Python-Markdown 2.1 Release Notes

We are pleased to release Python-Markdown 2.1 which makes many improvements on 2.0. In fact, we consider 2.1 to be what 2.0 should have been. While 2.1 consists mostly of bug fixes, bringing Python-Markdown more inline with other implementations, some internal improvements were made to the parser, a few new built-in extensions were added, and HTML5 support was added.

Python-Markdown supports Python versions 2.4, 2.5, 2.6, 2.7, 3.1, and 3.2 out of the box. In fact, the same code base installs on Python 3.1 and 3.2 with no extra work by the end user.

Backwards-incompatible Changes

While Python-Markdown has received only minor internal changes since the last release, there are a few backward-incompatible changes to note:

  • Support had been dropped for Python 2.3. No guarantees are made that the library will work in any version of Python lower than 2.4. Additionally, while the library had been tested with Python 2.4, consider Python 2.4 support to be depreciated. It is not likely that any future versions will continue to support any version of Python less than 2.5. Note that Python 3.0 is not supported due to a bug in its 2to3 tool. If you must use Python-Markdown with Python 3.0, it is suggested you manually use Python 3.1’s 2to3 tool to do a conversion.

  • Python-Markdown previously accepted positional arguments on its class and wrapper methods. It now expects keyword arguments. Currently, the positional arguments should continue to work, but the solution feels hacky and may be removed in a future version. All users are encouraged to use keyword arguments as documented in the Library Reference.

  • Past versions of Python-Markdown provided module level Global variables which controlled the behavior of a few different aspects of the parser. Those global variables have been replaced with attributes on the Markdown class. Additionally, those attributes are settable as keyword arguments when initializing a class instance. Therefore, if you were editing the global variables (either by editing the source or by overriding them in your code), you should now set them on the class. See the Library Reference for the options available.

  • If you have been using the HeaderId extension to define custom ids on headers, you will want to switch to using the new Attribute List extension. The HeaderId extension now only auto-generates ids on headers which have not already had ids defined. Note that the Extra extension has been switched to use Attribute Lists instead of HeaderId as it did previously.

  • Some code was moved into the markdown.util namespace which was previously in the markdown namespace. Extension authors may need to adjust a few import statements in their extensions to work with the changes.

  • The command line script name was changed to markdown_py. The previous name (markdown) was conflicting with people (and Linux package systems) who also had markdown.pl installed on there system as markdown.pl’s command line script was also named markdown. Be aware that installing Python-Markdown 2.1 will not remove the old versions of the script with different names. You may want to remove them yourself as they are unlikely to work properly.

What’s New in Python-Markdown 2.1

Three new extensions were added. Attribute Lists, which was inspired by Maruku’s feature of the same name, Newline to Break, which was inspired by GitHub Flavored Markdown, and Smart Strong, which fills a hole in the Extra extension.

HTML5 is now supported. All this really means is that new block level elements introduced in the HTML5 spec are now properly recognized as raw HTML. As valid HTML5 can consist of either HTML4 or XHTML1, there is no need to add a new HTML5 serializers. That said, html5 and xhtml5 have been added as aliases of the html4 and xhtml1 serializers respectively.

An XHTML serializer has been added. Previously, ElementTree’s XML serializer was being used for XHTML output. With the new serializer we are able to avoid more invalid output like empty elements (i.e., <p />) which can choke browsers.

Improved support for Python 3.x. Now when running setupy.py install in Python 3.1 or greater the 2to3 tool is run automatically. Note that Python 3.0 is not supported due to a bug in its 2to3 tool. If you must use Python-Markdown with Python 3.0, it is suggested you manually use Python 3.1’s 2to3 tool to do a conversion.

Methods on instances of the Markdown class that do not return results can now be changed allowing one to do md.reset().convert(moretext).

The Markdown class was refactored so that a subclass could define its own build_parser method which would build a completely different parser. In other words, one could use the basic machinery in the markdown library to build a parser of a different markup language without the overhead of building the markdown parser and throwing it away.

Import statements within markdown have been improved so that third party libraries can embed the markdown library if they desire (licensing permitting).

Added support for Python’s -m command line option. You can run the markdown package as a command line script. Do python -m markdown [options] [args]. Note that this is only fully supported in Python 2.7+. Python 2.5 & 2.6 require you to call the module directly (markdown.__main__) rather than the package (markdown). This does not work in Python 2.4.

The command line script has been renamed to markdown_py which avoids all the various problems we had with previous names. Also improved the command line script to accept input on stdin.

The testing framework has been completely rebuilt using the Nose testing framework. This provides a number of benefits including the ability to better test the built-in extensions and other options available to change the parsing behavior. See the Test Suite documentation for details.

Various bug fixes have been made, which are too numerous to list here. See the commit log for a complete history of the changes.