tv-anarchy/Sources/TVAnarchyCore/PornCollectionService.swift

import Foundation

#if ENABLE_ADULT

/// One porn collection with live fresh/total counts, surfaced in the Adult tab
/// and the play-queue popover. `id == name`.
public struct PornCollection: Identifiable, Sendable, Equatable, Decodable {
    public let name: String
    public let desc: String
    public let fresh: Int
    public let total: Int
    public var id: String { name }
    public init(name: String, desc: String, fresh: Int, total: Int) {
        self.name = name; self.desc = desc; self.fresh = fresh; self.total = total
    }
}

/// A named, virtual playlist over the porn pool: an *include* filter (match any
/// substring) with an optional *exclude*, layered on top of the global SKIP
/// floor. A file can belong to several collections; nothing is moved on disk.
/// Ported verbatim (data, not code) from portable-net-tv's `porn-rotation.py`.
public struct PornCollectionSpec: Sendable {
    public let name: String
    public let desc: String
    public let include: [String]
    public let exclude: [String]
    public init(_ name: String, _ desc: String, include: [String], exclude: [String] = []) {
        self.name = name; self.desc = desc; self.include = include; self.exclude = exclude
    }
}

/// Native, in-process port of portable-net-tv's `porn-rotation.py` freshness
/// rotation — no runtime shell-out. The file pool comes from the library index
/// (loose porn files land there as one-file movies under the `porn` category),
/// so it works offline and reuses black's fast index instead of re-walking NFS.
/// The pure resolvers (floor / collection / dedup / freshness) are split from the
/// play-log IO so they're unit-testable without a media mount.
///
/// Freshness is per *file* (basename → last-played ISO), in a single global
/// play-log, so a clip watched under `goon` won't resurface under `pmv` tomorrow.
public enum PornCollectionService {
    // MARK: - Definitions (ported from porn-rotation.py)

    /// The hard floor: titles never surfaced regardless of freshness or
    /// collection. No-JOI rule; cuck/chastity excluded (eporner's "chastity" tag
    /// skews heavily cuckhold — caged/keyholder/chastity are the same cluster).
    static let skipSubstrings = [
        "joi", "cei", "jerk off instruction",
        "cuck", "chastity", "keyholder", "caged",
    ]

    static let videoExts: Set<String> = [".mp4", ".mkv", ".m4v", ".avi", ".webm", ".wmv", ".mov"]

    /// Each collection: include substrings (OR-matched, lowercased); a candidate
    /// joins if it matches ANY include term (or include is empty → the whole
    /// pool). `exclude` drops matches within the collection only. Order is the
    /// display order. Kept in sync with `porn-rotation.py`'s COLLECTIONS dict.
    static let specs: [PornCollectionSpec] = [
        .init("all", "the whole library (no include filter)", include: []),
        .init("goon", "gooning / edging encouragement",
              include: ["goon", "edge", "edging", "encouragement", "mantra", "trance"]),
        .init("pmv", "PMV / HMV music edits",
              include: ["pmv", "hmv", "music video", "cock hero"]),
        .init("sissy", "sissy / trap / femboy",
              include: ["sissy", "sissif", "trap", "femboy", "feminiz"]),
        .init("bbc", "BBC / interracial / split-screen",
              include: ["bbc", "interracial", "blacked", "split screen", "splitscreen"]),
        .init("futa", "futa / futanari",
              include: ["futa", "futanari", "dickgirl"]),
        .init("anime", "hentai / 3D / game-parody animation",
              include: ["hentai", "uncensored", "animated", "animation", "parody",
                        "cartoon", "3d porn", "sfm", "anime"]),
        .init("comp", "compilations / split-screen montages",
              include: ["compilation", "split screen", "splitscreen", " comp ", "cumpilation"]),
        .init("breeding", "breeding / creampie",
              include: ["breed", "breeding", "creampie", "impregnat", "cum inside"]),
        .init("gangbang", "group / gangbang / orgy",
              include: ["gangbang", "gang bang", "orgy", "threesome", "foursome", "bukkake"]),
        .init("milf", "milf / mommy / mature",
              include: ["milf", "stepmom", "step mom", "mommy", "mature", "cougar"]),
        .init("cosplay", "cosplay", include: ["cosplay"]),
        .init("asmr", "ASMR / ear / whisper",
              include: ["asmr", "ear lick", "ear licking", "whisper", "moans in your ear"]),
        .init("naruto", "Naruto parody",
              include: ["naruto", "hinata", "tsunade", "sakura", "kushina", "boruto"]),
        .init("overwatch", "Overwatch parody",
              include: ["overwatch", "d.va", "dva", "widowmaker", "mercy", "tracer", "pharah"]),
    ]

