Update app.py

app.py

@@ -1,8 +1,8 @@
 # app.py
-# GPT Chatbot powered by HuggingFace Inference API
-# Clean, safe, production-ready for Hugging Face Spaces
+# GPT Chatbot with thinking mode
+# Users enter their HF_API_TOKEN in the UI
+# Ready for Hugging Face Spaces
 
-import os
 import time
 import gradio as gr
 from huggingface_hub import InferenceClient
@@ -10,86 +10,93 @@ from huggingface_hub import InferenceClient
 # -----------------------
 # Configuration
 # -----------------------
-HF_TOKEN = os.getenv("HF_API_TOKEN")
-GPT_MODEL_ID = os.getenv("GPT_MODEL_ID", "HuggingFaceH4/zephyr-7b-beta")
-
-if HF_TOKEN is None:
-    raise ValueError("❌ Please define HF_API_TOKEN in Hugging Face Secrets.")
-
-# Create inference client
-client = InferenceClient(
-    model=GPT_MODEL_ID,
-    token=HF_TOKEN,
-)
+GPT_MODEL_ID = "HuggingFaceH4/zephyr-7b-beta"  # default model, can change
 
 # -----------------------
 # Chatbot Core
 # -----------------------
-def gpt_chat(prompt: str, history, thinking_mode=False, thinking_time=2.0):
+def gpt_chat(prompt, history, thinking, thinking_time, hf_token):
     """
-    Generate a chat response using GPT model on HuggingFace Inference API.
-    Clean, without chain-of-thought exposure.
+    Generate a chat response using Hugging Face Inference API.
+    Users provide their own HF_API_TOKEN.
     """
+    if not hf_token or hf_token.strip() == "":
+        return history, "❌ Please enter your Hugging Face API Token."
 
     if not prompt.strip():
-        return history, "لطفاً پیام بنویسید."
+        return history, "Please write a message."
+
+    # Initialize client
+    try:
+        client = InferenceClient(model=GPT_MODEL_ID, token=hf_token)
+    except Exception as e:
+        return history, f"Error creating HF client: {str(e)}"
 
-    # Optional "thinking" simulation
-    if thinking_mode:
-        time.sleep(thinking_time)
+    # Append user message to history
+    history = history + [(prompt, None)]
+
+    # Simulate thinking
+    if thinking:
+        time.sleep(max(0.0, thinking_time))
 
     try:
-        # Build clean conversation prompt
+        # Build conversation prompt
         conversation_text = ""
         for user_msg, bot_msg in history:
-            conversation_text += f"User: {user_msg}\nAssistant: {bot_msg}\n"
+            if bot_msg is not None:
+                conversation_text += f"User: {user_msg}\nAssistant: {bot_msg}\n"
         conversation_text += f"User: {prompt}\nAssistant:"
 
-        # Query HuggingFace API
+        # Generate response
         response = client.text_generation(
             prompt=conversation_text,
             max_new_tokens=200,
             temperature=0.7,
-            do_sample=True,
+            do_sample=True
         )
 
-        # Clean output
         answer = response.replace(conversation_text, "").strip()
-        answer = answer.split("Assistant:")[-1].strip()
+        if "Assistant:" in answer:
+            answer = answer.split("Assistant:")[-1].strip()
 
-        # Update history
-        history.append((prompt, answer))
+        # Update last entry in history
+        history = history[:-1] + [(prompt, answer)]
         return history, "OK"
 
     except Exception as e:
-        print("Error:", e)
-        history.append((prompt, "⚠ خطا در ارتباط با مدل."))
+        history = history[:-1] + [(prompt, "⚠ Error generating response.")]
         return history, str(e)
 
 # -----------------------
 # Gradio UI
 # -----------------------
-with gr.Blocks(title="GPT Chatbot") as demo:
+with gr.Blocks(title="NovaTalk - Chatbot") as demo:
     gr.Markdown("""
-    # 🤖 GPT Chatbot (HuggingFace Inference API)
-    مدل: **{GPT_MODEL_ID}**
-    
-    این چت‌بات از API رسمی هاگینگ‌فیس استفاده می‌کند.
+    # 🤖 NovaTalk Chatbot
+    Enter your Hugging Face API Token below to start chatting.
+    Built with ❤️ by [JumpLander](https://jumplander.org)
     """)
 
-    chatbot = gr.Chatbot(label="Chat")
-    user_input = gr.Textbox(placeholder="پیام خود را بنویسید...", show_label=False)
-    thinking_mode = gr.Checkbox(label="حالت فکر کردن", value=True)
-    thinking_time = gr.Slider(0, 5, value=2, step=0.5, label="مدت فکر کردن (ثانیه)")
-    status_box = gr.Textbox(label="وضعیت")
-
-    def respond(history, msg, thinking, thinking_t):
-        return gpt_chat(msg, history, thinking, thinking_t)
-
-    user_input.submit(
-        respond,
-        inputs=[chatbot, user_input, thinking_mode, thinking_time],
-        outputs=[chatbot, status_box]
-    )
-
-demo.launch()
+    with gr.Row():
+        with gr.Column(scale=2):
+            chatbot = gr.Chatbot(label="Chat")
+            user_input = gr.Textbox(show_label=False, placeholder="Write a message and press Send")
+            hf_token_input = gr.Textbox(
+                label="Hugging Face API Token",
+                placeholder="Paste your HF_API_TOKEN here",
+                type="password"
+            )
+            thinking_mode = gr.Checkbox(label="Thinking mode", value=True)
+            thinking_time = gr.Slider(0, 5, value=2.0, step=0.5, label="Thinking time (seconds)")
+            status_box = gr.Textbox(label="Status")
+            send_btn = gr.Button("Send")
+
+    # handle send
+    def respond(history, msg, thinking, thinking_t, hf_token):
+        return gpt_chat(msg, history or [], thinking, thinking_t, hf_token)
+
+    send_btn.click(respond, inputs=[chatbot, user_input, thinking_mode, thinking_time, hf_token_input], outputs=[chatbot, status_box])
+    user_input.submit(respond, inputs=[chatbot, user_input, thinking_mode, thinking_time, hf_token_input], outputs=[chatbot, status_box])
+
+if __name__ == "__main__":
+    demo.launch()