django distill

Minimal configuration static site generator for Django

189
25
Python

django-distill

django-distill now has a website. Read more at:

🔗 https://django-distill.com/

django-distill is a minimal configuration static site generator and publisher
for Django. Most Django versions are supported, however up to date versions are
advised including the Django 3.x releases. django-distill as of the 1.7 release
only supports Python 3. Python 2 support has been dropped. If you require Python 2
support please pin django-distill to version 1.6 in your requirements.txt or
Pipfile. Python 3.6 or above is advised.

django-distill extends existing Django sites with the ability to export
fully functional static sites. It is suitable for sites such as blogs that have
a mostly static front end but you still want to use a CMS to manage the
content.

django-distill iterates over URLs in your Django project using easy to write
iterable functions to yield the parameters for whatever pages you want to save
as static HTML. These static files can be automatically uploaded to a bucket-style
remote container such as Amazon S3, Googe Cloud Files, Microsoft Azure Storage,
or, written to a local directory as a fully working local static version of
your project. The site generation, or distillation process, can be easily
integrated into CI/CD workflows to auto-deploy static sites on commit.
django-distill can be defined as an extension to Django to make Django
projects compatible with “Jamstack”-style site architecture.

django-distill plugs directly into the existing Django framework without the
need to write custom renderers or other more verbose code. You can also integrate
django-distill with existing dynamic sites and just generate static pages for
a small subsection of pages rather than the entire site.

For static files on CDNs you can use the following ‘cache buster’ library to
allow for fast static media updates when pushing changes:

🔗 meeb/django-cachekiller

There is a complete example site that creates a static blog and uses
django-distill with django-cachekiller via continuous deployment on Netlify
available here:

🔗 meeb/django-distill-example

Installation

Install from pip:

$ pip install django-distill

Add django_distill to your INSTALLED_APPS in your settings.py:

INSTALLED_APPS = [
    # ... other apps here ...
    'django_distill',
]

That’s it.

Limitations

django-distill generates static pages and therefore only views which allow
GET requests that return an HTTP 200 status code are supported.

It is assumed you are using URI parameters such as /blog/123-abc and not
querystring parameters such as /blog?post_id=123&title=abc. Querystring
parameters do not make sense for static page generation for obvious reasons.

Additionally With one-off static pages dynamic internationalisation won’t work
so all files are generated using the LANGUAGE_CODE value in your
settings.py.

Static media files such as images and style sheets are copied from your static
media directory defined in STATIC_ROOT. This means that you will want to run
./manage.py collectstatic before you run ./manage.py distill-local
if you have made changes to static media. django-distill doesn’t chain this
request by design, however you can enable it with the --collectstatic
argument.

Usage

Assuming you have an existing Django project, edit a urls.py to include the
distill_path function which replaces Django’s standard path function and
supports the new keyword arguments distill_func and distill_file.

The distill_func argument should be provided with a function or callable
class that returns an iterable or None.

The distill_file argument is entirely optional and allows you to override the
URL that would otherwise be generated from the reverse of the URL regex. This
allows you to rename URLs like /example to any other name like
example.html. As of v0.8 any URIs ending in a slash / are automatically
modified to end in /index.html. You can use format string parameters in the
distill_file to customise the file name, arg values from the URL will be
substituted in, for example {} for positional args or {param_name} for
named args.

An example distill setup for a theoretical blogging app would be:

# Replaces the standard django.conf.path, identical syntax
from django_distill import distill_path

# Views and models from a theoretical blogging app
from blog.views import PostIndex, PostView, PostYear
from blog.models import Post

def get_index():
    # The index URI path, '', contains no parameters, named or otherwise.
    # You can simply just return nothing here.
    return None

def get_all_blogposts():
    # This function needs to return an iterable of dictionaries. Dictionaries
    # are required as the URL this distill function is for has named parameters.
    # You can just export a small subset of values here if you wish to
    # limit what pages will be generated.
    for post in Post.objects.all():
        yield {'blog_id': post.id, 'blog_title': post.title}

def get_years():
    # You can also just return an iterable containing static strings if the
    # URL only has one argument and you are using positional URL parameters:
    return (2014, 2015)
    # This is really just shorthand for ((2014,), (2015,))

urlpatterns = (
    # e.g. / the blog index
    distill_path('',
                 PostIndex.as_view(),
                 name='blog-index',
                 # Note that for paths which have no paramters
                 # distill_func is optional
                 distill_func=get_index,
                 # '' is not a valid file name! override it to index.html
                 distill_file='index.html'),
    # e.g. /post/123-some-post-title using named parameters
    distill_path('post/<int:blog_id>-<slug:blog_title>.html',
                 PostView.as_view(),
                 name='blog-post',
                 distill_func=get_all_blogposts),
    # e.g. /posts-by-year/2015 using positional parameters
    # url ends in / so file path will have /index.html appended
    distill_path('posts-by-year/<int:year>/',
                 PostYear.as_view(),
                 name='blog-year',
                 distill_func=get_years),
)

Your site will still function identically with the above changes. Internally
the distill_func and distill_file parameters are removed and the URL is
passed back to Django for normal processing. This has no runtime performance
impact as this happens only once upon starting the application.

