targets-parameterized-bookdown

Example repository for rendering a bookdown book, compiling parameterized Rmarkdown documents, using the targets package.

This example extends the targets-minimal example. To mimic a directory of input files, we split the example data into two chunks.

library(tidyverse)
data <- read_csv('data/raw_data.csv')

nrows <- 50

data1 <- head(data, n = nrows)
data2 <- tail(data, n = nrows)

write_csv(data1, 'data/split/data1.csv')
write_csv(data2, 'data/split/data2.csv')

To setup a target for our input files, we use tar_files from tarchetypes (target: paths). Dynamic branching in subsequent targets is mapped over these files (targets: raw_data, data, hist, fit).

Parameterized RMarkdown documents are generated by mapping over the fitted model and histograms using a template document (template/template.Rmd) (target: report).

Finally, the output documents are merged into a book with bookdown. We use a custom wrapper function render_with_deps to explicitly declare the book's dependency on the output reports (target: book).

Notes

• We need to change the extension of the parameterized github_document outputs (target: report) from .md to .Rmd because bookdown doesn't pick up .md files unless they are explicitly listed with the rmd_files argument in _bookdown.yml. See here.

• The title and output_file for rendered RMarkdown documents are set in this example to the name of the branch (target: report). If adapted for your own use, this could alternatively be set to an id from the data, etc.

• We can't use the tar_render function from tarchetypes because we need to call bookdown::render_book and not rmarkdown::render (target: book). Therefore, we need to track the .Rmd files in separate targets.

• You may need to use pandoc >= 2.11.2 when generating the github_documents (target: report) so ATX style headings are used by default.