jellyfin-server/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs

769 lines
29 KiB
C#

#pragma warning disable CS1591
using System;
using System.Collections.Concurrent;
using System.Diagnostics;
using System.Diagnostics.CodeAnalysis;
using System.Globalization;
using System.IO;
using System.Linq;
using System.Net.Http;
using System.Text;
using System.Threading;
using System.Threading.Tasks;
using MediaBrowser.Common;
using MediaBrowser.Common.Configuration;
using MediaBrowser.Common.Extensions;
using MediaBrowser.Common.Net;
using MediaBrowser.Controller.Entities;
using MediaBrowser.Controller.Library;
using MediaBrowser.Controller.MediaEncoding;
using MediaBrowser.Model.Dto;
using MediaBrowser.Model.Entities;
using MediaBrowser.Model.IO;
using MediaBrowser.Model.MediaInfo;
using Microsoft.Extensions.Logging;
using UtfUnknown;
namespace MediaBrowser.MediaEncoding.Subtitles
{
public sealed class SubtitleEncoder : ISubtitleEncoder
{
private readonly ILogger<SubtitleEncoder> _logger;
private readonly IApplicationPaths _appPaths;
private readonly IFileSystem _fileSystem;
private readonly IMediaEncoder _mediaEncoder;
private readonly IHttpClientFactory _httpClientFactory;
private readonly IMediaSourceManager _mediaSourceManager;
private readonly ISubtitleParser _subtitleParser;
/// <summary>
/// The _semaphoreLocks.
/// </summary>
private readonly ConcurrentDictionary<string, SemaphoreSlim> _semaphoreLocks =
new ConcurrentDictionary<string, SemaphoreSlim>();
public SubtitleEncoder(
ILogger<SubtitleEncoder> logger,
IApplicationPaths appPaths,
IFileSystem fileSystem,
IMediaEncoder mediaEncoder,
IHttpClientFactory httpClientFactory,
IMediaSourceManager mediaSourceManager,
ISubtitleParser subtitleParser)
{
_logger = logger;
_appPaths = appPaths;
_fileSystem = fileSystem;
_mediaEncoder = mediaEncoder;
_httpClientFactory = httpClientFactory;
_mediaSourceManager = mediaSourceManager;
_subtitleParser = subtitleParser;
}
private string SubtitleCachePath => Path.Combine(_appPaths.DataPath, "subtitles");
private Stream ConvertSubtitles(
Stream stream,
string inputFormat,
string outputFormat,
long startTimeTicks,
long endTimeTicks,
bool preserveOriginalTimestamps,
CancellationToken cancellationToken)
{
var ms = new MemoryStream();
try
{
var trackInfo = _subtitleParser.Parse(stream, inputFormat);
FilterEvents(trackInfo, startTimeTicks, endTimeTicks, preserveOriginalTimestamps);
var writer = GetWriter(outputFormat);
writer.Write(trackInfo, ms, cancellationToken);
ms.Position = 0;
}
catch
{
ms.Dispose();
throw;
}
return ms;
}
private void FilterEvents(SubtitleTrackInfo track, long startPositionTicks, long endTimeTicks, bool preserveTimestamps)
{
// Drop subs that are earlier than what we're looking for
track.TrackEvents = track.TrackEvents
.SkipWhile(i => (i.StartPositionTicks - startPositionTicks) < 0 || (i.EndPositionTicks - startPositionTicks) < 0)
.ToArray();
if (endTimeTicks > 0)
{
track.TrackEvents = track.TrackEvents
.TakeWhile(i => i.StartPositionTicks <= endTimeTicks)
.ToArray();
}
if (!preserveTimestamps)
{
foreach (var trackEvent in track.TrackEvents)
{
trackEvent.EndPositionTicks -= startPositionTicks;
trackEvent.StartPositionTicks -= startPositionTicks;
}
}
}
async Task<Stream> ISubtitleEncoder.GetSubtitles(BaseItem item, string mediaSourceId, int subtitleStreamIndex, string outputFormat, long startTimeTicks, long endTimeTicks, bool preserveOriginalTimestamps, CancellationToken cancellationToken)
