[WIP] libarchive-based dearchiving

Directory.Packages.props

@@ -6,6 +6,7 @@
 		<PackageVersion Include="Google.FlatBuffers" Version="23.5.26" /> <!-- last version with .NET Standard 2.0 support -->
 		<PackageVersion Include="ImGui.NET" Version="1.90.6.1" />
 		<PackageVersion Include="JunitXml.TestLogger" Version="3.1.12" />
+		<PackageVersion Include="LibArchive.Net" Version="0.2.0-alpha.0.82" />
 		<PackageVersion Include="Magick.NET-Q8-AnyCPU" Version="14.8.2" />
 		<PackageVersion Include="Menees.Analyzers" Version="3.2.2" />
 		<PackageVersion Include="Meziantou.Analyzer" Version="2.0.206" />

NuGet.config

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="utf-8"?>
+<configuration>
+	<packageSources>
+		<add key="LocalPackages" value="./References" />
+	</packageSources>
+</configuration>

src/BizHawk.Common/BizHawk.Common.csproj

@@ -15,6 +15,7 @@
 		<PackageReference Include="System.ComponentModel.Annotations" />
 		<PackageReference Include="System.Memory" />
 		<PackageReference Include="System.Runtime.CompilerServices.Unsafe" />
+		<PackageReference Include="LibArchive.Net" />
 	</ItemGroup>
 	<ItemGroup>
 		<Analyzer Include="$(MSBuildProjectDirectory)/../../References/BizHawk.SrcGen.VersionInfo.dll" />

src/BizHawk.Common/LibarchiveArchiveFile.cs

@@ -0,0 +1,64 @@
+using System.Collections.Generic;
+using System.IO;
+using System.Linq;
+
+using BizHawk.Common;
+
+using LibArchive.Net;
+
+namespace BizHawk.Client.Common
+{
+	/// <see cref="LibarchiveDearchivalMethod"/>
+	public sealed class LibarchiveArchiveFile : IHawkArchiveFile
+	{
+		private LibArchiveReader? _handle;
+
+		private FileInfo? _tempOnDisk;
+
+		private (int ArchiveIndex, LibArchiveReader.Entry Entry)[]? field = null;
+
+		private (int ArchiveIndex, LibArchiveReader.Entry Entry)[] AllEntries
+			=> field ??= _handle?.Entries().Index().OrderBy(static tuple => tuple.Item.Name).ToArray()
+				?? throw new ObjectDisposedException(nameof(LibarchiveArchiveFile));
+
+		public LibarchiveArchiveFile(string path)
+			=> _handle = new(path);
+
+		public LibarchiveArchiveFile(Stream fileStream)
+		{
+			_tempOnDisk = new(TempFileManager.GetTempFilename("dearchive"));
+			using (FileStream fsCopy = new(_tempOnDisk.FullName, FileMode.Create)) fileStream.CopyTo(fsCopy);
+			try
+			{
+				_handle = new(_tempOnDisk.FullName);
+			}
+			catch (Exception e)
+			{
+				_tempOnDisk.Delete();
+				Console.WriteLine(e);
+				throw;
+			}
+		}
+
+		public void Dispose()
+		{
+			_handle?.Dispose();
+			_handle = null;
+			_tempOnDisk?.Delete();
+			_tempOnDisk = null;
+		}
+
+		public void ExtractFile(int index, Stream stream)
+		{
+			using var entryStream = AllEntries[index].Entry.Stream;
+			entryStream.CopyTo(stream);
+		}
+
+		public List<HawkArchiveFileItem>? Scan()
+			=> AllEntries.Select(static (tuple, i) => new HawkArchiveFileItem(
+				tuple.Entry.Name,
+				size: tuple.Entry.LengthBytes ?? 0,
+				index: i,
+				archiveIndex: tuple.ArchiveIndex)).ToList();
+	}
+}

src/BizHawk.Common/LibarchiveDearchivalMethod.cs

@@ -0,0 +1,71 @@
+using System.Collections.Generic;
+using System.IO;
+
+using BizHawk.Common;
+
+namespace BizHawk.Client.Common
+{
+	/// <summary>A <see cref="IFileDearchivalMethod{T}">dearchival method</see> for <see cref="HawkFile"/> implemented using <c>libarchive</c> (via <c>LibArchive.Net</c> bindings).</summary>
+	public class LibarchiveDearchivalMethod : IFileDearchivalMethod<LibarchiveArchiveFile>
+	{
+		public static readonly LibarchiveDearchivalMethod Instance = new();
+
+		public IReadOnlyCollection<string> AllowedArchiveExtensions { get; } = [
+			".7z",
+			".gz",
+			".rar",
+			".tar",
+			/*.tar*/".bz2", ".tb2", ".tbz", ".tbz2", ".tz2",
+			/*.tar.gz,*/ ".taz", ".tgz",
+			/*.tar*/".lz",
+			".zip",
+		];
+
+		private LibarchiveDearchivalMethod() {}
+
+		public bool CheckSignature(string fileName)
+		{
+			LibarchiveArchiveFile? file = null;
+			try
+			{
+				file = new(fileName);
+				return true;
+			}
+			catch (Exception)
+			{
+				return false;
+			}
+			finally
+			{
+				file?.Dispose();
+			}
+		}
+
+		public bool CheckSignature(Stream fileStream, string? filenameHint)
+		{
+			if (!fileStream.CanRead || !fileStream.CanSeek) return false;
+			var initialPosition = fileStream.Position;
+			LibarchiveArchiveFile? file = null;
+			try
+			{
+				file = new(fileStream);
+				return true;
+			}
+			catch (Exception)
+			{
+				return false;
+			}
+			finally
+			{
+				file?.Dispose();
+				fileStream.Seek(initialPosition, SeekOrigin.Begin);
+			}
+		}
+
+		public LibarchiveArchiveFile Construct(string path)
+			=> new(path);
+
+		public LibarchiveArchiveFile Construct(Stream fileStream)
+			=> new(fileStream);
+	}
+}

