ClickHouse · robot-ch-test-poll1 · Nov 14, 2025 · Nov 14, 2025 · Nov 14, 2025 · Nov 14, 2025
diff --git a/src/Processors/Formats/Impl/BSONEachRowRowInputFormat.cpp b/src/Processors/Formats/Impl/BSONEachRowRowInputFormat.cpp
@@ -433,14 +433,20 @@ void BSONEachRowRowInputFormat::readTuple(IColumn & column, const DataTypePtr &
 
     assertChar(BSON_DOCUMENT_END, *in);
 
-    if (read_nested_columns != data_type_tuple->getElements().size())
+    const auto elements_size = data_type_tuple->getElements().size();
+    if (read_nested_columns != elements_size)
         throw Exception(
                         ErrorCodes::INCORRECT_DATA,
                         "Cannot parse tuple column with type {} from BSON array/embedded document field, "
                         "the number of fields in tuple and BSON document doesn't match: {} != {}",
                         data_type->getName(),
-                        data_type_tuple->getElements().size(),
+                        elements_size,
                         read_nested_columns);
+
+    /// There are no nested columns to grow, so we must explicitly increment the column size.
+    /// Otherwise, `column.size()` will return 0 for empty tuples columns.
+    if (elements_size == 0)
+        tuple_column.addSize(1);
 }
 
 void BSONEachRowRowInputFormat::readMap(IColumn & column, const DataTypePtr & data_type, BSONType bson_type)

diff --git a/src/Processors/Formats/Impl/MsgPackRowInputFormat.cpp b/src/Processors/Formats/Impl/MsgPackRowInputFormat.cpp
@@ -407,8 +407,13 @@ bool MsgPackVisitor::start_array(size_t size) // NOLINT
 
         ColumnTuple & column_tuple = assert_cast<ColumnTuple &>(info_stack.top().column);
         /// Push nested columns into stack in reverse order.
-        for (ssize_t i = nested_types.size() - 1; i >= 0; --i)
+        for (ssize_t i = static_cast<ssize_t>(nested_types.size()) - 1; i >= 0; --i)
             info_stack.push(Info{column_tuple.getColumn(i), nested_types[i], true, std::nullopt, nullptr});
+
+        /// There are no nested columns to grow, so we must explicitly increment the column size.
+        /// Otherwise, `column.size()` will return 0 for empty tuples columns.
+        if (nested_types.empty())
+            column_tuple.addSize(1);
     }
     else
     {

diff --git a/tests/queries/0_stateless/03277_empty_tuple_formats.reference b/tests/queries/0_stateless/03277_empty_tuple_formats.reference
@@ -32,7 +32,9 @@ RowBinary
 Values
 ()
 BSONEachRow
+()
 MsgPack
+()
 Native
 ()
 TSV