{
if (item == null)
{
throw new ArgumentNullException(nameof(item));
}
if (string.IsNullOrWhiteSpace(mediaSourceId))
{
throw new ArgumentNullException(nameof(mediaSourceId));
}
var mediaSources = await _mediaSourceManager.GetPlaybackMediaSources(item, null, true, false, cancellationToken).ConfigureAwait(false);
var mediaSource = mediaSources
.First(i => string.Equals(i.Id, mediaSourceId, StringComparison.OrdinalIgnoreCase));
var subtitleStream = mediaSource.MediaStreams
.First(i => i.Type == MediaStreamType.Subtitle && i.Index == subtitleStreamIndex);
var subtitle = await GetSubtitleStream(mediaSource, subtitleStream, cancellationToken)
.ConfigureAwait(false);
var inputFormat = subtitle.Format;
// Return the original if the same format is being requested
// Character encoding was already handled in GetSubtitleStream
if (string.Equals(inputFormat, outputFormat, StringComparison.OrdinalIgnoreCase))
{
return subtitle.Stream;
}
using (var stream = subtitle.Stream)
{
return ConvertSubtitles(stream, inputFormat, outputFormat, startTimeTicks, endTimeTicks, preserveOriginalTimestamps, cancellationToken);
}
}
private async Task<(Stream Stream, string Format)> GetSubtitleStream(
MediaSourceInfo mediaSource,
MediaStream subtitleStream,
CancellationToken cancellationToken)
{
var fileInfo = await GetReadableFile(mediaSource, subtitleStream, cancellationToken).ConfigureAwait(false);
var stream = await GetSubtitleStream(fileInfo, cancellationToken).ConfigureAwait(false);
return (stream, fileInfo.Format);
}
private async Task<Stream> GetSubtitleStream(SubtitleInfo fileInfo, CancellationToken cancellationToken)
{
if (fileInfo.IsExternal)
{
using (var stream = await GetStream(fileInfo.Path, fileInfo.Protocol, cancellationToken).ConfigureAwait(false))
{
var result = CharsetDetector.DetectFromStream(stream).Detected;
stream.Position = 0;
if (result != null)
{
_logger.LogDebug("charset {CharSet} detected for {Path}", result.EncodingName, fileInfo.Path);
using var reader = new StreamReader(stream, result.Encoding);
var text = await reader.ReadToEndAsync().ConfigureAwait(false);
return new MemoryStream(Encoding.UTF8.GetBytes(text));
}
}
}
return AsyncFile.OpenRead(fileInfo.Path);
}
internal async Task<SubtitleInfo> GetReadableFile(
MediaSourceInfo mediaSource,
MediaStream subtitleStream,
CancellationToken cancellationToken)
{
if (!subtitleStream.IsExternal || subtitleStream.Path.EndsWith(".mks", StringComparison.OrdinalIgnoreCase))
{
string outputFormat;
string outputCodec;
if (string.Equals(subtitleStream.Codec, "ass", StringComparison.OrdinalIgnoreCase)
|| string.Equals(subtitleStream.Codec, "ssa", StringComparison.OrdinalIgnoreCase)
|| string.Equals(subtitleStream.Codec, "srt", StringComparison.OrdinalIgnoreCase))
{
// Extract
outputCodec = "copy";
outputFormat = subtitleStream.Codec;
}
else if (string.Equals(subtitleStream.Codec, "subrip", StringComparison.OrdinalIgnoreCase))
{
// Extract
outputCodec = "copy";
outputFormat = "srt";
}
else
{
// Extract
outputCodec = "srt";
outputFormat = "srt";
}
// Extract
var outputPath = GetSubtitleCachePath(mediaSource, subtitleStream.Index, "." + outputFormat);
await ExtractTextSubtitle(mediaSource, subtitleStream, outputCodec, outputPath, cancellationToken)
.ConfigureAwait(false);
return new SubtitleInfo(outputPath, MediaProtocol.File, outputFormat, false);
}
