Skip to content

Introduction

Starting from version 6.0.0, roxygen supports Markdown markup within most roxygen tags. Roxygen uses the commonmark package, which is based on the CommonMark Reference Implementation to parse these tags. See https://commonmark.org/help/ for more about the parser and the markdown language it supports. You can also still use the .Rd syntax, some of which we will present below in the Rd syntax section.

Turning on Markdown support

There are two ways to turn on Markdown support for a package: globally, at the package level, and locally at the block level.

To turn on Markdown for the whole package, insert this entry into the DESCRIPTION file of the package:

Roxygen: list(markdown = TRUE)

The position of the entry in the file does not matter. After this, all Roxygen documentation will be parsed as Markdown.

Alternatively, you can use the @md tag to turn on Markdown support for a single documentation chunk. This is a good option to write any new documentation1 for existing packages in markdown.

There is also a new @noMd tag. Use this if you turned on Markdown parsing globally, but need to avoid it for a single chunk. This tag is handy if the Markdown parser interferes with more complex Rd syntax.

Here is an example roxygen chunk that uses Markdown.

#' Use roxygen to document a package
#'
#' This function is a wrapper for the [roxygen2::roxygenize()] function from
#' the roxygen2 package. See the documentation and vignettes of
#' that package to learn how to use roxygen.
#'
#' @param pkg package description, can be path or package name.  See
#'   [as.package()] for more information
#' @param clean,reload Deprecated.
#' @inheritParams roxygen2::roxygenise
#' @seealso [roxygen2::roxygenize()], `browseVignettes("roxygen2")`
#' @export
#' @md

Syntax

Sections

The usual Markdown heading markup creates sections and subsections. Top level headings, i.e. ‘#’ create sections, via the \section{} Rd tag. ‘#’ may only appear after the @description and @details tags. Since @details can appear multiple times in a block, you can always precede a ‘#’ section with @details, if you prefer to place it towards the end of the block, after @return for example:

#' @details
#' Trim the leading and trailing whitespace from a character vector.
#'
#' @param x Character vector.
#' @return Character vector, with the whitespace trimmed.
#'
#' @details # This will be a new section
#' ...

Top level sections are always placed at a fixed position in the manual page, after the parameters and the details, but before \note{}, \seealso{} and the \examples{}. Their order will be the same as in the roxygen block.

Subsections

Headings at level two and above may appear inside any roxygen tag that formats lines of text. E.g. @description, @details, @return, etc. They create subsections, via the \subsection{} Rd tag. They are allowed within top level sections as well, i.e. after ‘#’. Subsections can be nested. Example:

#' @details
#' ## Subsection within details
#' ### Sub-subsection
#' ... text ...

Emphasis

Emphasis and strong (bold) text are supported. For emphasis, put the text between asterisks or underline characters. For strong text, use two asterisks at both sides.

#' @references
#' Robert E Tarjan and Mihalis Yannakakis. (1984). Simple
#' linear-time algorithms to test chordality of graphs, test acyclicity
#' of hypergraphs, and selectively reduce acyclic hypergraphs.
#' *SIAM Journal of Computation* **13**, 566-579.
#' See `::is_falsy` for the definition of what is _falsy_
#' and what is _truthy_.

Code

Inline code is supported via backticks.

#' @param ns Optionally, a named vector giving prefix-url pairs, as
#'   produced by `xml_ns`. If provided, all names will be explicitly
#'   qualified with the ns prefix, i.e. if the element `bar` is defined ...

For blocks of code, put your code between triple backticks:

#' ```
#' pkg <- make_packages(
#'   foo1 = { f <- function() print("hello!") ; d <- 1:10 },
#'   foo2 = { f <- function() print("hello again!") ; d <- 11:20 }
#' )
#' foo1::f()
#' foo2::f()
#' foo1::d
#' foo2::d
#' dispose_packages(pkg)
#' ```

Note that this is not needed in @examples, since its contents are formatted as R code, anyway.

You can use similar syntax to include a block of R code and/or its output in the manual page. See section ‘Dynamic R code’ below.

Lists

Regular Markdown lists are recognized and converted to \enumerate{} or \itemize{} lists:

#' There are two ways to use this function:
#' 1. If its first argument is not named, then it returns a function
#'    that can be used to color strings.
#' 1. If its first argument is named, then it also creates a
#'    style with the given name. This style can be used in
#'    `style`. One can still use the return value
#'    of the function, to create a style function.
#' The style (the `...` argument) can be anything of the
#' following:
#' * An R color name, see `colors()`.
#' * A 6- or 8-digit hexa color string, e.g. `#ff0000` means
#'   red. Transparency (alpha channel) values are ignored.
#' * A one-column matrix with three rows for the red, green
#'   and blue channels, as returned by [grDevices::col2rgb()]

Nested lists are also supported.

