v1.1.0

English

Added

• Batch scraping gains an optional AI name-resolution fallback: ROMs whose English title cannot be resolved locally are matched by your configured OpenAI-compatible LLM against the built-in No-Intro title list. Only titles copied verbatim from the list are accepted, results (including misses) are cached locally, and the per-run switch lives in the batch scrape dialog.
• Sega Saturn libraries now index Alcohol 120% images (.mds/.mdf), so official Traditional-Chinese discs stored in that format appear in the library.

Fixed

• Cartridge-header identification no longer returns No-Intro comma-article names (Lion King, The) as scrape queries; they are restored to natural order so translated SFC ROMs with intact headers finally match.
• The built-in Chinese title database now resolves far more translated ROMs: folder names take priority over cryptic release-group filenames for disc games, full-width brackets and trailing disc markers (CD1, Disc A) are stripped, MSU-1 markers are removed, and queries formed as “database title + extra subtitle” match via a normalized-prefix tier (e.g. 46亿年物语 向远方的伊甸 → E.V.O. - Search for Eden).
• Fuzzy matching validates digit signatures, eliminating confident mismatches such as 梦幻模拟战1&2 → Langrisser III or 光明力量3剧本2 → Marie no Atelier; exact multi-region hits now deterministically prefer informative variants (D之食卓 → D no Shokutaku instead of the unsearchable D).
• Saturn/Dreamcast disc-header titles are only trusted when the Chinese database cannot resolve a name, and polluted headers (rewritten by translation tools) are rejected instead of producing garbled queries.
• Injected-disc PlayStation translations no longer mislead Providers: the carrier disc's serial is ignored once the Chinese database resolves the real title, and lettered disc markers (Disc A) are cleaned.
• Batch scrape no longer reports “0 games” for libraries whose gamelist.xml contains empty tags: empty <desc/>/<image/> elements no longer count as scraped metadata or produce phantom media paths.
• Disc-based platforms (PS1, Saturn, Dreamcast, Mega-CD, PCE-CD, 3DO) index a single bootable descriptor per disc — CUE-referenced tracks and same-name bin/img/mdf/ccd/sub companions are no longer listed as separate games.
• BIOS and system files (*bios*, scph*, syscard*, neogeo.zip, pgm.zip, dc_boot/dc_flash) are excluded from scanning on every platform.

简体中文

新增

• 批量抓取新增可选的 AI 名称解析兜底：本地无法解析英文名的 ROM，交给你配置的 OpenAI 兼容 LLM 在内置 No-Intro 标题清单内匹配。只接受逐字来自清单的标题，结果（含未命中）本地缓存，开关位于批量抓取对话框、按次生效。
• 世嘉土星库现在会索引 Alcohol 120% 镜像（.mds/.mdf），官方繁体中文光盘不再缺席。

修复

• 卡带头识别不再把 No-Intro 逗号冠词名（Lion King, The）直接用作抓取查询词，而是还原自然语序，卡带头完好的汉化 SFC ROM 终于能够命中。
• 内置中文标题库解析能力大幅增强：光盘游戏优先使用中文文件夹名而非汉化组代号文件名，清理全角括号与尾部盘号（CD1、Disc A），剥离 MSU-1 标记，“库标题 + 额外副标题”式查询经规范化前缀层命中（如 46亿年物语 向远方的伊甸 → E.V.O. - Search for Eden）。
• 模糊匹配增加数字签名校验，消除“梦幻模拟战1&2 → Langrisser III”“光明力量3剧本2 → 玛莉的炼金工房”这类自信错配；多区域精确命中确定性优先取信息量足的变体（D之食卓 → D no Shokutaku 而非检索不到的 D）。
• 土星/DC 光盘头标题仅在中文库解析失败时才采信，被汉化工具改写污染的头部会被拒绝，不再产生乱码查询。
• 注入式 PlayStation 汉化盘不再误导 Provider：中文库解析出真实标题后忽略底盘序列号，并清理字母盘号（Disc A）。
• 修复 gamelist.xml 空标签导致批量抓取显示“0 个游戏”的问题：空 <desc/>/<image/> 不再被计为已抓取，也不再产生指向库目录的伪媒体路径。
• 光盘平台（PS1、土星、DC、Mega-CD、PCE-CD、3DO）一张盘只索引一个可引导描述文件——CUE 引用轨与同名 bin/img/mdf/ccd/sub 不再被列为独立游戏。
• 全平台扫描排除 BIOS 与系统文件（*bios*、scph*、syscard*、neogeo.zip、pgm.zip、dc_boot/dc_flash）。