Simple SEO tests with Python, Requests and Requests-HTML

seo-monitoring.py

import requests
from requests_html import requests_html

if __name__ == '__main__': 
    # Define test data
    test_data = [
        {
            "url": "https://www.example.com/",
            "status": 200,
            "title": "Example Domain",
            "h1": "Example Domain",
        },
    ]

    for item in test_data:
        try:
            r = requests.get(item["url"])
            html = HTML(html=r.text)

            # No redirects
            if r.history:
                assert len(r.history) == 0, json.dumps({
                    "error_type": "redirection",
                    "url": item["url"],
                    "status": r.history[0].status_code,
                })
            
            # Only HTTP 200
            assert r.status_code == 200, json.dumps({
                "error_type": "not_http_200",
                "url": item["url"],
                "status": r.status_code,
            })
            
            # Assert title
            assert len(html.xpath("//title")) == 1, json.dumps({
                "error_type": "multiple_or_no_title",
                "url": item["url"],
                "nb_titles": len(html.xpath("//title")),
            })
            assert html.xpath("//title", first=True).text == item["title"], json.dumps({
                "error_type": "wrong_title",
                "url": item["url"],
                "title": str(html.xpath("//title", first=True).text),
                "expected_title": str(item["title"]),
            })

            # Assert h1
            assert len(html.xpath("//h1")) == 1, json.dumps({
                "error_type": "multiple_or_no_h1",
                "url": item["url"],
                "nb_h1": len(html.xpath("//h1")),
            })
            assert html.xpath("//h1", first=True).text == item["h1"], json.dumps({
                "error_type": "wrong_h1",
                "url": item["url"],
                "h1": str(html.xpath("//h1", first=True).text),
                "expected_h1": str(item["h1"]),
            })
        
        except AssertionError as e:
            report.append(e)
            continue

    if len(report) > 0:
        for e in report:
            print(e)
    else:
        print("Everything's fine!")