adding statistics page for admin

web/app.py

@@ -21,6 +21,7 @@ from web.db import (
     set_user_keywords,
     get_all_regions,
     get_all_keywords,
+    stats_overview,
     upsert_region,
     upsert_keyword,
     list_regions_full,
@@ -487,6 +488,34 @@ def admin_taxonomy():
     return render_template('admin/taxonomy.html', title='Taxonomy', regions=regions, keywords=keywords)
 
 
+@app.route('/admin/stats', methods=['GET'])
+def admin_stats():
+    if not require_admin():
+        return redirect(url_for('login'))
+    # Optional filters via query params
+    keyword = request.args.get('keyword')
+    region = request.args.get('region')
+    try:
+        stats = stats_overview()
+        # For detailed jobs table, reuse get_all_jobs() and filter
+        jobs = get_all_jobs()
+        if keyword:
+            jobs = [j for j in jobs if (j.get('keyword') or '') == keyword]
+        if region:
+            jobs = [j for j in jobs if (j.get('region') or '') == region]
+    except Exception as e:
+        flash(f'Error computing stats: {e}')
+        stats = {
+            'total_jobs': 0,
+            'total_keywords': 0,
+            'total_regions': 0,
+            'jobs_per_keyword': [],
+            'jobs_per_region': []
+        }
+        jobs = []
+    return render_template('admin/stats.html', title='Statistics', stats=stats, jobs=jobs, regions=get_all_regions(), keywords=get_all_keywords())
+
+
 def init():
     """Main function to run the Flask app."""
     # Ensure DB is initialized

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,
+    }

web/templates/admin/stats.html

@@ -0,0 +1,46 @@
+{% extends 'base.html' %} {% block content %}
+<div id="admin-stats">
+  <h2>Database Statistics</h2>
+  <div class="stats-summary">
+    <p><strong>Total jobs:</strong> {{ stats.total_jobs }}</p>
+    <p><strong>Total keywords:</strong> {{ stats.total_keywords }}</p>
+    <p><strong>Total regions:</strong> {{ stats.total_regions }}</p>
+  </div>
+
+  <h3>Jobs per keyword</h3>
+  <table>
+    <thead>
+      <tr>
+        <th>Keyword</th>
+        <th>Count</th>
+      </tr>
+    </thead>
+    <tbody>
+      {% for row in stats.jobs_per_keyword %}
+      <tr>
+        <td>{{ row.keyword or '(empty)' }}</td>
+        <td>{{ row.count }}</td>
+      </tr>
+      {% endfor %}
+    </tbody>
+  </table>
+
+  <h3>Jobs per region</h3>
+  <table>
+    <thead>
+      <tr>
+        <th>Region</th>
+        <th>Count</th>
+      </tr>
+    </thead>
+    <tbody>
+      {% for row in stats.jobs_per_region %}
+      <tr>
+        <td>{{ row.region or '(empty)' }}</td>
+        <td>{{ row.count }}</td>
+      </tr>
+      {% endfor %}
+    </tbody>
+  </table>
+</div>
+{% endblock %}

web/templates/base.html

@@ -22,6 +22,7 @@
         {% if current_user and current_user.is_admin %} |
         <a href="{{ url_for('scrape_page') }}">Scrape Jobs</a> |
         <a href="{{ url_for('admin_taxonomy') }}">Taxonomy</a> |
+        <a href="{{ url_for('admin_stats') }}">Statistics</a> |
         <a href="{{ url_for('admin_users') }}">Users</a> {% endif %} {% if
         session.get('username') %} |
         <a href="{{ url_for('logout') }}">Logout</a> {% else %} |