Merge pull request #1172 from adamhathcock/copilot/fix-sevenzip-contiguous-streams

Fix SevenZipReader to maintain contiguous stream state for solid archives

Author: adamhathcock

src/SharpCompress/Archives/SevenZip/SevenZipArchive.cs

@@ -212,10 +212,31 @@ protected override IReader CreateReaderForSolidExtraction() =>
     public override long TotalSize =>
         _database?._packSizes.Aggregate(0L, (total, packSize) => total + packSize) ?? 0;
 
-    private sealed class SevenZipReader : AbstractReader<SevenZipEntry, SevenZipVolume>
+    internal sealed class SevenZipReader : AbstractReader<SevenZipEntry, SevenZipVolume>
     {
         private readonly SevenZipArchive _archive;
         private SevenZipEntry? _currentEntry;
+        private Stream? _currentFolderStream;
+        private CFolder? _currentFolder;
+
+        /// <summary>
+        /// Enables internal diagnostics for tests.
+        /// When disabled (default), diagnostics properties return null to avoid exposing internal state.
+        /// </summary>
+        internal bool DiagnosticsEnabled { get; set; }
+
+        /// <summary>
+        /// Current folder instance used to decide whether the solid folder stream should be reused.
+        /// Only available when <see cref="DiagnosticsEnabled"/> is true.
+        /// </summary>
+        internal object? DiagnosticsCurrentFolder => DiagnosticsEnabled ? _currentFolder : null;
+
+        /// <summary>
+        /// Current shared folder stream instance.
+        /// Only available when <see cref="DiagnosticsEnabled"/> is true.
+        /// </summary>
+        internal Stream? DiagnosticsCurrentFolderStream =>
+            DiagnosticsEnabled ? _currentFolderStream : null;
 
         internal SevenZipReader(ReaderOptions readerOptions, SevenZipArchive archive)
             : base(readerOptions, ArchiveType.SevenZip) => this._archive = archive;
@@ -231,9 +252,10 @@ protected override IEnumerable<SevenZipEntry> GetEntries(Stream stream)
                 _currentEntry = dir;
                 yield return dir;
             }
-            // For non-directory entries, yield them without creating shared streams
-            // Each call to GetEntryStream() will create a fresh decompression stream
-            // to avoid state corruption issues with async operations
+            // For solid archives (entries in the same folder share a compressed stream),
+            // we must iterate entries sequentially and maintain the folder stream state
+            // across entries in the same folder to avoid recreating the decompression
+            // stream for each file, which breaks contiguous streaming.
             foreach (var entry in entries.Where(x => !x.IsDirectory))
             {
                 _currentEntry = entry;
@@ -243,19 +265,53 @@ protected override IEnumerable<SevenZipEntry> GetEntries(Stream stream)
 
         protected override EntryStream GetEntryStream()
         {
-            // Create a fresh decompression stream for each file (no state sharing).
-            // However, the LZMA decoder has bugs in its async implementation that cause
-            // state corruption even on fresh streams. The SyncOnlyStream wrapper
-            // works around these bugs by forcing async operations to use sync equivalents.
-            //
-            // TODO: Fix the LZMA decoder async bugs (in LzmaStream, Decoder, OutWindow)
-            // so this wrapper is no longer necessary.
             var entry = _currentEntry.NotNull("currentEntry is not null");
             if (entry.IsDirectory)
             {
                 return CreateEntryStream(Stream.Null);
             }
-            return CreateEntryStream(new SyncOnlyStream(entry.FilePart.GetCompressedStream()));
+
+            var filePart = (SevenZipFilePart)entry.FilePart;
+            if (!filePart.Header.HasStream)
+            {
+                // Entries with no underlying stream (e.g., empty files or anti-items)
+                // should return an empty stream, matching previous behavior.
+                return CreateEntryStream(Stream.Null);
+            }
+
+            var folder = filePart.Folder;
+            // Check if we're starting a new folder - dispose old folder stream if needed
+            if (folder != _currentFolder)
+            {
+                _currentFolderStream?.Dispose();
+                _currentFolderStream = null;
+                _currentFolder = folder;
+            }
+
+            // Create the folder stream once per folder
+            if (_currentFolderStream is null)
+            {
+                _currentFolderStream = _archive._database!.GetFolderStream(
+                    _archive.Volumes.Single().Stream,
+                    folder!,
+                    _archive._database.PasswordProvider
+                );
+            }
+
+            // Wrap with SyncOnlyStream to work around LZMA async bugs
+            // Return a ReadOnlySubStream that reads from the shared folder stream
+            return CreateEntryStream(
+                new SyncOnlyStream(
+                    new ReadOnlySubStream(_currentFolderStream, entry.Size, leaveOpen: true)
+                )
+            );
+        }
+
+        public override void Dispose()
+        {
+            _currentFolderStream?.Dispose();
+            _currentFolderStream = null;
+            base.Dispose();
         }
     }
 

