fix(reader): split large txt into chapters (章 as boundaries, 卷 as optgroup labels, pseudo-sections when unmarked) rendered one at a time so the browser stops freezing; decode via strict-UTF-8-then-GB18030 with UTF-16 BOM sniffing to kill mojibake; progress locator {ch,scrollFraction} with old global-fraction fallback

docs/CHANGELOG_web.md

@@ -8,6 +8,13 @@ The format loosely follows Keep a Changelog and can be adapted to the team's hab
 
 ## [Unreleased]
 
+### Fixed / 修复
+
+- .txt/.md files encoded in GBK/GB2312/GB18030 or UTF-16 (typical web-novel exports) no longer render as mojibake: the reader strictly tries UTF-8 first and falls back to GB18030, recognizing UTF-16 by BOM.
+- GBK/GB2312/GB18030 或 UTF-16 编码的 .txt/.md（网文导出常见）不再乱码：阅读器先按 UTF-8 严格解码，失败自动回退 GB18030，UTF-16 靠 BOM 识别。
+- Reading a large .txt no longer freezes the browser: the text is split into chapters (「第X章」/Chapter X/序章… markers, pseudo-sections for unmarked files) and rendered one chapter at a time with a chapter selector and prev/next navigation; 「第X卷/部」 are volume groupings (dropdown optgroups), not chapter breaks, and books with only volumes split by volume; reading progress restores to the saved chapter.
+- 阅读大 txt 不再卡死浏览器：正文按章节切分（识别「第X章」/Chapter X/序章等标题，无标记的按段落切成小节），每次只渲染一章，并提供章节目录选择与上一章/下一章导航；「第X卷/部」按卷分组（下拉框分组标签）而非章节边界，只有卷没有章的书按卷切分；阅读进度会恢复到上次的章节。
+
 ### Changed / 变更
 
 - Creating a library on the admin Libraries page no longer asks for a server-side absolute path; just a name (the directory lands at `BOOKS_DIR/<name>` automatically).