FreeBSD Manual Pages
pandoc-crossref(1) General Commands Manual pandoc-crossref(1)
pandoc-crossref is a pandoc filter for numbering figures, equations,
tables and cross-references to them.
Bug tracker: <https://github.com/lierdakil/pandoc-crossref/issues>
• TOC {:toc}
Caveats
LaTeX input (a.k.a. converting LaTeX to docx/epub/etc)
The principal aim of pandoc-crossref is to add references to Markdown.
Other input formats may happen to work also (basically if you can make
Pandoc citation syntax to work), but that's more of a happy coincidence
than a deliberate design decision.
In particular, LaTeX syntax is generally not recognized, although under
some specific conditions it may seem to kinda-sorta work.
Converting TeX to anything editable in general is nigh impossible, with
TeX being a Turing-complete layout language rather than a markup lan-
guage. That was one of the motivations for pandoc-crossref in the
first place.
All that said, if you really need to use LaTeX as an input format, see
the discussions in <https://github.com/lierdakil/pandoc-crossref/is-
sues/250>.
For best results, consider using a purpose-built filter instead of pan-
doc-crossref. For example, pandoc-text-numbering
<https://github.com/fncokg/pandoc-tex-numbering>.
LaTeX output and --include-in-header
pandoc-crossref uses metadata variable header-includes to add LaTeX de-
finitions to output. However, Pandoc's command line option --in-
clude-in-header/-H overrides this variable. If you need to use --in-
clude-in-header, add pandoc-crossref-specific definitions as well. See
LaTeX customization for more information.
Note on non-standard LaTeX templates
If you're using non-standard pandoc LaTeX template (this includes the
cases where document class is not article or book close enough to
that), pandoc-crossref might not be able to configure that properly.
You might to configure LaTeX in the template manually.
Note on LaTeX and chapters option
Because pandoc-crossref offloads all numbering to LaTeX if it can,
chapters: true has no direct effect on LaTeX output. You have to spec-
ify Pandoc's --top-level-division=chapter option, which should hope-
fully configure LaTeX appropriately.
It's a good idea to specify --top-level-division=chapter for any output
format actually, because pandoc-crossref can't signal pandoc you want
to use chapters, and vice versa.
citeproc and pandoc-crossref
Since pandoc-crossref uses the same citation syntax as citeproc, you
have to run former before latter. For example:
pandoc -F pandoc-crossref --citeproc file.md -o file.html
Note on leading/trailing spaces in metadata options
Leading and trailing spaces in YAML metadata will most likely be
stripped by either YAML parser or Pandoc itself. If you need leading
and/or trailing spaces in pandoc-crossref metadata variables, use html
entity for space instead, i.e.  . For example, if you want refer-
ence ranges to be delimited by a dash with spaces (e.g. 2 - 5), include
the following in YAML metadata:
rangeDelim: ' - '
or pass -MrangeDelim=' - ' to pandoc on command line.
You can use other html entites of course, like etc.
Using raw blocks to define elements
It may be useful/convenient to define some elements as raw blocks, as
Pandoc Markdown, while very powerful, doesn't support everything Pandoc
can support internally (e.g. colspan/rowspan in tables).
Pandoc-crossref relies on Pandoc's AST representation, thus simply us-
ing raw blocks <https://pandoc.org/MANUAL.html#extension-raw_at-
tribute> won't work.
However, it's feasible if Pandoc can actually parse the contents of the
raw block with a little helper Lua filter. For example, something like
this:
function Div(div)
local id = div.attr.identifier
if (id:match "^tbl:" or id:match "^fig:") and div.content[2].t == "RawBlock"
then
local raw = div.content[2]
local elem = pandoc.read(raw.text, raw.format).blocks
elem[1].caption = pandoc.Caption(div.content[1])
elem[1].attr = div.attr
return elem
end
end
applied before pandoc-crossref, will reparse table definitions like
this:
:::{#tbl:foo}
Caption
```{=html}
<table>
<tr><th>Foo</th><th>Bar</th></tr>
<tr><td rowspan=2>1</td><td>Quux</td></tr>
<tr><td>Baz</td></tr>
</table>
```
:::
This also will work for figures, e.g.:
:::{#fig:foo}
Caption
```{=html}
<figure>
<p>This is just a paragraph instead of an image in a figure, nothing to see
here</p>
</figure>
```
:::
Bear in mind that if Pandoc fails to parse the raw block, pandoc-cross-
ref will fail to interpret it.
Syntax
Syntax is loosely based on discussion in <https://github.com/jgm/pan-
doc/issues/813>
Image labels
{#fig:label}
To label an (implicit) figure, append {#fig:label} (with label being
something unique to reference this figure by) immediately after image
definition.
This only works on implicit figures, i.e. an image occurring by itself
in a paragraph (which will be rendered as a figure with caption by pan-
doc)
Image block and label can not be separated by spaces.
