Sitemap Generator for Django Pattern Library Pages

sitemap-generator.py

# Please adapt the file paths according to your project conventions.

import glob
import re

base_url = "http://localhost:8000/pattern-library/render-pattern/patterns"
# /render-pattern/ loads the pattern in full page view without the pattern library UI

pattern_libray_pages = glob.glob('**/patterns/pages/**/*.html', recursive=True)

sitemap = """<?xml version='1.0' encoding='utf-8'?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
  xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9">

"""

for html_file in pattern_libray_pages:
  # removes all characters in the file path up to and including "patterns"
  url = base_url + re.sub(r'^.*?patterns', '', html_file)
  sitemap += "  <url>\n	   <loc>"
  sitemap += url
  sitemap += "</loc>\n  </url>\n\n"

sitemap += "</urlset>"

f = open("sitemap.xml", "w")
f.write(sitemap)
f.close()