Added Encode extension to BufferWriterSlim type

dotnet · Dec 4, 2024 · b96205d · b96205d
1 parent 0aa4c56
commit b96205d
Show file tree

Hide file tree

Showing 3 changed files with 190 additions and 13 deletions.
diff --git a/src/DotNext.IO/Buffers/BufferWriter.cs b/src/DotNext.IO/Buffers/BufferWriter.cs
@@ -60,9 +60,9 @@ static int Write7BitEncodedInteger(ref SpanWriter<byte> writer, int value)
 
     internal static int WriteLength(this IBufferWriter<byte> buffer, int length, LengthFormat lengthFormat)
     {
-        var writer = new SpanWriter<byte>(buffer.GetSpan(SevenBitEncodedInt.MaxSizeInBytes));
-        buffer.Advance(writer.WriteLength(length, lengthFormat));
-        return writer.WrittenCount;
+        var bytesWritten = WriteLength(buffer.GetSpan(SevenBitEncodedInt.MaxSizeInBytes), length, lengthFormat);
+        buffer.Advance(bytesWritten);
+        return bytesWritten;
     }
 
     internal static int WriteLength(Span<byte> buffer, int length, LengthFormat lengthFormat)
@@ -71,21 +71,28 @@ internal static int WriteLength(Span<byte> buffer, int length, LengthFormat leng
         return writer.WriteLength(length, lengthFormat);
     }
 
+    private static int WriteLength(this ref BufferWriterSlim<byte> buffer, int length, LengthFormat lengthFormat)
+    {
+        var bytesWritten = WriteLength(buffer.GetSpan(SevenBitEncodedInt.MaxSizeInBytes), length, lengthFormat);
+        buffer.Advance(bytesWritten);
+        return bytesWritten;
+    }
+
     /// <summary>
     /// Encodes string using the specified encoding.
     /// </summary>
     /// <param name="writer">The buffer writer.</param>
-    /// <param name="value">The sequence of characters.</param>
+    /// <param name="chars">The sequence of characters.</param>
     /// <param name="context">The encoding context.</param>
     /// <param name="lengthFormat">String length encoding format; or <see langword="null"/> to prevent encoding of string length.</param>
     /// <returns>The number of written bytes.</returns>
-    public static long Encode(this IBufferWriter<byte> writer, ReadOnlySpan<char> value, in EncodingContext context, LengthFormat? lengthFormat = null)
+    public static long Encode(this IBufferWriter<byte> writer, ReadOnlySpan<char> chars, in EncodingContext context, LengthFormat? lengthFormat = null)
     {
         var result = lengthFormat.HasValue
-            ? writer.WriteLength(context.Encoding.GetByteCount(value), lengthFormat.GetValueOrDefault())
+            ? writer.WriteLength(context.Encoding.GetByteCount(chars), lengthFormat.GetValueOrDefault())
             : 0L;
 
-        context.GetEncoder().Convert(value, writer, true, out var bytesWritten, out _);
+        context.GetEncoder().Convert(chars, writer, true, out var bytesWritten, out _);
         result += bytesWritten;
 
         return result;
@@ -95,19 +102,19 @@ public static long Encode(this IBufferWriter<byte> writer, ReadOnlySpan<char> va
     /// Encodes string using the specified encoding.
     /// </summary>
     /// <param name="writer">The buffer writer.</param>
-    /// <param name="value">The sequence of characters.</param>
+    /// <param name="chars">The sequence of characters.</param>
     /// <param name="context">The encoding context.</param>
     /// <param name="lengthFormat">String length encoding format; or <see langword="null"/> to prevent encoding of string length.</param>
     /// <returns>The number of written bytes.</returns>
-    public static int Encode(this ref SpanWriter<byte> writer, ReadOnlySpan<char> value, in EncodingContext context, LengthFormat? lengthFormat = null)
+    public static int Encode(this ref SpanWriter<byte> writer, scoped ReadOnlySpan<char> chars, in EncodingContext context, LengthFormat? lengthFormat = null)
     {
         var result = lengthFormat.HasValue
-            ? writer.WriteLength(context.Encoding.GetByteCount(value), lengthFormat.GetValueOrDefault())
+            ? writer.WriteLength(context.Encoding.GetByteCount(chars), lengthFormat.GetValueOrDefault())
             : 0;
 
         var bytesWritten = context.TryGetEncoder() is { } encoder
-            ? encoder.GetBytes(value, writer.RemainingSpan, flush: true)
-            : context.Encoding.GetBytes(value, writer.RemainingSpan);
+            ? encoder.GetBytes(chars, writer.RemainingSpan, flush: true)
+            : context.Encoding.GetBytes(chars, writer.RemainingSpan);
         result += bytesWritten;
         writer.Advance(bytesWritten);
 
