Fix StringBuilder MySqlParameter support for .NET 5.0. Fixes #977

- Use StringBuilder.GetChunks for binary parameters
- Fix UTF-8 encoding for surrogate code units split across chunks

Author: bgrainger

docs/content/tutorials/migrating-from-connector-net.md

@@ -282,3 +282,5 @@ The following bugs in Connector/NET are fixed by switching to MySqlConnector. (~
 * ~~[#101714](https://bugs.mysql.com/bug.php?id=101714): Extremely slow performance reading result sets~~
 * [#102593](https://bugs.mysql.com/bug.php?id=102593): Can't use `MemoryStream` as `MySqlParameter.Value`
 * [#103390](https://bugs.mysql.com/bug.php?id=103390): Can't query `CHAR(36)` column if `MySqlCommand` is prepared
+* [#103801](https://bugs.mysql.com/bug.php?id=103801): `TimeSpan` parameters lose microseconds with prepared statement
+* [#103819](https://bugs.mysql.com/bug.php?id=103819): Can't use `StringBuilder` containing non-BMP characters as `MySqlParameter.Value`

src/MySqlConnector/MySqlParameter.cs

@@ -413,6 +413,8 @@ internal void AppendSqlString(ByteBufferWriter writer, StatementPreparerOptions
 				writer.Write((byte) '\'');
 				foreach (var chunk in stringBuilder.GetChunks())
 					WriteString(writer, noBackslashEscapes, writeDelimiters: false, chunk.Span);
+				if (stringBuilder.Length != 0)
+					writer.Write("".AsSpan(), flush: true);
 				writer.Write((byte) '\'');
 #elif NET45 || NETSTANDARD1_3
 				WriteString(writer, noBackslashEscapes, stringBuilder.ToString());
@@ -483,13 +485,13 @@ static void WriteString(ByteBufferWriter writer, bool noBackslashEscapes, bool w
 					if (nextDelimiterIndex == -1)
 					{
 						// write the rest of the string
-						writer.Write(remainingValue);
+						writer.Write(remainingValue, flush: writeDelimiters);
 						charsWritten += remainingValue.Length;
 					}
 					else
 					{
 						// write up to (and including) the delimiter, then double it
-						writer.Write(remainingValue.Slice(0, nextDelimiterIndex + 1));
+						writer.Write(remainingValue.Slice(0, nextDelimiterIndex + 1), flush: true);
 						if (remainingValue[nextDelimiterIndex] == '\\' && !noBackslashEscapes)
 							writer.Write((byte) '\\');
 						else if (remainingValue[nextDelimiterIndex] == '\'')
@@ -671,7 +673,7 @@ internal void AppendBinary(ByteBufferWriter writer, StatementPreparerOptions opt
 #endif
 			else if (Value is StringBuilder stringBuilder)
 			{
-				writer.WriteLengthEncodedString(stringBuilder.ToString());
+				writer.WriteLengthEncodedString(stringBuilder);
 			}
 			else if (MySqlDbType == MySqlDbType.Int16)
 			{

src/MySqlConnector/Protocol/Serialization/ByteBufferWriter.cs

@@ -130,26 +130,76 @@ public unsafe void Write(string value, int offset, int length)
 			}
 		}
 #else
-		public void Write(string value) => Write(value.AsSpan());
-		public void Write(string value, int offset, int length) => Write(value.AsSpan(offset, length));
+		public void Write(string value) => Write(value.AsSpan(), flush: true);
+		public void Write(string value, int offset, int length) => Write(value.AsSpan(offset, length), flush: true);
 
-		public void Write(ReadOnlySpan<char> chars)
+		public void Write(ReadOnlySpan<char> chars, bool flush)
 		{
 			m_encoder ??= Encoding.UTF8.GetEncoder();
 			while (chars.Length > 0)
 			{
 				if (m_output.Length < 4)
 					Reallocate();
-				m_encoder.Convert(chars, m_output.Span, true, out var charsUsed, out var bytesUsed, out var completed);
+				m_encoder.Convert(chars, m_output.Span, flush: false, out var charsUsed, out var bytesUsed, out var completed);
 				chars = chars.Slice(charsUsed);
 				m_output = m_output.Slice(bytesUsed);
 				if (!completed)
 					Reallocate();
 				Debug.Assert(completed == (chars.Length == 0));
 			}
+
+			if (flush && m_encoder is not null)
+			{
+				if (m_output.Length < 4)
+					Reallocate();
+				m_encoder.Convert("".AsSpan(), m_output.Span, flush: true, out _, out var bytesUsed, out _);
+				m_output = m_output.Slice(bytesUsed);
+			}
 		}
 #endif
 
