Delete app.py with huggingface_hub

app.py

@@ -1,135 +0,0 @@
-import gradio as gr
-from fastapi import FastAPI, Request
-import uvicorn
-import base64
-import torch
-import os
-import tempfile
-import traceback
-import json
-import time
-
-# FORCE BUILD TRIGGER: 00:05:00 Jan 20 2026
-os.environ["COQUI_TOS_AGREED"] = "1"
-
-# Global models
-MODELS = {"stt": None, "translate": None, "tts": None, "tokenizer": None}
-
-def load_models():
-    global MODELS
-    if MODELS["stt"] is None:
-        print("🎙️ Loading Whisper large-v3...")
-        import whisper
-        MODELS["stt"] = whisper.load_model("large-v3")
-        if torch.cuda.is_available():
-            MODELS["stt"] = MODELS["stt"].half().to("cuda")
-    
-    if MODELS["translate"] is None:
-        print("🌍 Loading NLLB-200...")
-        from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-        MODELS["tokenizer"] = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M")
-        MODELS["translate"] = AutoModelForSeq2SeqLM.from_pretrained(
-            "facebook/nllb-200-distilled-600M",
-            torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
-        ).to("cuda" if torch.cuda.is_available() else "cpu")
-    
-    if MODELS["tts"] is None:
-        print("🔊 Loading XTTS-v2...")
-        from TTS.api import TTS
-        MODELS["tts"] = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(
-            "cuda" if torch.cuda.is_available() else "cpu"
-        )
-
-def process_logic(request):
-    """Core logic shared between Gradio and FastAPI"""
-    action = request.get("action")
-    print(f"⚙️ Logic Action: {action}")
-    
-    if action == "health":
-        return {"status": "ok", "gpu": torch.cuda.is_available(), "timestamp": time.time()}
-    
-    load_models()
-    
-    if action == "stt":
-        audio_b64 = request.get("file")
-        lang = request.get("lang")
-        audio_bytes = base64.b64decode(audio_b64)
-        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
-            f.write(audio_bytes)
-            temp_path = f.name
-        try:
-            result = MODELS["stt"].transcribe(temp_path, language=lang, fp16=torch.cuda.is_available())
-            return {"text": result["text"].strip()}
-        finally:
-            os.unlink(temp_path)
-    
-    elif action == "translate":
-        text = request.get("text")
-        target_lang = request.get("target_lang")
-        tgt_code = f"{target_lang}_Latn"
-        inputs = MODELS["tokenizer"](text, return_tensors="pt", padding=True).to(MODELS["translate"].device)
-        translated = MODELS["translate"].generate(**inputs, forced_bos_token_id=MODELS["tokenizer"].convert_tokens_to_ids(tgt_code), max_length=512)
-        result = MODELS["tokenizer"].decode(translated[0], skip_special_tokens=True)
-        return {"translated": result.strip()}
-    
-    elif action == "tts":
-        text = request.get("text")
-        lang = request.get("lang")
-        speaker_wav_b64 = request.get("speaker_wav")
-        speaker_wav_path = None
-        if speaker_wav_b64:
-            speaker_bytes = base64.b64decode(speaker_wav_b64)
-            with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
-                f.write(speaker_bytes)
-                speaker_wav_path = f.name
-        try:
-            with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as output_file:
-                output_path = output_file.name
-            MODELS["tts"].tts_to_file(text=text, language=lang, file_path=output_path, speaker_wav=speaker_wav_path)
-            with open(output_path, "rb") as f:
-                audio_b64 = base64.b64encode(f.read()).decode()
-            return {"audio": audio_b64}
-        finally:
-            if speaker_wav_path and os.path.exists(speaker_wav_path):
-                os.unlink(speaker_wav_path)
-            if os.path.exists(output_path):
-                os.unlink(output_path)
-                
-    return {"error": f"Unknown action: {action}"}
-
-# --- GRADIO SECTION ---
-def gradio_fn(request_json):
-    try:
-        req = json.loads(request_json)
-        res = process_logic(req)
-        return json.dumps(res)
-    except Exception as e:
-        return json.dumps({"error": str(e), "trace": traceback.format_exc()})
-
-demo = gr.Interface(
-    fn=gradio_fn,
-    inputs=gr.Textbox(label="JSON Request", lines=5),
-    outputs=gr.Textbox(label="JSON Response"),
-    title="🚀 Unified AI Engine"
-)
-
-# --- FASTAPI SECTION ---
-app = FastAPI()
-
-@app.post("/api/v1/process")
-async def api_v1_process(request: Request):
-    try:
-        body = await request.json()
-        return process_logic(body)
-    except Exception as e:
-        return {"error": str(e)}
-
-@app.get("/health")
-def health():
-    return {"status": "ok"}
-
-# Mount Gradio into FastAPI
-app = gr.mount_gradio_app(app, demo, path="/")
-
-if __name__ == "__main__":
-    uvicorn.run(app, host="0.0.0.0", port=7860)