src/BizHawk.Tests.Client.Common/dearchive/DearchivalTests.cs

@@ -1,5 +1,5 @@
-using System.IO;
-using System.Linq;
+using System.Collections.Generic;
+using System.IO;
 
 using BizHawk.Client.Common;
 using BizHawk.Common;
@@ -12,13 +12,14 @@ public class DearchivalTests
 	{
 		private const string EMBED_GROUP = "data.dearchive.";
 
-		private static readonly (string Filename, bool HasSharpCompressSupport)[] TestCases = {
-			("m3_scy_change.7z", true),
-			("m3_scy_change.gb.gz", true),
-			("m3_scy_change.rar", true),
-			("m3_scy_change.bsdtar.tar", true),
-			("m3_scy_change.gnutar.tar", true),
-			("m3_scy_change.zip", true),
+		private static IEnumerable<object?[]> TestCases { get; } = new[]
+		{
+			new object?[] { "m3_scy_change.7z", true },
+			new object?[] { "m3_scy_change.gb.gz", true },
+			new object?[] { "m3_scy_change.rar", true },
+			new object?[] { "m3_scy_change.bsdtar.tar", true },
+			new object?[] { "m3_scy_change.gnutar.tar", true },
+			new object?[] { "m3_scy_change.zip", true },
 		};
 
 		private readonly Lazy<byte[]> _rom = new(static () => ReflectionCache.EmbeddedResourceStream(EMBED_GROUP + "m3_scy_change.gb").ReadAllBytes());
@@ -30,16 +31,33 @@ private static readonly (string Filename, bool HasSharpCompressSupport)[] TestCa
 		public void SanityCheck() => Assert.AreEqual("SHA1:70DCA8E791878BDD32426391E4233EA52B47CDD1", SHA1Checksum.ComputePrefixedHex(Rom));
 #pragma warning restore BHI1600
 
+		[DynamicData(nameof(TestCases))]
+		[TestMethod]
+		public void TestLibarchive(string filename, bool hasSharpCompressSupport)
+		{
+			var sc = LibarchiveDearchivalMethod.Instance;
+			var archive = ReflectionCache.EmbeddedResourceStream(EMBED_GROUP + filename);
+			Assert.IsTrue(sc.CheckSignature(archive, filename), $"{filename} is an archive, but wasn't detected as such"); // puts the seek pos of the Stream param back where it was (in this case at the start)
+			using var af = sc.Construct(archive);
+			var items = af.Scan();
+			Assert.IsNotNull(items, $"{filename} contains 1 file, but it couldn't be enumerated correctly");
+			Assert.AreEqual(1, items!.Count, $"{filename} contains 1 file, but was detected as containing {items.Count} files");
+			using MemoryStream ms = new((int) items[0].Size);
+			af.ExtractFile(items[0].ArchiveIndex, ms);
+			ms.Seek(0L, SeekOrigin.Begin);
+			CollectionAssert.AreEqual(Rom, ms.ReadAllBytes(), $"the file extracted from {filename} doesn't match the uncompressed file");
+		}
+
+		[DynamicData(nameof(TestCases))]
 		[TestMethod]
-		public void TestSharpCompress()
+		public void TestSharpCompress(string filename, bool hasSharpCompressSupport)
 		{
+			if (!hasSharpCompressSupport) return;
 			var sc = SharpCompressDearchivalMethod.Instance;
-			foreach (var filename in TestCases.Where(testCase => testCase.HasSharpCompressSupport)
-				.Select(testCase => testCase.Filename))
-			{
+			/*scope*/{
 				var archive = ReflectionCache.EmbeddedResourceStream(EMBED_GROUP + filename);
 				Assert.IsTrue(sc.CheckSignature(archive, filename), $"{filename} is an archive, but wasn't detected as such"); // puts the seek pos of the Stream param back where it was (in this case at the start)
-				var af = sc.Construct(archive);
+				using var af = sc.Construct(archive);
 				var items = af.Scan();
 				Assert.IsNotNull(items, $"{filename} contains 1 file, but it couldn't be enumerated correctly");
 				Assert.AreEqual(1, items!.Count, $"{filename} contains 1 file, but was detected as containing {items.Count} files");