Why not django.contrib.sitemaps?

Django’s sitemaps framework didn’t support alternates when this package was written. Django 3.2 added i18n, languages and alternates, but alternate URLs are still derived by activating each language and reversing the same object’s URL. django-sitemaps instead lets you pass arbitrary alternate URLs for each entry, e.g. when languages live on different domains. URLs are added one-by-one from a regular view, without having to write Sitemap classes. Existing Django sitemaps can still be included using add_django_sitemap.

Installation

Simply pip install django-sitemaps. The package consists of a single python module, django_sitemaps, containing the single class; there’s no additional configuration necessary.

Usage

View:

from app.pages.sitemaps import PagesSitemap

def sitemap(request):
    sitemap = Sitemap(
        # All URLs are passed through build_absolute_uri.
        build_absolute_uri=request.build_absolute_uri,
    )

    # URLs can be added one-by-one. The only required argument
    # is the URL. All other arguments are keyword-only arguments.
    for p in Page.objects.active():
        url = p.get_absolute_url()
        sitemap.add(
            url,
            changefreq='weekly',
            priority=0.5,
            lastmod=p.modification_date,
            alternates={
                code: urljoin(domain, url)
                for code, domain in PAGE_DOMAINS[p.language].items()
            },
        )

    # Adding conventional Django sitemaps is supported. Pass either
    # the class or an instance. Their URLs are also passed through
    # build_absolute_uri, so django.contrib.sites isn't required.
    sitemap.add_django_sitemap(PagesSitemap)

    # You could get the serialized XML...
    # ... = sitemap.serialize([pretty_print=False])
    # ... or use the ``response`` helper to return a
    # ready-made ``HttpResponse``:
    return sitemap.response(
        # pretty_print is False by default
        pretty_print=settings.DEBUG,
    )

Some browsers display the sitemap as plain text because the xhtml:link alternates use a different namespace and no stylesheet is attached. The response is still valid XML served as application/xml; search engines don’t mind.

URLconf:

from django.urls import path
from django_sitemaps import robots_txt
from app.views import sitemap

urlpatterns = [
    path('sitemap.xml', sitemap),
    path('robots.txt', robots_txt(timeout=86400)),
    ...
]

The robots_txt function returns a view which can be used to generate a robots.txt file containing sitemap URLs. The default sitemap only contains:

User-agent: *
Sitemap: <protocol>://<host>/sitemap.xml

The list of sitemap URLs may be overridden by setting sitemaps:

from django.urls import reverse_lazy

urlpatterns = [
    path('robots.txt', robots_txt(
        timeout=86400,
        sitemaps=[
            '/sitemap.xml',
            reverse_lazy('articles-sitemap'),
            ...,
        ],
    )),
]