    /// Default freshness window: a file unplayed for this many days is "fresh".
    public static let defaultDays = 7

    // MARK: - Pure resolvers (unit-tested without a mount)

    /// Eporner exports the same clip at several qualities; the basename carries
    /// the stable id in [brackets]. Grouped by it for de-duplication.
    static let epornerID = try! NSRegularExpression(pattern: #"\[([0-9A-Za-z]{6,})\]"#)

    static func passesFloor(_ basename: String) -> Bool {
        let low = basename.lowercased()
        return !skipSubstrings.contains { low.contains($0) }
    }

    static func isPlayable(_ basename: String) -> Bool {
        let low = basename.lowercased()
        guard videoExts.contains(where: { low.hasSuffix($0) }) else { return false }
        // Skip in-flight downloads (.part/.tmp, or eporner's `.mp4_<n>` chunks).
        return !(low.hasSuffix(".part") || low.hasSuffix(".tmp") || low.contains(".mp4_"))
    }

    static func epornerKey(_ basename: String) -> String? {
        let range = NSRange(basename.startIndex..., in: basename)
        guard let m = epornerID.firstMatch(in: basename, range: range),
              let r = Range(m.range(at: 1), in: basename) else { return nil }
        return String(basename[r])
    }

    /// De-dupe eporner re-exports by their [id], keeping the FIRST seen. The index
    /// carries no file size, so we can't replicate the python's "keep largest" —
    /// counts still match (one entry per id), only the chosen representative may
    /// differ. Non-eporner files (no id) pass through untouched.
    static func dedupById(_ paths: [String]) -> [String] {
        var seenIDs = Set<String>()
        var out: [String] = []
        for p in paths {
            let base = (p as NSString).lastPathComponent
            if let key = epornerKey(base) {
                if seenIDs.insert(key).inserted { out.append(p) }
            } else {
                out.append(p)
            }
        }
        return out
    }

    static func inCollection(_ spec: PornCollectionSpec, basename: String) -> Bool {
        let low = basename.lowercased()
        if !spec.include.isEmpty, !spec.include.contains(where: { low.contains($0) }) { return false }
        if spec.exclude.contains(where: { low.contains($0) }) { return false }
        return true
    }

    /// True if `basename` was never played, or last played more than `days` ago.
    static func isFresh(_ basename: String, log: [String: String], days: Int, now: Date) -> Bool {
        guard let ts = log[basename], let last = ISO8601.date(ts) else { return true }
        return last < now.addingTimeInterval(-Double(days) * 86_400)
    }

    /// Floor-filtered, deduped, playable files from a raw library pool. The single
    /// candidate set every collection filters over.
    static func candidates(pool: [String]) -> [String] {
        dedupById(pool.filter {
            let base = ($0 as NSString).lastPathComponent
            return isPlayable(base) && passesFloor(base)
        })
    }

    // MARK: - Public API (drives the UI; `pool` = porn paths from the library index)

    /// The defined collections with fresh/total counts over `pool`. Empty `pool`
    /// (index not yet loaded) → all-zero rows, which the UI renders as disabled.
    public static func collections(pool: [String], days: Int = defaultDays) -> [PornCollection] {
        let log = PornPlayLog.load()
        let now = Date()
        let cands = candidates(pool: pool)
        return specs.map { spec in
            let inSpec = cands.filter { inCollection(spec, basename: ($0 as NSString).lastPathComponent) }
            let fresh = inSpec.filter { isFresh(($0 as NSString).lastPathComponent, log: log, days: days, now: now) }
            return PornCollection(name: spec.name, desc: spec.desc, fresh: fresh.count, total: inSpec.count)
        }
    }

