Sitemap in 43 lines of Jinja

[seanh]

Wanted to share the approach I'm using to generate my sitemap.xml file. I'm aware of the pelican-sitemap plugin, but my solution is <50 lines of template code and it supports articles, pages, the index page, the archives page, the tags, categories and authors pages, tag, category and author index pages, and period archives.

I added this to my pelicanconf.py:

# pelicanconf.py

DIRECT_TEMPLATES = ["index", "authors", "categories", "tags", "archives", "sitemap"]
THEME_TEMPLATES_OVERRIDES = ["templates"]
SITEMAP_SAVE_AS = "sitemap.xml"

Then added this templates/sitemap.html file:

{% macro render_url(path, lastmod, force=false) -%}
  {% if path or force %}
  <url>
    <loc>{{ (SITEURL + '/' + path)|escape }}</loc>
    {% if lastmod %}
    <lastmod>{{ lastmod.isoformat() }}</lastmod>
    {% endif %}
  </url>
  {% endif %}
{% endmacro -%}

<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  {% for article_or_page in articles + pages %}
  {{ render_url(article_or_page.url, article_or_page.modified or article_or_page.date) }}
  {% endfor %}

  {%- set latest_article_date = articles|map(attribute="date")|sort|last if articles else none %}
  {{ render_url("", latest_article_date, force=true) }}

  {{ render_url(ARCHIVES_URL, latest_article_date) }}

  {% macro tuples(url, tuples) %}
    {% set oldest_dates = [] %}
    {% for _, articles in tuples %}
      {% set _ = oldest_dates.append(articles|map(attribute="date")|sort|first) %}
    {% endfor %}
    {{ render_url(url, oldest_dates|sort|last) }}
  {% endmacro %}
  {{ tuples(AUTHORS_URL, authors) }}
  {{ tuples(CATEGORIES_URL, categories) }}
  {{ tuples(TAGS_URL, tags) }}

  {% for thing, articles in tags + categories + authors %}
  {{ render_url(thing.url, articles|map(attribute="date")|sort|last) }}
  {% endfor %}

  {% for archives in period_archives.values() %}
  {% for archive in archives %}
  {{ render_url(archive.url, archive.articles|map(attribute="date")|sort|last) }}
  {% endfor %}
  {% endfor %}
</urlset>

Notes:

• The filename extension needs to be templates/sitemap.html even though it will actually be rendered as /sitemap.xml (according to the SITEMAP_SAVE_AS setting): Pelican seems to expect everything in DIRECT_TEMPLATES to be .html, I couldn't get it to accept a templates/sitemap.xml file.

• I omitted <changefreq> and <priority> from the sitemap. These are optional in the sitemap spec and Google ignores them anyway, and I didn't have sensible values to set them to. These could be easily added but I think it's better to omit them.

• For articles the <lastmod> will be the article's modified time if it has one, otherwise its created time.

• Pages don't have a created time so <lastmod> will be the page's modified time if it has one or will be omitted (<lastmod> is optional in the sitemap spec).

• The <lastmod> of the index and archives pages is the created time of the most recent article on the site. (Modified times are deliberately ignored here.)

• The <lastmod> of tag, category, author and period archive pages is the created time of the latest article on that page.

• For the tags, categories and authors pages it first finds the oldest article for each tag/category/author and then takes the newest of these oldest articles, and uses that as the <lastmod> for the page. This is an attempt to set the tag/category/author page's <lastmod> to the last time that a new tag/category/author was created.

• If your site disables the archives, tags, categories, authors, tag index, category index, author index or period archives pages they'll be omitted from the sitemap.

• Pagination pages beyond the first page are omitted from the sitemap, which I think is fine.