From 6e3c1874936b767754cd9267749f35b27f2ae65f Mon Sep 17 00:00:00 2001 From: Piotr Niełacny Date: Wed, 15 Jul 2026 14:55:55 +0200 Subject: Fix race condition in concurrent subtitle conversion SubtitleEncoder.ConvertSubtitles parsed subtitles with libse's static Subtitle.Parse, which iterates a statically cached list of shared SubtitleFormat instances. Format parsers keep mutable per-parse state on the instance, so concurrent subtitle requests corrupted each other's output (cues mixed across streams and languages, truncated files) or failed with NullReferenceException when format detection broke down and Subtitle.Parse returned null. Parse through the injected ISubtitleParser instead. SubtitleEditParser instantiates a fresh format parser per call, so requests no longer share state. Its Parse method now returns the libse Subtitle directly (the SubtitleTrackInfo flattening was unused since the SubtitleEdit writer rework) so the writers keep full fidelity such as ASS styling. --- .../Subtitles/SubtitleEncoderTests.cs | 94 ++++++++++++++++++++++ 1 file changed, 94 insertions(+) (limited to 'tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs') diff --git a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs index ce1f005f40..2d0fa29c9a 100644 --- a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs +++ b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs @@ -1,3 +1,8 @@ +using System; +using System.Globalization; +using System.IO; +using System.Linq; +using System.Text; using System.Threading; using System.Threading.Tasks; using AutoFixture; @@ -6,12 +11,16 @@ using MediaBrowser.MediaEncoding.Subtitles; using MediaBrowser.Model.Dto; using MediaBrowser.Model.Entities; using MediaBrowser.Model.MediaInfo; +using Microsoft.Extensions.Logging.Abstractions; using Xunit; namespace Jellyfin.MediaEncoding.Subtitles.Tests { public class SubtitleEncoderTests { + private const int StreamCount = 8; + private const int CueCount = 500; + public static TheoryData GetReadableFile_Valid_TestData() { var data = new TheoryData(); @@ -103,5 +112,90 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests Assert.Equal(subtitleInfo.Format, result.Format); Assert.Equal(subtitleInfo.IsExternal, result.IsExternal); } + + [Fact] + public void ConvertSubtitles_SequentialCalls_AreDeterministic() + { + using var encoder = CreateEncoder(); + var sources = GenerateSources(); + + var first = ConvertAllSequential(encoder, sources); + var second = ConvertAllSequential(encoder, sources); + + for (var i = 0; i < StreamCount; i++) + { + Assert.Contains($"S{i}C{CueCount - 1}", first[i], StringComparison.Ordinal); + Assert.Equal(first[i], second[i]); + } + } + + [Fact] + public async Task ConvertSubtitles_ConcurrentCalls_MatchSequentialBaseline() + { + const int Iterations = 10; + + using var encoder = CreateEncoder(); + var sources = GenerateSources(); + var baseline = ConvertAllSequential(encoder, sources); + + for (var iteration = 0; iteration < Iterations; iteration++) + { + var results = await Task.WhenAll(Enumerable.Range(0, StreamCount) + .Select(i => Task.Run(() => Convert(encoder, sources[i], i))) + .ToArray()); + + for (var i = 0; i < StreamCount; i++) + { + Assert.True( + string.Equals(baseline[i], results[i], StringComparison.Ordinal), + $"Iteration {iteration}: stream {i} returned corrupted content ({results[i].Length} chars vs {baseline[i].Length} baseline)"); + } + } + } + + private static SubtitleEncoder CreateEncoder() + { + var fixture = new Fixture().Customize(new AutoMoqCustomization { ConfigureMembers = true }); + fixture.Inject(new SubtitleEditParser(NullLogger.Instance)); + return fixture.Create(); + } + + private static byte[][] GenerateSources() + { + return