Update fake.py

fake.py

@@ -1,15 +1,57 @@
 import gradio as gr
-import spaces
-import transformers_gradio
+import requests
 
-demo = gr.load(name="google/gemma-3-1b-it", src=transformers_gradio.registry)
-demo = gr.load(name="unsloth/gemma-3-1b-it-GGUF", src=transformers_gradio.registry)
+LLAMA_URL = "http://localhost:8000/completion"
 
+# 🔎 Wikipedia (دائمًا)
+def search_wikipedia(query):
+    try:
+        url = f"https://en.wikipedia.org/api/rest_v1/page/summary/{query.replace(' ', '_')}"
+        res = requests.get(url, timeout=5)
+        
+        if res.status_code == 200:
+            data = res.json()
+            return data.get("extract", "")[:1000]
+    except:
+        pass
+    
+    return "No Wikipedia information found."
 
-demo.fn = spaces.GPU()(demo.fn)
+# 💬 الدردشة
+def chat(message, history):
 
-for fn in demo.fns.values():
-    fn.api_name = False
+    # 🔎 دائمًا نجلب من Wikipedia
+    wiki_text = search_wikipedia(message)
+
+    # 🧠 Prompt
+    prompt = f"""
+You are a helpful assistant.
+
+Use the Wikipedia information below to answer accurately.
+
+Wikipedia:
+{wiki_text}
+
+"""
+
+    # إضافة المحادثة السابقة
+    for h in history:
+        prompt += f"User: {h[0]}\nAssistant: {h[1]}\n"
+
+    prompt += f"User: {message}\nAssistant:"
+
+    # 🔗 llama.cpp
+    response = requests.post(LLAMA_URL, json={
+        "prompt": prompt,
+        "temperature": 0.7,
+        "top_p": 0.9,
+        "n_predict": 256,
+        "stop": ["User:"]
+    })
+
+    return response.json()["content"]
+
+demo = gr.ChatInterface(chat)
 
 if __name__ == "__main__":
     demo.launch()