Draft of a script that removes all sorts of stuff from vendored rust crates to reduce size

shrink.R

#!/usr/bin/env Rscript
# we need to vendor the dependencies and then store the checksum
rextendr::vendor_pkgs()

# make the checksum
# writeLines(
#   tools::md5sum("src/rust/vendor.tar.xz"),
#   "./tools/vendor.md5"
# )

all_vendored_files <- list.files("src/rust/vendor", full.names = TRUE, recursive = TRUE, all.files = TRUE)
file_exts <- tools::file_ext(all_vendored_files)

# use this to identify directories we might want to remove
# list.dirs("src/rust/vendor", T, F) |>
#   lapply(list.dirs, F, F) |>
#   unlist() |>
#   table() |>
#   as.data.frame()

# identify all md files
mds <- file_exts == "md"

# identify all licenses
license_files <- stringr::str_detect(
  all_vendored_files,
  stringr::coll("LICENSE", ignore_case = TRUE)
)

# identify all files in benches, tests, examples, misc, ci, and ui-tests
dir_idx <- stringr::str_detect(
  all_vendored_files,
  "/benches/|/tests/|/examples/|/misc/|/ci/|/ui-tests/|/guide/"
)

# if any of these are true, we want to remove them
to_remove <- dir_idx | mds | license_files

# remove all of these files
file.remove(all_vendored_files[to_remove])

# identify all of rust files that havent been removed
all_rust_sources <- all_vendored_files[file_exts == "rs" & !to_remove]

# remove all of the comments from the source files
for (fp in all_rust_sources) {
  file_lines <- brio::read_lines(fp)
  brio::write_lines(file_lines[!stringr::str_detect(file_lines, "^\\s*//.*$")], fp)
}

# we need to recompute the hashes for all files in each .cargo-checksum.json
all_checksums <- all_vendored_files[stringr::str_detect(
  all_vendored_files,
  ".cargo-checksum.json$"
)]

# remove checksums 
for (fp in all_checksums) {
  # read the json file
  cs <- yyjsonr::read_json_file(fp)

  # calculate the new checksums
  cs[["files"]] <- lapply(cs$files, openssl::sha256)

  # handle null package
  if (is.null(cs[["package"]])) {
    cs[["package"]] <- "null"
  }

  yyjsonr::write_json_file(cs, fp, opts = list(auto_unbox = TRUE))

}

# recompress
src_dir <- rprojroot::find_package_root_file("src/rust")
compress_res <- withr::with_dir(src_dir, {
  processx::run(
    "tar", c(
      "-cJ", "--no-xattrs", "-f", "vendor.tar.xz", "vendor"
    )
  )
})