Merge pull request #17087 from dkanada/book-resolver

improve book resolution from filename

Emby.Naming/Book/BookFileNameParser.cs

@@ -1,3 +1,4 @@
+using System;
 using System.Text.RegularExpressions;
 
 namespace Emby.Naming.Book
@@ -5,7 +6,7 @@ namespace Emby.Naming.Book
     /// <summary>
     /// Helper class to retrieve basic metadata from a book filename.
     /// </summary>
-    public static class BookFileNameParser
+    public static partial class BookFileNameParser
     {
         private const string NameMatchGroup = "name";
         private const string IndexMatchGroup = "index";
@@ -15,14 +16,17 @@ namespace Emby.Naming.Book
         private static readonly Regex[] _nameMatches =
         [
             // seriesName (seriesYear) #index (of count) (year) where only seriesName and index are required
-            new Regex(@"^(?<seriesName>.+?)((\s\((?<seriesYear>[0-9]{4})\))?)\s#(?<index>[0-9]+)((\s\(of\s(?<count>[0-9]+)\))?)((\s\((?<year>[0-9]{4})\))?)$"),
-            new Regex(@"^(?<name>.+?)\s\((?<seriesName>.+?),\s#(?<index>[0-9]+)\)((\s\((?<year>[0-9]{4})\))?)$"),
-            new Regex(@"^(?<index>[0-9]+)\s\-\s(?<name>.+?)((\s\((?<year>[0-9]{4})\))?)$"),
+            new Regex(@"^(?<seriesName>.+?)((\s\((?<seriesYear>[0-9]{4})\))?)\s#(?<index>[0-9]+)(?:\.0)?((\s\(of\s(?<count>[0-9]+)\))?)((\s\((?<year>[0-9]{4})\))?)$"),
+            new Regex(@"^(?<name>.+?)\s\((?<seriesName>.+?),\s#(?<index>[0-9]+)\)(?:\.0)?((\s\((?<year>[0-9]{4})\))?)$"),
+            new Regex(@"^(?<index>[0-9]+)(?:\.0)?\s\-\s(?<name>.+?)((\s\((?<year>[0-9]{4})\))?)$"),
             new Regex(@"(?<name>.*)\((?<year>[0-9]{4})\)"),
             // last resort matches the whole string as the name
             new Regex(@"(?<name>.*)")
         ];
 
+        [GeneratedRegex(@"^(?<name>.+?)(\sv(?<volume>[0-9]+))?(\sc(?<chapter>[0-9]+))?$")]
+        private static partial Regex ComicRegex();
+
         /// <summary>
         /// Parse a filename name to retrieve the book name, series name, index, and year.
         /// </summary>
@@ -48,7 +52,22 @@ namespace Emby.Naming.Book
 
                 if (match.Groups.TryGetValue(NameMatchGroup, out Group? nameGroup) && nameGroup.Success)
                 {
-                    result.Name = nameGroup.Value.Trim();
+                    var comicMatch = ComicRegex().Match(nameGroup.Value.Trim());
+
+                    if (comicMatch.Success)
+                    {
+                        if (comicMatch.Groups.TryGetValue("volume", out Group? volumeGroup) && volumeGroup.Success && int.TryParse(volumeGroup.ValueSpan, out var volume))
+                        {
+                            result.ParentIndex = volume;
+                        }
+
+                        if (comicMatch.Groups.TryGetValue("chapter", out Group? chapterGroup) && chapterGroup.Success && int.TryParse(chapterGroup.ValueSpan, out var chapter))
+                        {
+                            result.Index = chapter;
+                        }
+                    }
+
+                    result.Name = nameGroup.ValueSpan.Trim().ToString();
                 }
 
                 if (match.Groups.TryGetValue(IndexMatchGroup, out Group? indexGroup) && indexGroup.Success && int.TryParse(indexGroup.Value, out var index))

Emby.Naming/Book/BookFileNameParserResult.cs

@@ -1,5 +1,3 @@
-using System;
-
 namespace Emby.Naming.Book
 {
     /// <summary>
@@ -14,6 +12,7 @@ namespace Emby.Naming.Book
         {
             Name = null;
             Index = null;
+            ParentIndex = null;
             Year = null;
             SeriesName = null;
         }
@@ -28,6 +27,11 @@ namespace Emby.Naming.Book
         /// </summary>
         public int? Index { get; set; }
 
+        /// <summary>
+        /// Gets or sets the parent index number.
+        /// </summary>
+        public int? ParentIndex { get; set; }
+
         /// <summary>
         /// Gets or sets the publication year.
         /// </summary>

Emby.Server.Implementations/Library/Resolvers/Books/BookResolver.cs

@@ -1,5 +1,3 @@
-#nullable disable
-
 #pragma warning disable CS1591
 
 using System;
@@ -18,7 +16,7 @@ namespace Emby.Server.Implementations.Library.Resolvers.Books
     {
         private readonly string[] _validExtensions = { ".azw", ".azw3", ".cb7", ".cbr", ".cbt", ".cbz", ".epub", ".mobi", ".pdf" };
 
