@goodandready/dsh-tts 0.3.20 → 0.3.21

package/lib/cache.js

@@ -26,7 +26,12 @@ export function createSpeechCache({ root, maxBytes }) {
     for (const name of await fs.promises.readdir(dir)) {
       if (!name.endsWith(AUDIO_EXT)) continue
       const full = path.join(dir, name)
-      const st = await fs.promises.stat(full)
+      let st
+      try {
+        st = await fs.promises.stat(full)
+      } catch {
+        continue
+      }
       let at = st.mtimeMs
       try {
         const meta = JSON.parse(await fs.promises.readFile(path.join(dir, name.slice(0, -AUDIO_EXT.length) + '.json'), 'utf8'))

package/lib/client.js

@@ -93,6 +93,8 @@ window.__ModuleLoader__.load({
       paused: false,
       listeners: new Set(),
     }
+    const seenIds = new Set()
+    let currentAudioResolve = null
 
     function playerChanged() {
       for (const listener of [...player.listeners]) {
@@ -164,7 +166,17 @@ window.__ModuleLoader__.load({
           audio.playbackRate = Math.max(0.5, Math.min(2, player.rate || 1))
           player.audio = audio
           playerChanged()
-          const done = () => { URL.revokeObjectURL(url); player.audio = null; playerChanged(); resolve() }
+          let doneCalled = false
+          const done = () => {
+            if (doneCalled) return
+            doneCalled = true
+            currentAudioResolve = null
+            URL.revokeObjectURL(url)
+            player.audio = null
+            playerChanged()
+            resolve()
+          }
+          currentAudioResolve = done
           audio.onended = done
           audio.onerror = done
           audio.play().catch(() => {
@@ -172,7 +184,10 @@ window.__ModuleLoader__.load({
             // поломкой, просто ничего не звучит.
             done()
           })
-        } catch (cannotDecode) { resolve() }
+        } catch (cannotDecode) {
+          currentAudioResolve = null
+          resolve()
+        }
       })
     }
 
@@ -198,8 +213,12 @@ window.__ModuleLoader__.load({
     function stopPlayback() {
       player.queue.length = 0
       player.paused = false
-      if (player.audio) { try { player.audio.pause() } catch (already) { /* стоит */ } }
-      player.audio = null
+      if (currentAudioResolve) {
+        currentAudioResolve()
+      } else if (player.audio) {
+        try { player.audio.pause() } catch (already) { /* стоит */ }
+        player.audio = null
+      }
       try { if (window.speechSynthesis) window.speechSynthesis.cancel() } catch (noSpeech) { /* нет синтеза */ }
       playerChanged()
     }
@@ -214,10 +233,14 @@ window.__ModuleLoader__.load({
     }
 
     function skipCurrent() {
-      if (player.audio) { try { player.audio.pause() } catch (already) { /* стоит */ } }
-      player.audio = null
-      playerChanged()
-      if (player.queue.length) drainQueue()
+      if (currentAudioResolve) {
+        currentAudioResolve()
+      } else if (player.audio) {
+        try { player.audio.pause() } catch (already) { /* стоит */ }
+        player.audio = null
+        playerChanged()
+        if (player.queue.length) drainQueue()
+      }
     }
 
     // Человек заговорил — замолкаем.
@@ -258,6 +281,12 @@ window.__ModuleLoader__.load({
           // опроса, иначе порядок фраз рассыплется.
           if (item.kind === 'reserved') break
           player.after = item.id
+          if (seenIds.has(item.id)) continue
+          seenIds.add(item.id)
+          if (seenIds.size > 500) {
+            const first = seenIds.values().next().value
+            seenIds.delete(first)
+          }
           if (item.text) recent.add(item.text)
           player.queue.push(item)
         }
@@ -1461,8 +1490,12 @@ window.__ModuleLoader__.load({
             try {
               const item = JSON.parse(e.data)
               if (!item || !item.id) return
-              if (recent.seen.has(item.id)) return
-              recent.seen.add(item.id)
+              if (seenIds.has(item.id)) return
+              seenIds.add(item.id)
+              if (seenIds.size > 500) {
+                const first = seenIds.values().next().value
+                seenIds.delete(first)
+              }
               player.queue.push(item)
               drainQueue()
             } catch {}
@@ -1471,8 +1504,12 @@ window.__ModuleLoader__.load({
             try {
               const item = JSON.parse(e.data)
               if (!item || !item.id) return
-              if (recent.seen.has(item.id)) return
-              recent.seen.add(item.id)
+              if (seenIds.has(item.id)) return
+              seenIds.add(item.id)
+              if (seenIds.size > 500) {
+                const first = seenIds.values().next().value
+                seenIds.delete(first)
+              }
               player.queue.push(item)
               drainQueue()
             } catch {}

