Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Parser for Markdown and Markdown Extra derived from the original by John Gruber.

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.

PHP Markdown

PHP Markdown Lib 1.3-beta1 - Sun 13 Jan 2013

by Michel Fortin

based on Markdown by John Gruber


This is a library package that includes both PHP Markdown and its sibling PHP Markdown Extra which additional features.

Markdown is a text-to-HTML conversion tool for web writers. Markdown allows you to write using an easy-to-read, easy-to-write plain text format, then convert it to structurally valid XHTML (or HTML).

"Markdown" is two things: a plain text markup syntax, and a software tool, written in Perl, that converts the plain text markup to HTML. PHP Markdown is a port to PHP of the original Markdown program by John Gruber.

PHP Markdown can work as a plug-in for WordPress, as a modifier for the Smarty templating engine, or as a replacement for Textile formatting in any software that supports Textile.

Full documentation of Markdown's syntax is available on John's Markdown page:

Installation and Requirement

This library package requires PHP 5.3 or later.

Note: The older plugin/library hybrid package for PHP Markdown and PHP Markdown Extra is still maintained and will work with PHP 4.0.5 and later.

Before PHP 5.3.7, pcre.backtrack_limit defaults to 100 000, which is too small in many situations. You might need to set it to higher values. Later PHP releases defaults to 1 000 000, which is usually fine.


You can use PHP Markdown easily in your PHP program. This library package is meant to be used with autoloading, so putting the 'Michelf' folder in your include path should be enough for this to work:

use \Michelf\Markdown;
$my_html = Markdown::defaultTransform($my_text);

PHP Markdown Extra is also available the same way:

use \Michelf\MarkdownExtra;
$my_html = MarkdownExtra::defaultTransform($my_text);

If you wish to use PHP Markdown with another text filter function built to parse HTML, you should filter the text after the Markdown function call. This is an example with [PHP SmartyPants][psp]:

use \Michelf\Markdown, \Michelf\SmartyPants;
$my_html = Markdown::defaultTransform($my_text);
$my_html = SmartyPants::defaultTransform($my_html);

All these examples are using the static markdown function found inside the parser class. If you want to customize the parser, you can also instantiate it directly and change some configuration variables:

use \Michelf\MarkdownExtra;
$parser = new MarkdownExtra;
$parser->fn_id_prefix = "post22-";
$my_html = $parser->transform($my_text);


By default, PHP Markdown produces XHTML output for tags with empty elements. E.g.:

<br />

Markdown can be configured to produce HTML-style tags; e.g.:


To do this, you must set the empty_element_suffix variable of the parser object to ">".


To file bug reports please send email to:

Please include with your report: (1) the example input; (2) the output you expected; (3) the output PHP Markdown actually produced.

If you have a problem where Markdown gives you an empty result, first check that the backtrack limit is not too low by running php --info | grep pcre. See Installation and Requirement above for details.

Version History

Current Lib:

  • Changed HTML output for footnotes: now instead of adding rel and rev attributes, footnotes links have the class name footnote-ref and backlinks footnote-backref. (This change only affect Lib branch.)

  • Corrected namespace capitalization in composer package definition file.

Lib 1.3-beta4 (21 Jan 2013):

  • Changed namespace name from michelf (lowercase) to Michelf (capitalized).

  • Fixed some regular expressions to make PCRE not shout warnings about POSIX collation classes (dependent on your version of PCRE).

Lib 1.3-beta2 (14 Jan 2013):

  • Fixed missing autoloading information for composer.

Lib 1.3-beta1 (13 Jan 2013):

Extra 1.2.6 (13 Jan 2013):

  • Headers can now have a class attribute. You can add a class inside the extra attribute block which can optionally be put after a header:

    ### Header ###  {#id .class1 .class2}

    Spaces between components in the brace is optional.

  • Fenced code blocks can also have a class and an id attribute. If you only need to apply a class (typically to indicate the language of a code snippet), you can write it like this:

    ~~~ html

    or like this:

    ~~~ .html

    There is a new configuration option MARKDOWN_CODE_CLASS_PREFIX you can use if you need to append a prefix to the class name.

    You might also opt to use an extra attribute block just like for headers:

    ~~~ {.html #id .codeclass}

    Note that class names added this way are not affected by the MARKDOWN_CODE_CLASS_PREFIX.