Enumerable.Range(0, StreamCount) + .Select(i => Encoding.UTF8.GetBytes(GenerateSrt(i, CueCount))) + .ToArray(); + } + + private static string Convert(SubtitleEncoder encoder, byte[] source, int streamIndex) + { + using var input = new MemoryStream(source); + var info = new SubtitleEncoder.SubtitleInfo { Path = $"track{streamIndex}.srt", Format = "srt" }; + using var output = encoder.ConvertSubtitles(input, info, "vtt", 0, 0, false); + return Encoding.UTF8.GetString(output.ToArray()); + } + + private static string[] ConvertAllSequential(SubtitleEncoder encoder, byte[][] sources) + { + return sources.Select((source, i) => Convert(encoder, source, i)).ToArray(); + } + + private static string GenerateSrt(int streamIndex, int cueCount) + { + var builder = new StringBuilder(); + for (var i = 0; i < cueCount; i++) + { + var start = TimeSpan.FromSeconds(i * 4); + var end = start + TimeSpan.FromSeconds(2); + builder.Append(i + 1).AppendLine() + .Append(start.ToString(@"hh\:mm\:ss\,fff", CultureInfo.InvariantCulture)) + .Append(" --> ") + .AppendLine(end.ToString(@"hh\:mm\:ss\,fff", CultureInfo.InvariantCulture)) + .Append('S').Append(streamIndex).Append('C').Append(i).AppendLine() + .AppendLine(); + } + + return builder.ToString(); + } } } -- cgit v1.2.3 From ed61acc19a958ceaf8de6f6f0f481b2b5370463a Mon Sep 17 00:00:00 2001 From: Tim Eisele Date: Tue, 21 Jul 2026 14:52:59 +0200 Subject: Fix subtitle encoding for local files (#17281) * Fix subtitle encoding * Add short-circuit * Use IsTextFormat * Update MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs Co-authored-by: Bond-009 --------- Co-authored-by: Bond-009 --- .../Subtitles/SubtitleEncoder.cs | 34 +-- .../Subtitles/SubtitleEncoderTests.cs | 244 +++++++++++++++------ 2 files changed, 201 insertions(+), 77 deletions(-) (limited to 'tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs') diff --git a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs index c568a74b01..bd516f0a9f 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs @@ -163,28 +163,36 @@ namespace MediaBrowser.MediaEncoding.Subtitles return (stream, fileInfo); } - private async Task GetSubtitleStream(SubtitleInfo fileInfo, CancellationToken cancellationToken) + internal async Task GetSubtitleStream(SubtitleInfo fileInfo, CancellationToken cancellationToken) { - if (fileInfo.Protocol == MediaProtocol.Http) + if (fileInfo.IsExternal && MediaStream.IsTextFormat(fileInfo.Format)) { var result = await DetectCharset(fileInfo.Path, cancellationToken).ConfigureAwait(false); var detected = result.Detected; - if (detected is not null) + var stream = fileInfo.Protocol == MediaProtocol.Http + ? await _httpClientFactory.CreateClient(NamedClient.Default) + .GetStreamAsync(new Uri(fileInfo.Path), cancellationToken) + .ConfigureAwait(false) + : AsyncFile.OpenRead(fileInfo.Path); + + // Short-circuit when the file is already UTF-8/ASCII. + if (detected is null + || string.Equals(detected.EncodingName, "utf-8", StringComparison.OrdinalIgnoreCase) + || string.Equals(detected.EncodingName, "ascii", StringComparison.OrdinalIgnoreCase) + || string.Equals(detected.EncodingName, "us-ascii", StringComparison.OrdinalIgnoreCase)) { - _logger.LogDebug("charset {CharSet} detected for {Path}", detected.EncodingName, fileInfo.Path); + return stream; + } - using var stream = await _httpClientFactory.CreateClient(NamedClient.Default) - .GetStreamAsync(new Uri(fileInfo.Path), cancellationToken) - .ConfigureAwait(false); + _logger.LogDebug("charset {CharSet} detected for {Path}", detected.EncodingName, fileInfo.Path); - await using (stream.ConfigureAwait(false)) - { - using var reader = new StreamReader(stream, detected.Encoding); - var text = await reader.ReadToEndAsync(cancellationToken).ConfigureAwait(false); + await