Count usages of data.table functions in Importing revdeps

dt_usage_revdeps.R

#!/usr/local/bin/Rscript
library(data.table)

imports_dt <- tools::dependsOnPkgs(
  "data.table", "Imports",
  recursive = TRUE,
  installed = available.packages()
)

tar_dir <- "/media/michael/69913553-793b-4435-ac82-0e7df8e34b9f/cran-mirror/src/contrib"
all_pkg <- list.files(tar_dir, pattern = "\\.tar\\.gz$", full.names = TRUE)

dt_funs <- grep("^[.:]", getNamespaceExports("data.table"), invert = TRUE, value = TRUE)
infix_idx <- grepl("^%.*%$", dt_funs)
dt_funs[!infix_idx] <- paste0("\\b", dt_funs[!infix_idx], "\\(")

counts <- data.table(fun = dt_funs, count = 0L, key = 'fun')

n_pkg <- 0L

for (pkg in imports_dt) {
  tmp <- file.path(tempdir(), pkg)
  
  pkg_path <- grep(sprintf("/%s_[0-9.]+\\.tar\\.gz", pkg), all_pkg, value = TRUE)
  
  if (!length(pkg_path)) next
  
  n_pkg <- n_pkg + 1L
  pkg_path <- pkg_path[1L]
  
  # --strip-components makes sure the output structure is
  # /path/to/tmp/pkg/ instead of /path/to/tmp/pkg/pkg
  utils::untar(pkg_path, exdir = tmp, extras="--strip-components=1")
  
  r_src <- lapply(list.files(file.path(tmp, "R"), pattern = "\\.R", full.names = TRUE), readLines, warn = FALSE)
  r_src <- unlist(r_src)
  
  pkg_counts <- sapply(dt_funs, function(fun) sum(grepl(fun, r_src)))
  counts[.(fun = names(pkg_counts)), count := count + pkg_counts]
  
  unlink(tmp, recursive = TRUE)
}

counts[ , fun := gsub("^[\\]b|[\\][(]$", "", fun)]
counts[ , usage_per_pkg := count/n_pkg]

counts[order(-count)]