Futonizer/Services/SubtitleMatcher.cs

using System.IO;
using System.Text.RegularExpressions;

namespace Futonizer.Services;

/// <summary>
/// Recognises subtitle files by extension and extracts episode information
/// from a filename — either a normalized season+episode code (e.g.
/// "S04E06") or, failing that, a bare episode number (e.g. "06", "6",
/// "16") — used to match dropped subtitle files to the video file for the
/// same episode.
/// </summary>
public static class SubtitleMatcher
{
    /// <summary>
    /// Every subtitle container/format extension mkvmerge (and therefore
    /// Futonizer) can mux in: SubRip, (Advanced) SubStation Alpha, VobSub,
    /// WebVTT, HDMV/PGS, Universal Subtitle Format, SAMI, TTML/DFXP, EBU-STL,
    /// MicroDVD/SubViewer (both commonly saved with a plain ".sub" extension),
    /// and Kate.
    /// </summary>
    public static readonly string[] SupportedExtensions =
    {
        ".srt", ".ass", ".ssa", ".sub", ".idx", ".vtt", ".webvtt",
        ".sup", ".pgs", ".usf", ".smi", ".sami", ".ttml", ".dfxp",
        ".stl", ".kate",
    };

    public static bool IsSubtitleFile(string path)
    {
        string ext = Path.GetExtension(path);
        return !string.IsNullOrEmpty(ext)
               && SupportedExtensions.Contains(ext, StringComparer.OrdinalIgnoreCase);
    }

    // "S04E06", "S4E6", "S04 E06", "S04-E06", "S04.E06" — the standard scene/anime tagging scheme.
    private static readonly Regex SeasonEpisodeRegex =
        new(@"S(\d{1,4})[\s._-]*E(\d{1,4})", RegexOptions.IgnoreCase | RegexOptions.Compiled);

    // Fallback: "4x06" style season x episode separator.
    private static readonly Regex SeasonXEpisodeRegex =
        new(@"(?<![\d.])(\d{1,2})x(\d{2,3})(?!\d)", RegexOptions.IgnoreCase | RegexOptions.Compiled);

    // Explicit episode marker with no season prefix: "E06", "EP06", "Episode 06".
    // Requires a non-alphanumeric character (or start of string) right before
    // the marker so it doesn't fire on the "e" inside an unrelated word.
    private static readonly Regex ExplicitEpisodeMarkerRegex =
        new(@"(?:^|[^A-Za-z0-9])(?:episode|ep|e)[\s._-]*(\d{1,3})(?!\d)", RegexOptions.IgnoreCase | RegexOptions.Compiled);

    // Whole filename stem is just a number, e.g. "06", "6", "16".
    private static readonly Regex BareWholeNumberRegex =
        new(@"^\d{1,4}$", RegexOptions.Compiled);

    // A number cleanly delimited at the very start of the stem by a
    // separator, e.g. "06 - Title", "06_Title", "06.Title".
    private static readonly Regex BareLeadingNumberRegex =
        new(@"^(\d{1,3})(?=[\s._-])", RegexOptions.Compiled);

    // A number delimited by a " - " (or "-") release-style dash separator
    // that isn't at the very end of the stem, e.g. "Show - 06 - Title",
    // "[Group] Show Name - 06 [1080p]". Requires whitespace immediately
    // before the dash so it doesn't fire on glued tokens like "x264-06".
    private static readonly Regex DashDelimitedNumberRegex =
        new(@"(?<=\s-\s)(\d{1,3})(?!\d)|(?<=\s-)(\d{1,3})(?!\d)", RegexOptions.Compiled);

    // A number cleanly delimited at the very end of the stem by a
    // separator, e.g. "Show - 06", "Show_06", "Show.06". A "." only counts
    // as a separator here when it isn't itself preceded by a digit, so
    // decimal-looking tokens like audio channel counts ("DD5.1") aren't
    // mistaken for an episode number.
    private static readonly Regex BareTrailingNumberRegex =
        new(@"(?<=[\s_-]|(?<!\d)\.)(\d{1,3})$", RegexOptions.Compiled);

    /// <summary>
    /// Extracts and normalizes the episode code from a filename (e.g.
    /// "Show.S4E6.1080p.mkv" and "Show - 04x06 - Title.srt" both return
    /// "S04E06"), so files can be matched regardless of zero-padding or
    /// separator style. Returns null if no season+episode code is found —
    /// use <see cref="ExtractEpisodeNumber"/> for the season-agnostic
    /// fallback (plain episode numbers like "06").
    /// </summary>
    public static string? ExtractEpisodeCode(string fileName)
    {
        var match = SeasonEpisodeRegex.Match(fileName);
        if (!match.Success)
            match = SeasonXEpisodeRegex.Match(fileName);

        if (!match.Success) return null;

        if (!int.TryParse(match.Groups[1].Value, out int season)) return null;
        if (!int.TryParse(match.Groups[2].Value, out int episode)) return null;

        return $"S{season:D2}E{episode:D2}";
    }

    /// <summary>
    /// Extracts just the episode number from a filename, ignoring season.
    /// Tries, in order: a full season+episode code (e.g. "S04E06" → 6), an
    /// explicit marker with no season ("E06", "EP06", "Episode 06" → 6), or
    /// a bare number that is either the entire filename stem ("06", "6",
    /// "16") or cleanly delimited from the rest of the name by a separator
    /// ("06 - Title", "Show - 06"). Bare numbers are capped at a few digits
    /// so resolutions (1080p), years, and bitrates aren't mistaken for an
    /// episode number. Returns null if nothing usable is found.
    /// </summary>
    public static int? ExtractEpisodeNumber(string fileName)
    {
        string stem = Path.GetFileNameWithoutExtension(fileName);

        var seasonMatch = SeasonEpisodeRegex.Match(stem);
        if (!seasonMatch.Success)
            seasonMatch = SeasonXEpisodeRegex.Match(stem);
        if (seasonMatch.Success && int.TryParse(seasonMatch.Groups[2].Value, out int fromSeasonCode))
            return fromSeasonCode;

        var markerMatch = ExplicitEpisodeMarkerRegex.Match(stem);
        if (markerMatch.Success && int.TryParse(markerMatch.Groups[1].Value, out int fromMarker))
            return fromMarker;

        if (BareWholeNumberRegex.IsMatch(stem) && int.TryParse(stem, out int wholeStem))
            return wholeStem;

        var leadingMatch = BareLeadingNumberRegex.Match(stem);
        if (leadingMatch.Success && int.TryParse(leadingMatch.Groups[1].Value, out int fromLeading))
            return fromLeading;

        var dashMatch = DashDelimitedNumberRegex.Match(stem);
        if (dashMatch.Success)
        {
            string dashValue = dashMatch.Groups[1].Success ? dashMatch.Groups[1].Value : dashMatch.Groups[2].Value;
            if (int.TryParse(dashValue, out int fromDash))
                return fromDash;
        }

        var trailingMatch = BareTrailingNumberRegex.Match(stem);
        if (trailingMatch.Success && int.TryParse(trailingMatch.Groups[1].Value, out int fromTrailing))
            return fromTrailing;

        return null;
    }
}