@@ -128,6 +135,66 @@ public static int Write(this ref SpanWriter<byte> writer, scoped ReadOnlySpan<by
 
         return result;
     }
+
+    /// <summary>
+    /// Encodes string using the specified encoding.
+    /// </summary>
+    /// <param name="writer">The buffer writer.</param>
+    /// <param name="chars">The sequence of characters.</param>
+    /// <param name="context">The encoding context.</param>
+    /// <param name="lengthFormat">String length encoding format; or <see langword="null"/> to prevent encoding of string length.</param>
+    /// <returns>The number of written bytes.</returns>
+    public static int Encode(this ref BufferWriterSlim<byte> writer, scoped ReadOnlySpan<char> chars, in EncodingContext context,
+        LengthFormat? lengthFormat = null)
+    {
+        Span<byte> buffer;
+        int byteCount, result;
+        if (lengthFormat.HasValue)
+        {
+            byteCount = context.Encoding.GetByteCount(chars);
+            result = writer.WriteLength(byteCount, lengthFormat.GetValueOrDefault());
+
+            buffer = writer.GetSpan(byteCount);
+            byteCount = context.TryGetEncoder() is { } encoder
+                ? encoder.GetBytes(chars, buffer, flush: true)
+                : context.Encoding.GetBytes(chars, buffer);
+
+            result += byteCount;
+            writer.Advance(byteCount);
+        }
+        else
+        {
+            result = 0;
+            var encoder = context.GetEncoder();
+            byteCount = context.Encoding.GetMaxByteCount(1);
+            for (int charsUsed, bytesWritten; !chars.IsEmpty; chars = chars.Slice(charsUsed), result += bytesWritten)
+            {
+                buffer = writer.GetSpan(byteCount);
+                var maxChars = buffer.Length / byteCount;
+
+                encoder.Convert(chars, buffer, chars.Length <= maxChars, out charsUsed, out bytesWritten, out _);
+                writer.Advance(bytesWritten);
+            }
+        }
+
+        return result;
+    }
+
+    /// <summary>
+    /// Writes a sequence of bytes prefixed with the length.
+    /// </summary>
+    /// <param name="writer">The buffer writer.</param>
+    /// <param name="value">A sequence of bytes to be written.</param>
+    /// <param name="lengthFormat">A format of the buffer length to be written.</param>
+    /// <returns>A number of bytes written.</returns>
+    public static int Write(this ref BufferWriterSlim<byte> writer, scoped ReadOnlySpan<byte> value, LengthFormat lengthFormat)
+    {
+        var result = writer.WriteLength(value.Length, lengthFormat);
+        writer.Write(value);
+        result += value.Length;
+
+        return result;
+    }
 
     private static bool TryFormat<T>(IBufferWriter<byte> writer, T value, Span<char> buffer, in EncodingContext context, LengthFormat? lengthFormat, ReadOnlySpan<char> format, IFormatProvider? provider, out long bytesWritten)
         where T : notnull, ISpanFormattable

diff --git a/src/DotNext.Tests/Buffers/BufferWriterSlimTests.cs b/src/DotNext.Tests/Buffers/BufferWriterSlimTests.cs
@@ -1,10 +1,13 @@
 using System.Numerics;
 using System.Text;
-using DotNext.Buffers.Binary;
 using static System.Globalization.CultureInfo;
 
 namespace DotNext.Buffers;
 
+using Binary;
+using IO;
+using static DotNext.Text.EncodingExtensions;
+
 public sealed class BufferWriterSlimTests : Test
 {
     [Fact]
@@ -360,4 +363,92 @@ public static void ReadWriteBigInteger()
 
