Update app.py

app.py

@@ -1,38 +1,99 @@
+import os, traceback, regex as re2
 import gradio as gr
+import pandas as pd
+from typing import List, Tuple, Dict
+
 from settings import HEALTHCARE_SETTINGS, GENERAL_CONVERSATION_PROMPT, USE_SCENARIO_ENGINE
+from audit_log import log_event
+from privacy import safety_filter, refusal_reply
 from data_registry import DataRegistry
 from upload_ingest import extract_text_from_files
 from healthcare_analysis import HealthcareAnalyzer
-from rag import RAGIndex
-from scenario_planner import plan_from_llm
+from scenario_planner import parse_to_plan
 from scenario_engine import ScenarioEngine
-from llm_router import cohere_chat
-
-def is_healthcare_scenario(text, files):
-    return any(k in text.lower() for k in HEALTHCARE_SETTINGS["healthcare_keywords"]) and bool(files)
-
-def handle(msg, history, files):
-    registry=DataRegistry()
-    for f in files or []: registry.add_path(f)
-    rag=RAGIndex(); rag.add(extract_text_from_files(files).get("chunks",[]))
-    if is_healthcare_scenario(msg, files) and USE_SCENARIO_ENGINE:
-        analyzer=HealthcareAnalyzer(registry)
-        results=analyzer.comprehensive_analysis(msg)
-        catalog={n:list(df.columns) for n,df in results.items() if hasattr(df,"columns")}
-        plan=plan_from_llm(msg, catalog)
-        structured=ScenarioEngine.render_plan(plan, results)
-        return history+[(msg, structured)], ""
-    else:
-        out=cohere_chat(f"{GENERAL_CONVERSATION_PROMPT}\n\nUser: {msg}\nAssistant:") or "..."
-        return history+[(msg, out)], ""
-
-with gr.Blocks() as demo:
-    chat=gr.Chatbot()
-    files=gr.Files(type="filepath", file_count="multiple")
-    msg=gr.Textbox()
-    btn=gr.Button("Send")
-    btn.click(handle,[msg,chat,files],[chat,msg])
-    msg.submit(handle,[msg,chat,files],[chat,msg])
-
-if __name__=="__main__":
-    demo.launch()
+from rag import RAGIndex
+from llm_router import generate_narrative, cohere_chat
+
+def _sanitize_text(s: str) -> str:
+    if not isinstance(s, str): return s
+    return re2.sub(r'[\p{C}--[\n\t]]+', '', s)
+
+def _dataset_catalog(results: Dict[str, any]) -> Dict[str, List[str]]:
+    cat = {}
+    for k, v in results.items():
+        if isinstance(v, pd.DataFrame):
+            cat[k] = v.columns.tolist()
+    return cat
+
+def is_healthcare_scenario(text: str, has_files: bool) -> bool:
+    t = (text or "").lower()
+    kws = HEALTHCARE_SETTINGS["healthcare_keywords"]
+    structured = any(s in t for s in ["background", "situation", "tasks", "deliverables"])
+    return has_files and (structured or any(k in t for k in kws))
+
+def handle(user_msg: str, history: list, files: list) -> Tuple[list, str]:
+    try:
+        safe_in, blocked_in, reason_in = safety_filter(user_msg, mode="input")
+        if blocked_in: return history + [(user_msg, refusal_reply(reason_in))], ""
+
+        # Normalize files -> paths
+        file_paths = [getattr(f, "name", None) or f for f in (files or [])]
+
+        # Register CSVs
+        registry = DataRegistry()
+        for p in file_paths:
+            try: registry.add_path(p)
+            except Exception as e: log_event("ingest_error", None, {"file": p, "err": str(e)})
+
+        # RAG ingest (safe on empty)
+        rag = RAGIndex()
+        ing = extract_text_from_files(file_paths)
+        rag.add(ing.get("chunks", []))
+
+        if is_healthcare_scenario(safe_in, bool(file_paths)) and USE_SCENARIO_ENGINE:
+            analyzer = HealthcareAnalyzer(registry)
+            datasets = analyzer.comprehensive_analysis(safe_in)
+            catalog = _dataset_catalog(datasets)
+
+            # LLM → plan (no hardcoding)
+            plan = parse_to_plan(safe_in, catalog)
+
+            # Deterministic execution
+            structured_md = ScenarioEngine.execute_plan(plan, datasets)
+
+            # Narrative with Canadian grounding
+            rag_hits = [txt for txt, _ in rag.retrieve(safe_in, k=6)]
+            narrative = generate_narrative(safe_in, structured_md, rag_hits)
+
+            final = f"{structured_md}\n\n# Narrative & Recommendations\n\n{narrative}"
+            return history + [(user_msg, _sanitize_text(final))], ""
+
+        # General conversation (Cohere primary, open-model fallback inside cohere_chat if needed)
+        prompt = f"{GENERAL_CONVERSATION_PROMPT}\n\nUser: {safe_in}\nAssistant:"
+        ans = cohere_chat(prompt) or "How can I help further?"
+        return history + [(user_msg, _sanitize_text(ans))], ""
+
+    except Exception as e:
+        tb = traceback.format_exc()
+        log_event("app_error", None, {"err": str(e), "tb": tb})
+        return history + [(user_msg, f"Error: {e}\n\n{tb}")], ""
+
+with gr.Blocks(analytics_enabled=False) as demo:
+    gr.Markdown("## Canadian Healthcare AI • Scenario-Agnostic (Cohere primary • Deterministic analytics)")
+    chat = gr.Chatbot(type="tuple", height=520)  # tuple mode (matches how we store history)
+    files = gr.Files(file_count="multiple", type="filepath", file_types=HEALTHCARE_SETTINGS["supported_file_types"])
+    msg = gr.Textbox(placeholder="Paste any scenario (Background / Situation / Tasks / Deliverables) or just chat.")
+    send = gr.Button("Send")
+    clear = gr.Button("Clear")
+
+    def _on_send(m, h, f):
+        h2, _ = handle(m, h or [], f or [])
+        return h2, ""
+
+    send.click(_on_send, inputs=[msg, chat, files], outputs=[chat, msg])
+    msg.submit(_on_send, inputs=[msg, chat, files], outputs=[chat, msg])
+    clear.click(lambda: ([], ""), outputs=[chat, msg])
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=int(os.getenv("PORT", "7860")))
+