kotej/Sources/KotejEngine/Tree.swift

import Foundation

/// Where a node's bytes live. Archives are browsed in place, so an entry inside a
/// JAR is addressed without ever unpacking the whole thing to disk.
public enum NodeSource: Sendable {
    case fileSystem(URL)
    case archiveEntry(archive: URL, path: String)
    /// A nested archive (a JAR inside an EAR): the bytes were extracted once so
    /// its own central directory could be read.
    case extracted(URL, origin: String)
}

/// One file or folder in a comparison tree. Archives appear as folders.
public final class Node: @unchecked Sendable {
    public let name: String
    /// Path relative to the root of its side, used to pair the two trees.
    public let path: String
    public let isDirectory: Bool
    public let size: UInt64
    /// CRC32 from the archive's central directory, when the node came from one.
    /// Present means "we already know the content fingerprint for free".
    public let crc32: UInt32?
    public let source: NodeSource
    /// True when this folder is really an archive shown as a folder.
    public let isArchive: Bool
    public private(set) var children: [Node]

    init(name: String, path: String, isDirectory: Bool, size: UInt64 = 0,
         crc32: UInt32? = nil, source: NodeSource, isArchive: Bool = false,
         children: [Node] = []) {
        self.name = name
        self.path = path
        self.isDirectory = isDirectory
        self.size = size
        self.crc32 = crc32
        self.source = source
        self.isArchive = isArchive
        self.children = children
    }

    func add(_ child: Node) { children.append(child) }
    func sortChildren() {
        children.sort { left, right in
            if left.isDirectory != right.isDirectory { return left.isDirectory }
            return left.name.localizedStandardCompare(right.name) == .orderedAscending
        }
        children.forEach { $0.sortChildren() }
    }

    /// Reads this node's bytes, inflating from its archive when needed.
    public func data() throws -> Data {
        switch source {
        case .fileSystem(let url), .extracted(let url, _):
            return try Data(contentsOf: url, options: .mappedIfSafe)
        case .archiveEntry(let archive, let entryPath):
            let reader = try ZipReader(url: archive)
            guard let entry = reader.entries.first(where: { $0.path == entryPath }) else {
                throw ZipReader.ZipError.corrupt("entry \(entryPath) vanished")
            }
            return try reader.data(for: entry)
        }
    }
}

/// Builds comparison trees from folders and archives.
public enum TreeScanner {
    /// Scans a folder or an archive into a tree. Archives — including archives
    /// nested inside archives (an EAR's JARs) — are expanded as folders, which is
    /// what makes comparing deployment artefacts useful.
    public static func scan(_ url: URL, expandNestedArchives: Bool = true) throws -> Node {
        var isDirectory: ObjCBool = false
        guard FileManager.default.fileExists(atPath: url.path, isDirectory: &isDirectory) else {
            throw CocoaError(.fileNoSuchFile)
        }
        if isDirectory.boolValue {
            return try scanDirectory(url, expandNestedArchives: expandNestedArchives)
        }
        if ZipReader.isArchive(url) {
            return try scanArchive(url, expandNestedArchives: expandNestedArchives)
        }
        let size = (try? url.resourceValues(forKeys: [.fileSizeKey]).fileSize) ?? 0
        return Node(name: url.lastPathComponent, path: url.lastPathComponent, isDirectory: false,
                    size: UInt64(size), source: .fileSystem(url))
    }

