAllColumnsDescription()](const auto & column_name){ return !existing_columns.has(column_name) && !existing_columns.hasSubcolumn(column_name); });
 
             auto child = createPlanForTable(
                 nested_storage_snapshot,

tests/queries/0_stateless/03457_merge_engine_subcolumns.reference

@@ -0,0 +1,21 @@
+1	(2,3)	s1
+4	(5,6)	s2
+2
+5
+3
+6
+1	2
+4	5
+s1	2
+s2	5
+2	3
+5	6
+1	2	3
+4	5	6
+s1	2	3
+s2	5	6
+1	2	3	s1
+4	5	6	s2
+42	42.42	[1,2,3]	2020-01-01
+\N	\N	\N	2020-01-02
+43	43.43	[4,5,6]	\N

tests/queries/0_stateless/03457_merge_engine_subcolumns.sql

@@ -0,0 +1,36 @@
+drop table if exists test1;
+drop table if exists test2;
+drop table if exists test_merge;
+
+create table test1 (x UInt64, t Tuple(a UInt32, b UInt32), y String) engine=Memory;
+create table test2 (x UInt64, t Tuple(a UInt32, b UInt32), y String) engine=Memory;
+create table test_merge (x UInt64, t Tuple(a UInt32, b UInt32), y String) engine=Merge(currentDatabase(), 'test');
+
+insert into test1 select 1, tuple(2, 3), 's1';
+insert into test2 select 4, tuple(5, 6), 's2';
+
+set allow_suspicious_types_in_order_by=1;
+
+select * from test_merge order by all;
+select t.a from test_merge order by all;
+select t.b from test_merge order by all;
+select x, t.a from test_merge order by all;
+select y, t.a from test_merge order by all;
+select t.a, t.b from test_merge order by all;
+select x, t.a, t.b from test_merge order by all;
+select y, t.a, t.b from test_merge order by all;
+select x, t.a, t.b, y from test_merge order by all;
+
+drop table test_merge;
+drop table test1;
+drop table test2;
+
+drop table if exists test;
+create table test (json JSON) engine=Memory;
+create table test_merge (json JSON) engine=Merge(currentDatabase(), 'test');
+insert into test values ('{"a" : {"b" : 42, "g" : 42.42}, "c" : [1, 2, 3], "d" : "2020-01-01"}'), ('{"f" : "Hello, World!", "d" : "2020-01-02"}'), ('{"a" : {"b" : 43, "e" : 10, "g" : 43.43}, "c" : [4, 5, 6]}');
+
+select json.a.b, json.a.g, json.c, json.d from test_merge;
+drop table test_merge;
+drop table test;
+