All text transformations are done with filters. You can think of a filter as a UNIX pipe: text goes in, magic happens, text goes out. You can chain filters, of course. In Acrylamid a filter can convert Markdown into HTML, render MathML from AsciiMathML, apply typographical enhancements or just increase headings by an offset.


There are three ways to apply filters: global, per view and per entry. During compilation, all filters will be chained, conflicting filters such as reST and Markdown at the same time are removed and then ordered by internal priorities [1]. Each entry evaluates its own filter chain.

Acrylamid defines a filter as a simple object containing one or more identifiers [2] and also has a list of filters that would conflict with this filter. Below you can find a complete list of all built-in filters.

global : [HTML, Markdown, ...]
If you prefer Markdown as markup language, you can set this as default filter like FILTERS = [Markdown, ] in your
view : HTML or [Markdown, summarize, ...]

Apply filters to specific views. For example you can provide a summarized feed.

'/rss/': {'view': 'rss', filters: 'summarize'}
entry : Jinja2 or [reST, ...]

Switch between Markdown and reStructuredText? Use one of them as default filter but override it in the given article’s metadata. For convenience both “filter” and “filters” can be used as key. An example YAML front matter:

title: Test
filter: reST

Now if your default filter converts Markdown, this article uses reST.

[1]the evaluation order depends on an internal priority value for each filter so we don’t confuse our users or produce unexpected behavior.
[2]an identifier is the name you use to enable this specific filter, most filters have multiple aliases for the same filter, like “reStructuredText” which you can also enable with “rst” or “reST”.

Additional Arguments

Filters may take additional arguments to use extensions like Markdown’s code-hilighting. Not every filter supports additional arguments, please refer to the actual filter documentation.


A normal usage of explicit filters in an article:

title: We write reStructuredText
filters: [reST, hyphenate]

With reStructuredText I can write :math:`x^2`, that's pretty cool, eh?

Filters with arguments:

filters: [markdown+mathml, summarize+100]
filters: [markdown+mathml+codehilite(css_class=highlight), ...]

Disabling Filters

Sometimes it is useful to disable filters per entry or per view. You can annul filters you have applied globally in per view filters or entry metadata. The syntax is the filter name (without any arguments) prefixed with “no”:

filters: nosummary

Built-in Filters

Acrylamid ships with good maintained filters but you are not restricted only to them. Simply create a directory like filters/ and add FILTERS_DIR += ['filters/'] to your and use your own filters. See Custom Filters.

A quick note to the following tables:

  • Requires indicates what you have to install to use this filter.
  • Alias is a list of alternate identifiers to this filter.
  • Conflicts shows what filters don’t work together (does not conflict if empty).
  • Arguments what arguments you can apply to this filter.


Lets you write using Markdown which simplifies HTML generation and is a lot easier. The Markdown filter uses the official implementation by John Gruber and all it’s available extensions. Note that pygments is required for codehilite.

Here’s an online service converting Markdown to HTML and providing a handy cheat sheet: Dingus.

Acrylamid features some additional extension:

  • inline math via AsciiMathML. The aliases are: asciimathml, mathml and math and require the python-asciimathml package. Note put your formula into single dollar signs like $a+b^2$!
  • super und subscript via sup or superscript as well as sub or subscript. The syntax for subscript is H~2~O and for superscript a^2^.
  • deletion and insertion syntax via delins. The syntax is ~~old~~ and ++new.
Requires markdown or (python-markdown) – already as a dependency implicitly installed
Aliases md, mkdown, markdown
Conflicts HTML, reStructuredText, Pandoc
Arguments asciimathml, sub, sup, delins, <built-in extensions>


reStructuredText lets you write (like the name says) in reStructuredText syntax instead of HTML and is more powerful and reliable than Markdown but also slower and slightly more difficult to use. See their quickref for syntax details.

Using a decent version of docutils (≥ 0.8) let you also write inline math with a subset of LaTeX math syntax, so there is no need of an additional extension like in Markdown. In addition to all standard built-in directives, acrylamid offers three additional one:

  • Pygments syntax highlighting via code-block, sourcecode or pygments. Here’s an example (linenos enables line numbering):

    .. code-block:: python
      #!/usr/bin/env python
      print "Hello World!
  • JavaScript-enabled syntax highlighting via code and additional scripts:

    .. source:: python
       #!/usr/bin/env python
       print "Hello, World!"
    .. raw:: html
        <script type="text/javascript" src=""></script>
        <script type="text/javascript" src=""></script>
        <link type="text/css" rel="stylesheet" href=""/>
        <script type="text/javascript">SyntaxHighlighter.defaults.toolbar=false; SyntaxHighlighter.all();</script>
  • GitHub:Gist embedding via gist optionally with a filename.

