tv-anarchy/Sources/TVAnarchyCore/Library/LibraryScanner.swift

import Foundation

/// Direct Swift scan of local media roots — a faithful port of tv-anarchy-mcp's
/// `media/library.ts` (SxxEyy parse, show-root bucketing, release-noise name
/// normalization). The primary library source is black's prebuilt index (see
/// `scanFromIndex`); this local walk only covers a configured, media-tree-structured
/// `MEDIA_ROOTS` dir. It never walks the old NFS `~/media` mount — this project does
/// not depend on NFS. (Loose downloaded files are matched separately by
/// `DownloadsIndex`, by filename.)
public enum LibraryScanner {
    private static let videoExt: Set<String> = ["mkv", "mp4", "m4v", "avi", "mov", "webm"]

    // Compiled once. `S(\d{1,2})E(\d{1,3})`, case-insensitive.
    private static let sxxeyy = try! NSRegularExpression(pattern: "S(\\d{1,2})E(\\d{1,3})",
                                                         options: [.caseInsensitive])

    /// Media-tree-structured roots for the offline library walk (`<root>/<category>/
    /// <show>/…`). Colon-separated `MEDIA_ROOTS`, else none — there is no `~/media`
    /// default. (Distinct from `DownloadsIndex`, which matches loose downloaded files
    /// by name and is NOT tree-structured.)
    public static func mediaRoots() -> [String] {
        if let env = ProcessInfo.processInfo.environment["MEDIA_ROOTS"], !env.isEmpty {
            return env.split(separator: ":").map(String.init).filter { !$0.isEmpty }
        }
        return []
    }

    /// True when at least one local root exists and holds content. (No autofs to
    /// coax now that `~/media` is gone, so a single readdir suffices.) MUST be
    /// called off the main actor — a directory read can block on slow disks.
    public static func rootsAvailable() -> Bool {
        let fm = FileManager.default
        return mediaRoots().contains { (try? fm.contentsOfDirectory(atPath: $0))?.isEmpty == false }
    }

    /// One video found on disk. `season`/`episode` are nil for non-episodic files
    /// (movies/clips); `size` is only populated for those (used to pick the main
    /// file in a movie folder) — episodic files skip the extra stat.
    private struct FoundFile { let path: String; let size: Int64; let season: Int?; let episode: Int?; let mtime: Date? }

    /// `onProgress` (if given) is called periodically with the running count of
    /// directories read — a live, honest progress proxy for the UI (each readdir is
    /// one NFS round-trip; there's no known total to make a determinate %).
    public static func scan(onProgress: ((Int) -> Void)? = nil) -> [CachedShow] {
        let fm = FileManager.default
        var grouped: [String: [FoundFile]] = [:]
        var mediaRootForKey: [String: String] = [:]

        for root in mediaRoots() {
            var isDir: ObjCBool = false
            guard fm.fileExists(atPath: root, isDirectory: &isDir), isDir.boolValue else { continue }
            for f in walkForVideos(rootURL: URL(fileURLWithPath: root, isDirectory: true),
                                   rootPath: root, maxDepth: 4, onProgress: onProgress) {
                let key = showRoot(for: f.path, mediaRoot: root)
                grouped[key, default: []].append(f)
                mediaRootForKey[key] = root
            }
        }

