Add a Generated Table of Contents Anywhere in RMarkdown

R Markdown

February 28, 2018

Links: Github Gist

GitHub user @stanstrup posted a question today on the blogdown GitHub repo about manually positioning a table of contents in blogdown:

When I use toc: true in a post the toc is inserted at the very top of the post. … If you could specify the position of the toc with some keyword you could work around it.

I don’t use the academic theme for Hugo (I use a modified version of hyde), so I’m not entirely sure if I can completely solve stanstrup’s problems, but I know I’ve run into something similar recently.

And while Yihui is probably right that the effort isn’t worth it when fiddling with trivial aesthetics, I use R Markdown in enough places and have run into this a few times. Knowing that someone else out there felt the same pain was enough to push me to code up a quick solution.

The function I’ve worked up is called render_toc() and it allows you to drop in a table of contents anywhere inside an R Markdown document. This means you can use it to manually position a table of contents in:

and many more places.

Get It

I’ve posted the function and an example document as a GitHub Gist. To use it in your document, choose one of the following:

  1. Download render_toc.R and source("render_toc.R") in your project or script

  2. Copy the function code into your RMarkdown document

  3. Source the function from GitHub using devtools:

                          filename = 'render_toc.R')

Use It

I included an example file in the GitHub Gist. Essentially, you just need to source render_toc.R somewhere (such as a setup chunk) and then call it in the document where you want to render the table of contents.

The output will just be a markdown list, so if you want to give the table of contents it’s own header, you’ll have to include that in the document.

Here’s what a simple R Markdown document would look like.

```{r setup, include=FALSE} 
knitr::opts_chunk$set(echo = TRUE)
                      filename = 'render_toc.R')

## Table of Contents

```{r toc, echo=FALSE} 

# Writing

## R Markdown

This is an R Markdown document...

```{r cars} 
# This is not a header

## Regular Code

# Regular markdown code (not run)

# Plots

## Including Plots {#plots-are-here .class-foo}

You can also embed plots, for example:

```{r pressure, echo=FALSE} 

which outputs as this document (click to view image).

Behind the Scenes

The function simply reads through the lines of the RMarkdown document and strips out any code blocks. The supported code fencing style is three or more ` characters in a row.

Then I extract the headers, which must be in the hashtag-style to work. In other words headers like this

## A Nice Header

work well, while headers like these won’t be processed

A Not So Nice Header

The function creates the header anchor if not manually specified – see the pandoc header identifiers help page for more information – or uses the identifier if it is included.

The example above would link to #a-nice-header and the example below links to #my-shortcut

## An Overly Wordy Header Title {#my-shortcut}

Any headers with a higher depth than the toc_depth parameter (default is 3) are discarded. Also any initial headers prior to the first base level header with higher levels (say ### when the base level is ##) are discarded as well.

Finally, if toc_header_name is set, the header with that name is discarded so that the TOC itself isn’t included in the TOC.

The end result is a simple markdown list that can be rendered anywhere!

                      filename = 'render_toc.R')

# `this_post` is set in the setup chunk,
# points to the Rmd file for this post

Which, underneath, is just markdown.

- [Get It](#get-it)
- [Use It](#use-it)
- [Behind the Scenes](#behind-the-scenes)

Let me know on twitter @grrrck if you found this helpful or run into any issues!