mirror of
https://github.com/Aminark/legado-ios.git
synced 2026-10-24 04:41:23 +08:00
Aminark
67bc212ffd
支持 EPUB 2/3 导入、目录、封面和章节插图，保留本地书管理与阅读进度。 修复 JSON 模板、动态替换与请求参数兼容，以及 RSS 翻页和连续滚动定位。 更新编辑器工具栏和回归测试，发布版本为 0.1.12（build 18）。
400 lines
21 KiB
Swift
400 lines
21 KiB
Swift
import Foundation
|
||
import ImageIO
|
||
import SwiftSoup
|
||
import ZIPFoundation
|
||
|
||
public enum EPUBError: Error, LocalizedError, Sendable {
|
||
case invalidArchive, invalidPackage, missingResource(String), encrypted, resourceTooLarge, invalidChapter
|
||
|
||
public var errorDescription: String? {
|
||
switch self {
|
||
case .invalidArchive: "EPUB 文件已损坏或不是有效的压缩包"
|
||
case .invalidPackage: "EPUB 缺少有效的书籍信息或阅读顺序"
|
||
case .missingResource(let path): "EPUB 中缺少文件：\(path)"
|
||
case .encrypted: "此 EPUB 的正文或图片已加密，暂不支持读取"
|
||
case .resourceTooLarge: "EPUB 单个章节或资源过大，无法安全读取"
|
||
case .invalidChapter: "EPUB 章节或目录锚点无效，请重新打开书籍"
|
||
}
|
||
}
|
||
}
|
||
|
||
public struct EPUBIllustration: Identifiable, Sendable {
|
||
public let id: Int
|
||
public let caption: String
|
||
public let data: Data
|
||
}
|
||
|
||
/// Reads resources on demand without unpacking untrusted archive paths to disk.
|
||
/// Navigation follows the spine; EPUB 2 NCX and EPUB 3 nav supply chapter labels/anchors.
|
||
struct EPUBBook {
|
||
private struct Item {
|
||
let id: String
|
||
let path: String
|
||
let mediaType: String
|
||
let properties: Set<String>
|
||
}
|
||
private struct Reference {
|
||
let path: String
|
||
let fragment: String?
|
||
let title: String
|
||
}
|
||
private let archive: Archive
|
||
private let entries: [String: Entry]
|
||
private let package: EPUBXMLNode
|
||
private let packagePath: String
|
||
private let manifest: [Item]
|
||
private let spine: [Item]
|
||
private let encryptedPaths: Set<String>
|
||
private static let resourceLimit = 32 * 1_024 * 1_024
|
||
|
||
init(url: URL) throws {
|
||
let archive: Archive
|
||
do { archive = try Archive(url: url, accessMode: .read) }
|
||
catch { throw EPUBError.invalidArchive }
|
||
var entries: [String: Entry] = [:]
|
||
for entry in archive {
|
||
guard entries.count < 50_000 else { throw EPUBError.resourceTooLarge }
|
||
// Directories and symlinks are never used as book resources.
|
||
guard entry.type == .file else { continue }
|
||
guard Self.safePath(entry.path), entries[entry.path] == nil else { throw EPUBError.invalidArchive }
|
||
entries[entry.path] = entry
|
||
}
|
||
self.archive = archive
|
||
self.entries = entries
|
||
let container = try EPUBXMLNode.parse(Self.read("META-INF/container.xml", archive: archive, entries: entries))
|
||
guard let rootfile = container.descendants("rootfile").first,
|
||
let fullPath = rootfile.attributes["full-path"],
|
||
let path = Self.resolve(fullPath, relativeTo: "")?.path else { throw EPUBError.invalidPackage }
|
||
packagePath = path
|
||
package = try EPUBXMLNode.parse(Self.read(path, archive: archive, entries: entries))
|
||
guard package.name == "package", let manifestNode = package.children.first(where: { $0.name == "manifest" }),
|
||
let spineNode = package.children.first(where: { $0.name == "spine" }) else { throw EPUBError.invalidPackage }
|
||
manifest = manifestNode.children.compactMap { node in
|
||
guard node.name == "item", let id = node.attributes["id"], let href = node.attributes["href"],
|
||
let ref = Self.resolve(href, relativeTo: path) else { return nil }
|
||
return Item(id: id, path: ref.path, mediaType: node.attributes["media-type"] ?? "",
|
||
properties: Set((node.attributes["properties"] ?? "").split(whereSeparator: \.isWhitespace).map(String.init)))
|
||
}
|
||
let items = manifest
|
||
let readable = spineNode.children.filter { $0.name == "itemref" && $0.attributes["linear"] != "no" }
|
||
spine = try readable.map { node in
|
||
guard let item = items.first(where: { $0.id == node.attributes["idref"] }),
|
||
["application/xhtml+xml", "text/html"].contains(item.mediaType), entries[item.path] != nil else {
|
||
throw EPUBError.invalidPackage
|
||
}
|
||
return item
|
||
}
|
||
guard !spine.isEmpty else { throw EPUBError.invalidPackage }
|
||
if entries["META-INF/encryption.xml"] != nil {
|
||
let encryption = try EPUBXMLNode.parse(Self.read("META-INF/encryption.xml", archive: archive, entries: entries))
|
||
encryptedPaths = Set(encryption.descendants("CipherReference").compactMap {
|
||
$0.attributes["URI"].flatMap { Self.resolve($0, relativeTo: "")?.path }
|
||
})
|
||
} else { encryptedPaths = [] }
|
||
// Font obfuscation is common and does not prevent text reading.
