tv-anarchy/Sources/TVAnarchyCore/Metadata/FilenameParser.swift

import Foundation

/// Regex-first filename → structured fields. This is the deterministic core of
/// the metadata pipeline (handles the overwhelming majority of releases); an
/// MLX `TitleRefiner` is consulted only for the messy tail (see `refiner`).
public enum FilenameParser {

    /// Optional model-backed refiner for titles regex can't cleanly extract.
    /// nil by default — the regex path stands alone. MLX plugs in here.
    public static var refiner: (any TitleRefiner)?

    public static func parse(path: String) -> ParsedFilename {
        let base = (((path as NSString).lastPathComponent) as NSString).deletingPathExtension
        return parse(filename: base)
    }

    public static func parse(filename: String) -> ParsedFilename {
        let s = filename
        let se = LibraryScanner.parseSxxEyy(s)
        // Episode releases give S+E; season packs give only "S01" / "Season 6".
        let season = se?.0
            ?? captureInt(s, "\\bS(\\d{1,2})(?![0-9])")
            ?? captureInt(s, "\\bSeason\\s*(\\d+)\\b")
        let episode = se?.1
        let year = firstInt(s, "\\b(19|20)\\d{2}\\b")
        let quality = firstGroup(s, "\\b(2160p|1080p|720p|480p)\\b")
        let codec = firstGroup(s, "\\b(x ?265|x ?264|h\\.?265|h\\.?264|hevc|xvid|divx|av1)\\b")
        let releaseSource = firstGroup(s, "\\b(blu-?ray|web-?dl|web-?rip|hdtv|dvdrip|brrip|bdrip|remux|hdrip)\\b")

        var title = extractTitle(s)
        if title.count < 2, let r = refiner, let refined = r.refineTitle(from: s), !refined.isEmpty {
            title = refined
        }

        return ParsedFilename(title: title, year: year, season: season, episode: episode,
                              quality: quality.map(normalizeQuality), codec: codec,
                              releaseSource: releaseSource)
    }

    /// Title = the text before the earliest "noise" marker (SxxEyy, year,
    /// quality, "Season N"), with separators tidied. Mirrors the library
    /// normalization but keeps the year out of the title.
    private static func extractTitle(_ s: String) -> String {
        let markers = [
            "S\\d{1,2}E\\d{1,3}",
            "\\bS\\d{1,2}\\b",
            "\\bSeason\\s*\\d+\\b",
            "\\b(19|20)\\d{2}\\b",
            "\\b(2160p|1080p|720p|480p)\\b",
        ]
        var cut = s.count
        for m in markers {
            if let r = s.range(of: m, options: [.regularExpression, .caseInsensitive]) {
                cut = min(cut, s.distance(from: s.startIndex, to: r.lowerBound))
            }
        }
        let head = String(s.prefix(cut))
        var title = head.replacingOccurrences(of: "[._-]+", with: " ", options: .regularExpression)
        title = title.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
        title = title.trimmingCharacters(in: .whitespaces)
        return title.isEmpty ? s : title
    }

    private static func normalizeQuality(_ q: String) -> String { q.lowercased() }

    private static func firstGroup(_ s: String, _ pattern: String) -> String? {
        guard let r = s.range(of: pattern, options: [.regularExpression, .caseInsensitive]) else { return nil }
        return String(s[r])
    }

    private static func firstInt(_ s: String, _ pattern: String) -> Int? {
        firstGroup(s, pattern).flatMap { Int($0) }
    }

    /// Capture group 1 of the first match, as Int (e.g. "S01" → 1).
    private static func captureInt(_ s: String, _ pattern: String) -> Int? {
        guard let re = try? NSRegularExpression(pattern: pattern, options: [.caseInsensitive]) else { return nil }
        let range = NSRange(s.startIndex..., in: s)
        guard let m = re.firstMatch(in: s, range: range),
              let r = Range(m.range(at: 1), in: s) else { return nil }
        return Int(s[r])
    }
}

/// Seam for a model-backed title cleaner (MLX). Implementations refine a title
/// from a messy filename when the regex path produced nothing useful.
public protocol TitleRefiner: Sendable {
    func refineTitle(from filename: String) -> String?
}