        return group(grouped: grouped, mediaRootForKey: mediaRootForKey)
    }

    /// Build the index TSV produced by black's `build_index.sh` (one
    /// `size⇥mtime-epoch⇥path` line per video, black-side absolute paths) into the
    /// same `CachedShow` list a local walk produces — running the IDENTICAL grouping
    /// rules so there's no second source of truth. Paths are kept black-side (the
    /// canonical identity); at launch a downloaded copy plays on plum's VLC and
    /// anything else routes to black (mpv) — see `PlayerController`. This is the
    /// primary, NFS-free library path: black builds the index out-of-band, plum
    /// just parses it.
    public static func scanFromIndex(_ tsv: String) -> [CachedShow] {
        let root = MediaPaths.remoteRoot
        var grouped: [String: [FoundFile]] = [:]
        var mediaRootForKey: [String: String] = [:]
        for raw in tsv.split(separator: "\n") {
            let cols = raw.split(separator: "\t", maxSplits: 2, omittingEmptySubsequences: false)
            guard cols.count == 3 else { continue }
            let size = Int64(cols[0]) ?? 0
            let mtime = Double(cols[1]).map { Date(timeIntervalSince1970: $0) }
            let path = String(cols[2])
            let name = (path as NSString).lastPathComponent
            let se = parseSxxEyy(name)
            let f = FoundFile(path: path, size: size, season: se?.0, episode: se?.1, mtime: mtime)
            let key = showRoot(for: path, mediaRoot: root)
            grouped[key, default: []].append(f)
            mediaRootForKey[key] = root
        }
        return group(grouped: grouped, mediaRootForKey: mediaRootForKey)
    }

    /// Shared post-gather step: turn grouped FoundFiles into shows (series vs movie),
    /// dedup episodes, capture year + newest-mtime, merge split-season siblings, sort.
    private static func group(grouped: [String: [FoundFile]],
                              mediaRootForKey: [String: String]) -> [CachedShow] {
        var out: [CachedShow] = []
        for (key, files) in grouped {
            let mediaRoot = mediaRootForKey[key] ?? ""
            let comps = componentsAfter(mediaRoot: mediaRoot, path: key)
            let cat = comps.first ?? ""
            // A series only when the key is a real show FOLDER (category/show, depth
            // ≥ 2). Loose files share the category-root key; one stray SxxEyy match
            // among them must NOT flip the whole pile into a single "series" and drop
            // the rest (this silently ate ~940 loose porn files). Those go to
            // movieItems, which makes each loose file its own movie.
            let isShowFolder = comps.count >= 2
            let episodic = files.filter { $0.season != nil }
            if isShowFolder, !episodic.isEmpty {
                let showName = normalizeShowName((key as NSString).lastPathComponent)
                var eps = episodic.map { f in
                    let resolved = LibraryDisplayNames.resolve(showName: showName, path: f.path)
                    return CachedEpisode(path: f.path, season: f.season!, episode: f.episode!,
                                         label: resolved.displayName,
                                         displayName: resolved.displayName,
                                         episodeTitle: resolved.episodeTitle)
                }
                eps.sort(by: episodeOrder)
                // Collapse duplicate episodes across releases (e.g. 1080p + 720p
                // both have S01E06) → one entry per season×episode. The first after
                // sort wins; alternate releases stay reachable via the player's
                // quality switcher.
                var seen = Set<Int>()
                eps = eps.filter { seen.insert($0.season * 1000 + $0.episode).inserted }
                let year = episodic.compactMap { parseYear($0.path) }.min()
                out.append(CachedShow(name: showName,
                                      rootDir: key, category: cat, kind: .series, episodes: eps, year: year,
                                      addedAt: files.compactMap(\.mtime).max()))
            } else {
                out.append(contentsOf: movieItems(key: key, files: files, mediaRoot: mediaRoot, category: cat))
            }
        }
        return mergeSeriesByName(out)
            .sorted { $0.name.localizedCaseInsensitiveCompare($1.name) == .orderedAscending }
    }

    /// Fold freshly-scanned show(s) into an existing library snapshot — union
    /// episodes per `rootDir`, then run the same series-by-name merge as a full scan.
    public static func mergeIngest(_ incoming: [CachedShow], into existing: [CachedShow]) -> [CachedShow] {
        var byRoot = Dictionary(existing.map { ($0.rootDir, $0) }, uniquingKeysWith: { a, _ in a })
        for show in incoming {
            if var prev = byRoot[show.rootDir] {
                prev.episodes.append(contentsOf: show.episodes)
                prev.episodes.sort(by: episodeOrder)
                var seen = Set<Int>()
                prev.episodes = prev.episodes.filter { seen.insert($0.season * 1000 + $0.episode).inserted }
                prev.year = [prev.year, show.year].compactMap { $0 }.min()
                if let added = show.addedAt { prev.addedAt = max(prev.addedAt ?? added, added) }
                byRoot[show.rootDir] = prev
            } else {
                byRoot[show.rootDir] = show
            }
        }
        return mergeSeriesByName(Array(byRoot.values))
            .sorted { $0.name.localizedCaseInsensitiveCompare($1.name) == .orderedAscending }
    }