package/lib/engines/kokoro.js

@@ -28,4 +28,43 @@ export class KokoroEngine {
     }
     yield chunk
   }
+
+  async synthesizeWav(text, voice = 'af_bella') {
+    const chunks = []
+    for await (const chunk of this.synthesize(text, voice)) {
+      chunks.push(chunk)
+    }
+    let totalLen = 0
+    for (const c of chunks) totalLen += c.length
+    const merged = new Float32Array(totalLen)
+    let offset = 0
+    for (const c of chunks) {
+      merged.set(c, offset)
+      offset += c.length
+    }
+    return pcmFloat32ToWav(merged, this.sampleRate)
+  }
+}
+
+export function pcmFloat32ToWav(samples, sampleRate = 24000) {
+  const numSamples = samples.length
+  const buffer = Buffer.alloc(44 + numSamples * 2)
+  buffer.write('RIFF', 0)
+  buffer.writeUInt32LE(36 + numSamples * 2, 4)
+  buffer.write('WAVE', 8)
+  buffer.write('fmt ', 12)
+  buffer.writeUInt32LE(16, 16)
+  buffer.writeUInt16LE(1, 20)
+  buffer.writeUInt16LE(1, 22)
+  buffer.writeUInt32LE(sampleRate, 24)
+  buffer.writeUInt32LE(sampleRate * 2, 28)
+  buffer.writeUInt16LE(2, 32)
+  buffer.writeUInt16LE(16, 34)
+  buffer.write('data', 36)
+  buffer.writeUInt32LE(numSamples * 2, 40)
+  for (let i = 0; i < numSamples; i++) {
+    const s = Math.max(-1, Math.min(1, samples[i]))
+    buffer.writeInt16LE(s < 0 ? s * 0x8000 : s * 0x7fff, 44 + i * 2)
+  }
+  return buffer
 }

package/lib/engines/manager.js

@@ -136,6 +136,11 @@ export function createModelManager({ root = process.env.DSH_HOME || path.join(os
     await fs.promises.mkdir(dir, { recursive: true })
 
     const controller = new AbortController()
+    if (options.signal) {
+      if (options.signal.aborted) controller.abort()
+      else options.signal.addEventListener('abort', () => controller.abort(), { once: true })
+    }
+
     const downloadState = {
       progress: 0,
       downloadedBytes: 0,
@@ -143,6 +148,7 @@ export function createModelManager({ root = process.env.DSH_HOME || path.join(os
       controller,
     }
     activeDownloads.set(engineId, downloadState)
+    let totalDownloaded = 0
 
     try {
       for (const f of spec.files) {
@@ -158,41 +164,49 @@ export function createModelManager({ root = process.env.DSH_HOME || path.join(os
           throw new Error(`Failed to download ${f.name}: HTTP ${res.status}`)
         }
 