If your path has no URI paramters, such as / or /some-static-url you do
not have to specify the distill_func parameter if you don’t want to. As for
paths with no parameters the distill_func always returns None, this is set
as the default behaviour for distill_funcs.

You can use the distill_re_path function as well, which replaces the default
django.urls.re_path function. Its usage is identical to the above:

from django_distill import distill_re_path

urlpatterns = (
    distill_re_path(r'some/regex'
                    SomeOtherView.as_view(),
                    name='url-other-view',
                    distill_func=some_other_func),
)

If you are using an older version of Django in the 1.x series you can use the
distill_url function instead which replaces the django.conf.urls.url or
django.urls.url functions. Its usage is identical to the above:

from django_distill import distill_url

urlpatterns = (
    distill_url(r'some/regex'
                SomeView.as_view(),
                name='url-view',
                distill_func=some_func),
)

Parameters in file names

You can standard Python string formatting in distill_file as well to enable
you to change the output file path for a file if you wish. Note this does not
update the URL used by Django so if you use this make sure your path pattern
matches the distill_file pattern or your links might not work in Django. An
example:

# Override file path with parameters. Values are taken from the URL pattern
urlpatterns = (
    distill_path('post/<int:blog_id>-<slug:blog_title>.html',
                 PostView.as_view(),
                 name='blog-post',
                 distill_func=get_all_blogposts,
                 distill_file="post/{blog_id}-{blog_title}.html"
)

Non-standard status codes

All views rendered by django-distill into static pages must return an HTTP 200 status
code. If for any reason you need to render a view which does not return an HTTP 200
status code, for example you also want to statically generate a 404 page which has a
view which (correctly) returns an HTTP 404 status code you can use the
distill_status_codes optional argument to a view. For example:

from django_distill import distill_url

urlpatterns = (
    distill_url(r'some/regex'
                SomeView.as_view(),
                name='url-view',
                distill_status_codes=(200, 404),
                distill_func=some_func),
)

The optional distill_status_codes argument accepts a tuple of status codes as integers
which are permitted for the view to return without raising an error. By default this is
set to (200,) but you can override it if you need to for your site.

Tracking Django’s URL function support

django-distill will mirror whatever your installed version of Django supports,
therefore at some point the distill_url function will cease working in the future
when Django 2.x itself depreciates the django.conf.urls.url and django.urls.url
functions. You can use distill_re_path as a drop-in replacement. It is advisable to
use distill_path or distill_re_path if you’re building a new site now.

The distill-local command

Once you have wrapped the URLs you want to generate statically you can now
generate a complete functioning static site with:

$ ./manage.py distill-local [optional /path/to/export/directory]

Under the hood this simply iterates all URLs registered with distill_url and
generates the pages for them using parts of the Django testing framework to
spoof requests. Once the site pages have been rendered then files from the
STATIC_ROOT are copied over. Existing files with the same name are replaced in
the target directory and orphan files are deleted.

distill-local supports the following optional arguments:

--collectstatic: Automatically run collectstatic on your site before
rendering, this is just a shortcut to save you typing an extra command.

--quiet: Disable all output other than asking confirmation questions.

--force: Assume ‘yes’ to all confirmation questions.

--exclude-staticfiles: Do not copy any static files at all, only render output from
Django views.

Note If any of your views contain a Python error then rendering will fail
then the stack trace will be printed to the terminal and the rendering command
will exit with a status code of 1.

The distill-publish command

$ ./manage.py distill-publish [optional destination here]

If you have configured at least one publishing destination (see below) you can
use the distill-publish command to publish the site to a remote location.

This will perform a full synchronisation, removing any remote files that are no
longer present in the generated static site and uploading any new or changed
files. The site will be built into a temporary directory locally first when
publishing which is deleted once the site has been published. Each file will be
checked that it has been published correctly by requesting it via the
PUBLIC_URL.

distill-publish supports the following optional arguments:

--collectstatic: Automatically run collectstatic on your site before
rendering, this is just a shortcut to save you typing an extra command.

--quiet: Disable all output other than asking confirmation questions.

--force: Assume ‘yes’ to all confirmation questions.

--exclude-staticfiles: Do not copy any static files at all, only render output from
Django views.

--skip-verify: Do not test if files are correctly uploaded on the server.

--ignore-remote-content: Do not fetch the list of remote files. It means that all
files will be uploaded, and no existing remote file will be deleted. This can be
useful if you have a lot of files on the remote server, and you know that you want
to update most of them, and you don’t care if old files remain on the server.

--parallel-publish [number of threads]: Publish files in parallel on multiple
threads, this can speed up publishing. Defaults to 1 thread.

Note that this means if you use --force and --quiet that the output
directory will have all files not part of the site export deleted without any
confirmation.

Note If any of your views contain a Python error then rendering will fail
then the stack trace will be printed to the terminal and the rendering command
will exit with a status code of 1.

The distill-test-publish command

$ ./manage.py distill-test-publish [optional destination here]

This will connect to your publishing target, authenticate to it, upload a
randomly named file, verify it exists on the PUBLIC_URL and then delete it
again. Use this to check your publishing settings are correct.

distill-test-publish has no arguments.