    /// Merge SERIES that share a normalized name + category but live in separate
    /// top-level folders (e.g. `Bridgerton.S01`, `Bridgerton.S02`, `Bridgerton.S03`
    /// as siblings → one "Bridgerton" with all seasons). Episodes are unioned and
    /// deduped by season×episode; the earliest year and first rootDir win. Movies
    /// are left distinct.
    static func mergeSeriesByName(_ shows: [CachedShow]) -> [CachedShow] {
        var byKey: [String: CachedShow] = [:]
        var order: [String] = []
        var out: [CachedShow] = []
        for show in shows {
            guard show.kind == .series else { out.append(show); continue }
            let key = show.category + "\u{1}" + show.name.lowercased()
            guard var existing = byKey[key] else { byKey[key] = show; order.append(key); continue }
            // Merge ONLY when the season sets are disjoint — one show split across
            // sibling season folders (Bridgerton S01 / S02 / S03). Overlapping
            // seasons (both start at S01) mean two DIFFERENT shows that merely share
            // a name + category (e.g. an anime and its live-action remake) → keep
            // them as separate entries rather than interleaving their episodes.
            let haveSeasons = Set(existing.episodes.map(\.season))
            let newSeasons = Set(show.episodes.map(\.season))
            guard haveSeasons.isDisjoint(with: newSeasons) else { out.append(show); continue }
            existing.episodes += show.episodes
            existing.episodes.sort(by: episodeOrder)
            var seen = Set<Int>()
            existing.episodes = existing.episodes.filter { seen.insert($0.season * 1000 + $0.episode).inserted }
            existing.year = [existing.year, show.year].compactMap { $0 }.min()
            byKey[key] = existing
        }
        out.append(contentsOf: order.compactMap { byKey[$0] })
        return out
    }

    /// Turn a group of non-episodic videos into movie items. A movie *folder*
    /// (key deeper than the category dir) yields one item — the largest non-sample
    /// file. Loose files sitting directly under the category dir each become their
    /// own item.
    private static func movieItems(key: String, files: [FoundFile],
                                   mediaRoot: String, category: String) -> [CachedShow] {
        let real = files.filter { !isSampleOrExtra($0.path) }
        let use = real.isEmpty ? files : real
        let depth = componentsAfter(mediaRoot: mediaRoot, path: key).count

        func movie(path: String, name: String, root: String, addedAt: Date?) -> CachedShow {
            CachedShow(name: normalizeShowName(name), rootDir: root, category: category, kind: .movie,
                       episodes: [CachedEpisode(path: path, season: 0, episode: 0, label: name)],
                       year: parseYear(path) ?? parseYear(root), addedAt: addedAt)
        }

        if depth <= 1 {
            // Loose files at the category root → one movie each (rootDir = file).
            return use.map { f in
                let base = ((f.path as NSString).lastPathComponent as NSString).deletingPathExtension
                return movie(path: f.path, name: base, root: f.path, addedAt: f.mtime)
            }
        }
        // Movie folder → the largest file represents it (rootDir = the folder).
        guard let main = use.max(by: { $0.size < $1.size }) else { return [] }
        return [movie(path: main.path, name: (key as NSString).lastPathComponent, root: key,
                      addedAt: files.compactMap(\.mtime).max())]
    }

    private static func episodeOrder(_ lhs: CachedEpisode, _ rhs: CachedEpisode) -> Bool {
        if lhs.season != rhs.season { return lhs.season < rhs.season }
        return lhs.episode < rhs.episode
    }

