adding statistics page for admin

web/db.py

@@ -783,3 +783,44 @@ def change_keyword_color(keyword_id: int, new_color: str) -> bool:
         except Exception:
             session.rollback()
             return False
+
+
+def stats_overview() -> Dict[str, Any]:
+    """Return an overview of job DB statistics.
+
+    Returns a dict with keys:
+      - total_jobs: int
+      - total_keywords: int (distinct keywords in listings)
+      - total_regions: int (distinct regions in listings)
+      - jobs_per_keyword: List[{"keyword": str, "count": int}]
+      - jobs_per_region: List[{"region": str, "count": int}]
+    """
+    with _ensure_session() as session:
+        total_jobs = session.execute(text(
+            "SELECT COUNT(*) FROM job_listings l INNER JOIN job_descriptions d ON l.job_id = d.job_id AND l.url = d.url"
+        )).scalar_one()
+        total_keywords = session.execute(text(
+            "SELECT COUNT(DISTINCT keyword) FROM job_listings WHERE keyword IS NOT NULL AND keyword != ''"
+        )).scalar_one()
+        total_regions = session.execute(text(
+            "SELECT COUNT(DISTINCT region) FROM job_listings WHERE region IS NOT NULL AND region != ''"
+        )).scalar_one()
+
+        rows = session.execute(text(
+            "SELECT COALESCE(keyword, '') AS keyword, COUNT(*) as cnt FROM job_listings l INNER JOIN job_descriptions d ON l.job_id = d.job_id AND l.url = d.url GROUP BY keyword ORDER BY cnt DESC"
+        )).fetchall()
+        jobs_per_keyword = [
+            {"keyword": r[0], "count": int(r[1])} for r in rows]
+
+        rows = session.execute(text(
+            "SELECT COALESCE(region, '') AS region, COUNT(*) as cnt FROM job_listings l INNER JOIN job_descriptions d ON l.job_id = d.job_id AND l.url = d.url GROUP BY region ORDER BY cnt DESC"
+        )).fetchall()
+        jobs_per_region = [{"region": r[0], "count": int(r[1])} for r in rows]
+
+    return {
+        "total_jobs": int(total_jobs or 0),
+        "total_keywords": int(total_keywords or 0),
+        "total_regions": int(total_regions or 0),
+        "jobs_per_keyword": jobs_per_keyword,
+        "jobs_per_region": jobs_per_region,
+    }