Note that you do not have to leave an empty line before the list. This is different from some Markdown parsers.

Tables

Use GFM table formatting:

| foo | bar |
| --- | --- |
| baz | bim |

By default, columns are left-aligned. Use colons to generate right and center aligned columns:

| left | center | right |
| :--- | :----: | ----: |
| 1    | 2      | 3     |

Markdown hyperlinks work as usual:

#' See more about the Markdown markup at the
#' [Commonmark web site](http://commonmark.org/help)

URLs inside angle brackets are also automatically converted to hyperlinks:

#' The main R web site is at <https://r-project.org>.

Markdown notation can also be used to create links to other help topics. There are two basic forms:

  • [ref]: The target topic and the link text are one and the same.
  • [text][ref]: Link text differs from the target.

First we explore the simplest form: [ref]. The presence of trailing parentheses, e.g., [func()], signals that the target func is a function, which causes two things to happen:

  • The link text func() is automatically typeset as code.
  • The parentheses are stripped in the derived Rd link target.
[ref]
examples
Links to help
topic for …
Notes
[func()]
[pkg::func()]
a function in same
package or in pkg
Always typeset as code.
Produces Rd: \code{\link[=func]{func()}}
or \code{\link[pkg:func]{pkg::func()}}
[thing]
[pkg::thing]
a topic in same
package or in pkg
Use for a topic that documents NULL and name is set
via @name, e.g., a dataset or concept.
Not typeset as code.
Produces Rd: \link{thing} or
\link[pkg:thing]{pkg::thing}
[`thing`]
[`pkg::thing`]
a topic in same
package or in pkg
Same as above, but explicit backticks
mean that it is typeset as code.
Good for documenting a class.
Produces Rd: \code{\link{thing}} or
\code{\link[pkg:thing]{pkg::thing}}

Use the second form [text][ref] to link to the topic specified by ref, but with text as the link text.

[text][ref]
examples
Links to help
topic for …
Notes
[text][func()]
[text][pkg::func()]
a function in same
package or in pkg
Text is not typeset as code.
Produces Rd: \link[=func]{text} or
\link[pkg:func]{text}
[text][thing]
[text][pkg::thing]
a topic in same
package or in pkg
Text is not typeset as code.
Use for a topic that documents NULL
and name is set via @name,
e.g., a dataset or concept.
Produces Rd: \link[=thing]{text} or
\link[pkg:thing]{text}

In the [text][ref], the link text is treated like normal text by default.

  • Use backticks to typeset the link text as code: [`text`][ref].

It is never appropriate to use backticks around the ref in this form.

  • No, do not do this: [text][`blah-blah`]
  • Yes, do this instead: [text][blah-blah]

S3 and S4 class not done yet

Examples Help topic
for what?
Notes
[abc-class]
[pkg::abc-class]
an S4 class named
“abc” in same package
or in pkg
In Rd: \linkS4class{abc} or
\link[pkg:abc-class]{pkg::abc}
[abc][abc-class] *is this a thing? ??? \link[=abc-class]{abc}

Images

Markdown syntax for inline images works. The image files must be in the man/figures directory:

#' Here is an example plot:
#' ![](example-plot.jpg "Example Plot Title")

Dynamic R code

Similarly to the knitr package, you can use the Markdown inline code markup or Markdown code blocks to evaluate R code and insert its output into the manual page.

Inline code

To insert code inline, enclose it in `r `. Roxygen will interpret the rest of the text within backticks as R code and evaluate it, and replace the backtick expression with its value. After all such substitutions, the text of the whole tag is interpreted as Markdown, as usual.

A simple example:

#' @title Title `r 1 + 1`
#' @description Description `r 2 + 2`
#' @md
foo <- function() NULL

will be turned into:

#' @title Title 2
#' @description Description 4
#' @md
foo <- function() NULL

The resulting text, together with the whole tag is interpreted as markdown, as usual. This means that you can use R to dynamically write markdown. To markup something as code from within R, you must replace the inner backticks with the unicode hex code "\x60", rather than an actual unicode backtick character.

