Get direct URL of Wiki Commons image by filename from Media Wiki API in R/rstats

Assume you have the file name of an image in Wikimedia, like MJK_16033_Rosa_von_Praunheim_(Berlinale_2018)_crop.jpg and you want to get the direct URL of that image. In this case, this URL is called https://upload.wikimedia.org/wikipedia/commons/8/85/MJK_16033_Rosa_von_Praunheim_%28Berlinale_2018%29_crop.jpg.
The part after commons os from the MD5 hash (see comment) That part after commons cannot be guessed, but can also retrieved from the Media Wiki API.

A function to that in R with httr2 package:

get_url_of_wikimedia_img_file <- function(wiki_media_file_name) {
  media_wiki_api <- URLencode(paste0("https://en.wikipedia.org/w/api.php?action=query&titles=File:", wiki_media_file_name, "&prop=imageinfo&iiprop=url&format=json"))
  res <- httr2::request(media_wiki_api) |> httr2::req_perform() |> httr2::resp_body_json()
  url <- res$query$pages$`-1`$imageinfo[[1]]$url
  url
}

P.S. Why do I need to do that? I want to automatically import those images in Omeka S – and if a URL is not working, the whole Omeka object won’t be created.