@johpaz/hive-sdk 0.0.12 → 0.0.15

package/packages/core/src/voice/index.ts

@@ -0,0 +1,656 @@
+import { getDb } from "../storage/SQLiteStorage.ts";
+import { decryptApiKey } from "../storage/crypto.ts";
+import { logger } from "../utils/logger.ts";
+
+export interface VoiceConfig {
+  voiceEnabled: boolean;
+  ttsEnabled: boolean;
+  sttProvider: string | null;
+  ttsProvider: string | null;
+  ttsVoiceId: string | null;
+}
+
+export interface AudioInput {
+  type: "buffer" | "url" | "base64";
+  data: Buffer | string;
+  mimeType?: string;
+}
+
+export interface AudioOutput {
+  type: "buffer" | "base64";
+  data: Buffer | string;
+  mimeType: string;
+}
+
+const log = logger.child("voice");
+
+/**
+ * Limpia texto para síntesis de voz (TTS)
+ * Elimina formato Markdown, emojis y otros elementos que no se pronuncian bien
+ */
+export function cleanTextForTTS(text: string): string {
+  if (!text) return "";
+  
+  return text
+    // Eliminar código en bloque (``` ... ```)
+    .replace(/```[\s\S]*?```/g, " ")
+    // Eliminar código inline (`texto`)
+    .replace(/`([^`]+)`/g, "$1")
+    // Eliminar enlaces [texto](url) → texto
+    .replace(/\[([^\]]+)\]\([^)]+\)/g, "$1")
+    // Eliminar imágenes ![alt](url) → alt
+    .replace(/!\[([^\]]*)\]\([^)]+\)/g, "$1")
+    // Eliminar negritas **texto** → texto
+    .replace(/\*\*([^*]+)\*\*/g, "$1")
+    // Eliminar cursivas *texto* o _texto_ → texto
+    .replace(/\*([^*]+)\*/g, "$1")
+    .replace(/_([^_]+)_/g, "$1")
+    // Eliminar tachado ~~texto~~ → texto
+    .replace(/~~([^~]+)~~/g, "$1")
+    // Eliminar negritas/cursivas combinadas ***texto*** → texto
+    .replace(/\*\*\*([^*]+)\*\*\*/g, "$1")
+    // Eliminar encabezados # texto → texto
+    .replace(/^#+\s+/gm, "")
+    // Eliminar listas con guión - texto → texto
+    .replace(/^[\-\*]\s+/gm, "")
+    // Eliminar listas numeradas 1. texto → texto
+    .replace(/^\d+\.\s+/gm, "")
+    // Eliminar citas > texto → texto
+    .replace(/^>\s+/gm, "")
+    // Eliminar emojis (rangos Unicode de emojis)
+    .replace(/[\p{Emoji}]/gu, "")
+    // Eliminar caracteres de control Unicode
+    .replace(/[\u200B-\u200D\uFEFF]/g, "")
+    // Eliminar espacios múltiples
+    .replace(/\s+/g, " ")
+    // Trim final
+    .trim();
+}
+
+class VoiceService {
+  private static instance: VoiceService;
+
+  private constructor() {}
+
+  static getInstance(): VoiceService {
+    if (!VoiceService.instance) {
+      VoiceService.instance = new VoiceService();
+    }
+    return VoiceService.instance;
+  }
+
+  getChannelVoiceConfig(channelId: string): VoiceConfig {
+    const db = getDb();
+    const result = db.query(`
+      SELECT voice_enabled, tts_enabled, stt_provider, tts_provider, tts_voice_id
+      FROM channels WHERE id = ?
