Use one-shot UTF-8 encoding when possible.

This avoids allocating a EncoderNLS object and making multiple passes over the source data when the entire string could be converted at once. Additionally, UTF8Encoding.GetBytes is more likely to be highly optimized in the framework.
mysql-net · Aug 13, 2023 · 461def2 · 461def2
1 parent 845461c
commit 461def2
Show file tree

Hide file tree

Showing 2 changed files with 59 additions and 15 deletions.
diff --git a/src/MySqlConnector/MySqlParameter.cs b/src/MySqlConnector/MySqlParameter.cs
@@ -197,15 +197,15 @@ internal void AppendSqlString(ByteBufferWriter writer, StatementPreparerOptions
 		}
 		else if (Value is string stringValue)
 		{
-			WriteString(writer, noBackslashEscapes, writeDelimiters: true, stringValue.AsSpan());
+			WriteString(writer, noBackslashEscapes, stringValue.AsSpan());
 		}
 		else if (Value is ReadOnlyMemory<char> readOnlyMemoryChar)
 		{
-			WriteString(writer, noBackslashEscapes, writeDelimiters: true, readOnlyMemoryChar.Span);
+			WriteString(writer, noBackslashEscapes, readOnlyMemoryChar.Span);
 		}
 		else if (Value is Memory<char> memoryChar)
 		{
-			WriteString(writer, noBackslashEscapes, writeDelimiters: true, memoryChar.Span);
+			WriteString(writer, noBackslashEscapes, memoryChar.Span);
 		}
 		else if (Value is char charValue)
 		{
@@ -447,12 +447,12 @@ internal void AppendSqlString(ByteBufferWriter writer, StatementPreparerOptions
 #if NETCOREAPP3_1_OR_GREATER
 			writer.Write((byte) '\'');
 			foreach (var chunk in stringBuilder.GetChunks())
-				WriteString(writer, noBackslashEscapes, writeDelimiters: false, chunk.Span);
+				WriteStringChunk(writer, noBackslashEscapes, chunk.Span);
 			if (stringBuilder.Length != 0)
 				writer.Write("".AsSpan(), flush: true);
 			writer.Write((byte) '\'');
 #else
-			WriteString(writer, noBackslashEscapes, writeDelimiters: true, stringBuilder.ToString().AsSpan());
+			WriteString(writer, noBackslashEscapes, stringBuilder.ToString().AsSpan());
 #endif
 		}
 		else if (MySqlDbType == MySqlDbType.Int16)
@@ -494,10 +494,9 @@ internal void AppendSqlString(ByteBufferWriter writer, StatementPreparerOptions
 			throw new NotSupportedException($"Parameter type {Value.GetType().Name} is not supported; see https://fl.vu/mysql-param-type. Value: {Value}");
 		}
 
-		static void WriteString(ByteBufferWriter writer, bool noBackslashEscapes, bool writeDelimiters, ReadOnlySpan<char> value)
+		static void WriteString(ByteBufferWriter writer, bool noBackslashEscapes, ReadOnlySpan<char> value)
 		{
-			if (writeDelimiters)
-				writer.Write((byte) '\'');
+			writer.Write((byte) '\'');
 
 			var charsWritten = 0;
 			while (charsWritten < value.Length)
@@ -507,13 +506,13 @@ static void WriteString(ByteBufferWriter writer, bool noBackslashEscapes, bool w
 				if (nextDelimiterIndex == -1)
 				{
 					// write the rest of the string
-					writer.Write(remainingValue, flush: writeDelimiters);
+					writer.Write(remainingValue);
 					charsWritten += remainingValue.Length;
 				}
 				else
 				{
 					// write up to (and including) the delimiter, then double it
-					writer.Write(remainingValue[..nextDelimiterIndex], flush: true);
+					writer.Write(remainingValue[..nextDelimiterIndex]);
 					if (remainingValue[nextDelimiterIndex] == '\\' && !noBackslashEscapes)
 						writer.Write((ushort) 0x5C5C); // \\
 					else if (remainingValue[nextDelimiterIndex] == '\\' && noBackslashEscapes)
@@ -528,9 +527,43 @@ static void WriteString(ByteBufferWriter writer, bool noBackslashEscapes, bool w
 				}
 			}
 
