Update app.py

app.py

@@ -2,7 +2,18 @@ import os
 import gradio as gr
 from huggingface_hub import InferenceClient
 
+# ----------------------------------------------------------------------
+# Helper: read a secret with a safe fallback (useful when you run the
+# script locally without a secrets file).
+# ----------------------------------------------------------------------
+def _secret(key: str, fallback: str) -> str:
+    """Return the value of a secret or the supplied fallback."""
+    return os.getenv(key, fallback)
 
+
+# ----------------------------------------------------------------------
+# Core chat logic – the system prompt now comes from the secret `prec_chat`.
+# ----------------------------------------------------------------------
 def respond(
     message: str,
     history: list[dict[str, str]],
@@ -14,31 +25,21 @@ def respond(
     """
     Generate a response using the HuggingFace Inference API.
 
-    The system prompt is taken from the secret **prec_chat** (defined in the
-    Gradio secrets file).  Users can no longer edit the system prompt.
+    The system prompt is taken from the secret **prec_chat**.
+    Users cannot edit it from the UI.
     """
-    # ----------------------------------------------------------------------
-    # 1️⃣  Load the system prompt from the secret.
-    # ----------------------------------------------------------------------
-    # If the secret is missing we fall back to a generic prompt so the app
-    # still works locally.
-    system_message = os.getenv("prec_chat", "You are a helpful assistant.")
+    # 1️⃣  Load the system prompt (fallback = generic assistant)
+    system_message = _secret("prec_chat", "You are a helpful assistant.")
 
-    # ----------------------------------------------------------------------
     # 2️⃣  Initialise the HF inference client.
-    # ----------------------------------------------------------------------
     client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
 
-    # ----------------------------------------------------------------------
     # 3️⃣  Build the message list for the chat completion endpoint.
-    # ----------------------------------------------------------------------
     messages = [{"role": "system", "content": system_message}]
     messages.extend(history)                     # previous conversation
     messages.append({"role": "user", "content": message})  # current query
 
-    # ----------------------------------------------------------------------
     # 4️⃣  Stream the response back to the UI.
-    # ----------------------------------------------------------------------
     response = ""
     for chunk in client.chat_completion(
         messages,
@@ -47,25 +48,22 @@ def respond(
         temperature=temperature,
         top_p=top_p,
     ):
-        # The API returns a list of choices – we only care about the first one.
         choices = chunk.choices
         token = ""
         if choices and choices[0].delta.content:
             token = choices[0].delta.content
-
         response += token
         yield response
 
 
-# --------------------------------------------------------------------------
+# ----------------------------------------------------------------------
 # UI definition – the system‑prompt textbox has been removed.
-# --------------------------------------------------------------------------
+# ----------------------------------------------------------------------
 chatbot = gr.ChatInterface(
     respond,
     type="messages",
     additional_inputs=[
-        # System prompt is now fixed via the secret, so we only expose the
-        # generation parameters.
+        # Only generation parameters are exposed now.
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
@@ -78,11 +76,31 @@ chatbot = gr.ChatInterface(
     ],
 )
 
+# ----------------------------------------------------------------------
+# Build the Blocks layout (no LoginButton – we use our own auth).
+# ----------------------------------------------------------------------
 with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
     chatbot.render()
 
 
+# ----------------------------------------------------------------------
+# Launch with **basic authentication**.
+# ----------------------------------------------------------------------
 if __name__ == "__main__":
-    demo.launch()
+    # Pull the allowed credentials from secrets (fallback = no access)
+    allowed_user = _secret("CHAT_USER", "")
+    allowed_pass = _secret("CHAT_PASS", "")
+
+    # If either is missing we refuse to start – this prevents an accidental
+    # open‑access deployment.
+    if not allowed_user or not allowed_pass:
+        raise RuntimeError(
+            "Authentication credentials not found in secrets. "
+            "Add CHAT_USER and CHAT_PASS to secrets.toml."
+        )
+
+    demo.launch(
+        auth=(allowed_user, allowed_pass),   # <-- Gradio's built‑in basic auth
+        # optional: you can also set `auth_message="Please log in"` or
+        # `prevent_thread_lock=True` depending on your deployment.
+    )