initial commit

app.py

@@ -1,189 +1,480 @@
-# ==========================================================
-#  KC ROBOT AI - APP.PY (V2.0 MAX FINAL)
-#  Cloud AI Robot with Gemini 2.5 Flash + ESP32 + Telegram
-# ==========================================================
-
-from flask import Flask, request, jsonify, render_template_string
-from google import genai
-import requests
+"""
+RobotAI v9.8 — Gemini Brain + gTTS + Telegram (Hugging Face ready)
+
+Features:
+- Keep original Gemini AI logic (uses google-generativeai)
+- Web UI: auto language detect (vi/en), clean punctuation for TTS
+- Only play the newest audio on the web (stop previous audio)
+- Telegram integration (polling): replies with text and sends audio file if possible
+- Audio caching and cleanup
+"""
+
 import os
+import json
+import uuid
+import re
 import time
+import logging
+import threading
+from flask import Flask, request, jsonify, render_template_string, send_file, redirect, url_for
+
+# Gemini SDK
+import google.generativeai as genai
+
+# gTTS for TTS (female-like voice)
 from gtts import gTTS
-from langdetect import detect
-import tempfile
-import base64
 
-# ==========================================================
-# CONFIGURATION
-# ==========================================================
+# Telegram support (try modern v20, fallback v13)
+try:
+    from telegram import Bot
+    from telegram.ext import ApplicationBuilder, MessageHandler, CommandHandler, filters
+    TELEGRAM_LIB = "v20"
+except Exception:
+    try:
+        from telegram import Bot
+        from telegram.ext import Updater, MessageHandler, Filters, CommandHandler
+        TELEGRAM_LIB = "v13"
+    except Exception:
+        TELEGRAM_LIB = None
 
-# Load environment variables from secrets (Cloud Run or Hugging Face)
-GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
-GEMINI_MODEL = os.getenv("GEMINI_MODEL", "gemini-2.5-flash")
-TELEGRAM_TOKEN = os.getenv("TELEGRAM_TOKEN")
-TELEGRAM_CHAT_ID = os.getenv("TELEGRAM_CHAT_ID")
+# ---------------- Config ----------------
+CONFIG_FILE = "config.json"
+AUDIO_DIR = "audio_cache"
+os.makedirs(AUDIO_DIR, exist_ok=True)
 
-# Create Flask app
 app = Flask(__name__)
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger("RobotAI")
+
+DEFAULT_MODEL = "gemini-2.5-flash"
+USE_GEMINI = False
+
+# ----------------- config helpers -----------------
+def load_config():
+    cfg = {
+        "GEMINI_API_KEY": os.environ.get("GEMINI_API_KEY", ""),
+        "GEMINI_MODEL": os.environ.get("GEMINI_MODEL", DEFAULT_MODEL),
+        "TELEGRAM_TOKEN": os.environ.get("TELEGRAM_TOKEN", ""),
+        "TELEGRAM_CHAT_ID": os.environ.get("TELEGRAM_CHAT_ID", "")
+    }
+    if os.path.exists(CONFIG_FILE):
+        try:
+            with open(CONFIG_FILE, "r", encoding="utf-8") as f:
+                data = json.load(f)
+                cfg.update(data)
+        except Exception as e:
+            logger.exception("Load config error")
+    return cfg
 
-# ==========================================================
-# SETUP GEMINI CLIENT
-# ==========================================================
-if not GEMINI_API_KEY:
-    print("❌ ERROR: No Gemini API Key found. Please add GEMINI_API_KEY in Secrets.")
-    client = None
-else:
-    client = genai.Client(api_key=GEMINI_API_KEY)
-
-# ==========================================================
-# TELEGRAM UTILITIES
-# ==========================================================
-def send_telegram_message(text):
-    if not TELEGRAM_TOKEN or not TELEGRAM_CHAT_ID:
-        print("⚠️ Telegram not configured.")
+def save_config(cfg):
+    try:
+        with open(CONFIG_FILE, "w", encoding="utf-8") as f:
+            json.dump(cfg, f, ensure_ascii=False, indent=2)
+        return True
+    except Exception:
+        logger.exception("Save config failed")
+        return False
+
+# ----------------- Gemini init + wrapper -----------------
+def init_gemini():
+    global USE_GEMINI
+    cfg = load_config()
+    key = cfg.get("GEMINI_API_KEY") or ""
+    if not key:
+        logger.warning("Gemini API key missing.")
+        USE_GEMINI = False
         return
-    url = f"https://api.telegram.org/bot{TELEGRAM_TOKEN}/sendMessage"
-    payload = {"chat_id": TELEGRAM_CHAT_ID, "text": text}
     try:
-        requests.post(url, json=payload, timeout=5)
-    except Exception as e:
-        print("Telegram Error:", e)
+        genai.configure(api_key=key)
+        USE_GEMINI = True
+        logger.info("✅ Gemini connected OK.")
+    except Exception:
+        USE_GEMINI = False
+        logger.exception("Gemini init error")
 
