A collection of JavaScript-based extensions for the pandoc universal document converter.
CoffeeScript CSS JavaScript
Switch branches/tags
Nothing to show
Latest commit a02f7ac Nov 1, 2013 Rod Waldhoff fix executable scripts to match revised packaging
Failed to load latest commit information.
test remove old test case Nov 1, 2013
README.md fix typo Nov 1, 2013



A collection of PANDOc eXtensions

Pandox is a small collection of utilities that extend John MacFarlane's Pandoc, the "universal document converter".


The Command Line Interface

The Pandox extensions are typically invoked by piping a JSON-formatted AST (generated via pandoc -t json FILENAME) in as stdin, and piping the transformed JSON-formatted AST to another pandoc invocation for rendering. For example:

> pandoc -t json README.md | coffee lib/up-caser.coffee | pandoc -f json -t html

will generate an HTML version of the README.md file, first applying the filter defined in up-caser.

The npm module includes directly executable scripts for each extension. Hence:

> pandoc -t json README.md | pandox-up-caser | pandoc -f json -t html

also works (following npm install -g pandox).

Use the --help or -h command line parameter for more information:

> node lib/pandoc-filter.js --help

Usage: node lib/pandoc-filter.js [OPTIONS] [FILE]

  -h, --help    Show this help message.
  -f, --filter  A filter to apply before this one. May be repeated.

  pandoc -t json FILE.md | node lib/pandoc-filter.js
  pandoc -t json FILE.md > FILE.json && node lib/pandoc-filter.js FILE.json

The API (Code-Level) Interface

Pandox processes the JSON-format abstact-syntax tree that pandoc can generate when given the -t json flag. The internal API is (intended to be) the same the Pandoc filters API that exists for Haskell and Python.

If you'd like to try your hand at writing custom JavaScript-based Pandoc filters, simply extend the PandocFilter class or supply a filtering method. For example:

var PandocFilter = require('pandoc-filter');

function upcase(type,content) {
  if(type==='Str') {
    return { t:type, c:content.toUpperCase();
  } else {
    return null;

var filter = new PandocFilter(upcase);

Several examples can be found in the ./lib directory.

See the comments in ./lib/pandoc-filter.coffee (or ./lib/pandoc-filter.js) for a detailed explanation of the API.

The Extensions


The UpCaser filter is a trivial extension primarily intended to serve as an example.

It will convert any string values to upper case, but ignores code blocks, urls, and other special cases.

Usage example:

> pandoc -t json README.md | pandox-up-caser | pandoc -f json -t html


By default, Pandoc's AST treats each (whitespace-delimited) word as an indepdent block. For instance, the string:

Hello World.

is represented in the AST as three separate blocks:

[ { t:'Str',c:'Hello' }, { t:'Space',c:[] }, { t:'Str',c:'World.' } ]

The StringCombiner filter will collapse these sequences of strings and spaces into a single (multi-word) string:

[ { t:'Str',c:'Hello World.' } ]

Usage example:

> pandoc -t json MY-FILE.md | pandox-string-combiner | pandoc -f json -t html


The CodeBlockProcessor filter adds several capabilities to the way in which Pandoc handles "fenced code blocks", such as:

This is a sample of text inside a "fenced" code block.

Pandoc supports several parameters that control the way in which a code block is rendered. The general form is:

This is a sample of text inside a "fenced" code block.


  • #THE-ID is used to identify the code block in things like HTML anchors and Latex cross-references.

  • .CLASS-ONEand .CLASS-TWO enumerate HTML classes to assign to the code block, and sometimes influence the rendering in other ways. For example, adding the class .numberLines will cause Pandoc to number the lines in the code block when rendering it.

  • NAME="VALUE" and NAME2="VALUE2" enumerate name-value pairs that can be used to modify the way in which the code block is rendered. For example, adding the pair startFrom=100 will cause Pandoc to number the lines starting with 100 rather than 1.

CodeBlockProcessor adds a few new parameters that can be controlled by name-value pairs.

  • input-file - replaces the body of the code block with the contents of the specified file.

  • input-cmd - replaces the body of the code block with output of the specified command.

  • exec - executes the body of the code block as it were a shell script

  • output-file - writes the body of the code block to the specified file.

  • output-cmd - pipes the body of the code block to the specified command.

Usage example:

> pandoc -t json MY-FILE.md | pandox-code-block-processor | pandoc -f json -t html


Pandox is made availble under an MIT-license. See license.txt for details.