Build Status codecov Pub CommonMark spec

mdown is fast and CommonMark-compliant Markdown parser.

Basic usage:

print(markdownToHtml('# Hello world!'));

Project main goal is create processing library for Markdown.


As there are not many Markdown parsers written in Dart out there, parsing speed is compared with markdown package. Progit was used as a source of markdown files in different languages. mdown appears to be 3 times faster in VM, 11 times faster in Chrome, 2.2 times faster in Safari and 3.7 times faster in Firefox.

Run benchmarks yourself or see details.

mdown make extensive use of String.codeUnitAt instead of RegExp. So you can see noticeable gain for non-latin languages (up to ×38 in Chrome for Japan language).


mdown supports some language extensions. You can specify enabled extensions using options parameter in markdownToHtml.

Options options = const Options(superscript: true);
String res = markdownToHtml('Hello world!\n===', options);

There three predefined sets of options:

  • Options.strict: all extensions are disabled
  • Options.commonmark: only smartPunctuation extension is enabled.
  • Options.defaults: smartPunctuation, strikeout, subscript, superscript, pipeTables, texMathDollars, rawTex are enabled.

To get correspondent parser/writer instance use static getter on class:

String res = markdownToHtml('Hello world!\n===', Options.strict);

If second parameter is not provided, Options.defaults is used.

Smart punctuation (Options.smartPunctuation)

Smart punctuation is automatic replacement of ..., ---, --, " and ' to "…", "—", "–" and curly versions of quote marks accordingly. It's only official extension to date.

NOTE: This extension uses Unicode chars. Make sure that your code supports it.

Extended attributes for fenced code (Options.fencedCodeAttributes)

Allows fenced code block to have arbitrary extended attributes.

``` {#someId .class1 .class2 key=value}

This will be rendered in HTML as

<pre id="someId" class="class1 class2" key="value"><code>code

Extended attributes for headings (Options.headingAttributes)

Allows headings to have arbitrary extended attributes.

# Heading 1 {#someId}

Heading 2 {.someClass}

This will be rendered in HTML as

<h1 id="someId">Heading 1</h1>
<h2 class="someClass">Heading 2</h2>

Extended attributes for inline code (Options.inlineCodeAttributes)

Adds extended attributes support to inline code.

`code`{#id .class key='value'}

Extended attributes for links and images (Options.linkAttributes)

Extended attributes for links and images. Both inline and reference links are supported.

![](image.jpg){width="800" height="600"}


[ref]: {#id}

This will be transformed into:

<p><img src="image.jpg" alt="" width="800" height="600" /></p>
<p><a href="" id="id">test</a></p>

Strikeout (Options.strikeout)

Strikeouts text (~~like this~~). Just wrap text with double tildes (~~).

Strikeouts text (~~like this~~).

Subscript (Options.subscript)

Support for subscript (H2O). Wrap text with tildes (~).


Subscript couldn't contain spaces. If you need to insert space into the subscript, escape space (\).

subscript~with\ spaces~

Superscript (Options.superscript)

Support for superscript (22=4). Wrap text with carets (^).


Superscript couldn't contain spaces. If you need to insert space into superscript, escape space (\).

superscript^with\ spaces^

Pipe tables (Options.pipeTables)

Allows to parse tables where cells are separated with vertical bars (|). Compatible with GitHub table syntax.

head | cells
body | cells
more | cells

Also supports cells alignment.

left aligned | center aligned | right aligned

TeX Math between dollars (Options.texMathDollars)

Anything between two $ characters will be treated as inline TeX math. The opening $ must have a non-space character immediately to its right, while the closing $ must have a non-space character immediately to its left, and must not be followed immediately by a digit. Thus, $20,000 and $30,000 won’t parse as math. If for some reason you need to enclose text in literal $ characters, backslash-escape them and they won’t be treated as math delimiters.

Anything between two $$ will be treated as display TeX math.

HTML writer generates markup for MathJax library. I.e. wraps content with \(...\) or \[...\] and additionally wraps it with <span class="math inline"> or <span class="math display">. If you need custom classes for span you can override them with Options.inlineTexMathClasses and Options.displayTexMathClasses.

TeX Math between backslashed () or [] (Options.texMathSingleBackslash)

Causes anything between \( and \) to be interpreted as inline TeX math and anything between \[ and \] to be interpreted as display TeX math.

NOTE 1: This extension breaks escaping of ( and [].

NOTE 2: This extension is disabled by default.

TeX Math between double backslashed () or [] (Options.texMathDoubleBackslash)

Causes anything between \\( and \\) to be interpreted as inline TeX math and anything between \\[ and \\] to be interpreted as display TeX math.

NOTE: This extension is disabled by default.

Raw TeX (Options.rawTex)

Allows to include raw TeX blocks into documents. Right now only environment blocks are supported. Everything between \begin{...} and \end{...} is treated as TeX and passed into resulting HTML as is.

Custom reference resolver

Custom reference resolver may be required when parsing document without implicitly defined references, for example, Dartdoc.

 * Throws a [StateError] if ...
 * similar to [anotherMethod], but ...

In that case, you could supply parser with the resolver, which should provide all missing links.

import 'package:mdown/mdown.dart';
import 'package:mdown/ast/standard_ast_factory.dart';

String library = "mdown";
String version = "0.11.0";
Target linkResolver(String normalizedReference, String reference) {
  if (reference.startsWith("new ")) {
    String className = reference.substring(4);
        "$library/$version/index.html#$li[email protected]_$className-",
  } else {
    return null;

String res = markdownToHtml('Hello world!\n===', new Options(linkResolver: linkResolver));