Read R’s C headers to identify where R functions are coming from

read-r-headers.R

library(dplyr)

api <- as_tibble(tools:::funAPI())


# list all header files from include dir
header_files <- list.files(
  R.home("include"),
  full.names = TRUE,
  recursive = TRUE,
  pattern = "*.h"
)

read_header <- function(.h) {
  # find include dir
  include_dir <- R.home("include")
  # generate the clang ar #thanksChatGPT
  system_cmd <- sprintf(
    "clang -Xclang -ast-dump=json -fsyntax-only -x c -I%s %s",
    include_dir,
    .h
  )

  # parse and read
  json <- paste(system(system_cmd, intern = TRUE), collapse = "")
  in_header <- yyjsonr::read_json_str(json)[[c("inner", "name")]]

  header_fns <- tibble(
    name = in_header,
    header_path = gsub(paste0(include_dir, "/"), "", .h)
  )
  header_fns
}

all_api_fns <- lapply(header_files, read_header) |>
  bind_rows()

res <- inner_join(api, all_api_fns, by = "name")