Treatments, Typography, and Markdown for SPIP


A quick and rather hackish way to replace SPIP's built-in text markup format with Markdown. Posted by Thomas Sutton on October 20, 2009

SPIP is a great content management system, but there are a few of it’s features that I dislike. One of the most directly user visible is it’s formatting language for text content (the “typographical shortcuts”). This language is effective enough in many circumstances but I find it a little ugly and it has some fairly important limitations (try doing multiple paragraphs in a list item without hackish <br/>s, for example). Thankfully, the “shortcuts” are implemented as a “treatment” for the #TEXTE, etc. tags and can be overridden quite easily.

In this post I’ll describe how SPIP’s “treatments” work and how they can be overridden. In the process, you’ll see how to replace the built-in typographical shortcuts with my favourite light-weight mark-up language – PHP Markdown Extra and Smartypants – as well as a few less intrusive “tweaks”.

SPIP’s template language is based on the idea of loops – which “loop” over the results of a database query – and “tags” – which output the value of a column from the current row1. If all you want to do is loop over the contents of a database table and output some values then you don’t have to “do” anything:

SPIP will do what you want without any additional code of configuration (assuming that your table is called my_table and has columns called acolumn and another_column). This “default” behaviour (assuming that loops and tags match with tables and columns) makes it very easy to add new loops and tags: just modify the database schema and it all “works”.

There are lots of circumstances, though, where some additional processing will need to be done on values before they are suitable for display. Perhaps some special characters need to be escaped or some formatting applied before the data it is suitable for output. You could just apply a filter to the tag each and every time you use it:

Or implement the tag in PHP:

Thankfully, you don’t have to filter the tag in every template, or implement the tag in PHP to use such “treatments”: you can just tell SPIP what function/s to call with the $table_des_traitements global variable.

The $table_des_traitements variable is an array of arrays (hence the “table”) which SPIP uses to figure out exactly how to process a raw database value into something that is safe to output in a page. The “first” dimension of the array is the tag name – TEXTE, DATE, etc. – and the second is the loop – documents, articles, etc. When SPIP is evaluating a tag (whether or not it has been implemented in PHP), it checks the $table_des_traitements array to see if there is a “treatment” for that tag/loop combination or, failing that, for the tag everywhere (the 0th element in the array). This happens in champs_traitements() and takes the “star” into account (where adding a * to the end of a tag suppresses automatic filtering).

You can see the default values in ecrire/public/interfaces.php, but lets consider the trivial example of removing the leading number from all #TITREs with the supprimer_numero filter (completely ignoring the normal processing):

Another trivial example might be to transform the #TEXTE of all sections (but not articles, etc) to uppercase2 (again ignoring the normal processing):

Getting back to my goal of replacing SPIP’s built-in typographical shortcuts language with the PHP Markdown and Smartypants. The TEXTE entry in interfaces.php has the value 'propre(%s, $connect)'. Replacing this call to propre() (which implements the typographical shortcuts) is simply a matter of installing the markdown.php and smartypants.php scripts and adding a similar line to our mes_options.php file:

Now the value set in interfaces.php will be $table_des_traitements['TEXTE'][1] instead of $table_des_traitements['TEXTE'][0] and our new value will be the default!

Alas, there is one very large drawback to doing this. While I’ve replaced the very limited typographical shortcuts with the much more powerful Markdown, this change has also removed the ability to embed models in the TEXTE of articles, sections, etc. It’s now harder to use images and documents within texte and completely impossible to embed forms and other interactive elements.

This is a serious drawback which I’ve still not figured out how to work around.


  1. Please note that I am grossly oversimplifying things here: loops can loop over other data, and tags can do much, much more than merely output values from a database. But this is the basis for much of the design and implementation and is the default behaviour.

  2. And thus probably breaking your document as XML tag and attribute names are case sensitive and XHTML defines them all as lower-case.

This post was published on October 20, 2009 and last modified on September 4, 2020. It is tagged with: php, spip.