Add cache for statistics

The statistics for past days will never change so we can savely
cache it and reduce the time we spent in sqlite dramatically

On my dev machine from 325ms to 14ms and the change might be
even more dramatically on the server which is quite slower.

app/monitoring.py

@@ -1,3 +1,4 @@
+from datetime import date
 import hashlib
 from dataclasses import dataclass
 from sqlite3 import Connection
@@ -24,7 +25,7 @@ def get_statistics(db: Connection) -> statistic:
     # Get daily active users
     cursor = db.cursor()
     cursor.execute(
-        """SELECT COUNT(*) 
+        """SELECT COUNT(*)
         FROM statistics_daily
         WHERE date == DATE('now');"""
     )
@@ -48,7 +49,30 @@ def get_statistics(db: Connection) -> statistic:
     return statistic(daily_users, monthly_users, total_users)
 
 
-def get_chart_data(db: Connection) -> list[Tuple[str, int]]:
+# Only the usage for today will change, so we can cache all past days.
+chart_cache: list[Tuple[str, int, int, int]] = []
+
+def get_chart_data(db: Connection) -> list[Tuple[str, int, int, int]]:
+    global chart_cache
+    since = chart_cache[-1][0] if len(chart_cache) > 0 else None
+
+    # Fetch the newest data
+    new_data = get_chart_data_since(db, since)
+    result = chart_cache + new_data
+
+    # Update the cache (never store the last day because we don't trust it)
+    # Well the day is not over so it will probably change.
+    chart_cache += new_data[:-1]
+
+    return result
+
+
+def get_chart_data_since(db: Connection, since: str|None = None) -> list[Tuple[str, int, int, int]]:
+    if since is None:
+        # This is just any day far in the past but, the exact day is just the day before
+        # it went into production.
+        since = "2023-07-16"
+
     # Get daily active users
     cursor = db.cursor()
     cursor.execute(
@@ -57,17 +81,18 @@ def get_chart_data(db: Connection) -> list[Tuple[str, int]]:
                     COUNT(*) AS 'daily',
                     (SELECT COUNT(DISTINCT token_hash)
                     FROM statistics_daily s2
-                    WHERE s2.date <= s.date 
+                    WHERE s2.date <= s.date
                     AND s2.date >= DATE( s.date, '-30 days')
                 ) AS 'monbthly',
                 (SELECT COUNT(DISTINCT token_hash)
                     FROM statistics_daily s3
                     WHERE s3.date <=  s.date
                 ) AS 'total'
             FROM statistics_daily s
+            WHERE day > ?
             GROUP BY day
-            ORDER BY day 
-        """
+            ORDER BY day
+        """, (since,)
     )
     rows = cursor.fetchall()
     return rows