Building filters

Basically, a pandoc filter will

1. read JSON data from the standard input with Yojson.Basic.from_channel{.ocaml}
2. convert it to the library's standard representation with Pandoc.of_json{.ocaml}
3. transform it using the Pandoc.map{.ocaml} function which recursively maps functions on various elements,
4. convert the result to JSON with Pandoc.of_json{.ocaml},
5. print the result on the standard output using Yojson.Basic.to_string{.ocaml}.

Once you have build your filter (say, myfilter) you can use it to transform pandoc documents while they are being processed by using the --filter commandline argument. For instance:

pandoc --filter=myfilter file.md -o file.pdf

Examples

Some examples are provided in the examples directory for you convenience (or because I needed those).

• pandoc-abbreviations: adds non-breakable spaces after abbreviations (listed in abbreviations file). Pandoc already does this natively, but it does not play well with other plugins.
• pandoc-crossref: handles cross-references in LaTeX. It replaces references of the form #chap:bla and #sec:bli to \cref{chap:bla} and ``cref{sec:bli}` respectively.

• pandoc-include: includes other documents. It replaces

  !include "file"

  by the contents of the file file and

    ```{.ocaml include="test.ml" from=2 to=5}
  ```

  by the contents of the file test.ml between lines 2 and 5 (negative values for to are counted from the end of the file). Useful for including small code snippets.

• pandoc-inspects: acts as the identity plugin, but prints the JSON output given by pandoc on the standard error. Useful for debugging and adding missing features to the library.

Bugs and features requests

The code is simple and stable enough for me but some features are missing. Feel free to fill bug reports or submit pull requests.