Merge pull request #5366 from abyss7/storage-merge-sample

Check that underlying tables support sampling for StorageMerge

dbms/src/Storages/StorageMerge.cpp

@@ -41,6 +41,7 @@ namespace ErrorCodes
     extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
     extern const int NO_SUCH_COLUMN_IN_TABLE;
     extern const int BLOCKS_HAVE_DIFFERENT_STRUCTURE;
+    extern const int SAMPLING_NOT_SUPPORTED;
 }
 
 
@@ -218,6 +219,7 @@ BlockInputStreams StorageMerge::read(
         query_info.query, has_table_virtual_column, true, context.getCurrentQueryId());
 
     if (selected_tables.empty())
+        /// FIXME: do we support sampling in this case?
         return createSourceStreams(
             query_info, processed_stage, max_block_size, header, {}, {}, real_column_names, modified_context, 0, has_table_virtual_column);
 
@@ -234,6 +236,10 @@ BlockInputStreams StorageMerge::read(
         StoragePtr storage = it->first;
         TableStructureReadLockHolder struct_lock = it->second;
 
+        /// If sampling requested, then check that table supports it.
+        if (query_info.query->as<ASTSelectQuery>()->sample_size() && !storage->supportsSampling())
+            throw Exception("Illegal SAMPLE: table doesn't support sampling", ErrorCodes::SAMPLING_NOT_SUPPORTED);
+
         BlockInputStreams source_streams;
 
         if (current_streams)

dbms/tests/queries/0_stateless/00578_merge_table_sampling.reference

+494
+331
+576
+709
+903
+378
+498
+102
+861
+97
+494
+331
+576
+709
+903
+378
+498
+102
+861
+97

dbms/tests/queries/0_stateless/00578_merge_table_sampling.sql

+DROP TABLE IF EXISTS test.numbers1;
+DROP TABLE IF EXISTS test.numbers2;
+
+CREATE TABLE test.numbers1 ENGINE = Memory AS SELECT number FROM numbers(1000);
+CREATE TABLE test.numbers2 ENGINE = Memory AS SELECT number FROM numbers(1000);
+
+SELECT * FROM merge(test, '^numbers\\d+$') SAMPLE 0.1; -- { serverError 141 }
+
+DROP TABLE test.numbers1;
+DROP TABLE test.numbers2;
+
+CREATE TABLE test.numbers1 ENGINE = MergeTree ORDER BY intHash32(number) SAMPLE BY intHash32(number) AS SELECT number FROM numbers(1000);
+CREATE TABLE test.numbers2 ENGINE = MergeTree ORDER BY intHash32(number) SAMPLE BY intHash32(number) AS SELECT number FROM numbers(1000);
+
+SELECT * FROM merge(test, '^numbers\\d+$') SAMPLE 0.01;
+
+DROP TABLE test.numbers1;
+DROP TABLE test.numbers2;