feat(local-files): 保存并阅读已有 HTML 与 Markdown 结构

Packages/PeerCore/Sources/PeerCore/InfoBase.swift

@@ -36,11 +36,13 @@ public enum InfoBase {
         throw PeerError.connection("信息库未返回资料标识。")
     }
 
-    public static func updateBlock(_ id: Int, content: String, on connection: PostgresConnection)
+    public static func updateBlock(
+        _ id: Int, content: String, resolver: String? = nil, on connection: PostgresConnection
+    )
         async throws
     {
         let rows = try await connection.query(
-            "UPDATE inkcre.blocks SET content=\(content) WHERE id=\(id) RETURNING id",
+            "UPDATE inkcre.blocks SET content=\(content),resolver=COALESCE(\(resolver),resolver) WHERE id=\(id) RETURNING id",
             logger: PeerDatabase.logger)
         for try await _ in rows.decode(Int.self) { return }
         throw PeerError.needsAction("资料已被移除。")

THIRD_PARTY_NOTICES.md

@@ -10,6 +10,10 @@ PostgresNIO 及传递依赖使用各自许可。Package.resolved 固定来源与
 
 InKCre UI 的 `InKCreLogoAssets` 用于本项目品牌；代码 MIT 许可不授予 InKCre 商标或品牌资源的使用权。打包从锁定的 UI checkout 收集 `assets/logo/README.zh-CN.md` 与 `PROVENANCE.json`，保留资源来源和使用边界；任一文件缺失仍使打包失败。其他依赖继续要求完整许可。应用图标的历史来源另保留于本机品牌目录。
 
-[SwiftSoup 2.14.3](https://github.com/scinfu/SwiftSoup/tree/2.14.3)（e7f85f4d3dd91ca9a1e94d3328aa1ed5120a8c0f）用于本地 HTML 文字解析，使用 MIT。构建收集上游完整 LICENSE，保留原 jsoup 作者与 Swift port 作者署名。
+[SwiftSoup 2.14.3](https://github.com/scinfu/SwiftSoup/tree/2.14.3)（e7f85f4d3dd91ca9a1e94d3328aa1ed5120a8c0f）用于本地 HTML 结构解析与安全清理，使用 MIT。构建收集上游完整 LICENSE，保留原 jsoup 作者与 Swift port 作者署名。
 
 [Hister](https://github.com/asciimoo/hister) 的目录监视与客户端提取行为为本地文件功能提供启发。参考版本 f1fff29683c8c8006a35d3387384d9b1e123cebe，项目使用 AGPL-3.0-or-later。本实现未复制或改写其源码，不运行其服务；本致谢不表示上游参与或背书。
+
+[swift-readability](https://github.com/neolee/swift-readability/tree/53be6c0d9c3723a7f9acb17324d435740347a872) 使用 MIT，固定修订用于已取得 HTML 的正文提取。[swift-cmark 0.9.0](https://github.com/swiftlang/swift-cmark/tree/0.9.0) 用于临时 GFM 阅读转换。打包从锁定源码收集完整 LICENSE／COPYING，包含各自上游作者声明。
+
+本轮页面内容与结构提取同时参考 Asciimoo 及 Hister 贡献者的工作，固定修订为 `2e017c05cf544b812b8fd509ff3fc07c5c7e464e`。实现保留已取得结构及已有来源事实，不复制、改写或运行 Hister 源码，不采用其全部内容删减策略。

Tests/LocalFilesTests/LocalFilesInterfaceTests.swift

@@ -5,6 +5,7 @@ import PeerCore
 import Security
 import SwiftUI
 import Testing
+import WebKit
 