-			if (writeDelimiters)
-				writer.Write((byte) '\'');
+			writer.Write((byte) '\'');
 		}
+
+#if NETCOREAPP3_1_OR_GREATER
+		// Writes a partial chunk of a string (that may end with half of a surrogate pair), escaping any delimiter characters.
+		static void WriteStringChunk(ByteBufferWriter writer, bool noBackslashEscapes, ReadOnlySpan<char> value)
+		{
+			var charsWritten = 0;
+			while (charsWritten < value.Length)
+			{
+				var remainingValue = value[charsWritten..];
+				var nextDelimiterIndex = remainingValue.IndexOfAny('\0', '\'', '\\');
+				if (nextDelimiterIndex == -1)
+				{
+					// write the rest of the string
+					writer.Write(remainingValue, flush: false);
+					charsWritten += remainingValue.Length;
+				}
+				else
+				{
+					// write up to (and including) the delimiter, then double it
+					writer.Write(remainingValue[..nextDelimiterIndex], flush: true);
+					if (remainingValue[nextDelimiterIndex] == '\\' && !noBackslashEscapes)
+						writer.Write((ushort) 0x5C5C); // \\
+					else if (remainingValue[nextDelimiterIndex] == '\\' && noBackslashEscapes)
+						writer.Write((byte) 0x5C); // \
+					else if (remainingValue[nextDelimiterIndex] == '\'')
+						writer.Write((ushort) 0x2727); // ''
+					else if (remainingValue[nextDelimiterIndex] == '\0' && !noBackslashEscapes)
+						writer.Write((ushort) 0x305C); // \0
+					else if (remainingValue[nextDelimiterIndex] == '\0' && noBackslashEscapes)
+						writer.Write((byte) 0x00); // (nul)
+					charsWritten += nextDelimiterIndex + 1;
+				}
+			}
+		}
+#endif
 	}
 
 	internal void AppendBinary(ByteBufferWriter writer, StatementPreparerOptions options)

diff --git a/src/MySqlConnector/Protocol/Serialization/ByteBufferWriter.cs b/src/MySqlConnector/Protocol/Serialization/ByteBufferWriter.cs
@@ -99,11 +99,22 @@ public void Write(ReadOnlySpan<byte> span)
 		m_output = m_output[span.Length..];
 	}
 
-	public void Write(string value) => Write(value.AsSpan(), flush: true);
+	public void Write(string value) => Write(value.AsSpan());
 
 	public void WriteAscii(string value) => WriteAscii(value.AsSpan());
 
-	public void Write(string value, int offset, int length) => Write(value.AsSpan(offset, length), flush: true);
+	public void Write(string value, int offset, int length) => Write(value.AsSpan(offset, length));
+
+	public void Write(ReadOnlySpan<char> chars)
+	{
+		if (m_output.Length < chars.Length * 3)
+		{
+			var neededBytes = Encoding.UTF8.GetByteCount(chars);
+			if (m_output.Length < neededBytes)
+				Reallocate(neededBytes);
+		}
+		m_output = m_output[Encoding.UTF8.GetBytes(chars, m_output.Span)..];
+	}
 
 	public void Write(ReadOnlySpan<char> chars, bool flush)
 	{
@@ -273,7 +284,7 @@ public static void WriteLengthEncodedString(this ByteBufferWriter writer, ReadOn
 	{
 		var byteCount = Encoding.UTF8.GetByteCount(value);
 		writer.WriteLengthEncodedInteger((ulong) byteCount);
-		writer.Write(value, flush: true);
+		writer.Write(value);
 	}
 
 	public static void WriteLengthEncodedAsciiString(this ByteBufferWriter writer, string value)