Optional configuration settings

You can set the following optional settings.py variables:

DISTILL_DIR: string, default directory to export to:

DISTILL_DIR = '/path/to/export/directory'

DISTILL_PUBLISH: dictionary, like Django’s settings.DATABASES, supports
default:

DISTILL_PUBLISH = {
    'default': {
        ... options ...
    },
    'some-other-target': {
        ... options ...
    },
}

DISTILL_SKIP_ADMIN_DIRS: bool, defaults to True

DISTILL_SKIP_ADMIN_DIRS = True

Set DISTILL_SKIP_ADMIN_DIRS to False if you want django-distill to also copy over
static files in the static/admin directory. Usually, these are not required or
desired for statically generated sites. The default behaviour is to skip static admin
files.

Writing single files

As of django-distill version 3.0.0 you can use the
django_distill.renderer.render_single_file method to write out a single file
to disk using django_distill. This is useful for writing out single files to disk,
for example, you have a Django site which has some static files in a directory
written by django_distill but the rest of the site is a normal dynamic Django site.
You can update a static HTML file every time a model instance is saved. You can
use single file writing with signals to achieve this. For example:

# in models.py
from django.db.models.signals import post_save
from django.dispatch import receiver
from django_distill.renderer import render_single_file

@receiver(post_save, sender=SomeBlogPostModel)
def write_blog_post_static_file_post_save(sender, **kwargs):
    render_single_file(
        '/path/to/output/directory',
        'blog-post-view-name',
        blog_id=sender.pk,
        blog_slug=sender.slug
    )

The syntax for render_single_file is similar to Django’s url.reverse. The full
usage interface is:

render_single_file(
    '/path/to/output/directory',
    'view-name-set-in-urls-py',
    *view_args,
    **view_kwargs
)

For example, if you had a blog post URL defined as:

    # in urls.py
    distill_path('post/<int:blog_id>_<slug:blog_slug>.html',
                 PostView.as_view(),
                 name='blog-post',
                 distill_func=get_all_blogposts),

Your usage would be:

render_single_file(
    '/path/to/output/directory',
    'blog-post',
    blog_id=123,
    blog_slug='blog-title-slug',
)

which would write out the contents of /post/123_blog-title-slug.html into
/path/to/output/directory as the file
/path/to/output/directory/post/123_blog-title-slug.html. Note any required
sub-directories (/path/to/output/directory/post in this example) will be
automatically created if they don’t already exist. All django-distill rules
apply, such as URLs ending in / will be saved as /index.html to make sense
for a physical file on disk.

Also note that render_single_file can only be imported and used into an
initialised Django project.

Publishing targets

You can automatically publish sites to various supported remote targets through
backends just like how you can use MySQL, SQLite, PostgreSQL etc. with
Django by changing the backend database engine. Currently the engines supported
by django-distill are:

django_distill.backends.amazon_s3: Publish to an Amazon S3 bucket. Requires
the Python library boto3 ($ pip install django-distill[amazon]). The bucket
must already exist (use the AWS control panel). Options:

'some-s3-container': {
    'ENGINE': 'django_distill.backends.amazon_s3',
    'PUBLIC_URL': 'http://.../',
    'ACCESS_KEY_ID': '...',
    'SECRET_ACCESS_KEY': '...',
    'BUCKET': '...',
},

django_distill.backends.google_storage: Publish to a Google Cloud Storage
bucket. Requires the Python libraries google-api-python-client and
google-cloud-storage
($ pip install django-distill[google]). The bucket
must already exist and be set up to host a public static website (use the
Google Cloud control panel). Options:

'some-google-storage-bucket': {
    'ENGINE': 'django_distill.backends.google_storage',
    'PUBLIC_URL': 'https://storage.googleapis.com/[bucket.name.here]/',
    'BUCKET': '[bucket.name.here]',
    'JSON_CREDENTIALS': '/path/to/some/credentials.json',
},

Note that JSON_CREDENTIALS is optional; if it is not specified, the google libraries
will try other authentication methods, in the search order described here:
https://cloud.google.com/docs/authentication/application-default-credentials (e.g. the
GOOGLE_APPLICATION_CREDENTIALS environment variable, attached service account, etc).

django_distill.backends.microsoft_azure_storage: Publish to a Microsoft
Azure Blob Storage container. Requires the Python library
azure-storage-blob ($ pip install django-distill[microsoft]). The storage
account must already exist and be set up to host a public static website
(use the Microsoft Azure control panel). Options:

'some-microsoft-storage-account': {
    'ENGINE': 'django_distill.backends.microsoft_azure_storage',
    'PUBLIC_URL': 'https://[storage-account-name]...windows.net/',
    'CONNECTION_STRING': '...',
},

Note that each Azure storage account supports one static website using the
magic container $web which is where django-distill will attempt to
publish your site.

Tests

There is a minimal test suite, you can run it by cloing this repository,
installing the required dependancies in requirements.txt then execuiting:

# ./run-tests.py

Contributing

All properly formatted and sensible pull requests, issues and comments are
welcome.