-# ==========================================================
-# GEMINI AI RESPONSE
-# ==========================================================
-def ask_gemini(prompt: str):
-    if not client:
-        return "⚠️ Gemini API key missing. Please configure in Secrets."
+init_gemini()
 
+def gemini_answer(prompt: str) -> str:
+    """
+    Call Gemini. Keep simple: use GenerativeModel if present (matches your v9.2).
+    """
+    cfg = load_config()
+    model = cfg.get("GEMINI_MODEL", DEFAULT_MODEL)
     try:
-        response = client.models.generate_content(
-            model=GEMINI_MODEL,
-            contents=prompt
-        )
-        if hasattr(response, "text"):
-            return response.text.strip()
-        elif "text" in response:
-            return response["text"].strip()
-        else:
-            return "⚠️ No response text from Gemini."
-    except Exception as e:
-        print("Gemini Error:", e)
-        return f"⚠️ Gemini Error: {e}"
-
-# ==========================================================
-# LANGUAGE DETECTION & TTS
-# ==========================================================
-def text_to_speech(text):
+        if hasattr(genai, "GenerativeModel"):
+            try:
+                m = genai.GenerativeModel(model)
+                resp = m.generate_content(prompt)
+                if hasattr(resp, "text"):
+                    return resp.text
+                return str(resp)
+            except Exception:
+                logger.debug("GenerativeModel failed, trying responses.generate", exc_info=True)
+        # fallback: try responses.create (newer)
+        if hasattr(genai, "responses") and hasattr(genai.responses, "create"):
+            r = genai.responses.create(model=model, input=prompt)
+            if hasattr(r, "output_text"):
+                return r.output_text
+            return str(r)
+    except Exception:
+        logger.exception("Gemini call error")
+    return "⚠️ Gemini không phản hồi — kiểm tra API key / library."
+
+# ----------------- Language detection & TTS cleaning -----------------
+VIET_CHARS = "ăâđêôơưáàảãạắằẳẵặấầẩẫậéèẻẽẹíìỉĩịóòỏõọốồổỗộớờởỡợúùủũụưứừửữựýỳỷỹỵ"
+
+def detect_lang(text: str) -> str:
+    return "vi" if any(ch in VIET_CHARS for ch in text.lower()) else "en"
+
+# Clean punctuation but keep letters & numbers and basic punctuation replacement -> single spaces.
+# We must not return empty string; fallback to short phrase.
+def clean_text_for_tts(text: str) -> str:
+    if not text:
+        return "Xin chào"
+    # replace typical punctuation with space
+    cleaned = re.sub(r"[.,!?;:()\"'“”‘’\[\]{}<>\/\\\|@#\$%\^&\*\+=~`–—\-]", " ", text)
+    cleaned = re.sub(r"\s+", " ", cleaned).strip()
+    if not cleaned:
+        return "Xin chào"
+    return cleaned
+
+def speak_text(text: str, lang: str = "vi") -> str:
+    """
+    Generate mp3 via gTTS, return local path or None.
+    """
     try:
-        lang = detect(text)
-        if lang not in ["vi", "en"]:
-            lang = "en"
-        tts = gTTS(text=text, lang=lang)
-        tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
-        tts.save(tmp.name)
-        with open(tmp.name, "rb") as f:
-            audio_b64 = base64.b64encode(f.read()).decode("utf-8")
-        os.unlink(tmp.name)
-        return audio_b64
-    except Exception as e:
-        print("TTS Error:", e)
+        tts_text = clean_text_for_tts(text)
+        fname = f"tts_{uuid.uuid4().hex[:8]}.mp3"
+        path = os.path.join(AUDIO_DIR, fname)
+        tts = gTTS(text=tts_text, lang=lang, slow=False)
+        tts.save(path)
+        logger.info("gTTS saved %s", path)
+        return path
+    except Exception:
+        logger.exception("TTS generation failed")
         return None
 
