Merge pull request #34124 from ClickHouse/backport/22.1/33956

Backport #33956 to 22.1: Fix reading of nested tuples
ClickHouse · Jan 29, 2022 · bc393af · bc393af
2 parents d15926f + 28d42c2
commit bc393af
Show file tree

Hide file tree

Showing 4 changed files with 64 additions and 2 deletions.
diff --git a/src/DataTypes/DataTypeTuple.cpp b/src/DataTypes/DataTypeTuple.cpp
@@ -32,6 +32,7 @@ namespace ErrorCodes
     extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
     extern const int SIZES_OF_COLUMNS_IN_TUPLE_DOESNT_MATCH;
     extern const int ILLEGAL_INDEX;
+    extern const int LOGICAL_ERROR;
 }
 
 
@@ -156,8 +157,19 @@ MutableColumnPtr DataTypeTuple::createColumn() const
 
 MutableColumnPtr DataTypeTuple::createColumn(const ISerialization & serialization) const
 {
-    const auto & element_serializations =
-        assert_cast<const SerializationTuple &>(serialization).getElementsSerializations();
+    /// If we read subcolumn of nested Tuple, it may be wrapped to SerializationNamed
+    /// several times to allow to reconstruct the substream path name.
+    /// Here we don't need substream path name, so we drop first several wrapper serializations.
+
+    const auto * current_serialization = &serialization;
+    while (const auto * serialization_named = typeid_cast<const SerializationNamed *>(current_serialization))
+        current_serialization = serialization_named->getNested().get();
+
+    const auto * serialization_tuple = typeid_cast<const SerializationTuple *>(current_serialization);
+    if (!serialization_tuple)
+        throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected serialization to create column of type Tuple");
+
+    const auto & element_serializations = serialization_tuple->getElementsSerializations();
 
     size_t size = elems.size();
     assert(element_serializations.size() == size);

diff --git a/src/DataTypes/Serializations/SerializationNamed.h b/src/DataTypes/Serializations/SerializationNamed.h
@@ -5,6 +5,11 @@
 namespace DB
 {
 
+/// Serialization wrapper that acts like nested serialization,
+/// but adds a passed name to the substream path like the
+/// read column was the tuple element with this name.
+/// It's used while reading subcolumns of complex types.
+/// In particular while reading components of named tuples.
 class SerializationNamed final : public SerializationWrapper
 {
 private:

diff --git a/tests/queries/0_stateless/02184_nested_tuple.reference b/tests/queries/0_stateless/02184_nested_tuple.reference
@@ -0,0 +1,7 @@
+{"endUserIDs":{"_experience":{"aaid":{"id":"id_1","namespace":{"code":"code_1"},"primary":1},"mcid":{"id":"id_2","namespace":{"code":"code_2"},"primary":2}}}}
+{"endUserIDs._experience":{"aaid":{"id":"id_1","namespace":{"code":"code_1"},"primary":1},"mcid":{"id":"id_2","namespace":{"code":"code_2"},"primary":2}}}
+{"endUserIDs._experience.aaid":{"id":"id_1","namespace":{"code":"code_1"},"primary":1}}
+{"endUserIDs._experience.aaid.id":"id_1"}
+{"endUserIDs._experience.aaid.namespace":{"code":"code_1"}}
+{"endUserIDs._experience.aaid.namespace.code":"code_1"}
+{"endUserIDs._experience.aaid.primary":1}
diff --git a/tests/queries/0_stateless/02184_nested_tuple.sql b/tests/queries/0_stateless/02184_nested_tuple.sql
@@ -0,0 +1,38 @@
+DROP TABLE IF EXISTS t_nested_tuple;
+
+CREATE TABLE t_nested_tuple
+(
+    endUserIDs Tuple(
+      _experience Tuple(
+          aaid Tuple(
+              id Nullable(String),
+              namespace Tuple(
+                  code LowCardinality(Nullable(String))
+              ),
+              primary LowCardinality(Nullable(UInt8))
+          ),
+          mcid Tuple(
+              id Nullable(String),
+              namespace Tuple(
+                  code LowCardinality(Nullable(String))
+              ),
+              primary LowCardinality(Nullable(UInt8))
+          )
+      )
+  )
+)
+ENGINE = MergeTree ORDER BY tuple();
+
+SET output_format_json_named_tuples_as_objects = 1;
+
+INSERT INTO t_nested_tuple FORMAT JSONEachRow {"endUserIDs":{"_experience":{"aaid":{"id":"id_1","namespace":{"code":"code_1"},"primary":1},"mcid":{"id":"id_2","namespace":{"code":"code_2"},"primary":2}}}};
+
+SELECT * FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience.aaid FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience.aaid.id FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience.aaid.namespace FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience.aaid.namespace.code FROM t_nested_tuple FORMAT JSONEachRow;
+SELECT endUserIDs._experience.aaid.primary FROM t_nested_tuple FORMAT JSONEachRow;
+
+DROP TABLE t_nested_tuple;