+    `).get(channelId) as {
+      voice_enabled: number;
+      tts_enabled: number;
+      stt_provider: string | null;
+      tts_provider: string | null;
+      tts_voice_id: string | null;
+    } | undefined;
+
+    if (!result) {
+      return {
+        voiceEnabled: false,
+        ttsEnabled: false,
+        sttProvider: null,
+        ttsProvider: null,
+        ttsVoiceId: null,
+      };
+    }
+
+    return {
+      voiceEnabled: result.voice_enabled === 1,
+      ttsEnabled: result.tts_enabled === 1,
+      sttProvider: result.stt_provider,
+      ttsProvider: result.tts_provider,
+      ttsVoiceId: result.tts_voice_id,
+    };
+  }
+
+  async transcribe(audio: AudioInput, modelId: string): Promise<string> {
+    const isGroq = modelId.startsWith("whisper");
+    const isOpenAi = modelId === "whisper-1";
+    
+    if (isGroq) {
+      return this.transcribeWithGroq(audio, modelId);
+    } else if (isOpenAi) {
+      return this.transcribeWithOpenAIWhisper(audio);
+    }
+    
+    log.warn(`Unknown STT provider ${modelId}, defaulting to Groq Whisper`);
+    return this.transcribeWithGroq(audio, "whisper-large-v3-turbo");
+  }
+
+  private async getProviderApiKey(providerId: string): Promise<string | null> {
+    const db = getDb();
+    const provider = db.query(`
+      SELECT api_key_encrypted, api_key_iv FROM providers WHERE id = ?
+    `).get(providerId) as { api_key_encrypted: string; api_key_iv: string } | undefined;
+
+    if (!provider?.api_key_encrypted) {
+      return null;
+    }
+
+    try {
+      return await decryptApiKey(provider.api_key_encrypted, provider.api_key_iv);
+    } catch (error) {
+      log.error(`Failed to decrypt API key for provider ${providerId}: ${(error as Error).message}`);
+      return null;
+    }
+  }
+
+  private async transcribeWithGroq(audio: AudioInput, modelId: string): Promise<string> {
+    const key = await this.getProviderApiKey("groq") || process.env.GROQ_API_KEY;
+    if (!key) {
+      throw new Error("GROQ_API_KEY not configured. Configúrala en Proveedores o en las variables de entorno.");
+    }
+
+    let audioData: ArrayBuffer | Uint8Array;
+    
+    if (audio.type === "buffer") {
+      audioData = new Uint8Array((audio.data as Buffer));
+    } else if (audio.type === "base64") {
+      const buf = Buffer.from(audio.data as string, "base64");
+      audioData = new Uint8Array(buf);
+    } else if (audio.type === "url") {
+      const response = await fetch(audio.data as string);
+      const ab = await response.arrayBuffer();
+      audioData = new Uint8Array(ab);
+    } else {
+      throw new Error("Invalid audio input type");
+    }
+
+    const mime = audio.mimeType || "audio/ogg";
+    const ext = mime.includes("webm") ? "webm"
+      : mime.includes("mp4") || mime.includes("m4a") ? "m4a"
+      : mime.includes("mp3") || mime.includes("mpeg") ? "mp3"
+      : mime.includes("wav") ? "wav"
+      : mime.includes("flac") ? "flac"
+      : "ogg";
+    const blob = new Blob([audioData as BlobPart], { type: mime });
+    const formData = new FormData();
+    formData.append("file", blob, `audio.${ext}`);
+    formData.append("model", modelId);
+    formData.append("response_format", "json");
+    formData.append("language", "es");
+
+    const result = await fetch("https://api.groq.com/openai/v1/audio/transcriptions", {
+      method: "POST",
+      headers: {
+        "Authorization": `Bearer ${key}`,
+      },
+      body: formData,
+    });
+
+    if (!result.ok) {
+      const error = await result.text();
+      throw new Error(`Groq Whisper transcription failed: ${error}`);
+    }
+
+    const data = await result.json() as { text: string };
+    return data.text;
+  }
+
+  private async transcribeWithOpenAIWhisper(audio: AudioInput): Promise<string> {
+    const key = await this.getProviderApiKey("openai") || process.env.OPENAI_API_KEY;
