@goodandready/dsh-voice 0.9.12 → 0.9.14

package/README.md

@@ -76,7 +76,7 @@ graph LR
 * 🎮 **Tactile Push-to-Talk**:
   * **Mouse**: Hold the wave button — releasing sends the message; dragging pointer away discards.
   * **Keyboard**: Hold <kbd>Ctrl</kbd> (or custom hotkey) for hands-free speaking; press <kbd>Esc</kbd> to cancel.
-* ⚡ **Zero-Latency In-Browser Captions (`browser`)**: Chrome Web Speech API recognition runs 100% locally with live floating captions as you speak.
+* ⚡ **Live In-Browser Captions (`browser`)**: Web Speech API speech-to-text with real-time floating captions as you speak (Note: standard browser recognition typically streams audio to vendor servers like Google in Chrome; for true 100% offline privacy, use local `whisper` or `sensevoice`).
 * 🛡️ **Ironclad Multi-Provider Fallbacks**: If your primary cloud provider runs out of credits or hits a 429 rate limit, requests seamlessly fail over down the chain.
 * 🧠 **Context Glossary Injection**: Automatically extracts code variables and identifiers from your composer draft to steer STT model accuracy on technical jargon.
 * 🎵 **Embedded Audio Player**: Preview, scrubber, and playback of your recorded voice message directly in chat and the composer dock.
@@ -151,7 +151,7 @@ Quality batch after the v0.8.18 review (Gitea #79–#87, PR #88):
 
 | Provider Key | Service Backend | Default Model | Credential Ref | Features & Notes |
 |---|---|---|---|---|
-| `browser` | Web Speech API | Native Browser | *None* | Zero latency, floating live captions in Chrome |
+| `browser` | Web Speech API | Native Browser | *None* | Zero latency, floating live captions in Chrome (vendor server-based; not offline) |
 | `deepgram` | Deepgram API | `nova-2` | `DEEPGRAM_API_KEY` | Ultra-fast cloud transcription |
 | `groq` | Groq Whisper | `whisper-large-v3-turbo` | `GROQ_API_KEY` | Near-instant inference speed |
 | `hf` | HuggingFace Inference | `openai/whisper-large-v3` | `HF_TOKEN` | High-accuracy open Whisper |

package/README.ru.md

@@ -76,7 +76,7 @@ graph LR
 * 🎮 **Тактильный Push-to-Talk**:
   * **Мышь**: зажмите кнопку волны — запись идёт пока кнопка зажата; отпускание отправляет, увод курсора отменяет запись.
   * **Клавиатура**: зажмите <kbd>Ctrl</kbd> для записи без мыши; нажмите <kbd>Esc</kbd> для отмены.
-* ⚡ **Субтитры браузера в реальном времени (`browser`)**: локальное распознавание Chrome Web Speech API без отправки звука на сервер с плавающими субтитрами.
+* ⚡ **Субтитры браузера в реальном времени (`browser`)**: распознавание Web Speech API с плавающими субтитрами в реальном времени (Внимание: стандартный браузерный движок в Chrome отправляет аудио на серверы Google; для полностью локальной приватной работы без внешних вызовов используйте локальный `whisper` или `sensevoice`).
 * 🛡️ **Надёжные цепочки фолбеков**: при исчерпании квоты или ошибке 429 плагин мгновенно обращается к следующему провайдеру в списке.
 * 🧠 **Контекстный словарь (Context Glossary)**: автоматическое извлечение переменных и технических терминов из черновика композера для точного распознавания редких слов и кода.
 * 🎵 **Встроенный аудиоплеер**: предпросмотр и воспроизведение записанного голосового сообщения в чате и доке перед отправкой или для переслушивания.
@@ -151,7 +151,7 @@ Quality-батч после ревью v0.8.18 (Gitea #79–#87, PR #88):
 