var currentFormat = (Path.GetExtension(subtitleStream.Path) ?? subtitleStream.Codec)
.TrimStart('.');
// Fallback to ffmpeg conversion
if (!_subtitleParser.SupportsFileExtension(currentFormat))
{
// Convert
var outputPath = GetSubtitleCachePath(mediaSource, subtitleStream.Index, ".srt");
await ConvertTextSubtitleToSrt(subtitleStream, mediaSource, outputPath, cancellationToken).ConfigureAwait(false);
return new SubtitleInfo(outputPath, MediaProtocol.File, "srt", true);
}
// It's possible that the subtitleStream and mediaSource don't share the same protocol (e.g. .STRM file with local subs)
return new SubtitleInfo(subtitleStream.Path, _mediaSourceManager.GetPathProtocol(subtitleStream.Path), currentFormat, true);
}
private bool TryGetWriter(string format, [NotNullWhen(true)] out ISubtitleWriter? value)
{
if (string.Equals(format, SubtitleFormat.ASS, StringComparison.OrdinalIgnoreCase))
{
value = new AssWriter();
return true;
}
if (string.IsNullOrEmpty(format))
{
throw new ArgumentNullException(nameof(format));
}
if (string.Equals(format, "json", StringComparison.OrdinalIgnoreCase))
{
value = new JsonWriter();
return true;
}
if (string.Equals(format, SubtitleFormat.SRT, StringComparison.OrdinalIgnoreCase) || string.Equals(format, SubtitleFormat.SUBRIP, StringComparison.OrdinalIgnoreCase))
{
value = new SrtWriter();
return true;
}
if (string.Equals(format, SubtitleFormat.SSA, StringComparison.OrdinalIgnoreCase))
{
value = new SsaWriter();
return true;
}
if (string.Equals(format, SubtitleFormat.VTT, StringComparison.OrdinalIgnoreCase))
{
value = new VttWriter();
return true;
}
if (string.Equals(format, SubtitleFormat.TTML, StringComparison.OrdinalIgnoreCase))
{
value = new TtmlWriter();
return true;
}
value = null;
return false;
}
private ISubtitleWriter GetWriter(string format)
{
if (TryGetWriter(format, out var writer))
{
return writer;
}
throw new ArgumentException("Unsupported format: " + format);
}
/// <summary>
/// Gets the lock.
/// </summary>
/// <param name="filename">The filename.</param>
/// <returns>System.Object.</returns>
private SemaphoreSlim GetLock(string filename)
{
return _semaphoreLocks.GetOrAdd(filename, _ => new SemaphoreSlim(1, 1));
}
/// <summary>
/// Converts the text subtitle to SRT.
/// </summary>
/// <param name="subtitleStream">The subtitle stream.</param>
/// <param name="mediaSource">The input mediaSource.</param>
/// <param name="outputPath">The output path.</param>
/// <param name="cancellationToken">The cancellation token.</param>
/// <returns>Task.</returns>
private async Task ConvertTextSubtitleToSrt(MediaStream subtitleStream, MediaSourceInfo mediaSource, string outputPath, CancellationToken cancellationToken)
{
var semaphore = GetLock(outputPath);
await semaphore.WaitAsync(cancellationToken).ConfigureAwait(false);
try
{
if (!File.Exists(outputPath))
{
await ConvertTextSubtitleToSrtInternal(subtitleStream, mediaSource, outputPath, cancellationToken).ConfigureAwait(false);
}
}
finally
{
semaphore.Release();
}
}
/// <summary>
/// Converts the text subtitle to SRT internal.
/// </summary>
/// <param name="subtitleStream">The subtitle stream.</param>
/// <param name="mediaSource">The input mediaSource.</param>
/// <param name="outputPath">The output path.</param>
/// <param name="cancellationToken">The cancellation token.</param>
/// <returns>Task.</returns>
/// <exception cref="ArgumentNullException">
/// The <c>inputPath</c> or <c>outputPath</c> is <c>null</c>.
