UHD_Media_Tool/packages/db/migrations/0047_identity_pool.ts

import type { Kysely } from "kysely";
import { execDdl, oneOf } from "../src/migration-helpers.ts";

// DEC-1 惰性识别 + 待识别池治理(REQ §4.2–4.3、§7.5、§13;对照 MP:RSS 资源只在匹配订阅 / 用户搜索时识别,识别失败不告警):
// - identity_candidate.reason 新增 unmatched:本地召回无结果、也不属于任何需要 TMDB 查询的触发(订阅别名 / 新作品收录规则 /
//   手动 / 搜索),只进池、不查 TMDB、不开待处理;
// - resolution_attempt.outcome 同步新增 unmatched;
// - identity_candidate.normalized_title:归一化标题(与 title_alias.normalized 同一口径),也是待处理合并分组键
//   (来源 + 归一化标题)的一部分;title_alias 插入触发器按它唤醒相近的池条目复查(新证据到来即重试);
// - 清理存量:未关闭的 identity_unresolved 事项里,资源未关联作品、池内无候选、标题也不命中任何本地别名的,
//   事项关闭为 dismissed(resolution.resolution = 'moved_to_pool'),池条目改为 unmatched 并解除事项关联。

const OLD_REASONS = ["ambiguous", "no_match", "proposed", "lookup_failed"] as const;
const OLD_OUTCOMES = [
  "auto",
  "proposed",
  "ambiguous",
  "no_match",
  "lookup_failed",
  "manual",
  "skipped_locked",
] as const;

// 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,
    `
alter table identity_candidate drop constraint identity_candidate_reason_check;
alter table identity_candidate add constraint identity_candidate_reason_check
  ${oneOf("reason", [...OLD_REASONS, "unmatched"])};
alter table resolution_attempt drop constraint resolution_attempt_outcome_check;
alter table resolution_attempt add constraint resolution_attempt_outcome_check
  ${oneOf("outcome", [...OLD_OUTCOMES, "unmatched"])};

alter table identity_candidate add column normalized_title text;
update identity_candidate set normalized_title = nullif(query->>'normalized', '');
create index identity_candidate_normalized_trgm_idx on identity_candidate
  using gin (normalized_title gin_trgm_ops) where status = 'pending';

-- 新证据唤醒:任何路径新增别名(建档 / 手动加别名 / 合并)后,归一化标题相同或足够相近(与本地召回同阈值 0.45)的
-- unmatched / no_match 池条目立即到期复查(复查仍是本地召回优先,命中别名才查 TMDB)
create function identity_pool_wake_on_alias() returns trigger language plpgsql as $$
begin
  if new.normalized is null or new.normalized = '' then return null; end if;
  update identity_candidate set next_review_at = now()
   where status = 'pending' and reason in ('unmatched', 'no_match') and normalized_title is not null
     and (next_review_at is null or next_review_at > now())
     and (normalized_title = new.normalized
          or (length(new.normalized) >= 3 and normalized_title % new.normalized
              and similarity(normalized_title, new.normalized) >= 0.45));
  return null;
end $$;
create trigger title_alias_wake_identity_pool after insert on title_alias
  for each row execute function identity_pool_wake_on_alias();

with moved as (
  select p.id as pool_id, d.id as decision_id
    from identity_candidate p
    join source_release r on r.id = p.source_release_id
    join decision d on d.id = p.decision_id or d.dedupe_key = 'identity:' || p.source_release_id::text
   where d.kind = 'identity_unresolved'
     and d.status in ('open', 'investigating', 'awaiting_user')
     and p.status = 'pending'
     and r.media_entity_id is null
     and jsonb_array_length(p.candidates) = 0
     and not exists (
       select 1 from title_alias a
        where p.normalized_title is not null
          and (a.normalized = p.normalized_title
               or (length(p.normalized_title) >= 3 and similarity(a.normalized, p.normalized_title) >= 0.45)))
), closed as (
  update decision set status = 'dismissed',
         resolution = jsonb_build_object('resolution', 'moved_to_pool', 'by', 'migration:0047_identity_pool'),
         resolved_by = 'system:identity-pool', resolved_at = now()
   where id in (select decision_id from moved)
  returning id
)
update identity_candidate set reason = 'unmatched', decision_id = null, next_review_at = null
 where id in (select pool_id from moved);
`,
  );
}

// biome-ignore lint/suspicious/noExplicitAny: 同上
export async function down(db: Kysely<any>): Promise<void> {
  await execDdl(
    db,
    `
update identity_candidate set reason = 'no_match' where reason = 'unmatched';
update resolution_attempt set outcome = 'no_match' where outcome = 'unmatched';
drop trigger if exists title_alias_wake_identity_pool on title_alias;
drop function if exists identity_pool_wake_on_alias();
drop index if exists identity_candidate_normalized_trgm_idx;
alter table identity_candidate drop column normalized_title;
alter table identity_candidate drop constraint identity_candidate_reason_check;
alter table identity_candidate add constraint identity_candidate_reason_check ${oneOf("reason", OLD_REASONS)};
alter table resolution_attempt drop constraint resolution_attempt_outcome_check;
alter table resolution_attempt add constraint resolution_attempt_outcome_check ${oneOf("outcome", OLD_OUTCOMES)};
`,
  );
}