Chameleon Partials

Simple reuse of partial HTML page templates in the Chameleon template language for Python web frameworks. (There is also a Jinja2/Flask version here.)

Overview

When building real-world web apps with Chameleon, it's easy to end up with repeated HTML fragments. Just like organizing code for reuse, it would be ideal to reuse smaller sections of HTML template code. That's what this library is all about.

Documentation

Full documentation lives at mkennedy.codes/docs/chameleon-partials. The user guide expands each section of this README:

• Getting started: install, register the template folder, wire render_partial into your framework, render a first fragment.
• Writing partials: how paths resolve, the model, nesting, loops, METAL layouts, rendering from Python, and choosing partials safely.
• Testing: a pytest fixture for the one-time registration and what the error types promise.
• Troubleshooting: every error you can hit in normal use, its cause, and the fix.
• API reference, generated from the library's docstrings.

Coding agents can read the same material from llms-full.txt or install the agent skill.

Example

This project comes with a sample Pyramid application (see the example folder). This app displays videos that can be played on YouTube. The image, author subtitle, and view count are reused throughout the app. Here's a visual:

Check out the demo / example application to see it in action.

Installation

It's just pip install chameleon-partials (or uv add chameleon-partials) and you're all set with this pure Python package. It needs Python 3.10 or newer and depends only on Chameleon.

Usage

Using the library is incredibly easy. The first step is to register the partial method with Chameleon. Do this once at app startup:

from pathlib import Path

from pyramid.config import Configurator

import chameleon_partials


def main(_, **settings):
    """This function returns a Pyramid WSGI application."""
    with Configurator(settings=settings) as config:
        config.include('pyramid_chameleon')
        config.include('.routes')
        config.scan()

        # Register the extension for working with Chameleon.
        folder = (Path(__file__).parent / 'templates').as_posix()
        chameleon_partials.register_extensions(folder, auto_reload=True, cache_init=True)

    return config.make_wsgi_app()

Set auto_reload=True while developing so template edits show up without a restart, and leave it at the default False in production. Calling register_extensions again after it has succeeded does nothing, so repeated startup calls are safe (see Testing for pointing tests at their own folder).

Next, you define your main HTML (Chameleon) templates as usual. Then define your partial templates. I recommend locating and naming them accordingly:

├── templates
│   ├── errors
│   │   └── 404.pt
│   ├── home
│   │   ├── index.pt
│   │   └── listing.pt
│   └── shared
│       ├── _layout.pt
│       └── partials
│           ├── video_image.pt
│           └── video_square.pt

Notice the partials subfolder in the templates/shared folder.

The templates are just HTML fragments. Here is a stand-alone one for the YouTube thumbnail from the example app:

<img src="https://img.youtube.com/vi/${ video.id }/maxresdefault.jpg"
     class="img img-responsive ${ ' '.join(classes or []) }"
     alt="${ video.title }"
     title="${ video.title }">

Notice that an object called video and a list of classes are passed in as the model.

Templates can also be nested. Here is the whole single video fragment with the image as well as other info linking out to YouTube:

<div>
    <a href="https://www.youtube.com/watch?v=${ video.id }" target="_blank">
        ${ render_partial('shared/partials/video_image.pt', video=video, classes=[]) }
    </a>
    <a href="https://www.youtube.com/watch?v=${ video.id }" target="_blank"
       class="author">${ video.author }</a>
    <div class="views">${ "{:,}".format(video.views) } views</div>
</div>

Now you see the render_partial() method. It takes the subpath into the templates folder and any model data passed in as keyword arguments. The path is always relative to the registered folder, even when the call is inside another partial. It returns an HTML object that Chameleon inserts unescaped; read its html_text attribute when you need the string, for example in a test or an HTMX response.

We can finally generate the list of video blocks as follows:

<span class="video" tal:repeat="v videos">
    ${ render_partial('shared/partials/video_square.pt', video=v) }
</span>

This time, we reframe each item in the list from the outer template (called v) as the video model in the inner HTML section.

Choosing Partials Safely

Choose the partial in your code, never from request data. render_partial() refuses a name that would reach outside the registered template folder (an absolute path, a .. out of the folder, or a package:path spec), but a request that picks the fragment should still be mapped to a name you wrote rather than used to build the path:

PARTIALS = {'card': 'shared/partials/video_square.pt', 'row': 'shared/partials/video_row.pt'}
html = render_partial(PARTIALS[request.params['style']], video=video)

Pass the model as named keyword arguments too. Spreading a request's parameters into it (render_partial(name, **request.params)) lets a parameter replace a name the template relies on, including render_partial itself.

The View Methods

In order to share the render_partial() function with your template, you'll need to pass it along to the template with your model (dictionary).

If you are using the Pyramid web framework, you can add this file as middleware. Just drop it into your views folder:

# views/partials_middleware.py
from pyramid.events import subscriber, BeforeRender

import chameleon_partials


@subscriber(BeforeRender)
def add_global(event):
    event['render_partial'] = chameleon_partials.render_partial

For other frameworks using Chameleon (e.g. FastAPI), you can add render_partial to the resulting dictionary. We've built a simple function to keep this fool-proof:

chameleon_partials.extend_model(model)

Here's a typical view method that uses render_partial, notice the use of extending the model before passing it to the template (the example app's views skip this because the middleware above handles it):

@view_config(route_name='listing', renderer='demo_chameleon_partials:templates/home/listing.pt')
def listing(_):
    videos = video_service.all_videos()
    model = dict(videos=videos)
    return chameleon_partials.extend_model(model)

Again: If you are using Pyramid, use the middleware. Otherwise, use the extend_model() method or something similar to the middleware in your framework.

Testing

Partials are plain functions over plain files, so tests need no web framework. Register the test template folder in a fixture and reset the registration flag afterwards, because a second register_extensions() call is a no-op by default:

@pytest.fixture
def registered_extension():
    folder = (Path(__file__).parent / 'test_templates').as_posix()
    chameleon_partials.register_extensions(folder, auto_reload=True)
    yield
    chameleon_partials.has_registered_extensions = False


def test_render_with_data(registered_extension):
    html = chameleon_partials.render_partial('render/with_data.pt', name='Sarah', age=32)
    assert '<span>Your name is Sarah and age is 32</span>' in html.html_text

The Testing page covers the error contract: PartialsException is raised for misuse only, and Chameleon's own errors (such as the ValueError for a missing template) pass through unchanged.

Troubleshooting

Every error you can hit in normal use, with its cause and fix, is on the Troubleshooting page: rendering before registration, Template not found, NameError: render_partial, markup that shows up escaped, and template edits that do not appear.