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]>
2026-10-03 01:44:31 +08:00
1 parent 1a76d2d700
commit eeec91e892
@@ -16,7 +16,7 @@ import {
  InMemoryProfileLock,
  imageDimensions,
  type PageContent,
-  parseTorrent,
+  parseTorrentIsolated,
  prepareImage,
  profileKeyFor,
  type TabRef,
@@ -2237,7 +2237,7 @@ export class BrowserService {
    };
    try {
      const bytes = new Uint8Array(await readFile(this.stagedPath(artifact.bytes.ref)));
-      const meta = parseTorrent(bytes);
+      const meta = await parseTorrentIsolated(bytes);
      verification = {
        valid: true,
        infohashV1: meta.infohashV1,
@@ -0,0 +1,113 @@
+// 文件预览的纯计算(沙箱任务 file.preview):文件头字节 → 预览结构;读文件在主进程(preview.ts)。
+import { type BencodeValue, bdecode, parseTorrent } from "@umt/adapters";
+import type { FilePreview } from "@umt/contracts";
+import {
+  CUE_LIMIT,
+  decodeText,
+  IMAGE_MIME,
+  imageSize,
+  maskTracker,
+  parseAss,
+  parseSrtLike,
+  parseSup,
+} from "./preview-parse.ts";
+
+export const TORRENT_FILES = 500;
+/** 不是有效 .torrent 时抛出的错误信息(主进程据此转成 400) */
+export const BAD_TORRENT = "BAD_TORRENT";
+
+const str = (v: BencodeValue | undefined) => (v instanceof Uint8Array ? new TextDecoder().decode(v) : null);
+
+function trackersOf(root: { [k: string]: BencodeValue }): string[] {
+  const out: string[] = [];
+  const tiers = root["announce-list"];
+  if (Array.isArray(tiers))
+    for (const tier of tiers) for (const t of Array.isArray(tier) ? tier : [tier]) out.push(str(t) ?? "");
+  out.push(str(root.announce) ?? "");
+  return [...new Set(out.filter(Boolean).map(maskTracker))];
+}
+
+export interface PreviewInput {
+  kind: "image" | "torrent" | "subtitle" | "text";
+  ext: string;
+  /** 文件头字节(base64) */
+  base64: string;
+  /** 文件实际大小 */
+  size: number;
+}
+
+export function buildPreview(input: PreviewInput): FilePreview {
+  const { kind, ext, size } = input;
+  const buf = Uint8Array.from(atob(input.base64), (c) => c.charCodeAt(0));
+  if (kind === "image") {
+    const dim = imageSize(buf);
+    return {
+      type: "image",
+      size,
+      mime: IMAGE_MIME[ext] ?? "application/octet-stream",
+      width: dim?.width ?? null,
+      height: dim?.height ?? null,
+    };
+  }
+  if (kind === "torrent") {
+    let meta: ReturnType<typeof parseTorrent>;
+    let root: { [k: string]: BencodeValue };
+    try {
+      meta = parseTorrent(buf);
+      root = bdecode(buf) as { [k: string]: BencodeValue };
+    } catch {
+      throw new Error(BAD_TORRENT);
+    }
+    const info = root.info as { [k: string]: BencodeValue } | undefined;
+    const created = typeof root["creation date"] === "number" ? root["creation date"] : null;
+    return {
+      type: "torrent",
+      size,
+      name: meta.name,
+      hash: meta.hash,
+      infohashV1: meta.infohashV1,
+      infohashV2: meta.infohashV2,
+      private: meta.private,
+      totalSize: meta.totalSize,
+      fileCount: meta.files.length,
+      files: meta.files.slice(0, TORRENT_FILES).map((f) => ({ path: f.path, length: Math.max(0, f.length) })),
+      trackers: trackersOf(root),
+      comment: str(root["comment.utf-8"]) ?? str(root.comment),
+      createdBy: str(root["created by"]),
+      createdAt: created && created > 0 ? new Date(created * 1000).toISOString() : null,
+      pieceLength: typeof info?.["piece length"] === "number" ? info["piece length"] : null,
+    };
+  }
+  const truncated = size > buf.length;
+  if (kind === "subtitle") {
+    if (ext === "sup") {
+      const all = parseSup(buf);
+      return {
+        type: "subtitle",
+        size,
+        format: "sup",
+        encoding: null,
+        cues: all.slice(0, CUE_LIMIT),
+        parsed: all.length,
+        truncated,
+        title: null,
+      };
+    }
+    const { text, encoding } = decodeText(buf, truncated);
+    const ass = ext === "ass" || ext === "ssa";
+    const parsed = ass ? parseAss(text) : { cues: parseSrtLike(text), title: null };
+    const cues = [...parsed.cues].sort((a, b) => a.start - b.start);
+    return {
+      type: "subtitle",
+      size,
+      format: ext as "srt" | "vtt" | "ass" | "ssa",
+      encoding,
+      cues: cues.slice(0, CUE_LIMIT),
+      parsed: cues.length,
+      truncated,
+      title: parsed.title,
+    };
+  }
+  const { text, encoding } = decodeText(buf, truncated);
+  return { type: "text", size, format: ext, encoding, text, truncated };
+}
@@ -1,21 +1,12 @@
 import { open } from "node:fs/promises";
 import { basename } from "node:path";
-import { BencodeError, type BencodeValue, bdecode, parseTorrent } from "@umt/adapters";
 import type { FilePreview, FilesPreviewQuery } from "@umt/contracts";
 import type { Principal } from "../../http/auth.ts";
+import { runSandbox, SandboxError } from "../../sandbox/client.ts";
+import { b64encode } from "../../sandbox/jobs.ts";
 import { FilesError, type FmDeps, isAdminSession, recheck, resolveIn } from "./common.ts";
-import {
-  CUE_LIMIT,
-  decodeText,
-  extOf,
-  IMAGE_MIME,
-  imageSize,
-  maskTracker,
-  parseAss,
-  parseSrtLike,
-  parseSup,
-  previewKindOf,
-} from "./preview-parse.ts";
+import { BAD_TORRENT } from "./preview-build.ts";
+import { extOf, previewKindOf } from "./preview-parse.ts";