+    if (!key) {
+      throw new Error("OPENAI_API_KEY not configured. Configúrala en Proveedores o en las variables de entorno.");
+    }
+
+    let audioData: ArrayBuffer | Uint8Array;
+    
+    if (audio.type === "buffer") {
+      audioData = new Uint8Array(audio.data as Buffer);
+    } else if (audio.type === "base64") {
+      const buf = Buffer.from(audio.data as string, "base64");
+      audioData = new Uint8Array(buf);
+    } else if (audio.type === "url") {
+      const response = await fetch(audio.data as string);
+      const ab = await response.arrayBuffer();
+      audioData = new Uint8Array(ab);
+    } else {
+      throw new Error("Invalid audio input type");
+    }
+
+    const blob = new Blob([audioData as BlobPart], { type: audio.mimeType || "audio/webm" });
+    const formData = new FormData();
+    formData.append("file", blob, "audio.webm");
+
+    formData.append("model", "whisper-1");
+    formData.append("response_format", "json");
+    formData.append("language", "es");
+
+    const result = await fetch("https://api.openai.com/v1/audio/transcriptions", {
+      method: "POST",
+      headers: {
+        "Authorization": `Bearer ${key}`,
+      },
+      body: formData,
+    });
+
+    if (!result.ok) {
+      const error = await result.text();
+      throw new Error(`OpenAI Whisper transcription failed: ${error}`);
+    }
+
+    const data = await result.json() as { text: string };
+    return data.text;
+  }
+
+  async speak(text: string, modelId: string, voiceId?: string): Promise<AudioOutput> {
+    const isElevenLabs = modelId.startsWith("eleven");
+    const isOpenAI = modelId.startsWith("tts-") || modelId.startsWith("gpt-");
+    const isGemini = modelId.startsWith("gemini");
+    const isQwen = modelId.startsWith("qwen");
+    const isPiper = modelId === "piper" || modelId === "piper-local";
+
+    if (isPiper) {
+      return this.speakWithPiper(text, voiceId);
+    } else if (isElevenLabs) {
+      return this.speakWithElevenLabs(text, modelId, voiceId);
+    } else if (isOpenAI) {
+      return this.speakWithOpenAI(text, modelId, voiceId);
+    } else if (isGemini) {
+      return this.speakWithGemini(text, modelId, voiceId);
+    } else if (isQwen) {
+      return this.speakWithQwen(text, modelId, voiceId);
+    }
+
+    log.warn(`Unknown TTS provider ${modelId}, defaulting to ElevenLabs Flash`);
+    return this.speakWithElevenLabs(text, "eleven_flash_v2_5", voiceId);
+  }
+
+  private async speakWithPiper(text: string, voiceId?: string): Promise<AudioOutput> {
+    const cleanText = cleanTextForTTS(text);
+    const port = Number(process.env.TTS_PORT ?? 5500);
+    const res = await fetch(`http://localhost:${port}/tts`, {
+      method: "POST",
+      headers: { "Content-Type": "application/json" },
+      body: JSON.stringify({ text: cleanText, voice: voiceId }),
+      signal: AbortSignal.timeout(15_000),
+    });
+    if (!res.ok) {
+      throw new Error(`Piper TTS error ${res.status}. ¿Está el servidor TTS corriendo? (Ajustes → Voz)`);
+    }
+    const wav = await res.arrayBuffer();
+    return {
+      type: "buffer",
+      data: Buffer.from(wav),
+      mimeType: "audio/wav",
+    };
+  }
+
+  private async speakWithElevenLabs(text: string, modelId: string, voiceId?: string): Promise<AudioOutput> {
+    const apiKey = await this.getProviderApiKey("elevenlabs");
+    const key = apiKey || process.env.ELEVENLABS_API_KEY;
+    
+    if (!key) {
+      throw new Error("ELEVENLABS_API_KEY not configured");
+    }
+
+    const voice = voiceId || "21m00Tcm4TlvDq8ikWAM";
+    
+    const response = await fetch(`https://api.elevenlabs.io/v1/text-to-speech/${voice}`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        "xi-api-key": key,
+      },
+      body: JSON.stringify({
+        text,