    A code block creates a pre HTML element containing a code element. The code HTML element is the one that receives the attribute. If for some reason you need attributes to be applied to the enclosing pre element instead, you can set the MARKDOWN_CODE_ATTR_ON_PRE configuration variable to true.

  • Fixed an issue were consecutive fenced code blocks containing HTML-like code would confuse the parser.

  • Multiple references to the same footnote are now allowed.

  • Fixed an issue where no_markup mode was ineffective.

1.0.1p (13 Jan 2013):

  • Created a library-style unified package that will work with namespace-based autoloading introduced in PHP 5.3. This is the same parser, but packaged differently. It is available separatedly from the two other available packages, and requires PHP 5.3.

  • The following HTML 5 elements are treated as block elements when at the root of an HTML block: article, section, nav, aside, hgroup, header, footer, and figure. svg too.

  • Fixed an issue where some XML-style empty tags (such as <br/>) were not recognized correctly as such when inserted into Markdown-formatted text.

1.0.1o (8 Jan 2012):

  • Silenced a new warning introduced around PHP 5.3 complaining about POSIX characters classes not being implemented. PHP Markdown does not use POSIX character classes, but it nevertheless trigged that warning.

Extra 1.2.5 (8 Jan 2012):

  • Fixed an issue preventing fenced code blocks indented inside lists items and elsewhere from being interpreted correctly.

  • Fixed an issue where HTML tags inside fenced code blocks were sometime not encoded with entities.

1.0.1n (10 Oct 2009):

  • Enabled reference-style shortcut links. Now you can write reference-style links with less brakets:

    This is [my website].
    [my website]:

    This was added in the 1.0.2 betas, but commented out in the 1.0.1 branch, waiting for the feature to be officialized. But half of the other Markdown implementations are supporting this syntax, so it makes sense for compatibility's sake to allow it in PHP Markdown too.

  • Now accepting many valid email addresses in autolinks that were previously rejected, such as:

    <"Fred Bloggs">
  • Now accepting spaces in URLs for inline and reference-style links. Such URLs need to be surrounded by angle brakets. For instance:

    [link text](<http://url/with space> "optional title")
    [link text][ref]
    [ref]: <http://url/with space> "optional title"

    There is still a quirk which may prevent this from working correctly with relative URLs in inline-style links however.

  • Fix for adjacent list of different kind where the second list could end as a sublist of the first when not separated by an empty line.

  • Fixed a bug where inline-style links wouldn't be recognized when the link definition contains a line break between the url and the title.

  • Fixed a bug where tags where the name contains an underscore aren't parsed correctly.

  • Fixed some corner-cases mixing underscore-ephasis and asterisk-emphasis.

Extra 1.2.4 (10 Oct 2009):

  • Fixed a problem where unterminated tags in indented code blocks could prevent proper escaping of characaters in the code block.

Extra 1.2.3 (31 Dec 2008):

  • In WordPress pages featuring more than one post, footnote id prefixes are now automatically applied with the current post ID to avoid clashes between footnotes belonging to different posts.

  • Fix for a bug introduced in Extra 1.2 where block-level HTML tags where not detected correctly, thus the addition of erroneous <p> tags and interpretation of their content as Markdown-formatted instead of HTML-formatted.

Extra 1.2.2 (21 Jun 2008):

  • Fixed a problem where abbreviation definitions, footnote definitions and link references were stripped inside fenced code blocks.

  • Fixed a bug where characters such as " in abbreviation definitions weren't properly encoded to HTML entities.

  • Fixed a bug where double quotes " were not correctly encoded as HTML entities when used inside a footnote reference id.

1.0.1m (21 Jun 2008):

  • Lists can now have empty items.

  • Rewrote the emphasis and strong emphasis parser to fix some issues with odly placed and overlong markers.

Extra 1.2.1 (27 May 2008):

  • Fixed a problem where Markdown headers and horizontal rules were transformed into their HTML equivalent inside fenced code blocks.

Extra 1.2 (11 May 2008):

  • Added fenced code block syntax which don't require indentation and can start and end with blank lines. A fenced code block starts with a line of consecutive tilde (~) and ends on the next line with the same number of consecutive tilde. Here's an example:

    Hello World!
  • Rewrote parts of the HTML block parser to better accomodate fenced code blocks.

  • Footnotes may now be referenced from within another footnote.

  • Added programatically-settable parser property predef_attr for predefined attribute definitions.

