feat: Support Zstandard-compressed assemblies on Android (.NET 11)

src/Sentry.Android.AssemblyReader/ArchiveUtils.cs

@@ -2,9 +2,12 @@ namespace Sentry.Android.AssemblyReader;
 
 internal static class ArchiveUtils
 {
+    internal const uint Lz4Magic = 0x5A4C4158; // 'XALZ', little-endian
+    internal const uint ZstandardMagic = 0x535A4158; // 'XAZS', little-endian
+
     internal static PEReader CreatePEReader(string assemblyName, MemoryStream inputStream, DebugLogger? logger)
     {
-        var decompressedStream = TryDecompressLZ4(assemblyName, inputStream, logger); // Returns null if not compressed
+        var decompressedStream = TryDecompress(assemblyName, inputStream, logger); // Returns null if not compressed
         return new PEReader(decompressedStream ?? inputStream);
     }
 
@@ -18,21 +21,22 @@ internal static MemoryStream Extract(this ZipArchiveEntry zipEntry)
     }
 
     /// <summary>
-    /// The DLL may be LZ4 compressed, see https://github.com/xamarin/xamarin-android/pull/4686
+    /// The DLL may be compressed, see https://github.com/xamarin/xamarin-android/pull/4686
     /// In particular: https://github.com/dotnet/android/blob/44c5c30d3da692c54ca27d4a41571ef20b73670f/src/Xamarin.Android.Build.Tasks/Utilities/AssemblyCompression.cs#L96-L104
     /// The format is:
-    ///    [ 4 byte magic header ] (XALZ)
+    ///    [ 4 byte magic header ] (XALZ for LZ4, XAZS for Zstandard)
     ///    [ 4 byte descriptor header index ]
     ///    [ 4 byte uncompressed payload length ]
-    ///    [rest: lz4 compressed payload]
+    ///    [rest: compressed payload]
+    /// .NET 11 switched from LZ4 to Zstandard: https://github.com/dotnet/android/pull/11730
     /// </summary>
-    /// <seealso href="https://github.com/xamarin/xamarin-android/blob/c92702619f5fabcff0ed88e09160baf9edd70f41/tools/decompress-assemblies/main.cs#L26" />
-    private static Stream? TryDecompressLZ4(string assemblyName, MemoryStream inputStream, DebugLogger? logger)
+    /// <seealso href="https://github.com/dotnet/android/blob/f1aecf9e6ae80fe3f3992ec1f52ef953dac7c06b/.github/skills/read-assembly-store/src/AssemblyStore/AssemblyCompression.cs" />
+    private static Stream? TryDecompress(string assemblyName, MemoryStream inputStream, DebugLogger? logger)
     {
-        const uint compressedDataMagic = 0x5A4C4158; // 'XALZ', little-endian
         const int payloadOffset = 12;
         var reader = new BinaryReader(inputStream);
-        if (reader.ReadUInt32() != compressedDataMagic)
+        var magic = reader.ReadUInt32();
+        if (magic is not (Lz4Magic or ZstandardMagic))
         {
             // Restore the input stream to the beginning if we're not decompressing.
             inputStream.Position = 0;
@@ -42,8 +46,9 @@ internal static MemoryStream Extract(this ZipArchiveEntry zipEntry)
         var decompressedLength = reader.ReadInt32();
         Debug.Assert(inputStream.Position == payloadOffset);
         var inputLength = (int)(inputStream.Length - payloadOffset);
+        var format = magic == Lz4Magic ? "LZ4" : "Zstandard";
 
-        logger?.Invoke(DebugLoggerLevel.Debug, "Decompressing assembly ({0} bytes uncompressed) using LZ4", decompressedLength);
+        logger?.Invoke(DebugLoggerLevel.Debug, "Decompressing assembly ({0} bytes uncompressed) using {1}", decompressedLength, format);
 
         var outputStream = new MemoryStream(decompressedLength);
 
@@ -53,10 +58,23 @@ internal static MemoryStream Extract(this ZipArchiveEntry zipEntry)
 
         var inputBuffer = inputStream is MemorySlice slice ? slice.FullBuffer : inputStream.GetBuffer();
         var offset = inputStream is MemorySlice memorySlice ? memorySlice.Offset + payloadOffset : payloadOffset;
-        var decoded = LZ4Codec.Decode(inputBuffer, offset, inputLength, outputBuffer, 0, decompressedLength);
+        int decoded;
+        if (magic == Lz4Magic)
+        {
+            decoded = LZ4Codec.Decode(inputBuffer, offset, inputLength, outputBuffer, 0, decompressedLength);
+        }
+        else
+        {
+#if NET11_0_OR_GREATER
+            decoded = ZstandardDecoder.TryDecompress(inputBuffer.AsSpan(offset, inputLength),
+                outputBuffer.AsSpan(0, decompressedLength), out var bytesWritten) ? bytesWritten : -1;
+#else
+            throw new NotSupportedException($"Assembly {assemblyName} is Zstandard compressed, which requires .NET 11 or later");
+#endif
+        }
         if (decoded != decompressedLength)
         {
-            throw new Exception($"Failed to decompress LZ4 data of assembly {assemblyName} - decoded {decoded} instead of expected {decompressedLength} bytes");
+            throw new Exception($"Failed to decompress {format} data of assembly {assemblyName} - decoded {decoded} instead of expected {decompressedLength} bytes");
         }
         return outputStream;
     }