+        model_id: modelId,
+        voice_settings: {
+          stability: 0.5,
+          similarity_boost: 0.75,
+        },
+      }),
+    });
+
+    if (!response.ok) {
+      const error = await response.text();
+      throw new Error(`ElevenLabs TTS failed: ${error}`);
+    }
+
+    const buffer = await response.arrayBuffer();
+    return {
+      type: "buffer",
+      data: Buffer.from(buffer),
+      mimeType: "audio/mpeg",
+    };
+  }
+
+  private async speakWithOpenAI(text: string, modelId: string = "gpt-4o-mini-tts", voiceId?: string): Promise<AudioOutput> {
+    const apiKey = await this.getProviderApiKey("openai-tts");
+    const key = apiKey || process.env.OPENAI_API_KEY;
+
+    if (!key) {
+      throw new Error("OPENAI_API_KEY not configured");
+    }
+
+    const voice = voiceId || "alloy";
+
+    const response = await fetch("https://api.openai.com/v1/audio/speech", {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        "Authorization": `Bearer ${key}`,
+      },
+      body: JSON.stringify({
+        model: modelId,
+        voice,
+        input: text,
+        response_format: "mp3",
+      }),
+    });
+
+    if (!response.ok) {
+      const error = await response.text();
+      throw new Error(`OpenAI TTS failed: ${error}`);
+    }
+
+    const buffer = await response.arrayBuffer();
+    return {
+      type: "buffer",
+      data: Buffer.from(buffer),
+      mimeType: "audio/mpeg",
+    };
+  }
+
+  private async speakWithGemini(text: string, modelId: string, voiceId?: string): Promise<AudioOutput> {
+    const key = process.env.GEMINI_API_KEY;
+
+    if (!key) {
+      throw new Error("GEMINI_API_KEY not configured");
+    }
+
+    const voiceName = voiceId || "Aoede";
+
+    const response = await fetch(`https://generativelanguage.googleapis.com/v1beta/models/${modelId}:generateContent?key=${key}`, {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+      },
+      body: JSON.stringify({
+        contents: [{
+          parts: [{
+            text: `Genera audio de este texto: ${text}`,
+          }]
+        }],
+        generationConfig: {
+          responseModalities: ["AUDIO"],
+          speechConfig: {
+            languageCode: "es-ES",
+            voiceConfig: {
+              prebuiltVoiceConfig: {
+                voiceName,
+              },
+            },
+          },
+        },
+      }),
+    });
+
+    if (!response.ok) {
+      const error = await response.text();
+      throw new Error(`Gemini TTS failed: ${error}`);
+    }
+
+    const data = await response.json() as { candidates?: Array<{ content?: { parts?: Array<{ inlineData?: { data: string } }> } }> };
+    const audioData = data.candidates?.[0]?.content?.parts?.[0]?.inlineData?.data;
+    
+    if (!audioData) {
+      throw new Error("No audio returned from Gemini");
+    }
+
+    const buffer = Buffer.from(audioData, "base64");
+    return {
+      type: "buffer",
+      data: buffer,
+      mimeType: "audio/mpeg",
+    };
+  }
+
+  private async speakWithQwen(text: string, modelId: string, voiceId?: string): Promise<AudioOutput> {
+    const key = process.env.DASHSCOPE_API_KEY;
+
+    if (!key) {
+      throw new Error("DASHSCOPE_API_KEY not configured");
+    }
+
+    const voice = voiceId || "ruoxi";
+
+    const response = await fetch("https://dashscope.aliyuncs.com/api/v1/services/audio/t2a/generation", {
+      method: "POST",
+      headers: {
+        "Content-Type": "application/json",
+        "Authorization": `Bearer ${key}`,
+      },
+      body: JSON.stringify({
+        model: modelId,
+        input: {
+          text,
+        },
+        parameters: {
+          voice,
+          format: "mp3",
+        },
+      }),
+    });
+
+    if (!response.ok) {
+      const error = await response.text();
+      throw new Error(`Qwen TTS failed: ${error}`);
