Python script to generate a XML sitemap from a list of URLs

make_sitemap.py

import argparse
from validator_collection import validators, errors # pip install validator-collection
import requests # pip install requests
import time

def main(args):    
    with open(args.input) as f:
        urls = f.readlines()

    sitemap_urls = []
    for url in urls:
        try:
            # Check if URL is valid
            url = validators.url(url)

            # Check if URL response code is 200
            r = requests.get(url)
            if r.status_code == 200:
                xml = f"""<url>\n    <loc>{url.strip()}</loc>    \n</url>\n"""
                sitemap_urls.append(xml)
            else:
                print(f"URL {url} returned {r.status_code} status code")
        except errors.InvalidURLError:
            print('Invalid URL: {}'.format(url))
        # Wait `delay` seconds before next request`
        time.sleep(args.delay)

    xml_sitemap = f'''<?xml version="1.0" encoding="UTF-8"?>
    <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n{''.join(sitemap_urls).strip()}\n</urlset>'''

    with open(args.output, 'w') as f:
        f.write(xml_sitemap)

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--input', type=str, default='urls.txt')
    parser.add_argument('--output', type=str, default='sitemap.xml')
    parser.add_argument('--delay', type=float, default=0.5)

    args = parser.parse_args()

    main(args)