Update app.py

app.py

@@ -1,30 +1,27 @@
 import os, asyncio, hashlib, json, httpx, uuid, time, uvicorn, random, shutil, tempfile
 from contextlib import asynccontextmanager
 from fastapi import FastAPI, Request, UploadFile, File
-from fastapi.responses import JSONResponse, StreamingResponse
+from fastapi.responses import JSONResponse, StreamingResponse, HTMLResponse
 from fastapi.middleware.cors import CORSMiddleware
 from faster_whisper import WhisperModel
 from gtts import gTTS
 
-# ====================================
-# CONFIGURASI MODEL (Gampang di-Manage)
-# ====================================
+# --- CONFIG ---
 MODEL_STRATEGY = {
     "deepseek": [
         {"provider": "ollama", "model": "deepseek-r1:7b"},
-        {"provider": "openrouter", "model": "deepseek/deepseek-r1"},
-        {"provider": "cloudflare", "model": "@cf/meta/llama-3-8b-instruct"}
+        {"provider": "openrouter", "model": "deepseek/deepseek-r1"}
     ],
-    "gpt": [{"provider": "openrouter", "model": "openai/gpt-4o-mini"}]
+    "gpt": [{"provider": "openrouter", "model": "openai/gpt-4o-mini"}],
+    "gemini": [{"provider": "openrouter", "model": "google/gemini-2.0-flash"}]
 }
-
-TIMEOUT = 120
+TIMEOUT = 180
 QUEUE = asyncio.Queue()
+
+# Load Whisper
 whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8", download_root="./models_cache")
 
-# ====================================
-# CORE AI ROUTER
-# ====================================
+# --- PROVIDERS ---
 async def call_ollama(messages, model):
     try:
         async with httpx.AsyncClient(timeout=TIMEOUT) as client:
@@ -35,7 +32,7 @@ async def call_ollama(messages, model):
 
 async def call_openrouter(messages, model):
     keys = os.getenv("OPENROUTER_KEYS", "").split(",")
-    if not keys[0]: return None
+    if not keys or not keys[0]: return None
     try:
         async with httpx.AsyncClient(timeout=TIMEOUT) as client:
             res = await client.post("https://openrouter.ai/api/v1/chat/completions",
@@ -48,17 +45,10 @@ async def smart_router(messages, alias):
     strategies = MODEL_STRATEGY.get(alias, MODEL_STRATEGY["deepseek"])
     for attempt in strategies:
         prov, mod = attempt["provider"], attempt["model"]
-        print(f"[*] Mencoba {prov} ({mod})...")
-        res = None
-        if prov == "ollama": res = await call_ollama(messages, mod)
-        elif prov == "openrouter": res = await call_openrouter(messages, mod)
-        
+        res = await call_ollama(messages, mod) if prov == "ollama" else await call_openrouter(messages, mod)
         if res: return res
     return None
 
-# ====================================
-# WORKER & FASTAPI
-# ====================================
 async def worker():
     while True:
         messages, alias, future = await QUEUE.get()
@@ -70,30 +60,46 @@ async def worker():
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    # Jalankan 4 worker paralel
     workers = [asyncio.create_task(worker()) for _ in range(4)]
     yield
     for w in workers: w.cancel()
 
-app = FastAPI(lifespan=lifespan)
+app = FastAPI(title="AI Gateway", lifespan=lifespan)
 app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"], allow_headers=["*"])
 
-@app.get("/")
-async def root():
-    return {"status": "AI Gateway Online", "engine": "Ollama + Fallback API"}
+# --- UI & ENDPOINTS ---
+@app.get("/", response_class=HTMLResponse)
+async def home():
+    return """
+    <body style="background:#0f172a; color:white; font-family:sans-serif; display:flex; flex-direction:column; align-items:center; justify-content:center; height:100vh;">
+        <h1 style="color:#38bdf8">AI Gateway Running</h1>
+        <p style="color:#94a3b8">OpenAI-Compatible API for YT-Short-Clipper</p>
+        <div style="background:#1e293b; padding:20px; border-radius:10px; border:1px solid #334155">
+            <code>Endpoint: /v1/chat/completions</code><br>
+            <code>Model: deepseek, gpt, gemini</code>
+        </div>
+    </body>
+    """
 
 @app.post("/v1/chat/completions")
 async def chat(request: Request):
     body = await request.json()
+    alias = body.get("model", "deepseek")
     future = asyncio.get_event_loop().create_future()
-    await QUEUE.put((body.get("messages", []), body.get("model", "deepseek"), future))
+    await QUEUE.put((body.get("messages", []), alias, future))
     result = await future
-    if not result: return JSONResponse({"error": "Semua model gagal"}, status_code=500)
-    return {"id": str(uuid.uuid4()), "choices": [{"message": {"role": "assistant", "content": result}}]}
+    if not result: return JSONResponse({"error": "Failed"}, status_code=500)
+    return {
+        "id": f"chatcmpl-{uuid.uuid4()}",
+        "object": "chat.completion",
+        "created": int(time.time()),
+        "model": alias,
+        "choices": [{"message": {"role": "assistant", "content": result}, "finish_reason": "stop"}]
+    }
 
 @app.post("/v1/audio/transcriptions")
 async def transcribe(file: UploadFile = File(...)):
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".tmp") as tmp:
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
         shutil.copyfileobj(file.file, tmp)
         path = tmp.name
     try:
@@ -102,5 +108,15 @@ async def transcribe(file: UploadFile = File(...)):
     finally:
         if os.path.exists(path): os.remove(path)
 
+@app.post("/v1/audio/speech")
+async def tts(request: Request):
+    body = await request.json()
+    tmp_path = f"/tmp/{uuid.uuid4()}.mp3"
+    gTTS(text=body.get("input", ""), lang=body.get("voice", "id")).save(tmp_path)
+    def iter():
+        with open(tmp_path, "rb") as f: yield from f
+        os.remove(tmp_path)
+    return StreamingResponse(iter(), media_type="audio/mpeg")
+
 if __name__ == "__main__":
     uvicorn.run(app, host="0.0.0.0", port=7860)