Update app.py

app.py

@@ -1,46 +1,78 @@
-import gradio as gr
-from huggingface_hub import InferenceClient
 import os
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+from huggingface_hub import InferenceClient
 
-# HF model
-REPO_ID = "Qwen/Qwen2.5-7B-Instruct"
-HF_TOKEN = os.getenv("HF_TOKEN")
+# =========================
+# MODEL (ÜCRETSİZ)
+# =========================
+MODEL_ID = "google/gemma-2b-it"
+HF_TOKEN = os.getenv("HF_TOKEN")  # HF ayarlardan ekle (READ yeterli)
 
-client = InferenceClient(
-    model=REPO_ID,
-    token=HF_TOKEN
-)
+client = InferenceClient(model=MODEL_ID, token=HF_TOKEN)
 
-def chat_fn(message, history):
-    if not message:
-        return "Bir şey yazmadın."
+# =========================
+# FASTAPI
+# =========================
+app = FastAPI(title="ZenkaMind API Test")
 
-    try:
-        messages = [
-            {"role": "system", "content": "Sen ZenkaMind isimli, sadece Türkçe konuşan bir asistansın."}
-        ]
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
 
-        for u, a in history:
-            messages.append({"role": "user", "content": u})
-            messages.append({"role": "assistant", "content": a})
+# =========================
+# MODELLER
+# =========================
+class ChatRequest(BaseModel):
+    message: str
 
-        messages.append({"role": "user", "content": message})
+# =========================
+# ENDPOINTLER
+# =========================
+@app.get("/")
+def root():
+    return {
+        "status": "ok",
+        "service": "ZenkaMind API Test",
+        "model": MODEL_ID
+    }
 
-        output = client.chat_completion(
-            messages=messages,
+@app.get("/health")
+def health():
+    return {
+        "status": "healthy",
+        "model": MODEL_ID,
+        "token_loaded": bool(HF_TOKEN)
+    }
+
+@app.post("/api/chat")
+def chat(req: ChatRequest):
+    try:
+        completion = client.chat_completion(
+            messages=[
+                {"role": "system", "content": "Sen ZenkaMind adlı Türkçe konuşan bir yapay zekasın."},
+                {"role": "user", "content": req.message},
+            ],
             max_tokens=300,
-            temperature=0.7
+            temperature=0.7,
         )
 
-        return output.choices[0].message.content
+        reply = completion.choices[0].message.content
+        return {"response": reply}
 
     except Exception as e:
-        return f"Hata oluştu: {str(e)}"
-
-demo = gr.ChatInterface(
-    fn=chat_fn,
-    title="ZenkaMind API Test",
-    description="Bu ekran sadece backend test içindir."
-)
+        return {
+            "response": "Model şu anda yanıt veremedi.",
+            "error": str(e)
+        }
 
-demo.launch()
+# =========================
+# LOCAL (HF IGNORE EDER AMA KALSIN)
+# =========================
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)