Extend Marko

Here is an example of supporting parsing GitHub wiki links: [[Page 2|Page 2]].

Create a new element

GitHub wiki link is an inline level element. For the difference between block elements and inline elements, please refer to the corresponding section of Commonmark’s spec.

Now subclass marko.inline.InlineElement to create a new element type:

from marko import inline

class GitHubWiki(inline.InlineElement):

    pattern = r'\[\[ *(.+?) *\| *(.+?) *\]\]'
    parse_children = True

Inline elements use the pattern attribute to look for the matches in the text. To get more control of the scan process, consider overriding find() method to return an iterable of matches. If parse_children is True, parser will parse the group given by parse_group of the match to produce inline elements, the default group is 1. See Elements for available attributes and methods to change the parsing behavior.

Now, write the __init__() method to control how the parsed result should map to element attributes. You don’t need to provide the parsed content since it is handled by parser automatically:

class GitHubWiki(inline.InlineElement):

    pattern = r'\[\[ *(.+?) *\| *(.+?) *\]\]'
    parse_children = True

    def __init__(self, match):
        self.target = match.group(2)

About the parsing priority

The parser respects element’s priority attribute to control the parsing precedence. It is 5 by default, which is the same as emphasis, links and images. A higher number means the element will be tried sooner. For elements of the same priority, what comes the first will be parsed:

*This is an [[emphasis*|target]]
# Parsed as: <em>This is an [[emphasis</em>|target]]

If we set a higher priority (e.g. 6), it will be tried sooner:

*This is an <a href="target">emphasis*</a>

About overriding default elements

Sometimes you may want to modify the functionality of existing elements, like changing the parsing process or providing more attributes, and want to replace the old one. In this case, you should add override = True to the element attribute.

Add a new render function

Marko uses mixins to add functionalities to renderer or parser. Parser controls the parsing logic which you don’t need to change at the most of time, while renderer mixins controll how to represent the elements by the element name. In our case:

class WikiRendererMixin(object):

    def render_git_hub_wiki(self, element):
        return '<a href="{}">{}</a>'.format(
            self.escape_url(element.target), self.render_children(element)
        )

Note the method name is composed of render_ prefix and the element name in snake-cased form. The snake case form of GitHubWiki is git_hub_wiki.

The renderer mixins will be combined together with marko’s default base renderer: HTMLRenderer, which you need in most cases, to create a marko.renderer.Renderer instance.

Besides of the HTML renderer, Marko also provides some AST renderers to inspect the parsed AST. They are useful to see how parsing works when you are developing your own parsing algorithm:

• marko.ast_renderer.ASTRenderer: renders elements as JSON objects.

• marko.ast_renderer.XMLRenderer: renders elements as XML format AST.

• marko.ext.latex_renderer.LatexRenderer: renders elements as LaTeX document.

Create an extension object

We need an additional extension object to sum these mixins up. An extension object can be made with the help of marko.helpers.MarkoExtension:

from marko.helpers import MarkoExtension

GitHubWiki = MarkoExtension(
    elements=[GitHubWiki],
    renderer_mixins=[WikiRendererMixin]
)

An optional parser_mixins can be also given if you want to customize the parser. The extension exposes a single object so that it can be distributed as a standalone package. We will come to how to use it in the later sections.

Register the extension

Now you have your own extension ready, let’s register it to the markdown parser:

from marko import Markdown

markdown = Markdown(extensions=[GitHubWiki])
# Alternatively, you can register extensions later.
markdown = Markdown()
markdown.use(GitHubWiki)
print(markdown(text))

Note

The extensions argument, or use() accepts multiple extension objects. You can also call it multiple times. The registration order matters in the way that the last registered has the highest priority in the MRO.

You can also choose a different base parser or renderer by:

markdown = Markdown(renderer=marko.ast_renderer.ASTRenderer)

Let’s have a look at how Marko creates the renderer with the extensions and base renderer class. The same applies for the parser.

Assume you choose HTMLRenderer as the base renderer class and have three extensions A, B, C registered in order:

class A:
    renderer_mixins = [ARendererMixin]

class B:
    renderer_mixins = [BRendererMixin]

class C:
    renderer_mixins = [CRendererMixin]

markdown = Markdown(extensions=[A, B, C])

Then the renderer is created like following:

class MyRenderer(CRendererMixin, BRendererMixin, ARendererMixin, HTMLRenderer):
    pass

Note the order of the multi inheriting.

Publish the extension as package

You can also refer to the extension without actually importing the extension object.

To do so, put a make_extension() function in the entry file which takes any arguments and returns an extension object:

def make_extension(arg):
    return GitHubWiki(arg)

Then you can refer to the extension via import string(assume the package name is marko_github_wiki):

markdown = Markdown(extensions=["marko_github_wiki"])