|
||
guard !spine.contains(where: { encryptedPaths.contains($0.path) }) else { throw EPUBError.encrypted }
|
||
}
|
||
|
||
func report(url: URL, originalName: String) throws -> LocalBookImportReport {
|
||
let metadata = package.children.first { $0.name == "metadata" }
|
||
let title = metadata?.descendants("title").first?.text.trimmingCharacters(in: .whitespacesAndNewlines) ?? ""
|
||
let authors = metadata?.descendants("creator").map { $0.text.trimmingCharacters(in: .whitespacesAndNewlines) }.filter { !$0.isEmpty } ?? []
|
||
var book = Book(bookUrl: url.absoluteString, name: title.isEmpty ? (originalName as NSString).deletingPathExtension : title,
|
||
author: authors.joined(separator: "、"), origin: "local")
|
||
book.originName = originalName
|
||
book.intro = MetadataTextSanitizer().clean(metadata?.descendants("description").first?.text)
|
||
book.canUpdate = false
|
||
var configuration = BookReadConfig()
|
||
configuration.useReplaceRule = false
|
||
book.readConfig = configuration
|
||
let coverID = metadata?.descendants("meta").first { $0.attributes["name"] == "cover" }?.attributes["content"]
|
||
if let cover = manifest.first(where: { $0.properties.contains("cover-image") }) ?? manifest.first(where: { $0.id == coverID }),
|
||
["image/jpeg", "image/png", "image/gif", "image/webp"].contains(cover.mediaType),
|
||
let entry = entries[cover.path], entry.uncompressedSize <= 2 * 1_024 * 1_024,
|
||
let data = try? read(cover.path), let thumbnail = Self.coverThumbnail(data) {
|
||
// A bounded embedded thumbnail survives file renames, backups and cache eviction.
|
||
book.coverUrl = "data:image/png;base64,\(thumbnail.base64EncodedString())"
|
||
}
|
||
let chapters = try chapters(book: book)
|
||
book.totalChapterNum = chapters.count
|
||
book.latestChapterTitle = chapters.last?.title
|
||
return .init(book: book, chapters: chapters)
|
||
}
|
||
|
||
func chapters(book: Book) throws -> [BookChapter] {
|
||
let references = navigation()
|
||
var chapters: [BookChapter] = []
|
||
for item in spine where !item.properties.contains("nav") {
|
||
let document = try document(item.path)
|
||
guard let body = document.body() else { continue }
|
||
let elements = try body.getAllElements().array()
|
||
var positions: [String: Int] = [:]
|
||
for (index, element) in elements.enumerated() {
|
||
for key in [element.id(), (try? element.attr("name")) ?? ""] where !key.isEmpty && positions[key] == nil {
|
||
positions[key] = index
|
||
}
|
||
}
|
||
var seen = Set<String>()
|
||
let refs = references.filter { $0.path == item.path }
|
||
.filter { ref in
|
||
guard ref.fragment == nil || positions[ref.fragment!] != nil else { return false }
|
||
return seen.insert(ref.fragment ?? "").inserted
|
||
}
|
||
.sorted { (positions[$0.fragment ?? ""] ?? -1) < (positions[$1.fragment ?? ""] ?? -1) }
|
||
let documentTitle = try document.title().trimmingCharacters(in: .whitespacesAndNewlines)
|
||
let heading = try body.select("h1,h2,h3").first()?.text() ?? ""
|
||
let fallback = !documentTitle.isEmpty ? documentTitle : (!heading.isEmpty ? heading : "第\(chapters.count + 1)章")
|
||
var boundaries = refs
|
||
if boundaries.isEmpty {
|
||
boundaries = [Reference(path: item.path, fragment: nil, title: fallback)]
|
||
} else if let first = boundaries.first, let fragment = first.fragment {
|
||
let prefix = try render(body: body, from: nil, to: fragment).text
|
||
if !prefix.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
|
||
boundaries.insert(Reference(path: item.path, fragment: nil, title: fallback), at: 0)
|
||
}
|
||
}
|
||
for (index, ref) in boundaries.enumerated() {
|
||
var chapter = BookChapter(url: Self.referenceURL(path: item.path, fragment: ref.fragment),
|
||
title: ref.title.isEmpty ? fallback : ref.title, bookUrl: book.bookUrl, index: chapters.count)
|
||
chapter.resourceUrl = item.path
|
