Improve series name matching

Add a series path resolver that attempts to extract only the series
name from a path that contains more information that just the name.

4 files changed, 158 insertions, 0 deletions

diff --git a/Emby.Naming/TV/SeriesInfo.cs b/Emby.Naming/TV/SeriesInfo.cs
new file mode 100644
index 000000000..5d6cb4bd3
--- /dev/null
+++ b/Emby.Naming/TV/SeriesInfo.cs
@@ -0,0 +1,29 @@
+namespace Emby.Naming.TV
+{
+    /// <summary>
+    /// Holder object for Series information.
+    /// </summary>
+    public class SeriesInfo
+    {
+        /// <summary>
+        /// Initializes a new instance of the <see cref="SeriesInfo"/> class.
+        /// </summary>
+        /// <param name="path">Path to the file.</param>
+        public SeriesInfo(string path)
+        {
+            Path = path;
+        }
+
+        /// <summary>
+        /// Gets or sets the path.
+        /// </summary>
+        /// <value>The path.</value>
+        public string Path { get; set; }
+
+        /// <summary>
+        /// Gets or sets the name of the series.
+        /// </summary>
+        /// <value>The name of the series.</value>
+        public string? Name { get; set; }
+    }
+}
diff --git a/Emby.Naming/TV/SeriesPathParser.cs b/Emby.Naming/TV/SeriesPathParser.cs
new file mode 100644
index 000000000..a62e5f4d6
--- /dev/null
+++ b/Emby.Naming/TV/SeriesPathParser.cs
@@ -0,0 +1,61 @@
+using System.Globalization;
+using Emby.Naming.Common;
+
+namespace Emby.Naming.TV
+{
+    /// <summary>
+    /// Used to parse information about series from paths containing more information that only the series name.
+    /// Uses the same regular expressions as the EpisodePathParser but have different success criteria.
+    /// </summary>
+    public static class SeriesPathParser
+    {
+        /// <summary>
+        /// Parses information about series from path.
+        /// </summary>
+        /// <param name="options"><see cref="NamingOptions"/> object containing EpisodeExpressions and MultipleEpisodeExpressions.</param>
+        /// <param name="path">Path.</param>
+        /// <returns>Returns <see cref="SeriesPathParserResult"/> object.</returns>
+        public static SeriesPathParserResult Parse(NamingOptions options, string path)
+        {
+            SeriesPathParserResult? result = null;
+
+            foreach (var expression in options.EpisodeExpressions)
+            {
+                var currentResult = Parse(path, expression);
+                if (currentResult.Success)
+                {
+                    result = currentResult;
+                    break;
+                }
+            }
+
+            if (result != null)
+            {
+                if (!string.IsNullOrEmpty(result.SeriesName))
+                {
+                    result.SeriesName = result.SeriesName.Trim(' ', '_', '.', '-');
+                }
+            }
+
+            return result ?? new SeriesPathParserResult();
+        }
+
+        private static SeriesPathParserResult Parse(string name, EpisodeExpression expression)
+        {
+            var result = new SeriesPathParserResult();
+
+            var match = expression.Regex.Match(name);
+
+            if (match.Success && match.Groups.Count >= 3)
+            {
+                if (expression.IsNamed)
+                {
+                    result.SeriesName = match.Groups["seriesname"].Value;
+                    result.Success = !string.IsNullOrEmpty(result.SeriesName) && !string.IsNullOrEmpty(match.Groups["seasonnumber"]?.Value);
+                }
+            }
+
+            return result;
+        }
+    }
+}
diff --git a/Emby.Naming/TV/SeriesPathParserResult.cs b/Emby.Naming/TV/SeriesPathParserResult.cs
new file mode 100644
index 000000000..44cd2fdfa
--- /dev/null
+++ b/Emby.Naming/TV/SeriesPathParserResult.cs
@@ -0,0 +1,19 @@
+namespace Emby.Naming.TV
+{
+    /// <summary>
+    /// Holder object for <see cref="SeriesPathParser"/> result.
+    /// </summary>
+    public class SeriesPathParserResult
+    {
+        /// <summary>
+        /// Gets or sets the name of the series.
+        /// </summary>
+        /// <value>The name of the series.</value>
+        public string? SeriesName { get; set; }
+
+        /// <summary>
+        /// Gets or sets a value indicating whether parsing was successful.
+        /// </summary>
+        public bool Success { get; set; }
+    }
+}
diff --git a/Emby.Naming/TV/SeriesResolver.cs b/Emby.Naming/TV/SeriesResolver.cs
new file mode 100644
index 000000000..156a03c9e
--- /dev/null
+++ b/Emby.Naming/TV/SeriesResolver.cs
@@ -0,0 +1,49 @@
+using System.IO;
+using System.Text.RegularExpressions;
+using Emby.Naming.Common;
+
+namespace Emby.Naming.TV
+{
+    /// <summary>
+    /// Used to resolve information about series from path.
+    /// </summary>
+    public static class SeriesResolver
+    {
+        /// <summary>
+        /// Regex that matches strings of at least 2 characters separated by a dot or underscore.
+        /// Used for removing separators between words, i.e turns "The_show" into "The show" while
+        /// preserving namings like "S.H.O.W".
+        /// </summary>
+        private static readonly Regex _seriesNameRegex = new Regex(@"((?<a>[^\._]{2,})[\._]*)|([\._](?<b>[^\._]{2,}))");
+
+        /// <summary>
+        /// Resolve information about series from path.
+        /// </summary>
+        /// <param name="options"><see cref="NamingOptions"/> object passed to <see cref="SeriesPathParser"/>.</param>
+        /// <param name="path">Path to series.</param>
+        /// <returns>SeriesInfo.</returns>
+        public static SeriesInfo Resolve(NamingOptions options, string path)
+        {
+            string seriesName = Path.GetFileName(path);
+
+            SeriesPathParserResult result = SeriesPathParser.Parse(options, path);
+            if (result.Success)
+            {
+                if (!string.IsNullOrEmpty(result.SeriesName))
+                {
+                    seriesName = result.SeriesName;
+                }
+            }
+
+            if (!string.IsNullOrEmpty(seriesName))
+            {
+                seriesName = _seriesNameRegex.Replace(seriesName, "${a} ${b}").Trim();
+            }
+
+            return new SeriesInfo(path)
+            {
+                Name = seriesName
+            };
+        }
+    }
+}