Redcarpet is written with sugar, spice and everything nice

Build Status Dependency Status

Redcarpet is a Ruby library for Markdown processing that smells like butterflies and popcorn.

This library is written by people

Redcarpet was written by Vicent Martí. It is maintained by Robin Dupret and Matt Rogers.

Redcarpet would not be possible without the Sundown library and its authors (Natacha Porté, Vicent Martí, and its many awesome contributors).

You can totally install it as a Gem

Redcarpet is readily available as a Ruby gem. It will build some native extensions, but the parser is standalone and requires no installed libraries. Starting with Redcarpet 3.0, the minimum required Ruby version is 1.9.2 (or Rubinius in 1.9 mode).

$ [sudo] gem install redcarpet

If you need to use it with Ruby 1.8.7, you will have to stick with 2.3.0:

$ [sudo] gem install redcarpet -v 2.3.0

The Redcarpet source is available at GitHub:

$ git clone git://

And it’s like really simple to use

The core of the Redcarpet library is the Redcarpet::Markdown class. Each instance of the class is attached to a Renderer object; the Markdown class performs parsing of a document and uses the attached renderer to generate output.

The Redcarpet::Markdown object is encouraged to be instantiated once with the required settings, and reused between parses.

~~~~~ ruby

Initializes a Markdown parser

markdown =, extensions = {})


Here, the renderer variable refers to a renderer object, inheriting from Redcarpet::Render::Base. If the given object has not been instantiated, the library will do it with default arguments.

Rendering with the Markdown object is done through Markdown#render. Unlike in the RedCloth API, the text to render is passed as an argument and not stored inside the Markdown instance, to encourage reusability. Example:

~~~~~ ruby markdown.render(“This is bongos, indeed.”)

=> “

This is bongos, indeed.


You can also specify a hash containing the Markdown extensions which the parser will identify. The following extensions are accepted:

  • :no_intra_emphasis: do not parse emphasis inside of words. Strings such as foo_bar_baz will not generate <em> tags.

  • :tables: parse tables, PHP-Markdown style.

  • :fenced_code_blocks: parse fenced code blocks, PHP-Markdown style. Blocks delimited with 3 or more ~ or backticks will be considered as code, without the need to be indented. An optional language name may be added at the end of the opening fence for the code block.

  • :autolink: parse links even when they are not enclosed in <> characters. Autolinks for the http, https and ftp protocols will be automatically detected. Email addresses and http links without protocol, but starting with www are also handled.

  • :disable_indented_code_blocks: do not parse usual markdown code blocks. Markdown converts text with four spaces at the front of each line to code blocks. This option prevents it from doing so. Recommended to use with fenced_code_blocks: true.

  • :strikethrough: parse strikethrough, PHP-Markdown style. Two ~ characters mark the start of a strikethrough, e.g. this is ~~good~~ bad.

  • :lax_spacing: HTML blocks do not require to be surrounded by an empty line as in the Markdown standard.

  • :space_after_headers: A space is always required between the hash at the beginning of a header and its name, e.g. #this is my header would not be a valid header.

  • :superscript: parse superscripts after the ^ character; contiguous superscripts are nested together, and complex values can be enclosed in parenthesis, e.g. this is the 2^(nd) time.

  • :underline: parse underscored emphasis as underlines. This is _underlined_ but this is still *italic*.

  • :highlight: parse highlights. This is ==highlighted==. It looks like this: <mark>highlighted</mark>

  • :quote: parse quotes. This is a "quote". It looks like this: <q>quote</q>

  • :footnotes: parse footnotes, PHP-Markdown style. A footnote works very much like a reference-style link: it consists of a marker next to the text (e.g. This is a sentence.[^1]) and a footnote definition on its own line anywhere within the document (e.g. [^1]: This is a footnote.).


~~~ruby markdown =, autolink: true, tables: true) ~~~~~

Darling, I packed you a couple renderers for lunch

Redcarpet comes with two built-in renderers, Redcarpet::Render::HTML and Redcarpet::Render::XHTML, which output HTML and XHTML, respectively. These renderers are actually implemented in C and hence offer brilliant performance — several degrees of magnitude faster than other Ruby Markdown solutions.

All the rendering flags that previously applied only to HTML output have now been moved to the Redcarpet::Render::HTML class, and may be enabled when instantiating the renderer:

~~~~~ ruby = {})


Initializes an HTML renderer. The following flags are available:

  • :filter_html: do not allow any user-inputted HTML in the output.

  • :no_images: do not generate any <img> tags.

  • :no_links: do not generate any <a> tags.

  • :no_styles: do not generate any <style> tags.

  • :escape_html: escape any HTML tags. This option has precedence over :no_styles, :no_links, :no_images and :filter_html which means that any existing tag will be escaped instead of being removed.

  • :safe_links_only: only generate links for protocols which are considered safe.

  • :with_toc_data: add HTML anchors to each header in the output HTML, to allow linking to each section.

  • :hard_wrap: insert HTML <br> tags inside paragraphs where the original Markdown document had newlines (by default, Markdown ignores these newlines).

  • :xhtml: output XHTML-conformant tags. This option is always enabled in the Render::XHTML renderer.

