Upload app.py with huggingface_hub

app.py

@@ -18,7 +18,7 @@ import logging
 from threading import Thread, Lock
 from huggingface_hub import snapshot_download
 
-# 🛡️ 1. SILENCE LOGS & WARNINGS (v101: Pure Power & Stability)
+# 🛡️ 1. SILENCE LOGS & WARNINGS (v102: Mission-Critical Silence & Stability)
 logging.getLogger("transformers").setLevel(logging.ERROR)
 logging.getLogger("TTS").setLevel(logging.ERROR)
 logging.getLogger("onnxruntime").setLevel(logging.ERROR)
@@ -81,8 +81,8 @@ except ImportError:
             if f is None: return lambda x: x
             return f
 
-# FORCE BUILD TRIGGER: 13:10:00 Jan 21 2026
-# v101: Docker SDK Transition. Absolute Port Isolation.
+# FORCE BUILD TRIGGER: 13:16:00 Jan 21 2026
+# v102: Gradio SDK Restoration. Synchronized Port Binding.
 
 os.environ["COQUI_TOS_AGREED"] = "1"
 MODELS = {"stt": None, "translate": None, "tts": None, "denoiser": None}
@@ -91,7 +91,7 @@ WARMUP_STATUS = {"complete": False, "in_progress": False}
 WARMUP_LOCK = Lock()
 
 def activate_gpu_models(action):
-    """v101: Mission-Critical GPU Mode"""
+    """v102: Mission-Critical GPU Mode"""
     global MODELS, WARMUP_STATUS
     local_only = WARMUP_STATUS["complete"]
     
@@ -100,7 +100,7 @@ def activate_gpu_models(action):
         try: stt_on_gpu = MODELS["stt"] is not None and MODELS["stt"].model.device == "cuda"
         except: pass
         if not stt_on_gpu:
-            print(f"🎙️ [v101] Activating Whisper (GPU)...")
+            print(f"🎙️ [v102] Activating Whisper (GPU)...")
             try:
                 if MODELS["stt"]: del MODELS["stt"]
                 gc.collect(); torch.cuda.empty_cache()
@@ -122,7 +122,7 @@ def activate_gpu_models(action):
             tts_on_gpu = "cuda" in curr
         except: pass
         if MODELS["tts"] is None or not tts_on_gpu:
-            print(f"🔊 [v101] Activating XTTS-v2 (GPU)...")
+            print(f"🔊 [v102] Activating XTTS-v2 (GPU)...")
             try:
                 if MODELS["tts"] is None:
                     MODELS["tts"] = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", gpu=True)
@@ -137,9 +137,9 @@ def activate_gpu_models(action):
     if MODELS["translate"] is None: MODELS["translate"] = "active"
 
 def release_gpu_models():
-    """v101: Clean Resident State"""
+    """v102: Clean Resident State"""
     global MODELS
-    print("🧹 [v101] Releasing GPU resources.")
+    print("🧹 [v102] Releasing GPU resources.")
     try:
         if MODELS["stt"] and MODELS["stt"].model.device == "cuda":
             del MODELS["stt"]
@@ -153,18 +153,18 @@ def release_gpu_models():
     if torch.cuda.is_available(): torch.cuda.empty_cache()
 
 def warmup_task():
-    """Silent Pre-loading (v101)"""
+    """Silent Pre-loading (v102)"""
     global WARMUP_STATUS
     with WARMUP_LOCK:
         if WARMUP_STATUS["complete"] or WARMUP_STATUS["in_progress"]: return
         WARMUP_STATUS["in_progress"] = True
-    print("\n🔥 --- V101: DOCKER DEPLOY WARMUP STARTED ---")
+    print("\n🔥 --- V102: GRADIO SYNC WARMUP STARTED ---")
     try:
         MODELS["stt"] = WhisperModel("large-v3", device="cpu", compute_type="int8")
         MODELS["tts"] = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", gpu=False)
         chatterbox_utils.warmup_chatterbox()
         WARMUP_STATUS["complete"] = True
-        print(f"✅ --- SYSTEM WARM: v101 --- \n")
+        print(f"✅ --- SYSTEM WARM: v102 --- \n")
     except Exception as e:
         print(f"❌ Warmup fail: {e}")
     finally: WARMUP_STATUS["in_progress"] = False
@@ -217,7 +217,7 @@ def _tts_logic(text, lang, speaker_wav_b64):
 def core_process(request_dict):
     action = request_dict.get("action")
     t1 = time.time()
-    print(f"--- [v101] 🚀 GPU SESSION: {action} ---")
+    print(f"--- [v102] 🚀 GPU SESSION: {action} ---")
     activate_gpu_models(action)
     try:
         if action == "stt": res = _stt_logic(request_dict)
@@ -230,13 +230,13 @@ def core_process(request_dict):
             res = {"text": stt_res.get("text"), "translated": translated, "audio": tts_res.get("audio")}
         else: res = {"error": f"Unknown action: {action}"}
     finally:
-        print(f"--- [v101] ✨ SUCCESS: {action} ({time.time()-t1:.2f}s) ---")
+        print(f"--- [v102] ✨ SUCCESS: {action} ({time.time()-t1:.2f}s) ---")
         release_gpu_models()
     return res
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    # DOCKER ENTRYPOINT TRIGGER
+    # GRADIO MANAGED LIFECYCLE
     Thread(target=warmup_task, daemon=True).start()
     yield
 
@@ -252,7 +252,7 @@ async def api_process(request: Request):
     except Exception as e: return {"error": str(e)}
 
 @app.get("/health")
-def health(): return {"status": "ok", "warm": WARMUP_STATUS["complete"], "v": "101"}
+def health(): return {"status": "ok", "warm": WARMUP_STATUS["complete"], "v": "102"}
 
 @app.post("/api/v1/clear_cache")
 async def clear_cache():
@@ -266,10 +266,18 @@ async def clear_cache():
         return {"status": "success"}
     except: return {"status": "error"}
 
-# GRADIO INTERFACE (v101)
-demo = gr.Interface(fn=lambda x: json.dumps(core_process(json.loads(x))), inputs="text", outputs="text")
+# 🚀 V102: SYNCHRONIZED ENTRY POINT
+# We mount everything into Gradio and let the SDK handle the binding.
+demo = gr.Interface(
+    fn=lambda x: json.dumps(core_process(json.loads(x))), 
+    inputs="text", 
+    outputs="text",
+    title="🚀 AI Engine v102"
+)
+
+# Use gr.mount_gradio_app to merge FastAPI and Gradio.
+# The Hugging Face Gradio SDK will automatically detect and launch the mounted app.
 app = gr.mount_gradio_app(app, demo, path="/")
 
-if __name__ == "__main__":
-    print("🚀 [v101] DOCKER SERVER STARTING ON 7860...")
-    uvicorn.run(app, host="0.0.0.0", port=7860, log_level="error", loop="asyncio")
+# No uvicorn.run here! The Hugging Face Gradio launcher handles it.
+# This prevents the "Address already in use" error.