    /// Fresh, shuffled file paths for a collection, marking them played so the
    /// freshness state advances (playback happens app-side, not here).
    public static func freshPaths(pool: [String], collection: String,
                                  count: Int, days: Int = defaultDays) -> [String] {
        guard let spec = specs.first(where: { $0.name == collection }) else { return [] }
        let log = PornPlayLog.load()
        let now = Date()
        let fresh = candidates(pool: pool)
            .filter { inCollection(spec, basename: ($0 as NSString).lastPathComponent) }
            .filter { isFresh(($0 as NSString).lastPathComponent, log: log, days: days, now: now) }
        let pick = Array(fresh.shuffled().prefix(count))
        if !pick.isEmpty { markPlayed(pick) }
        return pick
    }

    /// Stamp paths played now, persisting the shared freshness log.
    public static func markPlayed(_ paths: [String]) {
        guard !paths.isEmpty else { return }
        var log = PornPlayLog.load()
        let stamp = ISO8601.string(Date())
        for p in paths { log[(p as NSString).lastPathComponent] = stamp }
        PornPlayLog.save(log)
    }
}

/// Per-file porn play-log (basename → last-played ISO). App-owned at
/// `~/.local/state/tv-anarchy/porn-plays.json`; on first use it imports the
/// legacy portable-net-tv log so freshness history isn't lost when the runtime
/// shell-out is retired. `TV_ANARCHY_STATE_DIR` redirects it for tests (mirrors
/// QueueStore), so tests never touch the real log.
enum PornPlayLog {
    private static var stateDir: URL {
        if let dir = ProcessInfo.processInfo.environment["TV_ANARCHY_STATE_DIR"], !dir.isEmpty {
            return URL(fileURLWithPath: dir, isDirectory: true)
        }
        return FileManager.default.homeDirectoryForCurrentUser
            .appendingPathComponent(".local/state/tv-anarchy")
    }
    private static var url: URL { stateDir.appendingPathComponent("porn-plays.json") }
    private static var legacyURL: URL {
        FileManager.default.homeDirectoryForCurrentUser
            .appendingPathComponent(".local/state/portable-net-tv/porn-plays.json")
    }

    static func load() -> [String: String] {
        if let d = try? Data(contentsOf: url),
           let log = try? JSONDecoder().decode([String: String].self, from: d) { return log }
        // One-time import: seed from the legacy script's log so the don't-replay
        // history carries over. Only when our own file doesn't exist yet.
        if let d = try? Data(contentsOf: legacyURL),
           let log = try? JSONDecoder().decode([String: String].self, from: d) {
            save(log)
            return log
        }
        return [:]
    }

    static func save(_ log: [String: String]) {
        guard let d = try? JSONEncoder().encode(log) else { return }
        try? FileManager.default.createDirectory(at: stateDir, withIntermediateDirectories: true)
        try? d.write(to: url, options: .atomic)
    }
}

/// ISO-8601 helper for the play-log. The python log is written by
/// `datetime.now(timezone.utc).isoformat()` → **microsecond** precision
/// (`2026-05-29T19:59:47.649910+00:00`), which `ISO8601DateFormatter`'s
/// `.withFractionalSeconds` rejects (it accepts only 3 fractional digits). Since
/// freshness compares at day granularity, sub-second precision is irrelevant: we
/// parse just the `yyyy-MM-dd'T'HH:mm:ss` prefix in UTC, tolerant of any
/// fractional part and timezone suffix. That retroactively rescues the imported
/// legacy timestamps — no data migration needed.
enum ISO8601 {
    private static func utcFormatter(_ fmt: String) -> DateFormatter {
        let f = DateFormatter()
        f.locale = Locale(identifier: "en_US_POSIX")
        f.timeZone = TimeZone(identifier: "UTC")
        f.dateFormat = fmt
        return f
    }
    /// We write the same microsecond-style UTC stamp the python uses, so a log
    /// shared with (or re-read by) the legacy tool stays format-compatible.
    private static let writer = utcFormatter("yyyy-MM-dd'T'HH:mm:ss.SSSSSS'+00:00'")
    private static let secondsParser = utcFormatter("yyyy-MM-dd'T'HH:mm:ss")

    static func string(_ date: Date) -> String { writer.string(from: date) }

    /// Parse to the second, ignoring any fractional/offset tail. Treats the
    /// wall-clock as UTC (the python always writes UTC), which is exact for the
    /// day-granular freshness comparison.
    static func date(_ s: String) -> Date? {
        let head = s.prefix(19)   // "yyyy-MM-ddTHH:mm:ss"
        return secondsParser.date(from: String(head))
    }
}

#endif