test/Sentry.Android.AssemblyReader.Tests/ArchiveUtilsTests.cs

@@ -0,0 +1,101 @@
+using System.Reflection.Metadata;
+using K4os.Compression.LZ4;
+
+namespace Sentry.Android.AssemblyReader.Tests;
+
+// Assembly.Location is empty on Android; AndroidAssemblyReaderTests covers decompression on device
+#if !ANDROID
+public class ArchiveUtilsTests
+{
+    private static readonly byte[] Assembly = File.ReadAllBytes(typeof(ArchiveUtilsTests).Assembly.Location);
+
+    [Fact]
+    public void CreatePEReader_Uncompressed_ReadsAssembly()
+    {
+        using var peReader = ArchiveUtils.CreatePEReader("test.dll", new MemoryStream(Assembly), null);
+
+        AssertIsThisAssembly(peReader);
+    }
+
+    [Fact]
+    public void CreatePEReader_Lz4_ReadsAssembly()
+    {
+        var compressed = new byte[LZ4Codec.MaximumOutputSize(Assembly.Length)];
+        var length = LZ4Codec.Encode(Assembly, 0, Assembly.Length, compressed, 0, compressed.Length);
+
+        using var peReader = ArchiveUtils.CreatePEReader("test.dll", WithHeader(ArchiveUtils.Lz4Magic, compressed.AsSpan(0, length)), null);
+
+        AssertIsThisAssembly(peReader);
+    }
+
+#if NET11_0_OR_GREATER
+    [Fact]
+    public void CreatePEReader_Zstandard_ReadsAssembly()
+    {
+        var compressed = new byte[ZstandardEncoder.GetMaxCompressedLength(Assembly.Length)];
+        ZstandardEncoder.TryCompress(Assembly, compressed, out var length).Should().BeTrue();
+
+        using var peReader = ArchiveUtils.CreatePEReader("test.dll", WithHeader(ArchiveUtils.ZstandardMagic, compressed.AsSpan(0, length)), null);
+
+        AssertIsThisAssembly(peReader);
+    }
+
+    [Fact]
+    public void CreatePEReader_CorruptZstandard_Throws()
+    {
+        var garbage = new byte[64];
+
+        var act = () => ArchiveUtils.CreatePEReader("test.dll", WithHeader(ArchiveUtils.ZstandardMagic, garbage), null);
+
+        act.Should().Throw<Exception>().WithMessage("*Zstandard*test.dll*");
+    }
+#else
+    [Fact]
+    public void CreatePEReader_Zstandard_ThrowsNotSupported()
+    {
+        var act = () => ArchiveUtils.CreatePEReader("test.dll", WithHeader(ArchiveUtils.ZstandardMagic, new byte[64]), null);
+
+        act.Should().Throw<NotSupportedException>().WithMessage("*test.dll*Zstandard*");
+    }
+#endif
+
+    [Fact]
+    public void CreatePEReader_SliceOfLargerBuffer_ReadsAssembly()
+    {
+        var compressed = new byte[LZ4Codec.MaximumOutputSize(Assembly.Length)];
+        var length = LZ4Codec.Encode(Assembly, 0, Assembly.Length, compressed, 0, compressed.Length);
+        var entry = WithHeader(ArchiveUtils.Lz4Magic, compressed.AsSpan(0, length)).ToArray();
+
+        const int prefix = 100;
+        var store = new MemoryStream();
+        store.Write(new byte[prefix]);
+        store.Write(entry);
+        var slice = new ArchiveUtils.MemorySlice(store, prefix, entry.Length);
+
+        using var peReader = ArchiveUtils.CreatePEReader("test.dll", slice, null);
+
+        AssertIsThisAssembly(peReader);
+    }
+
+    private static MemoryStream WithHeader(uint magic, ReadOnlySpan<byte> payload)
+    {
+        var stream = new MemoryStream();
+        using (var writer = new BinaryWriter(stream, Encoding.UTF8, leaveOpen: true))
+        {
+            writer.Write(magic);
+            writer.Write(0u); // descriptor index
+            writer.Write(Assembly.Length);
+            writer.Write(payload);
+        }
+        stream.Position = 0;
+        return stream;
+    }
+
+    private static void AssertIsThisAssembly(PEReader peReader)
+    {
+        peReader.HasMetadata.Should().BeTrue();
+        peReader.GetMetadataReader().GetAssemblyDefinition().GetAssemblyName().Name
+            .Should().Be(typeof(ArchiveUtilsTests).Assembly.GetName().Name);
+    }
+}
+#endif