terebi/Sources/App/ResolvedSource.swift

import Foundation
import Darwin

struct ResolvedSubtitle: Decodable, Sendable {
    let url: String
    let format: String
    let language: String
    let label: String
    let isDefault: Bool
    private let headerFields: [String: String]
    var headers: String {
        headerFields.sorted { $0.key < $1.key }.map { "\($0.key): \($0.value)\r\n" }.joined()
    }
    fileprivate var validatedHeaders: [String: String] { headerFields }
    private enum CodingKeys: String, CodingKey {
        case url, format, language, label
        case isDefault = "default"
        case headerFields = "headers"
    }
}

struct ResolvedSource: Sendable {
    let url: String
    let headers: String
    let identity: String
    let title: String
    let subtitles: [ResolvedSubtitle]

    private struct Document: Decodable {
        struct Identity: Decodable {
            let provider: String
            let media_id: String
            let episode: Int
        }
        let version: Int
        let url: String
        let headers: [String: String]
        let identity: Identity
        let title: String?
        let subtitles: [ResolvedSubtitle]
    }

    private struct InvalidSource: LocalizedError {
        var errorDescription: String? { "Invalid or unreadable resolved source file." }
    }

    static func load(from file: URL) throws -> ResolvedSource {
        do {
            guard file.isFileURL else { throw InvalidSource() }
            let descriptor: Int32 = file.withUnsafeFileSystemRepresentation { path in
                guard let path else { return -1 }
                return Darwin.open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC)
            }
            guard descriptor >= 0 else { throw InvalidSource() }
            var info = stat()
            guard fstat(descriptor, &info) == 0, (info.st_mode & S_IFMT) == S_IFREG else {
                Darwin.close(descriptor)
                throw InvalidSource()
            }
            let handle = FileHandle(fileDescriptor: descriptor, closeOnDealloc: true)
            defer { try? handle.close() }
            let data = try handle.read(upToCount: 1_048_577) ?? Data()
            guard data.count <= 1_048_576 else { throw InvalidSource() }
            let document = try JSONDecoder().decode(Document.self, from: data)
            let id = document.identity
            guard document.version == 1, id.provider == "miruro",
                  !id.media_id.isEmpty, id.media_id.utf8.count <= 512,
                  id.media_id.utf8.allSatisfy({ isIDByte($0) }),
                  id.episode > 0, id.episode <= Int(Int32.max),
                  validURL(document.url), validHeaders(document.headers),
                  validText(document.title ?? "", limit: 4096),
                  document.subtitles.count <= 32,
                  document.subtitles.filter({ $0.isDefault }).count <= 1 else { throw InvalidSource() }
            for subtitle in document.subtitles {
                guard validURL(subtitle.url), subtitle.format == "vtt",
                      validText(subtitle.language, limit: 128), !subtitle.language.isEmpty,
                      validText(subtitle.label, limit: 4096), !subtitle.label.isEmpty,
                      validHeaders(subtitle.validatedHeaders) else { throw InvalidSource() }
            }
            let headers = document.headers.sorted { $0.key < $1.key }
                .map { "\($0.key): \($0.value)\r\n" }.joined()
            return ResolvedSource(url: document.url, headers: headers,
                identity: "miruro:\(id.media_id):\(id.episode)", title: document.title ?? "Miruro episode \(id.episode)",
                subtitles: document.subtitles)
        } catch { throw InvalidSource() }
    }

    private static func isIDByte(_ byte: UInt8) -> Bool {
        (65...90).contains(byte) || (97...122).contains(byte) ||
            (48...57).contains(byte) || byte == 45 || byte == 95
    }

    private static func validText(_ text: String, limit: Int) -> Bool {
        text.utf8.count <= limit && !text.unicodeScalars.contains { $0.value < 32 || $0.value == 127 }
    }

    private static func validURL(_ text: String) -> Bool {
        guard !text.isEmpty, validText(text, limit: 32768),
              !text.contains(where: { $0.isWhitespace }),
              let components = URLComponents(string: text),
              components.scheme == "https" || components.scheme == "http",
              let host = components.host, !host.isEmpty,
              components.user == nil, components.password == nil,
              components.fragment == nil, components.url != nil else { return false }
        return components.port == nil || (1...65535).contains(components.port!)
    }

    private static func validHeaders(_ headers: [String: String]) -> Bool {
        guard headers.count <= 64 else { return false }
        var names = Set<String>()
        var size = 0
        for (name, value) in headers {
            guard !name.isEmpty, name.utf8.count <= 256,
                  name.utf8.allSatisfy({ isIDByte($0) || "!#$%&'*+.^`|~".utf8.contains($0) }),
                  names.insert(name.lowercased()).inserted,
                  value.utf8.count <= 16384,
                  !value.unicodeScalars.contains(where: { $0.value < 32 || $0.value == 127 }) else { return false }
            size += name.utf8.count + value.utf8.count + 4
        }
        return size <= 65536
    }
}