Fix author-name collisions and series-set release searches

src/Chaptarr.Core.Test/DecisionEngine/ReleaseTitleMatchSpecificationFixture.cs

@@ -14,6 +14,29 @@ namespace Chaptarr.Core.Test.DecisionEngine
     [TestFixture]
     public class ReleaseTitleMatchSpecificationFixture
     {
+        [TestCase("Isaac Asimov Galactic Empire Series 2 Books Set EPUB")]
+        [TestCase("Isaac Asimov Galactic Empire Series Book 1 EPUB")]
+        [TestCase("Isaac Asimov Galactic Empire Series Pebble in the Sky EPUB")]
+        public void should_reject_partial_series_sets_with_default_balanced_matching(string releaseTitle)
+        {
+            var author = new Author { Name = "Isaac Asimov" };
+            var book = new Book { Title = "Galactic Empire Series 3 Books Set", Author = author };
+            var criteria = new BookSearchCriteria
+            {
+                Author = author,
+                Books = new List<Book> { book },
+                InteractiveSearch = true
+            };
+            var remoteBook = new RemoteBook
+            {
+                Release = new ReleaseInfo { Title = releaseTitle, Author = author.Name }
+            };
+
+            var spec = new ReleaseTitleMatchSpecification(LogManager.GetCurrentClassLogger());
+
+            Assert.That(spec.IsSatisfiedBy(remoteBook, criteria).Accepted, Is.False);
+        }
+
         [Test]
         public void should_accept_interactive_search_result_when_release_title_omits_author_but_release_author_hint_matches()
         {

src/Chaptarr.Core.Test/Indexers/ReleaseSearchServiceTitleSelectionFixture.cs

@@ -9,6 +9,24 @@ namespace Chaptarr.Core.Test.Indexers
     [TestFixture]
     public class ReleaseSearchServiceTitleSelectionFixture
     {
+        [TestCase("Galactic Empire Series 3 Books Set", "Galactic+Empire")]
+        [TestCase("Galactic Empire 3-Book Box Set", "Galactic+Empire")]
+        [TestCase("Galactic Empire 3 Books Set: Pebble in the Sky and more", "Galactic+Empire")]
+        [TestCase("The Three-Body Problem", "Three+Body+Problem")]
+        [TestCase("Foundation Book 3", "Foundation+Book+3")]
+        [TestCase("3 Books Set", "3+Books+Set")]
+        public void book_query_should_remove_only_trailing_box_set_packaging(string title, string query)
+        {
+            var criteria = new BookSearchCriteria
+            {
+                Author = new Author { Name = "Isaac Asimov" },
+                BookTitle = title
+            };
+
+            Assert.That(criteria.BookQuery, Is.EqualTo(query));
+            Assert.That(criteria.BookTitle, Is.EqualTo(title));
+        }
+
         [Test]
         public void should_use_selected_edition_title_when_any_edition_ok()
         {

