Cache inverted indexes

ConfusedPolarBear.Plugin.IntroSkipper.Tests/TestAudioFingerprinting.cs

@@ -2,6 +2,7 @@
  * which supports both chromaprint and the "-fp_format raw" flag.
  */
 
+using System;
 using System.Collections.Generic;
 using Xunit;
 using Microsoft.Extensions.Logging;
@@ -77,7 +78,7 @@ public class TestAudioFingerprinting
             {77, 5},
         };
 
-        var actual = Chromaprint.CreateInvertedIndex(fpr);
+        var actual = Chromaprint.CreateInvertedIndex(Guid.NewGuid(), fpr);
 
         Assert.Equal(expected, actual);
     }
@@ -111,6 +112,7 @@ public class TestAudioFingerprinting
     {
         return new QueuedEpisode()
         {
+            EpisodeId = Guid.NewGuid(),
             Path = "../../../" + path,
             FingerprintDuration = 60
         };

ConfusedPolarBear.Plugin.IntroSkipper/Chromaprint.cs

@@ -20,6 +20,8 @@ public static class Chromaprint
 
     private static Dictionary<string, string> ChromaprintLogs { get; set; } = new();
 
+    private static Dictionary<Guid, Dictionary<uint, int>> InvertedIndexCache { get; set; } = new();
+
     /// <summary>
     /// Check that the installed version of ffmpeg supports chromaprint.
     /// </summary>
@@ -124,10 +126,16 @@ public static class Chromaprint
     /// <summary>
     /// Transforms a Chromaprint into an inverted index of fingerprint points to the last index it appeared at.
     /// </summary>
+    /// <param name="id">Episode ID.</param>
     /// <param name="fingerprint">Chromaprint fingerprint.</param>
     /// <returns>Inverted index.</returns>
-    public static Dictionary<uint, int> CreateInvertedIndex(uint[] fingerprint)
+    public static Dictionary<uint, int> CreateInvertedIndex(Guid id, uint[] fingerprint)
     {
+        if (InvertedIndexCache.TryGetValue(id, out var cached))
+        {
+            return cached;
+        }
+
         var invIndex = new Dictionary<uint, int>();
 
         for (int i = 0; i < fingerprint.Length; i++)
@@ -139,6 +147,8 @@ public static class Chromaprint
             invIndex[point] = i;
         }
 
+        InvertedIndexCache[id] = invIndex;
+
         return invIndex;
     }
 

ConfusedPolarBear.Plugin.IntroSkipper/ScheduledTasks/AnalyzeEpisodesTask.cs

@@ -268,8 +268,6 @@ public class AnalyzeEpisodesTask : IScheduledTask
             }
         }
 
-        // TODO: cache inverted indexes
-
         // TODO: implementing bucketing
 
         // For all episodes
@@ -357,7 +355,7 @@ public class AnalyzeEpisodesTask : IScheduledTask
 
         // Creates an inverted fingerprint point index for both episodes.
         // For every point which is a 100% match, search for an introduction at that point.
-        var (lhsRanges, rhsRanges) = SearchInvertedIndex(lhsPoints, rhsPoints);
+        var (lhsRanges, rhsRanges) = SearchInvertedIndex(lhsId, lhsPoints, rhsId, rhsPoints);
 
         if (lhsRanges.Count > 0)
         {
@@ -417,19 +415,23 @@ public class AnalyzeEpisodesTask : IScheduledTask
     /// <summary>
     /// Search for a shared introduction sequence using inverted indexes.
     /// </summary>
+    /// <param name="lhsId">LHS ID.</param>
     /// <param name="lhsPoints">Left episode fingerprint points.</param>
+    /// <param name="rhsId">RHS ID.</param>
     /// <param name="rhsPoints">Right episode fingerprint points.</param>
     /// <returns>List of shared TimeRanges between the left and right episodes.</returns>
     private (List<TimeRange> Lhs, List<TimeRange> Rhs) SearchInvertedIndex(
+        Guid lhsId,
         uint[] lhsPoints,
+        Guid rhsId,
         uint[] rhsPoints)
     {
         var lhsRanges = new List<TimeRange>();
         var rhsRanges = new List<TimeRange>();
 
         // Generate inverted indexes for the left and right episodes.
-        var lhsIndex = Chromaprint.CreateInvertedIndex(lhsPoints);
+        var lhsIndex = Chromaprint.CreateInvertedIndex(lhsId, lhsPoints);
-        var rhsIndex = Chromaprint.CreateInvertedIndex(rhsPoints);
+        var rhsIndex = Chromaprint.CreateInvertedIndex(rhsId, rhsPoints);
         var indexShifts = new HashSet<int>();
 
         // For all audio points in the left episode, check if the right episode has a point which matches exactly.