    .. gist:: 4145152
       :file: transmission.rb
  • YouTube directive for easy embedding (:options: are optional).

    .. youtube:: ZPJlyRv_IGI
       :start: 34
       :align: center
       :height: 1280
       :width: 720
  • Vimeo directive for easy embedding (:options: are optional).

    .. vimeo:: 6455561
       :align: center
       :height: 1280
       :width: 720
       :border: 1px
       :color: ffffff
Requires docutils (or python-docutils), optional pygments for syntax highlighting
Aliases rst, rest, reST, restructuredtext
Conflicts HTML, Markdown, Pandoc


A textile filter if like the textile markup language. Note, that the python implementation of Textile has been not actively maintained for more than a year. Textile is the only text processor so far that adds some typographical enhancements automatically (but not every applied via typography).

Requires textile
Aliases Textile, textile, pytextile, PyTextile
Conflicts HTML, Markdown, Pandoc, reStructuredText


This is filter is a universal converter for various markup language such as Markdown, reStructuredText, Textile and LaTeX (including special extensions by pandoc) to HTML. A typical call would look like filters: [pandoc+Markdown+mathml+...]. You can find a complete list of pandocs improved (and bugfixed) Markdown implementation in the Pandoc User’s Guide.

Requires Pandoc – a universal document converter in PATH
Aliases Pandoc, pandoc
Conflicts reStructuredText, HTML, Markdown
Arguments First argument is the FORMAT like Markdown, textile and so on. All arguments after that are applied as additional long-opts to pandoc.


Discount – a C implementation of John Gruber’s Markdown including definition lists, pseudo protocols and Smartypants (makes typography obsolete).

Requires discount
Aliases Discount, discount
Conflicts reStructuredText, Markdown, Pandoc, PyTextile, Typography


No transformation applied. Useful if your text is already written in HTML.

Requires <built-in>
Aliases pass, plain, html, xhtml, HTML
Conflicts reStructuredText, Markdown, Pandoc

h, head_offset

This filter increases HTML headings tag by N whereas N is the suffix of this filter, e.g. h2 increases headers by two.

Requires <built-in>
Aliases h1, h2, h3, h4, h5


Summarizes content to make listings of text previews (used in tag/page by default). You can customize the ellipsis, CSS-class, link-text and the behaviour how the link appears in your Configuration. You can override the maximum words per entry using summarize.maxwords: 10 in your metadata.

With <!-- break --> you can end the summarizing process preliminary. For convenience excerpt, summary and more will also work as keyword.

Requires <built-in>
Aliases sum
Arguments Maximum words in summarize (an Integer), defaults to summarize+200.

You can define the following additional Configuration parameters for the summarize filter. You can overwrite all configuration values per entry with, summarize.mode and summarize.ignore.

SUMMARIZE_MODE : an integer value

  • 0 – inject the link directly after the tag, which content has exceeded maxwords.
  • 1 – inject link after certain blacklisted tags such as pre, a and b to avoid accidental miss-interpretion of the continuation link.
  • 2 – close currently open tags and insert link afterwards.

SUMMARIZE_LINK : a continuation string with a %s inside

String template for the continue reading link. Default uses an ellipsis (three typographical dots, …), a link with the css class continue and the text continue and a single dot afterwards. This string must contain %s where the link location will be inserted.

SUMMARIZE_IGNORE : a list of tags

Ignores given self-closed HTML tags in the output generation, defaults to ['img', 'video', 'audio']. With [] you can disable this behavior.


This filter is an alternative to the summarize filter mentioned above. With the latter it is harder to control what is shown in the entry listings; sometimes headings also appear in the summary if the first paragraph is short enough. This filter shows only up to N paragraphs.

You can overwrite the amount of paragraphs shown in each entry using intro.maxparagraphs: 3 in the metadata section.

Requires <built-in>
Arguments Maximum paragraphs (an Integer), defaults to intro+1.

Additional Configuration parameters for the introduction filter. You can overwrite both configuration values per entry with and intro.ignore respectively.

INTRO_LINK : a string (may be empty)

Same default value and usage like the SUMMARIZE_LINK but you can disable the intro link output, by setting INTRO_LINK=''.