Subfigures
It's possible to group figures as subfigures. Basic syntax is as fol-
lows:
<div id="fig:figureRef">
{#fig:figureRefA}
{#fig:figureRefB}
Caption of figure
</div>
To sum up, subfigures are made with a div having a figure id. Contents
of said div consist of several paragraphs. All but last paragraphs
contain one subfigure each, with captions, images and (optionally) ref-
erence attributes. Last paragraph contains figure caption.
If you put more than one figure in the paragraph, those will still be
rendered, but Pandoc will omit subfigure caption in most outputs (but
it will work as expected with LaTeX). You can use output-specific
hacks to work around that, or use subfigGrid or subfigColumn (see be-
low).
Output is customizable, with metadata fields. See Customization for
more information.
Default settings will produce the following equivalent Markdown from
example above:
<div id="fig:figureRef" class="subfigures">
{#fig:figureRefA}
{#fig:figureRefB}
Figure 1: Caption of figure. a -- subfigure 1 caption, b -- subfigure 2
caption
</div>
References to subfigures will be rendered as figureNumber (subfig-
ureNumber), e.g., in this particular example, [@fig:figureRefA] will
produce fig. 1 (a).
You can add nocaption class to an image to suppress subfigure caption
altogether. Note that it will still be counted.
Subfigure grid
If you need to align subfigures in a grid, and using output format
styles is not an option, you can use subfigGrid option. That will
typeset subfigures inside a table.
Rows are formed by different paragraphs, with each image in a separate
column.
Column widths will be taken from width attributes of corresponding im-
ages, e.g.
<div id="fig:coolFig">
{#fig:cfa width=30%}
{#fig:cfb width=60%}
{#fig:cfc width=10%}
{#fig:cfd}
{#fig:cfe}
{#fig:cff}
Cool figure!
</div>
will produce a table with columns of 30%, 60% and 10% respectively.
Only first row of images is considered for table width computation,
other rows are completely ignored.
Anything except images is silently ignored. So any text, spaces, soft
line breaks etc will silently disappear from output. That doesn't ap-
ply to caption paragraph, obviously.
All images will have width attribute automatically set to 100% in order
to fill whole column.
Specifying width in anything but % will throw an error.
If width for some images in first row is not specified, those will span
equally in the remaining space.
If width isn't specified for any image in first row, those will span
equally on 99% of page width (due to Pandoc otherwise omitting width
attribute for table).
This option is ignored with LaTeX output, but paragraph breaks should
produce similar effect, so images should be typeset correctly. TL;DR
you don't need subfigGrid enabled for it to work with LaTeX, but you
can still enable it.
Subfigure columns
Similar to subfigGrid, subfigColumns option will align each subfigure
row as Pandoc's columns environment. This primarily works well with
HTML output.
All the caveats applicable to subfigGrid are equally applicable to sub-
figColumns as well.
subfigGrid and subfigColumns are mutually exclusive. If both are spec-
ified, the former takes precedence.
Equation labels
$$ math $$ {#eq:label}
To label a display equation, append {#eq:label} (with label being some-
thing unique to reference this equation by) immediately after math
block.
Math block and label can be separated by one or more spaces.
You can also number all display equations with autoEqnLabels metadata
setting (see below). Note, however, that you won't be able to refer-
ence equations without explicit labels.
Equations numbers will be typeset inside math with \qquad before them.
If you want to use tables instead, use tableEqns option. Depending on
output format, tables might work better or worse than \qquad.
Alternatively, for formats that support it, you can use arbitrary LaTeX
command accepting a single argument (that is, label text) for typeset-
ting. A common example is \tag. Use equationNumberTeX metadata vari-
able for that (set to \qquad by default).
Beware that eqnIndexTemplate gets applied first, so you'll likely want
to set it to plain index as well.
For instance, to use \tag, you would have the following in your meta-
data:
equationNumberTeX: \\tag
eqnIndexTemplate: $$i$$
These options don't affect LaTeX output (which offloads numbering to
the LaTeX engine).
For advanced usage, see eqnInlineTemplate, eqnBlockTemplate.
Table labels
a b c
--- --- ---
1 2 3
4 5 6
: Caption {#tbl:label}
To label a table, append {#tbl:label} at the end of table caption (with
label being something unique to reference this table by). Caption and
label must be separated by at least one space.
Section labels
You can also reference sections of any level. Section labels use na-
tive pandoc syntax, but must start with "sec:", e.g.
Section {#sec:section}
You can also use autoSectionLabels variable to automatically prepend
all section labels (automatically generated with pandoc included) with
"sec:". Bear in mind that references can't contain periods, commas
etc, so some auto-generated labels will still be unusable.