  • Fixed an issue where an indented code block preceded by a blank line containing some other whitespace would confuse the HTML block parser into creating an HTML block when it should have been code.

1.0.1l (11 May 2008):

  • Now removing the UTF-8 BOM at the start of a document, if present.

  • Now accepting capitalized URI schemes (such as HTTP:) in automatic links, such as <HTTP://EXAMPLE.COM/>.

  • Fixed a problem where <> was seen as a horizontal rule instead of an automatic link.

  • Fixed an issue where some characters in Markdown-generated HTML attributes weren't properly escaped with entities.

  • Fix for code blocks as first element of a list item. Previously, this didn't create any code block for item 2:

    *   Item 1 (regular paragraph)
    *       Item 2 (code block)
  • A code block starting on the second line of a document wasn't seen as a code block. This has been fixed.

  • Added programatically-settable parser properties predef_urls and predef_titles for predefined URLs and titles for reference-style links. To use this, your PHP code must call the parser this way:

    $parser = new Markdwon_Parser;
    $parser->predef_urls = array('linkref' => '');
    $html = $parser->transform($text);

    You can then use the URL as a normal link reference:

    [my link][linkref]  
    [my link][linkRef]

    Reference names in the parser properties must be lowercase. Reference names in the Markdown source may have any case.

  • Added setup and teardown methods which can be used by subclassers as hook points to arrange the state of some parser variables before and after parsing.

Extra 1.1.7 (26 Sep 2007):

1.0.1k (26 Sep 2007):

  • Fixed a problem introduced in 1.0.1i where three or more identical uppercase letters, as well as a few other symbols, would trigger a horizontal line.

Extra 1.1.6 (4 Sep 2007):

1.0.1j (4 Sep 2007):

  • Fixed a problem introduced in 1.0.1i where the closing code and pre tags at the end of a code block were appearing in the wrong order.

  • Overriding configuration settings by defining constants from an external before markdown.php is included is now possible without producing a PHP warning.

Extra 1.1.5 (31 Aug 2007):

1.0.1i (31 Aug 2007):

  • Fixed a problem where an escaped backslash before a code span would prevent the code span from being created. This should now work as expected:

    Litteral backslash: \\`code span`
  • Overall speed improvements, especially with long documents.

Extra 1.1.4 (3 Aug 2007):

1.0.1h (3 Aug 2007):

  • Added two properties (no_markup and no_entities) to the parser allowing HTML tags and entities to be disabled.

  • Fix for a problem introduced in 1.0.1g where posting comments in WordPress would trigger PHP warnings and cause some markup to be incorrectly filtered by the kses filter in WordPress.

Extra 1.1.3 (3 Jul 2007):

  • Fixed a performance problem when parsing some invalid HTML as an HTML block which was resulting in too much recusion and a segmentation fault for long documents.

  • The markdown="" attribute now accepts unquoted values.

  • Fixed an issue where underscore-emphasis didn't work when applied on the first or the last word of an element having the markdown="1" or markdown="span" attribute set unless there was some surrounding whitespace. This didn't work:

    <p markdown="1">_Hello_ _world_</p>

    Now it does produce emphasis as expected.

  • Fixed an issue preventing footnotes from working when the parser's footnote id prefix variable (fn_id_prefix) is not empty.

  • Fixed a performance problem where the regular expression for strong emphasis introduced in version 1.1 could sometime be long to process, give slightly wrong results, and in some circumstances could remove entirely the content for a whole paragraph.

  • Fixed an issue were abbreviations tags could be incorrectly added inside URLs and title of links.

  • Placing footnote markers inside a link, resulting in two nested links, is no longer allowed.

1.0.1g (3 Jul 2007):

  • Fix for PHP 5 compiled without the mbstring module. Previous fix to calculate the length of UTF-8 strings in detab when mb_strlen is not available was only working with PHP 4.

  • Fixed a problem with WordPress 2.x where full-content posts in RSS feeds were not processed correctly by Markdown.

  • Now supports URLs containing literal parentheses for inline links and images, such as:


    Such parentheses may be arbitrarily nested, but must be balanced. Unbalenced parentheses are allowed however when the URL when escaped or when the URL is enclosed in angle brakets <>.

  • Fixed a performance problem where the regular expression for strong emphasis introduced in version 1.0.1d could sometime be long to process, give slightly wrong results, and in some circumstances could remove entirely the content for a whole paragraph.

