@goodandready/dsh-voice 0.8.19 → 0.8.21

package/lib/index.js

@@ -213,7 +213,10 @@ export function apply(ctx, baseConfig) {
     try {
       const controller = new AbortController()
       const t = setTimeout(() => controller.abort(), 2000)
-      const res = await fetch(live().whisperUrl.split('/inference')[0] + '/', { signal: controller.signal })
+      const rawUrl = String(live().whisperUrl || '').trim()
+      const base = rawUrl.includes('/inference') ? rawUrl.split('/inference')[0] : rawUrl.replace(/\/+$/, '')
+      if (!base) return false
+      const res = await fetch(base + '/', { signal: controller.signal })
       clearTimeout(t)
       return res.ok
     } catch { return false }
@@ -266,7 +269,8 @@ export function apply(ctx, baseConfig) {
     if (!cfg.sensevoiceModel) return false
     if (await sensevoiceAlive()) return true
     try {
-      const port = new URL(cfg.sensevoiceUrl).port || '6006'
+      let port = '6006'
+      try { port = new URL(cfg.sensevoiceUrl).port || '6006' } catch { /* invalid URL */ }
       const spec = ctx.shell.resolve({
         command: `${JSON.stringify(cfg.sensevoiceBin)}`
           + ` --sense-voice-model=${JSON.stringify(cfg.sensevoiceModel)}`
@@ -533,7 +537,7 @@ export function apply(ctx, baseConfig) {
         + 'Use for voice messages, recordings, interviews.',
       parameters: {
         file_path: { type: 'string', required: true, description: 'Absolute path to the audio file (wav, mp3, m4a, ogg, flac, webm).' },
-        language: { type: 'string', description: `Recognition language code. Default: ${baseConfig.message.language}.` },
+        language: { type: 'string', description: `Recognition language code. Default: ${(baseConfig && baseConfig.message && baseConfig.message.language) || 'ru'}.` },
       },
       output: {
         schema: {
@@ -549,7 +553,7 @@ export function apply(ctx, baseConfig) {
         },
       },
       isConcurrencySafe: () => false,
-      timeoutMs: baseConfig.timeoutMs * 3 + 5000,
+      timeoutMs: 365000,
       async execute(args, exec) {
         const cfg = live()
         const filePath = String(args.file_path || '').trim()

package/lib/providers.js

@@ -102,6 +102,17 @@ function isAutoLang(lang) {
   return !lang || lang === 'auto' || String(lang).includes(',')
 }
 
+async function readErrorDetail(res, defaultLabel) {
+  let detail = `HTTP ${res.status}`
+  try {
+    const e = await res.json()
+    if (e?.error) detail = typeof e.error === 'string' ? e.error : (e.error.message || JSON.stringify(e.error))
+    else if (e?.message) detail = e.message
+    else if (e?.err_msg) detail = e.err_msg
+  } catch { /* not json */ }
+  return `${defaultLabel} ${detail}`
+}
+
 export function makeProviders(deps, req) {
   const { resolveKey, fetchImpl, cfg } = deps
   const { bytes, mime, lang, signal, models } = req
@@ -122,7 +133,7 @@ export function makeProviders(deps, req) {
       body: bytes,
       signal,
     })
-    if (!res.ok) throw new Error(`Deepgram HTTP ${res.status}`)
+    if (!res.ok) throw new Error(await readErrorDetail(res, 'Deepgram'))
     const data = await res.json()
     const text = (data?.results?.channels?.[0]?.alternatives?.[0]?.transcript || '').trim()
     return { ok: text.length > 0, provider: 'deepgram', text, reason: text ? '' : 'empty transcript' }
@@ -142,7 +153,7 @@ export function makeProviders(deps, req) {
       body: form,
       signal,
     })
-    if (!res.ok) throw new Error(`Groq HTTP ${res.status}`)
+    if (!res.ok) throw new Error(await readErrorDetail(res, 'Groq'))
     const data = await res.json()
     const text = (data?.text || '').trim()
     return { ok: text.length > 0, provider: 'groq', text, reason: text ? '' : 'empty transcript' }
@@ -158,7 +169,7 @@ export function makeProviders(deps, req) {
       body: bytes,
       signal,
     })
-    if (!res.ok) throw new Error(`HF HTTP ${res.status}`)
+    if (!res.ok) throw new Error(await readErrorDetail(res, 'HF'))
     const data = await res.json()
     const text = (data?.text || '').trim()
     return { ok: text.length > 0, provider: 'hf', text, reason: text ? '' : 'empty transcript' }
@@ -286,7 +297,7 @@ export function makeProviders(deps, req) {
       const res = await fetchImpl(`${base}/audio/transcriptions`, {
         method: 'POST', headers, body: form, signal,
       })
-      if (!res.ok) throw new Error(`${label} HTTP ${res.status}`)
+      if (!res.ok) throw new Error(await readErrorDetail(res, label))
       const data = await res.json()
       return (data?.text || '').trim()
     }
@@ -319,7 +330,7 @@ export function makeProviders(deps, req) {
         }),
         signal,
       })
-      if (!res.ok) throw new Error(`${label} HTTP ${res.status}`)
+      if (!res.ok) throw new Error(await readErrorDetail(res, label))
       const data = await res.json()
       return String(data?.choices?.[0]?.message?.content || '').trim()
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@goodandready/dsh-voice",
-  "version": "0.8.19",
+  "version": "0.8.21",
   "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",