WARNING: With LaTeX output, you have to invoke pandoc with --num-
ber-sections, otherwise section labels won't work. It's also advised
with other output formats, since with no numbers in section titles, it
would be hard to navigate anyway.
Section numbering
Pandoc doesn't properly support numbering sections in some output for-
mats, and section reference labels (see below).
You can let pandoc-crossref handle section numbering instead. This is
done via numberSections and sectionsDepth metadata options.
numberSections controls if pandoc-crossref handles numbering sections,
while sectionsDepth controls what sections are numbered.
Additionally, with numberSections, if the first heading in your docu-
ment is level 2 or more, pandoc-crossref will assume you meant to have
implicit headings with previous levels, and will assign those phantom
implicit headings the index 1. Without numberSections, the behaviour
is consistent with pandoc, that is, missing headings will be assigned
the index 0.
Set sectionsDepth to 0 to make section numbering consistent with chap-
tersDepth.
If sectionsDepth value is lesser than 0, all sections will be numbered.
Otherwise, only header levels up to and including sectionsDepth will be
numbered.
You can also supply a custom section header template via secHeaderTem-
plate metadata option. The following variables are supported:
• $$i$$ - formatted section number, according to sectionsDepth
• $$t$$ - original section header text
• $$n$$ - 0-indexed section level (0 is the topmost)
See section on templates for more information
Reference labels
Not currently supported with LaTeX output
If you want to reference some object by a pre-defined label instead of
by number, you can specify attribute label, like this:
# Section {label="Custom Label"}
{fig:fig1 label="Custom label"}
Note that to use this with equations and tables, you need to use fenced
div/span syntax, not the short syntax:
:::{#tbl:table label="T"}
a b c
--- --- ---
1 2 3
4 5 6
: Caption
:::
[$$y = e^x$$]{#eq:equation label="E"}
This label will be used instead of a number in chapters output for sec-
tions and when referencing the element directly.
Note that with chapters output with depth>1, only the given section
will be referenced by the custom label, e.g. with
Chapter 1.
# Section with custom label {#sec:scl label="SCL"}
{#fig:figure}
@sec:scl will translate into sec. 1.SCL, and @fig:figure into fig.
1.SCL.1
Manual numbering adjustment
Not currently supported with LaTeX output
For cases when you need to manually adjust numbering, you can specify
the number attribute on the object. It will set the internal object
counter for the annotated object to the number specified, and all the
following objects of this type will count from that. Conceptually,
this is similar to document processors' "start from..." etc.
Same as with label attributes, to use this with equations and tables,
you need to use fenced div/span syntax, not the short syntax.
Code Block labels
There are a couple options to add code block labels. Those work only
if code block id starts with lst:, e.g. {#lst:label}
caption attribute
caption attribute will be treated as code block caption. If code block
has both id and caption attributes, it will be treated as numbered code
block.
Table-style captions
Enabled with codeBlockCaptions metadata option. If code block is imme-
diately adjacent to paragraph, starting with Listing: or :, said para-
graph will be treated as code block caption.
or
It also allows to specify label in caption, as do tables, for example:
Wrapping div
Wrapping code block without label in a div with id lst:... and class,
starting with listing, and adding paragraph before code block, but in-
side div, will treat said paragraph as code block caption.
Any additional attributes and classes on the wrapping div will be
merged with classes/attributes on the listing itself. In case of du-
plicate attributes, the behaviour is unspecified, but likely either
both values will end up in the output, or div attributes will take
precedence. This is important to keep in mind if you need to do some
additional post-processing.
References
[@fig:label1;@fig:label2;...] or [@eq:label1;@eq:label2;...] or [@tbl:label1;@tbl:label2;...] or @fig:label or @eq:label or @tbl:label
Reference syntax heavily relies on citation syntax. Basic reference is
created by writing @, then basically desired label with prefix. It is
also possible to reference a group of objects, by putting them into
brackets with ; as separator. Similar objects will be grouped in order
of them appearing in citation brackets, and sequential reference num-
bers will be shortened, e.g. 1,2,3 will be shortened to 1-3.
You can capitalize first reference character to get capitalized prefix,
e.g. [@Fig:label1] will produce Fig. ... by default. Capitalized pre-
fixes are derived automatically by capitalizing first letter of every
word in non-capitalized prefix, unless overridden with metadata set-
tings. See Customization for more information.
Linking references
To make references into hyperlinks to referenced element, enable
linkReferences metadata option. This has no effect on LaTeX output,
since in this case, hyperlinking references is handled with hyperref
LaTeX package.
Custom prefix per-reference
It's possible to provide your own prefix per-reference, f.ex. [Prefix
@reference] will replace default prefix (fig./sec./etc) with prefix
verbatim, e.g. [Prefix @fig:1] will be rendered as Prefix 1 instead of
fig. 1.
In citation group, citations with the same prefix will be grouped. So,
for example [A @fig:1; A @fig:2; B @fig:3] will turn into A 1, 2, B 3.