-# ==========================================================
-# SIMPLE HTML INTERFACE (for testing)
-# ==========================================================
-HTML_PAGE = """
-<!DOCTYPE html>
-<html>
-<head>
-<title>KC Robot AI v2.0</title>
+# ----------------- Audio cleanup -----------------
+def cleanup_audio_older_than(seconds: int = 3600):
+    now = time.time()
+    for f in os.listdir(AUDIO_DIR):
+        p = os.path.join(AUDIO_DIR, f)
+        try:
+            if os.path.isfile(p) and (now - os.path.getmtime(p) > seconds):
+                os.remove(p)
+                logger.info("Removed old audio: %s", p)
+        except Exception:
+            logger.exception("cleanup failed for %s", p)
+
+# ----------------- Web UI templates (client-side ensures only newest audio plays) -----------------
+INDEX_HTML = """<!doctype html><html><head>
+<meta charset="utf-8"><title>RobotAI v9.8</title>
 <style>
-body { font-family: Arial; text-align: center; background-color: #101010; color: white; }
-input, button { padding: 10px; font-size: 16px; margin: 5px; }
-#chat { max-width: 700px; margin: auto; text-align: left; background: #202020; padding: 20px; border-radius: 10px; }
-.msg-user { color: #4af; }
-.msg-bot { color: #fa4; margin-left: 20px; }
-audio { margin-top: 10px; }
+body{font-family:Arial;background:#f5faff;padding:12px}
+#chat{background:#fff;border:1px solid #ddd;padding:12px;min-height:260px;border-radius:8px;overflow:auto}
+.you{color:#0b66c3;margin:6px 0}
+.bot{color:#0b8a5f;margin:6px 0}
+.button{background:#0b66c3;color:white;border:none;padding:8px 12px;border-radius:6px;cursor:pointer}
+.audio-controls{margin-top:6px}
 </style>
-</head>
-<body>
-<h1>🤖 KC Robot AI v2.0 MAX FINAL</h1>
-<div id="chat"></div>
-<br>
-<input id="user_input" placeholder="Nói gì đó..." style="width:60%">
-<button onclick="sendMessage()">Gửi</button>
+</head><body>
+<h2>🤖 RobotAI v9.8 — Gemini Brain + Voice + Telegram</h2>
+<div>Gemini: <b>{{ gemini_status }}</b> | Model: <b>{{ model }}</b> | <a href="/config">Config</a></div>
+<hr>
+<textarea id="text" rows="4" style="width:100%;padding:8px;border-radius:6px;border:1px solid #ccc"></textarea><br><br>
+<button class="button" onclick="send()">Gửi</button> <button class="button" onclick="clearChat()">Xóa</button>
+<div id="chat" style="margin-top:12px"></div>
 
