feat(subtitles): 字幕站(SubHD、字幕库)在线搜索与下载;rar / 7z 解压;验证码与页面解析在沙箱里

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

apps/server/src/sandbox/archives.ts

@@ -0,0 +1,70 @@
+// 压缩包解压(沙箱内):zip 用 fflate,rar 用 node-unrar-js(WASM),7z 用 7z-wasm。
+// WASM 库要读自己包里的 .wasm 文件:prepareArchives() 在撤销权限前预热(载入内存),之后只在内存里解压。
+import SevenZip from "7z-wasm";
+import { unzipSync } from "fflate";
+import { createExtractorFromData } from "node-unrar-js";
+
+type Seven = Awaited<ReturnType<typeof SevenZip>>;
+let seven: Seven | null = null;
+let warmed = false;
+
+export async function prepareArchives(): Promise<void> {
+  if (warmed) return;
+  warmed = true;
+  try {
+    await createExtractorFromData({ data: new Uint8Array([0x52, 0x61, 0x72, 0x21]).buffer });
+  } catch {
+    // 只为载入 WASM,内容无效报错是预期的
+  }
+  seven = await SevenZip({ print: () => {}, printErr: () => {} });
+}
+
+export type AcceptEntry = (name: string, size: number) => boolean;
+
+export function archiveKindOf(name: string): "zip" | "rar" | "7z" | null {
+  const m = /\.(zip|rar|7z)$/i.exec(name);
+  return m ? (m[1]!.toLowerCase() as "zip" | "rar" | "7z") : null;
+}
+
+/** 解出通过 accept 的条目;压缩包损坏 / 加密抛错 */
+export async function extractArchive(
+  kind: "zip" | "rar" | "7z",
+  bytes: Uint8Array,
+  accept: AcceptEntry,
+): Promise<Record<string, Uint8Array>> {
+  if (kind === "zip")
+    return unzipSync(bytes, { filter: (e) => !e.name.endsWith("/") && accept(e.name, e.originalSize) });
+  if (kind === "rar") {
+    const ex = await createExtractorFromData({ data: bytes.slice().buffer });
+    const out: Record<string, Uint8Array> = {};
+    for (const f of ex.extract({ files: (h) => !h.flags.directory && accept(h.name, h.unpSize) }).files)
+      if (f.extraction) out[f.fileHeader.name] = f.extraction;
+    return out;
+  }
+  await prepareArchives();
+  const z = seven!;
+  const dir = `/x${crypto.randomUUID().slice(0, 8)}`;
+  z.FS.writeFile(`${dir}.7z`, bytes);
+  // callMain 出错时抛 ExitStatus 或静默不写文件:两种都当成损坏处理
+  z.callMain(["x", `${dir}.7z`, `-o${dir}`, "-y", "-p-"]);
+  try {
+    z.FS.stat(dir);
+  } catch {
+    throw new Error("7z 解压失败");
+  }
+  const out: Record<string, Uint8Array> = {};
+  const walk = (d: string, rel: string) => {
+    for (const n of z.FS.readdir(d) as string[]) {
+      if (n === "." || n === "..") continue;
+      const p = `${d}/${n}`;
+      const r = rel ? `${rel}/${n}` : n;
+      if (z.FS.isDir(z.FS.stat(p).mode)) walk(p, r);
+      else {
+        const data = z.FS.readFile(p) as Uint8Array;
+        if (accept(r, data.length)) out[r] = data;
+      }
+    }
+  };
+  walk(dir, "");
+  return out;
+}