+    }
+
+    const data = await response.json() as { output?: { audio?: string } };
+    const audioData = data.output?.audio;
+    
+    if (!audioData) {
+      throw new Error("No audio returned from Qwen");
+    }
+
+    const buffer = Buffer.from(audioData, "base64");
+    return {
+      type: "buffer",
+      data: buffer,
+      mimeType: "audio/mpeg",
+    };
+  }
+
+  getConfiguredVoiceProviders(): { groq: boolean; elevenlabs: boolean; openai: boolean; gemini: boolean; qwen: boolean } {
+    const db = getDb();
+    const hasDbKey = (providerId: string): boolean => {
+      const row = db.query(
+        `SELECT api_key_encrypted FROM providers WHERE id = ? AND api_key_encrypted IS NOT NULL AND api_key_encrypted != ''`
+      ).get(providerId) as { api_key_encrypted: string } | undefined;
+      return !!row;
+    };
+
+    return {
+      groq:       hasDbKey("groq")       || !!(process.env.GROQ_API_KEY),
+      elevenlabs: hasDbKey("elevenlabs") || !!(process.env.ELEVENLABS_API_KEY),
+      openai:     hasDbKey("openai")     || !!(process.env.OPENAI_API_KEY),
+      gemini:     hasDbKey("gemini")     || !!(process.env.GEMINI_API_KEY),
+      qwen:       hasDbKey("qwen")       || !!(process.env.DASHSCOPE_API_KEY),
+    };
+  }
+
+  getOpenAIVoices(): Array<{ id: string; name: string }> {
+    return [
+      { id: "alloy", name: "Alloy" },
+      { id: "echo", name: "Echo" },
+      { id: "fable", name: "Fable" },
+      { id: "onyx", name: "Onyx" },
+      { id: "nova", name: "Nova" },
+      { id: "shimmer", name: "Shimmer" },
+      { id: "ash", name: "Ash" },
+      { id: "ballad", name: "Ballad" },
+      { id: "coral", name: "Coral" },
+      { id: "sage", name: "Sage" },
+      { id: "verse", name: "Verse" },
+    ];
+  }
+
+  getGeminiVoices(): Array<{ id: string; name: string }> {
+    return [
+      { id: "Puck", name: "Puck" },
+      { id: "Charon", name: "Charon" },
+      { id: "Kore", name: "Kore" },
+      { id: "Fenrir", name: "Fenrir" },
+      { id: "Aoede", name: "Aoede" },
+      { id: "Orbit", name: "Orbit" },
+      { id: "Zephyr", name: "Zephyr" },
+      { id: "Autonoe", name: "Autonoe" },
+      { id: "Enceladus", name: "Enceladus" },
+      { id: "Iapetus", name: "Iapetus" },
+      { id: "Umbriel", name: "Umbriel" },
+      { id: "Algieba", name: "Algieba" },
+      { id: "Despina", name: "Despina" },
+      { id: "Erinome", name: "Erinome" },
+      { id: "Laomedeia", name: "Laomedeia" },
+      { id: "Achernar", name: "Achernar" },
+      { id: "Rasalgethi", name: "Rasalgethi" },
+      { id: "Schedar", name: "Schedar" },
+      { id: "Sulafat", name: "Sulafat" },
+      { id: "Vindemiatrix", name: "Vindemiatrix" },
+      { id: "Zubenelgenubi", name: "Zubenelgenubi" },
+      { id: "Pulcherrima", name: "Pulcherrima" },
+      { id: "Achird", name: "Achird" },
+      { id: "Zubeneschamali", name: "Zubeneschamali" },
+      { id: "Sadachbia", name: "Sadachbia" },
+      { id: "Sadaltager", name: "Sadaltager" },
+      { id: "Sheratan", name: "Sheratan" },
+    ];
+  }
+
+  getQwenVoices(): Array<{ id: string; name: string }> {
+    return [
+      { id: "ruoxi", name: "Ruoxi (F, Chinese)" },
+      { id: "longhua", name: "Longhua (M, Chinese)" },
+      { id: "lingli", name: "Lingli (F, Chinese)" },
+      { id: "zhiyan", name: "Zhiyan (F, Chinese)" },
+      { id: "aicheng", name: "Aicheng (F, Chinese)" },
+      { id: "aida", name: "Aida (F, Chinese)" },
+      { id: "yucheng", name: "Yucheng (M, Chinese)" },
+      { id: "yijia", name: "Yijia (F, Chinese)" },
+      { id: "yinan", name: "Yinan (M, Chinese)" },
+      { id: "sijia", name: "Sijia (F, Chinese)" },
+      { id: "sicheng", name: "Sicheng (M, Chinese)" },
+      { id: "siqi", name: "Siqi (F, Chinese)" },
+      { id: "aixia", name: "Aixia (F, Chinese)" },
+    ];
+  }
+
