Writing plugins

All flamingo plugins have to be a python class, defining hooks which are simple python methods.

All hooks get the context object passed in. Some hooks get flamingo.Content objects passed in. More informations on Content objects: Flamingo data model

class ExamplePlugin:
    # The name of your plugin is irrelevant to flamingo. You can name it
    # how you want

    def setup(self, context):
        # Gets called on plugin setup and is meant to setup plugin
        # specific state.
        # When the Live-Server is running this hook gets re executed
        # every time a rebuild is triggered.


    def settings_setup(self, context):
        # Gets called when all settings are setup.
        # You can change global settings here.


    def parser_setup(self, context):
        # Gets called after all parsers are setup.
        # Use this hook to register rst directives.


    def templating_engine_setup(self, context, templating_engine):
        # Gets called when the templating engine is ready.
        # Use this hook to register your Jinja2 filters.


    def media_added(self, context, content, media_content):
        # Gets called every time a new media content gets added using
        # context.add_media().
        # You can use this hook to change meta data on media contents or
        # scale images for example.
        # 'media_content' contains the media content that got added,
        # 'content' the related content object.


    def content_parsed(self, context, content):
        # Gets called every time a content file gets parsed.
        # The current content is available in context.content.


    def contents_parsed(self, context):
        # Gets called when all content files in CONTENT_ROOT are parsed.
        # All content objects are available in context.contents.


    def context_setup(self, context):
        # Gets called when the context is set up fully.


    def pre_build(self, context):
        # Gets called as last hook before building.


    def post_build(self, context):
        # Gets called after building.


    # Live-Server specific hooks
    # these hook only run if the Live-Server runs and are meant to be used
    # to increase performance by rendering only contents that have been
    # requested

    def render_content(self, context, content):
        # This hook gets called for every Content object that gets rendered
        # by the Live-Server.


    def render_media_content(self, context, media_content):
        # This hook gets called for every media Content object that gets
        # rendered by the Live-Server.


Flamingo plugins can be loaded from python packages or directly from files. They don't have to be importable or setuptools packaged.

# plugins/example_plugin.py

class ExamplePlugin:
    def contents_parsed(self, context):

# settings.py

Plugins = [

Error reporting

pythons logging module is highly integrated into flamingo and especially into flamingo Live-Server.

See Live-Server documentation for more details.

Define your own hooks

Besides the default hooks, you can define your own hooks by simply running them.


This can be useful if you want to write a complex parser with intermediate states. The reStructuredText parser, for instance, produces a docutils document tree before generating HTML.

# flamingo/plugins/rst/base.py

# [...]
context.plugins.run_hook('rst_document_parsed', document)
# [...]

Plugins with state

For some applications plugins need state. The reStructuredText plugin for example needs to hold state of the current parsed line for correct error reporting.

To setup your plugins state you can use the hook setup(). It gets executed once on setup and every time a rebuild is triggered, if the Live-Server is running.

If you need to access the state of another plugin you can get it by using context.plugins.get_plugin(class_name).

class ExamplePlugin:
    def setup(self, context):
        self.internal_counter = 0

    def content_parsed(self, context, content):
        self.internal_counter += 1

class ExamplePlugin2:
    def content_parsed(self, context, content):
        example_plugin = context.plugins.get_plugin('ExamplePlugin')
        example_plugin.internal_counter += 1


Flamingo uses the same data model for content objects like for content objects. When a media content gets added to a content object, it get stored in content['media'].

Lets say you want to create a plugin that adds an image with the same name as the content it is placed next to.

The context class defines add_media() for this manner. It takes a name, which has to be a path in file system, within your settings.CONTENT_ROOT and a Content object (media is always related to content, not to your project).

If name starts with an /, the path is considered an absolute path, relative to your settings.CONTENT_ROOT. If it doesn't, it has be relative to your content['path'].

You also can add your own custom meta data.

import os

class ExamplePlugin:
    def content_parsed(self, context, content):
        new_media_content = context.add_media(

            # some custom meta data

    def media_added(self, context, content, media_content):
        if media_content['foo']:


If your plugin comes with templates or static files, your plugin can also be a theme.

A theme contains two directories: templates and static.

For example the directory structure of the Tags plugin looks like this:

├── __init__.py
├── tags.py
└── theme
    └── templates
        ├── tag.html
        └── tags.html

To tell flamingo where to search for a theme THEME_PATHS is used:

import os

class Tags:
        THEME_PATHS = [os.path.join(os.path.dirname(__file__), 'theme')]

Live-Server aware plugins

Lets say you have a plugin that generates thumbnails of large images.

class Thumbnails:
    def media_added(self, context, content, media_content):
        generate_thumbnail(media_content['path'], 'image.thumb.jpg')

For plain rendering your project with make html this is totally fine, but when running the Live-Server, this will slow down your the startup of make server by every new image you add.

Flamingo has special hooks to address this.

class Thumbnails:
    def media_added(self, context, content, media_content):
        thumbnail = context.add_media(

            # we use this meta data as hint for later

        # The Live-Server sets this variable on startup
        if context.settings.LIVE_SERVER_RUNNING:

        generate_thumbnail(thumbnail['path'], 'image.thumb.jpg')

    def render_media_content(self, context, media_content):
        # the meta data we added before
        if media_content['thumbnail'] == True:
            if not os.path.exists(media_content['path']):