    /// Carry forward poster/overview from a prior snapshot onto a fresh scan, then
    /// backfill anything still missing from the `.meta` cache. Keys are normalized
    /// to black-side form so enrichment survives the legacy-plum → black-side path
    /// switch (a raw rootDir match would silently drop every poster). The `.meta`
    /// cache is the durable record: re-folding it here means one bad snapshot can
    /// never lose artwork permanently.
    public static func mergeEnrichment(_ scanned: [CachedShow], from previous: [CachedShow]) -> [CachedShow] {
        let prior = Dictionary(previous.map { (MediaPaths.toRemote($0.rootDir), $0) },
                               uniquingKeysWith: { a, _ in a })
        return scanned.map { show in
            var s = show
            let key = MediaPaths.toRemote(show.rootDir)
            if let old = prior[key] {
                s.posterPath = old.posterPath
                s.overview = old.overview
                // re-attach per-episode metaPath by (normalized) episode path
                let oldMeta = Dictionary(old.episodes.map { (MediaPaths.toRemote($0.path), $0.metaPath) },
                                         uniquingKeysWith: { a, _ in a })
                s.episodes = s.episodes.map { ep in
                    var e = ep
                    if let m = oldMeta[MediaPaths.toRemote(ep.path)] ?? nil { e.metaPath = m }
                    return e
                }
            }
            if s.posterPath == nil || s.overview == nil, let meta = MetaWriter.loadCache(forPath: key) {
                if s.posterPath == nil { s.posterPath = meta.posterURL }
                if s.overview == nil { s.overview = meta.overview }
            }
            return s
        }
    }

    // MARK: - walk

    private static func walkForVideos(rootURL: URL, rootPath: String, maxDepth: Int,
                                      onProgress: ((Int) -> Void)? = nil) -> [FoundFile] {
        let fm = FileManager.default
        var out: [FoundFile] = []
        var dirsVisited = 0
        // Prefetch is-dir + size with the directory read. Over NFS this is the hot
        // path: the old code did a separate `fileExists(isDirectory:)` stat PER
        // entry plus an `attributesOfItem` stat per movie file — two extra round
        // trips each. `contentsOfDirectory(at:includingPropertiesForKeys:)` batches
        // those attributes into the enumeration (readdirplus), and the values are
        // cached on the URL, so `resourceValues` below costs no further syscall.
        //
        // We build emitted paths as STRINGS appended to `rootPath` rather than
        // reading `url.path`, because `contentsOfDirectory(at:)` canonicalizes
        // symlinks/APFS firmlinks (e.g. /var → /private/var) — so `url.path` would
        // no longer be prefixed by the media root the rest of the scan compares
        // against (`componentsAfter`). Appending to `rootPath` keeps the prefix.
        let keys: [URLResourceKey] = [.isDirectoryKey, .fileSizeKey, .contentModificationDateKey]
        let keySet = Set(keys)
        var stack: [(url: URL, path: String, depth: Int)] = [(rootURL, rootPath, 0)]
        while let top = stack.popLast() {
            guard let entries = try? fm.contentsOfDirectory(
                at: top.url, includingPropertiesForKeys: keys, options: [.skipsHiddenFiles]
            ) else { continue }
            dirsVisited += 1
            if dirsVisited % 32 == 0 { onProgress?(dirsVisited) }
            for url in entries {
                let name = url.lastPathComponent
                let full = top.path + "/" + name
                let rv = try? url.resourceValues(forKeys: keySet)
                if rv?.isDirectory == true {
                    if top.depth < maxDepth { stack.append((url, full, top.depth + 1)) }
                    continue
                }
                guard videoExt.contains((name as NSString).pathExtension.lowercased()) else { continue }
                let mtime = rv?.contentModificationDate
                if let (s, e) = parseSxxEyy(name) {
                    out.append(FoundFile(path: full, size: 0, season: s, episode: e, mtime: mtime))
                } else {
                    // Non-episodic video (movie/clip) — size (prefetched, no extra
                    // stat) lets a movie folder pick the main file over samples.
                    out.append(FoundFile(path: full, size: Int64(rv?.fileSize ?? 0),
                                         season: nil, episode: nil, mtime: mtime))
                }
            }
        }
        return out
    }

