feat(sandbox): 不可信字节在沙箱子进程里解析(撤销权限、超时与内存强杀);种子、文件预览、字幕解析迁入

Co-Authored-By: Claude Opus 5.5 <[email protected]>

apps/server/src/sandbox/jobs.ts

@@ -0,0 +1,115 @@
+// 沙箱任务表(docs/adr/ADR-005-sandbox.md):只放「外部不可信字节 → 结构化数据」的纯函数。
+// 子进程加载完本表后撤销全部权限(文件 / 网络 / 环境变量 / 子进程),所以任务里不能有任何 I/O;
+// 输入输出必须能 JSON 化(二进制用 base64)。新增任务:在这里加一项,调用方用 runSandbox(名字, 输入)。
+import { parseTorrent } from "@umt/adapters";
+import { decodeSubtitle, detectSubtitleLanguage, subtitleEpisode, subtitleExtOf } from "@umt/domain";
+import { unzipSync } from "fflate";
+import { buildPreview } from "../modules/file-manager/preview-build.ts";
+
+const MAX_SUBTITLE_BYTES = 10 * 1024 * 1024;
+const MAX_ARCHIVE_ENTRIES = 300;
+
+export const b64encode = (bytes: Uint8Array) => {
+  let s = "";
+  for (let i = 0; i < bytes.length; i += 0x8000) s += String.fromCharCode(...bytes.subarray(i, i + 0x8000));
+  return btoa(s);
+};
+export const b64decode = (s: string) => Uint8Array.from(atob(s), (c) => c.charCodeAt(0));
+
+export interface ParsedSubtitleFile {
+  name: string;
+  archive: string | null;
+  ext: string;
+  language: string;
+  season: number | null;
+  seasonInferred: boolean;
+  episode: number | null;
+  encoding: string;
+  sizeBytes: number;
+  /** 转成 UTF-8 后的内容(base64) */
+  content: string;
+}
+
+/** 字幕 / zip:解压、解码转 UTF-8、识别语言与季集 */
+function subtitleParse(input: { files: { name: string; base64: string }[] }) {
+  const items: ParsedSubtitleFile[] = [];
+  const skipped: { name: string; reason: string }[] = [];
+  const one = (name: string, bytes: Uint8Array, archive: string | null) => {
+    const ext = subtitleExtOf(name);
+    if (!ext) return skipped.push({ name, reason: "不是字幕文件" });
+    if (bytes.length > MAX_SUBTITLE_BYTES) return skipped.push({ name, reason: "文件超过 10 MB" });
+    const { text, encoding } = decodeSubtitle(bytes);
+    const ep = subtitleEpisode(name);
+    items.push({
+      name,
+      archive,
+      ext,
+      language: detectSubtitleLanguage(name, text),
+      season: ep.season,
+      seasonInferred: ep.seasonInferred,
+      episode: ep.episode,
+      encoding,
+      sizeBytes: bytes.length,
+      content: b64encode(new TextEncoder().encode(text)),
+    });
+  };
+  for (const f of input.files) {
+    let bytes: Uint8Array;
+    try {
+      bytes = b64decode(f.base64);
+    } catch {
+      skipped.push({ name: f.name, reason: "内容不是有效的 base64" });
+      continue;
+    }
+    if (/\.zip$/i.test(f.name)) {
+      let entries: Record<string, Uint8Array>;
+      try {
+        let n = 0;
+        entries = unzipSync(bytes, {
+          filter: (e) =>
+            !e.name.endsWith("/") &&
+            !!subtitleExtOf(e.name) &&
+            e.originalSize <= MAX_SUBTITLE_BYTES &&
+            ++n <= MAX_ARCHIVE_ENTRIES,
+        });
+      } catch {
+        skipped.push({ name: f.name, reason: "压缩包损坏或加密" });
+        continue;
+      }
+      const names = Object.keys(entries);
+      if (!names.length) skipped.push({ name: f.name, reason: "压缩包里没有字幕文件" });
+      for (const n of names) one(n, entries[n]!, f.name);
+    } else if (/\.(rar|7z)$/i.test(f.name)) {
+      skipped.push({ name: f.name, reason: "暂不支持 rar / 7z，请解压后上传或改用 zip" });
+    } else one(f.name, bytes, null);
+  }
+  return { items, skipped };
+}
+
+/** 种子文件:info hash、名称、文件清单(TorrentMeta,可 JSON 化) */
+function torrentParse(input: { base64: string }) {
+  return parseTorrent(b64decode(input.base64));
+}
+
+/** 测试用:按指令卡死 / 吃内存 / 抛错,验证父进程的超时强杀与内存上限 */
+function selfTest(input: { mode: "ok" | "hang" | "memory" | "throw" | "io" }) {
+  if (input.mode === "hang") for (;;) {}
+  if (input.mode === "memory") {
+    const keep: Uint8Array[] = [];
+    for (;;) keep.push(new Uint8Array(16 * 1024 * 1024).fill(1));
+  }
+  if (input.mode === "throw") throw new Error("任务自己抛的错");
+  if (input.mode === "io") return { read: Deno.readTextFileSync("/etc/hostname") };
+  return { ok: true };
+}
+
+export const SANDBOX_JOBS = {
+  "subtitle.parse": subtitleParse,
+  "torrent.parse": torrentParse,
+  "file.preview": buildPreview,
+  "self.test": selfTest,
+} as const;
+
+export type SandboxJobName = keyof typeof SANDBOX_JOBS;
+export type SandboxInput<J extends SandboxJobName> = Parameters<(typeof SANDBOX_JOBS)[J]>[0];
+export type SandboxOutput<J extends SandboxJobName> = Awaited<ReturnType<(typeof SANDBOX_JOBS)[J]>>;