It can be used to an advantage, although it's a bit more cumbersome
than it should be, e.g. [Appendices @sec:A1; Appendices @sec:A2; Ap-
pendices @sec:A3] will turn into Appendices @A1-@A3 (with @A1 and @A3
being relevant section numbers). Note that non-contiguous sequences of
identical prefixes will not be grouped.
Not supported with cleveref LaTeX output.
Prefix suppression
Prepending - before @, like so [-@citation], will suppress default pre-
fix, e.g. [-@fig:1] will produce just 1 (or whatever number it happens
to be) without fig. prefix.
In citation group, citations with and without prefixes will be in dif-
ferent groups. So [-@fig:1; @fig:2; -@fig:3] will be rendered as 1,
fig. 2, 3, so be careful with this feature. Again, non-contiguous se-
quences are not grouped together.
Lists
It's possible to use raw latex commands \listoffigures, \listoftables
and \listoflistings, which will produce ordered list of figure/ta-
ble/listings titles, in order of appearance in document.
For LaTeX output, \listoflistings depends on other options, and is de-
fined in preamble, so it will work reliably only with standalone/pdf
output.
NOTE: With Pandoc 2.0.6 and up, you might have to explicitly separate
these commands if they are close together, at least when targeting
something besides LaTeX. So this might not work:
\listoffigures
\listoftables
\listoflistings
but this will:
\listoffigures
<!-- hack to split raw blocks -->
\listoftables
<!-- hack to split raw blocks -->
\listoflistings
With HTML-compatible output, lists are wrapped into a div with classes
list and list-of-<prefix>, where <prefix> is either fig, tbl or lst de-
pending on the type of the list. This allows for ad-hoc style over-
rides in HTML.
Hiding items from list-of-X lists
NB: Highly experimental with LaTeX output (that includes PDF).
Might not work as expected or break in some edge cases. Please
check your output and raise issues!
If you need to hide a figure/table/code block from the corresponding
\listofX, you can set hidden attribute on the item, e.g.
{#fig:hidden hidden="y"}
The following values are interpreted as "no", so setting hidden to
those will leave the object visible in the list-of-X: no, n, false, f,
off, 0, none, null, disable, disabled, unset. Any other value (includ-
ing empty string!) will hide the item from list-of-X.
Additionally, this attribute can be set on sections. Since there isn't
a list-of-sections per se, table of contents notwithstanding, instead
of hiding a section, this will implicitly hide all items inside that
section and all its subsections.
You can override subsections or individual items inside a hidden sec-
tion to be visible by explicitly setting hidden=no (or to any other
falsy value).
Same as with other attributes, to use this with tables, you need to use
fenced div syntax, not the short syntax.
List-of-X in Pandoc templates
If you'd prefer to include list-of-X things in your custom template,
rather than use \listofX directly in the document, set the listOfMeta-
data option to true, and add something like this to your template:
$if(list-of-figures)$
$list-of-figures$
$endif$
The following variables are supported:
• list-of-figures
• list-of-tables
• list-of-listings
Usage
Run pandoc with --filter option, passing path to pandoc-crossref exe-
cutable, or simply pandoc-crossref, if it's in PATH:
pandoc --filter pandoc-crossref
If you installed with cabal, it's most likely located in $HOME/.ca-
bal/bin on *NIX systems, $HOME/Library/Haskell/bin on Macs, or in %App-
Data%\cabal\bin on Windows.
Customization
There are several parameters that can be set via YAML metadata (either
by passing -M to pandoc, or by setting it in source markdown)
A list of variables follows.
General options
• cref: if True, latex export will use \cref from cleveref package.
Only relevant for LaTeX output. \usepackage{cleveref} will be auto-
matically added to header-includes.
• chapters: if True, number elements as chapter.item, and restart item
on each first-level heading. You might also need to run pandoc with
--top-level-division=chapter argument to signal it you want to use
chapters; whether it's actually required or not depends on the output
format, but it's always safe to include. Notice chapters and related
options are ignored in LaTeX output. See Note on LaTeX and chapters
option
• chaptersDepth, default 1: header level to treat as "chapter". If
chaptersDepth>1, then items will be prefixed with several numbers,
corresponding to header numbers, e.g. fig. 1.4.3.
• numberSections, default false: if True, pandoc-crossref will prepend
section number to section titles (as counted by pandoc-crossref it-
self). This also makes pandoc-crossref assign missing top-level
headings the index of 1 instead of 0 to avoid ugly sec. 0.0.1 refer-
ences.
• sectionsDepth, default 0:
• sectionsDepth < 0 - number all sections
• sectionsDepth == 0 - be consistent with chaptersDepths
• sectionsDepth > 0 - number section levels up to and including sec-
tionsDepth
• listings: if True, generate code blocks for listings package. Only
relevant for LaTeX output. \usepackage{listings} will be automati-
cally added to header-includes. You need to specify --syntax-high-
lighting=idiomatic option as well.