tests/SharpCompress.Test/SevenZip/SevenZipArchiveTests.cs

@@ -251,4 +251,98 @@ public void SevenZipArchive_TestSolidDetection()
         );
         Assert.False(nonSolidArchive.IsSolid);
     }
+
+    [Fact]
+    public void SevenZipArchive_Solid_ExtractAllEntries_Contiguous()
+    {
+        // This test verifies that solid archives iterate entries as contiguous streams
+        // rather than recreating the decompression stream for each entry
+        var testArchive = Path.Combine(TEST_ARCHIVES_PATH, "7Zip.solid.7z");
+        using var archive = SevenZipArchive.Open(testArchive);
+        Assert.True(archive.IsSolid);
+
+        using var reader = archive.ExtractAllEntries();
+        while (reader.MoveToNextEntry())
+        {
+            if (!reader.Entry.IsDirectory)
+            {
+                reader.WriteEntryToDirectory(
+                    SCRATCH_FILES_PATH,
+                    new ExtractionOptions { ExtractFullPath = true, Overwrite = true }
+                );
+            }
+        }
+
+        VerifyFiles();
+    }
+
+    [Fact]
+    public void SevenZipArchive_Solid_VerifyStreamReuse()
+    {
+        // This test verifies that the folder stream is reused within each folder
+        // and not recreated for each entry in solid archives
+        var testArchive = Path.Combine(TEST_ARCHIVES_PATH, "7Zip.solid.7z");
+        using var archive = SevenZipArchive.Open(testArchive);
+        Assert.True(archive.IsSolid);
+
+        using var reader = archive.ExtractAllEntries();
+
+        var sevenZipReader = Assert.IsType<SevenZipArchive.SevenZipReader>(reader);
+        sevenZipReader.DiagnosticsEnabled = true;
+
+        Stream? currentFolderStreamInstance = null;
+        object? currentFolder = null;
+        var entryCount = 0;
+        var entriesInCurrentFolder = 0;
+        var streamRecreationsWithinFolder = 0;
+
+        while (reader.MoveToNextEntry())
+        {
+            if (!reader.Entry.IsDirectory)
+            {
+                // Extract the entry to trigger GetEntryStream
+                using var entryStream = reader.OpenEntryStream();
+                var buffer = new byte[4096];
+                while (entryStream.Read(buffer, 0, buffer.Length) > 0)
+                {
+                    // Read the stream to completion
+                }
+
+                entryCount++;
+
+                var folderStream = sevenZipReader.DiagnosticsCurrentFolderStream;
+                var folder = sevenZipReader.DiagnosticsCurrentFolder;
+
+                Assert.NotNull(folderStream); // Folder stream should exist
+
+                // Check if we're in a new folder
+                if (currentFolder == null || !ReferenceEquals(currentFolder, folder))
+                {
+                    // Starting a new folder
+                    currentFolder = folder;
+                    currentFolderStreamInstance = folderStream;
+                    entriesInCurrentFolder = 1;
+                }
+                else
+                {
+                    // Same folder - verify stream wasn't recreated
+                    entriesInCurrentFolder++;
+
+                    if (!ReferenceEquals(currentFolderStreamInstance, folderStream))
+                    {
+                        // Stream was recreated within the same folder - this is the bug we're testing for!
+                        streamRecreationsWithinFolder++;
+                    }
+
+                    currentFolderStreamInstance = folderStream;
+                }
+            }
+        }
+
+        // Verify we actually tested multiple entries
+        Assert.True(entryCount > 1, "Test should have multiple entries to verify stream reuse");
+
+        // The critical check: within a single folder, the stream should NEVER be recreated
+        Assert.Equal(0, streamRecreationsWithinFolder); // Folder stream should remain the same for all entries in the same folder
+    }
 }