diff --git a/tests/queries/0_stateless/03716_bson_each_row_empty_tuple_column.reference b/tests/queries/0_stateless/03716_bson_each_row_empty_tuple_column.reference
@@ -0,0 +1,5 @@
+1	()
+1	()
+1	()
+1	()
+1	()
diff --git a/tests/queries/0_stateless/03716_bson_each_row_empty_tuple_column.sql b/tests/queries/0_stateless/03716_bson_each_row_empty_tuple_column.sql
@@ -0,0 +1,17 @@
+DROP TABLE IF EXISTS t0;
+DROP TABLE IF EXISTS random_filename;
+
+CREATE TABLE t0 (c0 Int32, c1 Tuple()) ENGINE = Memory;
+CREATE TABLE random_filename (name String) ENGINE = Memory;
+
+INSERT INTO random_filename SELECT concat('03716_test_bson_empty_tuple_', toString(generateUUIDv4()), '.bson');
+
+INSERT INTO FUNCTION file((SELECT name FROM random_filename LIMIT 1), 'BSONEachRow', 'c0 Int32, c1 Tuple()')
+SELECT 1, tuple() FROM numbers(5) SETTINGS engine_file_truncate_on_insert = 1;
+
+INSERT INTO t0 SELECT * FROM file((SELECT name FROM random_filename LIMIT 1), 'BSONEachRow', 'c0 Int32, c1 Tuple()');
+
+SELECT * FROM t0 ORDER BY c0;
+
+DROP TABLE t0;
+DROP TABLE random_filename;
diff --git a/tests/queries/0_stateless/03717_msgpack_empty_tuple_column.reference b/tests/queries/0_stateless/03717_msgpack_empty_tuple_column.reference
@@ -0,0 +1,5 @@
+1	()
+1	()
+1	()
+1	()
+1	()
diff --git a/tests/queries/0_stateless/03717_msgpack_empty_tuple_column.sql b/tests/queries/0_stateless/03717_msgpack_empty_tuple_column.sql
@@ -0,0 +1,20 @@
+-- Tags: no-fasttest
+-- no-fasttest: 'MsgPack` format is not supported
+
+DROP TABLE IF EXISTS t0;
+DROP TABLE IF EXISTS random_filename;
+
+CREATE TABLE t0 (c0 Int32, c1 Tuple()) ENGINE = Memory;
+CREATE TABLE random_filename (name String) ENGINE = Memory;
+
+INSERT INTO random_filename SELECT concat('03716_test_msgpack_empty_tuple_', toString(generateUUIDv4()), '.msgpack');
+
+INSERT INTO FUNCTION file((SELECT name FROM random_filename LIMIT 1), 'MsgPack', 'c0 Int32, c1 Tuple()')
+SELECT 1, tuple() FROM numbers(5) SETTINGS engine_file_truncate_on_insert = 1;
+
+INSERT INTO t0 SELECT * FROM file((SELECT name FROM random_filename LIMIT 1), 'MsgPack', 'c0 Int32, c1 Tuple()');
+
+SELECT * FROM t0 ORDER BY c0;
+
+DROP TABLE t0;
+DROP TABLE random_filename;
diff --git a/tests/queries/0_stateless/03719_multiple_column_empty_tuple_formats.reference b/tests/queries/0_stateless/03719_multiple_column_empty_tuple_formats.reference
@@ -0,0 +1,198 @@
+Native
+0	()
+1	()
+2	()
+3	()
+4	()
+TSV
+0	()
+1	()
+2	()
+3	()
+4	()
+CSV
+0	\N
+1	\N
+2	\N
+3	\N
+4	\N
+TSKV
+0	()
+1	()
+2	()
+3	()
+4	()
+JSON
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONCompact
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONEachRow
+0	[]
+1	[]
+2	[]
+3	[]
+4	[]
+JSONObjectEachRow
+0	[]
+1	[]
+2	[]
+3	[]
+4	[]
+JSONCompactEachRow
+0	[]
+1	[]
+2	[]
+3	[]
+4	[]
+JSONColumns
+0	[]
+1	[]
+2	[]
+3	[]
+4	[]
+JSONCompactColumns
+0	[]
+1	[]
+2	[]
+3	[]
+4	[]
+JSONColumnsWithMetadata
+0	()
+1	()
+2	()
+3	()
+4	()
+ORC
+0	()
+1	()
+2	()
+3	()
+4	()
+Arrow
+0	()
+1	()
+2	()
+3	()
+4	()
+Avro
+0	()
+1	()
+2	()
+3	()
+4	()
+RowBinary
+0	()
+1	()
+2	()
+3	()
+4	()
+Values
+0	()
+1	()
+2	()
+3	()
+4	()
+BSONEachRow
+0	()
+1	()
+2	()
+3	()
+4	()
+MsgPack
+0	()
+1	()
+2	()
+3	()
+4	()
+Native
+0	()
+1	()
+2	()
+3	()
+4	()
+TSV
+0	()
+1	()
+2	()
+3	()
+4	()
+CSV
+0	()
+1	()
+2	()
+3	()
+4	()
+TSKV
+0	()
+1	()
+2	()
+3	()
+4	()
+JSON
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONCompact
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONEachRow
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONObjectEachRow
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONCompactEachRow
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONColumns
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONCompactColumns
+0	()
+1	()
+2	()
+3	()
+4	()
+JSONColumnsWithMetadata
+0	()
+1	()
+2	()
+3	()
+4	()
+ORC
+0	()
+1	()
+2	()
+3	()
+4	()
+Arrow
+0	()
+1	()
+2	()
+3	()
+4	()
diff --git a/tests/queries/0_stateless/03719_multiple_column_empty_tuple_formats.sh b/tests/queries/0_stateless/03719_multiple_column_empty_tuple_formats.sh
@@ -0,0 +1,47 @@
+#!/usr/bin/env bash
+# Tags: no-fasttest
+
+CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CUR_DIR"/../shell_config.sh
+
+
+# Test that no formats crash or return LOGICAL_ERROR on empty tuple
+# This test is on the same spirit as 03277_empty_tuple_formats.sh except it has additional column in addition to empty tuple.
+# For example this logical error: Invalid number of rows in Chunk  Int32(size = 5) Tuple(size = 0) column Tuple() at position 1: expected 5, got 0
+# was triggered when there was another column in addition to the empty tuple. This test covers that case.
+
+FILE=03277_$CLICKHOUSE_DATABASE
+
+# With schema inference.
+for format in Native TSV CSV TSKV JSON JSONCompact JSONEachRow JSONObjectEachRow JSONCompactEachRow JSONColumns JSONCompactColumns JSONColumnsWithMetadata ORC Arrow
+do
+  echo $format
+  $CLICKHOUSE_LOCAL -q "
+    insert into function file('$FILE', '$format') select number, () from numbers(5) settings engine_file_truncate_on_insert=1;
+    select * from file('$FILE', '$format');"
+done
+
+# Picky about column names.
+echo Avro
+$CLICKHOUSE_LOCAL -q "
+  insert into function file('$FILE', 'Avro') select number as x, () as y from numbers(5) settings engine_file_truncate_on_insert=1;
+  select * from file('$FILE', 'Avro');"
+
+# Without schema inference.
+for format in RowBinary Values BSONEachRow MsgPack Native TSV CSV TSKV JSON JSONCompact JSONEachRow JSONObjectEachRow JSONCompactEachRow JSONColumns JSONCompactColumns JSONColumnsWithMetadata ORC Arrow
+do
+  echo $format
+  $CLICKHOUSE_LOCAL -q "
+    insert into function file('$FILE', '$format', 'x UInt64, y Tuple()') select number as x, () as y from numbers(5) settings engine_file_truncate_on_insert=1;
+    select * from file('$FILE', '$format', 'x UInt64, y Tuple()');"
+done
+
+# Formats that don't support empty tuples/multiple columns.
+$CLICKHOUSE_LOCAL -q "
+  insert into function file('$FILE', 'Parquet') select number, () from numbers(5) settings engine_file_truncate_on_insert=1; -- {serverError BAD_ARGUMENTS}
+  insert into function file('$FILE', 'Npy') select number, () from numbers(5) settings engine_file_truncate_on_insert=1; -- {serverError TOO_MANY_COLUMNS}
+  insert into function file('$FILE', 'CapnProto', 'x UInt64, y Tuple()') select number as x, () as y from numbers(5) settings engine_file_truncate_on_insert=1; -- {serverError CAPN_PROTO_BAD_CAST}
+  insert into function file('$FILE', 'RawBLOB') select number, () from numbers(5) settings engine_file_truncate_on_insert=1; -- {serverError NOT_IMPLEMENTED}"
+
+rm $FILE
-Original file line number
+Diff line change
@@ Expand Up / @@ -32,7 +32,9 @@ RowBinary @@
     Values
     ()
     BSONEachRow
+    ()
     MsgPack
+    ()
     Native
     ()
     TSV
@@ Expand Down @@