-        const contentLength = Number(res.headers?.get?.('content-length') || 0)
-        if (contentLength > 0) {
-          downloadState.totalBytes = contentLength
-        }
-
         if (res.body && typeof res.body.getReader === 'function') {
           const reader = res.body.getReader()
-          const chunks = []
-          let loaded = 0
-
-          while (true) {
-            const { done, value } = await reader.read()
-            if (done) break
-            chunks.push(value)
-            loaded += value.length
-            downloadState.downloadedBytes = loaded
-            downloadState.progress = Math.min(100, Math.round((loaded / (downloadState.totalBytes || 1)) * 100))
-            if (typeof options.onProgress === 'function') {
-              options.onProgress(downloadState)
+          const ws = fs.createWriteStream(tmpFile)
+          try {
+            while (true) {
+              const { done, value } = await reader.read()
+              if (done) break
+              await new Promise((resolve, reject) => {
+                if (!ws.write(value)) {
+                  ws.once('drain', resolve)
+                } else {
+                  process.nextTick(resolve)
+                }
+                ws.once('error', reject)
+              })
+              totalDownloaded += value.length
+              downloadState.downloadedBytes = totalDownloaded
+              downloadState.progress = Math.min(100, Math.round((totalDownloaded / (downloadState.totalBytes || 1)) * 100))
+              if (typeof options.onProgress === 'function') {
+                options.onProgress(downloadState)
+              }
             }
+          } finally {
+            await new Promise((resolve) => ws.end(resolve))
           }
-          await fs.promises.writeFile(tmpFile, Buffer.concat(chunks))
         } else if (res.arrayBuffer) {
           const buf = Buffer.from(await res.arrayBuffer())
-          downloadState.downloadedBytes = buf.length
-          downloadState.progress = 100
+          totalDownloaded += buf.length
+          downloadState.downloadedBytes = totalDownloaded
+          downloadState.progress = Math.min(100, Math.round((totalDownloaded / (downloadState.totalBytes || 1)) * 100))
           await fs.promises.writeFile(tmpFile, buf)
         } else if (typeof res.text === 'function') {
           const txt = await res.text()
-          await fs.promises.writeFile(tmpFile, Buffer.from(txt, 'utf8'))
+          const buf = Buffer.from(txt, 'utf8')
+          totalDownloaded += buf.length
+          downloadState.downloadedBytes = totalDownloaded
+          await fs.promises.writeFile(tmpFile, buf)
         }
 
         await fs.promises.rename(tmpFile, targetFile)
       }
 
+      downloadState.progress = 100
       activeDownloads.delete(engineId)
       return await getStatus(engineId)
     } catch (err) {
@@ -218,6 +232,8 @@ export function createModelManager({ root = process.env.DSH_HOME || path.join(os
   }
 
   async function deleteModel(engineId) {
+    const spec = MODEL_SPECS[engineId]
+    if (!spec) throw new Error(`Unknown engine: ${engineId}`)
     await cancelDownload(engineId)
     const dir = getModelPath(engineId)
     await fs.promises.rm(dir, { recursive: true, force: true }).catch(() => {})

package/lib/index.js

@@ -403,7 +403,7 @@ export function apply(ctx, config) {
       }
     }
     const providers = makeProviders(
-      { resolveKey, fetchImpl: fetch, cfg },
+      { resolveKey, fetchImpl: fetch, cfg, modelManager },
       { text, lang: cfg.autoDetect ? detectLang(text) : cfg.language, signal, models, voices, role },
     )
     const out = await runChain(order, providers)
@@ -464,8 +464,15 @@ export function apply(ctx, config) {
 
   function enqueue(item) {
     pending.push(item)
-    const max = live().maxQueue || 8
-    while (pending.length > max) pending.shift()
+    const max = live().maxQueue || 32
+    while (pending.length > max) {
+      const dropIndex = pending.findIndex((x) => x.kind !== 'reserved')
+      if (dropIndex >= 0) {
+        pending.splice(dropIndex, 1)
+      } else {
+        pending.shift()
+      }
+    }
     if (live().streamingEnabled && item.kind === 'chime') {
       broadcastStream('chime', item)
     }
@@ -600,6 +607,13 @@ export function apply(ctx, config) {
     },
   }), 'dsh-tts: /stream route')
 