 @testable import LocalFilesMac
 
@@ -58,8 +59,19 @@ struct LocalFilesInterfaceTests {
         let file = try #require(model.records.first { $0.name == "note.md" })
         model.selectedFile = model.records.first { $0.name == "note.pdf" }?.id
         model.selectedFile = file.id
-        #expect(model.selectedText == nil && model.loadingText)
-        try await until { model.selectedText == "# 标题\n\n第一段。\n\n- 列表\n" && !model.loadingText }
+        #expect(model.selectedContent == nil && model.loadingContent)
+        try await until {
+            model.selectedContent == "# 标题\n\n第一段。\n\n- 列表\n" && !model.loadingContent
+        }
+        try await until {
+            guard let view = window.contentView, let reader = self.webReader(in: view),
+                !reader.isLoading
+            else { return false }
+            return
+                (try? await reader.evaluateJavaScript("document.querySelector('h1')?.textContent"))
+                as? String == "标题"
+        }
+        #expect(!window.isVisible)
         let database = PeerDatabase.shared
         let pool = try await database.client(
             for: target, password: try #require(env["INKCRE_FILES_TEST_PASSWORD"]))
@@ -73,35 +85,67 @@ struct LocalFilesInterfaceTests {
         )
         .decode(String.self).first { @Sendable _ in true }
         #expect(catalogVersion == afterRefresh)
-        #expect(model.selectedText != nil && !model.loadingText)
+        #expect(model.selectedContent != nil && !model.loadingContent)
         try await snapshot(window, name: "files-body")
+        model.selectedFile = model.records.first { $0.name == "note.html" }?.id
+        try await until {
+            guard model.file?.bodyResolver == "core.html.v1", !model.loadingContent,
+                let view = window.contentView, let reader = self.webReader(in: view),
+                !reader.isLoading
+            else { return false }
+            return
+                (try? await reader.evaluateJavaScript("document.querySelector('td')?.textContent"))
+                as? String == "A"
+        }
+        let reader = try #require(window.contentView.flatMap(webReader))
+        let code = try await reader.evaluateJavaScript("document.querySelector('pre')?.textContent")
+        #expect((code as? String)?.contains("<node>& value\n  indented") == true)
+        #expect(
+            try await reader.evaluateJavaScript("typeof window.not_body") as? String == "undefined")
+        #expect(
+            try await reader.evaluateJavaScript(
+                "document.querySelectorAll('video,track,[src]').length") as? Int == 0)
+        #expect(
+            try await reader.evaluateJavaScript(
+                "!!document.querySelector('a[href=\"https://example.org/captions.vtt\"]')") as? Bool
+                == true)
+        try await snapshot(window, name: "files-html")
+        if let path = env["INKCRE_LAYOUT_EVIDENCE_DIR"] {
+            let image = try await reader.takeSnapshot(configuration: nil)
+            let data = try #require(image.tiffRepresentation)
+            try #require(NSBitmapImageRep(data: data)?.representation(using: .png, properties: [:]))
+                .write(
+                    to: URL(fileURLWithPath: path).appendingPathComponent("files-html-content.png"))
+        }
+        model.selectedFile = file.id
+        try await until { model.selectedContent == "# 标题\n\n第一段。\n\n- 列表\n" }
         let url = folder.appendingPathComponent("note.md")
         try Data("# 自动监视后的正文".utf8).write(to: url, options: .atomic)
-        try await until { model.selectedText == "# 自动监视后的正文" }
+        try await until { model.selectedContent == "# 自动监视后的正文" }
         let source = try #require(model.source)
         model.toggle(source)
         try await until { model.source?.config.enabled == false && !model.working }
         try Data("# 暂停时修改".utf8).write(to: url, options: .atomic)
         try await Task.sleep(for: .seconds(2))
-        #expect(model.selectedText == "# 自动监视后的正文")
+        #expect(model.selectedContent == "# 自动监视后的正文")
         try await snapshot(window, name: "files-paused")
         model.toggle(try #require(model.source))
-        try await until { model.selectedText == "# 暂停时修改" && !model.working }
+        try await until { model.selectedContent == "# 暂停时修改" && !model.working }
         await model.stop()
         try Data("# 停机时修改".utf8).write(to: url, options: .atomic)
         try model.start()
-        try await until { model.selectedText == "# 停机时修改" }
+        try await until { model.selectedContent == "# 停机时修改" }
         try FileManager.default.removeItem(at: url)
         try await until { model.file?.metadata.available == false }
-        #expect(model.selectedText == "# 停机时修改")
+        #expect(model.selectedContent == "# 停机时修改")
         try await snapshot(window, name: "files-original-missing")
         try Data("not a pdf".utf8).write(
             to: folder.appendingPathComponent("note.pdf"), options: .atomic)
         try await until {
             model.records.contains { $0.name == "note.pdf" && $0.metadata.issue != nil }
         }
         model.selectedFile = model.records.first { $0.name == "note.pdf" }?.id
-        try await until { model.selectedText != nil && !model.loadingText }
+        try await until { model.selectedContent != nil && !model.loadingContent }
         try await snapshot(window, name: "files-failure")
         let restoredSource = try #require(model.source)
         let client = LocalFileClient(
@@ -113,15 +157,21 @@ struct LocalFilesInterfaceTests {
         await model.stop()
     }
 