/// </exception>
private async Task ConvertTextSubtitleToSrtInternal(MediaStream subtitleStream, MediaSourceInfo mediaSource, string outputPath, CancellationToken cancellationToken)
{
var inputPath = subtitleStream.Path;
if (string.IsNullOrEmpty(inputPath))
{
throw new ArgumentNullException(nameof(inputPath));
}
if (string.IsNullOrEmpty(outputPath))
{
throw new ArgumentNullException(nameof(outputPath));
}
Directory.CreateDirectory(Path.GetDirectoryName(outputPath) ?? throw new ArgumentException($"Provided path ({outputPath}) is not valid.", nameof(outputPath)));
var encodingParam = await GetSubtitleFileCharacterSet(subtitleStream, subtitleStream.Language, mediaSource, cancellationToken).ConfigureAwait(false);
// FFmpeg automatically convert character encoding when it is UTF-16
// If we specify character encoding, it rejects with "do not specify a character encoding" and "Unable to recode subtitle event"
if ((inputPath.EndsWith(".smi", StringComparison.Ordinal) || inputPath.EndsWith(".sami", StringComparison.Ordinal)) &&
(encodingParam.Equals("UTF-16BE", StringComparison.OrdinalIgnoreCase) ||
encodingParam.Equals("UTF-16LE", StringComparison.OrdinalIgnoreCase)))
{
encodingParam = string.Empty;
}
else if (!string.IsNullOrEmpty(encodingParam))
{
encodingParam = " -sub_charenc " + encodingParam;
}
int exitCode;
using (var process = new Process
{
StartInfo = new ProcessStartInfo
{
CreateNoWindow = true,
UseShellExecute = false,
FileName = _mediaEncoder.EncoderPath,
Arguments = string.Format(CultureInfo.InvariantCulture, "{0} -i \"{1}\" -c:s srt \"{2}\"", encodingParam, inputPath, outputPath),
WindowStyle = ProcessWindowStyle.Hidden,
ErrorDialog = false
},
EnableRaisingEvents = true
})
{
_logger.LogInformation("{0} {1}", process.StartInfo.FileName, process.StartInfo.Arguments);
try
{
process.Start();
}
catch (Exception ex)
{
_logger.LogError(ex, "Error starting ffmpeg");
throw;
}
var ranToCompletion = await process.WaitForExitAsync(TimeSpan.FromMinutes(30)).ConfigureAwait(false);
if (!ranToCompletion)
{
try
{
_logger.LogInformation("Killing ffmpeg subtitle conversion process");
process.Kill();
}
catch (Exception ex)
{
_logger.LogError(ex, "Error killing subtitle conversion process");
}
}
exitCode = ranToCompletion ? process.ExitCode : -1;
}
var failed = false;
if (exitCode == -1)
{
failed = true;
if (File.Exists(outputPath))
{
try
{
_logger.LogInformation("Deleting converted subtitle due to failure: ", outputPath);
_fileSystem.DeleteFile(outputPath);
}
catch (IOException ex)
{
_logger.LogError(ex, "Error deleting converted subtitle {Path}", outputPath);
}
}
}
else if (!File.Exists(outputPath))
{
failed = true;
}
if (failed)
{
_logger.LogError("ffmpeg subtitle conversion failed for {Path}", inputPath);
throw new FfmpegException(
string.Format(CultureInfo.InvariantCulture, "ffmpeg subtitle conversion failed for {0}", inputPath));
}
await SetAssFont(outputPath, cancellationToken).ConfigureAwait(false);
_logger.LogInformation("ffmpeg subtitle conversion succeeded for {Path}", inputPath);
}
/// <summary>
/// Extracts the text subtitle.