• codeBlockCaptions: if True, parse table-style code block captions.
• autoSectionLabels, default false: Automatically prefix all section
labels with sec:. Note that this messes with pandoc's automatic
header references.
• autoEqnLabels, default false: Automatically number all display equa-
tions (i.e. ones defined using $$...$$/\[...\]). Note that you won't
be able to reference equations without explicit labels.
• tableEqns, default false: Typeset equations and equation numbers as
blocks instead of embedding numbers into equations themselves. De-
pending on output format, this might work better or worse. See also
the section on equation templates.
• setLabelAttribute, default false: set label attribute on objects to
actual number used for referencing. This can be useful for post-pro-
cessing.
• equationNumberTeX, default \\qquad: use a LaTeX command for typeset-
ting equation numbers. Remember that metadata is parsed as Markdown,
so you may need to escape backslashes. This option doesn't affect
LaTeX output (which offloads numbering to the LaTeX engine).
• listOfMetadata, default false. Enables setting list-of-figures &c
variables to pass through to the Pandoc's templating engine.
Item title format
• figureTitle, default Figure: Word(s) to prepend to figure titles,
e.g. Figure 1: Description
• tableTitle, default Table: Word(s) to prepend to table titles,
e.g. Table 1: Description
• listingTitle, default Listing: Word(s) to prepend to listing titles,
e.g. Listing 1: Description
• titleDelim, default :: What to put between object number and caption
text.
• secHeaderDelim, default (i.e. space): What to put between section
number and title when numberSections is true. Can be a list, in that
case it's indexed by heading level, 0-based.
Subfigure-specific
See Subfigures
• ccsDelim, default , : delimiter for collected subfigure cap-
tions. See Subfigures and Templates
• ccsLabelSep, default -- : delimiter used between subfigure
label and subfigure caption in collected captions. See Subfigures
and Templates
• subfigGrid, default false. If true, typeset subfigures inside a ta-
ble. Ignored with LaTeX output. See Subfigures
• subfigColumns, default false. If true, typeset subfigure rows as
column environment. Ignored with LaTeX output. See Subfigures
List titles
• lofTitle, default # List of Figures: Title for list of figures (lof)
• lotTitle, default # List of Tables: Title for list of tables (lot)
• lolTitle, default # List of Listings: Title for list of listings
(lol)
List items
See also List item templates
• lofItemTitle, default empty: Title each item in the list of figures,
i.e. each item will be prefixed by this string.
• lotItemTitle, default empty: Title each item in the list of tables,
i.e. each item will be prefixed by this string.
• lolItemTitle, default empty: Title each item in the list of listings,
i.e. each item will be prefixed by this string.
• listItemTitleDelim, default . (str "."): delimiter between list item
numbers and captions.
Reference format
• figPrefix, default fig., figs.: Prefix for references to figures,
e.g. figs. 1-3
• eqnPrefix, default eq., eqns.: Prefix for references to equations,
e.g. eqns. 3,4
• tblPrefix, default tbl., tbls.: Prefix for references to tables,
e.g. tbl. 2
• lstPrefix, default lst., lsts.: Prefix for references to lists,
e.g. lsts. 2,5
• secPrefix, default sec., secs.: Prefix for references to sections,
e.g. secs. 2,5
• chapDelim, default .: Delimiter between chapter number and item num-
ber.
• rangeDelim, default -: Delimiter between reference ranges, e.g. eq.
2-5
• pairDelim, default ,: Delimiter between pair of reference ranges,
e.g. eq. 2-5 <and> 7-9, or eq. 2 <and> 7, but eq. 2, 4, 6
• lastDelim, default ,: Delimiter between second-to-last and last ref-
erence ranges, e.g. eq. 2-5, 6-8 <and> 10. pairDelim overrides this
for cases of exactly two references/ranges.
• refDelim, default ,: Delimiter between references, e.g. eq. 2, 5, 7
or eq. 2-4, 6-8
• linkReferences, default false: Make references hyperlinks to the ref-
erenced element
• nameInLink, default false: For single-element references, inlcude
prefix into hyperlink (when using linkReferences)
Note that none of the *Delim options are honored with cleveref output.
Use cleveref's customization options instead.
figPrefix, eqnPrefix, tblPrefix, lstPrefix can be YAML arrays. That
way, value at index corresponds to total number of references in group,
f.ex.
figPrefix:
- "fig."
- "figs."
Will result in all single-value references prefixed with "fig.", and
all reference groups of two and more will be prefixed with "figs.":
[@fig:one] -> fig. 1
[@fig:one; @fig:two] -> figs. 1, 2
[@fig:one; @fig:two; @fig:three] -> figs. 1-3
They can be YAML strings as well. In that case, prefix would be the
same regardless of number of references.