 | Ключ | Сервис | Модель по умолчанию | Переменная секрета | Особенности |
 |---|---|---|---|---|
-| `browser` | Web Speech API | Нативная в браузере | *Не требуется* | Нулевая задержка, живые субтитры в Chrome |
+| `browser` | Web Speech API | Нативная в браузере | *Не требуется* | Нулевая задержка, живые субтитры в Chrome (обработка на серверах вендора; не является офлайн) |
 | `deepgram` | Deepgram API | `nova-2` | `DEEPGRAM_API_KEY` | Сверхбыстрая облачная транскрипция |
 | `groq` | Groq Whisper | `whisper-large-v3-turbo` | `GROQ_API_KEY` | Мгновенная скорость генерации |
 | `hf` | HuggingFace Inference | `openai/whisper-large-v3` | `HF_TOKEN` | Высокоточный облачный Whisper |

package/README.zh.md

@@ -76,7 +76,7 @@ graph LR
 * 🎮 **沉浸式 Push-to-Talk 对讲**：
   * **鼠标操作**：按住声波按钮开始录音，松开发送，拖离按钮取消。
   * **键盘操作**：按住 <kbd>Ctrl</kbd> 无需鼠标即刻说话，按 <kbd>Esc</kbd> 放弃本次录音。
-* ⚡ **浏览器本地零延迟同声字幕 (`browser`)**：Chrome Web Speech API 本地离线解析，说话同时浮动显示实时字幕。
+* ⚡ **浏览器实时同声字幕 (`browser`)**：Web Speech API 实时语音识别与浮动同声字幕（注意：Chrome 等标准浏览器通常将音频发送至服务商云端解析；若需 100% 本地离线隐私保护，请使用本地 `whisper` 或 `sensevoice`）。
 * 🛡️ **多服务商自动容灾切换**：首选 API 额度耗尽或遭遇 429 限流时，毫秒级顺位切换备用引擎。
 * 🧠 **上下文术语注入 (Context Glossary)**：自动从输入草稿中提取代码变量名与专业术语，引导 STT 模型精准转写专业词汇。
 * 🎵 **内嵌音频播放器**：在输入框与录音浮层中随时试听和回放刚刚录制的原始音频片段。
@@ -120,7 +120,7 @@ v0.8.18 质量审查后的修复批次（Gitea #79–#87，PR #88）：
 
 | 服务商标识 | 对应引擎 | 默认模型 | 环境变量凭据 | 特性说明 |
 |---|---|---|---|---|
-| `browser` | Web Speech API | 浏览器原生引擎 | *无需密钥* | 零延迟同声字幕输出 |
+| `browser` | Web Speech API | 浏览器原生引擎 | *无需密钥* | 零延迟同声字幕输出（依赖浏览器厂商云端解析；非离线本地） |
 | `deepgram` | Deepgram API | `nova-2` | `DEEPGRAM_API_KEY` | 极速高精云端转写 |
 | `groq` | Groq Whisper | `whisper-large-v3-turbo` | `GROQ_API_KEY` | 毫秒级极速推理 |
 | `hf` | HuggingFace Inference | `openai/whisper-large-v3` | `HF_TOKEN` | 经典高精度开源模型 |

package/lib/client.js

@@ -235,6 +235,10 @@ window.__ModuleLoader__.load({
       'jargonFrom': 'Spoken slang',
       'jargonTo': 'Correct spelling',
       'jargonPlaceholder': 'k8s -> kubernetes\npr -> pull request',
+      'voiceInput': 'Voice input',
+      'assistantSpeakingGated': 'Assistant speaking (gated mode)',
+      'jargonFormatHint': ' (spoken -> replacement)',
+      'errorPrefix': 'Error',
     }
 
     const zh = {
@@ -438,11 +442,15 @@ window.__ModuleLoader__.load({
       'jargonFrom': '口语词汇',
       'jargonTo': '标准拼写',
       'jargonPlaceholder': 'k8s -> kubernetes\npr -> pull request',
+      'voiceInput': '语音输入',
+      'assistantSpeakingGated': '助手正在发言（门控模式）',
+      'jargonFormatHint': ' (口语 -> 替换文本)',
+      'errorPrefix': '错误',
     }
 
