Skip to contents

Collect the package dependencies a set of code files declare

Source: R/reproducibility_check.R
repro_dependencies.Rd

Walks each code file with code_library_names() and returns the distinct packages the paper's code loads, tagged with how each was referenced and whether it is installable from CRAN, from a named GitHub/URL source, or is a base/recommended package already shipped with R (so a run need not install it). This is the input list a run would install; it is deliberately name-only, because static analysis cannot know which version was used — running against the latest packages and observing a break is itself a finding (it argues the authors should have pinned versions).

Usage

repro_dependencies(code_text, lang = "R")

Arguments

code_text

the code text for a single file (character vector), OR a list of such vectors (one per file) to pool across files

lang

the language (only R declares installable packages here)

Value

a data frame with columns package, source (cran, bioc, github, url, or base), ref (the source path/ref for github/url, else NA), and base (logical, TRUE for a base/recommended package). One row per distinct package. Empty frame (same columns) when none are found.

Details

GitHub/URL sources are read from the code text, not the package name: a call remotes::install_github("user/repo") names the source user/repo, which is matched back to the package by the repo's basename. When a script pins a ref (user/repo@v1.2), the ref is kept.

Examples

code_text <- c(
  "library(dplyr)",
  "remotes::install_github('tidyverse/ggplot2@v3.4.0')",
  "x <- stats::sd(1:10)"
)
repro_dependencies(code_text)
#>   package source  ref  base
#> 1   dplyr   cran <NA> FALSE
#> 2 remotes   cran <NA> FALSE
#> 3   stats   base <NA>  TRUE