Utilise the synergy of algorithms and heuristics to unlock knowledge of R script/package dependencies.
Your first forays into understanding someone else's R analysis often involve flailing around trying to obtain all the required dependencies. This package will manage that for you by identifying your missing and out of date dependencies and offering a choice to automatically install & update.
depl_check(source_path) will examine the code found at or below source_path and produce a dependency report:
> depl_check()
[deplearning] Starting dependency check.
[deplearning] Searching path .
[deplearning] Scanning 7 R source files... done.
[deplearning] Found 29 dependencies.
[deplearning] Fetching remote data... CRAN... GitHub... done.
[deplearning] ✔ 14 Installed and up to date.
purrr, memoise, jsonlite, rstudioapi, tibble, clisymbols, knitr, tidyr, packagemetrics, xml2, devtools, testthat
[deplearning] ✖ 2 Installed but behind CRAN release.
package installed_ver CRAN_ver
visdat 0.0.5.9000 0.1.0
dplyr 0.7.1 0.7.2
[deplearning] ✖ 8 Missing CRAN packages.
datapasta, A3, abc, switchr, dejaVu, Rborist, kpmt, fasjem
[deplearning] ✖ 4 Missing GitHub packages.
njtierney/naniar, maelle/rtimicropem, benmarwick/rrtools, MilesMcBain/packup
[deplearning] ✖ 1 Missing packages from untracked repositories.
notARealPackage
[deplearning] ✔ Minimum R version to update & install is 3.3.1, you have 3.4.0.
[deplearning] ℹ Update & install will include 18 new recursive dependencies.
maps, gistr, assertive.base, assertive.files, assertive.numbers, assertive.properties, assertive.reflection, assertive.strings, assertive.types, pbapply, abc.data, quantreg, locfit, SparseM, MatrixModels, RJSONIO, RcppArmadillo, matrixStats
[deplearning] Would you like to update & install old and missing dependencies?
1: Yes
2: No
Selection: 2
[deplearning] Dependency check finished without install & update.
There is also an RStudio addin hook to search the dependencies in the currently active window.
Dependencies are identified using by pattern matching against common methods: library(...), require(...), p_load(..., ..., ), ...::func()
The plan is to extend this idea to:
- Estimation of time until code can run based on analysis of recursive dependencies.
- jimhester for code I adapted from his package
autoinst. - gepuro for the closest thing R has to a package name resolution service, and for making me an API! 😍
- gmbecker for
switchr, without which testing this package would have been a cumbersome nightmare.