• Breaking change: Remove ListSyntax.removeLeadingEmptyLine, ListSyntax.removeTrailingEmptyLines, TableSyntax.parseAlignments, TableSyntax.parseRow.
  • Allow intra-word strikethrough in GFM (#300).
  • Breaking change: Change BlockSyntax.canEndBlock from a getter to a method accepting a BlockParser.


  • Deprecate the public methods ListSyntax.removeLeadingEmptyLine, ListSyntax.removeTrailingEmptyLines, TableSyntax.parseAlignments, TableSyntax.parseRow. These will be made private in a major version bump as early as 3.0.0.


  • Add dependency on the meta package


  • Fix for custom link resolvers (#295).
  • Add missing HTML 5 block-level items (#294).


  • Overhaul table row parsing. This does not have many consequences, except that whitespace around escaped pipes is handled better. (#287).


  • Correctly parse a reference link with a newline in the link reference part (#281).


  • Do not encode HTML in link URLs. Also do not encode HTML in link text when encodeHtml is false (e.g. when used in Flutter).


  • Drop support for Dart 2.0.0 through 2.1.0.
  • Recognize Unicode ellipsis (…) and other Unicode punctuation as punctuation when parsing potential emphasis.
  • Reduce time to parse a large HTML-block-free Markdown document (such as that in #271) by more than half.
  • Add a new optional parameter for InlineSyntax(), startCharacter, where a subclass can specify a single character to try to match, before matching with more expensive regular expressions.


  • Fix for encoding HTML for text string that contains <pre> (#263).


  • Improve strict spec compliance of > handling by always encoding as &gt; – unless preceded by /.
  • Improve strict spec compliance for blockquote by always putting the closing tag on a new line.
  • Improve strict spec compliance for code elements defined with “`”.
  • Properly encode <, >, and " as their respective HTML entities when interpreted as text.
  • Improve inline code parsing when using multiple backticks.
  • Do not encode HTML in indented code blocks when encodeHtml is false (e.g. when used in Flutter).


  • Render element attributes in the order they were defined. Aligns more closely with the strict spec definition.
  • Correctly render & within inline image titles.
  • Add 68 new GitHub emojis.
  • Escape HTML attribute for fenced code blocks, in the info string.


  • Set max SDK version to <3.0.0, and adjust other dependencies.


  • Require Dart 2.0.0-dev.


  • Breaking change: The Link class has been renamed LinkReference, and the Document field, refLinks, has been renamed linkReferences.

  • Breaking change: Remove the deprecated ExtensionSet.gitHub field. Use ExtensionSet.gitHubFlavored instead.

  • Breaking change: Make all of the fields on Document read-only.

  • Overhaul support for emphasis (*foo* and _foo_) and strong emphasis (**foo** and __foo__), dramatically improving CommonMark compliance.

  • Overhaul support for links and images, again dramatically improving CommonMark compliance.

  • Improve support for tab characters, and horizontal rules.

  • Add support for GitHub Flavored Markdown's Strikethrough extension. See the GFM spec.

  • The above fixes raise compliance with the CommonMark specs to 93%, and compliance with the GFM specs to 92%.

  • Add an encodeHtml parameter to Document, which defaults to true. When false, HTML entities (such as &copy; and the < character) will not be escaped, useful when rendering Markdown in some output format other than HTML.

  • Allow the binary script to take a --extension-set option.

    A reminder: You can run bin/markdown.dart from anywhere via:

    $ pub global activate markdown
    $ markdown


  • Add support for GitHub's colon-based Emoji syntax. :tada:! This is available in the gitHubWeb extension set.


  • Make the constructor for ExtensionSet public, for tools like dartdoc.
  • Split the gitHub ExtensionSet into two sets: gitHubFlavored, which represents the GitHub Flavored Markdown spec, and gitHubWeb, which represents what GitHub actually renders Markdown.


  • Fix issue where accept could cause an exception.
  • Remove deprecated escapeHtml function.
  • Fix compliance with auto-links, including support for email addresses.
  • Updated ExtensionSet.gitHub to more closely align with GitHub markdown.


  • Fix bug with lazy blockquote continuations (#162)
  • Fix bug with list item continuations (#156)


  • Deprecate escapeHtml. This code exists in dart:convert.


  • Fix reference code links inside blockquotes.
  • Add src/util.dart to exports.


  • Add version information:
    • dart bin/markdown.dart --version now shows the package version number.
    • The playground app now shows the version number.
  • Improve autolink parsing.
  • Add new table syntax: TableSyntax.
  • Add new ExtensionSet that includes the table syntax: ExtensionSet.gitHub.
  • For development: added tool/travis.sh.
  • Support multiline Setext headers.
  • Handle loose-vs-strict list items better.
  • Support ordered lists that start with a number other than 1.



  • Parse HTML blocks more accurately, according to CommonMark.
  • Support shortcut reference links.
  • Don't allow an indented code block to interrupt a paragraph.
  • Change definition of “loose” and “strict” lists (items wrapped in paragraph tags vs not) to CommonMark's. The primary difference is that any single list item can trigger the entire list to be marked as “loose”, rather than defining “looseness” on each specific item.
  • Fix paragraph continuations in blockquotes and list items.
  • Fix silly typing bug with tool/common_mark_stats.dart, which resulted in a dramatic overestimate of our CommonMark compliance.
  • There are now 427/613 (69%) passing CommonMark v0.25 specs.


  • Parse hard line breaks properly (#86). Thanks @mehaase!
  • Fix processing of [ ... ] syntax when no resolver is specified (#92).
  • There are now 401/613 (65%) passing CommonMark v0.24 specs. (Actually: after 0f64c8f the actual number of passing tests was 352/613 (57%).)


  • BREAKING: Now following the CommonMark spec for fenced code blocks. If a language (info string) is provided, it is added as a class to the code element with a language- prefix.
  • BREAKING: Now following the CommonMark spec for images. Previously, ![text](img.png) would compile too <a href="img.prg"><img src="img.prg" alt="text"></img></a>. That same code will now compile to <img src="img.png" alt="text" />.
  • Fix all strong mode errors.


  • BREAKING: The text [foo] (bar) no longer renders as an inline link (#53).
  • BREAKING: Change list parsing to allow lists to begin immediately after a preceding block element, without a blank line in between.
  • Formalize an API for Markdown extensions (#43).
  • Introduce ExtensionSets. FencedCodeBlock is considered an extension, but existing usage of markdownToHtml() and new Document() will use the default extension set, which is ExtensionSet.commonMark, which includes FencedCodeBlock.
  • Inline HTML syntax support; This is also considered an extension (#18).
  • The text [foo]() now renders as an inline link.
  • Whitespace now allowed between a link's destination and title (#65).
  • Header identifier support in the HeaderWithIdSyntax and SetextHeaderWithIdSyntax extensions.
  • Implement backslash-escaping so that Markdown syntax can be escaped, such as [foo]\(bar) ==> <p>[foo](bar)</p>.
  • Support for hard line breaks with either \\\n or \n (#30, #60).
  • New public method for BlockParser: peek(int linesAhead), meant for use in subclasses.
  • New public members for ListSyntax: blocksInList and determineBlockItems(), meant for use in subclasses.
  • Improve public docs (better, and more of them).


  • Breaking: Remove (probably unused) fields: LinkSyntax.resolved, InlineParser.currentSource.
  • Switch tests to use test instead of unittest.
  • Fix a few bugs in inline code syntax.
  • Ignore underscores inside words (#41).


  • Allow resolving links that contain inline syntax (#42).


  • Updated homepage.


  • Formatted code.

  • Updated readme.