+		public void WriteLengthEncodedString(StringBuilder stringBuilder)
+		{
+#if !NET45 && !NET461 && !NET471 && !NETSTANDARD1_3 && !NETSTANDARD2_0 && !NETSTANDARD2_1 && !NETCOREAPP2_1
+			// save where the length will be written
+			var lengthPosition = Position;
+			if (m_output.Length < 9)
+				Reallocate(9);
+			Advance(9);
+
+			// write all the text as UTF-8
+			m_encoder ??= Encoding.UTF8.GetEncoder();
+			foreach (var chunk in stringBuilder.GetChunks())
+			{
+				var currentSpan = chunk.Span;
+				while (currentSpan.Length > 0)
+				{
+					if (m_output.Length < 4)
+						Reallocate();
+					m_encoder.Convert(currentSpan, m_output.Span, false, out var charsUsed, out var bytesUsed, out var completed);
+					currentSpan = currentSpan.Slice(charsUsed);
+					m_output = m_output.Slice(bytesUsed);
+					if (!completed)
+						Reallocate();
+					Debug.Assert(completed == (currentSpan.Length == 0));
+				}
+			}
+
+			// flush the output
+			if (m_output.Length < 4)
+				Reallocate();
+			m_encoder.Convert("".AsSpan(), m_output.Span, true, out _, out var finalBytesUsed, out _);
+			m_output = m_output.Slice(finalBytesUsed);
+
+			// write the length (as a 64-bit integer) in the reserved space
+			var textLength = Position - (lengthPosition + 9);
+			m_buffer[lengthPosition] = 0xFE;
+			BinaryPrimitives.WriteUInt64LittleEndian(m_buffer.AsSpan(lengthPosition + 1), (ulong) textLength);
+#else
+			this.WriteLengthEncodedString(stringBuilder.ToString());
+#endif
+		}
+
 		public void WriteString(short value)
 		{
 			int bytesWritten;
@@ -255,7 +305,7 @@ public static void WriteLengthEncodedString(this ByteBufferWriter writer, ReadOn
 		{
 			var byteCount = Encoding.UTF8.GetByteCount(value);
 			writer.WriteLengthEncodedInteger((ulong) byteCount);
-			writer.Write(value);
+			writer.Write(value, flush: true);
 		}
 #endif
 

tests/SideBySide/InsertTests.cs

@@ -249,27 +249,38 @@ public void InsertMemoryStream(bool prepare)
 			Assert.Equal(new byte[] { 97, 98, 99, 100 }, reader.GetValue(1));
 		}
 
-		[Theory]
+		[SkippableTheory(Baseline = "https://bugs.mysql.com/bug.php?id=103819")]
 		[InlineData(false)]
 		[InlineData(true)]
 		public void InsertStringBuilder(bool prepare)
 		{
 			using var connection = new MySqlConnection(AppConfig.ConnectionString);
 			connection.Open();
 			connection.Execute(@"drop table if exists insert_string_builder;
-create table insert_string_builder(rowid integer not null primary key auto_increment, str text);");
+create table insert_string_builder(rowid integer not null primary key auto_increment, str text collate utf8mb4_bin);");
+
+			var value = new StringBuilder("\aAB\\12'ab\\'\\'");
+			for (var i = 0; i < 100; i++)
+#if !NETCOREAPP3_1
+				value.Append("\U0001F600\uD800\'\U0001F601\uD800");
+#else
+				// the netcoreapp3.1 implementation is broken when handling an unpaired surrogate; probably https://github.com/dotnet/runtime/issues/33817
+				value.Append("\U0001F600\U0001F601");
+#endif
 
-			var value = "\aAB\\12'ab\\'\\'";
 			using var cmd = connection.CreateCommand();
 			cmd.CommandText = @"insert into insert_string_builder(str) values(@str);";
-			cmd.Parameters.AddWithValue("@str", new StringBuilder(value));
+			cmd.Parameters.AddWithValue("@str", value);
 			if (prepare)
 				cmd.Prepare();
 			cmd.ExecuteNonQuery();
 
 			using var reader = connection.ExecuteReader(@"select str from insert_string_builder order by rowid;");
 			Assert.True(reader.Read());
-			Assert.Equal(value, reader.GetValue(0));
+
+			// all unpaired high-surrogates will be converted to the Unicode Replacement Character when converted to UTF-8 to be transmitted to the server
+			var expected = value.ToString().Replace('\uD800', '\uFFFD');
+			Assert.Equal(expected, reader.GetValue(0));
 		}
 
 		[Fact]