This workflow philosophy relies on RMarkdown files being run in a defined sequence. It follows that an RMarkdown file should not read from the data written by another RMarkdown file written later in the sequence. These functions help you implement this idea.
These functions are function factories; they themselves return functions. You can use those returned functions to access paths.
To make things a little easier, the template used by
proj_workflow_use_rmd() includes a calls to proj_path_source()
and proj_path_target(), with the name argument populated.
proj_path_target(name)
proj_path_source(name)character name of the workflow component.
function that acts like here::here(),
returning a character path.
Each RMarkdown file in the sequence has its own target directory, created
using proj_create_dir_target(). Once a target directory is created, use
these functions to create functions to access your target directory,
or previous RMarkdown files' target directories (as sources).
For example, use proj_path_target() to create a path-generating function
that uses your target directory. Whenever you need to provide a path to a
file in your target directory, e.g. the file argument to
write.csv(), use this path-generating function.
Similarly, you can use proj_path_source() to create a path-generating
function for your source directories, which must be earlier in the
workflow than your current RMarkdown file. The path-generating function
ckecks that the source directory is, in fact, earlier in the workflow.
# create path-generating functions
path_target <- proj_path_target("01-clean")
path_source <- proj_path_source("01-clean")
# not run because they depend on side effects
if (FALSE) {
# use path-generating functions
# returns full path to a file in one of your source directories
path_source("00-import", "old-data.csv")
# returns full path to a file in your target directory
path_target("new-data.csv")
}