Add global stats generation

scripts/generate_reports.py

@@ -62,6 +62,35 @@ def _render_snippet(report: Dict, out_dir: Path) -> None:
     snippet_path.write_text(template.render(report=report))
 
 
+def _write_stats() -> None:
+    """Query basic dataset stats and write them to ``output/global/stats.json``."""
+    conn = sqlite3.connect(DB_PATH)
+    cur = conn.cursor()
+
+    cur.execute("SELECT COUNT(*) FROM logs")
+    total_logs = cur.fetchone()[0] or 0
+
+    cur.execute("SELECT MIN(time), MAX(time) FROM logs")
+    row = cur.fetchone() or (None, None)
+    start_date = row[0] or ""
+    end_date = row[1] or ""
+
+    cur.execute("SELECT COUNT(DISTINCT host) FROM logs")
+    unique_domains = cur.fetchone()[0] or 0
+
+    conn.close()
+
+    stats = {
+        "total_logs": total_logs,
+        "start_date": start_date,
+        "end_date": end_date,
+        "unique_domains": unique_domains,
+    }
+
+    out_path = OUTPUT_DIR / "global" / "stats.json"
+    _save_json(out_path, stats)
+
+
 def _bucket_expr(interval: str) -> str:
     """Return the SQLite strftime expression for the given interval."""
     fmt = INTERVAL_FORMATS.get(interval)
@@ -199,6 +228,7 @@ def _generate_global() -> None:
         report_list.append(entry)
 
     _save_json(out_dir / "reports.json", report_list)
+    _write_stats()
     typer.echo("Generated global reports")