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")
 
 

templates/index.html

@@ -32,6 +32,12 @@
         <span class="icon is-small is-left"><i data-feather="server"></i></span>
       </div>
     </div>
+    <div id="overview" class="box mb-5">
+      <h2 class="subtitle">Overview</h2>
+      <p>Total logs: <span id="stat-total">-</span></p>
+      <p>Date range: <span id="stat-start">-</span> to <span id="stat-end">-</span></p>
+      <p>Unique domains: <span id="stat-domains">-</span></p>
+    </div>
     <div id="reports-container"></div>
   </div>
   <script src="https://cdn.jsdelivr.net/npm/feather-icons/dist/feather.min.js"></script>
@@ -42,6 +48,10 @@
     const intervalSelect = document.getElementById('interval-select');
     const domainSelect = document.getElementById('domain-select');
     const container = document.getElementById('reports-container');
+    const totalElem = document.getElementById('stat-total');
+    const startElem = document.getElementById('stat-start');
+    const endElem = document.getElementById('stat-end');
+    const domainsElem = document.getElementById('stat-domains');
 
     let currentInterval = intervalSelect.value;
     let currentDomain = domainSelect.value;
@@ -97,6 +107,17 @@
         });
     }
 
+    function loadStats() {
+      fetch('global/stats.json')
+        .then(r => r.json())
+        .then(stats => {
+          totalElem.textContent = stats.total_logs;
+          startElem.textContent = stats.start_date;
+          endElem.textContent = stats.end_date;
+          domainsElem.textContent = stats.unique_domains;
+        });
+    }
+
     function loadReports() {
       let path = currentInterval;
       if (currentDomain) {
@@ -128,6 +149,7 @@
     });
 
     loadReports();
+    loadStats();
     feather.replace();
   </script>
 </body>

tests/test_reports.py

@@ -189,3 +189,31 @@ def test_global_reports_once(tmp_path, sample_reports, monkeypatch):
     assert global_snippet.exists()
     assert not (tmp_path / "output" / "hourly" / "domain_totals.html").exists()
 
+
+def test_global_stats_file(tmp_path, sample_reports, monkeypatch):
+    db_path = tmp_path / "database" / "ngxstat.db"
+    setup_db(db_path)
+
+    monkeypatch.setattr(gr, "DB_PATH", db_path)
+    monkeypatch.setattr(gr, "OUTPUT_DIR", tmp_path / "output")
+    monkeypatch.setattr(gr, "REPORT_CONFIG", sample_reports)
+    monkeypatch.setattr(
+        gr, "TEMPLATE_DIR", Path(__file__).resolve().parents[1] / "templates"
+    )
+
+    gr._generate_global()
+
+    stats_path = tmp_path / "output" / "global" / "stats.json"
+    assert stats_path.exists()
+    stats = json.loads(stats_path.read_text())
+    assert set(stats.keys()) == {
+        "total_logs",
+        "start_date",
+        "end_date",
+        "unique_domains",
+    }
+    assert stats["total_logs"] == 2
+    assert stats["start_date"] == "2024-01-01 10:00:00"
+    assert stats["end_date"] == "2024-01-01 10:05:00"
+    assert stats["unique_domains"] == 1
+