Parse R code for required packages and install them. Optionally control which package versions to install using a deps.yaml file that can be created manually or automatically created based on a directory of R source files and the versions and sources of packages found in the local R library.

Install the latest stable version from CRAN with install.packages('automagic'). Install the latest development version from GitHub with remotes::install_github('cole-brokamp/automagic').

To make sure R has all the required packages before running any R code, run automagic::automagic(). If there is no deps.yaml file present, automagic searches all .R and .Rmd files in the current working directory for necessary packages and attempts to install them from CRAN. Unlike previous versions of the automagic package, if a package is not available on CRAN, it will not attempt to install it from GitHub based on a best guess.

It is possible that automagic might mistakenly install the wrong package from GitHub or you might need a different version of an R package for the code to work as intended. In this case, you can create a deps.yaml file with automagic::make_deps_file(). This function parses R code and then queries the local R package libraries to determine the exact source and version of each package to install. Currently, only CRAN and GitHub packages are supported.

An example deps.yaml file looks like

- Package: CB
  GithubUsername: cole-brokamp
  GithubRepo: CB
  GithubRef: master
  GithubSHA1: 0a56eadaf4282678c949bc8eedaacb5d6e0777fe
- Package: remotes
  Repository: CRAN
  Version: 1.0.0

The dependencies file could also be created or changed manually if necessary. For example, you could create a list of packages that you frequently depend on and when moving to a new machine or server, run automagic::automagic() to install them all.

GitHub packages are installed using the remotes package. Set the environment variable GITHUB_PAT to supply a personal access token (PAT) to install a package from a private repository or to increase the limit of calls to the GitHub API during installation of a large number of packages.

See https://github.com/cole-brokamp/rize for robust method of dockerizing Shiny applications within R. rize utilizes automagic for documenting and installing the packages that are needed to run the R Shiny app in a Docker container.