-    private func until(_ condition: @escaping @MainActor () -> Bool) async throws {
+    private func until(_ condition: @escaping @MainActor () async -> Bool) async throws {
         for _ in 0..<150 {
-            if condition() { return }
+            if await condition() { return }
             try await Task.sleep(for: .milliseconds(100))
         }
-        #expect(condition(), "等待本地文件主流程状态超时")
+        #expect(await condition(), "等待本地文件主流程状态超时")
         throw PeerError.needsAction("测试等待状态超时。")
     }
 
+    private func webReader(in view: NSView) -> WKWebView? {
+        if let reader = view as? WKWebView { return reader }
+        for child in view.subviews { if let reader = webReader(in: child) { return reader } }
+        return nil
+    }
+
     private func snapshot(_ window: NSWindow, name: String) async throws {
         for appearance in [NSAppearance.Name.aqua, .darkAqua] {
             window.appearance = NSAppearance(named: appearance)
@@ -139,3 +189,44 @@ struct LocalFilesInterfaceTests {
         }
     }
 }
+
+@Test func safeFileReadingKeepsStructureAndBlocksSourceCode() throws {
+    let source = """
+        # 标题
+
+        | Item | Value |
+        | --- | --- |
+        | A | 12 |
+
+        - [x] 完成
+        - [ ] 待办
+
+        ```html
+        <node>& value
+          indented
+        ```
+
+        ~~删除~~ https://example.org
+
+        <a href="#part">文内目录</a><h2 id="part">章节</h2>
+
+        <script>window.executed = true</script><iframe src="https://invalid.test/frame"></iframe>
+        <img src="https://example.org/image.png" alt="图片说明"><a href="javascript:alert(1)">bad</a>
+        <video src="https://example.org/video.mp4"><track src="https://example.org/captions.vtt" label="English"><p>已有文字稿 <a href="https://example.org/transcript">完整文字稿</a></p></video>
+        """
+    let html = try readingDocument(source, resolver: "core.markdown.v1")
+    #expect(html.contains("<table>") && html.contains("<del>删除</del>"))
+    #expect(html.contains("&lt;node&gt;&amp; value\n  indented"))
+    #expect(html.contains("checked") && html.contains("disabled"))
+    #expect(html.contains("https://example.org/image.png") && html.contains("图片说明"))
+    #expect(
+        !html.contains("window.executed") && !html.contains("<iframe")
+            && !html.contains("javascript:"))
+    #expect(
+        html.contains("https://example.org/video.mp4")
+            && html.contains("https://example.org/captions.vtt"))
+    #expect(!html.contains("<video") && !html.contains("<track"))
+    #expect(html.contains("已有文字稿") && html.contains("https://example.org/transcript"))
+    #expect(html.contains("default-src 'none'"))
+    #expect(html.contains("href=\"#part\"") && html.contains("id=\"part\""))
+}