  • Some change in version 1.0.1d made possible the incorrect nesting of anchors within each other. This is now fixed.

  • Fixed a rare issue where certain MD5 hashes in the content could be changed to their corresponding text. For instance, this:

    The MD5 value for "+" is "26b17225b626fb9238849fd60eabdf60".

    was incorrectly changed to this in previous versions of PHP Markdown:

    <p>The MD5 value for "+" is "+".</p>
  • Now convert escaped characters to their numeric character references equivalent.

    This fix an integration issue with SmartyPants and backslash escapes. Since Markdown and SmartyPants have some escapable characters in common, it was sometime necessary to escape them twice. Previously, two backslashes were sometime required to prevent Markdown from "eating" the backslash before SmartyPants sees it:

    Here are two hyphens: \\--

    Now, only one backslash will do:

    Here are two hyphens: \--

Extra 1.1.2 (7 Feb 2007)

  • Fixed an issue where headers preceded too closely by a paragraph (with no blank line separating them) where put inside the paragraph.

  • Added the missing TextileRestricted method that was added to regular PHP Markdown since 1.0.1d but which I forgot to add to Extra.

1.0.1f (7 Feb 2007):

  • Fixed an issue with WordPress where manually-entered excerpts, but not the auto-generated ones, would contain nested paragraphs.

  • Fixed an issue introduced in 1.0.1d where headers and blockquotes preceded too closely by a paragraph (not separated by a blank line) where incorrectly put inside the paragraph.

  • Fixed an issue introduced in 1.0.1d in the tokenizeHTML method where two consecutive code spans would be merged into one when together they form a valid tag in a multiline paragraph.

  • Fixed an long-prevailing issue where blank lines in code blocks would be doubled when the code block is in a list item.

    This was due to the list processing functions relying on artificially doubled blank lines to correctly determine when list items should contain block-level content. The list item processing model was thus changed to avoid the need for double blank lines.

  • Fixed an issue with <% asp-style %> instructions used as inline content where the opening < was encoded as &lt;.

  • Fixed a parse error occuring when PHP is configured to accept ASP-style delimiters as boundaries for PHP scripts.

  • Fixed a bug introduced in 1.0.1d where underscores in automatic links got swapped with emphasis tags.

Extra 1.1.1 (28 Dec 2006)

  • Fixed a problem where whitespace at the end of the line of an atx-style header would cause tailing # to appear as part of the header's content. This was caused by a small error in the regex that handles the definition for the id attribute in PHP Markdown Extra.

  • Fixed a problem where empty abbreviations definitions would eat the following line as its definition.

  • Fixed an issue with calling the Markdown parser repetitivly with text containing footnotes. The footnote hashes were not reinitialized properly.

1.0.1e (28 Dec 2006)

  • Added support for internationalized domain names for email addresses in automatic link. Improved the speed at which email addresses are converted to entities. Thanks to Milian Wolff for his optimisations.

  • Made deterministic the conversion to entities of email addresses in automatic links. This means that a given email address will always be encoded the same way.

  • PHP Markdown will now use its own function to calculate the length of an UTF-8 string in detab when mb_strlen is not available instead of giving a fatal error.

Extra 1.1 (1 Dec 2006)

  • Added a syntax for footnotes.

  • Added an experimental syntax to define abbreviations.

1.0.1d (1 Dec 2006)

  • Fixed a bug where inline images always had an empty title attribute. The title attribute is now present only when explicitly defined.

  • Link references definitions can now have an empty title, previously if the title was defined but left empty the link definition was ignored. This can be useful if you want an empty title attribute in images to hide the tooltip in Internet Explorer.

  • Made detab aware of UTF-8 characters. UTF-8 multi-byte sequences are now correctly mapped to one character instead of the number of bytes.

  • Fixed a small bug with WordPress where WordPress' default filter wpautop was not properly deactivated on comment text, resulting in hard line breaks where Markdown do not prescribes them.

  • Added a TextileRestrited method to the textile compatibility mode. There is no restriction however, as Markdown does not have a restricted mode at this point. This should make PHP Markdown work again in the latest versions of TextPattern.

  • Converted PHP Markdown to a object-oriented design.

  • Changed span and block gamut methods so that they loop over a customizable list of methods. This makes subclassing the parser a more interesting option for creating syntax extensions.

