Add PostgreSQL analytics dashboard prototype and realistic sample data

mockups/sample_report.py

@@ -0,0 +1,47 @@
+"""Read-only checks and a compact report for the generated sample histories."""
+import math
+
+from database import connect
+
+
+def report():
+    with connect() as conn:
+        for game in conn.execute("SELECT id,name FROM games WHERE properties->>'sampleVersion'='2' ORDER BY id").fetchall():
+            gid=game['id']
+            summary=conn.execute('''WITH counts AS(SELECT player_id,count(*) n FROM sessions WHERE game_id=%s GROUP BY player_id)
+              SELECT count(*) players,100.0*count(*) FILTER(WHERE n=1)/count(*) one_session FROM counts''',(gid,)).fetchone()
+            minutes=conn.execute('SELECT avg(extract(epoch FROM ended_at-started_at))/60 minutes FROM sessions WHERE game_id=%s',(gid,)).fetchone()['minutes']
+            returns=[]
+            for day in [1,7,30]:
+                row=conn.execute('''SELECT 100.0*count(*) FILTER(WHERE EXISTS(SELECT 1 FROM sessions s
+                  WHERE s.game_id=p.game_id AND s.player_id=p.id AND (s.started_at AT TIME ZONE 'UTC')::date=(p.first_seen AT TIME ZONE 'UTC')::date+%s)) / nullif(count(*),0) rate
+                  FROM players p WHERE p.game_id=%s AND (p.first_seen AT TIME ZONE 'UTC')::date+%s<current_date''',(day,gid,day)).fetchone()
+                returns.append(float(row['rate']))
+            spending=conn.execute('''SELECT player_id,sum(unit_price_robux*quantity) total FROM product_sales
+              WHERE game_id=%s AND unit_price_robux>0 GROUP BY player_id ORDER BY total DESC''',(gid,)).fetchall()
+            total=sum(r['total'] for r in spending)
+            top_share=100*sum(r['total'] for r in spending[:max(1,math.ceil(len(spending)*.01))])/total
+            bad_sales=conn.execute('''SELECT count(*) n FROM product_sales p JOIN sessions s ON s.game_id=p.game_id AND s.id=p.session_id
+              WHERE p.game_id=%s AND (p.purchased_at<s.started_at OR p.purchased_at>s.ended_at)''',(gid,)).fetchone()['n']
+            overlaps=conn.execute('''WITH ordered AS(SELECT started_at,lag(ended_at) OVER(PARTITION BY player_id ORDER BY started_at) previous_end
+              FROM sessions WHERE game_id=%s) SELECT count(*) n FROM ordered WHERE started_at<previous_end''',(gid,)).fetchone()['n']
+            bad_balances=conn.execute('''SELECT count(*) n FROM player_snapshots WHERE game_id=%s AND data ? 'cash'
+              AND (data->>'cash')::numeric<>(data->>'cashEarned')::numeric-(data->>'cashSpent')::numeric''',(gid,)).fetchone()['n']
+            bad_snapshots=conn.execute('''SELECT count(*) n FROM player_statistics WHERE game_id=%s AND latest_data IS NOT NULL
+              AND ((latest_data->>'robux')::numeric<>robux_recorded OR (latest_data->>'purchases')::numeric<>products_bought)''',(gid,)).fetchone()['n']
+            assert returns[0]>returns[1]>returns[2], (game,returns)
+            assert 3<float(minutes)<30, (game,minutes)
+            assert summary['one_session']>50, (game,summary)
+            assert len(spending)<summary['players']*.1, (game,len(spending))
+            assert top_share>10, (game,top_share)
+            assert bad_sales==0, (game,bad_sales)
+            assert overlaps==0, (game,overlaps)
+            assert bad_balances==0, (game,bad_balances)
+            assert bad_snapshots==0, (game,bad_snapshots)
+            print(f"{game['name']}: {summary['players']:,} players | one session {summary['one_session']:.1f}% | average session {minutes:.1f} min")
+            print(f"  D1/D7/D30: {' / '.join(f'{r:.2f}%' for r in returns)} | payers {len(spending):,} ({100*len(spending)/summary['players']:.1f}%) | top 1% payer revenue share {top_share:.1f}%")
+        print('Sample distribution and timeline checks passed.')
+
+
+if __name__=='__main__':
+    report()