feat(scrape): 任意目录就地刮削,按路径识别作品,每次刮削一条记录(SCR-2)

- POST /files/scrape 改为对目录 / 视频文件刮削:已登记用作品,否则按路径识别(本地 + TMDB),
  认不准返回候选,选定后再提交;未登记直接用 TMDB 字段,addToLibrary 顺手建作品
- 作品详情重新刮削、入库后自动刮削改为对作品目录刮削,同一实现
- 迁移 0058_scrape_record 替代 media_scrape;GET /files/scrape-records
- 与 MP ScrapingChain 的文件布局差分(7 个场景,3 处登记偏离)

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

packages/db/migrations/0058_scrape_record.ts

@@ -0,0 +1,72 @@
+import type { Kysely } from "kysely";
+import { execDdl } from "../src/migration-helpers.ts";
+
+// SCR-2 刮削照 MP 简化:刮削对象从"作品"改为"目录 / 视频文件"(不要求先登记),每次刮削一条记录:
+// 目录(登记目录 id + 相对路径)、识别结果(本地作品 / TMDB 条目)、写了哪些文件、跳过 / 失败原因。
+// 替代 SCR-1 的 media_scrape(一部作品一行);作品详情取该作品最近一条记录。
+
+const MEDIA_SCRAPE = `create table if not exists media_scrape (
+   media_entity_id uuid primary key references media_entity (id) on delete cascade,
+   status text not null default 'queued' check (status in ('queued', 'running', 'succeeded', 'partial', 'failed')),
+   trigger text not null default 'manual' check (trigger in ('auto', 'manual')),
+   mode text not null default 'missing' check (mode in ('missing', 'overwrite')),
+   record_id uuid references organization_record (id) on delete set null,
+   engine_run_id text,
+   requested_at timestamptz not null default now(),
+   started_at timestamptz,
+   finished_at timestamptz,
+   written integer not null default 0,
+   skipped integer not null default 0,
+   failed integer not null default 0,
+   reason text,
+   files jsonb not null default '[]'::jsonb,
+   created_at timestamptz not null default now(),
+   updated_at timestamptz not null default now()
+ );
+ drop trigger if exists media_scrape_touch_updated_at on media_scrape;
+ create trigger media_scrape_touch_updated_at before update on media_scrape
+   for each row execute function umt_touch_updated_at();`;
+
+// biome-ignore lint/suspicious/noExplicitAny: 迁移不依赖应用表类型
+export async function up(db: Kysely<any>): Promise<void> {
+  await execDdl(
+    db,
+    `create table if not exists scrape_record (
+       id uuid primary key default gen_random_uuid(),
+       root_id text not null,
+       root_path text not null,
+       library_root_id uuid references library_root (id) on delete set null,
+       path text not null default '',
+       media_entity_id uuid references media_entity (id) on delete set null,
+       match jsonb,
+       status text not null default 'queued' check (status in ('queued', 'running', 'succeeded', 'partial', 'failed')),
+       trigger text not null default 'manual' check (trigger in ('auto', 'manual')),
+       mode text not null default 'missing' check (mode in ('missing', 'overwrite')),
+       organization_record_id uuid references organization_record (id) on delete set null,
+       engine_run_id text,
+       actor_type text,
+       actor_id text,
+       requested_at timestamptz not null default now(),
+       started_at timestamptz,
+       finished_at timestamptz,
+       written integer not null default 0,
+       skipped integer not null default 0,
+       failed integer not null default 0,
+       reason text,
+       files jsonb not null default '[]'::jsonb,
+       created_at timestamptz not null default now(),
+       updated_at timestamptz not null default now()
+     );
+     create index if not exists scrape_record_entity_idx on scrape_record (media_entity_id, requested_at desc);
+     create index if not exists scrape_record_path_idx on scrape_record (root_id, path, requested_at desc);
+     drop trigger if exists scrape_record_touch_updated_at on scrape_record;
+     create trigger scrape_record_touch_updated_at before update on scrape_record
+       for each row execute function umt_touch_updated_at();
+     drop table if exists media_scrape;`,
+  );
+}
+
+// biome-ignore lint/suspicious/noExplicitAny: 同上
+export async function down(db: Kysely<any>): Promise<void> {
+  await execDdl(db, `drop table if exists scrape_record; ${MEDIA_SCRAPE}`);
+}