Merge pull request #57855 from kitaisreal/interpreter-create-query-sample-block-fix

InterpreterCreateQuery sample block fix

src/Interpreters/InterpreterCreateQuery.cpp

@@ -786,10 +786,28 @@ InterpreterCreateQuery::TableProperties InterpreterCreateQuery::getTableProperti
         }
         else
         {
-            as_select_sample = InterpreterSelectWithUnionQuery::getSampleBlock(create.select->clone(),
-                getContext(),
-                false /* is_subquery */,
-                create.isParameterizedView());
+            /** To get valid sample block we need to prepare query without only_analyze, because we need to execute scalar
+              * subqueries. Otherwise functions that expect only constant arguments will throw error during query analysis,
+              * because the result of scalar subquery is not a constant.
+              *
+              * Example:
+              * CREATE MATERIALIZED VIEW test_mv ENGINE=MergeTree ORDER BY arr
+              * AS
+              * WITH (SELECT '\d[a-z]') AS constant_value
+              * SELECT extractAll(concat(toString(number), 'a'), assumeNotNull(constant_value)) AS arr
+              * FROM test_table;
+              *
+              * For new analyzer this issue does not exists because we always execute scalar subqueries.
+              * We can improve this in new analyzer, and execute scalar subqueries only in contexts when we expect constant
+              * for example: LIMIT, OFFSET, functions parameters, functions constant only arguments.
+              */
+
+            SelectQueryOptions options;
+            if (create.isParameterizedView())
+                options = options.createParameterizedView();
+
+            InterpreterSelectWithUnionQuery interpreter(create.select->clone(), getContext(), options);
+            as_select_sample = interpreter.getSampleBlock();
         }
 
         properties.columns = ColumnsDescription(as_select_sample.getNamesAndTypesList());
@@ -1223,7 +1241,7 @@ BlockIO InterpreterCreateQuery::createTable(ASTCreateQuery & create)
             {
                 input_block = InterpreterSelectWithUnionQuery(create.select->clone(),
                     getContext(),
-                    SelectQueryOptions().analyze()).getSampleBlock();
+                    {}).getSampleBlock();
             }
 
             Block output_block = to_table->getInMemoryMetadataPtr()->getSampleBlock();

src/Processors/Transforms/buildPushingToViewsChain.cpp

@@ -341,7 +341,7 @@ Chain buildPushingToViewsChain(
             if (select_context->getSettingsRef().allow_experimental_analyzer)
                 header = InterpreterSelectQueryAnalyzer::getSampleBlock(query, select_context);
             else
-                header = InterpreterSelectQuery(query, select_context, SelectQueryOptions().analyze()).getSampleBlock();
+                header = InterpreterSelectQuery(query, select_context, SelectQueryOptions()).getSampleBlock();
 
             /// Insert only columns returned by select.
             Names insert_columns;

tests/queries/0_stateless/02174_cte_scalar_cache_mv.reference

@@ -18,7 +18,7 @@
 89	89	89	89	5
 94	94	94	94	5
 99	99	99	99	5
-02177_MV	3	80	26
+02177_MV	7	80	22
 10
 40
 70
@@ -60,4 +60,4 @@
 178
 188
 198
-02177_MV_3	19	0	2
+02177_MV_3	20	0	1

tests/queries/0_stateless/02943_create_query_interpreter_sample_block_fix.reference

@@ -0,0 +1,9 @@
+['0a']
+--
+['0a']
+['1a']
+--
+['0a']
+--
+['0a']
+['1a']

tests/queries/0_stateless/02943_create_query_interpreter_sample_block_fix.sql

@@ -0,0 +1,52 @@
+DROP TABLE IF EXISTS test_table;
+CREATE TABLE test_table
+(
+    number UInt64
+)
+ENGINE=MergeTree ORDER BY number;
+
+DROP VIEW IF EXISTS test_mv;
+CREATE MATERIALIZED VIEW test_mv ENGINE=MergeTree ORDER BY arr
+AS
+WITH (SELECT '\d[a-z]') AS constant_value
+SELECT extractAll(concat(toString(number), 'a'), assumeNotNull(constant_value)) AS arr
+FROM test_table;
+
+INSERT INTO test_table VALUES (0);
+SELECT * FROM test_mv ORDER BY arr;
+
+SELECT '--';
+
+INSERT INTO test_table VALUES (1);
+SELECT * FROM test_mv ORDER BY arr;
+
+SELECT '--';
+
+TRUNCATE test_table;
+
+DROP TABLE IF EXISTS regex_test_table;
+CREATE TABLE regex_test_table
+(
+    regex String
+)
+ENGINE = MergeTree ORDER BY regex;
+
+INSERT INTO regex_test_table VALUES ('\d[a-z]');
+
+DROP VIEW test_mv;
+CREATE MATERIALIZED VIEW test_mv ENGINE=MergeTree ORDER BY arr
+AS
+WITH (SELECT regex FROM regex_test_table) AS constant_value
+SELECT extractAll(concat(toString(number), 'a'), assumeNotNull(constant_value)) AS arr
+FROM test_table;
+
+INSERT INTO test_table VALUES (0);
+SELECT * FROM test_mv ORDER BY arr;
+
+SELECT '--';
+
+INSERT INTO test_table VALUES (1);
+SELECT * FROM test_mv ORDER BY arr;
+
+DROP VIEW test_mv;
+DROP TABLE test_table;