     // Strings are also needed outside components — in recording handlers and
     // slot labels — so the translator is module-level, not only via props.
-    let moduleT = (key) => key
+    let moduleT = (key) => en[key] || key
     const t = (key) => moduleT(key)
     // ------------------------------------------------------------------ css
     const CSS =
@@ -862,6 +870,50 @@ window.__ModuleLoader__.load({
       return main
     }
 
+    function hotkeyMatches(event, name) {
+      if (!event || !name) return false
+      if (typeof name === 'string' && name.includes('+')) {
+        const parts = name.split('+')
+        const main = parts[parts.length - 1]
+        const needCtrl = parts.includes('Control')
+        const needAlt = parts.includes('Alt')
+        const needShift = parts.includes('Shift')
+        const needMeta = parts.includes('Meta')
+        if (needCtrl && !event.ctrlKey) return false
+        if (needAlt && !event.altKey) return false
+        if (needShift && !event.shiftKey) return false
+        if (needMeta && !event.metaKey) return false
+        return event.code === main || event.key === main
+      }
+      if (name === 'Control') return event.key === 'Control'
+      if (name === 'Alt') return event.key === 'Alt'
+      if (name === 'Shift') return event.key === 'Shift'
+      if (name === 'Meta') return event.key === 'Meta'
+      return event.code === name || event.key === name
+    }
+
+    function isAllowedInEditable(name) {
+      if (!name) return false
+      if (['Control', 'Alt', 'Shift', 'Meta'].includes(name)) return true
+      if (/^F\d{1,2}$/.test(name)) return true
+      if (typeof name === 'string' && name.includes('+')) return true
+      if (['Tab', 'Pause', 'ScrollLock', 'Insert'].includes(name)) return true
+      return false
+    }
+
+    function isHotkeyRelease(event, name) {
+      if (!event || !name) return false
+      if (typeof name === 'string' && name.includes('+')) {
+        const parts = name.split('+')
+        if (event.code === parts[parts.length - 1] || event.key === parts[parts.length - 1]) return true
+        for (let i = 0; i < parts.length - 1; i++) {
+          if (event.key === parts[i] || event.code === parts[i]) return true
+        }
+        return false
+      }
+      return event.code === name || event.key === name
+    }
+
     // Announce that the user started speaking.
     //
     // Playback should mute immediately: listening and talking at once is
@@ -1216,7 +1268,7 @@ window.__ModuleLoader__.load({
         if (voice.settings.bargeIn) {
           triggerBargeIn()
         } else if (voice.settings.gatedTurnTaking !== false) {
-          voice.set({ phase: 'error', error: 'Assistant speaking (gated mode)' })
+          voice.set({ phase: 'error', error: t('assistantSpeakingGated') })
           return
         }
       }
@@ -1281,35 +1333,7 @@ window.__ModuleLoader__.load({
       if (cancelled) cancelCurrent()
       else stopCurrent()
     }
-    function hotkeyMatches(event, name) {
-      if (!event || !name) return false
-      if (typeof name === 'string' && name.includes('+')) {
-        const parts = name.split('+')
-        const main = parts[parts.length - 1]
-        const needCtrl = parts.includes('Control')
-        const needAlt = parts.includes('Alt')
-        const needShift = parts.includes('Shift')
-        const needMeta = parts.includes('Meta')
-        if (needCtrl && !event.ctrlKey) return false
-        if (needAlt && !event.altKey) return false
-        if (needShift && !event.shiftKey) return false
-        if (needMeta && !event.metaKey) return false
-        return event.code === main || event.key === main
-      }
-      if (name === 'Control') return event.key === 'Control'
-      if (name === 'Alt') return event.key === 'Alt'