         Equal(expected, new BigInteger(writer.WrittenSpan));
     }
+
+    private static void EncodeDecodeZeroAndMaxValue<T>()
+        where T : struct, IBinaryInteger<T>, IUnsignedNumber<T>
+    {
+        Span<byte> buffer = stackalloc byte[SevenBitEncodedInteger<T>.MaxSizeInBytes];
+        var writer = new BufferWriterSlim<byte>(buffer);
+        var reader = new SpanReader<byte>(buffer);
+
+        Equal(1, writer.Write7BitEncodedInteger(T.Zero));
+        Equal(T.Zero, reader.Read7BitEncodedInteger<T>());
+
+        writer.Clear(reuseBuffer: true);
+        reader.Reset();
+
+        Equal(SevenBitEncodedInteger<T>.MaxSizeInBytes, writer.Write7BitEncodedInteger(T.AllBitsSet));
+        Equal(T.AllBitsSet, reader.Read7BitEncodedInteger<T>());
+    }
+
+    [Fact]
+    public static void EncodeDecodeUInt32() => EncodeDecodeZeroAndMaxValue<uint>();
+
+    [Fact]
+    public static void EncodeDecodeUInt64() => EncodeDecodeZeroAndMaxValue<ulong>();
+
+    [InlineData(LengthFormat.BigEndian)]
+    [InlineData(LengthFormat.LittleEndian)]
+    [InlineData(LengthFormat.Compressed)]
+    [Theory]
+    public static void WriteLengthPrefixedBytes(LengthFormat format)
+    {
+        ReadOnlySpan<byte> expected = [1, 2, 3];
+
+        var writer = new BufferWriterSlim<byte>();
+        True(writer.Write(expected, format) > 0);
+
+        using var buffer = writer.DetachOrCopyBuffer();
+        var reader = IAsyncBinaryReader.Create(buffer.Memory);
+        using var actual = reader.ReadBlock(format);
+        Equal(expected, actual.Span);
+    }
+
+    [Theory]
+    [InlineData("UTF-8", null)]
+    [InlineData("UTF-8", LengthFormat.LittleEndian)]
+    [InlineData("UTF-8", LengthFormat.BigEndian)]
+    [InlineData("UTF-8", LengthFormat.Compressed)]
+    [InlineData("UTF-16LE", null)]
+    [InlineData("UTF-16LE", LengthFormat.LittleEndian)]
+    [InlineData("UTF-16LE", LengthFormat.BigEndian)]
+    [InlineData("UTF-16LE", LengthFormat.Compressed)]
+    [InlineData("UTF-16BE", null)]
+    [InlineData("UTF-16BE", LengthFormat.LittleEndian)]
+    [InlineData("UTF-16BE", LengthFormat.BigEndian)]
+    [InlineData("UTF-16BE", LengthFormat.Compressed)]
+    [InlineData("UTF-32LE", null)]
+    [InlineData("UTF-32LE", LengthFormat.LittleEndian)]
+    [InlineData("UTF-32LE", LengthFormat.BigEndian)]
+    [InlineData("UTF-32LE", LengthFormat.Compressed)]
+    [InlineData("UTF-32BE", null)]
+    [InlineData("UTF-32BE", LengthFormat.LittleEndian)]
+    [InlineData("UTF-32BE", LengthFormat.BigEndian)]
+    [InlineData("UTF-32BE", LengthFormat.Compressed)]
+    public static void EncodeDecodeString(string encodingName, LengthFormat? format)
+    {
+        var encoding = Encoding.GetEncoding(encodingName);
+        const string expected = "Hello, world!&*(@&*(fghjwgfwffgw Привет, мир!";
+        var writer = new BufferWriterSlim<byte>();
+
+        True(writer.Encode(expected, encoding, format) > 0);
+
+        using var buffer = writer.DetachOrCopyBuffer();
+        MemoryOwner<char> actual;
+        if (format.HasValue)
+        {
+            var reader = IAsyncBinaryReader.Create(buffer.Memory);
+            actual = reader.Decode(encoding, format.GetValueOrDefault());
+            Equal(expected, actual.Span);
+        }
+        else
+        {
+            actual = encoding.GetChars(buffer.Span);
+        }
+
+        using (actual)
+        {
+            Equal(expected, actual.Span);
+        }
+    }
 }
diff --git a/src/DotNext/Buffers/ByteBuffer.cs b/src/DotNext/Buffers/ByteBuffer.cs
@@ -342,6 +342,25 @@ public static int Write7BitEncodedInteger<T>(this ref SpanWriter<byte> writer, T
 
         return count;
     }
+
+    /// <summary>
+    /// Writes 32-bit integer in a compressed format.
+    /// </summary>
+    /// <param name="writer">The buffer writer.</param>
+    /// <param name="value">The integer to be written.</param>
+    /// <returns>A number of bytes written to the buffer.</returns>
+    public static int Write7BitEncodedInteger<T>(this ref BufferWriterSlim<byte> writer, T value)
+        where T : struct, IBinaryInteger<T>, IUnsignedNumber<T>
+    {
+        var count = 0;
+        foreach (var b in new SevenBitEncodedInteger<T>(value))
+        {
+            writer.Add() = b;
+            count += 1;
+        }
+
+        return count;
+    }
 
     /// <summary>
     /// Decodes an integer encoded as 7-bit octets.