Update streamlit_app.py

streamlit_app.py

@@ -5,7 +5,7 @@ import main
 from pathlib import Path
 from datetime import datetime
 
-# --- PATHING ---
+# --- PATHING LOGIC ---
 if Path("/data").exists():
     CSV_PATH = Path("/data/policy_tracker.csv")
 else:
@@ -24,6 +24,33 @@ def load_data():
         return df.sort_values(by="event_date", ascending=False)
     return None
 
+# --- RETENTION POLICY (UI CLEANER) ---
+def apply_retention_policy(df):
+    if df.empty:
+        return df
+        
+    today = pd.Timestamp.now().tz_localize(None).normalize()
+    
+    # 1. Legislation: Keep everything (No expiration)
+    leg_df = df[df['type'] == 'Legislation']
+    
+    # 2. News/Media & Exec Action: Keep last 30 days (and any future anomalies)
+    news_types = ['News/Media', 'Federal/Exec Action']
+    news_mask = (df['type'].isin(news_types)) & ((df['event_date'] >= today - pd.Timedelta(days=30)) | df['event_date'].isna())
+    news_df = df[news_mask]
+    
+    # 3. Schedules: Keep last 60 days (and all future scheduled events)
+    sched_types = ['Schedule/Hearing', 'Hearing/Markup']
+    sched_mask = (df['type'].isin(sched_types)) & ((df['event_date'] >= today - pd.Timedelta(days=60)) | df['event_date'].isna())
+    sched_df = df[sched_mask]
+    
+    # 4. Fallback for any undefined types
+    other_df = df[~df['type'].isin(['Legislation'] + news_types + sched_types)]
+    
+    # Combine the filtered datasets
+    active_df = pd.concat([leg_df, news_df, sched_df, other_df]).drop_duplicates(subset=['link'])
+    return active_df.sort_values(by="event_date", ascending=False)
+
 # --- UI SETUP ---
 st.set_page_config(page_title="PolicyPilot Intel", layout="wide")
 st.title("PolicyPilot Intelligence Dashboard")
@@ -45,12 +72,25 @@ with st.sidebar:
     
     if df is not None and not df.empty:
         # Dynamically pull every unique category type found in the CSV
-        available_types = df['type'].unique().tolist()
+        available_types = df['type'].dropna().unique().tolist()
         selected_types = st.multiselect(
             "Filter by Category:", 
             options=available_types, 
             default=available_types
         )
+        
+    st.divider()
+    st.header("Data Management")
+    if df is not None and not df.empty:
+        # Convert the raw, unfiltered dataframe to CSV for download
+        csv_data = df.to_csv(index=False).encode('utf-8')
+        st.download_button(
+            label="Download Full Historical Archive (CSV)",
+            data=csv_data,
+            file_name=f"policy_pilot_archive_{pd.Timestamp.now().strftime('%Y-%m-%d')}.csv",
+            mime="text/csv",
+            use_container_width=True
+        )
 
 # --- EXECUTIVE BRIEFING ---
 if df is not None and not df.empty:
@@ -63,8 +103,9 @@ if df is not None and not df.empty:
 
     if st.button("Generate Briefing"):
         with st.spinner("AI is synthesizing latest data..."):
-            # Pull top 10 items for context
-            top_items = df.head(10)
+            # Pull top 10 items from the filtered active view for context
+            active_data = apply_retention_policy(df.copy())
+            top_items = active_data.head(10)
             context_text = "\n".join([f"- {row['title']} (Source: {row['source']})" for _, row in top_items.iterrows()])
             
             summary_prompt = f"Provide a 3-bullet point executive summary highlighting the most critical shifts in these updates for a policy team. Use professional language:\n\n{context_text}"
@@ -132,8 +173,10 @@ def render_event_cards(display_df, is_radar=False):
 
 # --- TAB LOGIC ---
 if df is not None and not df.empty:
-    # Filter by selected types
-    filtered_df = df.copy()
+    # Apply our tiered retention rules to clean up the UI
+    filtered_df = apply_retention_policy(df.copy())
+    
+    # Filter by selected types from the sidebar
     if selected_types:
         filtered_df = filtered_df[filtered_df['type'].isin(selected_types)]