Pandoc ODT filters

Filters to improve Pandoc's conversion to ODT. Pandoc is a great tool to convert between document formats, but conversion to ODT (OpenDocument format) is poor in some aspects. These filters attempt to workaround those aspects.

At the beginning of each file (it's just a plain text file, you can open with any text editor), there's a little text explaining what the filter does. Please read carefully that little text, and decide by yourself if that filter works for your particular case. In case where customization is possible, there are local variables at the very beginning of code, after text and before all requires.

None of these filters are guaranteed to work, as I wrote them only to address Pandoc problems in ODT writer. Hopefully, in the near future none of them will be either necessary, as Pandoc's ODT writer become more robust --- in fact, some of them are becoming obsolete as Pandoc improves that.

One of these filters, context-table-span.lua, is targeted at ConTeXt writer, making fake row span. This also will be made obsolete as soon as Pandoc ConTeXt writer implements real row spans (already available at Pandoc AST).

Another one, abstract.lua, is targeted at all but ODT and DOCX writers. Formerly, this filter was targeted at ODT, but Pandoc now handles abstract sections better than my filter.

Requirements

• Pandoc, min. version: 2.14
• util.lua in the same directory of each filter (see below)

Installation

You can either go to Releases and download all files in a ZIP, or download just some of them via GitHub interface. In latter case, just don't forget to also download util.lua (see important note below).

Usage

You can use as many filters as you want. Each filter must have his own declaration in command line: --lua-filter <filter-name>.lua. Here's an example with just one filter:

pandoc -t odt+smart --lua-filter <filter-name>.lua <your-file>.md -o <destination-file>.odt

You can also use a Pandoc defaults file, declaring these filters as you would normally do. I use and recommend this approach.

Each filter addresses a particular problem, and few of them need other filters to be run to properly work. Look at the text in the beginning of file, on dependencies item, to see if the filter depends upon other(s) filter(s).

Near all of them use util.lua (where I put common filter tasks, to reuse code), so you need this file too.

IMPORTANT NOTE

All filters can go whatever directory you want, but an util.lua must be in the same directory of each filter that depends on it. For instance, if you put three filters on three different directories, you need to copy util.lua to each of these directories.

Available filters

Filter name	Description	Note
abstract.lua	[For all writers except docx and odt] Searches for abstract entry in metadata and creates a Div in the beginning of the document, containing markup from that entry. An abstract-title can also be set on metadata, so this filter will create a header (level 1, unnumbered) for the abstract.
odt-anchors.lua	Corrects anchors when converting to ODT, by adding bookmarks where anchors should come. This allows proper cross-referencing to these anchors with links along the text.	Currently, only figures and tables anchors are supported, where the anchor is inserted at the caption. Only figures and tables that have an id are processed (see pandoc-crossref for autoSectionLabels).
odt-bib-style	Corrects the style of bibliography when converting to ODT. This is necessary because Pandoc's ODT writer doesn't properly set this style.	odt-custom-styles.lua must also be used, and must be after this filter. Currently, all paragraphs in bibliography are turned into raw blocks with correct style. Because of this, only italics, bold and link markups are keep; all other markup in bibliography is lost. (see Pandoc's issue 3459).
odt-captions.lua	Corrects captions when converting to ODT, by adding sequence fields to the number of caption. This allows the creation of lists of figures, tables, etc.	Currently, only figure and table captions are supported. Only images and tables with caption are processed. Expected syntax of caption is the generated by pandoc-crossref. (see Pandoc's issue 2401)
odt-custom-styles.lua	Workaround to use custom styles when converting to ODT. This filter turns spans and headers with custom style into ODT raw inlines/blocks, with the ODT code using the custom style. If variable useClassAsCustomStyle is true, and element (span/header but also div) doesn't have a custom-style attribute, then first class is used as style.	Currently, the following elements are ignored by this filter: blockquotes, lists (see odt-lists.lua), tables and code blocks (for div styles), and citations, smallcaps (see odt-smallcaps.lua), images, quotes, strikeouts, super and subscript, math and code inlines (for span styles).
odt-lists.lua	Improves lists when converting to ODT, by adding list styles to lists, and apropriate paragraph styles to list items.	Only lists that are one level deep are supported; in lists with two or more levels, only the innermost level is improved, generating strange results. Currently, just italics, bold, links and line blocks are preserved in lists; all other markup is ignored.
odt-smallcaps.lua	Turns smallcaps into span with custom character style, when converting to ODT. This is necessary because LibreOffice default smallcaps is not true smallcaps, but rather reduced capitalised letters.	You need to use a reference-doc with the custom character style properly set. After this, change variable smallcapsStyle to the name of that custom character style. Example: use Myriad Pro:smcp as font in style configuration. smcp is the OpenType smallcaps feature.
util.lua	Utilities for use by other filters.	This file must be in the same directory of any filter that depends on it.

Contributing

Feel free to make pull requests. I've written these filters primarily for my needs, but I hope they can help other people.

TODO's

Below there are things I didn't resolve yet. Some of them are checked, either because current filters already make the work, or because I already make that work, but for some other problems didn't put that solution here.

• make TOC work:
  • filter to put all heading-links in metadata, to access in template
  • access heading-links in template
  • get page number of each link (impossible, I think...)
• make TOC position configurable, by [toc] markup:
  • filter to find [toc] occurrences and substitute by custom template variable
  • custom template that recognizes that variable (hard, I think...)
• make pre-textual styles work:
  • get unnumbered headings as textual headings (without numbering)
  • mark unnumbered headings, that come before first "normal" heading, with custom style
  • avoid errors with these headings

License

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version.

See LICENSE for details.

© 2021 José de Mattos Neto