using code lama

app.py

@@ -1,46 +1,46 @@
 import os
 import gradio as gr
-import requests
+from huggingface_hub import InferenceClient
 
-DEEPSEEK_API_KEY = os.getenv("DEEPSEEK_API_KEY")  # Get from environment variables
+def get_hf_token():
+    """Get the HF token from environment variables or Gradio secrets"""
+    try:
+        # Try Gradio secrets first (works on Spaces)
+        return gr.secrets["HF_TOKEN"]
+    except (AttributeError, KeyError):
+        # Fall back to environment variable (works locally)
+        return os.getenv("HF_TOKEN")
 
 def generate_code(prompt):
     try:
-        if not DEEPSEEK_API_KEY:
-            return "Error: Missing API key. Please set DEEPSEEK_API_KEY environment variable."
+        token = get_hf_token()
+        if not token:
+            return "Error: Missing Hugging Face token."
+            
+        client = InferenceClient(
+            model="codellama/CodeLlama-34b-Instruct-hf",  # Smaller alternative
+            token=token
+        )
         
-        headers = {
-            "Authorization": f"Bearer {DEEPSEEK_API_KEY}",
-            "Content-Type": "application/json"
-        }
+        system_prompt = """You are an expert Next.js, TypeScript and TailwindCSS developer. 
+        Generate clean, efficient code following these rules:
+        1. Use Next.js App Router
+        2. Strict TypeScript typing
+        3. Modern TailwindCSS classes
+        4. Include proper error boundaries
+        5. Add JSDoc comments"""
         
-        payload = {
-            "model": "deepseek-chat",
-            "messages": [
-                {
-                    "role": "system",
-                    "content": "You are an expert Next.js, TypeScript and TailwindCSS developer..."
-                },
-                {
-                    "role": "user",
-                    "content": prompt
-                }
+        response = client.chat_completion(
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": prompt}
             ],
-            "temperature": 0.3,
-            "max_tokens": 1500
-        }
-        
-        response = requests.post(
-            "https://api.deepseek.com/v1/chat/completions",
-            headers=headers,
-            json=payload
+            max_tokens=1500,
+            temperature=0.3
         )
         
-        if response.status_code == 200:
-            return response.json()["choices"][0]["message"]["content"]
-        else:
-            return f"API Error: {response.status_code} - {response.text}"
-            
+        return response.choices[0].message.content
+        
     except Exception as e:
         return f"Error generating code: {str(e)}"