@linxiraos/pi-stats 1.2.0 → 1.2.1

package/CHANGELOG.md

@@ -4,14 +4,14 @@
 
 ### Added
 
-- 看板按宿主语言渲染:CLI 的语言设置(或浏览器偏好)注入页面,侧栏、卡片、表头、图例、空状态与日期/数字格式全部跟随;首发简体中文,英文为回退。语言在启动时固定(页面内无切换器),因为看板随 `zetacode stats` 一起启动。
-- 「情绪压力」页:判据覆盖卡会区分「已评判 / 正则 / 未评分」——词典未覆盖的语言(如日语、韩语、西里尔文)显示为**未评分**而不是判定为平静。
+- The dashboard renders in the host language: the CLI's language setting (or the browser preference) is injected into the page, and the sidebar, cards, table headers, legends, empty states, and date/number formats all follow it; Simplified Chinese ships first, with English as the fallback. The language is fixed at startup (no in-page switcher) because the dashboard launches together with `zetacode stats`.
+- The "emotional pressure" page: the coverage card distinguishes "judged / regex / unscored" — languages the dictionary does not cover (e.g. Japanese, Korean, Cyrillic) show as **unscored** rather than being judged calm.
 
 ### Changed
 
 - Improved dashboard responsiveness and efficiency by reducing unnecessary data refreshes and re-rendering, speeding up session synchronization, database access, package imports, and usage, model, and time-series requests, and avoiding repeated downloads of unchanged traces.
 - Dashboard requests are now cancelled when no longer needed, improving responsiveness when switching sessions or closing trace views during loading.
-- 「情绪压力」的辨识层重做:新增中文与语音输入词典(无词界/不依赖标点,因为中文输入大量来自语音转写,脏话常呈同音字),新增强调信号(全角爆点、叠字、叠词)用于判定“发怒”,并新增**整条消息重复上一条**这一跨语言信号。不再依赖 emoji(用户输入几乎不用,模型才输出);英文词典保持不变。判定提示词改为语言中立,并适配单条消息内中英混杂。
+- Reworked the "emotional pressure" detection layer: added a Chinese and voice-input dictionary (no word boundaries and no punctuation reliance, since Chinese input often comes from speech transcription and profanity frequently appears as homophones), added emphasis signals (full-width exclamation runs, repeated characters, repeated words) for judging anger, and added a cross-language signal — **an entire message repeating the previous one**. It no longer relies on emoji (users barely type them; only models emit them); the English dictionary is unchanged. The judging prompt is now language-neutral and tolerates mixed Chinese/English within a single message.
 
 ## [1.1.25] - 2026-10-03
 

package/README.md

@@ -29,16 +29,16 @@ Subscription-backed models use matching public API prices when an exact public m
 
 ```bash
 # Start dashboard server (default: http://localhost:3847)
-zeta-c stats
+zetacode stats
 
 # Custom port
-zeta-c stats --port 8080
+zetacode stats --port 8080
 
 # Print summary to console
-zeta-c stats --summary
+zetacode stats --summary
 
 # Output as JSON (for scripting)
-zeta-c stats --json
+zetacode stats --json
 ```
 
 ### Programmatic