They can also be used with first character capitalized, i.e. FigPre-
fix, etc. In this case, these settings will override default reference
capitailzation settings.
Custom numbering
See Custom Numbering Schemes
• figLabels, default arabic: the numbering scheme for figures.
• subfigLabels, default alpha a: the numbering scheme for subfigures.
• eqLabels, default arabic: the numbering scheme for equations.
• tblLabels, default arabic: the numbering scheme for tables.
• lstLabels, default arabic: the numbering scheme for listings.
• secLabels, default arabic: the numbering scheme for sections.
• secLevelLabels, default unset: the numbering scheme for sections,
YAML array, indexed by heading level; will override secLabels if set.
Item title templates
See Templates
• figureTemplate, default $$figureTitle$$ $$i$$$$titleDelim$$ $$t$$:
template for figure captions
• tableTemplate, default $$tableTitle$$ $$i$$$$titleDelim$$ $$t$$: tem-
plate for table captions
• listingTemplate, default $$listingTitle$$ $$i$$$$titleDelim$$ $$t$$:
template for listing captions
• secHeaderTemplate, default $$i$$$$secHeaderDelim[n]$$$$t$$: template
for section header text when numberSections is true
Subfigure templates
See Subfigures
• subfigureTemplate, default $$figureTitle$$ $$i$$$$titleDelim$$ $$t$$.
$$ccs$$: template for subfigure divs captions.
• subfigureChildTemplate, default $$i$$: template for actual subfigure
captions.
• ccsTemplate, default $$i$$$$ccsLabelSep$$$$t$$: template for col-
lected subfigure captions.
Equation templates
• eqnIndexTemplate, default ($$i$$): used to adjust how equation index
is rendered. This is required due to it being rendered inside a math
block. Note that for the same reason formatting is mostly ignored.
• eqnInlineTemplate, default $$e$$$$equationNumberTeX$${$$i$$}
A template to typeset math when tableEqns is false. Similar to
eqnIndexTemplate, formatting is mostly ignored, due to it being type-
set inside a math environment. However, most LaTeX should work (but
backslashes need to be doubled). The following template variables
are known:
• e, the equation itself,
• t, the same as e, for backwards compatibility,
• i, index after applying eqnIndexTemplate,
• nmi, same as i, but see eqnDisplayTemplate,
• ri, "raw" index, before applying eqnIndexTemplate,
• nmri, here, same as ri, but see eqnDisplayTemplate.
eqnInlineTemplate is ignored if tableEqns is true.
• eqnInlineTableTemplate, default $$e$$
A counterpart of eqnInlineTemplate for when tableEqns is true. Be-
haves the same. The logic is split like this mostly for backwards
compatibility, but it also allows specifying unambiguous defaults.
eqnInlineTableTemplate is ignored if tableEqns is false.
• eqnDisplayTemplate, default $$e$$
A template to typeset the math element produced by applying eqnIn-
lineTemplate or eqnInlineTableTemplate.
The output of this template must be a sequence of inline elements.
If you want to produce block elements, see eqnBlockTemplate.
The same template variables from eqnInlineTemplate are available,
with the following changes:
• e (and t) is now the formatted equation (as per eqnInlineTemplate
or eqnInlineTableTemplate) wrapped in a math environment (display
or inline depending on eqnBlockInlineMath),
• i and ri are now wrapped in a math environment (same type as e).
eqnDisplayTemplate is ignored if tableEqns is true.
• eqnBlockTemplate, default
+:--------------------------------------------------------------:+-----:+
| $$t$$ | ```{=openxml}
| | <w:tcPr><w:vAlign w:val="center"/></w:tcPr>
| | ```
| | $$i$$
+----------------------------------------------------------------+-----+
When used with tableEqns, a block to use to format equations. A ta-
ble by default, but could be literally any block or a sequence of
blocks. The behaviour is similar to eqnDisplayTemplate, but the ele-
ments produced are block elements (as opposed to inline).
Note that the default contains a raw block to fix vertical alignment
in docx output. If you're not targeting docx, it will be ignored by
pandoc.
eqnBlockTemplate is ignored if tableEqns is false (the default).