 // GET /files/preview(FM-2):仅管理员会话;路径守卫同下载(realpath 在根内、符号链接目标也须在根内,打开前复核)。
 // 只读文件头:文本/字幕 ≤ 1 MiB、图片头 512 KiB;.torrent 整个读但上限 4 MiB。只回 JSON,不回原始内容。
@@ -23,7 +14,6 @@ import {
 export const TEXT_PREVIEW_MAX = 1024 * 1024;
 const IMAGE_HEAD = 512 * 1024;
 const TORRENT_MAX = 4 * 1024 * 1024;
-const TORRENT_FILES = 500;

 async function readHead(abs: string, max: number): Promise<{ buf: Uint8Array; size: number }> {
  const fh = await open(abs, "r");
@@ -44,17 +34,6 @@ async function readHead(abs: string, max: number): Promise<{ buf: Uint8Array; si
  }
 }

-const str = (v: BencodeValue | undefined) => (v instanceof Uint8Array ? new TextDecoder().decode(v) : null);
-
-function trackersOf(root: { [k: string]: BencodeValue }): string[] {
-  const out: string[] = [];
-  const tiers = root["announce-list"];
-  if (Array.isArray(tiers))
-    for (const tier of tiers) for (const t of Array.isArray(tier) ? tier : [tier]) out.push(str(t) ?? "");
-  out.push(str(root.announce) ?? "");
-  return [...new Set(out.filter(Boolean).map(maskTracker))];
-}
-
 export async function previewFile(
  d: FmDeps,
  principal: Principal | undefined,
@@ -69,82 +48,21 @@ export async function previewFile(
  if (!kind) throw new FilesError("forbidden", `不支持预览 .${ext || "无扩展名"} 文件`);
  const real = await recheck(r.abs, r.rootReal);

