refreshcss 0.6.0

Removes unused classes, ids, and element selectors from CSS.

RefreshCSS 🫧

RefreshCSS is a Python library that removes unused classes, ids, and element selectors from CSS.

Make your CSS so fresh, so clean.

🔧 Installation

To use the refreshcss library or use the django-compressor integration

uv add refreshcss

To run refreshcss on the command-line

uvx refreshcss[cli]

⭐️ Features

• Pure Python (no extra NodeJS build process needed)
• Fast and spec-compliant CSS and HTML parsing
• Filters out unused classes, ids, and elements from CSS based on HTML templates
• Supports modern CSS syntax, including nested rules, container queries, and media queries
• Handles Django/Jinja-style HTML templates with context-aware extraction
• Integrates with django-compressor to clean CSS during the compress command
• Command-line interface for use in CI/CD pipelines

⌨️ Command-line interface

uvx refreshcss[cli]

Usage: refreshcss [OPTIONS] CSS HTML...

  Remove classes, ids, and element selectors not used in HTML from CSS.

Options:
  -o, --output FILENAME  Write to file instead of stdout.
  -R, -r, --recursive    Recursively search subdirectories listed.
  --encoding TEXT        Character encoding to use when reading files. If not
                         specified, the encoding will be guessed.
  --version              Show the version and exit.
  --help                 Show this message and exit.

🗜️ Integrate with django-compressor

Add "refreshcss.filters.RefreshCSSFilter" to COMPRESS_FILTERS in the Django settings file.

COMPRESS_FILTERS = {
    "css": [
        "refreshcss.filters.RefreshCSSFilter",
        ...
    ],
    "js": [...],
}

⚙️ Library

from refreshcss import RefreshCSS, PathSite

def clean_css(css_path: Path):
    # Parse the HTML files
    site = PathSite(paths=["templates"], recursive=True)
    site.parse()

    # Get clean CSS based on the parsed HTML
    css_text = css_path.read_text()
    cleaned_css = RefreshCSS(site).clean(css_text)

    return cleaned_css

🤓 How does it work?

1. Catalogue classes, ids, and elements that are currently being used in found HTML templates
2. Catalogue classes, ids, elements, and at-rules in a particular CSS stylesheet
3. Return a new CSS stylesheet that only contains rules that are actively being used by the HTML

🧐 Why?

I wanted to have a filter for django-compressor that would purge unused CSS as part of the compress step when deploying for coltrane apps. After dealing with a manual process and attempting to integrate https://purgecss.com and https://github.com/uncss/uncss I thought "this couldn't be that hard to do in Python".

Which is always the thought at the beginning of every side project... 😅

🙋 FAQ

Will this work with SPAs?

Probably not. RefreshCSS only inspects HTML templates, so if CSS classes are being changed client-side then refreshcss will not know about it.

Does this work by crawling a website URL?

Currently no, although that is a possibility in the future. PRs appreciated. 😉

Does this support HTML written in the Django Template Language?

Yes! That was a primary reason I built my own solution. 😅 Jinja might also be possible to support with some small tweaks, although it is currently untested.

Is this what people mean when they say "treeshaking"?

Maybe. 🤷

I found a bug!

Thanks for trying RefreshCSS out! Please make a PR (pull request) with a small test that replicates the bug or, if that is not possible, create a new discussion.

🤘 Related libraries

Node

• uncss
• PurgeCSS

Python

• treeshake: I unfortunately could not get this to work on my local environment.
• cssutils: This unfortunately seemed to choke on more modern CSS when I tested on Bulma 1.0.
• css-optomizer

🙏 Thanks

• Django
• django-compressor
• CSS standards

❤️ Support

This project is supported by GitHub Sponsors and Digital Ocean.

🥳 Contributors

Tobias Bölz 💻 📖 ⚠️