 <script>
-async function sendMessage() {
-  const input = document.getElementById("user_input").value;
-  if (!input) return;
-  const chat = document.getElementById("chat");
-  chat.innerHTML += `<div class='msg-user'><b>Bạn:</b> ${input}</div>`;
-  document.getElementById("user_input").value = "";
-  const res = await fetch("/api/chat", {
-    method: "POST",
-    headers: {"Content-Type": "application/json"},
-    body: JSON.stringify({message: input})
-  });
-  const data = await res.json();
-  chat.innerHTML += `<div class='msg-bot'><b>Robot:</b> ${data.reply}</div>`;
-  if (data.audio) {
-    const audio = document.createElement("audio");
-    audio.src = "data:audio/mp3;base64," + data.audio;
-    audio.controls = true;
-    chat.appendChild(audio);
+let currentAudio = null;
+
+function append(cls, txt){
+  const c = document.getElementById('chat');
+  c.innerHTML += '<div class="'+cls+'">'+txt+'</div>';
+  c.scrollTop = c.scrollHeight;
+}
+
+function stopCurrentAudio(){
+  if(currentAudio){
+    try{ currentAudio.pause(); currentAudio.currentTime = 0; }catch(e){}
+    currentAudio = null;
+  }
+}
+
+function clearChat(){
+  document.getElementById('chat').innerHTML = '';
+  stopCurrentAudio();
+}
+
+async function send(){
+  let txt = document.getElementById('text').value.trim();
+  if(!txt) return;
+  append('you', 'Bạn: ' + txt);
+  document.getElementById('text').value = '';
+
+  // stop previous audio (so we only play the newest)
+  stopCurrentAudio();
+
+  try{
+    const res = await fetch('/api/chat', {
+      method: 'POST',
+      headers: {'Content-Type':'application/json'},
+      body: JSON.stringify({text: txt})
+    });
+    const j = await res.json();
+    append('bot', '🤖: ' + (j.reply || '(no reply)'));
+    if(j.tts_url){
+      currentAudio = new Audio(j.tts_url);
+      currentAudio.autoplay = true;
+      currentAudio.controls = false;
+      // optional: show an audio control for manual replay
+      const audioEl = document.createElement('audio');
+      audioEl.src = j.tts_url;
+      audioEl.controls = true;
+      audioEl.className = 'audio-controls';
+      document.getElementById('chat').appendChild(audioEl);
+      // play newest
+      try{ currentAudio.play(); }catch(e){}
+    }
+  }catch(e){
+    append('bot', '[Lỗi mạng] ' + e);
   }
-  chat.scrollTop = chat.scrollHeight;
 }
 </script>
-</body>
-</html>
+</body></html>
+"""
+
+CONFIG_HTML = """<!doctype html><html><head><meta charset="utf-8"><title>Config</title></head>
+<body style="font-family:Arial;padding:12px">
+<h3>⚙️ Config RobotAI</h3>
+<form method="post" action="/config">
+Gemini API Key:<br><textarea name="GEMINI_API_KEY" rows="2" cols="80">{{ GEMINI_API_KEY }}</textarea><br><br>
+Gemini Model:<br><input name="GEMINI_MODEL" value="{{ GEMINI_MODEL }}" size="50"><br><br>
+Telegram Token:<br><input name="TELEGRAM_TOKEN" value="{{ TELEGRAM_TOKEN }}" size="60"><br><br>
+Telegram Chat ID (optional):<br><input name="TELEGRAM_CHAT_ID" value="{{ TELEGRAM_CHAT_ID }}" size="30"><br><br>
+<button type="submit">Lưu</button>
+</form>
+<p><a href="/">⬅ Trở về</a></p>
+</body></html>
 """
 
+# ----------------- Flask routes -----------------
 @app.route("/")
 def home():
-    return render_template_string(HTML_PAGE)
+    cfg = load_config()
+    return render_template_string(INDEX_HTML,
+                                  gemini_status="✅ Kết nối" if USE_GEMINI else "❌ Chưa kết nối",
+                                  model=cfg.get("GEMINI_MODEL"))
 
-# ==========================================================
-# API ENDPOINTS
-# ==========================================================
+@app.route("/config", methods=["GET","POST"])
+def config_page():
+    if request.method == "POST":
+        data = {
+            "GEMINI_API_KEY": request.form.get("GEMINI_API_KEY","").strip(),
+            "GEMINI_MODEL": request.form.get("GEMINI_MODEL", DEFAULT_MODEL).strip(),
+            "TELEGRAM_TOKEN": request.form.get("TELEGRAM_TOKEN","").strip(),
+            "TELEGRAM_CHAT_ID": request.form.get("TELEGRAM_CHAT_ID","").strip()
+        }
+        save_config(data)
+        init_gemini()
+        # restart telegram thread if token provided
+        try:
+            start_telegram_bot_thread()
+        except Exception:
+            logger.exception("start telegram thread failed")
+        return redirect(url_for("config_page"))
+    return render_template_string(CONFIG_HTML, **load_config())
 
