[API Proposal]: MemoryExtensions.TryWrite for UTF8

[stephentoub]

Background and motivation

In .NET 6, as part of rolling out the improved support for string interpolation, we added some TryWrite extension methods to MemoryExtensions that enable interpolating directly into a Span<char>, e.g.

Span<char> span = ...;
bool wrote = span.TryWrite($"The current day/time is {DateTime.Now}", out int charsWritten);

We should consider enabling the same capability for writing UTF8 into a Span<byte>.

API Proposal

namespace System
{
    public static class MemoryExtensions
    {
+        public static bool TryWrite(this Span<byte> utf8Destination, [InterpolatedStringHandlerArgument("destination")] ref TryWriteUtf8InterpolatedStringHandler handler, out int bytesWritten);
+        public static bool TryWrite(this Span<byte> utf8Destination, IFormatProvider? provider, [InterpolatedStringHandlerArgument("destination", "provider")] ref TryWriteUtf8InterpolatedStringHandler handler, out int bytesWritten);

+        [EditorBrowsable(EditorBrowsableState.Never)]
+        [InterpolatedStringHandlerAttribute]
+        public ref struct TryWriteUtf8InterpolatedStringHandler
+        {
+            public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> utf8Destination, out bool shouldAppend);
+            public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> utf8Destination, IFormatProvider? provider, out bool shouldAppend);

+            public bool AppendLiteral(string value);
+            // public bool AppendLiteral(ReadOnlySpan<byte> value); // if the C# compiler supports interpolation with u8 literals

+            public bool AppendFormatted(scoped ReadOnlySpan<char> value);
+            public bool AppendFormatted(scoped ReadOnlySpan<char> value, int alignment = 0, string? format = null);

+            public bool AppendFormatted(scoped ReadOnlySpan<byte> utf8Value);
+            public bool AppendFormatted(scoped ReadOnlySpan<byte> utf8Value, int alignment = 0, string? format = null);

+            public bool AppendFormatted<T>(T value);
+            public bool AppendFormatted<T>(T value, string? format);
+            public bool AppendFormatted<T>(T value, int alignment);
+            public bool AppendFormatted<T>(T value, int alignment, string? format);

+            public bool AppendFormatted(object? value, int alignment = 0, string? format = null);

+            public bool AppendFormatted(string? value);
+            public bool AppendFormatted(string? value, int alignment = 0, string? format = null);
        }
}

This is essentially the exact same design as for MemoryExtensions.TryWrite, with a few specific differences:

• The TryWrite methods take a Span<byte> instead of a Span<char>.
• In addition to ReadOnlySpan<char>-based AppendFormatted methods, there are also ReadOnlySpan<byte>-based AppendFormatted methods, the latter of which is expected to be UTF8 data, like a u8 literal. The ReadOnlySpan<byte> data would be memcpy'd into the destination, whereas the ReadOnlySpan<char> data would be Encoding.UTF8-encoded into the destination.
• The implementation of Append<T> will check for IUtf8SpanFormattable before it checks for ISpanFormattable, prefering to use a type's built-in UTF8 formatting support if supplied.

Also note that the AppendLiteral method takes a string value, as that's what's supported by the C# language. If C# were to ever support u8 literals in string interpolation, we could add an appropriate AppendLiteral(ReadOnlySpan<byte> utf8Value) overload.

API Usage

Span<byte> utf8 = ...;
bool wrote = utf8.TryWrite($"The current day/time is {DateTime.Now}", out int bytesWritten);

Alternative Designs

No response

Risks

No response

Tagging subscribers to this area: @dotnet/area-system-buffers
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Background and motivation

In .NET 6, as part of rolling out the improved support for string interpolation, we added some TryWrite extension methods to MemoryExtensions that enable interpolating directly into a Span<char>, e.g.

Span<char> span = ...;
bool wrote = span.TryWrite($"The current day/time is {DateTime.Now}", out int charsWritten);

We should consider enabling the same capability for writing UTF8 into a Span<byte>.

API Proposal

namespace System.Buffers.Text
{
    public static partial class Utf8Formatter
    {
        public static bool TryWrite(this Span<byte> destination, [InterpolatedStringHandlerArgument("destination")] ref TryWriteInterpolatedStringHandler handler, out int bytesWritten);
        public static bool TryWrite(this Span<byte> destination, IFormatProvider? provider, [InterpolatedStringHandlerArgument("destination", "provider")] ref TryWriteInterpolatedStringHandler handler, out int bytesWritten);