src/Chaptarr.Core.Test/Parser/ReleaseTitleMatchScorerFixture.cs

@@ -10,6 +10,60 @@ namespace Chaptarr.Core.Test.Parser
     [TestFixture]
     public class ReleaseTitleMatchScorerFixture
     {
+        [TestCase("Galactic Empire series by Isaac Asimov EPUB", true)]
+        [TestCase("Isaac Asimov Galactic Empire Series M4B", true)]
+        [TestCase("Isaac Asimov Galactic Empire Series 3 Books Set EPUB", true)]
+        [TestCase("Isaac Asimov Galactic Empire Series 2 Books Set EPUB", false)]
+        [TestCase("Isaac Asimov Galactic Empire Series Book 1 EPUB", false)]
+        [TestCase("Isaac Asimov Galactic Empire EPUB", false)]
+        [TestCase("Isaac Asimov Galactic Empire Series Pebble in the Sky EPUB", false)]
+        public void should_match_series_set_without_retail_book_count_but_reject_explicit_partial_sets(string releaseTitle, bool accepted)
+        {
+            var author = new Author { Name = "Isaac Asimov" };
+            var target = new Book { Id = 1, Title = "Galactic Empire Series 3 Books Set", Author = author };
+            var namesake = new Book { Id = 2, Title = "Isaac Asimov", Author = author };
+
+            var result = ReleaseTitleMatchScorer.FindBestMatch(
+                releaseTitle, author.Name, new[] { target }, null, new[] { target, namesake });
+
+            Assert.That(result?.IsMatch == true, Is.EqualTo(accepted));
+        }
+
+        [TestCase("The Naked Sun", "Isaac Asimov The Naked Sun 2011 RETAiL EPUB eBook-NODE")]
+        [TestCase("Foundation", "Isaac Asimov Foundation EPUB")]
+        [TestCase("Foundation", "Foundation Isaac Asimov M4B")]
+        public void should_not_treat_author_credit_as_a_conflicting_book_title(string title, string releaseTitle)
+        {
+            var author = new Author { Name = "Isaac Asimov" };
+            var target = new Book { Id = 1, Title = title, Author = author };
+            var namesake = new Book { Id = 2, Title = "Isaac Asimov", Author = author };
+
+            var result = ReleaseTitleMatchScorer.FindBestMatch(
+                releaseTitle, author.Name, new[] { target }, null, new[] { target, namesake });
+
+            Assert.That(result, Is.Not.Null);
+            Assert.That(result.IsMatch, Is.True);
+            Assert.That(result.Problems, Is.Empty);
+        }
+
+        [Test]
+        public void should_still_reject_a_different_book_when_catalog_contains_author_namesake()
+        {
+            var author = new Author { Name = "Isaac Asimov" };
+            var target = new Book { Id = 1, Title = "Foundation", Author = author };
+            var namesake = new Book { Id = 2, Title = "Isaac Asimov", Author = author };
+            var sequel = new Book { Id = 3, Title = "Foundation and Empire", Author = author };
+
+            var result = ReleaseTitleMatchScorer.FindBestMatch(
+                "Isaac Asimov Foundation and Empire EPUB", author.Name, new[] { target }, null,
+                new[] { target, namesake, sequel });
+
+            Assert.That(result, Is.Not.Null);
+            Assert.That(result.IsMatch, Is.False);
+            Assert.That(result.ProblemCode, Is.EqualTo(TitleMatchProblemCode.SiblingTitleContradiction));
+            Assert.That(result.MeaningfulLeftovers, Does.Contain("Foundation and Empire"));
+        }
+
         [Test]
         public void should_return_source_spans_for_backend_title_tokens()
         {

src/NzbDrone.Core/IndexerSearch/Definitions/BookSearchCriteria.cs

@@ -1,11 +1,14 @@
 using System;
 using System.Collections.Generic;
+using System.Text.RegularExpressions;
 using NzbDrone.Core.Parser;
 
 namespace NzbDrone.Core.IndexerSearch.Definitions
 {
     public class BookSearchCriteria : SearchCriteriaBase
     {
+        private static readonly Regex BoxSetSuffix = new Regex(@"\s+(?:series\s+)?\d+[\s-]+books?\s+(?:box(?:ed)?\s+)?set\s*$", RegexOptions.Compiled | RegexOptions.IgnoreCase);
+
         public string BookTitle { get; set; }
         public int BookYear { get; set; }
         public string BookIsbn { get; set; }
@@ -26,7 +29,11 @@ internal static string GetMainSearchTitle(string title, string author)
             }
 
             var mainTitle = titleWithoutAuthor.SplitBookTitle(author).Item1;
-            return string.IsNullOrWhiteSpace(mainTitle) ? titleWithoutAuthor : mainTitle;
+            var searchTitle = string.IsNullOrWhiteSpace(mainTitle) ? titleWithoutAuthor : mainTitle;
+
+            // Retail packaging terms rarely appear in indexer titles. Broaden only the
+            // query; identity and pack checks still use the original selected title.
+            return BoxSetSuffix.Replace(searchTitle, string.Empty);
         }
 
         internal static string RemoveLeadingAuthorPrefix(string title, string author)