||
chapter.startFragmentId = ref.fragment
|
||
chapter.endFragmentId = boundaries.indices.contains(index + 1) ? boundaries[index + 1].fragment : nil
|
||
chapters.append(chapter)
|
||
}
|
||
}
|
||
guard !chapters.isEmpty else { throw EPUBError.invalidPackage }
|
||
return chapters
|
||
}
|
||
|
||
func content(chapter: BookChapter) throws -> ChapterContentReport {
|
||
let (body, _) = try body(chapter: chapter)
|
||
let value = try render(body: body, from: chapter.startFragmentId, to: chapter.endFragmentId)
|
||
return .init(title: chapter.title, content: value.text, pagesVisited: 1)
|
||
}
|
||
|
||
func illustrations(chapter: BookChapter) throws -> [EPUBIllustration] {
|
||
let (body, path) = try body(chapter: chapter)
|
||
let value = try render(body: body, from: chapter.startFragmentId, to: chapter.endFragmentId)
|
||
var totalBytes = 0
|
||
return try value.images.enumerated().map { index, image in
|
||
guard let ref = Self.resolve(image.src, relativeTo: path),
|
||
let item = manifest.first(where: { $0.path == ref.path }), item.mediaType.hasPrefix("image/") else {
|
||
throw EPUBError.missingResource(image.src)
|
||
}
|
||
let data = try read(ref.path)
|
||
totalBytes += data.count
|
||
guard totalBytes <= Self.resourceLimit else { throw EPUBError.resourceTooLarge }
|
||
return EPUBIllustration(id: index + 1, caption: image.alt, data: data)
|
||
}
|
||
}
|
||
|
||
private func body(chapter: BookChapter) throws -> (Element, String) {
|
||
guard let path = chapter.resourceUrl, spine.contains(where: { $0.path == path }),
|
||
let body = try document(path).body() else { throw EPUBError.invalidChapter }
|
||
return (body, path)
|
||
}
|
||
|
||
private func document(_ path: String) throws -> Document {
|
||
let data = try read(path)
|
||
let text: String
|
||
if data.starts(with: [0xff, 0xfe]) || data.starts(with: [0xfe, 0xff]), let utf16 = String(data: data, encoding: .utf16) {
|
||
text = utf16
|
||
} else if let utf8 = String(data: data, encoding: .utf8) { text = utf8 }
|
||
else { throw EPUBError.invalidPackage }
|
||
return try SwiftSoup.parse(text)
|
||
}
|
||
|
||
private func navigation() -> [Reference] {
|
||
if let nav = manifest.first(where: { $0.properties.contains("nav") }), let doc = try? document(nav.path),
|
||
let nodes = try? doc.select("nav").array() {
|
||
let toc = nodes.first { node in
|
||
let type = (try? node.attr("epub:type")) ?? ""
|
||
return type.split(whereSeparator: \.isWhitespace).contains("toc") || (try? node.attr("role")) == "doc-toc"
|
||
}
|
||
if let links = try? toc?.select("a[href]").array() {
|
||
let refs = links.compactMap { element -> Reference? in
|
||
guard let href = try? element.attr("href"), let ref = Self.resolve(href, relativeTo: nav.path) else { return nil }
|
||
return Reference(path: ref.path, fragment: ref.fragment, title: (try? element.text()) ?? "")
|
||
}
|
||
if !refs.isEmpty { return refs }
|
||
}
|
||
}
|
||
let tocID = package.children.first { $0.name == "spine" }?.attributes["toc"]
|
||
if let ncx = manifest.first(where: { $0.id == tocID }) ?? manifest.first(where: { $0.mediaType == "application/x-dtbncx+xml" }),
|
||
let data = try? read(ncx.path), let root = try? EPUBXMLNode.parse(data) {
|
||
return root.descendants("navPoint").compactMap { node in
|
||
guard let href = node.children.first(where: { $0.name == "content" })?.attributes["src"],
|
||
let ref = Self.resolve(href, relativeTo: ncx.path) else { return nil }
|
||
return Reference(path: ref.path, fragment: ref.fragment,
|
||
title: node.children.first(where: { $0.name == "navLabel" })?.text.trimmingCharacters(in: .whitespacesAndNewlines) ?? "")
|
||
}
|
||
}
|
||
return []
|
||
}
|
||
|
||
/// Walk DOM boundaries rather than slicing serialized HTML: anchors can be nested
|
||
/// in different parents, and all text after the end anchor belongs to the next chapter.
