chore: update AltStore source for v0.1.12

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）。

App/Compatibility.swift

@@ -207,7 +207,7 @@ struct ReadingPhotoPickerController: UIViewControllerRepresentable {
     }
 }
 
-private struct ReadingTXTFileImporterModifier: ViewModifier {
+private struct ReadingLocalBookFileImporterModifier: ViewModifier {
     @Binding var isPresented: Bool
     let onFiles: ([LocalBookImportFile]) -> Void
     let onError: (Error) -> Void
@@ -219,7 +219,7 @@ private struct ReadingTXTFileImporterModifier: ViewModifier {
             pickerHasDismissed = true
             deliverFilesWhenReady()
         }) {
-            ReadingTXTFilePicker { files in
+            ReadingLocalBookFilePicker { files in
                 pickedFiles = files
                 isPresented = false
                 deliverFilesWhenReady()
@@ -239,7 +239,7 @@ private struct ReadingTXTFileImporterModifier: ViewModifier {
         guard pickerHasDismissed, !pickedFiles.isEmpty else { return }
         let files = pickedFiles
         pickedFiles = []
-        guard files.allSatisfy({ $0.url.pathExtension.lowercased() == "txt" && !$0.url.hasDirectoryPath }) else {
+        guard files.allSatisfy({ LocalBookService.supports($0.url) && !$0.url.hasDirectoryPath }) else {
             onError(LocalBookError.unsupportedFormat)
             return
         }
@@ -247,15 +247,16 @@ private struct ReadingTXTFileImporterModifier: ViewModifier {
     }
 }
 
-private struct ReadingTXTFilePicker: UIViewControllerRepresentable {
+private struct ReadingLocalBookFilePicker: UIViewControllerRepresentable {
     let onPick: @MainActor ([LocalBookImportFile]) -> Void
     let onCancel: @MainActor () -> Void
 
     func makeCoordinator() -> Coordinator { Coordinator(parent: self) }
 
     func makeUIViewController(context: Context) -> UIDocumentPickerViewController {
-        // Plain text includes UTF-8/UTF-16 TXT, while .text and .data also allow JSON.
+        // Keep source JSON out of the local-book picker.
-        let picker = UIDocumentPickerViewController(forOpeningContentTypes: [.plainText], asCopy: true)
+        let epub = UTType(filenameExtension: "epub") ?? UTType(importedAs: "org.idpf.epub-container")
+        let picker = UIDocumentPickerViewController(forOpeningContentTypes: [.plainText, epub], asCopy: true)
         picker.allowsMultipleSelection = true
         picker.shouldShowFileExtensions = true
         picker.delegate = context.coordinator
@@ -266,9 +267,9 @@ private struct ReadingTXTFilePicker: UIViewControllerRepresentable {
 
     @MainActor
     final class Coordinator: NSObject, UIDocumentPickerDelegate {
-        let parent: ReadingTXTFilePicker
+        let parent: ReadingLocalBookFilePicker
 
-        init(parent: ReadingTXTFilePicker) { self.parent = parent }
+        init(parent: ReadingLocalBookFilePicker) { self.parent = parent }
 
         func documentPickerWasCancelled(_ controller: UIDocumentPickerViewController) { parent.onCancel() }
 
@@ -281,12 +282,12 @@ private struct ReadingTXTFilePicker: UIViewControllerRepresentable {
 enum ReadingSheetDetent { case medium, large }
 
 extension View {
-    func readingTXTFileImporter(
+    func readingLocalBookFileImporter(
         isPresented: Binding<Bool>,
         onFiles: @escaping ([LocalBookImportFile]) -> Void,
         onError: @escaping (Error) -> Void
     ) -> some View {
-        modifier(ReadingTXTFileImporterModifier(isPresented: isPresented, onFiles: onFiles, onError: onError))
+        modifier(ReadingLocalBookFileImporterModifier(isPresented: isPresented, onFiles: onFiles, onError: onError))
     }
 
     func readingPhotoPicker(

App/DesignSystem.swift

@@ -300,11 +300,20 @@ struct BookCoverArtwork: View {
     }
 
     var body: some View {
-        AsyncImage(url: remoteURL.flatMap(URL.init(string:))) { phase in
+        Group {
-            if case .success(let image) = phase {
+            if let remoteURL, remoteURL.hasPrefix("data:image/"),
-                image.resizable().scaledToFill()
+               let comma = remoteURL.firstIndex(of: ","),
+               let data = Data(base64Encoded: String(remoteURL[remoteURL.index(after: comma)...])),
+               let image = UIImage(data: data) {
+                Image(uiImage: image).resizable().scaledToFill()
             } else {
-                placeholder
+                AsyncImage(url: remoteURL.flatMap(URL.init(string:))) { phase in
+                    if case .success(let image) = phase {
+                        image.resizable().scaledToFill()
+                    } else {
+                        placeholder
+                    }
+                }
             }
         }
         .frame(width: width, height: height)

App/Features/Activity/FileManagementView.swift

@@ -126,7 +126,7 @@ struct FileManagementView: View {
                     Button("完成") { selecting = false; selection.removeAll() }
                 } else {
                     Menu {
-                        Button("导入 TXT 文件", systemImage: "square.and.arrow.down") { importsFile = true }
+                        Button("导入 TXT / EPUB", systemImage: "square.and.arrow.down") { importsFile = true }
                         if tab == 0 {
                             Button("选择文件", systemImage: "checkmark.circle") { selecting = true }
                                 .disabled(displayedFiles.isEmpty)
@@ -163,7 +163,7 @@ struct FileManagementView: View {
 
     private var importAndRenameView: some View {
         fileList
-        .readingTXTFileImporter(isPresented: $importsFile) { files in
+        .readingLocalBookFileImporter(isPresented: $importsFile) { files in
             Task { await model.importFiles(files, library: library) }
         } onError: { error in
             model.errorMessage = error.localizedDescription
@@ -177,7 +177,7 @@ struct FileManagementView: View {
                 Task { await model.run(library) { try await library.renameLocalFile(at: url, to: name) } }
             }
         } message: { _ in
-            Text("保留 TXT 格式，书架中的书名、书签和阅读进度不变。")
+            Text("保留原文件格式，书架中的书名、书签和阅读进度不变。")
         }
     }
 
@@ -239,8 +239,8 @@ struct FileManagementView: View {
         }
         Section("文件") {
             if displayedFiles.isEmpty {
-                ReadingUnavailableView(query.isEmpty ? "还没有本地文件" : "没有匹配的文件", systemImage: "doc.text", description: Text("可从系统“文件”导入 TXT，文件会复制到应用中。"))
+                ReadingUnavailableView(query.isEmpty ? "还没有本地文件" : "没有匹配的文件", systemImage: "doc.text", description: Text("可从系统“文件”导入 TXT 或 EPUB，文件会复制到应用中。"))
-                Button("导入 TXT 文件", systemImage: "plus") { importsFile = true }
+                Button("导入 TXT / EPUB", systemImage: "plus") { importsFile = true }
             }
             ForEach(displayedFiles) { file in
                 Group {

App/Features/Books/BookLibraryModel.swift

@@ -360,7 +360,7 @@ final class BookLibraryModel: ObservableObject {
 
     func addLocalFile(at url: URL) async throws {
         guard !books.contains(where: { $0.bookUrl == url.absoluteString }) else { return }
-        let report = try await Task.detached { [localService] in try localService.openTextFile(at: url) }.value
+        let report = try await Task.detached { [localService] in try localService.openFile(at: url) }.value
         guard !books.contains(where: { $0.bookUrl == url.absoluteString }) else { return }
         guard await persist(books + [report.book]) else {
             throw LocalFileError.saveFailed(errorMessage ?? "无法写入数据库")

App/Features/Books/LocalBookDetailView.swift

@@ -33,16 +33,20 @@ struct LocalBookDetailView: View {
         ScrollView {
             VStack(spacing: 16) {
                 VStack(spacing: 14) {
-                    BookCoverArtwork(title: book.name, author: book.author, width: 108, height: 154)
+                    BookCoverArtwork(title: book.name, author: book.author, remoteURL: book.customCoverUrl ?? book.coverUrl, width: 108, height: 154)
                     Text(book.name).font(.title2.bold())
+                    if !book.author.isEmpty { Text(book.author).font(.subheadline).foregroundStyle(.secondary) }
                     HStack {
-                        StatusPill(text: "本地 TXT", systemImage: "doc.text.fill")
+                        StatusPill(text: "本地 " + (URL(string: book.bookUrl)?.pathExtension.uppercased() ?? "TXT"), systemImage: "doc.text.fill")
                         if let charset = book.charset { StatusPill(text: charset, color: ReadingTheme.coral, systemImage: "character") }
                     }
                     Text(book.originName).font(.caption).foregroundStyle(.secondary)
                 }
                 .padding(.top, 12)
 
+                if let intro = book.intro, !intro.isEmpty {
+                    ReadingCard { Text(intro).font(.subheadline).frame(maxWidth: .infinity, alignment: .leading) }
+                }
                 ReadingCard {
                     VStack(alignment: .leading, spacing: 0) {
                         HStack {

App/Features/Books/SourceSwitchView.swift

@@ -151,6 +151,6 @@ struct SourceSwitchView: View {
             .padding(.vertical, 5)
         }
         .disabled(model.switchingID != nil)
-        .accessibilityIdentifier("sourceSwitch.candidate.\(candidate.book.origin)")
+        .accessibilityIdentifier("sourceSwitch.candidate.\(candidate.id)")
     }
 }

App/Features/RSS/RSSLibraryModel.swift

@@ -10,29 +10,42 @@ final class RSSLibraryModel: ObservableObject {
     @Published private(set) var isWorking = false
     @Published var errorMessage: String?
 
-    private let sourceRepository: SQLiteRssSourceRepository
+    private var sourceRepository: SQLiteRssSourceRepository { get throws { .init(database: try database) } }
-    private let articleRepository: SQLiteRssArticleRepository
+    private var articleRepository: SQLiteRssArticleRepository { get throws { .init(database: try database) } }
-    private let starRepository: SQLiteRssStarRepository
+    private var starRepository: SQLiteRssStarRepository { get throws { .init(database: try database) } }
-    private let recordRepository: SQLiteRssReadRecordRepository
+    private var recordRepository: SQLiteRssReadRecordRepository { get throws { .init(database: try database) } }
     private let service: RSSService
     private let importer = RSSSourceImporter()
-    private let database: LegadoDatabase
+    private let databaseResult: Result<LegadoDatabase, Error>
+    private var database: LegadoDatabase { get throws { try databaseResult.get() } }
     private var nextURL: String?
     private var page = 1
     private var order: Int64 = 0
+    private struct Feed: Equatable {
+        let origin: String
+        let sort: String
+        let key: String?
+    }
+    private var activeFeed: Feed?
+    private var articleRequestID = UUID()
+    private var loadingMoreRequestID: UUID?
+    private var activeOperations = 0
 
     init(database: LegadoDatabase, service: RSSService = .init()) {
-        self.database = database
+        databaseResult = .success(database)
-        sourceRepository = .init(database: database)
-        articleRepository = .init(database: database)
-        starRepository = .init(database: database)
-        recordRepository = .init(database: database)
         self.service = service
     }
 
-    static func live() -> RSSLibraryModel {
+    private init(databaseError: Error) {
-        let database = try! LegadoDatabase(url: LegadoDatabase.defaultURL())
+        let error = LegadoDatabaseError.open("RSS 数据库无法打开：\(databaseError.localizedDescription)。请检查存储空间后重新启动应用。")
-        return .init(database: database)
+        databaseResult = .failure(error)
+        service = .init()
+        errorMessage = error.localizedDescription
+    }
+
+    static func live(databaseURL: URL = LegadoDatabase.defaultURL()) -> RSSLibraryModel {
+        do { return .init(database: try LegadoDatabase(url: databaseURL)) }
+        catch { return .init(databaseError: error) }
     }
 
     func load() async {
@@ -62,15 +75,16 @@ final class RSSLibraryModel: ObservableObject {
     func importText(_ text: String) async {
         await perform {
             let trimmed = text.trimmingCharacters(in: .whitespacesAndNewlines)
+            var merged: [RssSource]
             if let url = URL(string: trimmed), ["http", "https"].contains(url.scheme?.lowercased() ?? "") {
-                self.sources = try await self.importRemote(trimmed, existing: self.sources)
+                merged = try await self.importRemote(trimmed, existing: self.sources)
             } else {
                 let payload = try self.importer.decode(Data(trimmed.utf8))
-                var merged = try self.importer.merge(payload.sources, into: self.sources)
+                merged = try self.importer.merge(payload.sources, into: self.sources)
                 for remote in payload.subscriptionURLs { merged = try await self.importRemote(remote, existing: merged) }
-                self.sources = merged
             }
-            try await self.sourceRepository.save(self.sources)
+            try await self.sourceRepository.save(merged)
+            self.sources = try await self.sourceRepository.all()
         }
     }
 
@@ -130,31 +144,53 @@ final class RSSLibraryModel: ObservableObject {
     }
 
     func refresh(_ source: RssSource, sortName: String? = nil, sortURL: String? = nil, key: String? = nil) async {
-        page = 1; order = Int64(Date().timeIntervalSince1970 * 1_000)
         let name = sortName ?? source.sourceName
         let url = sortURL ?? source.sortUrl?.nilIfBlank ?? source.sourceUrl
-        await perform {
+        let feed = Feed(origin: source.sourceUrl, sort: name, key: key)
+        let requestID = UUID()
+        articleRequestID = requestID
+        loadingMoreRequestID = nil
+        if activeFeed != feed { articles = [] }
+        activeFeed = feed
+        nextURL = nil
+        page = 1
+        await perform(isCurrent: { self.articleRequestID == requestID }) {
             let report = try await self.service.articles(sortName: name, sortURL: url, source: source, page: 1, key: key)
+            guard self.articleRequestID == requestID else { return }
+            var nextOrder = Int64(Date().timeIntervalSince1970 * 1_000)
             var values = report.result.articles
-            for index in values.indices { values[index].order = self.order; self.order -= 1 }
+            for index in values.indices { values[index].order = nextOrder; nextOrder -= 1 }
             try await self.articleRepository.put(values)
-            if source.ruleNextPage?.isEmpty == false { try await self.articleRepository.clearOld(origin: source.sourceUrl, sort: name, before: self.order) }
+            if source.ruleNextPage?.isEmpty == false { try await self.articleRepository.clearOld(origin: source.sourceUrl, sort: name, before: nextOrder) }
+            let articles = try await self.articleRepository.articles(origin: source.sourceUrl, sort: name)
+            guard self.articleRequestID == requestID else { return }
+            self.order = nextOrder
             self.nextURL = report.result.nextURL
-            self.articles = try await self.articleRepository.articles(origin: source.sourceUrl, sort: name)
+            self.articles = articles
         }
     }
 
     func loadMore(_ source: RssSource, sortName: String? = nil, key: String? = nil) async {
-        guard let nextURL, !nextURL.isEmpty else { return }
-        page += 1
         let name = sortName ?? source.sourceName
-        await perform {
+        guard activeFeed == Feed(origin: source.sourceUrl, sort: name, key: key),
-            let report = try await self.service.articles(sortName: name, sortURL: nextURL, source: source, page: self.page, key: key)
+              loadingMoreRequestID == nil, let nextURL, !nextURL.isEmpty else { return }
+        let requestID = articleRequestID
+        loadingMoreRequestID = requestID
+        defer { if loadingMoreRequestID == requestID { loadingMoreRequestID = nil } }
+        let nextPage = page + 1
+        await perform(isCurrent: { self.articleRequestID == requestID }) {
+            let report = try await self.service.articles(sortName: name, sortURL: nextURL, source: source, page: nextPage, key: key)
+            guard self.articleRequestID == requestID else { return }
+            var nextOrder = self.order
             var values = report.result.articles
-            for index in values.indices { values[index].order = self.order; self.order -= 1 }
+            for index in values.indices { values[index].order = nextOrder; nextOrder -= 1 }
             try await self.articleRepository.append(values)
+            let articles = try await self.articleRepository.articles(origin: source.sourceUrl, sort: name)
+            guard self.articleRequestID == requestID else { return }
+            self.page = nextPage
+            self.order = nextOrder
             self.nextURL = report.result.nextURL
-            self.articles = try await self.articleRepository.articles(origin: source.sourceUrl, sort: name)
+            self.articles = articles
         }
     }
 
@@ -231,9 +267,11 @@ final class RSSLibraryModel: ObservableObject {
         let payload = try importer.decode(data)
         return try importer.merge(payload.sources, into: existing)
     }
-    private func perform(_ operation: () async throws -> Void) async {
+    private func perform(isCurrent: () -> Bool = { true }, _ operation: () async throws -> Void) async {
-        isWorking = true; defer { isWorking = false }
+        activeOperations += 1
-        do { try await operation() } catch { errorMessage = error.localizedDescription }
+        isWorking = true
+        defer { activeOperations -= 1; isWorking = activeOperations > 0 }
+        do { try await operation() } catch { if isCurrent() { errorMessage = error.localizedDescription } }
     }
 }
 

App/Features/RSS/RSSViews.swift

@@ -151,14 +151,20 @@ private struct RSSArticleListView: View {
         .overlay { if library.articles.isEmpty, !library.isWorking { ReadingUnavailableView("没有文章", systemImage: "newspaper") } }
         .navigationTitle(source.sourceName)
         .searchable(text: $searchKey, prompt: "源内搜索")
+        .onSubmit(of: .search) { Task { await refresh() } }
         .refreshable { await refresh() }
         .toolbar {
             ToolbarItem(placement: .primaryAction) { Button("刷新", systemImage: "arrow.clockwise") { Task { await refresh() } } }
             ToolbarItem(placement: .primaryAction) { NavigationLink { RSSSourceEditorView(source: source) } label: { Label("编辑", systemImage: "pencil") } }
             ToolbarItem(placement: .primaryAction) { NavigationLink { RSSSourceDebugView(source: source) } label: { Label("调试", systemImage: "stethoscope") } }
-            ToolbarItem(placement: .bottomBar) { Button("加载下一页", systemImage: "arrow.down") { Task { await library.loadMore(source, sortName: activeName, key: searchKey.nilIfEmpty) } } }
+            ToolbarItem(placement: .bottomBar) {
+                Button("加载下一页", systemImage: "arrow.down") {
+                    Task { await library.loadMore(source, sortName: activeName, key: searchKey.nilIfEmpty) }
+                }.disabled(library.isWorking)
+            }
         }
         .task { sorts = await RSSSortResolver().resolve(source); await refresh() }
+        .readingTabBarHidden()
     }
 
     private var activeSort: RSSSort { sorts.indices.contains(selectedSort) ? sorts[selectedSort] : .init(name: "", url: source.sourceUrl) }
@@ -464,6 +470,32 @@ private struct RSSBinaryDocument: FileDocument {
     func fileWrapper(configuration: WriteConfiguration) throws -> FileWrapper { .init(regularFileWithContents: data) }
 }
 
+private struct RSSPhotoToolbar: ViewModifier {
+    let canSave: Bool
+    let save: () -> Void
+    let close: () -> Void
+
+    func body(content: Content) -> some View {
+        if #available(iOS 26.0, *) {
+            content.toolbar {
+                ToolbarItem(placement: .primaryAction) {
+                    Button("保存到照片", systemImage: "square.and.arrow.down", action: save).disabled(!canSave)
+                }.sharedBackgroundVisibility(.hidden)
+                ToolbarItem(placement: .confirmationAction) {
+                    Button("完成", action: close).buttonStyle(.plain)
+                }.sharedBackgroundVisibility(.hidden)
+            }
+        } else {
+            content.toolbar {
+                ToolbarItem(placement: .primaryAction) {
+                    Button("保存到照片", systemImage: "square.and.arrow.down", action: save).disabled(!canSave)
+                }
+                ToolbarItem(placement: .confirmationAction) { Button("完成", action: close) }
+            }
+        }
+    }
+}
+
 struct RSSPhotoView: View {
     @Environment(\.dismiss) private var dismiss
     let value: String
@@ -487,10 +519,7 @@ struct RSSPhotoView: View {
             }
             .navigationTitle(source.sourceName).navigationBarTitleDisplayMode(.inline)
             .readingNavigationBarBackground()
-            .toolbar {
+            .modifier(RSSPhotoToolbar(canSave: image != nil, save: save, close: { dismiss() }))
-                ToolbarItem(placement: .primaryAction) { Button("保存到照片", systemImage: "square.and.arrow.down") { save() }.disabled(image == nil) }
-                ToolbarItem(placement: .confirmationAction) { Button("完成") { dismiss() } }
-            }
             .task(id: value) { await load() }
             .alert("图片", isPresented: Binding(get: { message != nil }, set: { if !$0 { message = nil } })) { Button("好") { message = nil } } message: { Text(message ?? "") }
         }

App/Features/Reader/EPUBIllustrationsView.swift

@@ -0,0 +1,74 @@
+import SwiftUI
+import ImageIO
+
+struct EPUBIllustrationsView: View {
+    @Environment(\.dismiss) private var dismiss
+    let book: Book
+    let chapter: BookChapter
+    @State private var images: [EPUBIllustration] = []
+    @State private var isLoading = true
+    @State private var errorMessage: String?
+
+    var body: some View {
+        ReadingNavigationStack {
+            ScrollView {
+                LazyVStack(spacing: 24) {
+                    if isLoading {
+                        ProgressView("正在读取插图…").padding()
+                    } else if let errorMessage {
+                        ReadingUnavailableView("无法读取插图", systemImage: "photo", description: Text(errorMessage))
+                    } else if images.isEmpty {
+                        ReadingUnavailableView("本章没有插图", systemImage: "photo", description: Text(chapter.title))
+                    } else {
+                        ForEach(images) { image in
+                            VStack(spacing: 8) {
+                                EPUBIllustrationImage(data: image.data)
+                                Text("插图 \(image.id)\(image.caption.isEmpty ? "" : " · " + image.caption)")
+                                    .font(.caption).foregroundStyle(.secondary)
+                            }
+                        }
+                    }
+                }.padding()
+            }
+            .background(ReadingTheme.canvas)
+            .navigationTitle("本章插图")
+            .navigationBarTitleDisplayMode(.inline)
+            .toolbar { ToolbarItem(placement: .confirmationAction) { Button("完成") { dismiss() } } }
+            .task(id: chapter.id) {
+                isLoading = true
+                errorMessage = nil
+                do {
+                    images = try await Task.detached { try LocalBookService().loadIllustrations(book: book, chapter: chapter) }.value
+                } catch { errorMessage = error.localizedDescription }
+                isLoading = false
+            }
+        }
+    }
+}
+
+private struct EPUBIllustrationImage: View {
+    let data: Data
+    @State private var image: UIImage?
+
+    var body: some View {
+        Group {
+            if let image {
+                Image(uiImage: image).resizable().scaledToFit().accessibilityLabel("书籍插图")
+            } else {
+                Text("图片无法显示").foregroundStyle(.secondary)
+            }
+        }
+        .task {
+            // Decode a bounded thumbnail off the main thread, even for huge source dimensions.
+            image = await Task.detached {
+                guard let source = CGImageSourceCreateWithData(data as CFData, nil),
+                      let thumbnail = CGImageSourceCreateThumbnailAtIndex(source, 0, [
+                        kCGImageSourceCreateThumbnailFromImageAlways: true,
+                        kCGImageSourceThumbnailMaxPixelSize: 2_048,
+                        kCGImageSourceCreateThumbnailWithTransform: true
+                      ] as CFDictionary) else { return Optional<UIImage>.none }
+                return UIImage(cgImage: thumbnail)
+            }.value
+        }
+    }
+}