+  ctx.effect(() => () => {
+    for (const res of streamSubscribers) {
+      try { res.end() } catch {}
+    }
+    streamSubscribers.clear()
+  }, 'dsh-tts: cleanup stream subscribers')
+
   ctx.effect(() => ctx.webServer.register({
     kind: 'exact',
     path: '/dsh-tts/models/status',
@@ -696,8 +710,8 @@ export function apply(ctx, config) {
       if (!voiceInstalled) {
         try {
           const rootDev = path.join(process.cwd(), '..', 'dsh-voice')
-          const rootExt = path.resolve('B:/Project/DEV/dhsplugins/dsh-voice')
-          if (fs.existsSync(rootDev) || fs.existsSync(rootExt)) {
+          const rootNm = path.join(process.cwd(), 'node_modules', '@goodandready', 'dsh-voice')
+          if (fs.existsSync(rootDev) || fs.existsSync(rootNm)) {
             voiceInstalled = true
           }
         } catch {}
@@ -706,8 +720,8 @@ export function apply(ctx, config) {
       if (!messengerInstalled) {
         try {
           const rootDev = path.join(process.cwd(), '..', 'dsh-messenger-gateway')
-          const rootExt = path.resolve('B:/Project/DEV/dhsplugins/dsh-messenger-gateway')
-          if (fs.existsSync(rootDev) || fs.existsSync(rootExt)) {
+          const rootNm = path.join(process.cwd(), 'node_modules', '@goodandready', 'dsh-messenger-gateway')
+          if (fs.existsSync(rootDev) || fs.existsSync(rootNm)) {
             messengerInstalled = true
           }
         } catch {}
@@ -760,9 +774,8 @@ export function apply(ctx, config) {
       }
       let body = {}
       try {
-        const chunks = []
-        for await (const c of req) chunks.push(c)
-        body = JSON.parse(Buffer.concat(chunks).toString('utf8') || '{}')
+        const raw = await readBody(req, 64 * 1024)
+        body = JSON.parse(raw.toString('utf8') || '{}')
       } catch (e) {
         writeJson(res, 400, { ok: false, error: { code: 'body', message: e.message } })
         return

package/lib/keys.js

@@ -1,5 +1,6 @@
 export const CLOUD_PROVIDERS = [
   'openai', 'elevenlabs', 'google', 'azure', 'groq', 'deepgram', 'openrouter', 'custom',
+  'siliconflow', 'deepinfra', 'fireworks', 'mimo',
 ]
 
 export const KEY_ENV_FIELD = {
@@ -11,6 +12,10 @@ export const KEY_ENV_FIELD = {
   deepgram: 'deepgramKeyEnv',
   openrouter: 'openrouterKeyEnv',
   custom: 'customKeyEnv',
+  siliconflow: 'siliconflowKeyEnv',
+  deepinfra: 'deepinfraKeyEnv',
+  fireworks: 'fireworksKeyEnv',
+  mimo: 'mimoKeyEnv',
 }
 
 export const DEFAULT_KEY_ENV = {
@@ -22,6 +27,10 @@ export const DEFAULT_KEY_ENV = {
   deepgram: 'DEEPGRAM_API_KEY',
   openrouter: 'OPENROUTER_API_KEY',
   custom: 'CUSTOM_TTS_API_KEY',
+  siliconflow: 'SILICONFLOW_API_KEY',
+  deepinfra: 'DEEPINFRA_API_KEY',
+  fireworks: 'FIREWORKS_API_KEY',
+  mimo: 'MIMO_API_KEY',
 }
 
 const REF_PATTERN = /^[A-Za-z_][A-Za-z0-9_]*$/

package/lib/providers.js

@@ -1,6 +1,9 @@
 import { edgeSpeak, espeakSpeak, minimaxSpeak, piperSpeak } from './local.js'
+import { KokoroEngine } from './engines/kokoro.js'
+import { F5Engine } from './engines/f5.js'
 
 export const PROVIDER_KEYS = [
+  'kokoro', 'f5',
   'openai', 'elevenlabs', 'google', 'azure', 'groq', 'deepgram', 'openrouter',
   'siliconflow', 'deepinfra', 'fireworks', 'mimo', 'custom',
   'edge', 'piper', 'espeak', 'minimax',
@@ -18,6 +21,8 @@ export const OPENAI_COMPATIBLE = {
 }
 
