updated dependency because cloudflare changed their endpoint

analysis.py

@@ -0,0 +1,67 @@
+import sqlite3
+import os
+from datetime import datetime
+import pytz
+import time
+
+from config import DB_PATH
+
+def get_conn() -> sqlite3.Connection:
+    uri = f"file:{os.path.abspath(DB_PATH)}?cache=shared"
+    conn = sqlite3.connect(uri, uri=True, check_same_thread=False)
+    conn.row_factory = sqlite3.Row
+    conn.execute("PRAGMA journal_mode=WAL;")
+    conn.execute("PRAGMA synchronous=NORMAL;")
+    conn.execute("PRAGMA temp_store=MEMORY;")
+    conn.execute("PRAGMA cache_size=-20000;")  # ~20MB cache
+    return conn
+
+CONN = get_conn()
+
+# ---------------------- Utilities ----------------------
+
+def ts_to_iso(ts: float | int | str) -> str:
+    try:
+        t = float(ts)
+    except Exception:
+        return str(ts)
+    return datetime.fromtimestamp(t, tz=pytz.timezone("Europe/Berlin")).isoformat()
+
+# ---------------------- API Endpoints ----------------------
+
+def api_series(q_from: float, q_to: float):
+    params = []
+    where = []
+    if q_from is not None:
+        where.append("timestamp >= ?")
+        params.append(q_from)
+    if q_to is not None:
+        where.append("timestamp <= ?")
+        params.append(q_to)
+    where_sql = ("WHERE " + " AND ".join(where)) if where else ""
+
+    sql = (
+        f"SELECT timestamp, down_90th, up_90th FROM speed_tests {where_sql} "
+        "ORDER BY timestamp ASC;"
+    )
+
+    rows = []
+    for i, r in enumerate(CONN.execute(sql, params)):
+        rows.append({
+            "t": float(r["timestamp"]),
+            "t_iso": ts_to_iso(r["timestamp"]),
+            "down_90th": None if r["down_90th"] is None else float(r["down_90th"]),
+            "up_90th": None if r["up_90th"] is None else float(r["up_90th"]),
+        })
+
+    return rows
+
+rows = api_series(1756220400, time.time())
+
+does_not_work = 0
+
+for r in rows:
+    if r["down_90th"] is None or r["up_90th"] is None:
+        does_not_work += 1
+
+print(f"Working percentage: {100 - (does_not_work / len(rows))*100}% ({does_not_work}/{len(rows)})")