From ed8fce2dced5aa1e3db5426e2bdd0aaf756d0e6b Mon Sep 17 00:00:00 2001 From: Bond_009 Date: Fri, 8 Jan 2021 23:03:02 +0100 Subject: Use SubtitleEdit to parse subtitles --- MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs | 101 ++-------------------- 1 file changed, 6 insertions(+), 95 deletions(-) (limited to 'MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs') diff --git a/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs b/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs index ccef7eeeae..7a7196f6c3 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs @@ -1,102 +1,13 @@ -#pragma warning disable CS1591 +#nullable enable -using System; -using System.Collections.Generic; -using System.Globalization; -using System.IO; -using System.Text.RegularExpressions; -using System.Threading; -using MediaBrowser.Model.MediaInfo; -using Microsoft.Extensions.Logging; +using Nikse.SubtitleEdit.Core.SubtitleFormats; namespace MediaBrowser.MediaEncoding.Subtitles { - public class SrtParser : ISubtitleParser + /// + /// SubRip subtitle parser. + /// + public class SrtParser : SubtitleEditParser { - private readonly ILogger _logger; - - private readonly CultureInfo _usCulture = new CultureInfo("en-US"); - - public SrtParser(ILogger logger) - { - _logger = logger; - } - - /// - public SubtitleTrackInfo Parse(Stream stream, CancellationToken cancellationToken) - { - var trackInfo = new SubtitleTrackInfo(); - var trackEvents = new List(); - using (var reader = new StreamReader(stream)) - { - string line; - while ((line = reader.ReadLine()) != null) - { - cancellationToken.ThrowIfCancellationRequested(); - - if (string.IsNullOrWhiteSpace(line)) - { - continue; - } - - var subEvent = new SubtitleTrackEvent { Id = line }; - line = reader.ReadLine(); - - if (string.IsNullOrWhiteSpace(line)) - { - continue; - } - - var time = Regex.Split(line, @"[\t ]*-->[\t ]*"); - - if (time.Length < 2) - { - // This occurs when subtitle text has an empty line as part of the text. - // Need to adjust the break statement below to resolve this. - _logger.LogWarning("Unrecognized line in srt: {0}", line); - continue; - } - - subEvent.StartPositionTicks = GetTicks(time[0]); - var endTime = time[1].AsSpan(); - var idx = endTime.IndexOf(' '); - if (idx > 0) - { - endTime = endTime.Slice(0, idx); - } - - subEvent.EndPositionTicks = GetTicks(endTime); - var multiline = new List(); - while ((line = reader.ReadLine()) != null) - { - if (line.Length == 0) - { - break; - } - - multiline.Add(line); - } - - subEvent.Text = string.Join(ParserValues.NewLine, multiline); - subEvent.Text = subEvent.Text.Replace(@"\N", ParserValues.NewLine, StringComparison.OrdinalIgnoreCase); - subEvent.Text = Regex.Replace(subEvent.Text, @"\{(?:\\[0-9]?[\w.-]+(?:\([^\)]*\)|&H?[0-9A-Fa-f]+&|))+\}", string.Empty, RegexOptions.IgnoreCase); - subEvent.Text = Regex.Replace(subEvent.Text, "<", "<", RegexOptions.IgnoreCase); - subEvent.Text = Regex.Replace(subEvent.Text, ">", ">", RegexOptions.IgnoreCase); - subEvent.Text = Regex.Replace(subEvent.Text, "<(\\/?(font|b|u|i|s))((\\s+(\\w|\\w[\\w\\-]*\\w)(\\s*=\\s*(?:\\\".*?\\\"|'.*?'|[^'\\\">\\s]+))?)+\\s*|\\s*)(\\/?)>", "<$1$3$7>", RegexOptions.IgnoreCase); - trackEvents.Add(subEvent); - } - } - - trackInfo.TrackEvents = trackEvents; - return trackInfo; - } - - private long GetTicks(ReadOnlySpan time) - { - return TimeSpan.TryParseExact(time, @"hh\:mm\:ss\.fff", _usCulture, out var span) - ? span.Ticks - : (TimeSpan.TryParseExact(time, @"hh\:mm\:ss\,fff", _usCulture, out span) - ? span.Ticks : 0); - } } } -- cgit v1.2.3 From be965e35b680de9362a09eeee2879e8867911656 Mon Sep 17 00:00:00 2001 From: Bond_009 Date: Fri, 8 Jan 2021 23:21:50 +0100 Subject: Log subtitle errors --- MediaBrowser.MediaEncoding/Subtitles/AssParser.cs | 8 ++++++++ MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs | 8 ++++++++ MediaBrowser.MediaEncoding/Subtitles/SsaParser.cs | 8 ++++++++ .../Subtitles/SubtitleEditParser.cs | 22 ++++++++++++++++++++-- .../Subtitles/SubtitleEncoder.cs | 6 +++--- .../Subtitles/AssParserTests.cs | 3 ++- .../Subtitles/SrtParserTests.cs | 3 ++- .../Subtitles/SsaParserTests.cs | 3 ++- 8 files changed, 53 insertions(+), 8 deletions(-) (limited to 'MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs') diff --git a/MediaBrowser.MediaEncoding/Subtitles/AssParser.cs b/MediaBrowser.MediaEncoding/Subtitles/AssParser.cs index cdfd87bad0..8219aa7b47 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/AssParser.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/AssParser.cs @@ -1,5 +1,6 @@ #nullable enable +using Microsoft.Extensions.Logging; using Nikse.SubtitleEdit.Core.SubtitleFormats; namespace MediaBrowser.MediaEncoding.Subtitles @@ -9,5 +10,12 @@ namespace MediaBrowser.MediaEncoding.Subtitles /// public class AssParser : SubtitleEditParser { + /// + /// Initializes a new instance of the class. + /// + /// The logger. + public AssParser(ILogger logger) : base(logger) + { + } } } diff --git a/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs b/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs index 7a7196f6c3..19fb951dc5 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SrtParser.cs @@ -1,5 +1,6 @@ #nullable enable +using Microsoft.Extensions.Logging; using Nikse.SubtitleEdit.Core.SubtitleFormats; namespace MediaBrowser.MediaEncoding.Subtitles @@ -9,5 +10,12 @@ namespace MediaBrowser.MediaEncoding.Subtitles /// public class SrtParser : SubtitleEditParser { + /// + /// Initializes a new instance of the class. + /// + /// The logger. + public SrtParser(ILogger logger) : base(logger) + { + } } } diff --git a/MediaBrowser.MediaEncoding/Subtitles/SsaParser.cs b/MediaBrowser.MediaEncoding/Subtitles/SsaParser.cs index 8cd06194d9..36dc2e01fc 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SsaParser.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SsaParser.cs @@ -1,5 +1,6 @@ #nullable enable +using Microsoft.Extensions.Logging; using Nikse.SubtitleEdit.Core.SubtitleFormats; namespace MediaBrowser.MediaEncoding.Subtitles @@ -9,5 +10,12 @@ namespace MediaBrowser.MediaEncoding.Subtitles /// public class SsaParser : SubtitleEditParser { + /// + /// Initializes a new instance of the class. + /// + /// The logger. + public SsaParser(ILogger logger) : base(logger) + { + } } } diff --git a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEditParser.cs b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEditParser.cs index 441b9ce332..82ec6ca217 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEditParser.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEditParser.cs @@ -6,17 +6,31 @@ using System.Linq; using System.Threading; using MediaBrowser.Common.Extensions; using MediaBrowser.Model.MediaInfo; +using Microsoft.Extensions.Logging; using Nikse.SubtitleEdit.Core; +using ILogger = Microsoft.Extensions.Logging.ILogger; +using SubtitleFormat = Nikse.SubtitleEdit.Core.SubtitleFormats.SubtitleFormat; namespace MediaBrowser.MediaEncoding.Subtitles { /// /// SubStation Alpha subtitle parser. /// - /// The . + /// The . public abstract class SubtitleEditParser : ISubtitleParser - where T : Nikse.SubtitleEdit.Core.SubtitleFormats.SubtitleFormat, new() + where T : SubtitleFormat, new() { + private readonly ILogger _logger; + + /// + /// Initializes a new instance of the class. + /// + /// The logger. + protected SubtitleEditParser(ILogger logger) + { + _logger = logger; + } + /// public SubtitleTrackInfo Parse(Stream stream, CancellationToken cancellationToken) { @@ -24,6 +38,10 @@ namespace MediaBrowser.MediaEncoding.Subtitles var subRip = new T(); var lines = stream.ReadAllLines().ToList(); subRip.LoadSubtitle(subtitle, lines, "untitled"); + if (subRip.ErrorCount > 0) + { + _logger.LogError("{ErrorCount} errors encountered while parsing subtitle."); + } var trackInfo = new SubtitleTrackInfo(); int len = subtitle.Paragraphs.Count; diff --git a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs index 62ac83c91b..a9d118ef58 100644 --- a/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs +++ b/MediaBrowser.MediaEncoding/Subtitles/SubtitleEncoder.cs @@ -271,17 +271,17 @@ namespace MediaBrowser.MediaEncoding.Subtitles if (string.Equals(format, SubtitleFormat.SRT, StringComparison.OrdinalIgnoreCase)) { - return new SrtParser(); + return new SrtParser(_logger); } if (string.Equals(format, SubtitleFormat.SSA, StringComparison.OrdinalIgnoreCase)) { - return new SsaParser(); + return new SsaParser(_logger); } if (string.Equals(format, SubtitleFormat.ASS, StringComparison.OrdinalIgnoreCase)) { - return new AssParser(); + return new AssParser(_logger); } if (throwIfMissing) diff --git a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/AssParserTests.cs b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/AssParserTests.cs index 457ef45446..f38fc23d5e 100644 --- a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/AssParserTests.cs +++ b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/AssParserTests.cs @@ -3,6 +3,7 @@ using System.Globalization; using System.IO; using System.Threading; using MediaBrowser.MediaEncoding.Subtitles; +using Microsoft.Extensions.Logging.Abstractions; using Xunit; namespace Jellyfin.MediaEncoding.Subtitles.Tests @@ -14,7 +15,7 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests { using (var stream = File.OpenRead("Test Data/example.ass")) { - var parsed = new AssParser().Parse(stream, CancellationToken.None); + var parsed = new AssParser(new NullLogger()).Parse(stream, CancellationToken.None); Assert.Single(parsed.TrackEvents); var trackEvent = parsed.TrackEvents[0]; diff --git a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SrtParserTests.cs b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SrtParserTests.cs index 6bfc426cb1..25b8b47461 100644 --- a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SrtParserTests.cs +++ b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SrtParserTests.cs @@ -3,6 +3,7 @@ using System.Globalization; using System.IO; using System.Threading; using MediaBrowser.MediaEncoding.Subtitles; +using Microsoft.Extensions.Logging.Abstractions; using Xunit; namespace Jellyfin.MediaEncoding.Subtitles.Tests @@ -14,7 +15,7 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests { using (var stream = File.OpenRead("Test Data/example.srt")) { - var parsed = new SrtParser().Parse(stream, CancellationToken.None); + var parsed = new SrtParser(new NullLogger()).Parse(stream, CancellationToken.None); Assert.Equal(2, parsed.TrackEvents.Count); var trackEvent1 = parsed.TrackEvents[0]; diff --git a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SsaParserTests.cs b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SsaParserTests.cs index 660b7f1be6..0ff7cb1ec3 100644 --- a/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SsaParserTests.cs +++ b/tests/Jellyfin.MediaEncoding.Tests/Subtitles/SsaParserTests.cs @@ -3,6 +3,7 @@ using System.Globalization; using System.IO; using System.Threading; using MediaBrowser.MediaEncoding.Subtitles; +using Microsoft.Extensions.Logging.Abstractions; using Xunit; namespace Jellyfin.MediaEncoding.Subtitles.Tests @@ -14,7 +15,7 @@ namespace Jellyfin.MediaEncoding.Subtitles.Tests { using (var stream = File.OpenRead("Test Data/example.ssa")) { - var parsed = new SsaParser().Parse(stream, CancellationToken.None); + var parsed = new SsaParser(new NullLogger()).Parse(stream, CancellationToken.None); Assert.Single(parsed.TrackEvents); var trackEvent = parsed.TrackEvents[0]; -- cgit v1.2.3