INTRO_IGNORE : a list of tags



Hyphenates words greater than 10 characters using Frank Liang’s algorithm. Hyphenation pattern depends on the current language of an article (defaulting to system’s locale). Only en, de and fr dictionaries are provided by Acrylamid. Example usage:

filters: [Markdown, hyphenate, ]
lang: en

If you need an additional language, download both, hyph-*.chr.txt and hyph-*.pat.txt, to `sys.prefix`/lib/python/site-packages/acrylamid/filters/hyph/.

Requires language patterns (ships with de, en and fr patterns)
Aliases hyphenate, hyph
Arguments Minimum length before this filter hyphenates the word (smallest possible value is four), defaults to hyphenate+10.


Enables typographical transformation to your written content. This includes no widows, typographical quotes and special css-classes for words written in CAPS and & (ampersand) to render an italic styled ampersand. See the original project for more information.

By default amp, widont, smartypants, caps are applied. all, typo and typogrify applyies widont, smartypants, caps, amp, initial_quotes. All filters are applied in the order as they are written down.

TYPOGRAPHY_MODE = "2"  # in your conf.oy

Smarty Pants has modes that let you customize the modification. See their options for reference. Acrylamid adds a custom mode "a" that behaves like "2" but does not educate dashes like --bare or bare--.

Requires smartypants
Aliases typography, typo, smartypants
Arguments all, typo, typogrify, amp, widont, smartypants, caps, initial_quotes, number_suffix. Defaults to typography+amp+widont+smartypants+caps.


This filter is a direct port of Pyblosxom’s acrynoms plugin, that marks acronyms and abbreviations in your text based on either a built-in acronyms list or a user-specified. To use a custom list just add the FILE to your like this:

ACRONYMS_FILE = '/path/to/my/acronyms.txt'

The built-in list of acronyms differs from Pyblosxom’s (see filters/ on GitHub). See the original description of how to make an acronyms file!

Requires <built-in>
Aliases Acronym(s), abbr (both case insensitive)
Arguments zero to N keys to use from acronyms file, no arguments by default (= all acronyms are used)


In addition to HTML+jinja2 templating you can also use Jinja2 in your postings, which may be useful when implementing a image gallery or other repeative tasks.

Within jinja you have a custom system-filter which allows you to call something like ls directly in your content (use it with care, when you rebuilt this content, the output might differ).

title: "Jinja2's system filter"
filters: jinja2

Take a look at my code:

.. code-block:: python

    {{ "cat ~/work/project/" | system | indent(4) }}

You can find my previous article "{{ env.prev.title }}" here_. Not
interesting enough? How about lorem ipsum?

{{ lipsum(5) }}

.. _here: {{ env.prev }}

Environment variables are the same as in Templating plus some imported modules from Python namely: time, datetime and urllib because you can’t import anything from Jinja2. You can also access the root templating environment when Jinja2. This means, you can import and inherit from templates located in your theme folder.

For convenience, the Jinja2 filter automatically imports every macro from macros.html into your post context, so there is no need for a {% from 'macros.html' import foo %}.

Requires <built-in>
Aliases Jinja2, jinja2


Just like Jinja2 filtering but using Mako. You have also system filter available within Mako. Unlike Jinja2 Mako can import python modules during runtime, therefore no additional modules are imported into the namespace.

Requires mako
Aliases Mako, mako


Some extension may generate relative references such as footnotes. While this is a good practise, it can get ambiguous when multiple posts with footnotes are included in an overview such as the index view does it. This ambiguity can be easily solved with the relative filter.

Requires <built-in>
Aliases relative


This also applies to feeds and many feed readers can’t/won’t resolve relative urls. This is where the absolute filter comes into play. This filter just expands a relative path to a valid URI. Important: if you ever change your domain, you have to force compilation otherwise this filter won’t notice this change

Requires <built-in>
Aliases absolute


Strip tags and attributes from HTML to produce a clean text version. Primary used by the static site search. By default, this filter includes everything between <tag>...</tag> but you can supply additional arguments to remove code listings wrapped in <pre> from the site search.

Requires <built-in>
Aliases strip
Arguments ignored tags (such as pre)


  • 90.0 : pre

    Jinja2, Mako

  • 70.0 : markup

    HTML, Markdown, pandoc, reST, textile

  • 50.0 : default

    metalogo, head_offset

  • 25.0 : post


  • 20.0 : post (conflict with typography)

    acronyms, hyphenate

  • 15.0 : shorten HTML

    intro, summarize

  • 10.0 : fix links

    relative, absolute

  • 0.0 : last


Custom Filters

To write your own filter, take a look at the code of already existing filters shipped with acrylamid and also visit Extending Acrylamid.