  • Also added a "document" gamut loop which can be used to hook document-level methods (like for striping link definitions).

  • Changed all methods which were inserting HTML code so that they now return a hashed representation of the code. New methods hashSpan and hashBlock are used to hash respectivly span- and block-level generated content. This has a couple of significant effects:

    1. It prevents invalid nesting of Markdown-generated elements which could occur occuring with constructs like *something [link*][1].
    2. It prevents problems occuring with deeply nested lists on which paragraphs were ill-formed.
    3. It removes the need to call hashHTMLBlocks twice during the the block gamut.

    Hashes are turned back to HTML prior output.

  • Made the block-level HTML parser smarter using a specially-crafted regular expression capable of handling nested tags.

  • Solved backtick issues in tag attributes by rewriting the HTML tokenizer to be aware of code spans. All these lines should work correctly now:

    <span attr='`ticks`'>bar</span>
    <span attr='``double ticks``'>bar</span>
    `<test a="` content of attribute `">`
  • Changed the parsing of HTML comments to match simply from <!-- to --> instead using of the more complicated SGML-style rule with paired --. This is how most browsers parse comments and how XML defines them too.

  • <address> has been added to the list of block-level elements and is now treated as an HTML block instead of being wrapped within paragraph tags.

  • Now only trim trailing newlines from code blocks, instead of trimming all trailing whitespace characters.

  • Fixed bug where this:

    [text]( "title" )

    wasn't working as expected, because the parser wasn't allowing for spaces before the closing paren.

  • Filthy hack to support markdown='1' in div tags.

  • _DoAutoLinks() now supports the 'dict://' URL scheme.

  • PHP- and ASP-style processor instructions are now protected as raw HTML blocks.

    <? ... ?>
    <% ... %>
  • Fix for escaped backticks still triggering code spans:

    There are two raw backticks here: \` and here: \`, not a code span

Extra 1.0 - 5 September 2005

  • Added support for setting the id attributes for headers like this:

    Header 1            {#header1}
    ## Header 2 ##      {#header2}

    This only work only for headers for now.

  • Tables will now work correctly as the first element of a definition list. For example, this input:

    :   Header  | Header
        ------- | -------
        Cell    | Cell

    used to produce no definition list and a table where the first header was named ": Header". This is now fixed.

  • Fix for a problem where a paragraph following a table was not placed between <p> tags.

Extra 1.0b4 - 1 August 2005

  • Fixed some issues where whitespace around HTML blocks were trigging empty paragraph tags.

  • Fixed an HTML block parsing issue that would cause a block element following a code span or block with unmatched opening bracket to be placed inside a paragraph.

  • Removed some PHP notices that could appear when parsing definition lists and tables with PHP notice reporting flag set.

Extra 1.0b3 - 29 July 2005

  • Definition lists now require a blank line before each term. Solves an ambiguity where the last line of lazy-indented definitions could be mistaken by PHP Markdown as a new term in the list.

  • Definition lists now support multiple terms per definition.

  • Some special tags were replaced in the output by their md5 hash key. Things such as this now work as expected:

    ## Header <?php echo $number ?> ##

Extra 1.0b2 - 26 July 2005

  • Definition lists can now take two or more definitions for one term. This should have been the case before, but a bug prevented this from working right.

  • Fixed a problem where single column table with a pipe only at the end where not parsed as table. Here is such a table:

    | header
    | ------
    | cell
  • Fixed problems with empty cells in the first column of a table with no leading pipe, like this one:

    header header
  • Code spans containing pipes did not within a table. This is now fixed by parsing code spans before splitting rows into cells.

  • Added the pipe character to the backlash escape character lists.

Extra 1.0b1 (25 Jun 2005)

  • First public release of PHP Markdown Extra.

Copyright and License

PHP Markdown & Extra
Copyright (c) 2004-2013 Michel Fortin
All rights reserved.

Based on Markdown
Copyright (c) 2003-2005 John Gruber
All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

  • Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

  • Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

  • Neither the name "Markdown" nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

This software is provided by the copyright holders and contributors "as is" and any express or implied warranties, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose are disclaimed. In no event shall the copyright owner or contributors be liable for any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited to, procurement of substitute goods or services; loss of use, data, or profits; or business interruption) however caused and on any theory of liability, whether in contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use of this software, even if advised of the possibility of such damage.

Something went wrong with that request. Please try again.