Distribution · text

Text::Markdown

Works

A Markdown parser with an HTML backend — headings, emphasis, code, quotes and links — written as a grammar, and covering rather less of Markdown than its name suggests.

Version: 1.1.1 zef:JJMERELO
Depends: HTML::Escape
License: MIT
Its own test suite: 5 files, green
Checked: 2026-09-16 against Raku++ 3.28.0 and Rakudo 2026.08
Where it lives: raku.land · source

Install it #

$ rakupp install Text::Markdown

zef install Text::Markdown writes the same store; either installer leaves the module usable by both engines.

What it is for #

parse-markdown returns a document object; .to-html renders it. That is the whole surface for the common case, and it is the reason seven distributions depend on this one rather than shelling out to a C library.

Rendering a document #

File

use Text::Markdown;

my $md = q:to/MD/;
# A heading

Some *emphasis*, some **strong**, and a `literal`.

- one
- two

> a quote

    a code block
MD

say parse-markdown($md).to-html;

Output

<h1>A heading</h1><p>Some <em>emphasis</em>, some <strong>strong</strong>, and a <code>literal</code>.</p><p>- one - two</p><blockquote><p>a quote</p></blockquote><pre><code>a code block</code></pre>

Headings, emphasis, strong, inline code, block quotes and indented code blocks all come out as you would expect.

What it does not do #

Look closely at the list in the middle of that output. The two - items came back as a single paragraph — <p>- one - two</p> — not a <ul>. Unordered lists written with - are not part of what this parser recognises, and neither is the * spelling.

That is worth knowing before you choose it: this is a parser for the original Markdown core, and a good deal of what people now write — lists with -, fenced code blocks with backticks, tables, task lists, anything from CommonMark's later additions — passes through as literal paragraph text rather than failing loudly. The output is valid HTML either way, which is exactly what makes the gap easy to miss.

The document object is inspectable, so a program that needs to know whether a construct was understood can look at the parsed structure instead of the rendered string.

Where the two engines differ #

Nowhere. The rendered HTML is byte-identical on Raku++ and Rakudo, and all five test files pass on both.

One note about the output format rather than the engines: .to-html emits no newlines between block elements — the whole document is one line. That is fine for a browser and awkward for a diff, so if you are snapshot-testing the result, compare parsed structures or normalise the HTML first.