feat(sites): extract douban IDs from list rows (audiences) (ID-3)

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

packages/adapters/src/sites/definitions/audiences.json

@@ -152,6 +152,16 @@
           }
         ]
       },
+      "doubanid": {
+        "selector": "a[href*=\"douban.com/subject/\"]",
+        "attribute": "href",
+        "filters": [
+          {
+            "name": "re_search",
+            "args": ["subject/(\\d+)", 1]
+          }
+        ]
+      },
       "freedate": {
         "selector": "img[class^=\"pro_\"]",
         "attribute": "title",

packages/adapters/src/sites/spider/map.ts

@@ -78,5 +78,6 @@ export function spiderTorrentToRelease(
   };
   if (t.description) release.subtitle = t.description;
   if (t.imdbid) release.imdbId = t.imdbid;
+  if (t.doubanid) release.doubanId = t.doubanid;
   return release;
 }

packages/adapters/src/sites/spider/spider.ts

@@ -67,6 +67,8 @@ export interface SpiderTorrent {
   seeders?: number;
   size?: number;
   imdbid?: string;
+  /** UMT 扩展(ID-3,MP 没有):列表行里的豆瓣 subject id,字段 `doubanid` 与 imdbid 同样的选择器写法 */
+  doubanid?: string;
   downloadvolumefactor?: number;
   uploadvolumefactor?: number;
   pubdate?: string;
@@ -476,6 +478,7 @@ export class SiteSpider {
       this.getSeeders(row);
       this.getSize(row);
       this.getImdbid(row);
+      this.getDoubanid(row);
       this.getVolumeFactor(row, "downloadvolumefactor");
       this.getVolumeFactor(row, "uploadvolumefactor");
       this.getPubdate(row);
@@ -659,6 +662,13 @@ export class SiteSpider {
     }
   }
 
+  private getDoubanid(row: Sel) {
+    if (!("doubanid" in this.fields)) return;
+    const selector = this.fields.doubanid ?? {};
+    const v = this.filterText(this.safeQuery(row, selector), selector.filters);
+    if (v != null && /^\d{4,10}$/.test(String(v))) this.info.doubanid = String(v);
+  }
+
   private getImdbid(row: Sel) {
     if (!("imdbid" in this.fields)) return;
     const selector = this.fields.imdbid ?? {};

packages/adapters/test/differential/list-ids.test.ts

@@ -0,0 +1,30 @@
+/**
+ * ID-3 列表页外部 ID:观众列表行右侧的评分徽标带 IMDb / 豆瓣链接(DET-2 列表样本 rl-audiences-list,脱敏)。
+ * imdbid 是 MP 字段;doubanid 是 UMT 扩展字段(MP SiteSpider 不读,索引差分只比 MP 字段,不受影响)。
+ */
+import { describe, expect, it } from "vitest";
+import { audiencesDefinition, SiteSpider } from "../../src/sites/index.ts";
+import { spiderTorrentToRelease } from "../../src/sites/spider/map.ts";
+import { readInput } from "./diff-util.ts";
+
+describe("观众列表页:IMDb / 豆瓣 ID(真实样本)", () => {
+  const html = readInput("inputs/indexer/rl-audiences-list.html.txt");
+  const rows = new SiteSpider(audiencesDefinition, {}).parse(html).torrents;
+  const byId = (id: string) => rows.find((r) => r.page_url?.includes(`id=${id}&`));
+
+  it("有豆瓣徽标的行都抽到豆瓣 ID,有 IMDb 徽标的抽到 IMDb", () => {
+    expect(byId("725858")).toMatchObject({ imdbid: "tt25399154", doubanid: "35322132" });
+    expect(byId("726249")).toMatchObject({ imdbid: "tt0039689", doubanid: "1302338" });
+    // 剧集行只有豆瓣(IMDb 在详情页里)
+    const prophet = byId("727077")!;
+    expect(prophet.doubanid).toBe("37297001");
+    expect(prophet.imdbid).toBeUndefined();
+    expect(rows.filter((r) => r.doubanid).length).toBe(10);
+  });
+
+  it("映射到 SiteRelease.doubanId / imdbId", () => {
+    const r = spiderTorrentToRelease(byId("727077")!, audiencesDefinition, "2026-09-30T00:00:00.000Z");
+    expect(r).toMatchObject({ siteReleaseId: "727077", doubanId: "37297001" });
+    expect(r?.imdbId).toBeUndefined();
+  });
+});