App/Features/Reader/ReaderScrollView.swift

@@ -129,7 +129,8 @@ struct ReaderScrollView: UIViewRepresentable {
                 let layout = view.collectionViewLayout as! ReaderContinuousLayout
                 let geometry = ReaderContinuousLayout.geometry(
                     blocks: newBlocks, width: view.bounds.width,
-                    top: CGFloat(config.paddingTop), bottom: CGFloat(config.paddingBottom), viewportHeight: view.bounds.height
+                    top: CGFloat(config.paddingTop), bottom: CGFloat(config.paddingBottom),
+                    viewportHeight: view.bounds.height, scale: view.traitCollection.displayScale
                 )
                 let targetY = anchor.flatMap { offset(for: $0, blocks: newBlocks, frames: geometry.frames) }
                 let correction = targetY.map { $0 - view.contentOffset.y } ?? 0
@@ -277,10 +278,14 @@ struct ReaderScrollView: UIViewRepresentable {
         private func captureAnchor() -> Anchor? {
             guard let view, !blocks.isEmpty else { return nil }
             let layout = view.collectionViewLayout as! ReaderContinuousLayout
-            let y = max(view.contentOffset.y, layout.frames.first?.minY ?? 0)
+            // The reading position starts below the configured top padding.
-            guard let row = layout.row(at: y) else { return nil }
+            // Preserve that reference point when earlier chapters are inserted.
+            let y = max(view.contentOffset.y + CGFloat(max(parent.configuration.paddingTop, 0)),
+                        layout.frames.first?.minY ?? 0)
+            let scale = max(view.traitCollection.displayScale, 1)
+            guard let row = layout.row(at: (y * scale).rounded() / scale) else { return nil }
             let block = blocks[row], chapter = block.chapter
-            let localY = y - layout.frames[row].minY + block.top
+            let localY = max(y - layout.frames[row].minY, 0) + block.top
             let glyph = chapter.manager.glyphIndex(for: CGPoint(x: 0, y: localY + 0.5), in: chapter.container)
             let safeGlyph = min(glyph, max(chapter.manager.numberOfGlyphs - 1, 0))
             let character = chapter.manager.characterIndexForGlyph(at: safeGlyph)
@@ -385,17 +390,21 @@ final class ReaderContinuousLayout: UICollectionViewLayout {
     private var size = CGSize.zero
     override var collectionViewContentSize: CGSize { size }
 
-    static func geometry(blocks: [ReaderScrollBlock], width: CGFloat, top: CGFloat, bottom: CGFloat, viewportHeight: CGFloat) -> (frames: [CGRect], size: CGSize) {
+    static func geometry(blocks: [ReaderScrollBlock], width: CGFloat, top: CGFloat, bottom: CGFloat, viewportHeight: CGFloat, scale: CGFloat) -> (frames: [CGRect], size: CGSize) {
-        var y = top
+        let scale = max(scale, 1)
+        // Keep insertion offsets in whole physical pixels. UIKit rounds scroll
+        // offsets, so fractional chapter heights otherwise accumulate drift.
+        var pixels = (top * scale).rounded()
         let frames = blocks.map { block -> CGRect in
-            let frame = CGRect(x: 0, y: y, width: width, height: block.height)
+            let height = ceil(block.height * scale)
-            y += block.height
+            let frame = CGRect(x: 0, y: pixels / scale, width: width, height: height / scale)
+            pixels += height
             return frame
         }
         let finalChapter = blocks.last?.chapter.chapter.id
         let finalStart = blocks.indices.first(where: { blocks[$0].chapter.chapter.id == finalChapter })
             .map { frames[$0].minY } ?? top
-        return (frames, CGSize(width: width, height: max(y, finalStart + viewportHeight) + bottom))
+        return (frames, CGSize(width: width, height: max(pixels / scale, finalStart + viewportHeight) + bottom))
     }
 
     func install(frames: [CGRect], size: CGSize, correction: CGFloat) {
@@ -412,9 +421,9 @@ final class ReaderContinuousLayout: UICollectionViewLayout {
         var low = 0, high = frames.count
         while low < high {
             let middle = (low + high) / 2
-            if frames[middle].maxY <= y { low = middle + 1 } else { high = middle }
+            if frames[middle].minY <= y { low = middle + 1 } else { high = middle }
         }
-        return min(low, frames.count - 1)
+        return max(low - 1, 0)
     }
 
     override func layoutAttributesForElements(in rect: CGRect) -> [UICollectionViewLayoutAttributes]? {

App/Features/Reader/ReaderView.swift

@@ -18,6 +18,7 @@ struct ReaderView: View {
     @StateObject private var speech = ReaderSpeechController()
     @State private var showsTableOfContents = false
     @State private var showsBookmarks = false
+    @State private var illustrationChapter: BookChapter?
     @State private var showsAutoReadControls = false
     @State private var showsStyleControls = false
     @State private var isAutoReading = false
@@ -257,6 +258,9 @@ struct ReaderView: View {
         }
         .sheet(isPresented: $showsTableOfContents) { tableOfContents }
         .sheet(isPresented: $showsBookmarks) { bookmarksSheet }
+        .sheet(item: $illustrationChapter) { chapter in
+            EPUBIllustrationsView(book: book, chapter: chapter)
+        }
         .sheet(isPresented: $showsAutoReadControls) { autoReadControls }
         .sheet(isPresented: $showsStyleControls) { styleControls }
         .sheet(isPresented: $showsContentSearch) { contentSearchSheet }
@@ -698,6 +702,7 @@ struct ReaderView: View {
             || showsReaderMenu
             || showsTableOfContents
             || showsBookmarks
+            || illustrationChapter != nil
             || showsAutoReadControls
             || showsStyleControls
             || showsContentSearch
@@ -834,6 +839,12 @@ struct ReaderView: View {
                 showsReaderMenu = false
                 showsContentSearch = true
             }
+            if book.origin == "local", URL(string: book.bookUrl)?.pathExtension.lowercased() == "epub" {
+                Button("本章插图", systemImage: "photo.on.rectangle") {
+                    showsReaderMenu = false
+                    illustrationChapter = model.currentChapter
+                }
+            }
             Button("书签列表", systemImage: "bookmark") {
                 showsReaderMenu = false
                 showsBookmarks = true
@@ -1575,7 +1586,7 @@ struct ReaderView: View {
                     Text("正文")
                 } footer: {
                     if !model.canEditContent {
-                        Text("本地 TXT 直接来自原文件；为避免破坏章节偏移，本章编辑与网络刷新仅对网络书籍开放。")
+                        Text("本地 TXT / EPUB 直接读取原文件。本章编辑与网络刷新仅对网络书籍开放。")
                     }
                 }
 

App/Features/Rules/Editor/SyntaxCodeEditor.swift

@@ -59,7 +59,12 @@ struct SyntaxCodeEditor: UIViewRepresentable {
             UIBarButtonItem(title: symbol == "\n" ? "↵" : symbol == "\t" ? "⇥" : symbol,
                 primaryAction: UIAction(title: "", subtitle: nil) { [weak view] _ in view?.insertText(symbol) })
         } + [UIBarButtonItem(systemItem: .flexibleSpace), UIBarButtonItem(title: "完成", primaryAction: UIAction(title: "", subtitle: nil) { [weak view] _ in view?.resignFirstResponder() })]
-        view.inputAccessoryView = bar
+        bar.items?.last?.accessibilityIdentifier = "syntaxCodeEditor.done"
+        if #available(iOS 26.0, *) {
+            view.inputAccessoryView = CodeInputAccessory(items: bar.items ?? [])
+        } else {
+            view.inputAccessoryView = bar
+        }
         view.text = text
         context.coordinator.highlight(view)
         return view
@@ -184,6 +189,66 @@ struct SyntaxCodeEditor: UIViewRepresentable {
     }
 }
 
+// Keep Done visible while allowing the editing shortcuts to scroll on narrow keyboards.
+private final class CodeInputAccessory: UIInputView {
+    init(items: [UIBarButtonItem]) {
+        super.init(frame: CGRect(x: 0, y: 0, width: 0, height: 44), inputViewStyle: .keyboard)
+        autoresizingMask = [.flexibleWidth]
+        backgroundColor = .secondarySystemGroupedBackground
+        let scroll = UIScrollView()
+        scroll.showsHorizontalScrollIndicator = false
+        scroll.delaysContentTouches = false
+        scroll.translatesAutoresizingMaskIntoConstraints = false
+        let stack = UIStackView()
+        stack.axis = .horizontal
+        stack.spacing = 4
+        stack.translatesAutoresizingMaskIntoConstraints = false
+        let actions = items.filter { $0.primaryAction != nil }
+        var done: UIButton?
+        for (index, item) in actions.enumerated() {
+            let button = UIButton(type: .system)
+            if let action = item.primaryAction { button.addAction(action, for: .touchUpInside) }
+            button.setImage(item.image, for: .normal)
+            button.setTitle(item.title, for: .normal)
+            button.tintColor = UIColor(ReadingTheme.accent)
+            button.titleLabel?.font = .systemFont(ofSize: 16)
+            button.accessibilityIdentifier = item.accessibilityIdentifier
+            button.accessibilityLabel = item.accessibilityLabel
+            button.translatesAutoresizingMaskIntoConstraints = false
+            if index == actions.count - 1 {
+                button.titleLabel?.font = .systemFont(ofSize: 16, weight: .semibold)
+                done = button
+                addSubview(button)
+            } else {
+                if index < 3 { button.accessibilityLabel = ["查找与替换", "撤销", "重做"][index] }
+                stack.addArrangedSubview(button)
+                button.widthAnchor.constraint(equalToConstant: 36).isActive = true
+            }
+        }
+        addSubview(scroll)
+        scroll.addSubview(stack)
+        guard let done else { return }
+        NSLayoutConstraint.activate([
+            done.trailingAnchor.constraint(equalTo: trailingAnchor, constant: -8),
+            done.widthAnchor.constraint(equalToConstant: 60),
+            done.topAnchor.constraint(equalTo: topAnchor),
+            done.bottomAnchor.constraint(equalTo: bottomAnchor),
+            scroll.leadingAnchor.constraint(equalTo: leadingAnchor, constant: 4),
+            scroll.trailingAnchor.constraint(equalTo: done.leadingAnchor, constant: -4),
+            scroll.topAnchor.constraint(equalTo: topAnchor),
+            scroll.bottomAnchor.constraint(equalTo: bottomAnchor),
+            stack.leadingAnchor.constraint(equalTo: scroll.contentLayoutGuide.leadingAnchor),
+            stack.trailingAnchor.constraint(equalTo: scroll.contentLayoutGuide.trailingAnchor),
+            stack.topAnchor.constraint(equalTo: scroll.contentLayoutGuide.topAnchor),
+            stack.bottomAnchor.constraint(equalTo: scroll.contentLayoutGuide.bottomAnchor),
+            stack.heightAnchor.constraint(equalTo: scroll.frameLayoutGuide.heightAnchor)
+        ])
+    }
+
+    required init?(coder: NSCoder) { fatalError("init(coder:) has not been implemented") }
+    override var intrinsicContentSize: CGSize { CGSize(width: UIView.noIntrinsicMetric, height: 44) }
+}
+
 struct RuleCodeFieldView: View {
     let title: String
     let path: String

App/Features/Rules/ReplaceRuleEditorView.swift

@@ -92,9 +92,10 @@ struct ReplaceRuleEditorView: View {
                     RuleCodeFieldView(title: "匹配内容", path: "pattern", text: $rule.pattern, language: .regex)
                 } label: { fieldPreview("匹配内容", value: rule.pattern) }
                 NavigationLink {
-                    RuleCodeFieldView(title: "替换为", path: "replacement", text: $rule.replacement, language: .regex)
+                    RuleCodeFieldView(title: "替换为", path: "replacement", text: $rule.replacement,
+                        language: rule.isRegex && rule.replacement.hasPrefix("@js:") ? .javascript : .regex)
                 } label: { fieldPreview("替换为（留空即删除）", value: rule.replacement) }
-                Text("正则替换使用 $1、$2 引用捕获组；普通文本替换按字面匹配。")
+                Text("正则替换使用 $1、$2 引用捕获组；以 @js: 开头可运行 JavaScript，result 为当前匹配文本，可读取 book 和 chapter。普通文本替换按字面匹配。")
                     .font(.footnote).foregroundStyle(.secondary)
             }
             Section("作用范围") {

App/Features/Search/SearchViewModel.swift

@@ -64,7 +64,8 @@ final class SearchViewModel: ObservableObject {
         pendingSourceIDs = Set(pendingSources.map(\.id))
         pagedSources = pendingSources.filter {
             let url = $0.searchUrl ?? ""
-            return url.contains("{{page}}") || url.contains("searchPage") || url.contains("<")
+            return url.range(of: #"\{\{[^}]*\bpage\b[^}]*\}\}"#, options: .regularExpression) != nil
+                || url.contains("searchPage") || url.contains("<")
         }
         sourceNames = Dictionary(pendingSources.map { ($0.id, $0.bookSourceName) }, uniquingKeysWith: { first, _ in first })
         accumulator = .init()

App/Features/Settings/AboutView.swift

@@ -80,6 +80,7 @@ private struct OpenSourceLicensesView: View {
             license("小浣熊阅读 / Legado", resource: "Legado-LICENSE", detail: "GNU GPL v3")
             license("SwiftSoup", resource: "SwiftSoup-LICENSE", detail: "MIT License")
             license("Kanna", resource: "Kanna-LICENSE", detail: "MIT License")
+            license("ZIPFoundation", resource: "ZIPFoundation-LICENSE", detail: "MIT License")
         }.navigationTitle("开源许可").navigationBarTitleDisplayMode(.inline)
     }
 

App/Info.plist

@@ -22,6 +22,24 @@
 				<string>public.data</string>
 			</array>
 		</dict>
+		<dict>
+			<key>CFBundleTypeName</key><string>EPUB 电子书</string>
+			<key>CFBundleTypeRole</key><string>Viewer</string>
+			<key>LSHandlerRank</key><string>Alternate</string>
+			<key>LSItemContentTypes</key><array><string>org.idpf.epub-container</string></array>
+		</dict>
+	</array>
+	<key>UTImportedTypeDeclarations</key>
+	<array>
+		<dict>
+			<key>UTTypeIdentifier</key><string>org.idpf.epub-container</string>
+			<key>UTTypeDescription</key><string>EPUB 电子书</string>
+			<key>UTTypeConformsTo</key><array><string>public.data</string></array>
+			<key>UTTypeTagSpecification</key><dict>
+				<key>public.filename-extension</key><array><string>epub</string></array>
+				<key>public.mime-type</key><string>application/epub+zip</string>
+			</dict>
+		</dict>
 	</array>
 	<key>CFBundleExecutable</key>
 	<string>$(EXECUTABLE_NAME)</string>

App/LegadoApp.swift

@@ -52,6 +52,7 @@ struct LegadoApp: App {
                     await bookLibrary.load()
                     await replaceRuleLibrary.load()
                     await readingActivity.load()
+                    await rssLibrary.load()
                     await ruleSubscriptions.load()
                     if UserDefaults.standard.bool(forKey: "auto_refresh") {
                         await bookLibrary.checkForUpdates(sources: sourceLibrary.sources, onlyRead: UserDefaults.standard.bool(forKey: "onlyUpdateRead"))
@@ -60,6 +61,15 @@ struct LegadoApp: App {
                 }
                 .onOpenURL { url in
                     Task {
+                        if url.isFileURL && url.pathExtension.lowercased() == "epub" {
+                            let file = LocalBookImportFile(url: url)
+                            await bookLibrary.load()
+                            await bookLibrary.importPickedLocalFiles([file])
+                            sourceLibrary.notice = .init(kind: bookLibrary.errorMessage == nil ? .success : .error,
+                                                         title: bookLibrary.errorMessage == nil ? "EPUB 已加入书架" : "导入失败",
+                                                         detail: bookLibrary.errorMessage ?? url.lastPathComponent)
+                            return
+                        }
                         await sourceLibrary.load()
                         if url.scheme?.lowercased() == "legado", url.path.lowercased() == "/replacerule" {
                             await replaceRuleLibrary.load()

App/Resources/ZIPFoundation-LICENSE.txt

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2017-2025 Thomas Zoechling (https://www.peakstep.com)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

App/RootTabView.swift

@@ -227,7 +227,7 @@ private struct BookshelfView: View {
         }
         .background(ReadingTheme.canvas)
         .navigationBarHidden(true)
-        .readingTXTFileImporter(isPresented: $showsLocalImporter) { files in
+        .readingLocalBookFileImporter(isPresented: $showsLocalImporter) { files in
             pendingLocalFiles = files
         } onError: { error in
             library.reportError(error)
@@ -449,7 +449,7 @@ private struct BookshelfBookDestination: View {
     let source: BookSource?
 
     var body: some View {
-        if ProcessInfo.processInfo.arguments.contains("-uiDemo") {
+        if ProcessInfo.processInfo.arguments.contains("-uiDemo"), book.origin != "local" {
             ReaderView(demoBook: book, chapters: DemoContent.readerChapters, initialIndex: 0, content: DemoContent.readerText)
         } else {
             BookshelfReadView(book: book, source: source)
@@ -912,6 +912,8 @@ struct SearchView: View {
                 } label: {
                     Label("搜索选项", systemImage: "slider.horizontal.3")
                 }
+                .accessibilityLabel("搜索选项")
+                .accessibilityIdentifier("search.options")
             }
         }
         .sheet(isPresented: $showsScope) {

CHANGELOG.md

@@ -1,5 +1,16 @@
 # 更新记录
 
+## 0.1.12（2026-10-24）
+
+- 增加无 DRM EPUB 2/3 导入和阅读，识别元数据、封面、NCX/nav 目录及同页章节锚点，并按书脊顺序保留未列入目录的正文。
+- EPUB 沿用阅读进度、书签、朗读和阅读样式；正文中的插图可从「更多 → 本章插图」查看。
+- 本地文件管理支持 EPUB 重命名、原格式导出、删除及从系统文件应用打开。
+
+- 改善 JSON 规则模板、候选选择、组合替换和请求参数兼容，修复 JavaScript 数字类型与动态净化替换。
+- 修复 RSS 翻页失败后的重试、切换订阅时的过期结果和重复翻页请求。
+- 改善连续滚动中的章节插入定位，调整 iOS 26 代码编辑工具栏。
+- 应用版本为 `0.1.12`，构建号为 `18`。
+
 ## 0.1.11（2026-10-22）
 
 - 修复书架和文件管理中的本地 TXT 选择、导入确认与文件访问权限，支持大小写扩展名和多选，并排除 JSON 书源。

Legado.xcodeproj/project.pbxproj

@@ -14,6 +14,7 @@
 		B10000000000000000000029 /* ReaderScrollView.swift in Sources */ = {isa = PBXBuildFile; fileRef = B20000000000000000000029 /* ReaderScrollView.swift */; };
 		F10000000000000000000001 /* SwiftSoup in Frameworks */ = {isa = PBXBuildFile; productRef = F20000000000000000000001 /* SwiftSoup */; };
 		F10000000000000000000002 /* Kanna in Frameworks */ = {isa = PBXBuildFile; productRef = F20000000000000000000002 /* Kanna */; };
+		F10000000000000000000003 /* ZIPFoundation in Frameworks */ = {isa = PBXBuildFile; productRef = F20000000000000000000003 /* ZIPFoundation */; };
 		A10000000000000000000001 /* LegadoApp.swift in Sources */ = {isa = PBXBuildFile; fileRef = A20000000000000000000001 /* LegadoApp.swift */; };
 		A10000000000000000000002 /* RootTabView.swift in Sources */ = {isa = PBXBuildFile; fileRef = A20000000000000000000002 /* RootTabView.swift */; };
 		A10000000000000000000003 /* SourceLibraryModel.swift in Sources */ = {isa = PBXBuildFile; fileRef = A20000000000000000000003 /* SourceLibraryModel.swift */; };
@@ -123,6 +124,9 @@
 		AA1100000000000000000008 = {isa = PBXBuildFile; fileRef = AA2200000000000000000008; };
 		AA1100000000000000000009 = {isa = PBXBuildFile; fileRef = AA2200000000000000000009; };
 		AA1100000000000000000010 = {isa = PBXBuildFile; fileRef = AA2200000000000000000010; };
+		EB1000000000000000000001 /* EPUBBook.swift in Sources */ = {isa = PBXBuildFile; fileRef = EB2000000000000000000001; };
+		EB1000000000000000000002 /* ZIPFoundation-LICENSE.txt in Resources */ = {isa = PBXBuildFile; fileRef = EB2000000000000000000002; };
+		EB1000000000000000000003 /* EPUBIllustrationsView.swift in Sources */ = {isa = PBXBuildFile; fileRef = EB2000000000000000000003; };
 /* End PBXBuildFile section */
 
 /* Begin PBXFileReference section */
@@ -245,6 +249,9 @@
 		AA2200000000000000000008 = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Sources/LegadoCore/Import/QRCodeDecoder.swift; sourceTree = SOURCE_ROOT; };
 		AA2200000000000000000009 = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Tests/LegadoReaderUITests/RuleEditorUITests.swift; sourceTree = SOURCE_ROOT; };
 		AA2200000000000000000010 = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Tests/LegadoReaderUITests/LiveReadingFlowUITests.swift; sourceTree = SOURCE_ROOT; };
+		EB2000000000000000000001 /* EPUBBook.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Sources/LegadoCore/LocalBook/EPUBBook.swift; sourceTree = SOURCE_ROOT; };
+		EB2000000000000000000002 /* ZIPFoundation-LICENSE.txt */ = {isa = PBXFileReference; lastKnownFileType = text; path = App/Resources/ZIPFoundation-LICENSE.txt; sourceTree = SOURCE_ROOT; };
+		EB2000000000000000000003 /* EPUBIllustrationsView.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = App/Features/Reader/EPUBIllustrationsView.swift; sourceTree = SOURCE_ROOT; };
 /* End PBXFileReference section */
 