|
||
private func render(body: Element, from start: String?, to end: String?) throws -> (text: String, images: [(src: String, alt: String)]) {
|
||
var active = start == nil
|
||
var foundStart = start == nil
|
||
var foundEnd = end == nil
|
||
var finished = false
|
||
var output = ""
|
||
var images: [(src: String, alt: String)] = []
|
||
let blocks: Set<String> = ["p", "div", "section", "article", "li", "ul", "ol", "h1", "h2", "h3", "h4", "h5", "h6", "blockquote", "pre", "tr", "hr"]
|
||
func visit(_ node: Node, depth: Int) throws {
|
||
guard depth < 512 else { throw EPUBError.invalidPackage }
|
||
guard !finished else { return }
|
||
guard let element = node as? Element else {
|
||
if active, let text = node as? TextNode {
|
||
output += text.getWholeText().replacingOccurrences(of: #"\s+"#, with: " ", options: .regularExpression)
|
||
}
|
||
return
|
||
}
|
||
let id = element.id()
|
||
let name = try element.attr("name")
|
||
if let end, id == end || name == end { active = false; finished = true; foundEnd = true; return }
|
||
if let start, id == start || name == start { active = true; foundStart = true }
|
||
let tag = element.tagName().lowercased()
|
||
let style = try element.attr("style").lowercased().replacingOccurrences(of: #"\s+"#, with: "", options: .regularExpression)
|
||
if ["script", "style", "noscript", "rt", "rp", "title"].contains(tag) || element.hasAttr("hidden") || style.contains("display:none") { return }
|
||
if active, blocks.contains(tag) || tag == "br" { output += "\n" }
|
||
if active, tag == "img" || tag == "image" {
|
||
var src = try element.attr(tag == "img" ? "src" : "xlink:href")
|
||
if src.isEmpty, tag == "image" { src = try element.attr("href") }
|
||
let alt = try element.attr("alt")
|
||
if !src.isEmpty {
|
||
images.append((src, alt))
|
||
output += "\n【插图 \(images.count)\(alt.isEmpty ? "" : "：" + alt)】\n"
|
||
}
|
||
}
|
||
for child in element.getChildNodes() { try visit(child, depth: depth + 1) }
|
||
if active, blocks.contains(tag) { output += "\n" }
|
||
if active, tag == "td" || tag == "th" { output += " " }
|
||
}
|
||
try visit(body, depth: 0)
|
||
guard foundStart && foundEnd else { throw EPUBError.invalidChapter }
|
||
let text = output.components(separatedBy: .newlines).map { $0.trimmingCharacters(in: .whitespaces) }.filter { !$0.isEmpty }.joined(separator: "\n")
|
||
return (text, images)
|
||
}
|
||
|
||
private func read(_ path: String) throws -> Data {
|
||
guard !encryptedPaths.contains(path) else { throw EPUBError.encrypted }
|
||
return try Self.read(path, archive: archive, entries: entries)
|
||
}
|
||
|
||
private static func read(_ path: String, archive: Archive, entries: [String: Entry]) throws -> Data {
|
||
guard let entry = entries[path] else { throw EPUBError.missingResource(path) }
|
||
guard entry.uncompressedSize <= resourceLimit else { throw EPUBError.resourceTooLarge }
|
||
var result = Data()
|
||
let checksum = try archive.extract(entry) { chunk in
|
||
guard result.count + chunk.count <= resourceLimit else { throw EPUBError.resourceTooLarge }
|
||
result.append(chunk)
|
||
}
|
||
guard checksum == entry.checksum, result.count == entry.uncompressedSize else { throw EPUBError.invalidArchive }
|
||
return result
|
||
}
|
||
|
||
private static func coverThumbnail(_ data: Data) -> Data? {
|
||
guard let source = CGImageSourceCreateWithData(data as CFData, nil),
|
||
let image = CGImageSourceCreateThumbnailAtIndex(source, 0, [
|
||
kCGImageSourceCreateThumbnailFromImageAlways: true,
|
||
kCGImageSourceThumbnailMaxPixelSize: 512,
|
||
kCGImageSourceCreateThumbnailWithTransform: true
|
||
] as CFDictionary) else { return nil }
|
||
let output = NSMutableData()
|
||
guard let destination = CGImageDestinationCreateWithData(output, "public.png" as CFString, 1, nil) else { return nil }
|
||
CGImageDestinationAddImage(destination, image, nil)
|
||
guard CGImageDestinationFinalize(destination) else { return nil }
|
||
return output as Data
|
||
}
|
||
|
||
private static func safePath(_ path: String) -> Bool {
|
||
!path.isEmpty && !path.hasPrefix("/") && !path.contains("\\") && !path.contains("\0") && !path.split(separator: "/").contains("..")