-  if (kind === "image") {
-    const { buf, size } = await readHead(real, IMAGE_HEAD);
-    const dim = imageSize(buf);
-    return {
-      type: "image",
-      size,
-      mime: IMAGE_MIME[ext] ?? "application/octet-stream",
-      width: dim?.width ?? null,
-      height: dim?.height ?? null,
-    };
+  const head = kind === "image" ? IMAGE_HEAD : kind === "torrent" ? TORRENT_MAX : TEXT_PREVIEW_MAX;
+  const { buf, size } = await readHead(real, head);
+  if (kind === "torrent" && size > TORRENT_MAX)
+    throw new FilesError("too_large", `.torrent 文件 ${size} 字节，超过预览上限 4 MiB`);
+  // 文件内容不可信:解析放进沙箱子进程(docs/adr/ADR-005-sandbox.md)
+  try {
+    return (await runSandbox(
+      "file.preview",
+      { kind, ext, base64: b64encode(buf), size },
+      { timeoutMs: 15_000 },
+    )) as FilePreview;
+  } catch (e) {
+    if (e instanceof SandboxError && e.kind === "job" && e.message.includes(BAD_TORRENT))
+      throw new FilesError("bad_request", "不是有效的 .torrent 文件");
+    if (e instanceof SandboxError) throw new FilesError("bad_request", `预览失败：${e.message}`);
+    throw e;
  }
-
-  if (kind === "torrent") {
-    const { buf, size } = await readHead(real, TORRENT_MAX);
-    if (size > TORRENT_MAX) throw new FilesError("too_large", `.torrent 文件 ${size} 字节，超过预览上限 4 MiB`);
-    try {
-      const meta = parseTorrent(buf);
-      const root = bdecode(buf) as { [k: string]: BencodeValue };
-      const info = root.info as { [k: string]: BencodeValue } | undefined;
-      const created = typeof root["creation date"] === "number" ? root["creation date"] : null;
-      return {
-        type: "torrent",
-        size,
-        name: meta.name,
-        hash: meta.hash,
-        infohashV1: meta.infohashV1,
-        infohashV2: meta.infohashV2,
-        private: meta.private,
-        totalSize: meta.totalSize,
-        fileCount: meta.files.length,
-        files: meta.files.slice(0, TORRENT_FILES).map((f) => ({ path: f.path, length: Math.max(0, f.length) })),
-        trackers: trackersOf(root),
-        comment: str(root["comment.utf-8"]) ?? str(root.comment),
-        createdBy: str(root["created by"]),
-        createdAt: created && created > 0 ? new Date(created * 1000).toISOString() : null,
-        pieceLength: typeof info?.["piece length"] === "number" ? info["piece length"] : null,
-      };
-    } catch (e) {
-      if (e instanceof BencodeError) throw new FilesError("bad_request", "不是有效的 .torrent 文件");
-      throw e;
-    }
-  }
-
-  const { buf, size } = await readHead(real, TEXT_PREVIEW_MAX);
-  const truncated = size > buf.length;
-
-  if (kind === "subtitle") {
-    if (ext === "sup") {
-      const all = parseSup(buf);
-      return {
-        type: "subtitle",
-        size,
-        format: "sup",
-        encoding: null,
-        cues: all.slice(0, CUE_LIMIT),
-        parsed: all.length,
-        truncated,
-        title: null,
-      };
-    }
-    const { text, encoding } = decodeText(buf, truncated);
-    const ass = ext === "ass" || ext === "ssa";
-    const parsed = ass ? parseAss(text) : { cues: parseSrtLike(text), title: null };
-    const cues = [...parsed.cues].sort((a, b) => a.start - b.start);
-    return {
-      type: "subtitle",
-      size,
-      format: ext as "srt" | "vtt" | "ass" | "ssa",
-      encoding,
-      cues: cues.slice(0, CUE_LIMIT),
-      parsed: cues.length,
-      truncated,
-      title: parsed.title,
-    };
-  }
-
-  const { text, encoding } = decodeText(buf, truncated);
-  return { type: "text", size, format: ext, encoding, text, truncated };
 }
@@ -14,32 +14,22 @@ import type {
 import type { SubtitleWriteTable } from "@umt/db";
 import {
  type DefaultSubtitleLang,
-  decodeSubtitle,
  detectSubtitleLanguage,
  isSidecarOf,
  pairSubtitle,
  type SubtitleExt,
-  subtitleEpisode,
  subtitleExtOf,
  subtitleFileName,
 } from "@umt/domain";
-import { unzipSync } from "fflate";
 import type { Selectable } from "kysely";
 import { dataPaths } from "../../config.ts";
 import { ApiException } from "../../http/errors.ts";
+import { runSandbox, SandboxError } from "../../sandbox/client.ts";
 import type { ServerContext } from "../types.ts";