  • :prettify: add prettyprint classes to <code> tags for google-code-prettify.

  • :link_attributes: hash of extra attributes to add to links.


~~~~~ ruby renderer = true, hard_wrap: true)


The HTML renderer has an alternate version, Redcarpet::Render::HTML_TOC, which will output a table of contents in HTML based on the headers of the Markdown document.

When instantiating this render object, you can optionally pass a nesting_level option which takes an integer and allows you to make it render only headers until a specific level.

Redcarpet also includes a plaintext renderer, Redcarpet::Render::StripDown, that strips out all the formatting:

~~~~ ruby require ‘redcarpet’ require ‘redcarpet/render_strip’

markdown =

markdown.render(“This is an example.”)

=> “This is an example (”


And you can even cook your own

Custom renderers are created by inheriting from an existing renderer. The built-in renderers, HTML and XHTML may be extended as such:

~~~~~ ruby

Create a custom renderer that sets a custom class for block-quotes.

class CustomRender < Redcarpet::Render::HTML def block_quote(quote) %


) end end

markdown =, fenced_code_blocks: true)


But new renderers can also be created from scratch by extending the abstract base class Redcarpet::Render::Base (see lib/redcarpet/render_man.rb for an example implementation of a Manpage renderer):

~~~~~~ ruby class ManPage < Redcarpet::Render::Base # you get the drill – keep going from here end


The following instance methods may be implemented by the renderer:

Block-level calls

If the return value of the method is nil, the block will be skipped. Therefore, make sure that your renderer has at least a paragraph method implemented. If the method for a document element is not implemented, the block will be skipped.


~~~~ ruby class RenderWithoutCode < Redcarpet::Render::HTML def block_code(code, language) nil end end


  • block_code(code, language)
  • block_quote(quote)
  • block_html(raw_html)
  • footnotes(content)
  • footnote_def(content, number)
  • header(text, header_level)
  • hrule()
  • list(contents, list_type)
  • list_item(text, list_type)
  • paragraph(text)
  • table(header, body)
  • table_row(content)
  • table_cell(content, alignment)

Span-level calls

A return value of nil will not output any data. If the method for a document element is not implemented, the contents of the span will be copied verbatim:

  • autolink(link, link_type)
  • codespan(code)
  • double_emphasis(text)
  • emphasis(text)
  • image(link, title, alt_text)
  • linebreak()
  • link(link, title, content)
  • raw_html(raw_html)
  • triple_emphasis(text)
  • strikethrough(text)
  • superscript(text)
  • underline(text)
  • highlight(text)
  • quote(text)
  • footnote_ref(number)

Note: When overriding a renderer’s method, be sure to return a HTML element with a level that matches the level of that method (e.g. return a block element when overriding a block-level callback). Otherwise, the output may be unexpected.

Low level rendering

  • entity(text)
  • normal_text(text)

Header of the document

Rendered before any another elements:

  • doc_header()

Rendered after all the other elements:

  • doc_footer()


Special callback: preprocess or postprocess the whole document before or after the rendering process begins:

  • preprocess(full_document)
  • postprocess(full_document)

You can look at “How to extend the Redcarpet 2 Markdown library?” for some more explanations.

Also, now our Pants are much smarter

Redcarpet 2 comes with a standalone SmartyPants implementation. It is fully compliant with the original implementation. It is the fastest SmartyPants parser there is, with a difference of several orders of magnitude.

The SmartyPants parser can be found in Redcarpet::Render::SmartyPants. It has been implemented as a module, so it can be used standalone or as a mixin.

When mixed with a Renderer class, it will override the postprocess method to perform SmartyPants replacements once the rendering is complete.

~~~~ ruby


class HTMLWithPants < Redcarpet::Render::HTML include Redcarpet::Render::SmartyPants end



Oh SmartyPants, you're so crazy...



SmartyPants works on top of already-rendered HTML, and will ignore replacements inside the content of HTML tags and inside specific HTML blocks such as <code> or <pre>.

What? You really want to mix Markdown renderers?

Redcarpet used to be a drop-in replacement for Redcloth. This is no longer the case since version 2 – it now has its own API, but retains the old name. Yes, that does mean that Redcarpet is not backwards-compatible with the 1.X versions.

Each renderer has its own API and its own set of extensions: you should choose one (it doesn’t have to be Redcarpet, though that would be great!), write your software accordingly, and force your users to install it. That’s the only way to have reliable and predictable Markdown output on your program.

Markdown is already ill-specified enough; if you create software that is renderer-independent, the results will be completely unreliable!

Still, if major forces (let’s say, tornadoes or other natural disasters) force you to keep a Markdown-compatibility layer, Redcarpet also supports this:

~~~~~ ruby require ‘redcarpet/compat’


Requiring the compatibility library will declare a Markdown class with the classical RedCloth API, e.g.

~~~~~ ruby‘this is my text’).to_html


This class renders 100% standards compliant Markdown with 0 extensions. Nada. Don’t even try to enable extensions with a compatibility layer, because that’s a maintenance nightmare and won’t work.

On a related topic: if your Markdown gem has a lib/markdown.rb file that monkeypatches the Markdown class, you’re a terrible human being. Just saying.

Copyright (c) 2011-2016, Vicent Martí

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.