|
||
}
|
||
|
||
private static func resolve(_ href: String, relativeTo base: String) -> (path: String, fragment: String?)? {
|
||
guard !href.hasPrefix("/"), !href.contains("\\"),
|
||
URLComponents(string: href)?.scheme == nil else { return nil }
|
||
let parts = href.split(separator: "#", maxSplits: 1, omittingEmptySubsequences: false)
|
||
guard let pathPart = String(parts[0]).removingPercentEncoding else { return nil }
|
||
var components = pathPart.isEmpty ? base.split(separator: "/").map(String.init) : base.split(separator: "/").dropLast().map(String.init)
|
||
if !pathPart.isEmpty {
|
||
for part in pathPart.split(separator: "/") {
|
||
if part == "." { continue }
|
||
if part == ".." { guard !components.isEmpty else { return nil }; components.removeLast() }
|
||
else { components.append(String(part)) }
|
||
}
|
||
}
|
||
let path = components.joined(separator: "/")
|
||
guard safePath(path), !pathPart.hasPrefix("/") else { return nil }
|
||
let fragment = parts.count > 1 ? String(parts[1]).removingPercentEncoding : nil
|
||
return (path, fragment?.isEmpty == false ? fragment : nil)
|
||
}
|
||
|
||
private static func referenceURL(path: String, fragment: String?) -> String {
|
||
var url = URLComponents()
|
||
url.scheme = "epub"
|
||
url.host = "book"
|
||
url.path = "/" + path
|
||
url.fragment = fragment
|
||
return url.string ?? path
|
||
}
|
||
}
|
||
|
||
/// Small namespace-independent XML tree for package metadata and navigation.
|
||
private final class EPUBXMLNode: NSObject, XMLParserDelegate {
|
||
var name = ""
|
||
var attributes: [String: String] = [:]
|
||
var children: [EPUBXMLNode] = []
|
||
var text = ""
|
||
private var stack: [EPUBXMLNode] = []
|
||
private var count = 0
|
||
|
||
func descendants(_ name: String) -> [EPUBXMLNode] {
|
||
children.flatMap { ($0.name == name ? [$0] : []) + $0.descendants(name) }
|
||
}
|
||
|
||
static func parse(_ data: Data) throws -> EPUBXMLNode {
|
||
let delegate = EPUBXMLNode()
|
||
let parser = XMLParser(data: data)
|
||
parser.shouldResolveExternalEntities = false
|
||
parser.delegate = delegate
|
||
guard parser.parse(), let root = delegate.children.first else { throw EPUBError.invalidPackage }
|
||
return root
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, didStartElement elementName: String, namespaceURI: String?, qualifiedName qName: String?, attributes attributeDict: [String: String]) {
|
||
count += 1
|
||
guard stack.count < 128, count <= 50_000 else { parser.abortParsing(); return }
|
||
let node = EPUBXMLNode()
|
||
node.name = elementName.split(separator: ":").last.map(String.init) ?? elementName
|
||
node.attributes = attributeDict
|
||
(stack.last ?? self).children.append(node)
|
||
stack.append(node)
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, foundCharacters string: String) {
|
||
for node in stack {
|
||
guard node.text.utf8.count + string.utf8.count <= 4 * 1_024 * 1_024 else { parser.abortParsing(); return }
|
||
node.text += string
|
||
}
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, foundInternalEntityDeclarationWithName name: String, value: String?) {
|
||
parser.abortParsing()
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, foundExternalEntityDeclarationWithName name: String, publicID: String?, systemID: String?) {
|
||
parser.abortParsing()
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, foundCDATA CDATABlock: Data) {
|
||
if let value = String(data: CDATABlock, encoding: .utf8) { self.parser(parser, foundCharacters: value) }
|
||
}
|
||
|
||
func parser(_ parser: XMLParser, didEndElement elementName: String, namespaceURI: String?, qualifiedName qName: String?) {
|
||
_ = stack.popLast()
|
||
}
|
||
}
|