 @app.route("/api/chat", methods=["POST"])
 def api_chat():
-    data = request.get_json()
-    if not data or "message" not in data:
-        return jsonify({"error": "Missing 'message'"}), 400
-
-    user_message = data["message"]
-    print(f"🧠 User said: {user_message}")
-    send_telegram_message(f"User: {user_message}")
-
-    ai_reply = ask_gemini(user_message)
-    send_telegram_message(f"Robot: {ai_reply}")
-
-    audio_b64 = text_to_speech(ai_reply)
-    return jsonify({"reply": ai_reply, "audio": audio_b64})
-
-# ESP32 sensor endpoint
-@app.route("/api/sensor", methods=["POST"])
-def sensor_data():
-    data = request.get_json()
-    if not data:
-        return jsonify({"error": "No data"}), 400
-    msg = f"👁️ ESP32 Sensor update: {data}"
-    send_telegram_message(msg)
-    return jsonify({"status": "received"})
-
-# Health check
-@app.route("/ping")
-def ping():
-    return jsonify({"status": "ok", "model": GEMINI_MODEL})
-
-# ==========================================================
-# MAIN ENTRY POINT
-# ==========================================================
+    payload = request.get_json(force=True)
+    text = (payload.get("text") or "").strip()
+    if not text:
+        return jsonify({"error":"empty"}), 400
+
+    lang = detect_lang(text)
+    try:
+        if USE_GEMINI:
+            prefix = "Trả lời bằng tiếng Việt:" if lang == "vi" else "Answer in English:"
+            reply = gemini_answer(prefix + "\n" + text)
+        else:
+            reply = "⚠️ Chưa kết nối Gemini. Kiểm tra API Key."
+    except Exception:
+        logger.exception("gemini call")
+        reply = "⚠️ Lỗi khi gọi Gemini."
+
+    tts_path = None
+    try:
+        if reply:
+            tts_path = speak_text(reply, lang)
+    except Exception:
+        logger.exception("tts generation failed")
+
+    tts_url = f"/api/tts/{os.path.basename(tts_path)}" if tts_path else None
+
+    # send to telegram (background) if configured
+    try:
+        cfg = load_config()
+        token = cfg.get("TELEGRAM_TOKEN","")
+        chat_id = cfg.get("TELEGRAM_CHAT_ID","")
+        if token and chat_id:
+            threading.Thread(target=send_to_telegram_sync, args=(token, chat_id, reply, tts_path), daemon=True).start()
+    except Exception:
+        logger.exception("telegram dispatch failed")
+
+    # cleanup audio in background
+    try:
+        threading.Thread(target=cleanup_audio_older_than, kwargs={"seconds":3600}, daemon=True).start()
+    except Exception:
+        pass
+
+    return jsonify({"reply": reply, "tts_url": tts_url})
+
+@app.route("/api/tts/<fname>")
+def get_tts(fname):
+    path = os.path.join(AUDIO_DIR, fname)
+    if not os.path.exists(path):
+        return jsonify({"error":"not found"}), 404
+    return send_file(path, mimetype="audio/mpeg")
+
+# ----------------- Telegram integration -----------------
+TG_THREAD = None
+
+def send_to_telegram_sync(token: str, chat_id: str, text: str, tts_path: str = None):
+    """
+    Send text and (optionally) audio to Telegram chat.
+    """
+    try:
+        bot = Bot(token=token)
+        try:
+            bot.send_message(chat_id=chat_id, text=text)
+        except Exception:
+            logger.exception("telegram send_message failed")
+        if tts_path and os.path.exists(tts_path):
+            try:
+                with open(tts_path, "rb") as fh:
+                    bot.send_audio(chat_id=chat_id, audio=fh)
+            except Exception:
+                logger.exception("telegram send_audio failed")
+    except Exception:
+        logger.exception("telegram overall send failed")
+
+def start_telegram_bot_thread():
+    global TG_THREAD
+    cfg = load_config()
+    token = cfg.get("TELEGRAM_TOKEN","")
+    if not token:
+        logger.info("Telegram token not configured; skipping")
+        return
+
+    if TG_THREAD and TG_THREAD.is_alive():
+        logger.info("Telegram thread already running.")
+        return
+
+    def _runner_v20(token):
+        try:
+            app_builder = ApplicationBuilder().token(token).build()
+
+            async def start_cmd(update, context):
+                await update.message.reply_text("RobotAI is running. Send me text and I'll reply with Gemini + TTS.")
+
+            async def handle_msg(update, context):
