Merge pull request #46 from wagesj45/codex/add-report-generation-time-statistics

Add report generation timing stats

scripts/generate_reports.py

@@ -3,6 +3,8 @@ import sqlite3
 from pathlib import Path
 import shutil
 from typing import List, Dict, Optional
+from datetime import datetime
+import time
 
 import yaml
 
@@ -74,7 +76,9 @@ def _render_snippet(report: Dict, out_dir: Path) -> None:
     snippet_path.write_text(template.render(report=report))
 
 
-def _write_stats() -> None:
+def _write_stats(
+    generated_at: Optional[str] = None, generation_seconds: Optional[float] = None
+) -> None:
     """Query basic dataset stats and write them to ``output/global/stats.json``."""
     conn = sqlite3.connect(DB_PATH)
     cur = conn.cursor()
@@ -98,6 +102,10 @@ def _write_stats() -> None:
         "end_date": end_date,
         "unique_domains": unique_domains,
     }
+    if generated_at:
+        stats["generated_at"] = generated_at
+    if generation_seconds is not None:
+        stats["generation_seconds"] = generation_seconds
 
     out_path = OUTPUT_DIR / "global" / "stats.json"
     _save_json(out_path, stats)
@@ -197,14 +205,9 @@ def _generate_root_index() -> None:
     """Render the top-level index listing all intervals and domains."""
     _copy_icons()
     intervals = sorted(
-        [
-            name
-            for name in INTERVAL_FORMATS
-            if (OUTPUT_DIR / name).is_dir()
-        ]
+        [name for name in INTERVAL_FORMATS if (OUTPUT_DIR / name).is_dir()]
     )
 
-
     domains_dir = OUTPUT_DIR / "domains"
     domains: List[str] = []
     if domains_dir.is_dir():
@@ -227,6 +230,9 @@ def _generate_global() -> None:
         typer.echo("No report definitions found")
         return
 
+    start_time = time.time()
+    generated_at = datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S")
+
     _copy_icons()
 
     conn = sqlite3.connect(DB_PATH)
@@ -269,7 +275,8 @@ def _generate_global() -> None:
         report_list.append(entry)
 
     _save_json(out_dir / "reports.json", report_list)
-    _write_stats()
+    elapsed = round(time.time() - start_time, 2)
+    _write_stats(generated_at, elapsed)
     typer.echo("Generated global reports")
 
 

templates/index.html

@@ -49,6 +49,8 @@
         <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>
+        <p>Last generated: <span id="stat-generated">-</span></p>
+        <p>Generation time: <span id="stat-elapsed">-</span> seconds</p>
       </div>
       <div id="overview-reports"></div>
     </div>
@@ -96,6 +98,8 @@
     const startElem = document.getElementById('stat-start');
     const endElem = document.getElementById('stat-end');
     const domainsElem = document.getElementById('stat-domains');
+    const generatedElem = document.getElementById('stat-generated');
+    const elapsedElem = document.getElementById('stat-elapsed');
 
     let currentInterval = intervalSelect.value;
     let currentDomain = domainSelect.value;
@@ -161,6 +165,9 @@
           startElem.textContent = stats.start_date;
           endElem.textContent = stats.end_date;
           domainsElem.textContent = stats.unique_domains;
+          generatedElem.textContent = stats.generated_at || '-';
+          elapsedElem.textContent =
+            stats.generation_seconds !== undefined ? stats.generation_seconds : '-';
         });
     }
 

tests/test_reports.py

@@ -148,20 +148,19 @@ def test_generate_interval_domain_filter(tmp_path, sample_reports, monkeypatch):
     gr._generate_interval("hourly", "example.com")
 
     hits = json.loads(
-        (tmp_path / "output" / "domains" / "example.com" / "hourly" / "hits.json").read_text()
+        (
+            tmp_path / "output" / "domains" / "example.com" / "hourly" / "hits.json"
+        ).read_text()
     )
     assert hits[0]["value"] == 2
     reports = json.loads(
-        (tmp_path / "output" / "domains" / "example.com" / "hourly" / "reports.json").read_text()
+        (
+            tmp_path / "output" / "domains" / "example.com" / "hourly" / "reports.json"
+        ).read_text()
     )
     assert {r["name"] for r in reports} == {"hits", "error_rate"}
     assert not (
-        tmp_path
-        / "output"
-        / "domains"
-        / "example.com"
-        / "hourly"
-        / "skip_report.json"
+        tmp_path / "output" / "domains" / "example.com" / "hourly" / "skip_report.json"
     ).exists()
 
 
@@ -245,9 +244,12 @@ def test_global_stats_file(tmp_path, sample_reports, monkeypatch):
         "start_date",
         "end_date",
         "unique_domains",
+        "generated_at",
+        "generation_seconds",
     }
     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
-
+    assert isinstance(stats["generated_at"], str)
+    assert stats["generation_seconds"] >= 0