Update app.py

app.py

@@ -2,47 +2,111 @@ import os
 import requests
 import gradio as gr
 
-# 1. Retrieve the token from the environment
+# 1) Grab your personal (read-access) token from HF Secrets
 HF_TOKEN = os.environ.get("HF_TOKEN")
+HEADERS = {"Authorization": f"Bearer {HF_TOKEN}"}
 
-# 2. Function that calls the private Space
-def call_private_space(user_input):
+# 2) Root URL for your PRIVATE space (jtdearmon/rag_dal_budget)
+#    Note how the subdomain is: jtdearmon--rag_dal_budget.hf.space
+PRIVATE_SPACE_ROOT = "https://jtdearmon--rag_dal_budget.hf.space"
+
+def call_private_space(function_index, data_list):
+    """
+    A helper that sends a POST request to one of the Gradio endpoints
+    in the private rag_dal_budget space.
+
+    By default:
+      - The 1st .click() function is at /run/predict
+      - The 2nd .click() function is at /run/predict_1
+      - 3rd at /run/predict_2
+      etc.
+
+    Inputs:
+      function_index: 0 for run_query, 1 for handle_feedback, etc.
+      data_list:      The 'data' array to send, matching the function’s inputs.
+
+    Returns:
+      The first output in the JSON payload (result["data"][0]) or an error message.
+    """
+    # Build the correct endpoint: /run/predict, /run/predict_1, /run/predict_2...
+    url = f"{PRIVATE_SPACE_ROOT}/run/predict"
+    if function_index > 0:
+        url += f"_{function_index}"
+
+    # Gradio expects a JSON body of the form {"data": [...]}
+    payload = {"data": data_list}
+    try:
+        resp = requests.post(url, headers=HEADERS, json=payload)
+        resp.raise_for_status()
+        json_resp = resp.json()
+        # Typically, the first (and often only) return value is in json_resp["data"][0]
+        return json_resp["data"][0]
+    except Exception as e:
+        return f"Error: {e}"
+
+# -------------------------------------------------------------------------
+# Correspond to the private Space’s two main interactive functions:
+#    1) run_query(user_query, use_summary)
+#    2) handle_feedback(user_query, answer, feedback_option)
+# -------------------------------------------------------------------------
+def run_query(user_query, use_summary):
+    """
+    Calls the first Gradio function (index 0) in the private space,
+    which is the 'run_query' pipeline.
     """
-    Sends a POST request to the private Space's Gradio endpoint.
-    Adjust the URL or payload if your private Space uses something different.
+    return call_private_space(0, [user_query, use_summary])
+
+def handle_feedback(user_query, answer, feedback_option):
+    """
+    Calls the second Gradio function (index 1) in the private space,
+    which is 'handle_feedback'.
     """
-    # The private Space's URL - note 'jtdearmon--rag_dal_budget' matches the subdomain pattern
-    private_url = "https://jtdearmon--rag_dal_budget.hf.space/run/predict"
-    
-    headers = {
-        "Authorization": f"Bearer {HF_TOKEN}"
-    }
-    
-    # Typical Gradio JSON payload for a single text input
-    payload = {
-        "data": [user_input]
-    }
-    
-    # Send request
-    response = requests.post(private_url, headers=headers, json=payload)
-    response_data = response.json()
-    
-    # This depends on how your Gradio app in the private space returns output
-    # Usually it might be response_data["data"][0] if it returns one text output
-    result = response_data["data"][0]
-    
-    return result
-
-
-# 3. Create a Gradio interface
-interface = gr.Interface(
-    fn=call_private_space,
-    inputs="text",
-    outputs="text",
-    title="Public Demo for rag_dal_budget (Private)",
-    description="This public Space calls the private `rag_dal_budget` behind the scenes."
-)
-
-# 4. Launch the interface
-if __name__ == "__main__":
-    interface.launch()
+    return call_private_space(1, [user_query, answer, feedback_option])
+
+# -------------------------------------------------------------------------
+# Build a minimal Gradio UI that calls the above 2 endpoints
+# -------------------------------------------------------------------------
+with gr.Blocks() as demo:
+    gr.Markdown("## Public Interface → Private `rag_dal_budget`")
+
+    gr.Markdown(
+        "This public Space relays requests to a **private** Gradio app "
+        "(`rag_dal_budget`) so the underlying code is hidden."
+    )
+
+    # Inputs for the first function
+    user_query = gr.Textbox(
+        label="Your Query",
+        placeholder="Ask about the Dallas budget..."
+    )
+    use_summary_chk = gr.Checkbox(
+        value=False,
+        label="Use Graph Summaries?"
+    )
+    get_answer_btn = gr.Button("Get Answer")
+
+    # Output from the first function
+    answer_output = gr.Markdown(label="AI Answer")
+
+    # Feedback components for the second function
+    feedback_radio = gr.Radio(
+        choices=["👍", "⚖️", "👎"],
+        value="⚖️",
+        label="Feedback"
+    )
+    feedback_btn = gr.Button("Submit Feedback")
+    feedback_result = gr.Markdown()
+
+    # Wire up the .click() events
+    get_answer_btn.click(
+        fn=run_query,
+        inputs=[user_query, use_summary_chk],
+        outputs=answer_output
+    )
+    feedback_btn.click(
+        fn=handle_feedback,
+        inputs=[user_query, answer_output, feedback_radio],
+        outputs=feedback_result
+    )
+
+demo.launch()