 /* Begin PBXFrameworksBuildPhase section */
@@ -256,6 +263,7 @@
 			files = (
 				F10000000000000000000001 /* SwiftSoup in Frameworks */,
 				F10000000000000000000002 /* Kanna in Frameworks */,
+				F10000000000000000000003 /* ZIPFoundation in Frameworks */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};
@@ -353,6 +361,7 @@
 				A20000000000000000000029 /* ReaderSpeechController.swift */,
 				A20000000000000000000028 /* ReaderAppearance.swift */,
 				A2000000000000000000002B /* ReaderSupport.swift */,
+				EB2000000000000000000003 /* EPUBIllustrationsView.swift */,
 			);
 			path = Features/Reader;
 			sourceTree = "<group>";
@@ -527,6 +536,7 @@
 			children = (
 				F21000000000000000000002 /* LocalFileService.swift */,
 				C20000000000000000000018 /* LocalBookService.swift */,
+				EB2000000000000000000001 /* EPUBBook.swift */,
 			);
 			path = LocalBook;
 			sourceTree = "<group>";
@@ -589,6 +599,7 @@
 				E20000000000000000000008 /* Legado-LICENSE.txt */,
 				E20000000000000000000009 /* SwiftSoup-LICENSE.txt */,
 				E2000000000000000000000A /* Kanna-LICENSE.txt */,
+				EB2000000000000000000002 /* ZIPFoundation-LICENSE.txt */,
 			);
 			path = Resources;
 			sourceTree = "<group>";
@@ -622,6 +633,7 @@
 			packageProductDependencies = (
 				F20000000000000000000001 /* SwiftSoup */,
 				F20000000000000000000002 /* Kanna */,
+				F20000000000000000000003 /* ZIPFoundation */,
 			);
 			productName = Legado;
 			productReference = D20000000000000000000001 /* Legado.app */;
@@ -659,6 +671,7 @@
 			packageReferences = (
 				F30000000000000000000001 /* XCRemoteSwiftPackageReference "SwiftSoup" */,
 				F30000000000000000000002 /* XCRemoteSwiftPackageReference "Kanna" */,
+				F30000000000000000000003 /* XCRemoteSwiftPackageReference "ZIPFoundation" */,
 			);
 			productRefGroup = D40000000000000000000002 /* Products */;
 			projectDirPath = "";
@@ -677,6 +690,7 @@
 				910000000000000000000008 /* Legado-LICENSE.txt in Resources */,
 				A10000000000000000000020 /* Assets.xcassets in Resources */,
 				A1000000000000000000001D /* txtTocRule.json in Resources */,
+				EB1000000000000000000002 /* ZIPFoundation-LICENSE.txt in Resources */,
 				A1000000000000000000001E /* httpTTS.json in Resources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
@@ -770,6 +784,8 @@
 				C10000000000000000000016 /* BookExploreService.swift in Sources */,
 				C10000000000000000000017 /* BookSourceDebugger.swift in Sources */,
 				C10000000000000000000018 /* LocalBookService.swift in Sources */,
+				EB1000000000000000000003 /* EPUBIllustrationsView.swift in Sources */,
+				EB1000000000000000000001 /* EPUBBook.swift in Sources */,
 				C10000000000000000000019 /* ContentSanitizer.swift in Sources */,
 				C1000000000000000000001A /* ReplaceRuleRepository.swift in Sources */,
 				C1000000000000000000001B /* ReadingActivityRepository.swift in Sources */,
@@ -864,12 +880,12 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CODE_SIGN_STYLE = Automatic;
-				CURRENT_PROJECT_VERSION = 17;
+				CURRENT_PROJECT_VERSION = 18;
 				DEVELOPMENT_ASSET_PATHS = "";
 				GENERATE_INFOPLIST_FILE = NO;
 				INFOPLIST_FILE = App/Info.plist;
 				IPHONEOS_DEPLOYMENT_TARGET = 15.0;
-				MARKETING_VERSION = 0.1.11;
+				MARKETING_VERSION = 0.1.12;
 				OTHER_LDFLAGS = "$(inherited) -lsqlite3";
 				PRODUCT_BUNDLE_IDENTIFIER = com.epiboly.legado;
 				PRODUCT_NAME = "$(TARGET_NAME)";
@@ -885,11 +901,11 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CODE_SIGN_STYLE = Automatic;
-				CURRENT_PROJECT_VERSION = 17;
+				CURRENT_PROJECT_VERSION = 18;
 				GENERATE_INFOPLIST_FILE = NO;
 				INFOPLIST_FILE = App/Info.plist;
 				IPHONEOS_DEPLOYMENT_TARGET = 15.0;
-				MARKETING_VERSION = 0.1.11;
+				MARKETING_VERSION = 0.1.12;
 				OTHER_LDFLAGS = "$(inherited) -lsqlite3";
 				PRODUCT_BUNDLE_IDENTIFIER = com.epiboly.legado;
 				PRODUCT_NAME = "$(TARGET_NAME)";
@@ -930,6 +946,11 @@
 			repositoryURL = "https://github.com/tid-kijyun/Kanna.git";
 			requirement = { kind = exactVersion; version = 6.1.0; };
 		};
+		F30000000000000000000003 /* XCRemoteSwiftPackageReference "ZIPFoundation" */ = {
+			isa = XCRemoteSwiftPackageReference;
+			repositoryURL = "https://github.com/weichsel/ZIPFoundation.git";
+			requirement = { kind = exactVersion; version = 0.9.20; };
+		};
 /* End XCRemoteSwiftPackageReference section */
 
 /* Begin XCSwiftPackageProductDependency section */
@@ -943,6 +964,11 @@
 			package = F30000000000000000000002 /* XCRemoteSwiftPackageReference "Kanna" */;
 			productName = Kanna;
 		};
+		F20000000000000000000003 /* ZIPFoundation */ = {
+			isa = XCSwiftPackageProductDependency;
+			package = F30000000000000000000003 /* XCRemoteSwiftPackageReference "ZIPFoundation" */;
+			productName = ZIPFoundation;
+		};
 /* End XCSwiftPackageProductDependency section */
 	};
 	rootObject = D70000000000000000000001 /* Project object */;

Legado.xcodeproj/project.xcworkspace/xcshareddata/swiftpm/Package.resolved

@@ -1,5 +1,5 @@
 {
-  "originHash" : "5cf2eb5dea5b10523b3afc86a164d9a3315a7269a1aefdfe8c91f0530c3f3522",
+  "originHash" : "a20ba014518540ad01fe4db62ea950304db8c0b8eb880b62b87a5f1d19569a3c",
   "pins" : [
     {
       "identity" : "kanna",
@@ -18,6 +18,15 @@
         "revision" : "49dcadd93161f4a44b4994d3a3e8de9f085aface",
         "version" : "2.13.5"
       }
+    },
+    {
+      "identity" : "zipfoundation",
+      "kind" : "remoteSourceControl",
+      "location" : "https://github.com/weichsel/ZIPFoundation.git",
+      "state" : {
+        "revision" : "22787ffb59de99e5dc1fbfe80b19c97a904ad48d",
+        "version" : "0.9.20"
+      }
     }
   ],
   "version" : 3

Legado.xcodeproj/xcshareddata/xcschemes/LegadoRSSFlowUITests.xcscheme

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Scheme LastUpgradeVersion="2610" version="1.7">
+  <BuildAction parallelizeBuildables="YES" buildImplicitDependencies="YES"><BuildActionEntries><BuildActionEntry buildForTesting="YES" buildForRunning="NO" buildForProfiling="NO" buildForArchiving="NO" buildForAnalyzing="YES"><BuildableReference BuildableIdentifier="primary" BlueprintIdentifier="B50000000000000000000050" BuildableName="LegadoReaderUITests.xctest" BlueprintName="LegadoReaderUITests" ReferencedContainer="container:Legado.xcodeproj"/></BuildActionEntry></BuildActionEntries></BuildAction>
+  <TestAction buildConfiguration="Debug" selectedDebuggerIdentifier="Xcode.DebuggerFoundation.Debugger.LLDB" selectedLauncherIdentifier="Xcode.DebuggerFoundation.Launcher.LLDB" shouldUseLaunchSchemeArgsEnv="NO">
+    <Testables><TestableReference skipped="NO"><BuildableReference BuildableIdentifier="primary" BlueprintIdentifier="B50000000000000000000050" BuildableName="LegadoReaderUITests.xctest" BlueprintName="LegadoReaderUITests" ReferencedContainer="container:Legado.xcodeproj"/><SelectedTests><Test Identifier="RuleEditorUITests/testRSSSearchSubmissionPaginationAndFavorites()"/></SelectedTests></TestableReference></Testables>
+    <EnvironmentVariables><EnvironmentVariable key="LEGADO_RSS_UI_FIXTURE_URL" value="http://127.0.0.1:18763/source.json" isEnabled="YES"/></EnvironmentVariables>
+  </TestAction>
+</Scheme>

Package.resolved

@@ -1,5 +1,5 @@
 {
-  "originHash" : "7f0bb8624f715c1c036934addf347a13f5c877af397c008dd678bb519c74d937",
+  "originHash" : "9d25b5f7ff3ac25767fea9b5152f5607bdb185299d3ebda18cd9b9255a772731",
   "pins" : [
     {
       "identity" : "kanna",
@@ -18,6 +18,15 @@
         "revision" : "49dcadd93161f4a44b4994d3a3e8de9f085aface",
         "version" : "2.13.5"
       }
+    },
+    {
+      "identity" : "zipfoundation",
+      "kind" : "remoteSourceControl",
+      "location" : "https://github.com/weichsel/ZIPFoundation.git",
+      "state" : {
+        "revision" : "22787ffb59de99e5dc1fbfe80b19c97a904ad48d",
+        "version" : "0.9.20"
+      }
     }
   ],
   "version" : 3

Package.swift

@@ -12,14 +12,16 @@ let package = Package(
     ],
     dependencies: [
         .package(url: "https://github.com/scinfu/SwiftSoup.git", exact: "2.13.5"),
-        .package(url: "https://github.com/tid-kijyun/Kanna.git", exact: "6.1.0")
+        .package(url: "https://github.com/tid-kijyun/Kanna.git", exact: "6.1.0"),
+        .package(url: "https://github.com/weichsel/ZIPFoundation.git", exact: "0.9.20")
     ],
     targets: [
         .target(
             name: "LegadoCore",
             dependencies: [
                 .product(name: "SwiftSoup", package: "SwiftSoup"),
-                .product(name: "Kanna", package: "Kanna")
+                .product(name: "Kanna", package: "Kanna"),
+                .product(name: "ZIPFoundation", package: "ZIPFoundation")
             ],
             linkerSettings: [
                 .linkedFramework("JavaScriptCore"),
@@ -29,7 +31,7 @@ let package = Package(
         ),
         .testTarget(
             name: "LegadoCoreTests",
-            dependencies: ["LegadoCore"],
+            dependencies: ["LegadoCore", .product(name: "ZIPFoundation", package: "ZIPFoundation")],
             resources: [.copy("Fixtures")]
         ),
         .target(
@@ -42,7 +44,7 @@ let package = Package(
                 "Rules/ReplaceRuleManagementView.swift", "Rules/ReplaceRuleEditorView.swift", "Rules/RuleSubscriptionView.swift", "Rules/TxtTocRuleManagementView.swift", "Rules/Editor",
                 "Books/BookDetailView.swift", "Books/LocalBookDetailView.swift", "Books/SourceSwitchView.swift",
                 "Reader/ReaderAppearance.swift", "Reader/ReaderSpeechController.swift", "Reader/ReaderSupport.swift",
-                "Reader/ReaderTextView.swift", "Reader/ReaderScrollView.swift", "Reader/ReaderView.swift"
+                "Reader/EPUBIllustrationsView.swift", "Reader/ReaderTextView.swift", "Reader/ReaderScrollView.swift", "Reader/ReaderView.swift"
             ],
             sources: [
                 "Sources/SourceLibraryModel.swift", "Rules/ReplaceRuleLibraryModel.swift", "Search/SearchViewModel.swift",
@@ -52,7 +54,7 @@ let package = Package(
         ),
         .testTarget(
             name: "LegadoAppModelsTests",
-            dependencies: ["LegadoAppModels", "LegadoCore"]
+            dependencies: ["LegadoAppModels", "LegadoCore", .product(name: "ZIPFoundation", package: "ZIPFoundation")]
         )
     ]
 )

README.md

@@ -20,7 +20,7 @@
 
 项目仍在边用边改，书源兼容和阅读体验会继续完善。
 
-**小浣熊阅读不提供书源或在线书库。** 书源需由用户自行从网络公开分享获取，或自行编写后导入；也可以直接阅读本地 TXT。
+**小浣熊阅读不提供书源或在线书库。** 书源需由用户自行从网络公开分享获取，或自行编写后导入；也可以直接阅读本地 TXT / EPUB。
 
 ## 下载安装
 
@@ -42,7 +42,7 @@
 | 快速搜书 | 多书源并发搜索，结果陆续展示；重复搜索可先显示有效缓存，同书多源合并 |
 | 书架管理 | 列表、网格、分组、章节更新、换源和本机阅读进度 |
 | 阅读排版 | 字号、间距、边距、日夜主题、滚动与分页阅读、目录和书签 |
-| 本地 TXT | 导入本地文件、自定义目录规则和章节阅读 |
+| 本地 TXT / EPUB | TXT 自定义目录规则；EPUB 2/3 目录、封面、分章阅读与插图查看 |
 | 语音朗读 | 系统朗读与 HTTP TTS 配置 |
 | 阅读工具 | 替换净化、阅读记录、时长统计和缓存管理 |
 | RSS 订阅 | 订阅源管理与内容阅读 |
@@ -85,9 +85,12 @@
 
 1. 自行准备你有权使用的阅读书源，安装后进入「我的 → 书源管理」导入。
 2. 在书架点击搜索，输入书名或作者，打开书籍详情后开始阅读。
-3. 将书籍加入书架，方便下次继续阅读；也可以从书架直接导入本地 TXT。
+3. 将书籍加入书架，方便下次继续阅读；也可以从书架直接导入本地 TXT 或 EPUB。
+
 4. 在阅读页调整排版、主题、书签和朗读设置。
 
+EPUB 导入入口：书架 → 更多 → 本地书，或「我的 → 文件管理 → 导入 TXT / EPUB」。支持无 DRM 的 EPUB 2/3，识别书名、作者、封面、NCX/nav 目录和章节锚点。正文使用现有阅读器重新排版，支持阅读进度、书签、朗读和样式设置；插图从阅读菜单「更多 → 本章插图」查看。暂不还原原书 CSS、固定版式和交互内容。
+
 应用支持 `legado://` 规则导入链接。来源网站改版或停止服务时，书源规则可能需要更新。
 
 ## 书源说明

Sources/LegadoCore/Book/ContentSanitizer.swift

@@ -115,9 +115,9 @@ public struct ContentSanitizer: Sendable {
         var diagnostics: [RuleDiagnostic] = []
 
         for rule in ordered where rule.scopeTitle && applies(rule, to: book) {
-            let result = apply(rule, to: title)
+            let result = apply(rule, to: title, book: book, chapter: chapter)
             diagnostics += result.diagnostics
-            if result.value != title {
+            if result.value != title, !result.value.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
                 title = result.value
                 effective.append(rule.id)
             }
@@ -146,7 +146,7 @@ public struct ContentSanitizer: Sendable {
         }
 
         for rule in ordered where rule.scopeContent && applies(rule, to: book) {
-            let result = apply(rule, to: body)
+            let result = apply(rule, to: body, book: book, chapter: chapter)
             diagnostics += result.diagnostics
             if result.value != body {
                 body = result.value
@@ -172,10 +172,16 @@ public struct ContentSanitizer: Sendable {
         return (!book.name.isEmpty && scope.contains(book.name)) || (!book.origin.isEmpty && scope.contains(book.origin))
     }
 
-    private func apply(_ rule: ReplaceRule, to input: String) -> (value: String, diagnostics: [RuleDiagnostic]) {
+    private func apply(_ rule: ReplaceRule, to input: String, book: Book, chapter: BookChapter) -> (value: String, diagnostics: [RuleDiagnostic]) {
         guard !rule.pattern.isEmpty else { return (input, []) }
         do {
             if rule.isRegex {
+                if rule.replacement.hasPrefix("@js:") {
+                    let evaluate = JavaScriptRuleEvaluator().replacementEvaluator(
+                        script: String(rule.replacement.dropFirst(4)), book: book, chapter: chapter)
+                    return (try regex.replacingStrict(pattern: rule.pattern, in: input,
+                        timeoutMilliseconds: rule.timeoutMillisecond, replacement: evaluate), [])
+                }
                 return (try regex.replacingStrict(pattern: rule.pattern, in: input, with: rule.replacement, timeoutMilliseconds: rule.timeoutMillisecond), [])
             }
             return (input.replacingOccurrences(of: rule.pattern, with: rule.replacement), [])

Sources/LegadoCore/LocalBook/EPUBBook.swift

@@ -0,0 +1,399 @@
+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()
+    }
+}

Sources/LegadoCore/LocalBook/LocalBookService.swift

@@ -10,7 +10,7 @@ public enum LocalBookError: Error, LocalizedError, Sendable {
 
     public var errorDescription: String? {
         switch self {
-        case .unsupportedFormat: "当前仅支持 TXT 本地书"
+        case .unsupportedFormat: "当前支持 TXT 和 EPUB 本地书"
         case .oversized: "本地书超过当前配置允许的大小"
         case .unreadableEncoding: "无法识别 TXT 编码"
         case .missingFile: "本地书文件已丢失"
@@ -104,6 +104,49 @@ public struct LocalBookService: Sendable {
         LocalBookService(tocRules: TxtTocRuleDefaults.load())
     }
 
+    public static func supports(_ url: URL) -> Bool {
+        url.isFileURL && ["txt", "epub"].contains(url.pathExtension.lowercased())
+    }
+
+    public func openFile(at url: URL) throws -> LocalBookImportReport {
+        guard Self.supports(url) else { throw LocalBookError.unsupportedFormat }
+        if url.pathExtension.lowercased() == "epub" {
+            try validateFile(url)
+            return try EPUBBook(url: url).report(url: url, originalName: url.lastPathComponent)
+        }
+        return try openTextFile(at: url)
+    }
+
+    public func importFile(at source: URL, into directory: URL) throws -> LocalBookImportReport {
+        guard Self.supports(source) else { throw LocalBookError.unsupportedFormat }
+        try validateFile(source)
+        try FileManager.default.createDirectory(at: directory, withIntermediateDirectories: true)
+        let destination = directory.appendingPathComponent(UUID().uuidString + "." + source.pathExtension.lowercased())
+        try FileManager.default.copyItem(at: source, to: destination)
+        do {
+            if source.pathExtension.lowercased() == "epub" {
+                return try EPUBBook(url: destination).report(url: destination, originalName: source.lastPathComponent)
+            }
+            return try textFileReport(data: Data(contentsOf: destination, options: .mappedIfSafe), sourceURL: source, destination: destination)
+        } catch {
+            try? FileManager.default.removeItem(at: destination)
+            throw error
+        }
+    }
+
+    public func loadIllustrations(book: Book, chapter: BookChapter) throws -> [EPUBIllustration] {
+        guard let url = URL(string: book.bookUrl), url.isFileURL, url.pathExtension.lowercased() == "epub" else { return [] }
+        try validateFile(url)
+        return try EPUBBook(url: url).illustrations(chapter: chapter)
+    }
+
+    private func validateFile(_ url: URL) throws {
+        guard FileManager.default.fileExists(atPath: url.path) else { throw LocalBookError.missingFile }
+        let values = try url.resourceValues(forKeys: [.isRegularFileKey, .fileSizeKey])
+        guard values.isRegularFile == true else { throw LocalBookError.missingFile }
+        guard Int64(values.fileSize ?? 0) <= maximumFileSize else { throw LocalBookError.oversized }
+    }
+
     public func importTextFile(at sourceURL: URL, into directory: URL) throws -> LocalBookImportReport {
         guard sourceURL.pathExtension.lowercased() == "txt" else { throw LocalBookError.unsupportedFormat }
         let values = try sourceURL.resourceValues(forKeys: [.fileSizeKey, .isRegularFileKey])
@@ -181,6 +224,10 @@ public struct LocalBookService: Sendable {
     }
 
     public func loadChapters(book: Book) throws -> ChapterListReport {
+        if let url = URL(string: book.bookUrl), url.isFileURL, url.pathExtension.lowercased() == "epub" {
+            try validateFile(url)
+            return try .init(chapters: EPUBBook(url: url).chapters(book: book))
+        }
         let decoded = try read(book)
         let selectionVolumeState = TxtVolumeState()
         let probeText = textProbe(decoded.data, encoding: decoded.encoding)
@@ -201,6 +248,10 @@ public struct LocalBookService: Sendable {
     }
 
     public func loadContent(book: Book, chapter: BookChapter) throws -> ChapterContentReport {
+        if let url = URL(string: book.bookUrl), url.isFileURL, url.pathExtension.lowercased() == "epub" {
+            try validateFile(url)
+            return try EPUBBook(url: url).content(chapter: chapter)
+        }
         guard let url = URL(string: book.bookUrl), url.isFileURL,
               FileManager.default.fileExists(atPath: url.path) else { throw LocalBookError.missingFile }
         let fileSize = (try? url.resourceValues(forKeys: [.fileSizeKey]).fileSize) ?? 0

Sources/LegadoCore/LocalBook/LocalFileService.swift

@@ -32,7 +32,7 @@ public struct StagedLocalFile: Sendable {
 public actor LocalFileService {
     public let directory: URL
 
-    public init(directory: URL) { self.directory = directory.standardizedFileURL }
+    public init(directory: URL) { self.directory = URL(fileURLWithPath: directory.path, isDirectory: true).standardizedFileURL }
 
     public func list() throws -> [LocalFileEntry] {
         guard FileManager.default.fileExists(atPath: directory.path) else { return [] }
@@ -41,13 +41,13 @@ public actor LocalFileService {
             options: [.skipsHiddenFiles]
         ).compactMap { url in
             let values = try url.resourceValues(forKeys: [.isRegularFileKey, .isSymbolicLinkKey, .fileSizeKey, .contentModificationDateKey])
-            guard values.isRegularFile == true, values.isSymbolicLink != true, url.pathExtension.lowercased() == "txt" else { return nil }
+            guard values.isRegularFile == true, values.isSymbolicLink != true, LocalBookService.supports(url) else { return nil }
             return .init(url: url, bytes: Int64(values.fileSize ?? 0), modified: values.contentModificationDate)
         }.sorted { $0.url.lastPathComponent.localizedStandardCompare($1.url.lastPathComponent) == .orderedAscending }
     }
 
     public func importFile(from source: URL) throws -> URL {
-        guard source.pathExtension.lowercased() == "txt" else { throw LocalBookError.unsupportedFormat }
+        guard LocalBookService.supports(source) else { throw LocalBookError.unsupportedFormat }
         let values = try source.resourceValues(forKeys: [.isRegularFileKey])
         guard values.isRegularFile == true else { throw LocalFileError.missingFile }
         try FileManager.default.createDirectory(at: directory, withIntermediateDirectories: true)
@@ -55,7 +55,7 @@ public actor LocalFileService {
         var target = directory.appendingPathComponent(source.lastPathComponent)
         var suffix = 2
         while FileManager.default.fileExists(atPath: target.path) {
-            target = directory.appendingPathComponent("\(stem) (\(suffix)).txt")
+            target = directory.appendingPathComponent("\(stem) (\(suffix)).\(source.pathExtension)")
             suffix += 1
         }
         try FileManager.default.copyItem(at: source, to: target)
@@ -69,7 +69,8 @@ public actor LocalFileService {
               !trimmed.contains(where: { $0 == "/" || $0 == "\\" || $0.unicodeScalars.contains(where: CharacterSet.controlCharacters.contains) }) else {
             throw LocalFileError.invalidName
         }
-        let filename = trimmed.lowercased().hasSuffix(".txt") ? trimmed : trimmed + ".txt"
+        let suffix = "." + source.pathExtension
+        let filename = trimmed.lowercased().hasSuffix(suffix.lowercased()) ? trimmed : trimmed + suffix
         guard filename.utf8.count <= 255 else { throw LocalFileError.invalidName }
         let target = directory.appendingPathComponent(filename)
         if source.standardizedFileURL == target.standardizedFileURL { return source }
@@ -83,7 +84,7 @@ public actor LocalFileService {
         guard FileManager.default.fileExists(atPath: source.path) else { return nil }
         let trash = directory.appendingPathComponent(".Trash", isDirectory: true)
         try FileManager.default.createDirectory(at: trash, withIntermediateDirectories: true)
-        let staged = trash.appendingPathComponent(UUID().uuidString + ".txt")
+        let staged = trash.appendingPathComponent(UUID().uuidString + "." + source.pathExtension)
         try FileManager.default.moveItem(at: source, to: staged)
         return .init(original: source, staged: staged)
     }
@@ -104,7 +105,7 @@ public actor LocalFileService {
     }
 
     private func validate(_ url: URL, allowMissing: Bool = false) throws {
-        guard url.isFileURL, url.pathExtension.lowercased() == "txt",
+        guard url.isFileURL, LocalBookService.supports(url),
               url.standardizedFileURL.deletingLastPathComponent() == directory,
               url.resolvingSymlinksInPath().deletingLastPathComponent() == directory.resolvingSymlinksInPath() else {
             throw LocalFileError.outsideLibrary
@@ -131,7 +132,7 @@ public actor BookFileExporter {
     }
 
     public func originalFile(_ source: URL, filename: String) throws -> BookFileExport {
-        let target = try destination(filename: filename)
+        let target = try destination(filename: filename, extension: source.pathExtension.lowercased())
         do {
             try FileManager.default.copyItem(at: source, to: target)
             return .init(url: target, cachedChapters: 0, totalChapters: 0)
@@ -178,11 +179,11 @@ public actor BookFileExporter {
         try FileManager.default.removeItem(at: export.url.deletingLastPathComponent())
     }
 