The interpretation from the source all the way to the final Rd documentation goes through three steps. For example:

#' Title
#' Description `r paste0('\x60', 'bar', '\x60')`
#' @md
foo <- function() NULL
  1. is run through knitr to evaluate R, yielding:

    #' Title
    #' Description `bar`
    #' @md
    foo <- function() NULL
  2. is then run through roxygen’s markdown engine, yielding:

    #' Title
    #' Description \code{bar}
    #' @md
    foo <- function() NULL
  3. is finally turned into Rd documentation using roxygen, yielding:

     % Generated by roxygen2: do not edit by hand
    % Please edit documentation in ./<text>
    \name{foo}
    \alias{foo}
    \title{Title
    Description \code{bar}}
    \usage{
    foo()
    }
    \description{
    Title
    Description \code{bar}
    } 

Code blocks

To insert entire blocks of code, enclose them in ```{r}, just like in knitr documents. They go through the same three steps described above for inline code.

For example:

#' @title Title
#' @details Details
#' ```{r lorem}
#' 1+1
#' ```
#' @md
foo <- function() NULL

becomes:

 % Generated by roxygen2: do not edit by hand
% Please edit documentation in ./<text>
\name{foo}
\alias{foo}
\title{Title}
\usage{
foo()
}
\description{
Title
}
\details{
Details

\if{html}{\out{<div class="sourceCode r">}}\preformatted{1+1
#> [1] 2
}\if{html}{\out{</div>}}
} 

Code blocks support knitr chunk options, e.g. to keep the output of several expressions together, you can specify results="hold":

#' ```{r results="hold"}
#' names(mtcars)
#' nrow(mtcars)
#' ```

Plots will create .png files in the man/figures directory. Note that plots in man/figures may quickly increase the size of your package. The file names are created from the chunk names.

#' ```{r iris-pairs-plot}
#' pairs(iris[1:4], main = "Anderson's Iris Data -- 3 species",
#'   pch = 21, bg = c("red", "green3", "blue")[unclass(iris$Species)])
#' ```

Details

This feature is powered by the knitr package.

This causes some limitations and differences to keep in mind:

  • Inline and fenced code blocks from the same roxygen block share the same evaluation environment. As a result, variables created in one code block can be reused in another one. For example:

    #' Title `r baz <- 420` `r baz`
    #'
    #' Description `r exists('baz', inherits = FALSE)`
    #' @md
    bar <- function() NULL
    
    #' Title
    #' 
    #' Description `r exists('baz', inherits = FALSE)`
    #' @md
    zap <- function() NULL

    becomes:

     % Generated by roxygen2: do not edit by hand
    % Please edit documentation in ./<text>
    \name{bar}
    \alias{bar}
    \title{Title  420}
    \usage{
    bar()
    }
    \description{
    Description TRUE
    } % Generated by roxygen2: do not edit by hand
    % Please edit documentation in ./<text>
    \name{zap}
    \alias{zap}
    \title{Title}
    \usage{
    zap()
    }
    \description{
    Description FALSE
    } 

    Notice how inside the documentation for zap(), baz does not exist (“Description FALSE”).

  • The R code is evaluated in a new environment that is the child of the package environment of the package you are documenting. This means that you can call (internal or exported) functions of the package. For example, packageName() will report the name of the current package:

    #' To insert the name of the current package: `packageName()`.
  • The code blocks are run every time you call roxygenize() (or devtools::document()) to generated the Rd files.

  • Some knitr chunk options are reset at the start of every code block, so if you want to change these, you’ll have to specify them for every chunk. These are currently error, fig.path, fig.process.

  • Some knitr options might not create meaningful output.

  • The Markdown code runs every time you call roxygenize() (or devtools::document()) to generate the Rd files. This potentially makes roxygenize() (much) slower. You can turn on knitr caching as usual, but make sure to omit the cache from the package.

  • knitr calls the appropriate print() or (if available) knitr::knit_print() methods on the result. The resulting markdown from some of these methods may not be supported by roxygen’s subset of markdown (see vignette("rd.Rmd")). You can always override the automatic methods, and have your R calls return valid markdown as a character vector, wrapped in knitr::asis_output().

Also recall that everything outside of inline or fenced code is not handled by rmarkdown, but by roxygen. Not all features from rmarkdown are available.

Roxygen and Rd tags not parsed as Markdown

Some of the roxygen tags are not parsed as Markdown. Most of these are unlikely to contain text that needs markup, so this is not an important restriction. Tags without Markdown support: @aliases, @backref, @docType, @encoding, @evalRd, @example, @examples, @family, @inheritParams, @keywords, @method @name, @md, @noMd, @noRd, @rdname, @rawRd, @usage.

When mixing Rd and Markdown notation, most Rd tags may contain Markdown markup, the ones that can not are: \acronym, \code, \command, \CRANpkg, \deqn, \doi, \dontrun, \dontshow, \donttest, \email, \env, \eqn, \figure, \file, \if, \ifelse, \kbd, \link, \linkS4class, \method, \mjeqn, \mjdeqn, \mjseqn, \mjsdeqn, \mjteqn, \mjtdeqn, \newcommand, \option, \out, \packageAuthor, \packageDescription, \packageDESCRIPTION, \packageIndices, \packageMaintainer, \packageTitle, \pkg, \PR, \preformatted, \renewcommand, \S3method, \S4method, \samp, \special, \testonly, \url, \var, \verb.