/// </summary>
/// <param name="mediaSource">The mediaSource.</param>
/// <param name="subtitleStream">The subtitle stream.</param>
/// <param name="outputCodec">The output codec.</param>
/// <param name="outputPath">The output path.</param>
/// <param name="cancellationToken">The cancellation token.</param>
/// <returns>Task.</returns>
/// <exception cref="ArgumentException">Must use inputPath list overload.</exception>
private async Task ExtractTextSubtitle(
MediaSourceInfo mediaSource,
MediaStream subtitleStream,
string outputCodec,
string outputPath,
CancellationToken cancellationToken)
{
var semaphore = GetLock(outputPath);
await semaphore.WaitAsync(cancellationToken).ConfigureAwait(false);
var subtitleStreamIndex = EncodingHelper.FindIndex(mediaSource.MediaStreams, subtitleStream);
try
{
if (!File.Exists(outputPath))
{
var args = _mediaEncoder.GetInputArgument(mediaSource.Path, mediaSource);
if (subtitleStream.IsExternal)
{
args = _mediaEncoder.GetExternalSubtitleInputArgument(subtitleStream.Path);
}
await ExtractTextSubtitleInternal(
args,
subtitleStreamIndex,
outputCodec,
outputPath,
cancellationToken).ConfigureAwait(false);
}
}
finally
{
semaphore.Release();
}
}
private async Task ExtractTextSubtitleInternal(
string inputPath,
int subtitleStreamIndex,
string outputCodec,
string outputPath,
CancellationToken cancellationToken)
{
if (string.IsNullOrEmpty(inputPath))
{
throw new ArgumentNullException(nameof(inputPath));
}
if (string.IsNullOrEmpty(outputPath))
{
throw new ArgumentNullException(nameof(outputPath));
}
Directory.CreateDirectory(Path.GetDirectoryName(outputPath) ?? throw new ArgumentException($"Provided path ({outputPath}) is not valid.", nameof(outputPath)));
var processArgs = string.Format(
CultureInfo.InvariantCulture,
"-i {0} -copyts -map 0:{1} -an -vn -c:s {2} \"{3}\"",
inputPath,
subtitleStreamIndex,
outputCodec,
outputPath);
int exitCode;
using (var process = new Process
{
StartInfo = new ProcessStartInfo
{
CreateNoWindow = true,
UseShellExecute = false,
FileName = _mediaEncoder.EncoderPath,
Arguments = processArgs,
WindowStyle = ProcessWindowStyle.Hidden,
ErrorDialog = false
},
EnableRaisingEvents = true
})
{
_logger.LogInformation("{File} {Arguments}", process.StartInfo.FileName, process.StartInfo.Arguments);
try
{
process.Start();
}
catch (Exception ex)
{
_logger.LogError(ex, "Error starting ffmpeg");
throw;
}
var ranToCompletion = await process.WaitForExitAsync(TimeSpan.FromMinutes(30)).ConfigureAwait(false);
if (!ranToCompletion)
{
try
{
_logger.LogWarning("Killing ffmpeg subtitle extraction process");
process.Kill();
}
catch (Exception ex)
{
_logger.LogError(ex, "Error killing subtitle extraction process");
}
}
exitCode = ranToCompletion ? process.ExitCode : -1;
}
var failed = false;
if (exitCode == -1)
{
failed = true;
try
{
_logger.LogWarning("Deleting extracted subtitle due to failure: {Path}", outputPath);
_fileSystem.DeleteFile(outputPath);
}
catch (FileNotFoundException)
{
}
catch (IOException ex)
{
_logger.LogError(ex, "Error deleting extracted subtitle {Path}", outputPath);
}
}
else if (!File.Exists(outputPath))
{
failed = true;
}
if (failed)
{
_logger.LogError("ffmpeg subtitle extraction failed for {InputPath} to {OutputPath}", inputPath, outputPath);
throw new FfmpegException(
string.Format(CultureInfo.InvariantCulture, "ffmpeg subtitle extraction failed for {0} to {1}", inputPath, outputPath));
}
else
{
_logger.LogInformation("ffmpeg subtitle extraction completed for {InputPath} to {OutputPath}", inputPath, outputPath);
}
if (string.Equals(outputCodec, "ass", StringComparison.OrdinalIgnoreCase))
{
await SetAssFont(outputPath, cancellationToken).ConfigureAwait(false);
}
}
/// <summary>
/// Sets the ass font.
