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/jobs.ts

@@ -3,8 +3,10 @@
 // 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";
+import { PAGE_PARSERS, type PageKind } from "../modules/subtitles/online/parse.ts";
+import { archiveKindOf, extractArchive, prepareArchives } from "./archives.ts";
+import { solveSubhdCaptcha, solveZimukuCaptcha } from "./captcha.ts";
 
 const MAX_SUBTITLE_BYTES = 10 * 1024 * 1024;
 const MAX_ARCHIVE_ENTRIES = 300;
@@ -30,8 +32,13 @@ export interface ParsedSubtitleFile {
   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 }[] }) {
+/** 子进程撤销权限前调用:预热要读文件的 WASM 库 */
+export async function prepareJobs(): Promise<void> {
+  await prepareArchives();
+}
+
+/** 字幕 / zip / rar / 7z:解压、解码转 UTF-8、识别语言与季集 */
+async 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) => {
@@ -61,17 +68,16 @@ function subtitleParse(input: { files: { name: string; base64: string }[] }) {
       skipped.push({ name: f.name, reason: "内容不是有效的 base64" });
       continue;
     }
-    if (/\.zip$/i.test(f.name)) {
+    const kind = archiveKindOf(f.name);
+    if (kind) {
       let entries: Record<string, Uint8Array>;
       try {
         let n = 0;
-        entries = unzipSync(bytes, {
-          filter: (e) =>
-            !e.name.endsWith("/") &&
-            !!subtitleExtOf(e.name) &&
-            e.originalSize <= MAX_SUBTITLE_BYTES &&
-            ++n <= MAX_ARCHIVE_ENTRIES,
-        });
+        entries = await extractArchive(
+          kind,
+          bytes,
+          (name, size) => !!subtitleExtOf(name) && size <= MAX_SUBTITLE_BYTES && ++n <= MAX_ARCHIVE_ENTRIES,
+        );
       } catch {
         skipped.push({ name: f.name, reason: "压缩包损坏或加密" });
         continue;
@@ -79,8 +85,6 @@ function subtitleParse(input: { files: { name: string; base64: string }[] }) {
       const names = Object.keys(entries);
       if (!names.length) skipped.push({ name: f.name, reason: "压缩包里没有字幕文件" });
       for (const n of names) one(n, entries[n]!, f.name);
-    } else if (/\.(rar|7z)$/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 };
@@ -103,10 +107,26 @@ function selfTest(input: { mode: "ok" | "hang" | "memory" | "throw" | "io" }) {
   return { ok: true };
 }
 
+/** SubHD SVG / 字幕库 BMP 验证码 */
+function captchaSolve(input: { kind: "subhd"; svg: string } | { kind: "zimuku"; bmpBase64: string }) {
+  return {
+    code: input.kind === "subhd" ? solveSubhdCaptcha(input.svg) : solveZimukuCaptcha(b64decode(input.bmpBase64)),
+  };
+}
+
+/** 字幕站页面 → 结构化数据(网页同样不可信) */
+function subtitlePage(input: { kind: PageKind; html: string }) {
+  const fn = PAGE_PARSERS[input.kind] as (html: string) => unknown;
+  if (!fn) throw new Error(`没有这种页面解析：${input.kind}`);
+  return fn(input.html);
+}
+
 export const SANDBOX_JOBS = {
   "subtitle.parse": subtitleParse,
   "torrent.parse": torrentParse,
   "file.preview": buildPreview,
+  "captcha.solve": captchaSolve,
+  "subtitle.page": subtitlePage,
   "self.test": selfTest,
 } as const;