Possible problems

Mixing Markdown and Rd markup

Note that turning on Markdown does not turn off the standard Rd syntax. We suggest that you use the regular Rd tags in a Markdown roxygen chunk only if necessary. The two parsers do occasionally interact, and the Markdown parser can pick up and reformat Rd syntax, causing an error, or corrupted manuals.

Leading whitespace

Leading whitespace is interpreted by the commonmark parser, whereas it is ignored by the Rd parser (except in \preformatted{}). Make sure that you only include leading whitespace intentionally, for example, for nested lists.

Spurious lists

The Commonmark parser does not require an empty line before lists, and this might lead to unintended lists if a line starts with a number followed by a dot, or with an asterisk followed by whitespace:

#' You can see more about this topic in the book cited below, on page
#' 42. Clearly, the numbered list that starts here is not intentional.

Links to operators or objects that contain special characters, do not work currently. E.g. to link to the %>% operator in the magrittr package, instead of [magrittr::%>%], you will need to use the Rd notation: \code{\link[magrittr]{\%>\%}}.

Rd syntax

Within roxygen tags, you can use .Rd syntax to format text. Below we show you examples of the most important .Rd markup commands. The full details are described in R extensions. Before roxygen version 6.0.0 this was the only supported syntax. Now all of the formatting described below can be achieved more easily with Markdown syntax, with the important exception of mathematical expressions.

Note that \ and % are special characters. To insert literals, escape with a backslash: \\, \%.

Character formatting

  • \emph{italics}

  • \strong{bold}

  • \code{r_function_call(with = "arguments")}, \code{NULL}, \code{TRUE}

  • \pkg{package_name}

To other documentation:

  • \code{\link{function}}: function in this package

  • \code{\link[MASS]{abbey}}: function in another package

  • \link[=dest]{name}: link to dest, but show name

  • \code{\link[MASS:abbey]{name}}: link to function in another package, but show name.

  • \linkS4class{abc}: link to an S4 class

To the web:

  • \url{http://rstudio.com}

  • \href{http://rstudio.com}{Rstudio}

  • \email{hadley@@rstudio.com} (note the doubled @)

Lists

  • Ordered (numbered) lists:

    #' \enumerate{
    #'   \item First item
    #'   \item Second item
    #' }
  • Unordered (bulleted) lists

    #' \itemize{
    #'   \item First item
    #'   \item Second item
    #' }
  • Definition (named) lists

    #' \describe{
    #'   \item{One}{First item}
    #'   \item{Two}{Second item}
    #' }

Mathematics

Standard LaTeX (with no extensions):

  • \eqn{a + b}: inline equation

  • \deqn{a + b}: display (block) equation

Tables

Tables are created with \tabular{}. It has two arguments:

  1. Column alignment, specified by letter for each column (l = left, r = right, c = centre.)

  2. Table contents, with columns separated by \tab and rows by \cr.

The following function turns an R data frame into the correct format, adding a row consisting of the (bolded) column names and prepending each row with #' for pasting directly into the documentation.

tabular <- function(df, ...) {
  stopifnot(is.data.frame(df))

  align <- function(x) if (is.numeric(x)) "r" else "l"
  col_align <- purrr::map_chr(df, align)

  cols <- lapply(df, format, ...)
  contents <- do.call("paste",
    c(cols, list(sep = " \\tab ", collapse = "\\cr\n#'   ")))

  paste("#' \\tabular{", paste(col_align, collapse = ""), "}{\n#'   ",
    paste0("\\strong{", names(df), "}", sep = "", collapse = " \\tab "), " \\cr\n#'   ",
    contents, "\n#' }\n", sep = "")
}

cat(tabular(mtcars[1:5, 1:5]))
#> #' \tabular{rrrrr}{
#> #'   \strong{mpg} \tab \strong{cyl} \tab \strong{disp} \tab \strong{hp} \tab \strong{drat} \cr
#> #'   21.0 \tab 6 \tab 160 \tab 110 \tab 3.90\cr
#> #'   21.0 \tab 6 \tab 160 \tab 110 \tab 3.90\cr
#> #'   22.8 \tab 4 \tab 108 \tab  93 \tab 3.85\cr
#> #'   21.4 \tab 6 \tab 258 \tab 110 \tab 3.08\cr
#> #'   18.7 \tab 8 \tab 360 \tab 175 \tab 3.15
#> #' }