using (stream.ConfigureAwait(false)) + { + using var reader = new StreamReader(stream, detected.Encoding); + var text = await reader.ReadToEndAsync(cancellationToken).ConfigureAwait(false); - return new MemoryStream(Encoding.UTF8.GetBytes(text)); - } + return new MemoryStream(Encoding.UTF8.GetBytes(text)); } } diff --git a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs index 2d0fa29c9a..48850b2f67 100644 --- a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs +++ b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SubtitleEncoderTests.cs @@ -21,81 +21,110 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests private const int StreamCount = 8; private const int CueCount = 500; - public static TheoryData GetReadableFile_Valid_TestData() + // A Greek line that requires a non-UTF-8 legacy encoding to reproduce the bug. The accented + // characters (ά, έ, ή, ί, ό, ύ, ώ) share the same code points in windows-1253 and iso-8859-7, + // so a Greek-vs-Greek charset misdetection still round-trips correctly. + private const string GreekText = "Καλημέρα κόσμε, αυτό είναι ένας υπότιτλος."; + + static SubtitleEncoderTests() { - var data = new TheoryData(); + // Mirrors Jellyfin.Server startup so legacy code pages (e.g. Greek windows-1253) are available. + Encoding.RegisterProvider(CodePagesEncodingProvider.Instance); + } - data.Add( - new MediaSourceInfo() - { - Protocol = MediaProtocol.File - }, - new MediaStream() - { - Path = "/media/sub.ass", - IsExternal = true - }, - new SubtitleEncoder.SubtitleInfo() - { - Path = "/media/sub.ass", - Protocol = MediaProtocol.File, - Format = "ass", - IsExternal = true - }); + // Enough Greek text to give the charset detector a strong, unambiguous signal. + private static string BuildGreekSrt() + { + var builder = new StringBuilder(); + for (var i = 1; i <= 8; i++) + { + builder.Append(i.ToString(CultureInfo.InvariantCulture)).Append('\n'); + builder.Append("00:00:0").Append(i.ToString(CultureInfo.InvariantCulture)) + .Append(",000 --> 00:00:0").Append((i + 1).ToString(CultureInfo.InvariantCulture)).Append(",000\n"); + builder.Append(GreekText).Append('\n'); + builder.Append("Η γρήγορη καφέ αλεπού πηδάει πάνω από το τεμπέλικο σκυλί.\n\n"); + } - data.Add( - new MediaSourceInfo() - { - Protocol = MediaProtocol.File - }, - new MediaStream() - { - Path = "/media/sub.ssa", - IsExternal = true - }, - new SubtitleEncoder.SubtitleInfo() - { - Path = "/media/sub.ssa", - Protocol = MediaProtocol.File, - Format = "ssa", - IsExternal = true - }); + return builder.ToString(); + } - data.Add( - new MediaSourceInfo() + public static TheoryData GetReadableFile_Valid_TestData() + { + var data = new TheoryData + { { - Protocol = MediaProtocol.File + new MediaSourceInfo() + { + Protocol = MediaProtocol.File + }, + new MediaStream() + { + Path = "/media/sub.ass", + IsExternal = true + }, + new SubtitleEncoder.SubtitleInfo() + { + Path = "/media/sub.ass", + Protocol = MediaProtocol.File, + Format = "ass", + IsExternal = true + } }, - new MediaStream() { - Path = "/media/sub.srt", - IsExternal = true + new MediaSourceInfo() + { + Protocol = MediaProtocol.File + }, + new MediaStream() + { + Path = "/media/sub.ssa", + IsExternal = true + }, + new SubtitleEncoder.SubtitleInfo() + { + Path = "/media/sub.ssa", + Protocol = MediaProtocol.File, + Format = "ssa", + IsExternal = true + } }, - new SubtitleEncoder.SubtitleInfo() { - Path = "/media/sub.srt", - Protocol = MediaProtocol.File, - Format = "srt", - IsExternal = true - }); - - data.Add( - new MediaSourceInfo() - { - Protocol = MediaProtocol.Http + new MediaSourceInfo() + { + Protocol = MediaProtocol.File + }, + new MediaStream() + { + Path = "/media/sub.srt", + IsExternal = true + }, + new SubtitleEncoder.SubtitleInfo() + { + Path = "/media/sub.srt", + Protocol = MediaProtocol.File, + Format = "srt", + IsExternal = true + } }, - new MediaStream() { - Path = "/media/sub.ass", - IsExternal = true - }, - new SubtitleEncoder.SubtitleInfo() - { - Path = "/media/sub.ass", - Protocol = MediaProtocol.File, - Format = "ass", - IsExternal = true - }); + new MediaSourceInfo() + { + Protocol = MediaProtocol.Http + }, + new MediaStream() + { + Path = "/media/sub.ass", + IsExternal = true + }, + new SubtitleEncoder.SubtitleInfo() + { + Path = "/media/sub.ass", + Protocol = MediaProtocol.File, + Format = "ass", + IsExternal = true + } + } + }; return data; } @@ -113,6 +142,55 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests Assert.Equal(subtitleInfo.IsExternal, result.IsExternal); } + public static TheoryData GetSubtitleStream_NonUtf8LocalFile_TestData() + { + return + [ + // Greek legacy encodings – the exact scenario reported in issue #17267. + Encoding.GetEncoding("windows-1253"), + Encoding.GetEncoding("iso-8859-7"), + // Wide encoding with a BOM. + new UnicodeEncoding(bigEndian: false, byteOrderMark: true), + ]; + } + + [Theory] + [MemberData(nameof(GetSubtitleStream_NonUtf8LocalFile_TestData))] + public async Task GetSubtitleStream_NonUtf8LocalFile_ConvertedToUtf8(Encoding sourceEncoding) + { + var cancellationToken = TestContext.Current.CancellationToken; + var srt = BuildGreekSrt(); + var path = Path.GetTempFileName(); + try + { + await File.WriteAllTextAsync(path, srt, sourceEncoding, cancellationToken); + + var fixture = new Fixture().Customize(new AutoMoqCustomization { ConfigureMembers = true }); + var subtitleEncoder = fixture.Create(); + + var fileInfo = new SubtitleEncoder.SubtitleInfo + { + Path = path, + Protocol = MediaProtocol.File, + Format = "srt", + IsExternal = true + }; + + using var stream = await subtitleEncoder.GetSubtitleStream(fileInfo, cancellationToken); + using var reader = new StreamReader(stream, new UTF8Encoding(false)); + var text = await reader.ReadToEndAsync(cancellationToken); + + // The Greek text must survive round-trip and contain no replacement characters. + Assert.Contains(GreekText, text, StringComparison.Ordinal); + Assert.DoesNotContain('�', text); + Assert.DoesNotContain('?', text); + } + finally + { + File.Delete(path); + } + } + [Fact] public void ConvertSubtitles_SequentialCalls_AreDeterministic() { @@ -129,6 +207,44 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests } } + [Fact] + public async Task GetSubtitleStream_Utf8LocalFile_PreservesContent() + { + var cancellationToken = TestContext.Current.CancellationToken; + var srt = BuildGreekSrt(); + var path = Path.GetTempFileName(); + try + { + await File.WriteAllTextAsync(path, srt, new UTF8Encoding(false), cancellationToken); + + var fixture = new Fixture().Customize(new AutoMoqCustomization { ConfigureMembers = true }); + var subtitleEncoder = fixture.Create(); + + var fileInfo = new SubtitleEncoder.SubtitleInfo + { + Path = path, + Protocol = MediaProtocol.File, + Format = "srt", + IsExternal = true + }; + + using var stream = await subtitleEncoder.GetSubtitleStream(fileInfo, cancellationToken); + + // An already-UTF-8 file must be short-circuited and served directly from disk, + // not read into memory and re-encoded (which would produce a MemoryStream). + Assert.IsNotType(stream); + + using var reader = new StreamReader(stream, new UTF8Encoding(false)); + var text = await reader.ReadToEndAsync(cancellationToken); + + Assert.Contains(GreekText, text, StringComparison.Ordinal); + } + finally + { + File.Delete(path); + } + } + [Fact] public async Task ConvertSubtitles_ConcurrentCalls_MatchSequentialBaseline() { -- cgit v1.2.3