/// </summary>
/// <param name="file">The file.</param>
/// <param name="cancellationToken">The token to monitor for cancellation requests. The default value is <c>System.Threading.CancellationToken.None</c>.</param>
/// <returns>Task.</returns>
private async Task SetAssFont(string file, CancellationToken cancellationToken = default)
{
_logger.LogInformation("Setting ass font within {File}", file);
string text;
Encoding encoding;
using (var fileStream = AsyncFile.OpenRead(file))
using (var reader = new StreamReader(fileStream, true))
{
encoding = reader.CurrentEncoding;
text = await reader.ReadToEndAsync().ConfigureAwait(false);
}
var newText = text.Replace(",Arial,", ",Arial Unicode MS,", StringComparison.Ordinal);
if (!string.Equals(text, newText, StringComparison.Ordinal))
{
var fileStream = new FileStream(file, FileMode.Create, FileAccess.Write, FileShare.None, IODefaults.FileStreamBufferSize, FileOptions.Asynchronous);
await using (fileStream.ConfigureAwait(false))
{
var writer = new StreamWriter(fileStream, encoding);
await using (writer.ConfigureAwait(false))
{
await writer.WriteAsync(newText.AsMemory(), cancellationToken).ConfigureAwait(false);
}
}
}
}
private string GetSubtitleCachePath(MediaSourceInfo mediaSource, int subtitleStreamIndex, string outputSubtitleExtension)
{
if (mediaSource.Protocol == MediaProtocol.File)
{
var ticksParam = string.Empty;
var date = _fileSystem.GetLastWriteTimeUtc(mediaSource.Path);
ReadOnlySpan<char> filename = (mediaSource.Path + "_" + subtitleStreamIndex.ToString(CultureInfo.InvariantCulture) + "_" + date.Ticks.ToString(CultureInfo.InvariantCulture) + ticksParam).GetMD5() + outputSubtitleExtension;
var prefix = filename.Slice(0, 1);
return Path.Join(SubtitleCachePath, prefix, filename);
}
else
{
ReadOnlySpan<char> filename = (mediaSource.Path + "_" + subtitleStreamIndex.ToString(CultureInfo.InvariantCulture)).GetMD5() + outputSubtitleExtension;
var prefix = filename.Slice(0, 1);
return Path.Join(SubtitleCachePath, prefix, filename);
}
}
/// <inheritdoc />
public async Task<string> GetSubtitleFileCharacterSet(MediaStream subtitleStream, string language, MediaSourceInfo mediaSource, CancellationToken cancellationToken)
{
var subtitleCodec = subtitleStream.Codec;
var path = subtitleStream.Path;
if (path.EndsWith(".mks", StringComparison.OrdinalIgnoreCase))
{
path = GetSubtitleCachePath(mediaSource, subtitleStream.Index, "." + subtitleCodec);
await ExtractTextSubtitle(mediaSource, subtitleStream, subtitleCodec, path, cancellationToken)
.ConfigureAwait(false);
}
using (var stream = await GetStream(path, mediaSource.Protocol, cancellationToken).ConfigureAwait(false))
{
var charset = CharsetDetector.DetectFromStream(stream).Detected?.EncodingName ?? string.Empty;
// UTF16 is automatically converted to UTF8 by FFmpeg, do not specify a character encoding
if ((path.EndsWith(".ass", StringComparison.Ordinal) || path.EndsWith(".ssa", StringComparison.Ordinal) || path.EndsWith(".srt", StringComparison.Ordinal))
&& (string.Equals(charset, "utf-16le", StringComparison.OrdinalIgnoreCase)
|| string.Equals(charset, "utf-16be", StringComparison.OrdinalIgnoreCase)))
{
charset = string.Empty;
}
_logger.LogDebug("charset {0} detected for {Path}", charset, path);
return charset;
}
}
private async Task<Stream> GetStream(string path, MediaProtocol protocol, CancellationToken cancellationToken)
{
switch (protocol)
{
case MediaProtocol.Http:
{
using var response = await _httpClientFactory.CreateClient(NamedClient.Default)
.GetAsync(new Uri(path), cancellationToken)
.ConfigureAwait(false);
return await response.Content.ReadAsStreamAsync(cancellationToken).ConfigureAwait(false);
}
case MediaProtocol.File:
return AsyncFile.OpenRead(path);
default:
throw new ArgumentOutOfRangeException(nameof(protocol));
}
}
internal readonly struct SubtitleInfo
{
public SubtitleInfo(string path, MediaProtocol protocol, string format, bool isExternal)
{
Path = path;
Protocol = protocol;
Format = format;
IsExternal = isExternal;
}
public string Path { get; }
public MediaProtocol Protocol { get; }
public string Format { get; }
public bool IsExternal { get; }
}
}
}