• eqnBlockInlineMath, default False: if you need to use inline math
while rendering equation templates. Useful, e.g., if you're using
raw ooxml and tabstops to align equations in docx. For example,
tableEqns: true
eqnBlockTemplate: |
`<w:pPr><w:tabs><w:tab w:val="center" w:leader="none" w:pos="4680" /><w:tab w:val="right" w:leader="none" w:pos="9360" /></w:tabs></w:pPr><w:r><w:tab /></w:r>`{=openxml} $$t$$ `<w:r><w:tab /></w:r>`{=openxml} $$i$$
eqnBlockInlineMath: true
(tweak w:pos for the two tabstops to taste, the unit is 1/20th of an
inch, sample values are appropriate for portrait A4 with 1-inch mar-
gins)
Reference templates
See Templates
• figPrefixTemplate, default $$p$$ $$i$$ - figure reference tem-
plate
• eqnPrefixTemplate, default $$p$$ $$i$$ - equation reference tem-
plate
• tblPrefixTemplate, default $$p$$ $$i$$ - table reference tem-
plate
• lstPrefixTemplate, default $$p$$ $$i$$ - listing reference tem-
plate
• secPrefixTemplate, default $$p$$ $$i$$ - section reference tem-
plate
• refIndexTemplate, default $$i$$$$suf$$ - individual reference index
template
• subfigureRefIndexTemplate, default $$i$$$$suf$$ ($$s$$) - subfigure
reference index template
List item templates
See Templates
• lofItemTemplate, default $$lofItemTitle$$$$i$$$$listItemTitleDelim$$
$$t$$\\\n - list-of-figures item template
• lotItemTemplate, default $$lotItemTitle$$$$i$$$$listItemTitleDelim$$
$$t$$\\\n)
• lolItemTemplate, default $$lolItemTitle$$$$i$$$$listItemTitleDelim$$
$$t$$\\\n)
Special handling is enabled for templates that are either ordered or
bullet lists: items will be merged into a single list. Thus, for exam-
ple, given
lofItemTemplate: |
1. $$t$$
the list of figures will be formatted as
# List of Figures
::: {.list .list-of-fig}
1. Figure 1 caption
2. Figure 2 caption
3. Figure 3 caption
4. ...
:::
Variables available in these templates:
• i - item number with chapter (if chapters are enabled), formatted ac-
cording to refIndexTemplate or subfigureRefIndexTemplate if item is a
subfigure
• ri - "raw" item number, before applying formatting (but with chapter
if chapters are enabled)
• lt - title, linked back to the item (i.e. wrapped in an anchor ele-
ment)
• t - just the title itself, without any links
• s - subfigure identifier, if any; empty if not a subfigure
LaTeX customization
Support for above variables with LaTeX/PDF output is limited. In par-
ticular, the following variables are honored:
• figureTitle
• tableTitle
• listingTitle
• lofTitle - ignores formatting
• lotTitle - ignores formatting
• lolTitle - ignores formatting
• *Prefix, upper-/lowercase and single/plural form. Note that with
cleveref output, if *Prefix is array, only first two items are used,
and the rest is ignored.
Templates are not supported.
You can add arbitrary LaTeX commands to document header, however, using
header-includes metadata field. Please bear in mind, that pan-
doc-crossref up to and including 0.1.2.1 requires header-includes to be
YAML array, e.g.
header-includes:
- "\\newcommand{\\pcdoc}{Pandoc-crossref documentation}"
This will be added before any customization applied by pandoc-crossref.
For a complete list of what is added to template, consult ModifyMeta.hs
<https://github.com/lierdakil/pandoc-crossref/blob/master/lib/Text/Pan-
doc/CrossRef/Util/ModifyMeta.hs>.
Templates
pandoc-crossref supports advanced caption customization via caption
templates. Templates are specified as YAML metadata variables (see
Customization), and are parsed as default Pandoc Markdown. Variables
are specified with display math syntax, i.e. $$var$$ in a template
will be replaced with value of variable var. Variables can be speci-
fied in YAML metadata block, or from command line (with -M switch).
There are two special variables, that are set internally:
• i - object number, possibly with chapter number (if chapter=True)
• t - object caption, as given in source Markdown
Also there is a number of specific variables that are meaningful only
in certain contexts:
• ccs - collected subfigure captions. Only applicable to subfigureTem-
plate. Collected captions will be separated by ccsDelim and individ-
ual captions will be printed with ccsTemplate. See Subfigures
• suf - reference suffix, applicable to refIndexTemplate, subfigureRe-
fIndexTemplate
• s - subfigure index, applicable to subfigureRefIndexTemplate
xPrefixTemplate, where x is fig, eqn, etc, are a special case. Those
don't have t variable, since there is no caption in source markdown,
but instead have p variable, that binds to relevant xPrefix. This is
done this way, since actual prefix value can depend on i. In xPrefix-
Template, i references formatted object numbers, i.e. if given a list
of references like [@fig:1; @fig:2; @fig:3], here i will contain some-
thing like 1-3.
refIndexTemplate is the template for the individual reference index.
It can be either a plain template, or can be a YAML object with keys
corresponding to different prefixes, and a special key default used as
a fallback, e.g.
refIndexTemplate:
sec: $$i$$$$suf$$ ($$t$$)
default: $$i$$$$suf$$
refIndexTemplate has the following internal variables defined:
• i - formatted object index (possibly with chapter number)
• suf - literal suffix used in the reference, e.g. given [@fig:1 some
suffix], suf will contain literally some suffix (complete with the
leading space)
• t - object title, if any, or empty if the object has no title
subfigureRefIndexTemplate is roughly the same as refIndexTemplate but
is used specifically for subfigures. It additionally has s variable
defined, which is described above.