    /// Group key for an episode/file: the **top-level show folder** under the
    /// category (`/media/tv/Psych`), collapsing every release + season subfolder
    /// beneath it into ONE show. (Was: the release/season dir, which made Psych
    /// appear once per release.) Loose files directly under the category fall back
    /// to the category dir so `movieItems` splits them per-file.
    private static func showRoot(for filePath: String, mediaRoot: String) -> String {
        let comps = componentsAfter(mediaRoot: mediaRoot, path: filePath)
        if comps.count >= 3 {            // <category>/<show>/…/<file>
            return mediaRoot + "/" + comps[0] + "/" + comps[1]
        }
        return (filePath as NSString).deletingLastPathComponent
    }

    // MARK: - parsing

    public static func parseSxxEyy(_ name: String) -> (Int, Int)? {
        let range = NSRange(name.startIndex..., in: name)
        guard let m = sxxeyy.firstMatch(in: name, range: range),
              let sR = Range(m.range(at: 1), in: name),
              let eR = Range(m.range(at: 2), in: name),
              let s = Int(name[sR]), let e = Int(name[eR]) else { return nil }
        return (s, e)
    }

    private static func episodeLabel(_ path: String) -> String {
        let base = (path as NSString).lastPathComponent
        return (base as NSString).deletingPathExtension
    }

    private static let yearRe = try! NSRegularExpression(pattern: "\\b(19|20)\\d{2}\\b")
    /// First 19xx/20xx year in the path — the release/air year for franchise order.
    static func parseYear(_ path: String) -> Int? {
        let r = NSRange(path.startIndex..., in: path)
        guard let m = yearRe.firstMatch(in: path, range: r), let rr = Range(m.range, in: path) else { return nil }
        return Int(path[rr])
    }

    /// Path components of `path` below `mediaRoot` — e.g. `/media/movies/Inception`
    /// under `/media` → `["movies", "Inception"]`. First element is the category.
    static func componentsAfter(mediaRoot: String, path: String) -> [String] {
        guard path.hasPrefix(mediaRoot) else { return [] }
        var rest = String(path.dropFirst(mediaRoot.count))
        while rest.hasPrefix("/") { rest = String(rest.dropFirst()) }
        return rest.split(separator: "/").map(String.init)
    }

    /// Sample reels, trailers and extras shouldn't represent a movie folder.
    static func isSampleOrExtra(_ path: String) -> Bool {
        let base = (path as NSString).lastPathComponent.lowercased()
        return matches(base, "\\bsample\\b")
            || matches(base, "\\b(extras?|featurettes?|trailers?|behind[ ._-]the[ ._-]scenes)\\b")
    }

    /// Strip bracketed groups, year-and-after, release-noise-and-after, then tidy
    /// separators. Mirrors `normalizeShowName` in library.ts.
    public static func normalizeShowName(_ dirName: String) -> String {
        var s = dirName
        s = replace(s, "\\[[^\\]]*\\]", " ")
        s = replace(s, "\\([^)]*\\)", " ")
        s = replace(s, "\\b(19|20)\\d{2}\\b.*$", "")
        s = replace(s, "\\b(season\\s*\\d+|s\\d{1,2}|complete|series|repack|bluray|webrip|web-dl|hdtv|dvdrip|x264|x265|h\\.?26[45]|hevc|1080p|720p|480p|tvrip|extras?|batch|commentary)\\b.*$", "")
        s = replace(s, "[._-]+", " ")
        s = replace(s, "\\s+", " ").trimmingCharacters(in: .whitespaces)
        return s.isEmpty ? dirName : s
    }

    // MARK: - regex helpers

    private static func matches(_ s: String, _ pattern: String) -> Bool {
        s.range(of: pattern, options: [.regularExpression, .caseInsensitive]) != nil
    }

    private static func replace(_ s: String, _ pattern: String, _ with: String) -> String {
        guard let re = try? NSRegularExpression(pattern: pattern, options: [.caseInsensitive]) else { return s }
        let range = NSRange(s.startIndex..., in: s)
        return re.stringByReplacingMatches(in: s, range: range, withTemplate: with)
    }
}