+  async getElevenLabsVoices(): Promise<Array<{ id: string; name: string; category: string }>> {
+    const apiKey = await this.getProviderApiKey("elevenlabs");
+    const key = apiKey || process.env.ELEVENLABS_API_KEY;
+    
+    if (!key) {
+      throw new Error("ELEVENLABS_API_KEY not configured");
+    }
+
+    const response = await fetch("https://api.elevenlabs.io/v1/voices", {
+      headers: {
+        "xi-api-key": key,
+      },
+    });
+
+    if (!response.ok) {
+      const error = await response.text();
+      throw new Error(`Failed to fetch ElevenLabs voices: ${error}`);
+    }
+
+    const data = await response.json() as { voices: Array<{ voice_id: string; name: string; category: string }> };
+    return data.voices.map(v => ({
+      id: v.voice_id,
+      name: v.name,
+      category: v.category,
+    }));
+  }
+
+  normalizeAudioFromChannel(channelType: string, audioData: unknown): AudioInput {
+    switch (channelType) {
+      case "telegram":
+        return this.normalizeTelegramAudio(audioData);
+      case "discord":
+        return this.normalizeDiscordAudio(audioData);
+      case "whatsapp":
+        return this.normalizeWhatsAppAudio(audioData);
+      case "slack":
+        return this.normalizeSlackAudio(audioData);
+      case "webchat":
+        return this.normalizeWebChatAudio(audioData);
+      default:
+        throw new Error(`Unknown channel type: ${channelType}`);
+    }
+  }
+
+  private normalizeTelegramAudio(audioData: unknown): AudioInput {
+    const data = audioData as { fileId?: string; buffer?: Buffer; url?: string };
+    
+    if (data.buffer) {
+      return { type: "buffer", data: data.buffer, mimeType: "audio/ogg" };
+    }
+    if (data.url) {
+      return { type: "url", data: data.url, mimeType: "audio/ogg" };
+    }
+    throw new Error("Telegram audio missing buffer or URL");
+  }
+
+  private normalizeDiscordAudio(audioData: unknown): AudioInput {
+    const data = audioData as { buffer?: Buffer; url?: string; mimeType?: string };
+    
+    if (data.buffer) {
+      return { type: "buffer", data: data.buffer, mimeType: data.mimeType || "audio/webm" };
+    }
+    if (data.url) {
+      return { type: "url", data: data.url, mimeType: data.mimeType || "audio/webm" };
+    }
+    throw new Error("Discord audio missing buffer or URL");
+  }
+
+  private normalizeWhatsAppAudio(audioData: unknown): AudioInput {
+    const data = audioData as { buffer?: Buffer; url?: string; base64?: string };
+
+    if (data.buffer) {
+      return { type: "buffer", data: data.buffer, mimeType: "audio/ogg" };
+    }
+    if (data.base64) {
+      return { type: "base64", data: data.base64, mimeType: "audio/ogg" };
+    }
+    if (data.url) {
+      return { type: "url", data: data.url, mimeType: "audio/ogg" };
+    }
+    throw new Error("WhatsApp audio: buffer not available — download may have failed");
+  }
+
+  private normalizeSlackAudio(audioData: unknown): AudioInput {
+    const data = audioData as { buffer?: Buffer; url?: string; mimeType?: string };
+    
+    if (data.buffer) {
+      return { type: "buffer", data: data.buffer, mimeType: data.mimeType || "audio/webm" };
+    }
+    if (data.url) {
+      return { type: "url", data: data.url, mimeType: data.mimeType || "audio/webm" };
+    }
+    throw new Error("Slack audio missing buffer or URL");
+  }
+
+  private normalizeWebChatAudio(audioData: unknown): AudioInput {
+    const data = audioData as { base64?: string; buffer?: Buffer };
+    
+    if (data.base64) {
+      return { type: "base64", data: data.base64, mimeType: "audio/webm" };
+    }
+    if (data.buffer) {
+      return { type: "buffer", data: data.buffer, mimeType: "audio/webm" };
+    }
+    throw new Error("WebChat audio missing base64 or buffer");
+  }
+}
+
+export const voiceService = VoiceService.getInstance();