src/NzbDrone.Core/Parser/ReleaseTitleMatchScorer.cs

@@ -57,6 +57,7 @@ internal sealed class BookTitleMatchContext
         public string PrimaryTitle { get; set; }
         public string SeriesName { get; set; }
         public string SeriesPosition { get; set; }
+        public string SeriesSetVariant { get; set; }
         public List<string> PrimaryVariants { get; } = new List<string>();
         public HashSet<string> PrefixAllowanceTokens { get; } = new HashSet<string>(StringComparer.OrdinalIgnoreCase);
     }
@@ -80,6 +81,7 @@ public static class ReleaseTitleMatchScorer
         private static readonly Regex SubtitleArticleInsertionPointRegex = new Regex(@"(?<prefix>[:;\-\u2013\u2014]\s+)(?<head>[\p{L}\p{Nd}])", RegexOptions.Compiled | RegexOptions.IgnoreCase);
         private static readonly Regex LeadingOptionalArticleRegex = new Regex(@"^(?:a|an|the)\s+", RegexOptions.Compiled | RegexOptions.IgnoreCase);
         private static readonly Regex WhitespaceRegex = new Regex(@"\s+", RegexOptions.Compiled);
+        private static readonly Regex SeriesSetTitleRegex = new Regex(@"^(?<series>.+\s+series)\s+\d+[\s-]+books?\s+(?:box(?:ed)?\s+)?set$", RegexOptions.Compiled | RegexOptions.IgnoreCase);
         private static readonly Regex SpaceBeforePunctuationRegex = new Regex(@"\s+([:;,])", RegexOptions.Compiled);
         private static readonly Regex SpaceAfterPunctuationRegex = new Regex(@"([:;,])(?=\S)", RegexOptions.Compiled);
         private static readonly Regex YearTokenRegex = new Regex(@"^(?:18\d{2}|19\d{2}|20\d{2}|21\d{2})$", RegexOptions.Compiled);
@@ -233,6 +235,12 @@ private static TitleMatchResult ScoreAgainstTitleVariant(IReadOnlyList<string> r
 
             foreach (var span in spans)
             {
+                if (TokenizedEquals(bookTitleVariant, context?.SeriesSetVariant) &&
+                    HasSeriesSetExtras(releaseTokens, span.Start, span.End, book?.Author?.Name))
+                {
+                    continue;
+                }
+
                 var problems = GetProblems(releaseTokens, span.Start, span.End, titleTokens.Count, hasAuthorInTitle, book?.Author?.Name, context, contradictoryVariants);
                 var leftovers = problems.Select(problem => problem.Value).ToList();
 
@@ -256,6 +264,14 @@ private static TitleMatchResult ScoreAgainstTitleVariant(IReadOnlyList<string> r
             return best;
         }
 
+        private static bool HasSeriesSetExtras(IReadOnlyList<string> releaseTokens, int matchedStart, int matchedEnd, string authorName)
+        {
+            var authorTokens = Tokenize(authorName);
+            return releaseTokens.Where((_, index) => index < matchedStart || index > matchedEnd).Any(token =>
+                (IsNumericToken(token) && !YearTokenRegex.IsMatch(token)) ||
+                (!IsMetadataToken(token) && !authorTokens.Any(authorToken => TokensMatch(authorToken, token))));
+        }
+
         private static bool IsLongAuthorlessYearTitleMatch(IReadOnlyList<string> releaseTokens, int matchedStart, int matchedEnd, IReadOnlyCollection<string> titleTokens, IReadOnlyCollection<TitleMatchProblem> problems)
         {
             var yearIndex = matchedEnd + 1;
@@ -317,7 +333,10 @@ private static List<TitleMatchProblem> GetProblems(IReadOnlyList<string> release
                         continue;
                     }
 
-                    if (IsTargetSeriesContext(contradiction.Title, context))
+                    // A catalogue can contain a book named after its author. The author
+                    // credit in a release is not evidence that it contains that book.
+                    if (TokenizedEquals(contradiction.Title, authorName) ||
+                        IsTargetSeriesContext(contradiction.Title, context))
                     {
                         continue;
                     }
@@ -886,6 +905,16 @@ internal static BookTitleMatchContext GetBookTitleMatchContext(Book book)
             AddTitleVariants(context.PrimaryVariants, primaryTitle);
             context.PrimaryTitle = primaryTitle;
 
+            // Keep "series" in this variant so a bare title or an individual volume
+            // cannot stand in for the set. Explicit counts and extra titles are checked
+            // before scoring this variant, including in relaxed matching modes.
+            var seriesSet = SeriesSetTitleRegex.Match(primaryTitle);
+            if (seriesSet.Success)
+            {
+                context.SeriesSetVariant = seriesSet.Groups["series"].Value;
+                AddTitleVariants(context.PrimaryVariants, context.SeriesSetVariant);
+            }
+
             foreach (var variant in context.PrimaryVariants)
             {
                 AddTokens(context.PrefixAllowanceTokens, variant);