tar_bookdown: Target factory for bookdown

In edalfon/flowme: Add boilerplate code to your project for some specific workflows

Target factory for bookdown

Description

{targets} and {tarchetypes} make it straightforward to include individual "dependency-aware R Markdown reports inside the individual targets of a pipeline" But a {bookdown}-based approach can be tricky. This function creates targets to let you concisely (e.g. tar_bookdown("report")) include a bookdown report in your pipeline, letting targets do its magic handling dependencies.

Usage

tar_bookdown(
  input_dir = "report",
  input_files = ".",
  output_dir = NULL,
  output_format = NULL,
  preview = FALSE
)

Arguments

input_dir	the main directory of the book
input_files	character vector with input files, in case you do not want to render them all (the default)
output_dir	The output directory. If NULL, a field named output_dir in the configuration file ‘_bookdown.yml’ will be used (possibly not specified, either, in which case a directory name ‘_book’ will be used).
output_format	as in bookdown::render_book
preview	Whether to render and preview the input files specified by the input argument. Previewing a certain chapter may save compilation time as you actively work on this chapter, but the output may not be accurate (e.g. cross-references to other chapters will not work).

Details

So, this function

• receives input_dir, typically a subfolder in your project that includes all .Rmd files and {bookdown} config files

• creates one target per .Rmd file in input_dir, making sure to include its dependencies (e.g. tar_read, tar_load calls within .Rmd files, detected via tarchetypes::tar_knitr_deps)

• includes also a single target for the {bookdown} report, that depends on all the individual targets corresponding to .Rmd files

• returns all the targets above as a list to be included in your pipeline (please note all targets are format = "file")

Note that when running your pipeline, the bookdown target renders the book using flowme::bookme which is simply a wrapper around bookdown::render_book to deal with {bookdown}'s restrictions and be able to render in subdirectories. This collides with the stricter (compared to .drake) policy of {targets} to find the data store (_targets/) at the project root.

EDIT: now it seems {targets} is more flexible and you can configure the location using tar_config_set(). Yet, the discussion below still applies because we need to deal with temporarily changing working directory to make bookdown work).

flowme::bookme changes working directory to input_dir to render the book. Thus, when you call tar_read in an .Rmd, {targets} will look for the data store in input_dir and not in your project root where it probably lies, leading your pipeline to fail. To circumvent this issue, there are a few alternatives you can use when retrieving targets in your .Rmd.

• change the store directly to a hard-coded value, which is probably the less verbose alternative, but it is not robust or portable (e.g. if you nest the bookdown folder into other directories)

  • tar_read(data, store = "..")

• also change the store, but using here to help you find the data store

  • tar_read(data, store = here::here(targets::tar_config_get("store")))

• change the working directory again, temporarily

  • xfun::in_dir(here::here(), tar_read(data)) # or withr::with_dir

• TODO: there might be a better way to do it. Try and find it. Meanwhile, this works.

Value

a list of targets, including one target for each .Rmd file in input_dir and one target for the bookdown output, that depends on all .Rmd files

Examples

## Not run: 
tar_bookdown("report")

## End(Not run)

edalfon/flowme documentation built on April 22, 2024, 7:37 p.m.