Exports your Draw.io diagrams at build time for easier embedding into your documentation.
First install the package:
pip install mkdocs-drawio-exporter
Then enable it:
plugins:
- drawio-exporter
For the default configuration, just add the plugin to the plugins
key:
plugins:
- drawio-exporter
You can override the default configuration; values shown are defaults:
plugins:
- drawio-exporter:
# Diagrams are cached to speed up site generation. The default path is
# drawio-exporter, relative to the documentation directory.
cache_dir: 'drawio-exporter'
# Path to the Draw.io executable:
# * drawio on Linux
# * draw.io on macOS
# * or draw.io.exe on Windows
# We'll look for it on your system's PATH, then default installation
# paths. If we can't find it we'll warn you.
drawio_executable: null
# Additional Draw.io CLI args
# * --embed-svg-images will embed external images in SVGS, if format is "svg".
drawio_args: []
# Output format (see draw.io --help | grep format)
format: svg
# Embed format
# * The default is to embed via the <img> tag.
# * Consider <object type="image/svg+xml" data="{img_src}"></object>
# to enable interactive elements (like hyperlinks) in SVGs.
# * Consider {content} to inline SVGs into documents directly, useful
# for styling with CSS, preserving interactivity, and improving
# search by indexing diagram text.
embed_format: '<img alt="{img_alt}" src="https://github.com/LukeCarrier/mkdocs-drawio-exporter/raw/main/{img_src}">'
# Glob pattern for matching source files
sources: '*.drawio'
With the plugin configured, you can now proceed to embed images by simply embedding the *.drawio
diagram file as you would with any image file:
![My alt text](my-diagram.drawio)
If you're working with multi-page documents, append the index of the page as an anchor in the URL:
![Page 1](my-diagram.drawio#0)
The plugin will export the diagram to the format
specified in your configuration and will rewrite the <img>
tag in the generated page to match. To speed up your documentation rebuilds, the generated output will be placed into cache_dir
and then copied to the desired destination. The cached images will only be updated if the source diagram's modification date is newer than the cached export. Thus, bear in mind caching works per file - with large multi-page documents a change to one page will rebuild all pages, which will be slower than separate files per page.
See this guide.
In addition to the above, if you're running in a headless environment (e.g. in integration, or inside a Docker container), you may need to ensure a display server is running and that the necessary dependencies are installed.
On Debian and Ubuntu, the following should install the dependencies:
sudo apt install libasound2 xvfb
To run MkDocs with an automatically assigned X display, wrap the command as follows:
xvfb-run -a mkdocs build
If you're seeing messages like the following it's likely that you're running MkDocs as root:
[22:0418/231827.169035:FATAL:electron_main_delegate.cc(211)] Running as root without --no-sandbox is not supported. See https://crbug.com/638180.
If possible, consider running MkDocs as a non-privileged user. Depending on the circumstances (e.g. running within an unprivileged container) it may be appropriate to disable the Chrome sandbox by adding the following option to mkdocs.yml
:
plugins:
- drawio-exporter:
drawio_args:
- --no-sandbox
Optionally, use Nix for a development shell with all the necessary dependencies:
nix develop
We use Poetry for dependency management:
poetry install
To get completion working in your editor, set up a virtual environment in the root of this repository and install MkDocs:
poetry install --with dev
To install the plugin onto a local MkDocs site in editable form:
poetry add --editable /path/to/mkdocs-drawio-exporter
Note that you'll need to repeat this step if you make any changes to the [tool.poetry.plugins.*]
sections listed in pyproject.toml
.
Run the tests with the test
script:
poetry run test
To upgrade the dependencies, first make any necessary changes to the constraints expressed in the [tool.poetry.dependencies]
section of pyproject.toml
, then have Poetry update them:
poetry update
Build the distributable package:
poetry build
Push it to the PyPI test instance:
twine upload --repository-url https://test.pypi.org/legacy/ dist/*
Test it inside a virtual environment:
pip install --index-url https://test.pypi.org/simple/ --no-deps mkdocs-drawio-exporter
Let's go live:
twine upload dist/*