Merge pull request #33021 from DevTeamBK/Issue135

Fix null pointer dereference in low cardinality data

src/DataTypes/Serializations/SerializationLowCardinality.cpp

@@ -636,6 +636,9 @@ void SerializationLowCardinality::deserializeBinaryBulkWithMultipleStreams(
 
         if (!low_cardinality_state->index_type.need_global_dictionary)
         {
+            if(additional_keys == nullptr)
+                throw Exception("No additional keys found.", ErrorCodes::INCORRECT_DATA);
+
             ColumnPtr keys_column = additional_keys;
             if (low_cardinality_state->null_map)
                 keys_column = ColumnNullable::create(additional_keys, low_cardinality_state->null_map);
@@ -662,6 +665,9 @@ void SerializationLowCardinality::deserializeBinaryBulkWithMultipleStreams(
 
             if (!maps.additional_keys_map->empty())
             {
+                if(additional_keys == nullptr)
+                    throw Exception("No additional keys found.", ErrorCodes::INCORRECT_DATA);
+
                 auto used_add_keys = additional_keys->index(*maps.additional_keys_map, 0);
 
                 if (dictionary_type->isNullable())

tests/queries/0_stateless/02010_lc_native.python

@@ -302,11 +302,44 @@ def insertLowCardinalityRowWithIncorrectDictType():
         print(readException(s))
         s.close()
 
+def insertLowCardinalityRowWithIncorrectAdditionalKeys():
+    with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+        s.settimeout(30)
+        s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT))
+        sendHello(s)
+        receiveHello(s)
+        sendQuery(s, 'insert into {}.tab format TSV'.format(CLICKHOUSE_DATABASE))
+
+        # external tables
+        sendEmptyBlock(s)
+        readHeader(s)
+
+        # Data
+        ba = bytearray()
+        writeVarUInt(2, ba) # Data
+        writeStringBinary('', ba)
+        serializeBlockInfo(ba)
+        writeVarUInt(1, ba) # rows
+        writeVarUInt(1, ba) # columns
+        writeStringBinary('x', ba)
+        writeStringBinary('LowCardinality(String)', ba)
+        ba.extend([1] + [0] * 7) # SharedDictionariesWithAdditionalKeys
+        ba.extend([3, 0] + [0] * 6) # indexes type: UInt64 [3], with  NO additional keys [0]
+        ba.extend([1] + [0] * 7) # num_keys in dict
+        writeStringBinary('hello', ba) # key
+        ba.extend([1] + [0] * 7) # num_indexes
+        ba.extend([0] * 8) # UInt64 index (0 for 'hello')
+        s.sendall(ba)
+
+        assertPacket(readVarUInt(s), 2)
+        print(readException(s))
+        s.close()
 
 def main():
     insertValidLowCardinalityRow()
     insertLowCardinalityRowWithIndexOverflow()
     insertLowCardinalityRowWithIncorrectDictType()
+    insertLowCardinalityRowWithIncorrectAdditionalKeys()
 
 if __name__ == "__main__":
     main()

tests/queries/0_stateless/02010_lc_native.reference

@@ -6,3 +6,6 @@ code 117:  Index for LowCardinality is out of range. Dictionary size is 1, but f
 Rows 0 Columns 1
 Column x type LowCardinality(String)
 code 117:  LowCardinality indexes serialization type for Native format cannot use global dictionary
+Rows 0 Columns 1
+Column x type LowCardinality(String)
+code 117:  No additional keys found.