 // 补字幕(docs/plan/subtitles.md;对照 MP 插件「海拉鲁字幕大师」upload / matching / subtitle_writer):
 // 解析(编码转 UTF-8、语言、季集)→ 配对到作品的视频 → 写到视频同目录(主干名 + 语言后缀)→ 登记字幕资产 + 历史,可撤销。

-const MAX_FILE_BYTES = 10 * 1024 * 1024;
-const MAX_ARCHIVE_ENTRIES = 300;
-
-const b64 = (bytes: Uint8Array) => {
-  let s = "";
-  for (let i = 0; i < bytes.length; i += 0x8000) s += String.fromCharCode(...bytes.subarray(i, i + 0x8000));
-  return btoa(s);
-};
 const unb64 = (s: string) => Uint8Array.from(atob(s), (c) => c.charCodeAt(0));
 const fileName = (p: string) => p.replace(/\\/g, "/").split("/").pop() ?? p;

@@ -133,60 +123,25 @@ export async function parseSubtitles(
        episode: t.episode,
      }))
    : [];
-  const items: SubtitleParsedItem[] = [];
-  const skipped: SubtitleParseResponse["skipped"] = [];
-  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_FILE_BYTES) return skipped.push({ name, reason: "文件超过 10 MB" });
-    const { text, encoding } = decodeSubtitle(bytes);
-    const ep = subtitleEpisode(name);
-    items.push({
-      id: crypto.randomUUID(),
-      name,
-      archive,
-      ext,
-      language: detectSubtitleLanguage(name, text) as SubtitleLanguage,
-      season: ep.season,
-      seasonInferred: ep.seasonInferred,
-      episode: ep.episode,
-      encoding,
-      sizeBytes: bytes.length,
-      content: b64(new TextEncoder().encode(text)),
-      targetLocationId: targets.length ? pairSubtitle(ep, targets) : null,
-    });
-  };
-  for (const f of files) {
-    let bytes: Uint8Array;
-    try {
-      bytes = unb64(f.base64);
-    } catch {
-      skipped.push({ name: f.name, reason: "内容不是有效的 base64" });
-      continue;
-    }
-    if (/\.zip$/i.test(f.name)) {
-      let entries: Record<string, Uint8Array>;
-      try {
-        let n = 0;
-        entries = unzipSync(bytes, {
-          filter: (e) =>
-            !e.name.endsWith("/") &&
-            !!subtitleExtOf(e.name) &&
-            e.originalSize <= MAX_FILE_BYTES &&
-            ++n <= MAX_ARCHIVE_ENTRIES,
-        });
-      } catch {
-        skipped.push({ name: f.name, reason: "压缩包损坏或加密" });
-        continue;
-      }
-      const names = Object.keys(entries);
-      if (!names.length) skipped.push({ name: f.name, reason: "压缩包里没有字幕文件" });
-      for (const n of names) one(n, entries[n]!, f.name);
-    } else if (/\.(rar|7z)$/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);
+  // 上传 / 下载的字幕与压缩包不可信:解压、解码、识别都在沙箱子进程里做(docs/adr/ADR-005-sandbox.md)
+  let parsed: Awaited<ReturnType<typeof runSandbox<"subtitle.parse">>>;
+  try {
+    parsed = await runSandbox("subtitle.parse", { files }, { timeoutMs: 30_000 });
+  } catch (e) {
+    if (e instanceof SandboxError) throw new ApiException("bad_request", `字幕解析失败：${e.message}`);
+    throw e;
  }
-  return { items, skipped };
+  const items: SubtitleParsedItem[] = parsed.items.map((it) => {
+    const ep = { season: it.season, seasonInferred: it.seasonInferred, episode: it.episode };
+    return {
+      ...it,
+      id: crypto.randomUUID(),
+      ext: it.ext as SubtitleExt,
+      language: it.language as SubtitleLanguage,
+      targetLocationId: targets.length ? pairSubtitle(ep, targets) : null,
+    };
+  });
+  return { items, skipped: parsed.skipped };
 }

 // ── 写入 ──