-    private func destination(filename: String) throws -> URL {
+    private func destination(filename: String, extension fileExtension: String = "txt") throws -> URL {
         let folder = directory.appendingPathComponent(UUID().uuidString, isDirectory: true)
         try FileManager.default.createDirectory(at: folder, withIntermediateDirectories: true)
         let safe = filename.components(separatedBy: CharacterSet(charactersIn: "/\\:").union(.controlCharacters)).joined(separator: "_")
-        let stem = String(safe.replacingOccurrences(of: ".txt", with: "", options: [.caseInsensitive, .anchored, .backwards]).prefix(70))
+        let stem = String(safe.replacingOccurrences(of: "." + fileExtension, with: "", options: [.caseInsensitive, .anchored, .backwards]).prefix(70))
-        return folder.appendingPathComponent((stem.isEmpty ? "书籍" : stem) + ".txt")
+        return folder.appendingPathComponent((stem.isEmpty ? "书籍" : stem) + "." + fileExtension)
     }
 }

Sources/LegadoCore/Networking/LegadoURLBuilder.swift

@@ -548,17 +548,32 @@ public struct LegadoURLBuilder: Sendable {
             case "{", ",":
                 output.append(character)
                 expectingKey = true
-            case ":":
+            case ":", "=":
-                output.append(character)
+                // Gson also accepts = and => between an object key and value.
+                // Quoted strings are consumed above, so Cookie/URL values stay intact.
+                output.append(":")
+                if character == "=", peek() == ">" { index += 1 }
                 expectingKey = false
-                if quoteUnquotedValues {
+                do {
                     var start = index
                     while start < characters.count, characters[start].isWhitespace { start += 1 }
                     if start < characters.count, !["\"", "'", "{", "[", "}", ","].contains(characters[start]) {
                         var end = start
-                        while end < characters.count, characters[end] != ",", characters[end] != "}" { end += 1 }
+                        if quoteUnquotedValues {
-                        output.append(escapedJSONString(String(characters[start..<end]).trimmingCharacters(in: .whitespacesAndNewlines)))
+                            // @put values are rules, including unquoted JSONPath expressions.
-                        index = end
+                            while end < characters.count, characters[end] != ",", characters[end] != "}" { end += 1 }
+                        } else {
+                            while end < characters.count, !characters[end].isWhitespace,
+                                  ![",", "}", "]", "{", "[", ":", ";", "=", "/", "#", "\\"].contains(characters[end]) { end += 1 }
+                        }
+                        if end > start {
+                            let token = String(characters[start..<end]).trimmingCharacters(in: .whitespacesAndNewlines)
+                            let literal = token.lowercased()
+                            let isLiteral = ["true", "false", "null"].contains(literal)
+                                || token.range(of: #"^-?(0|[1-9][0-9]*)(\.[0-9]+)?([eE][+-]?[0-9]+)?$"#, options: .regularExpression) != nil
+                            output.append(!quoteUnquotedValues && isLiteral ? literal : escapedJSONString(token))
+                            index = end
+                        }
                     }
                 }
             case "}", "]":
@@ -580,7 +595,7 @@ public struct LegadoURLBuilder: Sendable {
                     while peekPosition < characters.count, characters[peekPosition].isWhitespace {
                         peekPosition += 1
                     }
-                    if peekPosition < characters.count, characters[peekPosition] == ":" {
+                    if peekPosition < characters.count, [":", "="].contains(characters[peekPosition]) {
                         output.removeLast()
                         output.append(escapedJSONString(name))
                         index = position
@@ -611,7 +626,7 @@ public struct LegadoURLBuilder: Sendable {
         let parsed: JSONValue
         if case .string(let source) = value {
             do {
-                parsed = try JSONValue.decode(data: Data(source.utf8))
+                parsed = try Self.decodeLenientJSON(source)
             } catch {
                 throw RuleEngineError.malformedRule(rule: rule, detail: "headers JSON 无效")
             }

Sources/LegadoCore/RSS/RSSRuleParser.swift

@@ -84,6 +84,13 @@ public struct RSSRuleParser: Sendable {
     private func flatten(_ values: [JSONValue]) -> [JSONValue] {
         values.flatMap { value in
             if case .array(let items) = value { return items }
+            // Text rules serialize selected JSON arrays. Expand them before
+            // evaluating article fields, just as for an already parsed array.
+            if case .string(let text) = value,
+               text.trimmingCharacters(in: .whitespacesAndNewlines).hasPrefix("["),
+               let items = try? JSONDecoder().decode([JSONValue].self, from: Data(text.utf8)) {
+                return items
+            }
             return [value]
         }
     }

Sources/LegadoCore/RuleEngine/JSONPathEvaluator.swift

@@ -17,6 +17,48 @@ public struct JSONPathEvaluator: Sendable {
 
     public init() {}
 
+    // Legado's JSON rules also allow literal templates such as /book/{$.id}.
+    // Balance braces and quoted strings instead of stopping at the first }.
+    func evaluateRule(_ rule: String, in root: JSONValue) throws -> [JSONValue] {
+        var cursor = rule.startIndex
+        var output = ""
+        var foundTemplate = false
+        while let opening = rule.range(of: "{$.", range: cursor..<rule.endIndex) {
+            foundTemplate = true
+            output += rule[cursor..<opening.lowerBound]
+            let start = rule.index(after: opening.lowerBound)
+            var end = start
+            var depth = 1
+            var quote: Character?
+            var escaped = false
+            while end < rule.endIndex {
+                let character = rule[end]
+                if escaped { escaped = false }
+                else if character == "\\" { escaped = true }
+                else if let activeQuote = quote {
+                    if character == activeQuote { quote = nil }
+                } else if character == "\"" || character == "'" { quote = character }
+                else if character == "{" { depth += 1 }
+                else if character == "}" {
+                    depth -= 1
+                    if depth == 0 { break }
+                }
+                end = rule.index(after: end)
+            }
+            guard end < rule.endIndex else {
+                throw RuleEngineError.malformedRule(rule: rule, detail: "JSONPath 模板大括号未闭合")
+            }
+            let values = try evaluate(path: String(rule[start..<end]), in: root)
+            output += RuleEvaluation(values: values).strings.joined(separator: "\n")
+            cursor = rule.index(after: end)
+        }
+        if foundTemplate {
+            output += rule[cursor...]
+            return [.string(output)]
+        }
+        return try evaluate(path: rule.hasPrefix("$") ? rule : "$." + rule, in: root)
+    }
+
     public func evaluate(path: String, in root: JSONValue) throws -> [JSONValue] {
         let segments = try parse(path)
         var current = [root]
@@ -59,7 +101,10 @@ public struct JSONPathEvaluator: Sendable {
                 guard next < rule.endIndex else {
                     throw RuleEngineError.malformedRule(rule: path, detail: "JSONPath 不能以点结尾")
                 }
-                if rule[next] == "." {
+                if rule[next] == "[" {
+                    // Legacy sources use $.[*] as well as $[*].
+                    index = next
+                } else if rule[next] == "." {
                     index = rule.index(after: next)
                     let parsed = try parseDotSelector(rule, from: index, original: path)
                     segments.append(.recursive(parsed.selector))

Sources/LegadoCore/RuleEngine/JSONValue.swift

@@ -84,7 +84,8 @@ public extension JSONValue {
         if value is NSNull { return .null }
         if let value = value as? JSONValue { return value }
         if let value = value as? String { return .string(value) }
-        if let value = value as? Bool { return .bool(value) }
+        // NSNumber(0/1) also casts to Bool. Inspect its CF type first so
+        // JavaScript page arithmetic and numeric IDs remain numbers.
         if let value = value as? NSNumber {
             if CFGetTypeID(value) == CFBooleanGetTypeID() { return .bool(value.boolValue) }
             return .number(value.doubleValue)

Sources/LegadoCore/RuleEngine/JavaScriptRuleEvaluator.swift

@@ -208,6 +208,33 @@ public struct JavaScriptRuleEvaluator: Sendable {
         script: String,
         result: JSONValue,
         executionContext: RuleExecutionContext
+    ) throws -> JavaScriptRuleResult {
+        try evaluate(script: script, result: result, executionContext: executionContext,
+                     virtualMachine: JSVirtualMachine(), stringifyResult: false)
+    }
+
+    /// A rule shares java.get/put storage, but each match gets fresh JS globals.
+    /// Keep this closure on the synchronous replacement call's thread.
+    func replacementEvaluator(script: String, book: Book, chapter: BookChapter) -> (String) throws -> String {
+        var virtualMachine: JSVirtualMachine?
+        var executionContext = RuleExecutionContext(book: book, chapter: chapter)
+        return { match in
+            if virtualMachine == nil { virtualMachine = JSVirtualMachine() }
+            let result = try autoreleasepool {
+                try evaluate(script: script, result: .string(match), executionContext: executionContext,
+                             virtualMachine: virtualMachine!, stringifyResult: true)
+            }
+            executionContext.variables = result.variables
+            return result.value.stringValue
+        }
+    }
+
+    private func evaluate(
+        script: String,
+        result: JSONValue,
+        executionContext: RuleExecutionContext,
+        virtualMachine: JSVirtualMachine,
+        stringifyResult: Bool
     ) throws -> JavaScriptRuleResult {
         guard script.utf8.count <= maximumScriptLength else {
             throw RuleEngineError.resourceLimit(rule: script, detail: "JavaScript 规则超过长度上限")
@@ -215,8 +242,7 @@ public struct JavaScriptRuleEvaluator: Sendable {
         try rejectUnboundedExecution(script)
 
         let box = JSVariableBox(executionContext.variables)
-        let vm = JSVirtualMachine()
+        guard let context = JSContext(virtualMachine: virtualMachine) else {
-        guard let context = JSContext(virtualMachine: vm) else {
             throw RuleEngineError.invalidInput(rule: script, expected: "无法创建 JavaScriptCore 上下文")
         }
         var exceptionMessage: String?
@@ -271,7 +297,28 @@ public struct JavaScriptRuleEvaluator: Sendable {
         let normalized = evaluated.flatMap {
             context.objectForKeyedSubscript("__legadoUnboxString")?.call(withArguments: [$0])
         }
-        let value = JSONValue.bridge(normalized?.toObject())
+        let value: JSONValue
+        if stringifyResult {
+            // Rhino unwraps undefined to null before RegexExtensions.kt calls
+            // toString(). Preserve strings directly, rather than JSON encoding.
+            if let evaluated, !evaluated.isUndefined, !evaluated.isNull {
+                let text = evaluated.toString() ?? "null"
+                if evaluated.isNumber {
+                    // The reference runs Rhino in interpreted mode, yielding JVM Doubles.
+                    let number = evaluated.toDouble()
+                    value = .string(number.isFinite ? String(number) : text)
+                } else {
+                    value = .string(text)
+                }
+            } else {
+                value = .string("null")
+            }
+        } else {
+            value = JSONValue.bridge(normalized?.toObject())
+        }
+        if let exceptionMessage {
+            throw RuleEngineError.malformedRule(rule: script, detail: "JavaScript 执行失败：\(exceptionMessage)")
+        }
         guard value.stringValue.utf8.count <= maximumResultLength else {
             throw RuleEngineError.resourceLimit(rule: script, detail: "JavaScript 结果超过长度上限")
         }
@@ -974,53 +1021,48 @@ public struct JavaScriptRuleEvaluator: Sendable {
             return resolveJavaScriptRuleURL(direct.stringValue, enabled: isURL, context: executionContext)
         }
 
-        let plan = try RuleExpressionParser().parse(trimmedRule)
+        let jsonContent: JSONValue?
+        if case .string(let text) = content { jsonContent = try? JSONValue.decode(data: Data(text.utf8)) }
+        else { jsonContent = content }
+        let plan = try RuleExpressionParser().parse(trimmedRule, inputLooksLikeJSON: jsonContent != nil)
         var branchResults: [[String]] = []
         for branch in plan.branches {
             var values: [String]
-            switch branch.mode {
+            if branch.expression.isEmpty {
-            case .css:
+                values = [content.stringValue]
-                values = try HTMLRuleEvaluator().extractCSS(
+            } else {
-                    branch.expression,
+                switch branch.mode {
-                    html: content.stringValue,
+                case .css:
-                    baseURL: executionContext.baseURL,
+                    values = try HTMLRuleEvaluator().extractCSS(
-                    isFragment: executionContext.isHTMLFragment
+                        branch.expression,
-                )
+                        html: content.stringValue,
-            case .xpath:
+                        baseURL: executionContext.baseURL,
-                values = try HTMLRuleEvaluator().extractXPath(branch.expression, html: content.stringValue, isFragment: executionContext.isHTMLFragment)
+                        isFragment: executionContext.isHTMLFragment
-            case .jsonPath:
-                let json: JSONValue
-                if case .string(let text) = content {
-                    json = try JSONValue.decode(data: Data(text.utf8))
-                } else {
-                    json = content
-                }
-                let path = branch.expression.hasPrefix("$") ? branch.expression : "." + branch.expression
-                let normalizedPath = path.hasPrefix("$") ? path : "$" + path
-                values = try JSONPathEvaluator().evaluate(path: normalizedPath, in: json)
-                    .map(\.stringValue)
-            case .regex:
-                values = try RegexEvaluator().captureGroups(pattern: branch.expression, input: content.stringValue)
-                    .map { $0.count > 1 ? $0.dropFirst().joined() : $0.first ?? "" }
-            case .javaScript:
-                values = [try evaluate(
-                    script: branch.expression,
-                    result: content,
-                    executionContext: executionContext
-                ).value.stringValue]
-            case .webJavaScript:
-                throw RuleEngineError.unsupportedRule(rule: branch.expression, feature: "java.getString WebView JavaScript")
-            }
-            if let pattern = branch.replacePattern, !pattern.isEmpty {
-                values = try values.map {
-                    try RegexEvaluator().replacing(
-                        pattern: pattern,
-                        in: $0,
-                        with: branch.replacement,
-                        firstOnly: branch.replaceFirst
                     )
+                case .xpath:
+                    values = try HTMLRuleEvaluator().extractXPath(branch.expression, html: content.stringValue, isFragment: executionContext.isHTMLFragment)
+                case .jsonPath:
+                    let json: JSONValue
+                    if case .string(let text) = content {
+                        json = try JSONValue.decode(data: Data(text.utf8))
+                    } else {
+                        json = content
+                    }
+                    values = RuleEvaluation(values: try JSONPathEvaluator().evaluateRule(branch.expression, in: json)).strings
+                case .regex:
+                    values = try RegexEvaluator().captureGroups(pattern: branch.expression, input: content.stringValue)
+                        .map { $0.count > 1 ? $0.dropFirst().joined() : $0.first ?? "" }
+                case .javaScript:
+                    values = [try evaluate(
+                        script: branch.expression,
+                        result: content,
+                        executionContext: executionContext
+                    ).value.stringValue]
+                case .webJavaScript:
+                    throw RuleEngineError.unsupportedRule(rule: branch.expression, feature: "java.getString WebView JavaScript")
                 }
             }
+            if plan.combination == .fallback, values.joined(separator: "\n").isEmpty { values = [] }
             branchResults.append(values)
             if plan.combination == .fallback, !values.isEmpty { break }
         }
@@ -1037,7 +1079,11 @@ public struct JavaScriptRuleEvaluator: Sendable {
                 branchResults.compactMap { $0.indices.contains(index) ? $0[index] : nil }
             }
         }
-        return resolveJavaScriptRuleURL(combined.joined(separator: "\n"), enabled: isURL, context: executionContext)
+        var result = combined.joined(separator: "\n")
+        if let branch = plan.branches.first, let pattern = branch.replacePattern, !pattern.isEmpty {
+            result = try RegexEvaluator().replacing(pattern: pattern, in: result, with: branch.replacement, firstOnly: branch.replaceFirst)
+        }
+        return resolveJavaScriptRuleURL(result, enabled: isURL, context: executionContext)
     }
 
     private func resolveJavaScriptRuleURL(

Sources/LegadoCore/RuleEngine/RegexEvaluator.swift

@@ -112,6 +112,25 @@ public struct RegexEvaluator: Sendable {
 
     public func replacingStrict(pattern: String, in input: String, with replacement: String,
                                 timeoutMilliseconds: Int64 = 3_000) throws -> String {
+        try replacingStrict(pattern: pattern, in: input, timeoutMilliseconds: timeoutMilliseconds) { expression, match in
+            expression.replacementString(for: match, in: input, offset: 0, template: replacement)
+        }
+    }
+
+    /// Legado's JS replacement still expands capture references, while quoting
+    /// backslashes returned by the script (quoteReplacementJs in the reference).
+    func replacingStrict(pattern: String, in input: String, timeoutMilliseconds: Int64 = 3_000,
+                         replacement: (String) throws -> String) throws -> String {
+        let original = input as NSString
+        return try replacingStrict(pattern: pattern, in: input, timeoutMilliseconds: timeoutMilliseconds) { expression, match in
+            let template = try replacement(original.substring(with: match.range))
+                .replacingOccurrences(of: "\\", with: "\\\\")
+            return expression.replacementString(for: match, in: input, offset: 0, template: template)
+        }
+    }
+
+    private func replacingStrict(pattern: String, in input: String, timeoutMilliseconds: Int64,
+                                 replacement: (NSRegularExpression, NSTextCheckingResult) throws -> String) throws -> String {
         guard pattern.count <= limits.maximumPatternLength, input.utf16.count <= limits.maximumInputLength,
               !hasObviousNestedQuantifier(pattern) else {
             throw RuleEngineError.resourceLimit(rule: pattern, detail: "正则超过资源限制或包含高风险嵌套量词")
@@ -145,30 +164,51 @@ public struct RegexEvaluator: Sendable {
                 throw RuleEngineError.resourceLimit(rule: pattern, detail: "正则替换超时（\(milliseconds) ms）")
             }
             let untouched = original.substring(with: NSRange(location: offset, length: match.range.location - offset))
-            let replaced = expression.replacementString(for: match, in: input, offset: 0, template: replacement)
+            let replaced = try replacement(expression, match)
-            output += untouched + replaced
+            guard DispatchTime.now().uptimeNanoseconds < deadline else {
+                throw RuleEngineError.resourceLimit(rule: pattern, detail: "正则替换超时（\(milliseconds) ms）")
+            }
             outputLength += untouched.utf16.count + replaced.utf16.count
             offset = NSMaxRange(match.range)
             guard outputLength <= limits.maximumInputLength * 2 else {
                 throw RuleEngineError.resourceLimit(rule: pattern, detail: "替换结果超过长度限制")
             }
+            output += untouched + replaced
         }
-        output += original.substring(from: offset)
+        let tail = original.substring(from: offset)
+        guard outputLength + tail.utf16.count <= limits.maximumInputLength * 2 else {
+            throw RuleEngineError.resourceLimit(rule: pattern, detail: "替换结果超过长度限制")
+        }
+        output += tail
         return output
     }
 
     private func hasObviousNestedQuantifier(_ pattern: String) -> Bool {
-        let risky = #"\(((?:[^()\\]|\\.)*[+*](?:[^()\\]|\\.)*)\)([+*]|\{[^}]+\})"#
+        // Reject clear cases such as (a+)+ and (\w*)*. A '+' inside a character
-        guard let expression = try? NSRegularExpression(pattern: risky) else { return false }
+        // class/escape is not a quantifier, and bounded or delimited repeated
-        return expression.matches(in: pattern, range: NSRange(pattern.startIndex..., in: pattern)).contains { match in
+        // groups are not proof of catastrophic backtracking. Other expressions
-            guard let innerRange = Range(match.range(at: 1), in: pattern),
+        // remain protected by enumerateMatches' progress/deadline checks.
-                  let outerRange = Range(match.range(at: 2), in: pattern) else { return true }
+        let atom = #"^(?:\?:)?(?:\\[dDsSwW]|\\[^pPkKgG0-9]|\[(?:\\.|[^\]\\])*\]|[^\[\]()\\?*+{}|^$])(?:[+*]|\{[0-9]+,\})\??$"#
-            let inner = String(pattern[innerRange])
+        guard let repeatedAtom = try? NSRegularExpression(pattern: atom),
-            let outer = String(pattern[outerRange])
+              let unbounded = try? NSRegularExpression(pattern: #"^(?:[+*]|\{[0-9]+,\})(?!\+)"#) else { return false }
-            if outer.range(of: #"^\{[1-9]\}$"#, options: .regularExpression) != nil { return false }
+        var groups: [String.Index] = []
-            if !pattern.contains("(?s"),
+        var classDepth = 0
-               inner.range(of: #"^(?:\\n[^\[\]()*+?\\]*\.\*|\.\+\\n)$"#, options: .regularExpression) != nil { return false }
+        var escaped = false
-            return true
+        for index in pattern.indices {
+            let character = pattern[index]
+            if escaped { escaped = false; continue }
+            if character == "\\" { escaped = true; continue }
+            if character == "[" { classDepth += 1; continue }
+            if character == "]", classDepth > 0 { classDepth -= 1; continue }
+            guard classDepth == 0 else { continue }
+            if character == "(" { groups.append(pattern.index(after: index)) }
+            if character == ")", let start = groups.popLast() {
+                let inner = String(pattern[start..<index])
+                let suffix = String(pattern[pattern.index(after: index)...])
+                if repeatedAtom.firstMatch(in: inner, range: NSRange(inner.startIndex..., in: inner)) != nil,
+                   unbounded.firstMatch(in: suffix, range: NSRange(suffix.startIndex..., in: suffix)) != nil { return true }
+            }
         }
+        return false
     }
 }