        [EditorBrowsable(EditorBrowsableState.Never)]
        [InterpolatedStringHandlerAttribute]
        public ref struct TryWriteInterpolatedStringHandler
        {
            public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> destination, out bool shouldAppend);
            public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> destination, IFormatProvider? provider, out bool shouldAppend);

            public bool AppendLiteral(string value);

            public bool AppendFormatted(scoped ReadOnlySpan<char> value);
            public bool AppendFormatted(scoped ReadOnlySpan<char> value, int alignment = 0, string? format = null);

            public bool AppendFormatted(scoped ReadOnlySpan<byte> value);
            public bool AppendFormatted(scoped ReadOnlySpan<byte> value, int alignment = 0, string? format = null);

            public bool AppendFormatted<T>(T value);
            public bool AppendFormatted<T>(T value, string? format);
            public bool AppendFormatted<T>(T value, int alignment);
            public bool AppendFormatted<T>(T value, int alignment, string? format);

            public bool AppendFormatted(object? value, int alignment = 0, string? format = null);

            public bool AppendFormatted(string? value) { throw null; }
            public bool AppendFormatted(string? value, int alignment = 0, string? format = null);
        }
}

This is essentially the exact same design as for MemoryExtensions.TryWrite, with a few specific differences:

• The TryWrite methods take a Span<byte> instead of a Span<char>.
• In addition to ReadOnlySpan<char>-based AppendFormatted methods, there are also ReadOnlySpan<byte>-based AppendFormatted methods, the latter of which is expected to be UTF8 data, like a u8 literal. The ReadOnlySpan<byte> data would be memcpy'd into the destination, whereas the ReadOnlySpan<char> data would be Encoding.UTF8-encoded into the destination.

Also note that the AppendLiteral method takes a string value, as that's what's supported by the C# language. If C# were to ever support u8 literals in string interpolation, we could add an appropriate AppendLiteral(ReadOnlySpan<byte> value) overload.

API Usage

Span<byte> span = ...;
bool wrote = span.TryWrite($"The current day/time is {DateTime.Now}", out int bytesWritten);

Alternative Designs

No response

Risks

No response

Author:	stephentoub
Assignees:	-
Labels:	api-suggestion, area-System.Buffers
Milestone:	-

[stephentoub]

cc: @333fred for string interpolation
cc: @tannergooding for generic math interfaces, since we discussed possible future UTF8 support for something ISpanFormattable-like

[Zintom]

Love it!

[333fred]

@stephentoub was there a specific thing you wanted me to comment on, or just awareness?

[stephentoub]

Just awareness.

[stephentoub]

Prototype:
main...stephentoub:runtime:memextutf8trywrite

[terrajobst]

Video

• We should rename the method to make it clear that the buffer receives the UTF8 encoding, not the UTF16 encoding
• We're OK with supporting both UTF8 and UTF16, with UTF16 potentially generating a warning, telling the user that should use a u8 literal to avoid the transcoding
• Will C# support interpolating strings with the u8 suffix? Seems like it should? @dotnet/ldm
• We should consider making it work that one removes the $ that the code still compiles, but this would apply for the existing UTF16 version as well

namespace System;

public static partial class MemoryExtensions
{
    public static bool TryWriteUtf8(this Span<byte> destination, [InterpolatedStringHandlerArgument("destination")] ref TryWriteUtf8InterpolatedStringHandler handler, out int bytesWritten);
    public static bool TryWriteUtf8(this Span<byte> destination, IFormatProvider? provider, [InterpolatedStringHandlerArgument("destination", "provider")] ref TryWriteUtf8InterpolatedStringHandler handler, out int bytesWritten);

    [EditorBrowsable(EditorBrowsableState.Never)]
    [InterpolatedStringHandlerAttribute]
    public ref struct TryWriteUtf8InterpolatedStringHandler
    {
        public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> utf8Destination, out bool shouldAppend);
        public TryWriteInterpolatedStringHandler(int literalLength, int formattedCount, Span<byte> utf8Destination, IFormatProvider? provider, out bool shouldAppend);

        public bool AppendLiteral(string value);
        // public bool AppendLiteral(ReadOnlySpan<byte> value); // if the C# compiler supports interpolation with u8 literals

        public bool AppendFormatted(scoped ReadOnlySpan<char> value);
        public bool AppendFormatted(scoped ReadOnlySpan<char> value, int alignment = 0, string? format = null);

        public bool AppendFormatted(scoped ReadOnlySpan<byte> utf8Value);
        public bool AppendFormatted(scoped ReadOnlySpan<byte> utf8Value, int alignment = 0, string? format = null);

        public bool AppendFormatted<T>(T value);
        public bool AppendFormatted<T>(T value, string? format);
        public bool AppendFormatted<T>(T value, int alignment);
        public bool AppendFormatted<T>(T value, int alignment, string? format);

        public bool AppendFormatted(object? value, int alignment = 0, string? format = null);

        public bool AppendFormatted(string? value);
        public bool AppendFormatted(string? value, int alignment = 0, string? format = null);
    }
}