From 8e37aa2e7815523fb18ba9aa34d493eabb9262bf Mon Sep 17 00:00:00 2001
From: ta264 <ta264@users.noreply.github.com>
Date: Sun, 6 Feb 2022 14:20:20 +0000
Subject: [PATCH] Fixed: Take publisher and format into account when choosing
 best edition

---
 .../BookImport/Identification/Distance.cs     |  6 ++--
 .../Identification/DistanceCalculator.cs      | 33 +++++++++++++++++++
 2 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/src/NzbDrone.Core/MediaFiles/BookImport/Identification/Distance.cs b/src/NzbDrone.Core/MediaFiles/BookImport/Identification/Distance.cs
index 246c95d1d..5d1fcda9a 100644
--- a/src/NzbDrone.Core/MediaFiles/BookImport/Identification/Distance.cs
+++ b/src/NzbDrone.Core/MediaFiles/BookImport/Identification/Distance.cs
@@ -18,11 +18,13 @@ public class Distance
             { "asin", 10.0 },
             { "asin_missing", 0.1 },
             { "media_count", 1.0 },
-            { "media_format", 1.0 },
+            { "ebook_format", 0.1 },
+            { "audio_format", 0.1 },
+            { "wrong_format", 5.0 },
             { "year", 1.0 },
             { "country", 0.5 },
             { "language", 5.0 },
-            { "label", 0.5 },
+            { "publisher", 0.5 },
             { "catalog_number", 0.5 },
             { "book_disambiguation", 0.5 },
             { "book_id", 5.0 },
diff --git a/src/NzbDrone.Core/MediaFiles/BookImport/Identification/DistanceCalculator.cs b/src/NzbDrone.Core/MediaFiles/BookImport/Identification/DistanceCalculator.cs
index ed0eaeebc..2ca89825d 100644
--- a/src/NzbDrone.Core/MediaFiles/BookImport/Identification/DistanceCalculator.cs
+++ b/src/NzbDrone.Core/MediaFiles/BookImport/Identification/DistanceCalculator.cs
@@ -22,6 +22,10 @@ public static class DistanceCalculator
 
         private static readonly RegexReplace CleanTitleCruft = new RegexReplace(@"\((?:unabridged)\)", string.Empty, RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
+        private static readonly List<string> EbookFormats = new List<string> { "Kindle Edition", "Nook", "ebook" };
+
+        private static readonly List<string> AudiobookFormats = new List<string> { "Audiobook", "Audio CD", "Audio Cassette", "Audible Audio", "CD-ROM", "MP3 CD" };
+
         public static Distance BookDistance(List<LocalBook> localTracks, Edition edition)
         {
             var dist = new Distance();
@@ -123,6 +127,35 @@ public static Distance BookDistance(List<LocalBook> localTracks, Edition edition
                 Logger.Trace($"language: {localLanguage} vs {editionLanguage}; {dist.NormalizedDistance()}");
             }
 
+            // Publisher - only if set for both the local book and remote edition
+            var localPublisher = localTracks.MostCommon(x => x.FileTrackInfo.Publisher);
+            var editionPublisher = edition.Publisher;
+            if (localPublisher.IsNotNullOrWhiteSpace() && editionPublisher.IsNotNullOrWhiteSpace())
+            {
+                dist.AddString("publisher", localPublisher, editionPublisher);
+                Logger.Trace($"publisher: {localPublisher} vs {editionPublisher}; {dist.NormalizedDistance()}");
+            }
+
+            // try to tilt it towards the correct "type" of release
+            var isAudio = MediaFileExtensions.AudioExtensions.Contains(localTracks.First().Path.GetPathExtension());
+
+            if (edition.Format.IsNotNullOrWhiteSpace())
+            {
+                if (!isAudio)
+                {
+                    // text books should prefer ebook formats
+                    dist.AddBool("ebook_format", !EbookFormats.Contains(edition.Format));
+
+                    // text books should not match audio entries
+                    dist.AddBool("wrong_format", AudiobookFormats.Contains(edition.Format));
+                }
+                else
+                {
+                    // audio books should prefer audio formats
+                    dist.AddBool("audio_format", !AudiobookFormats.Contains(edition.Format));
+                }
+            }
+
             return dist;
         }