Sources/LegadoCore/RuleEngine/RuleEngine.swift

@@ -57,11 +57,11 @@ public struct RuleEngine: Sendable {
                 return literalEvaluation(rendered.rule, context: rendered.context)
             }
             if rendered.context.variables != context.variables {
-                var result = await evaluatePrepared(rendered.rule, in: input, context: rendered.context)
+                var result = await evaluatePrepared(rendered.rule, in: input, listMode: listMode, context: rendered.context)
                 result.variables = rendered.context.variables.merging(result.variables) { _, new in new }
                 return result
             }
-            return await evaluatePrepared(rendered.rule, in: input, context: rendered.context)
+            return await evaluatePrepared(rendered.rule, in: input, listMode: listMode, context: rendered.context)
         } catch let error as RuleEngineError {
             return .init(diagnostics: [error.diagnostic], variables: context.variables)
         } catch {
@@ -72,6 +72,7 @@ public struct RuleEngine: Sendable {
     private func evaluatePrepared(
         _ rule: String,
         in input: JSONValue,
+        listMode: Bool,
         context: RuleExecutionContext
     ) async -> RuleEvaluation {
         do {
@@ -80,7 +81,8 @@ public struct RuleEngine: Sendable {
             var diagnostics: [RuleDiagnostic] = []
             for branch in plan.branches {
                 do {
-                    let values = try await evaluate(branch, in: input, context: context)
+                    var values = try await evaluate(branch, in: input, context: context)
+                    if plan.combination == .fallback, isEmpty(values, listMode: listMode) { values = [] }
                     results.append(values)
                     if plan.combination == .fallback, !values.isEmpty { break }
                 } catch let error as RuleEngineError {
@@ -97,7 +99,7 @@ public struct RuleEngine: Sendable {
                 }
             }
             return .init(
-                values: combine(results, with: plan.combination),
+                values: try applyingReplacement(to: combine(results, with: plan.combination), plan: plan, listMode: listMode),
                 diagnostics: diagnostics,
                 variables: context.variables
             )
@@ -174,6 +176,11 @@ public struct RuleEngine: Sendable {
         listMode: Bool,
         context: RuleExecutionContext
     ) async -> RuleEvaluation {
+        if let root = try? JSONValue.decode(data: Data(input.utf8)),
+           let plan = try? parser.parse(rule, inputLooksLikeJSON: true),
+           plan.branches.allSatisfy({ $0.mode == .jsonPath }) {
+            return await evaluatePrepared(rule, in: root, listMode: listMode, context: context)
+        }
         if rule.hasPrefix(":"), rule.count > 1 {
             do {
                 let patterns = rule.dropFirst().components(separatedBy: "&&").filter { !$0.isEmpty }
@@ -217,7 +224,7 @@ public struct RuleEngine: Sendable {
                 }
             }
             return .init(
-                values: combine(results, with: plan.combination),
+                values: try applyingReplacement(to: combine(results, with: plan.combination), plan: plan, listMode: listMode),
                 diagnostics: diagnostics,
                 variables: context.variables
             )
@@ -241,15 +248,11 @@ public struct RuleEngine: Sendable {
         try Task.checkCancellation()
         let rawValues: [JSONValue]
         if branch.expression.isEmpty {
-            return try [JSONValue.string(regex.replacing(
+            return [input]
-                pattern: branch.replacePattern ?? "", in: input.stringValue,
-                with: branch.replacement, firstOnly: branch.replaceFirst
-            ))]
         }
         switch branch.mode {
         case .jsonPath:
-            let path = branch.expression.hasPrefix("$") ? branch.expression : "$.\(branch.expression)"
+            rawValues = try jsonPath.evaluateRule(branch.expression, in: input)
-            rawValues = try jsonPath.evaluate(path: path, in: input)
         case .regex:
             let matches = try regex.captureGroups(pattern: branch.expression, input: input.stringValue)
             rawValues = matches.map { .array($0.map(JSONValue.string)) }
@@ -266,15 +269,7 @@ public struct RuleEngine: Sendable {
         case .webJavaScript:
             throw RuleEngineError.unsupportedRule(rule: branch.expression, feature: "WKWebView JavaScript")
         }
-        guard let pattern = branch.replacePattern, !pattern.isEmpty else { return rawValues }
+        return rawValues
-        return try rawValues.map {
-            .string(try regex.replacing(
-                pattern: pattern,
-                in: $0.stringValue,
-                with: branch.replacement,
-                firstOnly: branch.replaceFirst
-            ))
-        }
     }
 
     private func evaluateTextBranch(
@@ -287,10 +282,7 @@ public struct RuleEngine: Sendable {
         try Task.checkCancellation()
         let raw: [String]
         if branch.expression.isEmpty {
-            return try [regex.replacing(
+            return [input]
-                pattern: branch.replacePattern ?? "", in: input,
-                with: branch.replacement, firstOnly: branch.replaceFirst
-            )]
         }
         switch branch.mode {
         case .css:
@@ -308,8 +300,7 @@ public struct RuleEngine: Sendable {
             } catch {
                 throw RuleEngineError.invalidInput(rule: branch.expression, expected: "JSON")
             }
-            let path = branch.expression.hasPrefix("$") ? branch.expression : "$.\(branch.expression)"
+            raw = try jsonPath.evaluateRule(branch.expression, in: root).map(\.stringValue)
-            raw = try jsonPath.evaluate(path: path, in: root).map(\.stringValue)
         case .regex:
             raw = try regex.captureGroups(pattern: branch.expression, input: input).map { captures in
                 captures.count > 1 ? captures.dropFirst().joined() : captures.first ?? ""
@@ -323,14 +314,22 @@ public struct RuleEngine: Sendable {
         case .webJavaScript:
             throw RuleEngineError.unsupportedRule(rule: branch.expression, feature: "WKWebView JavaScript")
         }
-        guard let pattern = branch.replacePattern, !pattern.isEmpty else { return raw }
+        return raw
-        return try raw.map {
+    }
-            try regex.replacing(
+
-                pattern: pattern,
+    private func isEmpty(_ values: [JSONValue], listMode: Bool) -> Bool {
-                in: $0,
+        let result = RuleEvaluation(values: values)
-                with: branch.replacement,
+        return listMode
-                firstOnly: branch.replaceFirst
+            ? result.listItems(rule: "").filter { $0 != .null }.isEmpty
-            )
+            : result.strings.joined(separator: "\n").isEmpty
+    }
+
+    private func applyingReplacement(to values: [JSONValue], plan: RulePlan, listMode: Bool) throws -> [JSONValue] {
+        guard let branch = plan.branches.first, let pattern = branch.replacePattern, !pattern.isEmpty else { return values }
+        let strings = RuleEvaluation(values: values).strings
+        let inputs = listMode ? strings : [strings.joined(separator: "\n")]
+        return try inputs.map {
+            .string(try regex.replacing(pattern: pattern, in: $0, with: branch.replacement, firstOnly: branch.replaceFirst))
         }
     }
 

Sources/LegadoCore/RuleEngine/RuleExpression.swift

@@ -52,8 +52,12 @@ public struct RuleExpressionParser: Sendable {
         guard !normalized.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty else {
             throw RuleEngineError.malformedRule(rule: source, detail: "规则不能为空")
         }
-        let split = try splitAtTopLevel(normalized)
+        // ## belongs to the whole selector stage. Regex/replacement text may
-        let branches = try split.parts.map { try parseBranch($0, inputLooksLikeJSON: inputLooksLikeJSON) }
+        // contain quotes, brackets and && that are not selector operators.
+        let replacementStart = normalized.range(of: "##")?.lowerBound ?? normalized.endIndex
+        let split = try splitAtTopLevel(String(normalized[..<replacementStart]))
+        let suffix = String(normalized[replacementStart...])
+        let branches = try split.parts.map { try parseBranch($0 + suffix, inputLooksLikeJSON: inputLooksLikeJSON) }
         return .init(source: source, combination: split.combination, branches: branches)
     }
 

THIRD_PARTY_NOTICES.md

@@ -21,6 +21,12 @@
 - 用途：XPath 解析。
 - 许可：MIT，完整版权及许可见 [Kanna-LICENSE.txt](App/Resources/Kanna-LICENSE.txt)。
 
+## ZIPFoundation
+
+- 项目：<https://github.com/weichsel/ZIPFoundation>，固定版本 `0.9.20`。
+- 用途：按需读取 EPUB ZIP 容器与校验资源 CRC。
+- 许可：MIT，完整版权及许可见 [ZIPFoundation-LICENSE.txt](App/Resources/ZIPFoundation-LICENSE.txt)。
+
 ## 测试数据与演示内容
 
 测试中的规则样本仅用于兼容验证，不打包为 App 默认书源。具体说明见 [测试样本说明](Tests/LegadoCoreTests/Fixtures/README.md)。演示模式使用虚构书籍元数据和项目内演示文本；发布截图不使用个人书架、登录信息或付费章节。

Tests/LegadoAppModelsTests/EPUBReadingRegressionTests.swift

@@ -0,0 +1,81 @@
+import Foundation
+import Testing
+import ZIPFoundation
+import LegadoCore
+@testable import LegadoAppModels
+
+@MainActor
+struct EPUBReadingRegressionTests {
+    private func fixture(at root: URL) throws -> URL {
+        let url = root.appendingPathComponent("示例.epub")
+        let archive = try Archive(url: url, accessMode: .create)
+        let files = [
+            ("mimetype", "application/epub+zip"),
+            ("META-INF/container.xml", #"<container><rootfiles><rootfile full-path="book.opf"/></rootfiles></container>"#),
+            ("book.opf", #"<package><metadata xmlns:dc="http://purl.org/dc/elements/1.1/"><dc:title>回归书籍</dc:title><dc:creator>测试作者</dc:creator></metadata><manifest><item id="first" href="first.xhtml" media-type="application/xhtml+xml"/><item id="last" href="last.xhtml" media-type="application/xhtml+xml"/></manifest><spine><itemref idref="first"/><itemref idref="last"/></spine></package>"#),
+            ("first.xhtml", "<html><head><title>第一章</title></head><body><h1>第一章</h1><p>第一章测试正文</p></body></html>"),
+            ("last.xhtml", "<html><head><title>第二章</title></head><body><h1>第二章</h1><p>第二章测试正文</p></body></html>")
+        ]
+        for (path, value) in files {
+            let data = Data(value.utf8)
+            try archive.addEntry(with: path, type: .file, uncompressedSize: Int64(data.count), compressionMethod: .deflate) { position, size in
+                data.subdata(in: Int(position)..<Int(position) + size)
+            }
+        }
+        return url
+    }
+
+    @Test func importedEPUBReadsNavigatesAndResumesAfterRenamingAndReloadingShelf() async throws {
+        let root = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString, isDirectory: true)
+        try FileManager.default.createDirectory(at: root, withIntermediateDirectories: true)
+        defer { try? FileManager.default.removeItem(at: root) }
+        let url = try fixture(at: root)
+        let repository = InMemoryBookRepository()
+        let managed = root.appendingPathComponent("managed", isDirectory: true)
+        let library = BookLibraryModel(repository: repository, localDirectory: managed, chapterRepository: InMemoryBookChapterRepository())
+        await library.load()
+        await library.importLocalFiles([url])
+        #expect(library.errorMessage == nil)
+        let book = try #require(library.books.first)
+        let shelfReader = BookshelfReadViewModel()
+        await shelfReader.load(book: book, source: nil)
+        #expect(shelfReader.errorMessage == nil && shelfReader.chapters.count == 2)
+        let reader = ReaderViewModel(book: book, chapters: shelfReader.chapters, initialIndex: 0, cache: .init(directory: root.appendingPathComponent("cache")))
+        #expect(!reader.canEditContent && !reader.usesReplaceRules)
+        await reader.load(rules: [])
+        #expect(reader.content == "第一章测试正文")
+        await reader.prepareContinuousReading(rules: [])
+        #expect(reader.scrollChapters.map(\.content) == ["第一章测试正文", "第二章测试正文"])
+        reader.goNext()
+        await reader.load(rules: [])
+        #expect(reader.content == "第二章测试正文")
+        await library.updateProgress(for: book, chapter: try #require(reader.currentChapter), position: 4)
+        try await library.renameLocalFile(at: URL(string: book.bookUrl)!, to: "改名")
+        let reloaded = BookLibraryModel(repository: repository, localDirectory: managed, chapterRepository: InMemoryBookChapterRepository())
+        await reloaded.load()
+        let saved = try #require(reloaded.books.first)
+        #expect(saved.name == "回归书籍" && saved.originName == "改名.epub")
+        #expect(saved.durChapterIndex == 1 && saved.durChapterPos == 4)
+        let chapters = try LocalBookService().loadChapters(book: saved).chapters
+        let resumed = ReaderViewModel(book: saved, chapters: chapters, initialIndex: saved.durChapterIndex)
+        await resumed.load(rules: [])
+        #expect(resumed.currentIndex == 1 && resumed.requestedPosition == 4)
+        #expect(resumed.content == "第二章测试正文")
+        #expect(try await reloaded.localFiles.list().count == 1)
+    }
+
+    @Test func corruptEPUBImportRollsBackFileAndContinuesOtherImports() async throws {
+        let root = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString, isDirectory: true)
+        try FileManager.default.createDirectory(at: root, withIntermediateDirectories: true)
+        defer { try? FileManager.default.removeItem(at: root) }
+        let good = try fixture(at: root)
+        let bad = root.appendingPathComponent("损坏.epub")
+        try Data("bad epub".utf8).write(to: bad)
+        let library = BookLibraryModel(repository: InMemoryBookRepository(), localDirectory: root.appendingPathComponent("managed", isDirectory: true), chapterRepository: InMemoryBookChapterRepository())
+        await library.load()
+        await library.importLocalFiles([bad, good])
+        #expect(library.books.count == 1)
+        #expect(library.errorMessage?.contains("损坏.epub") == true)
+        #expect(try await library.localFiles.list().map { $0.url.lastPathComponent } == ["示例.epub"])
+    }
+}

Tests/LegadoAppModelsTests/SearchRegressionTests.swift

@@ -5,6 +5,26 @@ import LegadoCore
 
 @MainActor
 struct SearchRegressionTests {
+    @Test func arithmeticPageExpressionCanLoadTheSecondPage() async throws {
+        let transport = GatedModelSearchTransport()
+        await transport.release()
+        let model = model(transport: transport, repository: InMemorySearchBookRepository())
+        defer { model.cancel(reset: true) }
+        var source = source("formula")
+        source.searchUrl = #"/search,{"method":"POST","body":"keyword={{key}}&page={{page-1}}"}"#
+        model.search(keyword: "书", sources: [source])
+        try await waitUntil { model.state == .finished }
+        #expect(model.hasMore)
+        model.loadMore()
+        try await waitUntil { !model.isLoadingMore }
+        let requests = await transport.requests()
+        #expect(requests.count == 2)
+        #expect(requests.compactMap(\.httpBody).map { String(decoding: $0, as: UTF8.self) } == [
+            "keyword=%E4%B9%A6&page=0", "keyword=%E4%B9%A6&page=1"
+        ])
+        #expect(!model.hasMore)
+    }
+
     @Test func cachedResultsAppearBeforeNetworkAndFreshResultsReplaceTheirMetadata() async throws {
         let source = source("slow")
         let cached = SearchBook(bookURL: source.id + "/cached", origin: source.id, originName: "源", type: 0, name: "书", author: "作者")

Tests/LegadoCoreTests/BookSourceRemoteProbe.swift

@@ -0,0 +1,89 @@
+import Foundation
+import Testing
+@testable import LegadoCore
+
+// Opt-in only: point LEGADO_SOURCE_PROBE_FILE at a downloaded JSON source pack.
+// No remote source or response is included in the application or normal test run.
+struct BookSourceRemoteProbe {
+    @Test(.enabled(if: ProcessInfo.processInfo.environment["LEGADO_SOURCE_CHAIN_FILE"] != nil))
+    func sampleReadingPipelines() async throws {
+        let path = try #require(ProcessInfo.processInfo.environment["LEGADO_SOURCE_CHAIN_FILE"])
+        let sources = try BookSourceImporter().decode(Data(contentsOf: URL(fileURLWithPath: path)))
+        var reports: [BookSourceDebugReport] = []
+        let transport = ProbeTransport()
+        let keyword = ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_KEYWORD"] ?? "斗罗大陆"
+        for source in sources {
+            var report = await BookSourceDebugger(transport: transport).run(source: source, keyword: keyword)
+            report.contentPreview = nil
+            reports.append(report)
+            print("SOURCE_CHAIN \(source.bookSourceName) " + String(decoding: try JSONEncoder().encode(report), as: UTF8.self))
+        }
+        if let output = ProcessInfo.processInfo.environment["LEGADO_SOURCE_CHAIN_OUTPUT"] {
+            try JSONEncoder().encode(reports).write(to: URL(fileURLWithPath: output))
+            try JSONEncoder().encode(await transport.records()).write(to: URL(fileURLWithPath: output + ".requests.json"))
+        }
+        #expect(reports.count == sources.count)
+    }
+
+    @Test(.enabled(if: ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_FILE"] != nil))
+    func sampleSearches() async throws {
+        guard let path = ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_FILE"] else { return }
+        let sources = try BookSourceImporter().decode(Data(contentsOf: URL(fileURLWithPath: path)))
+        let transport = ProbeTransport()
+        let service = BookSearchService(transport: transport, networkPolicy: .init(), history: .init())
+        let keyword = ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_KEYWORD"] ?? "斗罗大陆"
+        let page = Int(ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_PAGE"] ?? "1") ?? 1
+        var rows: [[String: JSONValue]] = []
+        for await outcome in service.searchOutcomes(keyword: keyword, page: page, sources: sources, maximumConcurrentSources: 3, includeDisabled: true) {
+            let source = try #require(sources.first { $0.id == outcome.sourceID })
+            let row: [String: JSONValue] = [
+                "name": .string(source.bookSourceName), "source": .string(source.id),
+                "count": .number(Double(outcome.books.count)),
+                "books": .array(outcome.books.prefix(3).map { .object([
+                    "name": .string($0.name), "url": .string($0.bookURL), "author": .string($0.author),
+                    "cover": .string($0.coverURL ?? ""), "wordCount": .string($0.wordCount ?? "")
+                ]) }),
+                "diagnostics": .array(outcome.diagnostics.map { .object(["code": .string($0.code), "message": .string($0.message), "rule": .string($0.rule)]) })
+            ]
+            rows.append(row)
+            print("SOURCE_PROBE " + JSONValue.object(row).stringValue)
+        }
+        let report = JSONValue.object(["keyword": .string(keyword), "page": .number(Double(page)), "outcomes": .array(rows.map(JSONValue.object)), "requests": .array(await transport.records())])
+        if let output = ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_OUTPUT"] {
+            try JSONEncoder().encode(report).write(to: URL(fileURLWithPath: output))
+        }
+        #expect(rows.count == Set(sources.map(\.id)).count)
+    }
+}
+
+private actor ProbeTransport: HTTPTransport {
+    private var values: [JSONValue] = []
+    private let transport = URLSessionHTTPTransport()
+
+    func data(for request: URLRequest) async throws -> HTTPResponse {
+        do {
+            let response = try await transport.data(for: request)
+            var record: [String: JSONValue] = [
+                "url": .string(request.url?.absoluteString ?? ""),
+                "finalURL": .string(response.finalURL.absoluteString),
+                "status": .number(Double(response.statusCode)),
+                "bytes": .number(Double(response.data.count)),
+                "preview": .string(String(response.decodedText.prefix(240)))
+            ]
+            if let directory = ProcessInfo.processInfo.environment["LEGADO_SOURCE_PROBE_SNAPSHOT_DIR"] {
+                let folder = URL(fileURLWithPath: directory, isDirectory: true)
+                try FileManager.default.createDirectory(at: folder, withIntermediateDirectories: true)
+                let file = folder.appendingPathComponent(UUID().uuidString + ".response")
+                try response.data.write(to: file)
+                record["snapshot"] = .string(file.path)
+            }
+            values.append(.object(record))
+            return response
+        } catch {
+            values.append(.object(["url": .string(request.url?.absoluteString ?? ""), "error": .string(error.localizedDescription)]))
+            throw error
+        }
+    }
+
+    func records() -> [JSONValue] { values }
+}

Tests/LegadoCoreTests/EPUBBookTests.swift

@@ -0,0 +1,193 @@
+import Foundation
+import Testing
+import ZIPFoundation
+@testable import LegadoCore
+
+struct EPUBBookTests {
+    private let service = LocalBookService()
+    private let png = Data(base64Encoded: "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")!
+
+    private func archive(_ files: [(String, Data)]) throws -> URL {
+        let directory = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        try FileManager.default.createDirectory(at: directory, withIntermediateDirectories: true)
+        let url = directory.appendingPathComponent("示例.EPUB")
+        let archive = try Archive(url: url, accessMode: .create)
+        for (path, data) in files {
+            try archive.addEntry(with: path, type: .file, uncompressedSize: Int64(data.count), compressionMethod: path == "mimetype" ? .none : .deflate) { position, size in
+                data.subdata(in: Int(position)..<Int(position) + size)
+            }
+        }
+        return url
+    }
+
+    private func fixture(epub3: Bool = true, navigation: Bool = true, extras: [(String, Data)] = [], chapter: String? = nil) throws -> URL {
+        let navItem = epub3
+            ? #"<item id="nav" href="目录/nav.xhtml" media-type="application/xhtml+xml" properties="nav"/>"#
+            : #"<item id="nav" href="目录/toc.ncx" media-type="application/x-dtbncx+xml"/>"#
+        let package = """
+        <?xml version="1.0" encoding="UTF-8"?>
+        <package xmlns="http://www.idpf.org/2007/opf" version="\(epub3 ? "3.0" : "2.0")">
+          <metadata xmlns:dc="http://purl.org/dc/elements/1.1/">
+            <dc:title>测试书籍</dc:title><dc:creator>作者甲</dc:creator><dc:creator>作者乙</dc:creator>
+            <dc:description>&lt;p&gt;测试简介 &amp;amp; 内容&lt;/p&gt;</dc:description>
+            <meta name="cover" content="cover"/>
+          </metadata>
+          <manifest>
+            <item id="last" href="Text/end.xhtml" media-type="application/xhtml+xml"/>
+            \(navigation ? navItem : "")
+            <item id="main" href="Text/中文%20章节.xhtml" media-type="application/xhtml+xml"/>
+            <item id="intro" href="Text/intro.xhtml" media-type="application/xhtml+xml"/>
+            <item id="cover" href="Images/封面.png" media-type="image/png" properties="cover-image"/>
+            <item id="extra" href="Text/extra.xhtml" media-type="application/xhtml+xml"/>
+          </manifest>
+          <spine toc="nav"><itemref idref="intro"/><itemref idref="main"/><itemref idref="extra"/><itemref idref="last"/></spine>
+        </package>
+        """
+        let body = chapter ?? """
+        <html><head><title>正文</title><style>不要显示样式</style></head><body>
+          <p>锚点前的引言</p><section><h1 id="one">第一章</h1><p>甲 &amp; 乙<em>相遇</em>。</p>
+          <script>不要显示脚本</script><p style="display: none">隐藏内容</p>
+          <p><ruby>字<rt>zi</rt><rp>(zi)</rp></ruby>形</p><img src="../Images/封面.png" alt="风景"/></section>
+          <section><div><h1 id="two">第二章</h1></div><p>第二章正文</p></section>
+        </body></html>
+        """
+        let nav = """
+        <html xmlns:epub="http://www.idpf.org/2007/ops"><body>
+        <nav epub:type="landmarks"><a href="../Text/end.xhtml">错误地标</a></nav>
+        <nav epub:type="toc"><ol><li><span>第一卷</span><ol>
+        <li><a href="../Text/中文%20章节.xhtml#one">第一章</a></li>
+        <li><a href="../Text/中文%20章节.xhtml#two">第二章</a></li></ol></li>
+        <li><a href="../Text/end.xhtml">尾声</a></li></ol></nav></body></html>
+        """
+        let ncx = """
+        <ncx xmlns="http://www.daisy.org/z3986/2005/ncx/"><navMap>
+        <navPoint><navLabel><text>第一卷</text></navLabel>
+          <navPoint><navLabel><text>第一章</text></navLabel><content src="../Text/中文%20章节.xhtml#one"/></navPoint>
+          <navPoint><navLabel><text>第二章</text></navLabel><content src="../Text/中文%20章节.xhtml#two"/></navPoint>
+        </navPoint><navPoint><navLabel><text>尾声</text></navLabel><content src="../Text/end.xhtml"/></navPoint>
+        </navMap></ncx>
+        """
+        var files: [(String, Data)] = [
+            ("mimetype", Data("application/epub+zip".utf8)),
+            ("META-INF/container.xml", Data(#"<container xmlns="urn:oasis:names:tc:opendocument:xmlns:container"><rootfiles><rootfile full-path="OPS/book.opf" media-type="application/oebps-package+xml"/></rootfiles></container>"#.utf8)),
+            ("OPS/book.opf", Data(package.utf8)),
+            ("OPS/Text/intro.xhtml", Data("<html><head><title>序言</title></head><body><p>开篇内容</p></body></html>".utf8)),
+            ("OPS/Text/中文 章节.xhtml", Data(body.utf8)),
+            ("OPS/Text/extra.xhtml", Data("<html><body><h1>未列入目录</h1><p>不应遗漏正文</p></body></html>".utf8)),
+            ("OPS/Text/end.xhtml", Data("<html><body><p>完结正文</p></body></html>".utf8)),
+            ("OPS/Images/封面.png", png)
+        ]
+        if navigation { files.append((epub3 ? "OPS/目录/nav.xhtml" : "OPS/目录/toc.ncx", Data((epub3 ? nav : ncx).utf8))) }
+        return try archive(files + extras)
+    }
+
+    @Test(arguments: [true, false]) func readsEPUB2And3MetadataSpineAnchorsAndIllustrations(epub3: Bool) throws {
+        let url = try fixture(epub3: epub3)
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let report = try service.openFile(at: url)
+        #expect(report.book.name == "测试书籍")
+        #expect(report.book.author == "作者甲、作者乙")
+        #expect(report.book.intro == "测试简介 & 内容")
+        #expect(report.book.coverUrl?.hasPrefix("data:image/png;base64,") == true)
+        #expect(report.book.readConfig?.useReplaceRule == false)
+        #expect(!report.book.canUpdate)
+        #expect(report.chapters.map(\.title) == ["序言", "正文", "第一章", "第二章", "未列入目录", "尾声"])
+        #expect(report.chapters.map(\.index) == Array(0..<6))
+        #expect(Set(report.chapters.map(\.id)).count == 6)
+        #expect(try service.loadChapters(book: report.book).chapters == report.chapters)
+        let contents = try report.chapters.map { try service.loadContent(book: report.book, chapter: $0).content }
+        #expect(contents[0] == "开篇内容")
+        #expect(contents[1] == "锚点前的引言")
+        #expect(contents[2] == "第一章\n甲 & 乙相遇。\n字形\n【插图 1：风景】")
+        #expect(contents[3] == "第二章\n第二章正文")
+        #expect(contents[4].contains("不应遗漏正文"))
+        #expect(contents[5] == "完结正文")
+        let images = try service.loadIllustrations(book: report.book, chapter: report.chapters[2])
+        #expect(images.count == 1 && images.first?.data == png && images.first?.caption == "风景")
+        #expect(try service.loadIllustrations(book: report.book, chapter: report.chapters[3]).isEmpty)
+    }
+
+    @Test func fallsBackToSpineWithoutNavigationAndDoesNotUseManifestOrder() throws {
+        let url = try fixture(navigation: false)
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let report = try service.openFile(at: url)
+        #expect(report.chapters.count == 4)
+        #expect(report.chapters.first?.title == "序言")
+        let body = try service.loadContent(book: report.book, chapter: report.chapters[1]).content
+        #expect(body.contains("锚点前的引言") && body.contains("第一章") && body.contains("第二章正文"))
+    }
+
+    @Test func importsCopyPreservingMetadataAndOriginalFile() throws {
+        let url = try fixture()
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let report = try service.importFile(at: url, into: url.deletingLastPathComponent().appendingPathComponent("managed"))
+        let destination = try #require(URL(string: report.book.bookUrl))
+        #expect(destination != url && destination.pathExtension == "epub")
+        #expect(report.book.originName == "示例.EPUB")
+        #expect(try Data(contentsOf: destination) == Data(contentsOf: url))
+        #expect(try service.loadContent(book: report.book, chapter: report.chapters[2]).content.contains("甲 & 乙"))
+    }
+
+    @Test func fileManagementPreservesEPUBOnRenameExportAndRemoval() async throws {
+        let url = try fixture()
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let root = url.deletingLastPathComponent()
+        let files = LocalFileService(directory: root.appendingPathComponent("managed"))
+        let first = try await files.importFile(from: url)
+        let second = try await files.importFile(from: url)
+        #expect(second.lastPathComponent == "示例 (2).EPUB")
+        let renamed = try await files.rename(first, to: "新名字")
+        #expect(renamed.pathExtension == "EPUB")
+        #expect(try await files.list().count == 2)
+        let exporter = BookFileExporter(directory: root.appendingPathComponent("exports"))
+        let export = try await exporter.originalFile(renamed, filename: "测试书籍.epub")
+        #expect(export.url.lastPathComponent == "测试书籍.epub")
+        #expect(try Data(contentsOf: export.url) == Data(contentsOf: url))
+        let staged = try #require(try await files.stageRemoval(renamed))
+        try await files.restore(staged)
+        #expect(try service.openFile(at: renamed).book.name == "测试书籍")
+        let removed = try #require(try await files.stageRemoval(renamed))
+        try await files.finishRemoval(removed)
+        #expect(try await files.list().count == 1)
+    }
+
+    @Test func rejectsCorruptionTraversalEncryptionAndOversizeWithoutLeavingImportedFiles() throws {
+        let good = try fixture()
+        defer { try? FileManager.default.removeItem(at: good.deletingLastPathComponent()) }
+        #expect(throws: LocalBookError.self) { try LocalBookService(maximumFileSize: 1).openFile(at: good) }
+        let broken = good.deletingLastPathComponent().appendingPathComponent("损坏.epub")
+        try Data("不是压缩包".utf8).write(to: broken)
+        let managed = good.deletingLastPathComponent().appendingPathComponent("managed")
+        #expect(throws: EPUBError.self) { try service.importFile(at: broken, into: managed) }
+        #expect(try FileManager.default.contentsOfDirectory(atPath: managed.path).isEmpty)
+        let malicious = try fixture(extras: [("../escape", Data("outside".utf8))])
+        defer { try? FileManager.default.removeItem(at: malicious.deletingLastPathComponent()) }
+        #expect(throws: EPUBError.self) { try service.openFile(at: malicious) }
+        let encrypted = try fixture(extras: [("META-INF/encryption.xml", Data(#"<encryption><EncryptedData><CipherData><CipherReference URI="OPS/Text/中文%20章节.xhtml"/></CipherData></EncryptedData></encryption>"#.utf8))])
+        defer { try? FileManager.default.removeItem(at: encrypted.deletingLastPathComponent()) }
+        #expect(throws: EPUBError.self) { try service.openFile(at: encrypted) }
+        let fontOnly = try fixture(extras: [("META-INF/encryption.xml", Data(#"<encryption><EncryptedData><CipherData><CipherReference URI="OPS/Fonts/font.otf"/></CipherData></EncryptedData></encryption>"#.utf8))])
+        defer { try? FileManager.default.removeItem(at: fontOnly.deletingLastPathComponent()) }
+        #expect(try service.openFile(at: fontOnly).chapters.count == 6)
+    }
+
+    @Test func rejectsForgedChapterResourceAndMissingAnchors() throws {
+        let url = try fixture()
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let report = try service.openFile(at: url)
+        var chapter = report.chapters[2]
+        chapter.resourceUrl = "../../outside"
+        #expect(throws: EPUBError.self) { try service.loadContent(book: report.book, chapter: chapter) }
+        chapter = report.chapters[2]
+        chapter.startFragmentId = "missing"
+        #expect(throws: EPUBError.self) { try service.loadContent(book: report.book, chapter: chapter) }
+    }
+
+    @Test func ignoresBadNavigationAnchorsAndOrdersValidAnchorsByDocumentPosition() throws {
+        let url = try fixture(chapter: "<html><body><h1 id='two'>第二章</h1><p>先读</p><h1 id='one'>第一章</h1><p>后读</p></body></html>")
+        defer { try? FileManager.default.removeItem(at: url.deletingLastPathComponent()) }
+        let report = try service.openFile(at: url)
+        #expect(report.chapters.map(\.title) == ["序言", "第二章", "第一章", "未列入目录", "尾声"])
+        #expect(try service.loadContent(book: report.book, chapter: report.chapters[1]).content == "第二章\n先读")
+    }
+}

