Update app.py

app.py

@@ -28,20 +28,40 @@ def extract_questions_handler(guideline_pdf, surgery_type, suggestions):
     return questions, df
 
 
-def answer_questions_handler(patient_pdf, questions_state):
+def preview_questions_handler(qs: List[str]):
+    """Return a read-only preview table for any list of questions."""
+    if not qs:
+        return pd.DataFrame({"#": [], "Question": []}), "No questions available."
+    df = pd.DataFrame({"#": list(range(1, len(qs) + 1)), "Question": qs})
+    return df, f"{len(qs)} questions in the list."
+
+
+def lock_questions_handler(qs: List[str]):
+    """
+    Freeze the current questions into a locked snapshot to ensure we ask the 'earlier' ones.
+    Returns (locked_list, locked_df, msg)
+    """
+    if not qs:
+        raise gr.Error("No questions to lock. Extract questions in Tab 1 first.")
+    df = pd.DataFrame({"#": list(range(1, len(qs) + 1)), "Question": qs})
+    return qs, df, f"🔒 Locked {len(qs)} questions. These will be used for answering until you lock again."
+
+
+def answer_questions_handler(patient_pdf, locked_qs: List[str]):
     """
     Returns a DataFrame: Question | Answer | Rationale
+    (Answers only the LOCKED questions to guarantee earlier list is used.)
     """
     if patient_pdf is None:
         raise gr.Error("Please upload a patient chart PDF.")
-    if not questions_state:
-        raise gr.Error("No questions available. Extract questions first.")
+    if not locked_qs:
+        raise gr.Error("No locked questions. Click “🔒 Lock questions for answering” first.")
 
     pdf_path = patient_pdf.name if hasattr(patient_pdf, "name") else str(patient_pdf)
     a_engine = AnsweringEngine(pdf_path)
 
     rows = []
-    for q in questions_state:
+    for q in locked_qs:
         res: Dict[str, Dict[str, str]] = a_engine.answer_one(q)
         ans = res.get(q, {})
         rows.append({
@@ -57,6 +77,10 @@ def answer_questions_handler(patient_pdf, questions_state):
 with gr.Blocks(title="Guideline → Questions → Chart Answers", theme="soft") as demo:
     gr.Markdown("# 🏥 Guideline Q&A (Yes/No) — with Iterative Feedback")
 
+    # States:
+    questions_state = gr.State([])        # live list (can be regenerated in Tab 1)
+    questions_locked = gr.State([])       # frozen snapshot used for answering
+
     with gr.Tabs():
         with gr.Tab("1) Extract Questions"):
             with gr.Row():
@@ -68,7 +92,6 @@ with gr.Blocks(title="Guideline → Questions → Chart Answers", theme="soft")
             )
             extract_btn = gr.Button("Extract Yes/No Questions", variant="primary")
 
-            questions_state = gr.State([])  # list[str]
             questions_df = gr.Dataframe(headers=["#", "Question"], interactive=False, wrap=True)
             gr.Markdown("If the questions need refinement, update 'Suggestions' and click the button again.")
 
@@ -83,13 +106,42 @@ with gr.Blocks(title="Guideline → Questions → Chart Answers", theme="soft")
             )
 
         with gr.Tab("2) Answer from Patient Chart"):
+            gr.Markdown("#### Step A — Confirm the questions to be asked")
+
+            # Preview current (live) questions
+            preview_btn = gr.Button("Preview current questions from Tab 1")
+            preview_df = gr.Dataframe(headers=["#", "Question"], interactive=False, wrap=True)
+            preview_msg = gr.Markdown()
+
+            preview_btn.click(
+                fn=preview_questions_handler,
+                inputs=[questions_state],
+                outputs=[preview_df, preview_msg],
+            )
+
+            # Lock questions (freeze snapshot)
+            lock_btn = gr.Button("🔒 Lock questions for answering", variant="secondary")
+            locked_df = gr.Dataframe(headers=["#", "Question"], interactive=False, wrap=True)
+            locked_msg = gr.Markdown()
+
+            lock_btn.click(
+                fn=lock_questions_handler,
+                inputs=[questions_state],
+                outputs=[questions_locked, locked_df, locked_msg],
+            )
+
+            gr.Markdown("#### Step B — Answer the locked questions from the patient chart")
+
             patient_pdf = gr.File(label="Upload Patient Chart PDF (e.g., JAY_MORGAN.pdf)", file_count="single")
-            answer_btn = gr.Button("Answer All Questions", variant="primary")
+            answer_btn = gr.Button("Answer All Locked Questions", variant="primary")
             answers_df = gr.Dataframe(headers=["Question", "Answer", "Rationale"], interactive=False, wrap=True)
 
+            def answer_locked_handler(patient_file, locked_qs):
+                return answer_questions_handler(patient_file, locked_qs)
+
             answer_btn.click(
-                fn=answer_questions_handler,
-                inputs=[patient_pdf, questions_state],
+                fn=answer_locked_handler,
+                inputs=[patient_pdf, questions_locked],
                 outputs=answers_df,
             )
 
@@ -98,6 +150,7 @@ with gr.Blocks(title="Guideline → Questions → Chart Answers", theme="soft")
         "- Embeddings are cached under `./embeddings/<PDF-name>/` to avoid recomputation.\n"
         "- Set your OpenAI key: `export OPENAI_API_KEY=...` before running.\n"
         "- If a PDF is scanned/image-only, text extraction may be poor (consider OCR pre-processing).\n"
+        "- Use the **lock** step to ensure the earlier questions are exactly the ones being answered.\n"
     )
 
 if __name__ == "__main__":