feat(identity): weighted evidence table, sequel numbers, generic-name guard (ID-2)

Configurable DEFAULT_MATCH_WEIGHTS; Chinese/English names both count and
corroborate; Chapter/Part/第N章 sequel conflicts reject; missing season
penalised; category words and single-word aliases colliding in the pool
cannot confirm alone; alias-only hits need year or second-script support;
same-level ties broken by exact year then matched name scripts.

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

packages/domain/src/identity/normalize.ts

@@ -35,3 +35,69 @@ export function normalizeTitle(s: string): string {
     .replace(/&/g, "and")
     .replace(/[^\p{L}\p{N}]+/gu, "");
 }
+
+/** 分类词(站点分类 / RSS 标题前缀)与常见停用词:不能当片名或别名用于身份比对(ID-2) */
+const GENERIC_NAMES = new Set(
+  [
+    "电影",
+    "电视剧",
+    "剧集",
+    "连续剧",
+    "动漫",
+    "动画",
+    "番剧",
+    "国漫",
+    "日漫",
+    "综艺",
+    "纪录片",
+    "记录片",
+    "音乐",
+    "短剧",
+    "微短剧",
+    "体育",
+    "游戏",
+    "软件",
+    "学习",
+    "其他",
+    "其它",
+    "未知",
+    "新",
+    "合集",
+    "全集",
+    "中字",
+    "国语",
+    "粤语",
+    "高清",
+    "蓝光",
+    "原盘",
+    "movie",
+    "movies",
+    "tv",
+    "tvseries",
+    "series",
+    "anime",
+    "animation",
+    "documentary",
+    "music",
+    "variety",
+    "show",
+    "the",
+    "season",
+    "complete",
+    "unknown",
+    "other",
+  ].map((s) => s.toLowerCase()),
+);
+
+/**
+ * 归一化后的名称是否太"通用"不能用于身份比对:分类词、停用词、单个汉字、两个以内的拉丁字符、纯数字。
+ * 参数是 normalizeTitle 之后的形态。
+ */
+export function isGenericAliasName(norm: string): boolean {
+  if (!norm) return true;
+  if (GENERIC_NAMES.has(norm)) return true;
+  if (/^\d+$/.test(norm)) return true;
+  const han = norm.match(/\p{Script=Han}/gu)?.length ?? 0;
+  if (han > 0) return [...norm].length < 2;
+  return norm.length <= 2;
+}