Tests/LegadoCoreTests/LegadoDatabaseTests.swift

@@ -393,6 +393,20 @@ struct LegadoDatabaseTests {
         #expect(report.result.nextURL == "https://rss.example/list?page=1")
     }
 
+    @Test func rssJSONListWithoutWildcardExtractsIndividualArticles() async throws {
+        var source = RssSource(sourceUrl: "https://rss.example", sourceName: "JSON 订阅")
+        source.ruleArticles = "$.items"
+        source.ruleTitle = "$.title"
+        source.ruleLink = "$.link"
+        let report = try await RSSRuleParser().parse(
+            sortName: "全部", sortURL: "https://rss.example/feed", redirectURL: "https://rss.example/feed",
+            body: #"{"items":[{"title":"第一篇","link":"/one"},{"title":"第二篇","link":"/two"}]}"#,
+            source: source
+        )
+        #expect(report.result.articles.map(\.title) == ["第一篇", "第二篇"])
+        #expect(report.result.articles.map(\.link) == ["https://rss.example/one", "https://rss.example/two"])
+    }
+
     @Test func rssSourceImporterSupportsObjectArrayManifestAndUpdateSelection() throws {
         let importer = RSSSourceImporter()
         let object = try importer.decode(Data(#"{"sourceUrl":"https://one","sourceName":"一"}"#.utf8))

Tests/LegadoCoreTests/ReplaceRuleExecutionTests.swift

@@ -0,0 +1,141 @@
+import Foundation
+import Testing
+@testable import LegadoCore
+
+struct ReplaceRuleExecutionTests {
+    private func process(_ content: String, rules: [ReplaceRule], title: String = "第十二章") -> ContentProcessingReport {
+        let book = Book(bookUrl: "book", name: "测试书", origin: "https://source.test")
+        let chapter = BookChapter(url: "chapter", title: title, bookUrl: book.bookUrl, index: 11)
+        return ContentSanitizer().process(book: book, chapter: chapter, content: content, rules: rules,
+            removeDuplicateTitle: false, resegment: false)
+    }
+
+    @Test func evaluatesJavaScriptForEveryMatchIncludingZeroWidthMatches() {
+        let width = ReplaceRule(id: 1, name: "全角", pattern: "[Ａ-Ｚ０-９]",
+            replacement: "@js:String.fromCharCode(result.charCodeAt(0) - 65248)")
+        var punctuation = ReplaceRule(id: 2, name: "标点", pattern: #"(?<=说道)(?=“)|\.{3,}"#,
+            replacement: #"@js:if (result === '') { '：' } else { '…' }"#)
+        punctuation.sortOrder = 1
+        let result = process("ＡＢＣ１２３\n他说道“等待...”", rules: [punctuation, width])
+        #expect(result.content == "ABC123\n他说道：“等待…”")
+        #expect(result.effectiveRuleIDs == [1, 2])
+        #expect(result.diagnostics.isEmpty)
+    }
+
+    @Test func bindsBookAndChapterAndSharesOnlyJavaVariablesBetweenMatches() {
+        let rule = ReplaceRule(id: 1, name: "上下文", pattern: "[ab]", replacement: """
+            @js:n = typeof n === 'undefined' ? 1 : n + 1;
+            java.put('count', String(Number(java.get('count') || 0) + 1));
+            [book.name, book.getName(), chapter.title, chapter.getIndex(), result, n, java.get('count')].join(':')
+            """)
+        let report = process("a b", rules: [rule])
+        #expect(report.content == "测试书:测试书:第十二章:11:a:1:1 测试书:测试书:第十二章:11:b:1:2")
+        #expect(report.diagnostics.isEmpty)
+        // A new application of the same rule must not inherit another chapter's variables.
+        #expect(process("a", rules: [rule]).content == "测试书:测试书:第十二章:11:a:1:1")
+    }
+
+    @Test(arguments: [
+        (#"'$1'"#, "x"),
+        (#"String.fromCharCode(92) + 'n'"#, #"\n"#),
+        ("null", "null"),
+        ("undefined", "null"),
+        ("['a', 'b']", "a,b"),
+        ("123", "123.0"),
+        ("false", "false")
+    ])
+    func convertsScriptCompletionAndPreservesReplacementEscaping(_ sample: (String, String)) {
+        let rule = ReplaceRule(id: 1, name: "返回值", pattern: "(x)", replacement: "@js:" + sample.0)
+        let report = process("x", rules: [rule])
+        #expect(report.content == sample.1)
+        #expect(report.diagnostics.isEmpty)
+    }
+
+    @Test func failedScriptDiscardsAllPartialReplacementsAndLaterRulesStillRun() {
+        let failing = ReplaceRule(id: 1, name: "失败", pattern: "[ab]",
+            replacement: "@js:if (result === 'b') { throw new Error('bad match') }; 'changed'")
+        var following = ReplaceRule(id: 2, name: "后续", pattern: "tail", replacement: "后续成功")
+        following.sortOrder = 1
+        let report = process("ab tail", rules: [failing, following])
+        #expect(report.content == "ab 后续成功")
+        #expect(report.effectiveRuleIDs == [2])
+        #expect(report.diagnostics.count == 1)
+        #expect(report.diagnostics.first?.code == "rule.malformed")
+    }
+
+    @Test func skipsDisabledExcludedAndUnmatchedScriptsAndKeepsLiteralReplacementsLiteral() {
+        var excluded = ReplaceRule(id: 1, name: "排除", pattern: "x", replacement: "@js:throw new Error('excluded')")
+        excluded.excludeScope = "测试书"
+        var disabled = excluded
+        disabled.id = 2
+        disabled.excludeScope = nil
+        disabled.isEnabled = false
+        let unmatched = ReplaceRule(id: 3, name: "未匹配", pattern: "z", replacement: "@js:throw new Error('unmatched')")
+        var literal = ReplaceRule(id: 4, name: "普通文本", pattern: "x", replacement: "@js:'literal'")
+        literal.isRegex = false
+        let report = process("x", rules: [excluded, disabled, unmatched, literal])
+        #expect(report.content == "@js:'literal'")
+        #expect(report.diagnostics.isEmpty)
+    }
+
+    @Test func preservesLastNonblankTitleButAllowsDeletingBody() {
+        var rename = ReplaceRule(id: 1, name: "标题", pattern: "十二", replacement: "12")
+        rename.scopeTitle = true
+        rename.scopeContent = false
+        var erase = ReplaceRule(id: 2, name: "空白", pattern: "^.+$", replacement: "@js:' 　\\n'")
+        erase.scopeTitle = true
+        erase.sortOrder = 1
+        let report = process("正文", rules: [erase, rename])
+        #expect(report.title == "第12章")
+        #expect(report.content == " 　\n")
+        #expect(report.effectiveRuleIDs == [1, 2])
+        erase.replacement = ""
+        erase.scopeContent = false
+        let titleOnly = process("正文", rules: [erase])
+        #expect(titleOnly.title == "第十二章")
+        #expect(titleOnly.effectiveRuleIDs.isEmpty)
+    }
+
+    @Test(arguments: [
+        (#"(?:\n[^\n]+){0,15}$"#, "正文\nPS：说明\n感谢支持", "正文"),
+        (#"^(?:[^\n\w]\w+){0,2}"#, "，第一章，开始\n正文", "\n正文"),
+        (#"＂((?:[\w，。]|‘\w+’|\b[%\+\-])+)[＂]"#, "＂你好，世界。＂", "“你好，世界。”"),
+        (#"(?:[+*]|\\)+"#, #"+*\"#, "")
+    ])
+    func acceptsBoundedDelimitedAndEscapedQuantifiers(_ sample: (String, String, String)) throws {
+        let replacement = sample.0.hasPrefix("＂") ? "“$1”" : ""
+        #expect(try RegexEvaluator().replacingStrict(pattern: sample.0, in: sample.1, with: replacement) == sample.2)
+    }
+
+    @Test(arguments: ["(a+)+$", #"(\w*)*"#, "(?:[ab]+){2,}"])
+    func retainsChecksForClearlyNestedUnboundedRepetition(_ pattern: String) {
+        #expect(throws: RuleEngineError.self) {
+            try RegexEvaluator().replacingStrict(pattern: pattern, in: "aaa!", with: "changed")
+        }
+    }
+
+    @Test func dynamicReplacementRespectsMatchOutputAndWholeOperationTimeLimits() throws {
+        let limited = RegexEvaluator(limits: .init(maximumInputLength: 4, maximumMatches: 2))
+        #expect(throws: RuleEngineError.self) {
+            try limited.replacingStrict(pattern: "a", in: "aaa", replacement: { _ in "b" })
+        }
+        #expect(try limited.replacingStrict(pattern: "a", in: "aa", replacement: { _ in "b" }) == "bb")
+        // Include untouched tail in the output limit, not only replaced spans.
+        #expect(throws: RuleEngineError.self) {
+            try limited.replacingStrict(pattern: "a", in: "abcd", replacement: { _ in "1234567" })
+        }
+        #expect(throws: RuleEngineError.self) {
+            try limited.replacingStrict(pattern: "a", in: "a", timeoutMilliseconds: 1, replacement: { _ in
+                Thread.sleep(forTimeInterval: 0.02)
+                return "b"
+            })
+        }
+    }
+
+    @Test func rejectsKnownUnboundedJavaScriptAndReportsErrorWithoutChangingInput() {
+        let rule = ReplaceRule(id: 1, name: "循环", pattern: "x", replacement: "@js:while(true){}")
+        let report = process("x", rules: [rule])
+        #expect(report.content == "x")
+        #expect(report.diagnostics.first?.code == "rule.resource_limit")
+    }
+}

Tests/LegadoCoreTests/SampledRequestRegressionTests.swift

@@ -0,0 +1,89 @@
+import Foundation
+import Testing
+@testable import LegadoCore
+
+// The payloads and host below are synthetic and never use network access.
+struct SampledRequestRegressionTests {
+    @Test func equalsSeparatedSourceHeaderReachesSearchTransport() async throws {
+        var source = source()
+        source.header = #"{"type"="app"}"#
+        source.searchUrl = #"/search,{"method":"POST","body":"keyword={{key}}&page={{page-1}}"}"#
+        let transport = SampledRequestTransport()
+        let result = await BookSearchService(transport: transport).search(keyword: "斗罗大陆", page: 1, source: source)
+        #expect(result.diagnostics.isEmpty)
+        #expect(result.books.count == 1)
+        let request = try #require(await transport.requests().first)
+        #expect(request.value(forHTTPHeaderField: "type") == "app")
+        #expect(request.httpBody.map { String(decoding: $0, as: UTF8.self) } == "keyword=%E6%96%97%E7%BD%97%E5%A4%A7%E9%99%86&page=0")
+    }
+
+    @Test func stringEncodedLenientHeadersReachSearchTransport() async throws {
+        var source = source()
+        source.searchUrl = #"/search?word={{key}},{"headers":"{os:'pc'}"}"#
+        let transport = SampledRequestTransport()
+        let result = await BookSearchService(transport: transport).search(keyword: "斗罗大陆", page: 1, source: source)
+        #expect(result.diagnostics.isEmpty)
+        #expect(result.books.count == 1)
+        let request = try #require(await transport.requests().first)
+        #expect(request.value(forHTTPHeaderField: "os") == "pc")
+    }
+
+    @Test func unquotedKeywordInObjectBodyPreservesJSONValueTypes() async throws {
+        var source = source()
+        source.searchUrl = #"/search,{"method":"POST","body":{"keyword":{{key}},"page_index":{{page}},"page_size":20,"enabled":true,"cursor":null}}"#
+        let transport = SampledRequestTransport()
+        let result = await BookSearchService(transport: transport).search(keyword: "斗罗大陆", page: 1, source: source)
+        #expect(result.diagnostics.isEmpty)
+        #expect(result.books.count == 1)
+        let request = try #require(await transport.requests().first)
+        let body = try JSONValue.decode(data: #require(request.httpBody))
+        #expect(body["keyword"] == .string("斗罗大陆"))
+        #expect(body["page_index"] == .number(1))
+        #expect(body["page_size"] == .number(20))
+        #expect(body["enabled"] == .bool(true))
+        #expect(body["cursor"] == .null)
+        #expect(request.value(forHTTPHeaderField: "Content-Type")?.hasPrefix("application/json") == true)
+    }
+
+    @Test func lenientSeparatorsDoNotRewriteQuotedCookieOrURLValues() throws {
+        let value = try LegadoURLBuilder.decodeLenientJSON(#"{type=>app, 'Cookie':'token=a=b; mode=pc', referer:'https://fixture.invalid/?a=b', retry:2, enabled:false, empty:null}"#)
+        #expect(value["type"] == .string("app"))
+        #expect(value["Cookie"] == .string("token=a=b; mode=pc"))
+        #expect(value["referer"] == .string("https://fixture.invalid/?a=b"))
+        #expect(value["retry"] == .number(2))
+        #expect(value["enabled"] == .bool(false))
+        #expect(value["empty"] == .null)
+    }
+
+    @Test func unquotedWebViewFlagUsedByAudioSourceParses() throws {
+        let request = try LegadoURLBuilder().prepare("https://request.fixture.invalid/audio,{webView:“true”}")
+        #expect(request.useWebView)
+    }
+
+    @Test func lenientValuesKeepCommentsAndPutRuleExpressionsWorking() throws {
+        let options = try LegadoURLBuilder.decodeLenientJSON("{retry: 2 /* comment */, enabled: FALSE, word: 小说}")
+        #expect(options["retry"] == .number(2))
+        #expect(options["enabled"] == .bool(false))
+        #expect(options["word"] == .string("小说"))
+        let put = try LegadoURLBuilder.decodeLenientJSON("{id:$.books[0].id, name:name}", quoteUnquotedValues: true)
+        #expect(put["id"] == .string("$.books[0].id"))
+        #expect(put["name"] == .string("name"))
+        #expect(throws: (any Error).self) { try LegadoURLBuilder.decodeLenientJSON("{broken:}") }
+    }
+
+    private func source() -> BookSource {
+        var source = BookSource(bookSourceUrl: "https://request.fixture.invalid", bookSourceName: "请求兼容测试")
+        source.ruleSearch = .init(bookList: "$.books", name: "$.name", bookUrl: "$.url")
+        return source
+    }
+}
+
+private actor SampledRequestTransport: HTTPTransport {
+    private var values: [URLRequest] = []
+    func data(for request: URLRequest) async throws -> HTTPResponse {
+        values.append(request)
+        return .init(data: Data(#"{"books":[{"name":"测试书","url":"/book/42"}]}"#.utf8),
+                     statusCode: 200, headers: [:], finalURL: try #require(request.url))
+    }
+    func requests() -> [URLRequest] { values }
+}

Tests/LegadoCoreTests/SampledSourceRegressionTests.swift

@@ -0,0 +1,104 @@
+import Foundation
+import Testing
+@testable import LegadoCore
+
+// All responses and hosts are synthetic; these tests never contact source websites.
+struct SampledSourceRegressionTests {
+    @Test func JavaScriptNumbersZeroAndOneRemainNumeric() throws {
+        let evaluator = JavaScriptRuleEvaluator()
+        let value = try evaluator.evaluate(script: "({zero: 0, one: 1, yes: true, no: false, list: [0, 1, true]})", result: .null, executionContext: .init()).value
+        #expect(value["zero"] == .number(0))
+        #expect(value["one"] == .number(1))
+        #expect(value["yes"] == .bool(true))
+        #expect(value["no"] == .bool(false))
+        #expect(value["list"] == .array([.number(0), .number(1), .bool(true)]))
+    }
+
+    @Test func JavaScriptGetStringUsesTheSameJSONTemplateAndReplacementSemantics() async throws {
+        let input = try JSONValue.decode(data: Data(#"{"id":42,"name":"测试小说","alternate":"备用小说"}"#.utf8))
+        let value = await RuleEngine().evaluate("@js:java.getString('/book/{$.id}') + ':' + java.getString('$.name||$.alternate##小说')", in: input)
+        #expect(value.diagnostics.isEmpty)
+        #expect(value.strings == ["/book/42:测试"])
+    }
+
+    @Test func legacyJSONTemplatesKeepEachSearchBookURLDistinct() async throws {
+        var source = BookSource(bookSourceUrl: "https://fixture.invalid", bookSourceName: "旧式 JSON 模板")
+        source.searchUrl = "/search?word={{key}}"
+        source.ruleSearch = .init(bookList: "data.bookList", name: "name", bookUrl: "/pc/book/{$.id}/catalog")
+        let outcome = await BookSearchService(transport: SampledResponseTransport(body:
+            #"{"data":{"bookList":[{"id":42,"name":"测试甲"},{"id":43,"name":"测试乙"}]}}"#
+        )).search(keyword: "测试", page: 1, source: source)
+        #expect(outcome.diagnostics.isEmpty)
+        #expect(outcome.books.map(\.bookURL) == ["https://fixture.invalid/pc/book/42/catalog", "https://fixture.invalid/pc/book/43/catalog"])
+    }
+
+    @Test func legacyJSONRootWildcardAndRecursiveURLTemplate() async throws {
+        var source = BookSource(bookSourceUrl: "https://fixture.invalid", bookSourceName: "旧式 JSON 列表")
+        source.searchUrl = "/search?word={{key}}"
+        source.ruleSearch = .init(bookList: "$.[*]", name: "$..name", bookUrl: "https://fixture.invalid{$..url}")
+        let outcome = await BookSearchService(transport: SampledResponseTransport(body:
+            #"[{"name":"测试甲","url":"/book/42"},{"name":"测试乙","url":"/book/43"}]"#
+        )).search(keyword: "测试", page: 1, source: source)
+        #expect(outcome.diagnostics.isEmpty)
+        #expect(outcome.books.map(\.bookURL) == ["https://fixture.invalid/book/42", "https://fixture.invalid/book/43"])
+    }
+
+    @Test func legacyJSONTemplatesRespectQuotedBracesAndReplacement() async throws {
+        let input = try JSONValue.decode(data: Data(#"{"id":42,"words":1200,"meta":{"label}":"玄幻"},"tags":["甲","乙"]}"#.utf8))
+        let value = await RuleEngine().evaluate(#"{$.meta['label}']}/{$.words}字/{$.tags[*]}##1200##1.2千"#, in: input)
+        #expect(value.diagnostics.isEmpty)
+        #expect(value.strings == ["玄幻/1.2千字/甲\n乙"])
+        let fromText = await RuleEngine().evaluate("/book/{$.id}", inText: input.stringValue)
+        #expect(fromText.diagnostics.isEmpty)
+        #expect(fromText.strings == ["/book/42"])
+    }
+
+    @Test func JSONFallbackSkipsEmptyArraysAndEmptyNames() async throws {
+        let input = try JSONValue.decode(data: Data(#"{"empty":[],"books":[{"name":"备用书"}],"name":"","alternate":"备用书","zero":0}"#.utf8))
+        for useText in [false, true] {
+            let list = if useText {
+                await RuleEngine().evaluate("$.empty||$.books", inText: input.stringValue, listMode: true)
+            } else {
+                await RuleEngine().evaluate("$.empty||$.books", in: input, listMode: true)
+            }
+            #expect(list.diagnostics.isEmpty)
+            #expect(list.listItems(rule: "$.empty||$.books").count == 1)
+            let name = await RuleEngine().evaluate("$.name||$.alternate", in: input)
+            #expect(name.strings == ["备用书"])
+            let zero = await RuleEngine().evaluate("$.zero||$.alternate", in: input)
+            #expect(zero.strings == ["0"])
+        }
+    }
+
+    @Test func replacementAppliesToTheCombinedSelectorResult() async throws {
+        // Search names in the sampled fallback sources and multi-part metadata
+        // use one trailing ## expression for the entire selector combination.
+        let input = try JSONValue.decode(data: Data(#"{"name":"测试小说","fallback":"备用小说","a":"A1","b":"B2"}"#.utf8))
+        let fallback = await RuleEngine().evaluate("$.name||$.fallback##小说", in: input)
+        #expect(fallback.strings == ["测试"])
+        let combined = await RuleEngine().evaluate("$.a&&$.b##[0-9]", in: input)
+        #expect(combined.strings.joined(separator: "\n") == "A\nB")
+        let anchored = await RuleEngine().evaluate(#"$.a&&$.b##^A1\nB2$##整段"#, in: input)
+        #expect(anchored.strings == ["整段"])
+        let list = await RuleEngine().evaluate("$.a&&$.b##[0-9]", in: input, listMode: true)
+        #expect(list.strings == ["A", "B"])
+        let html = await RuleEngine().evaluate("h1@text||h2@text##小说", inText: "<h1>测试小说</h1>")
+        #expect(html.strings == ["测试"])
+    }
+
+    @Test func replacementTextIsNotParsedAsSelectorSyntax() async {
+        let result = await RuleEngine().evaluate(#"h1@text##["']|[()]"#, inText: "<h1>书名(测试)'</h1>")
+        #expect(result.diagnostics.isEmpty)
+        #expect(result.strings == ["书名测试"])
+        let replacement = await RuleEngine().evaluate("h1@text##书##A&&B", inText: "<h1>书</h1>")
+        #expect(replacement.diagnostics.isEmpty)
+        #expect(replacement.strings == ["A&&B"])
+    }
+}
+
+private struct SampledResponseTransport: HTTPTransport {
+    let body: String
+    func data(for request: URLRequest) async throws -> HTTPResponse {
+        .init(data: Data(body.utf8), statusCode: 200, headers: [:], finalURL: try #require(request.url))
+    }
+}