-      if (name === 'Shift') return event.key === 'Shift'
-      if (name === 'Meta') return event.key === 'Meta'
-      return event.code === name || event.key === name
-    }
-    function isAllowedInEditable(name) {
-      if (!name) return false
-      if (['Control', 'Alt', 'Shift', 'Meta'].includes(name)) return true
-      if (/^F\d{1,2}$/.test(name)) return true
-      if (typeof name === 'string' && name.includes('+')) return true
-      if (['Tab', 'Pause', 'ScrollLock', 'Insert'].includes(name)) return true
-      return false
-    }
+
     let activeHotkeyCleanup = null
     function clearGlobalHotkey() {
       if (typeof activeHotkeyCleanup === 'function') {
@@ -1336,18 +1360,7 @@ window.__ModuleLoader__.load({
           beginHold(mode)
         }
       }
-    function isHotkeyRelease(event, name) {
-      if (!event || !name) return false
-      if (typeof name === 'string' && name.includes('+')) {
-        const parts = name.split('+')
-        if (event.code === parts[parts.length - 1] || event.key === parts[parts.length - 1]) return true
-        for (let i = 0; i < parts.length - 1; i++) {
-          if (event.key === parts[i] || event.code === parts[i]) return true
-        }
-        return false
-      }
-      return event.code === name || event.key === name
-    }
+
       const up = (event) => {
         if (hold.armed && isHotkeyRelease(event, keyName)) {
           event.preventDefault()
@@ -2386,7 +2399,7 @@ window.__ModuleLoader__.load({
                     }
                   }
                   const rate = item.attempts > 0 ? Math.round((item.successes / item.attempts) * 100) : 100
-                  return React.createElement('div', { key, className: 'dvo-dash-item', title: item.lastError ? `Error: ${item.lastError}` : '' },
+                  return React.createElement('div', { key, className: 'dvo-dash-item', title: item.lastError ? `${t('errorPrefix')}: ${item.lastError}` : '' },
                     React.createElement('div', { className: 'dvo-dash-name' }, key),
                     React.createElement('div', { className: 'dvo-dash-row' },
                       React.createElement('span', { className: 'dvo-badge ' + badgeClass }, badgeText),
@@ -2757,7 +2770,7 @@ window.__ModuleLoader__.load({
           if (!res.ok || data.ok === false) {
             throw new Error(data.error || `HTTP ${res.status}`)
           }
-          setUpdaterMsg(data.message || 'Update completed successfully.')
+          setUpdaterMsg(data.message || t('updateSuccess'))
           checkUpdater()
         } catch (err) {
           setUpdaterMsg(err && err.message ? err.message : String(err))
@@ -2791,7 +2804,7 @@ window.__ModuleLoader__.load({
           t('techJargonCorrection'),
         ),
         React.createElement('div', { className: 'cb-field' },
-          React.createElement('label', null, t('jargonAdd') + ' (spoken -> replacement)'),
+          React.createElement('label', null, t('jargonAdd') + t('jargonFormatHint')),
           React.createElement('textarea', {
             rows: 3,
             disabled: !writable,
@@ -3551,7 +3564,7 @@ window.__ModuleLoader__.load({
             name: 'plugins.item',
             id: ROW_ID,
             order: 60,
-            label: () => 'Voice input',
+            label: () => t('voiceInput'),
             locale: NS,
             inject: () => ({ ctx: ctx }),
           },
@@ -3607,6 +3620,16 @@ window.__ModuleLoader__.load({
       registerComposer(ctx)
       registerSettings(ctx)
     }
+    exports._test = {
+      voice,
+      keyLabel,
+      keyFromEvent,
+      hotkeyMatches,
+      isAllowedInEditable,
+      isHotkeyRelease,
+      extractContextKeywords,
+      resolveBrowserRecognitionLang,
+    }
     return module.exports
   },
 })