    // MARK: Folders

    private static func scanDirectory(_ url: URL, expandNestedArchives: Bool) throws -> Node {
        let root = Node(name: url.lastPathComponent, path: "", isDirectory: true, source: .fileSystem(url))
        try fill(root, at: url, prefix: "", expandNestedArchives: expandNestedArchives)
        root.sortChildren()
        return root
    }

    private static func fill(_ parent: Node, at url: URL, prefix: String, expandNestedArchives: Bool) throws {
        let keys: [URLResourceKey] = [.isDirectoryKey, .fileSizeKey]
        let entries = try FileManager.default.contentsOfDirectory(at: url, includingPropertiesForKeys: keys,
                                                                  options: [.skipsHiddenFiles])
        for entry in entries {
            let values = try? entry.resourceValues(forKeys: Set(keys))
            let childPath = prefix.isEmpty ? entry.lastPathComponent : prefix + "/" + entry.lastPathComponent

            if values?.isDirectory == true {
                let node = Node(name: entry.lastPathComponent, path: childPath, isDirectory: true,
                                source: .fileSystem(entry))
                parent.add(node)
                try fill(node, at: entry, prefix: childPath, expandNestedArchives: expandNestedArchives)
            } else if expandNestedArchives, ZipReader.isArchive(entry) {
                // Show the archive as a folder, keeping its own name.
                if let archiveNode = try? scanArchive(entry, expandNestedArchives: expandNestedArchives,
                                                      path: childPath) {
                    parent.add(archiveNode)
                } else {
                    parent.add(fileNode(entry, path: childPath, size: values?.fileSize ?? 0))
                }
            } else {
                parent.add(fileNode(entry, path: childPath, size: values?.fileSize ?? 0))
            }
        }
    }

    private static func fileNode(_ url: URL, path: String, size: Int) -> Node {
        Node(name: url.lastPathComponent, path: path, isDirectory: false,
             size: UInt64(size), source: .fileSystem(url))
    }

    // MARK: Archives

    /// Turns an archive's flat entry list into a tree of folders and files.
    static func scanArchive(_ url: URL, expandNestedArchives: Bool, path: String? = nil) throws -> Node {
        let reader = try ZipReader(url: url)
        let rootPath = path ?? url.lastPathComponent
        let root = Node(name: url.lastPathComponent, path: rootPath, isDirectory: true,
                        source: .fileSystem(url), isArchive: true)

        var folders: [String: Node] = ["": root]

        func folder(_ folderPath: String) -> Node {
            if let existing = folders[folderPath] { return existing }
            let parentPath = (folderPath as NSString).deletingLastPathComponent
            let parent = folder(parentPath)
            let node = Node(name: (folderPath as NSString).lastPathComponent,
                            path: rootPath + "/" + folderPath, isDirectory: true,
                            source: .archiveEntry(archive: url, path: folderPath))
            parent.add(node)
            folders[folderPath] = node
            return node
        }

        for entry in reader.entries where !entry.isDirectory {
            let parentPath = (entry.path as NSString).deletingLastPathComponent
            let parent = folder(parentPath)
            let childPath = rootPath + "/" + entry.path
            let name = (entry.path as NSString).lastPathComponent

            // A JAR inside an EAR: extract once so its own entries can be listed.
            // A JAR inside an EAR, or any renamed ZIP: expand it as a folder.
            if expandNestedArchives,
               ZipReader.mayBeNestedArchive(name: name, uncompressedSize: entry.uncompressedSize) {
                if let nested = try? extractNested(reader: reader, entry: entry, name: name),
                   ZipReader.looksLikeArchive(fileAt: nested),
                   let nestedNode = try? scanArchive(nested, expandNestedArchives: expandNestedArchives,
                                                     path: childPath) {
                    parent.add(nestedNode)
                    continue
                }
            }
            parent.add(Node(name: name, path: childPath, isDirectory: false,
                            size: entry.uncompressedSize, crc32: entry.crc32,
                            source: .archiveEntry(archive: url, path: entry.path)))
        }
        root.sortChildren()
        return root
    }

    private static func extractNested(reader: ZipReader, entry: ZipReader.Entry, name: String) throws -> URL {
        let directory = URL(fileURLWithPath: NSTemporaryDirectory())
            .appendingPathComponent("kotej-nested-\(UUID().uuidString)")
        try FileManager.default.createDirectory(at: directory, withIntermediateDirectories: true)
        let url = directory.appendingPathComponent(name)
        try reader.data(for: entry).write(to: url)
        return url
    }
}