Tests/LegadoReaderTests/ReaderScrollGeometryTests.swift

@@ -246,6 +246,54 @@ final class ReaderScrollGeometryTests: XCTestCase {
         XCTAssertEqual(fixture.view.contentOffset.y, layout.frames[row].minY, accuracy: 1 / fixture.window.screen.scale)
     }
 
+    func testRepeatedShortChapterPrependingDoesNotAccumulateOffsetDrift() async throws {
+        func short(_ index: Int) -> ReaderScrollChapter {
+            .init(id: index, title: "第\(index + 1)章 · 连续阅读", content: "短章正文，继续向下可以自然读到下一章。")
+        }
+        let fixture = host(chapters: (20...22).map(short), active: 20)
+        defer { fixture.coordinator.displayLink?.invalidate(); fixture.window.isHidden = true }
+        for first in stride(from: 19, through: 0, by: -1) {
+            fixture.coordinator.parent = surface(chapters: (first...22).map(short), active: 20)
+            fixture.coordinator.update(fixture.view)
+            try await Task.sleep(nanoseconds: 10_000_000)
+        }
+        let layout = fixture.view.collectionViewLayout as! ReaderContinuousLayout
+        let row = try XCTUnwrap(fixture.coordinator.blocks.firstIndex { $0.chapter.chapter.id == 20 })
+        XCTAssertEqual(fixture.view.contentOffset.y, layout.frames[row].minY,
+                       accuracy: 1 / fixture.window.screen.scale)
+        XCTAssertTrue(fixture.view.accessibilityValue?.contains("chapter=20 ") == true,
+                      fixture.view.accessibilityValue ?? "Missing viewport report")
+    }
+
+    func testFractionalPixelBoundariesBelongToTheFollowingRow() {
+        let layout = ReaderContinuousLayout()
+        let frames = (0..<100).map { CGRect(x: 0, y: CGFloat($0 * 5) / 3, width: 390, height: 5.0 / 3) }
+        layout.install(frames: frames, size: CGSize(width: 390, height: 500.0 / 3), correction: 0)
+        for index in frames.indices {
+            XCTAssertEqual(layout.row(at: frames[index].minY), index)
+        }
+    }
+
+    func testPrependingPreservesTheChapterBelowTopPadding() async throws {
+        func short(_ index: Int) -> ReaderScrollChapter {
+            .init(id: index, title: "短章\(index)", content: "短章正文")
+        }
+        let fixture = host(chapters: [short(20)], active: 20)
+        defer { fixture.coordinator.displayLink?.invalidate(); fixture.window.isHidden = true }
+        // A short initial chapter can settle at offset zero with its heading
+        // still below the reading area's top padding.
+        fixture.view.setContentOffset(.zero, animated: false)
+        fixture.coordinator.parent = surface(chapters: (19...21).map(short), active: 20)
+        fixture.coordinator.update(fixture.view)
+        try await Task.sleep(nanoseconds: 10_000_000)
+        let layout = fixture.view.collectionViewLayout as! ReaderContinuousLayout
+        let row = try XCTUnwrap(fixture.coordinator.blocks.firstIndex { $0.chapter.chapter.id == 20 })
+        XCTAssertEqual(layout.frames[row].minY - fixture.view.contentOffset.y, 16,
+                       accuracy: 1 / fixture.window.screen.scale)
+        XCTAssertTrue(fixture.view.accessibilityValue?.contains("chapter=20 ") == true,
+                      fixture.view.accessibilityValue ?? "Missing viewport report")
+    }
+
     private func chapter(_ index: Int) -> ReaderScrollChapter {
         .init(id: index, title: "第\(index)章", content: String(repeating: "正文包含中文、英文 words 与家庭表情👨‍👩‍👧‍👦，用于验证连续阅读的排版和位置。\n", count: 100))
     }

Tests/LegadoReaderTests/WebReaderBridgeTests.swift

@@ -88,3 +88,159 @@ final class WebReaderBridgeTests: XCTestCase {
         XCTAssertEqual(result as? String, "已保存|规则正文")
     }
 }
+
+private actor RSSPaginationTransport: HTTPTransport {
+    private var requests: [URL] = []
+    private var shouldFailSecondPage = true
+
+    func data(for request: URLRequest) async throws -> HTTPResponse {
+        let url = try XCTUnwrap(request.url)
+        requests.append(url)
+        if url.host == "unavailable.test" { throw URLError(.cannotConnectToHost) }
+        let page = URLComponents(url: url, resolvingAgainstBaseURL: false)?.queryItems?
+            .first { $0.name == "page" }?.value ?? "1"
+        if page == "2", shouldFailSecondPage {
+            shouldFailSecondPage = false
+            throw URLError(.networkConnectionLost)
+        }
+        let body = "{\"items\":[{\"title\":\"Page \(page)\",\"link\":\"https://feed.test/article/\(page)\"}]}"
+        return .init(data: Data(body.utf8), statusCode: 200, headers: [:], finalURL: url)
+    }
+
+    func requestedPages() -> [String] {
+        requests.compactMap { URLComponents(url: $0, resolvingAgainstBaseURL: false)?.queryItems?.first { $0.name == "page" }?.value }
+    }
+
+    func requestCount() -> Int { requests.count }
+}
+
+private actor RSSDelayedTransport: HTTPTransport {
+    let pausedHost: String
+    let pausedPage: String
+    private var requestTotal = 0
+    private var pending: CheckedContinuation<Void, Never>?
+
+    init(host: String, page: String) { pausedHost = host; pausedPage = page }
+
+    func data(for request: URLRequest) async throws -> HTTPResponse {
+        let url = try XCTUnwrap(request.url)
+        let page = URLComponents(url: url, resolvingAgainstBaseURL: false)?.queryItems?
+            .first { $0.name == "page" }?.value ?? "1"
+        requestTotal += 1
+        if url.host == pausedHost, page == pausedPage {
+            await withCheckedContinuation { continuation in
+                pending = continuation
+            }
+        }
+        let body = "{\"items\":[{\"title\":\"Page \(page)\",\"link\":\"https://feed.test/article/\(page)\"}]}"
+        return .init(data: Data(body.utf8), statusCode: 200, headers: [:], finalURL: url)
+    }
+
+    func waitUntilPaused() async throws {
+        for _ in 0..<500 {
+            if pending != nil { return }
+            try await Task.sleep(nanoseconds: 10_000_000)
+        }
+        throw URLError(.timedOut)
+    }
+
+    func resume() { pending?.resume(); pending = nil }
+    func requestCount() -> Int { requestTotal }
+}
+
+@MainActor
+final class RSSLibraryRegressionTests: XCTestCase {
+    func testUnavailableDatabaseDoesNotCrashOrPublishUnsavedImports() async throws {
+        let blocked = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        try Data("existing file".utf8).write(to: blocked)
+        defer { try? FileManager.default.removeItem(at: blocked) }
+        let model = RSSLibraryModel.live(databaseURL: blocked.appendingPathComponent("rss.sqlite3"))
+        XCTAssertNotNil(model.errorMessage)
+        await model.load()
+        XCTAssertNotNil(model.errorMessage)
+        let saved = await model.save(source(), replacing: nil)
+        XCTAssertFalse(saved)
+        await model.importText(#"{"sourceUrl":"https://feed.test","sourceName":"未保存的订阅"}"#)
+        XCTAssertTrue(model.sources.isEmpty)
+        XCTAssertFalse(model.isWorking)
+        XCTAssertEqual(try Data(contentsOf: blocked), Data("existing file".utf8))
+    }
+
+    private func source(host: String = "feed.test") -> RssSource {
+        var source = RssSource(sourceUrl: "https://\(host)", sourceName: host)
+        source.sortUrl = "https://\(host)/list?page={{page}}"
+        source.ruleArticles = "$.items"
+        source.ruleTitle = "$.title"
+        source.ruleLink = "$.link"
+        source.ruleNextPage = "PAGE"
+        return source
+    }
+
+    func testFailedNextPageRetriesTheSamePageWithoutSkippingArticles() async throws {
+        let directory = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        defer { try? FileManager.default.removeItem(at: directory) }
+        let database = try LegadoDatabase(url: directory.appendingPathComponent("rss.sqlite3"))
+        let transport = RSSPaginationTransport()
+        let model = RSSLibraryModel(database: database, service: RSSService(transport: transport))
+        let feed = source()
+        await model.refresh(feed)
+        XCTAssertEqual(model.articles.map(\.title), ["Page 1"])
+        await model.loadMore(feed)
+        XCTAssertNotNil(model.errorMessage)
+        XCTAssertEqual(model.articles.map(\.title), ["Page 1"])
+        await model.loadMore(feed)
+        let pages = await transport.requestedPages()
+        XCTAssertEqual(pages, ["1", "2", "2"])
+        XCTAssertEqual(model.articles.map(\.title), ["Page 1", "Page 2"])
+    }
+
+    func testFailedSourceChangeDoesNotReusePreviousSourcesArticlesOrNextPage() async throws {
+        let directory = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        defer { try? FileManager.default.removeItem(at: directory) }
+        let database = try LegadoDatabase(url: directory.appendingPathComponent("rss.sqlite3"))
+        let transport = RSSPaginationTransport()
+        let model = RSSLibraryModel(database: database, service: RSSService(transport: transport))
+        await model.refresh(source())
+        let unavailable = source(host: "unavailable.test")
+        await model.refresh(unavailable)
+        XCTAssertTrue(model.articles.isEmpty)
+        let count = await transport.requestCount()
+        await model.loadMore(unavailable)
+        let afterLoadMore = await transport.requestCount()
+        XCTAssertEqual(afterLoadMore, count)
+    }
+
+    func testSlowPreviousSourceCannotReplaceTheCurrentFeed() async throws {
+        let directory = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        defer { try? FileManager.default.removeItem(at: directory) }
+        let database = try LegadoDatabase(url: directory.appendingPathComponent("rss.sqlite3"))
+        let transport = RSSDelayedTransport(host: "slow.test", page: "1")
+        let model = RSSLibraryModel(database: database, service: RSSService(transport: transport))
+        let slow = Task { await model.refresh(source(host: "slow.test")) }
+        try await transport.waitUntilPaused()
+        await model.refresh(source())
+        XCTAssertTrue(model.isWorking)
+        await transport.resume()
+        await slow.value
+        XCTAssertEqual(model.articles.map(\.origin), ["https://feed.test"])
+        XCTAssertFalse(model.isWorking)
+    }
+
+    func testRepeatedLoadMoreWhileWaitingRequestsOnlyOnePage() async throws {
+        let directory = FileManager.default.temporaryDirectory.appendingPathComponent(UUID().uuidString)
+        defer { try? FileManager.default.removeItem(at: directory) }
+        let database = try LegadoDatabase(url: directory.appendingPathComponent("rss.sqlite3"))
+        let transport = RSSDelayedTransport(host: "feed.test", page: "2")
+        let model = RSSLibraryModel(database: database, service: RSSService(transport: transport))
+        let feed = source()
+        await model.refresh(feed)
+        let pending = Task { await model.loadMore(feed) }
+        try await transport.waitUntilPaused()
+        await model.loadMore(feed)
+        let count = await transport.requestCount()
+        XCTAssertEqual(count, 2)
+        await transport.resume()
+        await pending.value
+        XCTAssertEqual(model.articles.map(\.title), ["Page 1", "Page 2"])
+    }
+}

Tests/LegadoReaderUITests/DarkModeUITests.swift

@@ -80,11 +80,14 @@ final class DarkModeUITests: XCTestCase {
             let keyboard = try luminances(app.keyboards.firstMatch, in: app)
             XCTAssertLessThan(keyboard[keyboard.count / 2], 0.15)
         }
-        let accessory = app.buttons["完成"]
+        let accessory = app.buttons["syntaxCodeEditor.done"]
-        let accessoryFrame = accessory.frame.intersection(app.frame)
+        XCTAssertTrue(accessory.exists && accessory.isHittable)
-        if accessory.isHittable && !accessoryFrame.isNull && accessoryFrame.width > 20 && accessoryFrame.height > 20 {
+        try assertReadableDarkFill(accessory, in: app)
-            try assertReadableDarkFill(accessory, in: app)
+        app.buttons["@"].tap()
-        }
+        XCTAssertTrue((app.textViews["syntaxCodeEditor"].value as? String)?.contains("@") == true)
+        accessory.tap()
+        let dismissed = expectation(for: NSPredicate { _, _ in !accessory.exists || !accessory.isHittable }, evaluatedWith: nil)
+        wait(for: [dismissed], timeout: 5)
     }
 
     func testDarkSettingsRulesAndManagementPages() throws {
@@ -146,8 +149,8 @@ final class DarkModeUITests: XCTestCase {
         let app = launch(["-uiDemoTab", "search", "-themeMode", "2", "-uiAutomationSearch", "深色外观检查"])
         let cancelSearch = app.buttons.matching(NSPredicate(format: "label IN %@", ["取消", "Cancel"])).firstMatch
         if cancelSearch.waitForExistence(timeout: 3) { cancelSearch.tap() }
-        XCTAssertTrue(app.buttons["搜索选项"].waitForExistence(timeout: 10))
+        XCTAssertTrue(app.buttons["search.options"].waitForExistence(timeout: 10))
-        app.buttons["搜索选项"].tap()
+        app.buttons["search.options"].tap()
         app.buttons["分组或书源"].tap()
         XCTAssertTrue(app.navigationBars["分组或书源"].waitForExistence(timeout: 5))
         try captureDarkPage(app, name: "Search group filter")

Tests/LegadoReaderUITests/LiveReadingFlowUITests.swift

@@ -108,7 +108,10 @@ final class LiveReadingFlowUITests: XCTestCase {
         showReaderMenu()
         app.buttons["换源"].tap()
         XCTAssertTrue(app.navigationBars["为《极道天魔》换源"].waitForExistence(timeout: 10))
-        let replacement = app.buttons["sourceSwitch.candidate." + replacementID]
+        let replacement = app.buttons.matching(NSPredicate(
+            format: "identifier BEGINSWITH %@ AND label CONTAINS %@",
+            "sourceSwitch.candidate." + replacementID + "|", "滚开"
+        )).firstMatch
         XCTAssertTrue(replacement.waitForExistence(timeout: 240))
         attach("07-source-candidates")
         replacement.tap()

Tests/LegadoReaderUITests/ReaderScrollGestureTests.swift

@@ -354,8 +354,13 @@ final class ReaderScrollGestureTests: XCTestCase {
         app.launch()
         let reader = app.collectionViews["reader.continuous"]
         XCTAssertTrue(reader.waitForExistence(timeout: 20))
+        var lastInitialState = "<not reported>"
         let ready = expectation(for: NSPredicate { _, _ in
-            let text = reader.value as? String ?? ""
+            let text = reader.value as? String ?? "<missing>"
+            if text != lastInitialState {
+                print("[ContinuousReaderInitial] \(text)")
+                lastInitialState = text
+            }
             return text.contains("chapter=20 ") && !text.contains("chapters=1 ")
         }, evaluatedWith: nil)
         wait(for: [ready], timeout: 20)

Tests/LegadoReaderUITests/RuleEditorUITests.swift

@@ -2,34 +2,89 @@ import XCTest
 
 @MainActor
 final class RuleEditorUITests: XCTestCase {
-    func testBookshelfTXTFilePickerImportsTXTAndDisablesJSON() throws {
+    func testRSSSearchSubmissionPaginationAndFavorites() throws {
-        try exerciseLocalTXTImport(fromFileManagement: false)
+        guard let fixture = ProcessInfo.processInfo.environment["LEGADO_RSS_UI_FIXTURE_URL"] else {
-    }
+            throw XCTSkip("Set LEGADO_RSS_UI_FIXTURE_URL to the local RSS fixture source URL")
-
+        }
-    func testFileManagementTXTFilePickerImportsTXTAndDisablesJSON() throws {
-        try exerciseLocalTXTImport(fromFileManagement: true)
-    }
-
-    private func exerciseLocalTXTImport(fromFileManagement: Bool) throws {
         continueAfterFailure = false
         let app = XCUIApplication()
-        app.launchArguments = ["-uiDemo", "-themeMode", "1", "-bookshelfLayout", "grid3"]
+        app.launchArguments = ["-uiDemo", "-uiDemoTab", "rss", "-themeMode", "1"]
+        app.launch()
+        XCTAssertTrue(app.navigationBars["RSS"].waitForExistence(timeout: 20))
+        app.buttons["RSS操作"].tap()
+        app.buttons["URL或粘贴JSON"].tap()
+        let editor = app.textViews.firstMatch
+        XCTAssertTrue(editor.waitForExistence(timeout: 5))
+        editor.tap()
+        editor.typeText(fixture)
+        app.navigationBars["导入RSS源"].buttons["导入"].tap()
+        let source = app.staticTexts["RSS 全流程测试"]
+        XCTAssertTrue(source.waitForExistence(timeout: 10))
+        source.tap()
+        XCTAssertTrue(app.staticTexts["首页文章"].waitForExistence(timeout: 10))
+        let search = app.searchFields["源内搜索"]
+        if !search.exists { app.swipeDown() }
+        XCTAssertTrue(search.waitForExistence(timeout: 5))
+        search.tap()
+        search.typeText("needle\n")
+        let result = app.staticTexts["搜索结果 needle 第1页"]
+        XCTAssertTrue(result.waitForExistence(timeout: 10))
+        let nextPage = app.buttons["加载下一页"]
+        let ready = expectation(for: NSPredicate { _, _ in nextPage.exists && nextPage.isEnabled }, evaluatedWith: nil)
+        wait(for: [ready], timeout: 10)
+        nextPage.tap()
+        XCTAssertTrue(app.staticTexts["搜索结果 needle 第2页"].waitForExistence(timeout: 10))
+        result.tap()
+        XCTAssertTrue(app.navigationBars["文章"].waitForExistence(timeout: 5))
+        if app.buttons["取消收藏"].exists { app.buttons["取消收藏"].tap() }
+        XCTAssertTrue(app.buttons["收藏"].waitForExistence(timeout: 5))
+        app.buttons["收藏"].tap()
+        XCTAssertTrue(app.buttons["取消收藏"].waitForExistence(timeout: 5))
+        app.navigationBars["文章"].buttons.element(boundBy: 0).tap()
+        let cancelSearch = app.navigationBars["RSS 全流程测试"].buttons.matching(NSPredicate(
+            format: "label IN %@", ["Cancel", "取消", "Close", "关闭"]
+        )).firstMatch
+        if cancelSearch.exists { cancelSearch.tap() }
+        app.navigationBars["RSS 全流程测试"].buttons["RSS"].tap()
+        XCTAssertTrue(app.navigationBars["RSS"].waitForExistence(timeout: 5))
+        app.segmentedControls.buttons["收藏"].tap()
+        XCTAssertTrue(result.waitForExistence(timeout: 5))
+        attach(app, name: "RSS search article saved to favorites")
+        app.terminate()
+        app.launch()
+        XCTAssertTrue(app.staticTexts["RSS 全流程测试"].waitForExistence(timeout: 10))
+        app.segmentedControls.buttons["收藏"].tap()
+        XCTAssertTrue(result.waitForExistence(timeout: 5))
+    }
+
+    func testBookshelfFilePickerImportsTXTAndEPUBAndDisablesJSON() throws {
+        try exerciseLocalBookImport(fromFileManagement: false)
+    }
+
+    func testFileManagementFilePickerImportsTXTAndEPUBAndDisablesJSON() throws {
+        try exerciseLocalBookImport(fromFileManagement: true)
+    }
+
+    private func exerciseLocalBookImport(fromFileManagement: Bool) throws {
+        continueAfterFailure = false
+        let app = XCUIApplication()
+        app.launchArguments = ["-uiDemo", "-themeMode", "1", "-bookshelfLayout", "grid3", "-pageAnim", "0"]
         if fromFileManagement { app.launchArguments += ["-uiDemoTab", "settings", "-uiDemoFiles"] }
         app.launch()
-        let importTitle = fromFileManagement ? "导入 TXT 文件" : "本地书"
+        let importTitle = fromFileManagement ? "导入 TXT / EPUB" : "本地书"
         let more = fromFileManagement
             ? app.navigationBars["文件管理"].buttons["更多"].firstMatch
             : app.buttons["更多"].firstMatch
         XCTAssertTrue(more.waitForExistence(timeout: 20))
         more.tap()
-        app.buttons[importTitle].tap()
+        app.buttons[importTitle].firstMatch.tap()
         let cancel = app.buttons.matching(NSPredicate(format: "label IN %@", ["Cancel", "取消"])).firstMatch
         XCTAssertTrue(cancel.waitForExistence(timeout: 10))
         cancel.tap()
         let closed = expectation(for: NSPredicate { _, _ in !cancel.exists }, evaluatedWith: nil)
         wait(for: [closed], timeout: 10)
         more.tap()
-        app.buttons[importTitle].tap()
+        app.buttons[importTitle].firstMatch.tap()
         XCTAssertTrue(cancel.waitForExistence(timeout: 10))
 