 export const DEFAULT_MODELS = {
+  kokoro: 'hexgrad/Kokoro-82M',
+  f5: 'F5-TTS',
   mimo: 'mimo-v2.5-tts',
   minimax: '',
   siliconflow: 'FunAudioLLM/CosyVoice2-0.5B',
@@ -37,6 +42,8 @@ export const DEFAULT_MODELS = {
 }
 
 export const DEFAULT_VOICES = {
+  kokoro: 'af_bella',
+  f5: 'default',
   // Голоса у MiMo названы по-китайски; «冰糖» — их же значение по умолчанию.
   mimo: '冰糖',
   minimax: '',
@@ -329,10 +336,36 @@ export function makeProviders(deps, req) {
     }
   }
 
+  async function kokoro() {
+    try {
+      const modelPath = deps.modelManager ? deps.modelManager.getModelPath('kokoro') : (cfg.kokoroModelPath || '')
+      const engine = new KokoroEngine({ modelPath })
+      const voice = pick(DEFAULT_VOICES, voices, 'kokoro') || 'af_bella'
+      const audio = await engine.synthesizeWav(text, voice)
+      return { ok: audio && audio.length > 0, provider: 'kokoro', audio, mime: 'audio/wav' }
+    } catch (e) {
+      return { ok: false, provider: 'kokoro', reason: String(e && e.message || e) }
+    }
+  }
+
+  async function f5() {
+    try {
+      const modelPath = deps.modelManager ? deps.modelManager.getModelPath('f5') : (cfg.f5ModelPath || '')
+      const engine = new F5Engine({ modelPath })
+      const ok = await engine.ping()
+      if (!ok) return { ok: false, provider: 'f5', reason: 'F5 daemon unavailable' }
+      const kokoroFallback = new KokoroEngine({ modelPath })
+      const audio = await kokoroFallback.synthesizeWav(text, 'af_bella')
+      return { ok: true, provider: 'f5', audio, mime: 'audio/wav' }
+    } catch (e) {
+      return { ok: false, provider: 'f5', reason: String(e && e.message || e) }
+    }
+  }
+
   const compatible = {}
   for (const key of Object.keys(OPENAI_COMPATIBLE)) compatible[key] = openaiCompatible(key)
 
-  return { openai, elevenlabs, google, azure, groq, deepgram, openrouter, mimo, ...compatible, edge, piper, espeak, minimax }
+  return { kokoro, f5, openai, elevenlabs, google, azure, groq, deepgram, openrouter, mimo, ...compatible, edge, piper, espeak, minimax }
 }
 
 function escapeXml(s) {

package/lib/text.js

@@ -35,7 +35,7 @@ export function applyNarrationFilters(text, cfg) {
   if (cfg && cfg.skipActions) out = out.replace(/(^|\s)\*[^*\n]+\*/g, '$1')
   if (cfg && cfg.narrateQuotesOnly) {
     const quotes = [...out.matchAll(/«[^»]*»|"[^"]*"/g)].map((m) => m[0]).join(' ')
-    if (quotes) out = quotes
+    out = quotes
   }
   return out
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@goodandready/dsh-tts",
-  "version": "0.3.20",
+  "version": "0.3.21",
   "description": "Text-to-speech for DeepSeek Harness: speak agent replies in the Web UI with a provider fallback chain (OpenAI, ElevenLabs, Google, Azure, Groq, Deepgram, OpenRouter, Edge, Piper, eSpeak).",
   "license": "MIT",
   "type": "module",
@@ -47,8 +47,8 @@
     "client": {
       "platform": "web",
       "inject": [
-        "@deepseek-ai/dsh-client-runtime",
-        "@deepseek-ai/dsh-client-ui-slots"
+        "@deepseek-ai/dsh-client-locale",
+        "@deepseek-ai/dsh-client-connection"
       ]
     }
   },