package/lib/config-routes.js

@@ -51,6 +51,10 @@ export function registerConfigRoutes(ctx, {
       try {
         raw = await readBody(req, 1024 * 1024)
       } catch (e) {
+        if (e && (e.statusCode === 413 || e.message === 'body too large')) {
+          writeJson(res, 413, { ok: false, error: { code: 'too-large', message: 'payload too large' } })
+          return
+        }
         writeJson(res, 400, { ok: false, error: { code: 'body', message: e.message } })
         return
       }

package/lib/http-util.js

@@ -1,9 +1,30 @@
 // Pure HTTP helpers shared by host routes. No cordis, no network.
 
+export class BodyTooLargeError extends Error {
+  constructor(message = 'body too large') {
+    super(message)
+    this.name = 'BodyTooLargeError'
+    this.statusCode = 413
+    this.code = 'ERR_BODY_TOO_LARGE'
+  }
+}
+
 export function writeJson(res, code, body) {
   try {
-    res.writeHead(code, { 'Content-Type': 'application/json', 'Cache-Control': 'no-store' })
+    const headers = { 'Content-Type': 'application/json', 'Cache-Control': 'no-store' }
+    if (code === 413) {
+      headers['Connection'] = 'close'
+    }
+    res.writeHead(code, headers)
     res.end(JSON.stringify(body))
+    if (code === 413 && res && typeof res.on === 'function') {
+      res.on('finish', () => {
+        const sock = res.socket || (res.req && res.req.socket)
+        if (sock && !sock.destroyed && typeof sock.destroy === 'function') {
+          sock.destroy()
+        }
+      })
+    }
   } catch { /* socket may already be closed */ }
 }
 
@@ -17,13 +38,30 @@ export function readBody(req, maxBytes) {
   return new Promise((resolve, reject) => {
     const chunks = []
     let size = 0
-    req.on('data', (c) => {
+    let exceeded = false
+    const onData = (c) => {
+      if (exceeded) return
       size += c.length
-      if (size > maxBytes) { reject(new Error('body too large')); req.destroy(); return }
+      if (size > maxBytes) {
+        exceeded = true
+        if (typeof req.removeListener === 'function') {
+          req.removeListener('data', onData)
+        }
+        if (typeof req.resume === 'function') {
+          req.resume()
+        }
+        reject(new BodyTooLargeError('body too large'))
+        return
+      }
       chunks.push(c)
+    }
+    req.on('data', onData)
+    req.on('end', () => {
+      if (!exceeded) resolve(Buffer.concat(chunks))
+    })
+    req.on('error', (err) => {
+      if (!exceeded) reject(err)
     })
-    req.on('end', () => resolve(Buffer.concat(chunks)))
-    req.on('error', reject)
   })
 }
 

package/lib/index.js

@@ -421,7 +421,14 @@ export function apply(ctx, baseConfig) {
       if (!isTrustedCaller(req)) { writeJson(res, 403, { ok: false, error: { code: 'forbidden', message: 'forbidden source origin' } }); return }
       const cfg = live()
       let raw
-      try { raw = await readBody(req, 1024 * 1024) } catch (e) { writeJson(res, 400, { ok: false, error: { code: 'body', message: e.message } }); return }
+      try { raw = await readBody(req, 1024 * 1024) } catch (e) {
+        if (e && (e.statusCode === 413 || e.message === 'body too large')) {
+          writeJson(res, 413, { ok: false, error: { code: 'too-large', message: 'body exceeds limit of 1MB' } })
+          return
+        }
+        writeJson(res, 400, { ok: false, error: { code: 'body', message: e.message } })
+        return
+      }
       let payload = {}
       try { payload = JSON.parse(raw.toString('utf8') || '{}') } catch { /* empty */ }
       const text = typeof payload.text === 'string' ? payload.text.trim() : ''

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@goodandready/dsh-voice",
-  "version": "0.9.12",
+  "version": "0.9.14",
   "description": "Voice input for DeepSeek Harness: dictation chunked by pauses and voice messages, each with its own provider fallback chain (Deepgram, Groq, HuggingFace, local whisper.cpp, plus any OpenAI-compatible API of your own).",
   "license": "MIT",
   "type": "module",
@@ -44,8 +44,8 @@
     "test": "node --test test/*.test.mjs",
     "build:client": "node scripts/build-client.mjs",
     "build:check": "node scripts/build-client.mjs --check",
-    "pretest": "node scripts/build-client.mjs",
-    "lint": "node --check lib/*.js scripts/*.mjs",
+    "pretest": "node scripts/build-client.mjs && node scripts/lint.mjs && node scripts/pack-check.mjs",
+    "lint": "node scripts/lint.mjs",
     "pack:check": "node scripts/pack-check.mjs"
   },
   "dsh": {