         let pickerFiles = app.collectionViews["File View"].cells
@@ -66,6 +121,10 @@ final class RuleEditorUITests: XCTestCase {
         let uppercase = pickerFiles.matching(NSPredicate(format: "label BEGINSWITH %@", "TXT-Uppercase")).firstMatch
         XCTAssertTrue(uppercase.waitForExistence(timeout: 5))
         uppercase.tap()
+        let epub = pickerFiles.matching(NSPredicate(format: "label BEGINSWITH %@", "EPUB-Book")).firstMatch
+        XCTAssertTrue(epub.waitForExistence(timeout: 5))
+        XCTAssertTrue(epub.isEnabled)
+        epub.tap()
         XCTAssertTrue(open.waitForExistence(timeout: 5))
         XCTAssertTrue(open.isEnabled)
         attach(app, name: "TXT selectable and JSON excluded")
@@ -78,14 +137,34 @@ final class RuleEditorUITests: XCTestCase {
             XCTAssertTrue(app.navigationBars["导入本地书"].waitForExistence(timeout: 15))
             XCTAssertFalse(app.staticTexts["JSON-Source.json"].exists)
             XCTAssertFalse(app.staticTexts["JSON-Uppercase.JSON"].exists)
-            XCTAssertTrue(app.staticTexts["TXT-Book.txt"].exists)
+            XCTAssertTrue(app.staticTexts["TXT-Book.txt"].waitForExistence(timeout: 10), app.debugDescription)
-            XCTAssertTrue(app.staticTexts["TXT-Uppercase.TXT"].exists)
+            XCTAssertTrue(app.staticTexts["TXT-Uppercase.TXT"].waitForExistence(timeout: 10), app.debugDescription)
             app.buttons["导入"].tap()
         }
         // Repeated runs preserve managed files and correctly add a numeric suffix.
         XCTAssertTrue(app.staticTexts.matching(NSPredicate(format: "label BEGINSWITH %@", "TXT-Book")).firstMatch.waitForExistence(timeout: 15))
         XCTAssertTrue(app.staticTexts.matching(NSPredicate(format: "label BEGINSWITH %@", "TXT-Uppercase")).firstMatch.waitForExistence(timeout: 15))
-        attach(app, name: "Local TXT books imported onto shelf")
+        let epubBook = app.staticTexts["EPUB 导入测试"].firstMatch
+        XCTAssertTrue(epubBook.waitForExistence(timeout: 15))
+        attach(app, name: "Local TXT and EPUB books imported onto shelf")
+        epubBook.tap()
+        let reader = app.textViews["reader.text"]
+        XCTAssertTrue(reader.waitForExistence(timeout: 15))
+        XCTAssertTrue((reader.value as? String)?.contains("这是 EPUB 导入后的第一章正文") == true)
+        reader.coordinate(withNormalizedOffset: CGVector(dx: 0.5, dy: 0.5)).tap()
+        app.buttons["更多"].firstMatch.tap()
+        app.buttons["本章插图"].tap()
+        XCTAssertTrue(app.navigationBars["本章插图"].waitForExistence(timeout: 5))
+        XCTAssertTrue(app.images["书籍插图"].waitForExistence(timeout: 10))
+        app.buttons["完成"].tap()
+        reader.coordinate(withNormalizedOffset: CGVector(dx: 0.5, dy: 0.5)).tap()
+        app.buttons["目录"].tap()
+        app.staticTexts["第二章 EPUB"].firstMatch.tap()
+        let loaded = expectation(for: NSPredicate { _, _ in
+            (reader.value as? String)?.contains("目录锚点跳转成功") == true
+        }, evaluatedWith: nil)
+        wait(for: [loaded], timeout: 10)
+        attach(app, name: "EPUB second chapter after TOC navigation")
     }
 
     func testJSONFilePickerOffersOpenAndImportsSelectedFiles() throws {
@@ -146,7 +225,7 @@ final class RuleEditorUITests: XCTestCase {
         attach(app, name: "JSON import review after file picker dismissal")
         app.buttons["导入"].tap()
         if app.alerts.firstMatch.waitForExistence(timeout: 5) { app.alerts.buttons["好"].tap() }
-        XCTAssertTrue(app.staticTexts["文件导入 JSON 测试"].waitForExistence(timeout: 10))
+        XCTAssertTrue(app.staticTexts["文件导入 JSON 测试"].waitForExistence(timeout: 10), app.debugDescription)
         XCTAssertTrue(app.staticTexts["文件导入大写扩展名测试"].exists)
     }
 

docs/DEVELOPMENT.md

@@ -41,7 +41,7 @@ sh scripts/self-check.sh
 
 阅读器测试在 Xcode 选择 `LegadoReaderTests` / `LegadoReaderUITests` Scheme 和具体模拟器运行。联网书源测试需要显式配置对应环境变量，默认核心回归主要采用自造响应，不能将其通过解读为所有线上书源均可用。
 
-书源和本地 TXT 文件选择回归需要先在已启动的测试模拟器中准备文件共享应用（将下方 UUID 换成实际值）：
+书源和本地 TXT / EPUB 文件选择回归需要先在已启动的测试模拟器中准备文件共享应用（将下方 UUID 换成实际值）：
 
 ```sh
 LEGADO_TEST_DEVICE="模拟器 UUID"
@@ -49,12 +49,14 @@ sh scripts/prepare-source-import-fixture.sh "$LEGADO_TEST_DEVICE"
 xcodebuild -project Legado.xcodeproj -scheme LegadoReaderUITests \
   -destination "platform=iOS Simulator,id=$LEGADO_TEST_DEVICE" \
   -only-testing:LegadoReaderUITests/RuleEditorUITests/testJSONFilePickerOffersOpenAndImportsSelectedFiles \
-  -only-testing:LegadoReaderUITests/RuleEditorUITests/testBookshelfTXTFilePickerImportsTXTAndDisablesJSON \
+  -only-testing:LegadoReaderUITests/RuleEditorUITests/testBookshelfFilePickerImportsTXTAndEPUBAndDisablesJSON \
-  -only-testing:LegadoReaderUITests/RuleEditorUITests/testFileManagementTXTFilePickerImportsTXTAndDisablesJSON \
+  -only-testing:LegadoReaderUITests/RuleEditorUITests/testFileManagementFilePickerImportsTXTAndEPUBAndDisablesJSON \
   -collect-test-diagnostics never CODE_SIGNING_ALLOWED=NO test
 ```
 
-独立测试应用提供 JSON 书源和 `.txt`、`.TXT` 小说样本，不会被打包进正式 App。此回归检查取消后重新打开、书源 JSON 多选、本地 TXT 可选且 JSON 不可选、导入确认及最终保存。
+独立测试应用提供 JSON 书源和 `.txt`、`.TXT`、`.epub` 小说样本，不会被打包进正式 App。此回归检查取消后重新打开、书源 JSON 多选、本地 TXT / EPUB 可选且 JSON 不可选、导入确认、最终保存、EPUB 正文、插图与目录跳转。
+
+RSS 的导入、源内搜索、分页、正文、收藏以及重启恢复可通过本地自造响应回归，避免依赖第三方网站。先在一个终端运行 `python3 scripts/fixtures/rss-flow-server.py`，再在 Xcode 选择 `LegadoRSSFlowUITests` Scheme 和模拟器运行；也可用 `xcodebuild -project Legado.xcodeproj -scheme LegadoRSSFlowUITests -destination 'platform=iOS Simulator,id=模拟器UUID' CODE_SIGNING_ALLOWED=NO test`。该服务仅监听本机 `127.0.0.1:18763`，测试结束后按 Ctrl-C 停止。普通 UI Scheme 默认跳过此用例。
 
 ## 兼容性验证
 
@@ -101,7 +103,7 @@ source_release_tag=$(python3 -c 'import json,sys; print(json.load(open(sys.argv[
 gh release download "$source_release_tag" --repo Aminark/legado-ios \
   --pattern '*.ipa' --dir "$source_update_dir"
 python3 scripts/update-altstore-source.py \
-  --ipa "$source_update_dir/legado-ios-v0.1.11-build17-unsigned.ipa" \
+  --ipa "$source_update_dir/legado-ios-v0.1.12-build18-unsigned.ipa" \
   --release "$source_update_dir/release.json"
 python3 -m unittest discover -s scripts/tests -v
 ```
@@ -111,3 +113,9 @@ python3 -m unittest discover -s scripts/tests -v
 发布后检查固定地址可匿名访问、图标和截图能正常加载，并下载源中 IPA 再核对大小、摘要和版本。客户端的添加源、安装、更新与签名流程需分别实测；源文件校验通过不代表已完成真机安装验证。
 
 格式说明：[AltStore 源规范](https://faq.altstore.io/developers/make-a-source)、[SideStore 源字段](https://sidestore.io/sidestore-source-types/interfaces/Source.html)、[LiveContainer 源用法](https://livecontainer.github.io/docs/guides/sources)。
+
+## EPUB 回归
+
+`swift test --disable-sandbox --filter EPUB` 覆盖 EPUB 2/3 元数据、中文资源路径、同页锚点、无目录降级、书脊顺序、插图、损坏与加密文件、重命名导出、导入失败回滚和阅读进度恢复。样本均由测试自行生成，不依赖网络或第三方书籍。
+
+解析器按需读取 ZIP 资源，不解压到磁盘；单资源上限为 32 MiB，正文阅读不执行脚本或加载外部资源。字体混淆不影响文字阅读，加密正文会显示明确错误。原书 CSS、固定版式和交互内容不在当前重排阅读模式范围内。

docs/altstore.json

@@ -36,20 +36,20 @@
       ],
       "versions": [
         {
-          "version": "0.1.10",
+          "version": "0.1.12",
-          "buildVersion": "16",
+          "buildVersion": "18",
-          "date": "2026-10-20T15:34:03Z",
+          "date": "2026-10-23T17:02:37Z",
-          "localizedDescription": "小浣熊阅读 **0.1.10**，构建号 **16**，支持 **iOS / iPadOS 15.0 及以上**。\n\n## 更新内容\n\n- 修复书源文件选择及导入流程：支持 JSON 多选，调整文件读取权限和导入预览展示时机（#1）。\n- 支持进入书源分组查看书源、打开详情及切换启用状态。\n- 修复阅读外观中的背景图片和字体选择；限制背景图片解码尺寸，降低内存占用。\n- 完善背景图片与预设主题切换，关闭背景图时保留已选图片。\n- 修复退出后重新进入阅读器的书签状态，避免重复添加书签。\n\n## 下载安装\n\n下载附件 **`legado-ios-v0.1.10-build16-unsigned.ipa`**，使用自己的签名工具安装。支持 TrollStore 的设备可自行安装；本项目暂不提供 App Store 或 TestFlight 分发。\n\n`Source code (zip / tar.gz)` 是源码，不能直接安装。附件 `SHA256SUMS` 可用于核对 IPA 完整性。\n\n## 验证与兼容\n\n本次通过 299 项核心及页面模型测试、12 项 iOS 阅读器测试、4 项文件导入与阅读操作 UI 回归，以及模拟器构建。IPA 由本次源码重新进行 iPhoneOS Release 构建。\n\niOS 15.7 真机尚未复测。若在该系统上仍遇到文件无法选择的问题，请继续在 [Issue #1](https://github.com/Aminark/legado-ios/issues/1) 反馈。\n\n**小浣熊阅读不提供、内置或分发网络书源，也不提供在线书库；书源需自行导入。**\n",
+          "localizedDescription": "小浣熊阅读 **0.1.12**，构建号 **18**，支持 **iOS / iPadOS 15.0 及以上**。\n\n## 更新内容\n\n- 增加无 DRM EPUB 2/3 导入与阅读，识别书名、作者、封面、NCX/nav 目录及章节锚点，按书脊顺序保留正文。\n- EPUB 支持阅读进度、书签、朗读和阅读样式；在阅读菜单「更多 → 本章插图」查看插图。\n- 本地文件管理支持 EPUB 重命名、原格式导出、删除，以及从系统文件应用打开。\n- 改善 JSON 规则模板、候选选择、组合替换和请求参数兼容，修复 JavaScript 数字类型及动态净化替换。\n- 修复 RSS 翻页失败重试、切换订阅时的过期结果和重复翻页请求。\n- 改善连续滚动中的章节插入定位，调整 iOS 26 代码编辑与图片查看工具栏。\n\n## 下载安装\n\n下载附件 **`legado-ios-v0.1.12-build18-unsigned.ipa`**，使用自己的签名工具安装。支持 TrollStore 的设备可自行尝试安装；目前不提供 App Store 或 TestFlight 分发。\n\n`Source code (zip / tar.gz)` 是源码，不能直接安装。附件 `SHA256SUMS` 用于核对 IPA 完整性。\n\n## 兼容与验证\n\nEPUB 正文使用现有阅读器重新排版，暂不还原原书 CSS、固定版式及交互内容，不支持加密正文。\n\n安装包由对应源码重新进行 iPhoneOS Release 构建，并核对包内版本及完整性。本轮文件导入界面回归仍有待确认项，尚未完成 UI 回归和真机安装验证；如遇问题，请在 [Issues](https://github.com/Aminark/legado-ios/issues) 附上设备、系统版本及复现步骤。\n\n**小浣熊阅读不提供、内置或分发网络书源，也不提供在线书库；书源需自行导入。**\n",
-          "downloadURL": "https://github.com/Aminark/legado-ios/releases/download/v0.1.10/legado-ios-v0.1.10-build16-unsigned.ipa",
+          "downloadURL": "https://github.com/Aminark/legado-ios/releases/download/v0.1.12/legado-ios-v0.1.12-build18-unsigned.ipa",
-          "size": 12080513,
+          "size": 12638772,
           "minOSVersion": "15.0"
         }
       ],
-      "version": "0.1.10",
+      "version": "0.1.12",
-      "versionDate": "2026-10-20T15:34:03Z",
+      "versionDate": "2026-10-23T17:02:37Z",
-      "versionDescription": "小浣熊阅读 **0.1.10**，构建号 **16**，支持 **iOS / iPadOS 15.0 及以上**。\n\n## 更新内容\n\n- 修复书源文件选择及导入流程：支持 JSON 多选，调整文件读取权限和导入预览展示时机（#1）。\n- 支持进入书源分组查看书源、打开详情及切换启用状态。\n- 修复阅读外观中的背景图片和字体选择；限制背景图片解码尺寸，降低内存占用。\n- 完善背景图片与预设主题切换，关闭背景图时保留已选图片。\n- 修复退出后重新进入阅读器的书签状态，避免重复添加书签。\n\n## 下载安装\n\n下载附件 **`legado-ios-v0.1.10-build16-unsigned.ipa`**，使用自己的签名工具安装。支持 TrollStore 的设备可自行安装；本项目暂不提供 App Store 或 TestFlight 分发。\n\n`Source code (zip / tar.gz)` 是源码，不能直接安装。附件 `SHA256SUMS` 可用于核对 IPA 完整性。\n\n## 验证与兼容\n\n本次通过 299 项核心及页面模型测试、12 项 iOS 阅读器测试、4 项文件导入与阅读操作 UI 回归，以及模拟器构建。IPA 由本次源码重新进行 iPhoneOS Release 构建。\n\niOS 15.7 真机尚未复测。若在该系统上仍遇到文件无法选择的问题，请继续在 [Issue #1](https://github.com/Aminark/legado-ios/issues/1) 反馈。\n\n**小浣熊阅读不提供、内置或分发网络书源，也不提供在线书库；书源需自行导入。**\n",
+      "versionDescription": "小浣熊阅读 **0.1.12**，构建号 **18**，支持 **iOS / iPadOS 15.0 及以上**。\n\n## 更新内容\n\n- 增加无 DRM EPUB 2/3 导入与阅读，识别书名、作者、封面、NCX/nav 目录及章节锚点，按书脊顺序保留正文。\n- EPUB 支持阅读进度、书签、朗读和阅读样式；在阅读菜单「更多 → 本章插图」查看插图。\n- 本地文件管理支持 EPUB 重命名、原格式导出、删除，以及从系统文件应用打开。\n- 改善 JSON 规则模板、候选选择、组合替换和请求参数兼容，修复 JavaScript 数字类型及动态净化替换。\n- 修复 RSS 翻页失败重试、切换订阅时的过期结果和重复翻页请求。\n- 改善连续滚动中的章节插入定位，调整 iOS 26 代码编辑与图片查看工具栏。\n\n## 下载安装\n\n下载附件 **`legado-ios-v0.1.12-build18-unsigned.ipa`**，使用自己的签名工具安装。支持 TrollStore 的设备可自行尝试安装；目前不提供 App Store 或 TestFlight 分发。\n\n`Source code (zip / tar.gz)` 是源码，不能直接安装。附件 `SHA256SUMS` 用于核对 IPA 完整性。\n\n## 兼容与验证\n\nEPUB 正文使用现有阅读器重新排版，暂不还原原书 CSS、固定版式及交互内容，不支持加密正文。\n\n安装包由对应源码重新进行 iPhoneOS Release 构建，并核对包内版本及完整性。本轮文件导入界面回归仍有待确认项，尚未完成 UI 回归和真机安装验证；如遇问题，请在 [Issues](https://github.com/Aminark/legado-ios/issues) 附上设备、系统版本及复现步骤。\n\n**小浣熊阅读不提供、内置或分发网络书源，也不提供在线书库；书源需自行导入。**\n",
-      "downloadURL": "https://github.com/Aminark/legado-ios/releases/download/v0.1.10/legado-ios-v0.1.10-build16-unsigned.ipa",
+      "downloadURL": "https://github.com/Aminark/legado-ios/releases/download/v0.1.12/legado-ios-v0.1.12-build18-unsigned.ipa",
-      "size": 12080513,
+      "size": 12638772,
       "appPermissions": {
         "entitlements": [],
         "privacy": {

scripts/fixtures/SourceImportFixture.swift

@@ -23,6 +23,10 @@ final class FixtureAppDelegate: UIResponder, UIApplicationDelegate {
             try? Data("第一章 开始\n这是用于验证本地 TXT 导入的小说正文。\n第二章 后续\n正文继续。".utf8)
                 .write(to: documents.appendingPathComponent(name), options: .atomic)
         }
+        if let epub = Bundle.main.url(forResource: "EPUB-Book", withExtension: "epub"),
+           let data = try? Data(contentsOf: epub) {
+            try? data.write(to: documents.appendingPathComponent("EPUB-Book.epub"), options: .atomic)
+        }
         let controller = UIViewController()
         controller.view.backgroundColor = .systemBackground
         let label = UILabel()

scripts/fixtures/make-epub.py

@@ -0,0 +1,23 @@
+#!/usr/bin/env python3
+"""Generate an original EPUB 3 fixture for Files import and reader UI tests."""
+import base64
+import sys
+import zipfile
+from pathlib import Path
+
+output = Path(sys.argv[1])
+output.parent.mkdir(parents=True, exist_ok=True)
+with zipfile.ZipFile(output, "w") as archive:
+    archive.writestr("mimetype", "application/epub+zip", compress_type=zipfile.ZIP_STORED)
+    files = {
+        "META-INF/container.xml": '<container xmlns="urn:oasis:names:tc:opendocument:xmlns:container"><rootfiles><rootfile full-path="OPS/book.opf"/></rootfiles></container>',
+        "OPS/book.opf": '''<package xmlns="http://www.idpf.org/2007/opf" version="3.0" unique-identifier="id">
+          <metadata xmlns:dc="http://purl.org/dc/elements/1.1/"><dc:identifier id="id">legado-ui-epub</dc:identifier><dc:title>EPUB 导入测试</dc:title><dc:creator>测试作者</dc:creator><dc:language>zh</dc:language><meta property="dcterms:modified">2026-10-23T00:00:00Z</meta></metadata>
+          <manifest><item id="nav" href="nav.xhtml" media-type="application/xhtml+xml" properties="nav"/><item id="body" href="chapter.xhtml" media-type="application/xhtml+xml"/><item id="image" href="cover.png" media-type="image/png" properties="cover-image"/></manifest>
+          <spine><itemref idref="body"/></spine></package>''',
+        "OPS/nav.xhtml": '''<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops"><head><title>目录</title></head><body><nav epub:type="toc"><ol><li><a href="chapter.xhtml#one">第一章 EPUB</a></li><li><a href="chapter.xhtml#two">第二章 EPUB</a></li></ol></nav></body></html>''',
+        "OPS/chapter.xhtml": '''<html xmlns="http://www.w3.org/1999/xhtml"><head><title>EPUB 正文</title></head><body><section><h1 id="one">第一章 EPUB</h1><p>这是 EPUB 导入后的第一章正文。</p><img src="cover.png" alt="测试插图"/></section><section><h1 id="two">第二章 EPUB</h1><p>这是 EPUB 的第二章正文，目录锚点跳转成功。</p></section></body></html>''',
+    }
+    for name, contents in files.items():
+        archive.writestr(name, contents.encode("utf-8"), compress_type=zipfile.ZIP_DEFLATED)
+    archive.writestr("OPS/cover.png", base64.b64decode("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"))

scripts/fixtures/rss-flow-server.py

@@ -0,0 +1,44 @@
+"""Local, synthetic RSS responses for the LegadoRSSFlowUITests scheme."""
+
+import http.server
+import json
+import urllib.parse
+
+BASE_URL = "http://127.0.0.1:18763"
+
+
+class Fixture(http.server.BaseHTTPRequestHandler):
+    def do_GET(self):
+        parsed = urllib.parse.urlsplit(self.path)
+        query = urllib.parse.parse_qs(parsed.query)
+        key = query.get("key", [""])[0]
+        page = query.get("page", ["1"])[0]
+        if parsed.path == "/source.json":
+            payload = {
+                "sourceUrl": BASE_URL,
+                "sourceName": "RSS 全流程测试",
+                "sortUrl": "全部::" + BASE_URL + "/feed?page={{page}}",
+                "searchUrl": BASE_URL + "/feed?key={{key}}&page={{page}}",
+                "ruleArticles": "$.items",
+                "ruleTitle": "$.title",
+                "ruleLink": "$.link",
+                "ruleDescription": "$.description",
+                "ruleNextPage": "PAGE",
+            }
+        else:
+            title = f"搜索结果 {key} 第{page}页" if key else "首页文章"
+            payload = {"items": [{
+                "title": title,
+                "link": BASE_URL + "/article/" + key + "/" + page,
+                "description": "<p>这是用于全流程测试的自造订阅正文。</p>",
+            }]}
+        body = json.dumps(payload, ensure_ascii=False).encode()
+        self.send_response(200)
+        self.send_header("Content-Type", "application/json; charset=utf-8")
+        self.send_header("Content-Length", str(len(body)))
+        self.end_headers()
+        self.wfile.write(body)
+
+
+if __name__ == "__main__":
+    http.server.ThreadingHTTPServer(("127.0.0.1", 18763), Fixture).serve_forever()

scripts/prepare-source-import-fixture.sh

@@ -40,6 +40,7 @@ xcrun --sdk iphonesimulator swiftc -parse-as-library -swift-version 6 \
     -module-cache-path "$legado_fixture_repo/DerivedData/ModuleCache.noindex" \
     "$legado_fixture_repo/scripts/fixtures/SourceImportFixture.swift" \
     -o "$legado_fixture_app/SourceImportFixtures"
+python3 "$legado_fixture_repo/scripts/fixtures/make-epub.py" "$legado_fixture_app/EPUB-Book.epub"
 xcrun simctl install "$legado_fixture_device" "$legado_fixture_app"
 xcrun simctl launch --terminate-running-process "$legado_fixture_device" com.epiboly.legado.source-import-fixtures
-printf 'JSON and TXT import fixtures are ready in Files > On My iPhone > 书源导入测试.\n'
+printf 'JSON, TXT and EPUB import fixtures are ready in Files > On My iPhone > 书源导入测试.\n'