+                text = update.message.text or ""
+                if not text.strip():
+                    await update.message.reply_text("Không nhận được nội dung.")
+                    return
+                lang = detect_lang(text)
+                if USE_GEMINI:
+                    prefix = "Trả lời bằng tiếng Việt:" if lang == "vi" else "Answer in English:"
+                    reply = gemini_answer(prefix + "\n" + text)
+                else:
+                    reply = "⚠️ Chưa kết nối Gemini."
+
+                # generate tts using same cleaning logic
+                tts_path = None
+                try:
+                    if reply:
+                        tts_path = speak_text(reply, lang)
+                except Exception:
+                    logger.exception("tts for telegram failed")
+
+                await update.message.reply_text(reply)
+                if tts_path and os.path.exists(tts_path):
+                    try:
+                        with open(tts_path, "rb") as fh:
+                            await context.bot.send_audio(chat_id=update.effective_chat.id, audio=fh)
+                    except Exception:
+                        logger.exception("send audio to telegram (v20) failed")
+
+            app_builder.add_handler(CommandHandler("start", start_cmd))
+            app_builder.add_handler(MessageHandler(filters.TEXT & (~filters.COMMAND), handle_msg))
+            logger.info("Starting Telegram polling (v20)...")
+            app_builder.run_polling()
+        except Exception:
+            logger.exception("telegram v20 runner exception")
+
+    def _runner_v13(token):
+        try:
+            updater = Updater(token=token, use_context=True)
+            dp = updater.dispatcher
+
+            def start_cmd(update, context):
+                update.message.reply_text("RobotAI is running. Send me text and I'll reply with Gemini + TTS.")
+
+            def handle_msg(update, context):
+                text = update.message.text or ""
+                if not text.strip():
+                    update.message.reply_text("Không nhận được nội dung.")
+                    return
+                lang = detect_lang(text)
+                if USE_GEMINI:
+                    prefix = "Trả lời bằng tiếng Việt:" if lang == "vi" else "Answer in English:"
+                    reply = gemini_answer(prefix + "\n" + text)
+                else:
+                    reply = "⚠️ Chưa kết nối Gemini."
+
+                tts_path = None
+                try:
+                    if reply:
+                        tts_path = speak_text(reply, lang)
+                except Exception:
+                    logger.exception("tts for telegram failed")
+
+                update.message.reply_text(reply)
+                if tts_path and os.path.exists(tts_path):
+                    try:
+                        context.bot.send_audio(chat_id=update.effective_chat.id, audio=open(tts_path, "rb"))
+                    except Exception:
+                        logger.exception("send audio to telegram (v13) failed")
+
+            dp.add_handler(CommandHandler("start", start_cmd))
+            dp.add_handler(MessageHandler(Filters.text & (~Filters.command), handle_msg))
+            updater.start_polling()
+            updater.idle()
+        except Exception:
+            logger.exception("telegram v13 runner exception")
+
+    def runner():
+        if TELEGRAM_LIB == "v20":
+            _runner_v20(token)
+        elif TELEGRAM_LIB == "v13":
+            _runner_v13(token)
+        else:
+            logger.warning("python-telegram-bot not available; skipping telegram polling")
+
+    TG_THREAD = threading.Thread(target=runner, daemon=True)
+    TG_THREAD.start()
+    logger.info("Telegram thread launched (if configured).")
+
+# start telegram thread at startup if token present
+try:
+    start_telegram_bot_thread()
+except Exception:
+    logger.exception("Starting telegram thread failed")
+
+# ----------------- Run -----------------
 if __name__ == "__main__":
-    port = int(os.getenv("PORT", 8080))
-    print(f"🚀 KC Robot AI v2.0 running on port {port}")
+    port = int(os.environ.get("PORT", 7860))
+    logger.info("Starting RobotAI v9.8 on 0.0.0.0:%s", port)
     app.run(host="0.0.0.0", port=port)
+