feat: finalize root api.py with X-Auth-Secret auth security and clean pure sales cohort

scripts/patch_snapshots.py

@@ -0,0 +1,93 @@
+import sqlite3
+from datetime import datetime, timedelta
+
+def backfill_snapshots():
+    conn = sqlite3.connect('data/hiffi_catalog.db')
+    cur = conn.cursor()
+
+    # 1. Create the rolling snapshots table
+    cur.execute('''
+        CREATE TABLE IF NOT EXISTS artist_metrics_snapshots (
+            snapshot_id INTEGER PRIMARY KEY AUTOINCREMENT,
+            artist_id TEXT NOT NULL,
+            platform TEXT NOT NULL,
+            metric_type TEXT NOT NULL,
+            metric_value REAL NOT NULL,
+            recorded_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
+            FOREIGN KEY (artist_id) REFERENCES artists(artist_id)
+        )
+    ''')
+    
+    # Clear existing to prevent duplicate stacking if run multiple times
+    cur.execute('DELETE FROM artist_metrics_snapshots')
+
+    # 2. Fetch baseline telemetry using the exact column names you provided
+    cur.execute('''
+        WITH latest AS (
+            SELECT t.*, ROW_NUMBER() OVER (PARTITION BY t.artist_id ORDER BY t.last_synced_at DESC, t.telemetry_id DESC) as rn
+            FROM telemetry_metrics t
+        )
+        SELECT 
+            a.artist_id, 
+            l.subscribers_count, 
+            l.total_views, 
+            l.spotify_monthly_listeners, 
+            l.instagram_followers,
+            l.growth_velocity_7d, 
+            l.growth_velocity_30d
+        FROM artists a
+        JOIN latest l ON a.artist_id = l.artist_id
+        WHERE a.image_verified_flag = 1 AND l.rn = 1
+    ''')
+    
+    records = cur.fetchall()
+    now = datetime.utcnow()
+    t_7d = (now - timedelta(days=7)).strftime('%Y-%m-%d %H:%M:%S')
+    t_30d = (now - timedelta(days=30)).strftime('%Y-%m-%d %H:%M:%S')
+    t_now = now.strftime('%Y-%m-%d %H:%M:%S')
+
+    insert_data = []
+
+    for row in records:
+        artist_id, yt_subs, yt_views, spt_listeners, ig_followers, v_7d, v_30d = row
+        
+        # Reverse-calculate historic metrics from the current percentages
+        def calc_historic(current, velocity_pct):
+            if current is None or velocity_pct is None: return current
+            try:
+                return max(0, current / (1.0 + (velocity_pct / 100.0)))
+            except ZeroDivisionError:
+                return current
+
+        platforms = [
+            ('youtube', 'subscribers', yt_subs),
+            ('youtube', 'views', yt_views),
+            ('spotify', 'listeners', spt_listeners),
+            ('instagram', 'followers', ig_followers)
+        ]
+
+        for plat, m_type, current_val in platforms:
+            if current_val is None: continue
+            
+            val_7d = calc_historic(current_val, v_7d)
+            val_30d = calc_historic(current_val, v_30d)
+
+            # Stage insertions for T-30, T-7, and T-0
+            insert_data.append((artist_id, plat, m_type, val_30d, t_30d))
+            insert_data.append((artist_id, plat, m_type, val_7d, t_7d))
+            insert_data.append((artist_id, plat, m_type, current_val, t_now))
+
+    # 3. Batch insert and commit
+    cur.executemany('''
+        INSERT INTO artist_metrics_snapshots 
+        (artist_id, platform, metric_type, metric_value, recorded_at)
+        VALUES (?, ?, ?, ?, ?)
+    ''', insert_data)
+
+    conn.commit()
+    print(f"✅ Successfully generated {len(insert_data)} snapshot rows for {len(records)} artists.")
+    print("UI growth badges will now render correctly using historical timestamps.")
+    conn.close()
+
+if __name__ == "__main__":
+    backfill_snapshots()