ReadViewSDK/Sources/RDReaderView/EPUBCore/RDEPUBParser.swift

import Foundation

public final class RDEPUBParser {

    public internal(set) var metadata = RDEPUBMetadata()

    public internal(set) var manifest: [String: RDEPUBManifestItem] = [:]

    public internal(set) var spine: [RDEPUBSpineItem] = []

    public internal(set) var tableOfContents: [EPUBTableOfContentsItem] = []

    public internal(set) var extractionRootURL: URL?

    public internal(set) var opfURL: URL?

    public var opfDirectoryURL: URL? {
        opfURL?.deletingLastPathComponent()
    }

    public init() {}

    // MARK: - Cache Management

    /// The base directory used for EPUB extraction caches.
    public static var extractionCacheBaseDirectory: URL {
        FileManager.default.urls(for: .cachesDirectory, in: .userDomainMask).first?
            .appendingPathComponent("ssreaderview-epub", isDirectory: true)
            ?? FileManager.default.temporaryDirectory.appendingPathComponent("ssreaderview-epub", isDirectory: true)
    }

    /// Removes all EPUB extraction caches from disk.
    /// Call this when the host app wants to free disk space, e.g. on didReceiveMemoryWarning
    /// or during cleanup when the reader is no longer needed.
    public static func cleanExtractionCache() throws {
        let fileManager = FileManager.default
        let baseURL = extractionCacheBaseDirectory
        guard fileManager.fileExists(atPath: baseURL.path) else { return }
        try fileManager.removeItem(at: baseURL)
    }

    /// Evicts extraction caches until total disk usage is below the given threshold,
    /// removing least-recently-accessed directories first.
    /// - Parameter maxBytes: Maximum total bytes for all extraction caches. Defaults to 500 MB.
    public static func evictExtractionCache(maxBytes: UInt64 = 500 * 1024 * 1024) throws {
        let fileManager = FileManager.default
        let baseURL = extractionCacheBaseDirectory
        guard fileManager.fileExists(atPath: baseURL.path) else { return }

        let contents = try fileManager.contentsOfDirectory(
            at: baseURL,
            includingPropertiesForKeys: [.contentAccessDateKey, .isDirectoryKey],
            options: .skipsHiddenFiles
        )

        var entries: [(url: URL, accessDate: Date, size: UInt64)] = []
        var totalSize: UInt64 = 0

        for directoryURL in contents {
            guard let resourceValues = try? directoryURL.resourceValues(forKeys: [.isDirectoryKey, .contentAccessDateKey]),
                  resourceValues.isDirectory == true else {
                continue
            }
            let directorySize = directoryURL.directorySize
            let accessDate = resourceValues.contentAccessDate ?? Date.distantPast
            entries.append((url: directoryURL, accessDate: accessDate, size: directorySize))
            totalSize += directorySize
        }

        guard totalSize > maxBytes else { return }

        entries.sort { $0.accessDate < $1.accessDate }

        for entry in entries {
            guard totalSize > maxBytes else { break }
            try? fileManager.removeItem(at: entry.url)
            totalSize -= entry.size
        }
    }

    public func makePublication() -> RDEPUBPublication {
        RDEPUBPublication(parser: self)
    }

    public func parse(epubURL: URL) throws {
        reset()

        let extractionURL = try extractArchiveIfNeeded(epubURL: epubURL)
        extractionRootURL = extractionURL

        let containerURL = extractionURL.appendingPathComponent("META-INF/container.xml")
        guard FileManager.default.fileExists(atPath: containerURL.path) else {
            throw RDEPUBParserError.missingContainerXML
        }

        let rootFilePath = try parseContainerRootFile(at: containerURL)
        let packageURL = extractionURL.appendingPathComponent(rootFilePath)
        guard FileManager.default.fileExists(atPath: packageURL.path) else {
            throw RDEPUBParserError.invalidRootFilePath(rootFilePath)
        }

        try parseOPF(at: packageURL)
    }

    public func parseOPF(at opfURL: URL) throws {
        resetPublicationState()

        guard let parser = XMLParser(contentsOf: opfURL) else {
            throw RDEPUBParserError.invalidXML(opfURL)
        }

        let delegate = OPFPackageParserDelegate()
        parser.shouldProcessNamespaces = false
        parser.delegate = delegate

        guard parser.parse() else {
            throw parser.parserError ?? RDEPUBParserError.invalidXML(opfURL)
        }

        let packageDocument = delegate.packageDocument()
        self.opfURL = opfURL
        self.metadata = packageDocument.metadata
        self.manifest = Dictionary(uniqueKeysWithValues: packageDocument.manifest.map { ($0.id, $0) })
        self.spine = try buildSpine(from: packageDocument, opfURL: opfURL)
        self.tableOfContents = parseTOC(from: packageDocument, opfURL: opfURL)
    }

    public func parseTOC() -> [EPUBTableOfContentsItem] {
        tableOfContents
    }

    public func parseNavDocument(_ navURL: URL, baseURL: URL? = nil) -> [EPUBTableOfContentsItem] {
        parseNavDocumentItems(at: navURL, baseURL: baseURL ?? navURL.deletingLastPathComponent())
    }
}

// MARK: - URL Directory Size Extension

private extension URL {
    var directorySize: UInt64 {
        let fileManager = FileManager.default
        guard let enumerator = fileManager.enumerator(at: self, includingPropertiesForKeys: [.fileSizeKey], options: [.skipsHiddenFiles], errorHandler: nil) else {
            return 0
        }
        var totalSize: UInt64 = 0
        for case let fileURL as URL in enumerator {
            guard let resourceValues = try? fileURL.resourceValues(forKeys: [.fileSizeKey]),
                  let fileSize = resourceValues.fileSize else {
                continue
            }
            totalSize += UInt64(fileSize)
        }
        return totalSize
    }
}