Additionally, a special syntax is provided for indexed access to array
metadata variables: arrayVariable[indexVariable], where arrayVariable
is an array-like metadata variable, and indexVariable is an inte-
ger-typed template variable. If indexVariable is larger than length of
arrayVariable, then the last element in arrayVariable is used.
Indexed access can be useful with secHeaderTemplate for example, where
you might want to add a custom prefix depending on the header level.
For example, with this YAML metadata:
secHeaderTemplate: $$secHeaderPrefix[n]$$$$i$$. $$t$$
secHeaderPrefix:
- "Chapter "
- "Section "
- ""
sectionsDepth: -1
numberSections: true
top-level sections will be prefixed with Chapter, second-level sections
will be prefixed with Section and the rest won't be prefixed with any-
thing.
Please note that at the moment, templates are not supported with La-
TeX/PDF output.
Custom Numbering Schemes
It's possible to use other numbering schemes apart from arabic. This
is controlled by several metadata options, consult Customization for a
list. Possible values are:
• arabic - arabic numbers (1, 2, 3 ...)
• roman - roman numbers (I, II, III, IV, ...)
• lowercase roman - lowercase roman numbers (i, ii, iii, iv, ...)
• alpha x, where x is first letter to start from. This will work for
any letter, but will use UTF-8 codepage to determine what's next, so
using something strange is not advised. For example, you can safely
use alpha a to get lowercase latin letters for 26 figures. After
that, it will get weird (since basic latin alphabet has 26 letters).
Specifically, it will go into characters space ({, |, }, etc). You
can consult http://www.fileformat.info/info/unicode/block/ba-
sic_latin/utf8test.htm for general idea on letter progression.
• list of strings. You can define a YAML array for numbers. Mapping
is 1:1. For example, figLabels: [, , , , ] will give first object
label `', second - `', etc, up until the fifths.
Note that it will repeat last item in list indefinitely if there are
more references than items in list, i.e. in the example above, sixths
object and all after that will also have label `'.
Settings file
It is also possible to set variables used by pandoc-crossref with a
separate YAML file. If a given variable is not set in metadata, then
pandoc-crossref will attempt to read it from file specified by cross-
refYaml metadata variable, or, if not set, from pandoc-crossref.yaml
from current working directory. This allows for reusable configura-
tions. One possible application is ad-hoc internationalization.
For example, consider $HOME/misc/pandoc-crossref-es.yaml:
figureTitle: "Figura"
tableTitle: "Tabla"
figPrefix: "fig."
eqnPrefix: "ec."
tblPrefix: "tbl."
loftitle: "# Lista de figuras"
lotTitle: "# Lista de tablas"
pandoc-crossref will send this data to pandoc wrapped in lines of ---.
The YAML file's first line should specify a variable; it will not pass
the variables if it is --- or a blank line.
One could use this with pandoc-crossref as follows:
pandoc -F pandoc-crossref.hs -M "crossrefYaml=$HOME/misc/pandoc-cross-
ref-es.yaml"
You can also use global configuration files, which are expected in
$HOME/.pandoc-crossref/config.yaml and $HOME/.pandoc-crossref/con-
fig-$FORMAT.yaml, where $FORMAT is output format, f.ex. latex or epub.
On Windows, $HOME in general resolves to user's root directory,
e.g. C:\Users\username\.
Priorities are as follows (from highest to lowest):
• document metadata
• crossrefYaml/$CWD/pandoc-crossref.yaml
• $HOME/.pandoc-crossref/config-$FORMAT.yaml
• $HOME/.pandoc-crossref/config.yaml
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 2 of the License, or (at your
option) any later version.
This program includes code from roman-numerals library, covered by the
following terms:
Copyright 2009-2014 Roel van Dijk
All rights reserved.
Redistribution and use in source and binary forms, with or with-
out modification, are permitted provided that the following con-
ditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above
copyright notice, this list of conditions and the following
disclaimer in the documentation and/or other materials provided
with the distribution.
* The names of contributors may not be used to endorse or promote
products derived from this software without specific prior
written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBU-
TORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING,
BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSE-
QUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIA-
BILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUD-
ING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
DAMAGE.
AUTHORS
Nikolay Yakimov.
December 2024 pandoc-crossref(1)
Caveats | Syntax | Usage | License | AUTHORS
Want to link to this manual page? Use this URL:
<https://man.freebsd.org/cgi/man.cgi?query=pandoc-crossref&sektion=1&manpath=FreeBSD+Ports+15.0.quarterly>