-        protected override Book Resolve(ItemResolveArgs args)
+        protected override Book? Resolve(ItemResolveArgs args)
         {
             var collectionType = args.GetCollectionType();
 
@@ -47,13 +45,14 @@ namespace Emby.Server.Implementations.Library.Resolvers.Books
                 Path = args.Path,
                 Name = result.Name ?? string.Empty,
                 IndexNumber = result.Index,
+                ParentIndexNumber = result.ParentIndex,
                 ProductionYear = result.Year,
                 SeriesName = result.SeriesName ?? Path.GetFileName(Path.GetDirectoryName(args.Path)),
                 IsInMixedFolder = true,
             };
         }
 
-        private Book GetBook(ItemResolveArgs args)
+        private Book? GetBook(ItemResolveArgs args)
         {
             var bookFiles = args.FileSystemChildren.Where(f =>
             {
@@ -78,6 +77,7 @@ namespace Emby.Server.Implementations.Library.Resolvers.Books
                 Path = bookFiles[0].FullName,
                 Name = result.Name ?? string.Empty,
                 IndexNumber = result.Index,
+                ParentIndexNumber = result.ParentIndex,
                 ProductionYear = result.Year,
                 SeriesName = result.SeriesName ?? string.Empty,
             };

tests/Jellyfin.Naming.Tests/Book/BookResolverTests.cs

@@ -0,0 +1,58 @@
+using Emby.Naming.Book;
+using Xunit;
+
+namespace Jellyfin.Naming.Tests.Book;
+
+public class BookResolverTests
+{
+    [Theory]
+    // seriesName (seriesYear?) #index (of count?) (year?)
+    [InlineData("Sherlock Holmes (1887) #1 (of 4) (1887)", null, "Sherlock Holmes", 1, 1887)]
+    [InlineData("Sherlock Holmes #2", null, "Sherlock Holmes", 2, null)]
+    [InlineData("Sherlock Holmes (1887) #1", null, "Sherlock Holmes", 1, null)]
+    [InlineData("Sherlock Holmes #2 (1890)", null, "Sherlock Holmes", 2, 1890)]
+    // name (seriesName, #index) (year?)
+    [InlineData("A Study in Scarlet (Sherlock Holmes, #1) (1887)", "A Study in Scarlet", "Sherlock Holmes", 1, 1887)]
+    [InlineData("The Adventures of Sherlock Holmes (Sherlock Holmes, #5)", "The Adventures of Sherlock Holmes", "Sherlock Holmes", 5, null)]
+    // name (year)
+    [InlineData("The Sign of the Four (1890)", "The Sign of the Four", null, null, 1890)]
+    [InlineData("The Valley of Fear (1915)", "The Valley of Fear", null, null, 1915)]
+    // index - name (year?)
+    [InlineData("2 - The Sign of the Four (1890)", "The Sign of the Four", null, 2, 1890)]
+    [InlineData("4 - The Valley of Fear", "The Valley of Fear", null, 4, null)]
+    // parse entire string as book name
+    [InlineData("A Study in Scarlet", "A Study in Scarlet", null, null, null)]
+    [InlineData("The Adventures of Sherlock Holmes", "The Adventures of Sherlock Holmes", null, null, null)]
+    // leading zeros on index number
+    [InlineData("00 - Dracula's Guest (1914)", "Dracula's Guest", null, 0, 1914)]
+    [InlineData("01 - Dracula (1897)", "Dracula", null, 1, 1897)]
+    // basic decimal support for prequels and novellas
+    [InlineData("2.0 - Twenty Thousand Leagues Under the Sea", "Twenty Thousand Leagues Under the Sea", null, 2, null)]
+    // TODO decide how to process non-zero decimals
+    [InlineData("2.1 - The Blockade Runners", "2.1 - The Blockade Runners", null, null, null)]
+    public void Resolve_Books(string input, string? name, string? series, int? index, int? year)
+    {
+        var result = BookFileNameParser.Parse(input);
+
+        Assert.Equal(name, result.Name);
+        Assert.Equal(series, result.SeriesName);
+        Assert.Equal(index, result.Index);
+        Assert.Equal(year, result.Year);
+    }
+
+    [Theory]
+    // name volume? chapter? (year?)
+    [InlineData("Captain Marvel Adventures v01 (1941)", "Captain Marvel Adventures v01", null, null, 1, 1941)]
+    [InlineData("Captain Marvel Adventures c120", "Captain Marvel Adventures c120", null, 120, null, null)]
+    [InlineData("Captain Marvel Adventures v01 c120", "Captain Marvel Adventures v01 c120", null, 120, 1, null)]
+    public void Resolve_Comics(string input, string? name, string? series, int? chapter, int? volume, int? year)
+    {
+        var result = BookFileNameParser.Parse(input);
+
+        Assert.Equal(name, result.Name);
+        Assert.Equal(series, result.SeriesName);
+        Assert.Equal(chapter, result.Index);
+        Assert.Equal(volume, result.ParentIndex);
+        Assert.Equal(year, result.Year);
+    }
+}