From 5867bb378f5f4078c7f86f71897d1fb193146e51 Mon Sep 17 00:00:00 2001 From: ConfusedPolarBear <33811686+ConfusedPolarBear@users.noreply.github.com> Date: Thu, 25 Aug 2022 22:11:39 -0500 Subject: [PATCH] Implement new search algorithm --- .../TestAudioFingerprinting.cs | 2 +- .../Data/Intro.cs | 7 ++ .../ScheduledTasks/AnalyzeEpisodesTask.cs | 106 +++++++++++------- 3 files changed, 76 insertions(+), 39 deletions(-) diff --git a/ConfusedPolarBear.Plugin.IntroSkipper.Tests/TestAudioFingerprinting.cs b/ConfusedPolarBear.Plugin.IntroSkipper.Tests/TestAudioFingerprinting.cs index c6e2570..dbd4fab 100644 --- a/ConfusedPolarBear.Plugin.IntroSkipper.Tests/TestAudioFingerprinting.cs +++ b/ConfusedPolarBear.Plugin.IntroSkipper.Tests/TestAudioFingerprinting.cs @@ -90,7 +90,7 @@ public class TestAudioFingerprinting var lhsEpisode = queueEpisode("audio/big_buck_bunny_intro.mp3"); var rhsEpisode = queueEpisode("audio/big_buck_bunny_clip.mp3"); - var (lhs, rhs) = task.FingerprintEpisodes(lhsEpisode, rhsEpisode); + var (lhs, rhs) = task.CompareEpisodes(lhsEpisode, rhsEpisode); Assert.True(lhs.Valid); Assert.Equal(0, lhs.IntroStart); diff --git a/ConfusedPolarBear.Plugin.IntroSkipper/Data/Intro.cs b/ConfusedPolarBear.Plugin.IntroSkipper/Data/Intro.cs index 0875b80..4cd88e5 100644 --- a/ConfusedPolarBear.Plugin.IntroSkipper/Data/Intro.cs +++ b/ConfusedPolarBear.Plugin.IntroSkipper/Data/Intro.cs @@ -1,4 +1,5 @@ using System; +using System.Text.Json.Serialization; namespace ConfusedPolarBear.Plugin.IntroSkipper; @@ -49,6 +50,12 @@ public class Intro /// public bool Valid => IntroEnd > 0; + /// + /// Gets the duration of this intro. + /// + [JsonIgnore] + public double Duration => IntroEnd - IntroStart; + /// /// Gets or sets the introduction sequence start time. /// diff --git a/ConfusedPolarBear.Plugin.IntroSkipper/ScheduledTasks/AnalyzeEpisodesTask.cs b/ConfusedPolarBear.Plugin.IntroSkipper/ScheduledTasks/AnalyzeEpisodesTask.cs index c66b5e9..d66cae7 100644 --- a/ConfusedPolarBear.Plugin.IntroSkipper/ScheduledTasks/AnalyzeEpisodesTask.cs +++ b/ConfusedPolarBear.Plugin.IntroSkipper/ScheduledTasks/AnalyzeEpisodesTask.cs @@ -161,7 +161,7 @@ public class AnalyzeEpisodesTask : IScheduledTask { // Increment totalProcessed by the number of episodes in this season that were actually analyzed // (instead of just using the number of episodes in the current season). - var analyzed = AnalyzeSeason(season, cancellationToken); + var analyzed = AnalyzeSeason(season.Value, cancellationToken); Interlocked.Add(ref totalProcessed, analyzed); writeEdl = analyzed > 0 || Plugin.Instance!.Configuration.RegenerateEdlFiles; } @@ -241,66 +241,94 @@ public class AnalyzeEpisodesTask : IScheduledTask return previous; } + // TODO: restore warning +#pragma warning disable CA1002 + /// /// Fingerprints all episodes in the provided season and stores the timestamps of all introductions. /// - /// Pairing of season GUID to a list of QueuedEpisode objects. + /// Episodes in this season. /// Cancellation token provided by the scheduled task. /// Number of episodes from the provided season that were analyzed. private int AnalyzeSeason( - KeyValuePair> season, + List episodes, CancellationToken cancellationToken) { var seasonIntros = new Dictionary(); - var episodes = season.Value; - var first = episodes[0]; /* Don't analyze specials or seasons with an insufficient number of episodes. * A season with only 1 episode can't be analyzed as it would compare the episode to itself, * which would result in the entire episode being marked as an introduction, as the audio is identical. */ - if (season.Value.Count < 2 || first.SeasonNumber == 0) + if (episodes.Count < 2 || episodes[0].SeasonNumber == 0) { return episodes.Count; } + var first = episodes[0]; + _logger.LogInformation( "Analyzing {Count} episodes from {Name} season {Season}", - season.Value.Count, + episodes.Count, first.SeriesName, first.SeasonNumber); - // Ensure there are an even number of episodes - if (episodes.Count % 2 != 0) + // TODO: cache fingerprints and inverted indexes + + // TODO: implementing bucketing + + // For all episodes + foreach (var outer in episodes) { - episodes.Add(episodes[episodes.Count - 2]); + // Compare the outer episode to all other episodes + foreach (var inner in episodes) + { + // Don't compare the episode to itself + if (outer.EpisodeId == inner.EpisodeId) + { + continue; + } + + // Fingerprint both episodes + Intro outerIntro; + Intro innerIntro; + + try + { + (outerIntro, innerIntro) = CompareEpisodes(outer, inner); + } + catch (FingerprintException ex) + { + // TODO: remove the episode that threw the error from additional processing + _logger.LogWarning("Caught fingerprint error: {Ex}", ex); + continue; + } + + if (!outerIntro.Valid) + { + continue; + } + + // Save this intro if: + // - it is the first one we've seen for this episode + // - OR it is longer than the previous one + if ( + !seasonIntros.TryGetValue(outer.EpisodeId, out var currentOuterIntro) || + outerIntro.Duration > currentOuterIntro.Duration) + { + seasonIntros[outer.EpisodeId] = outerIntro; + } + + if ( + !seasonIntros.TryGetValue(inner.EpisodeId, out var currentInnerIntro) || + innerIntro.Duration > currentInnerIntro.Duration) + { + seasonIntros[inner.EpisodeId] = innerIntro; + } + } } - // Analyze each pair of episodes in the current season - for (var i = 0; i < episodes.Count; i += 2) - { - if (cancellationToken.IsCancellationRequested) - { - break; - } - - var lhs = episodes[i]; - var rhs = episodes[i + 1]; - - try - { - _logger.LogTrace("Analyzing {LHS} and {RHS}", lhs.Path, rhs.Path); - - var (lhsIntro, rhsIntro) = FingerprintEpisodes(lhs, rhs); - seasonIntros[lhsIntro.EpisodeId] = lhsIntro; - seasonIntros[rhsIntro.EpisodeId] = rhsIntro; - analysisStatistics.TotalAnalyzedEpisodes.Add(2); - } - catch (FingerprintException ex) - { - _logger.LogError("Caught fingerprint error: {Ex}", ex); - } - } + // TODO: analysisStatistics.TotalAnalyzedEpisodes.Add(2); // Ensure only one thread at a time can update the shared intro dictionary. lock (_introsLock) @@ -319,13 +347,15 @@ public class AnalyzeEpisodesTask : IScheduledTask return episodes.Count; } +#pragma warning restore CA1002 + /// /// Analyze two episodes to find an introduction sequence shared between them. /// /// First episode to analyze. /// Second episode to analyze. /// Intros for the first and second episodes. - public (Intro Lhs, Intro Rhs) FingerprintEpisodes(QueuedEpisode lhsEpisode, QueuedEpisode rhsEpisode) + public (Intro Lhs, Intro Rhs) CompareEpisodes(QueuedEpisode lhsEpisode, QueuedEpisode rhsEpisode) { var start = DateTime.Now; var lhsFingerprint = Chromaprint.Fingerprint(lhsEpisode); @@ -339,7 +369,7 @@ public class AnalyzeEpisodesTask : IScheduledTask _fingerprintCache[rhsEpisode.EpisodeId] = rhsFingerprint; } - return FingerprintEpisodes( + return CompareEpisodes( lhsEpisode.EpisodeId, lhsFingerprint, rhsEpisode.EpisodeId, @@ -356,7 +386,7 @@ public class AnalyzeEpisodesTask : IScheduledTask /// Second episode fingerprint points. /// If this was called as part of the first analysis pass, add the elapsed time to the statistics. /// Intros for the first and second episodes. - public (Intro Lhs, Intro Rhs) FingerprintEpisodes( + public (Intro Lhs, Intro Rhs) CompareEpisodes( Guid lhsId, uint[] lhsPoints, Guid rhsId,