pkgfilecache: Manage optional data for GNU R packages

pkgfilecache: Manage optional data for GNU R packages

Audience

This is intended for GNU R package developers who want to give package users the option to effortlessly download and mange optional data for their package.

Background

When package authors want to ship data for their package, they will quickly hit the package size limit on CRAN (which is 5 MB as of September 2019). The solution is to host the data elsewhere and download it on demand when the user requests it, then store it for future use. This is what pkgfilecache allows you to do. You can put your files onto a web server of your choice, take the MD5 sums, and have pkgfilecache download them locally unless they already exist and have the correct MD5 hash. Users can then access the data in a convenient way, similar to accessing files shipped in inst/extdata via system.file. They can also erase the data if it is no longer needed.

Technically, this package permanently stores the files under a subdir of the directory returned by rappdirs::user_data_dir. For my Ubuntu Linux system, that is /home/myuser/.local/share, but the exact location is operating system-dependent and you should not care about it.

Prequisites

You have put some file onto a server that can be accessed by the public via HTTP or HTTPS. Optionally, you know their MD5 checksums.

In the following example, we will assume that the package you develop (the one that requires the extra data in the package cache), is called yourpackage. And that you have these two files on the server:

Using the functions in your script

You can use this package to manage any files you want in normal application code, and this is explained here. If you are a package author and want to give your users the option to effortlessly download optional data for your package, make sure to also read the section ‘Giving users the option to download extra package data’ below.

Making files from a remote server available locally (on the user computer)

Let’s first make files hosted on our server available on the client, in the package cache. First, define the files you want:

library("pkgfilecache")

pkg_info = get_pkg_info("yourpackage");   # Something to identify the package that uses the package file cache.

local_filenames = c("file1.txt", "file2.txt");    # How the files should be called in the local package file cache
urls = c("https://your.server/yourpackage/large_file1.txt", "https://your.server/yourpackage/large_file2.txt"); # Remote URLs where to download files from
md5sums = c("35261471bcd198583c3805ee2a543b1f", "85ffec2e6efb476f1ee1e3e7fddd86de");    # MD5 checksums. Optional but recommended.

Now it is time to make them available:

res = ensure_files_available(pkg_info, local_filenames, urls, md5sums=md5sums);

The return value res is a named list with several entries. The field res$file_status is a logical vector indicating for each file whether it now exists locally. Note that the function will first check to see whether the files are already in the package cache. If you supplied md5sums, they will also be checked. Only files which did not pass the check will be downloaded, so it is save to call this function every time you want to make sure that the files exist. (E.g., in example code you give).

You can also get a list of the filename which are now available (no matter whether they have been downloaded or were already available) from res$available. And you can get a list of file which should have been downloaded, but for which retrieval failed, from res$missing.

Storing files in local subdirectories of the package cache

The example above stores all files in the base directory of the package cache. To store a file in a local subdirectory, pass a list of vectors instead. Each entry in the list will be interpreted as the relative path to the local file:

local_relative_filenames = list(c("dir1", "file1.txt"), c("dir2", "file2.txt"));

That example will download the files to dir1/file1.txtand dir2/file2.txt. The directories will be created automatically if they do not exist.

Accessing a file from the local package cache

Let’s now get the full path for a file in the package cash, so that we can use it in our application:

wanted_local_file = "file1.txt";
file_path = get_filepath(pkg_info, wanted_local_file, mustWork=TRUE);

If the file is in a local sub directory, use a list with one element, a vector of path elements:

file_path = get_filepath(pkg_info, list(c("test", "file1")), mustWork=TRUE);

Removing files you do not need anymore from the package cache

local_filenames = c("file1.txt", "file2.txt");
deleted = remove_cached_files(pkg_info, local_filenames);

The return value deleted is a logical vector indicating for each file whether it was deleted. IMPORTANT: Files that did not exist in the first place were not deleted. To check which files really exist, read on.

Manually checking whether a file exists in the package cache

Do one of the following, depending on whether you want MD5 sum checks:

files_exist = are_files_available(pkg_info, local_filenames);  # no MD5 check
files_exist_and_have_correct_md5 = are_files_available(pkg_info, local_filenames, md5sums=md5sums);  # with MD5 check

The return values are logical vectors indicating for each file whether it exists (and, in the second example, whether the MD5 sum is as expected).

Giving users the option to download extra package data

This part is for package authors and gives a suggestion for an interface that allows users to download optional data for your package. Make sure you have read the section ‘Using the functions in your script’ before reading on.

Interface suggestion

I recommend to have the following public functions in your package, which users can then call if they need to manage optional package data:

Example implementation of the interface for your package

Here are example implementations for the functions above. You can copy and paste them, all you need to do afterwards is:

Here are the functions:

#' @title Download optional data for this package if required.
#'
#' @description Ensure that the optioanl data is available locally in the package cache. Will try to download the data only if it is not available.
#'
#' @return Named list. The list has entries: "available": vector of strings. The names of the files that are available in the local file cache. You can access them using get_optional_data_file(). "missing": vector of strings. The names of the files that this function was unable to retrieve.
#'
#' @export
download_optional_data <- function() {
  pkg_info = pkgfilecache::get_pkg_info("yourpackage");        # to identify the package using the cache

  # Replace these with your optional data files.
  local_filenames = c("file1.txt", "file2.txt");    # How the files should be called in the local package file cache
  urls = c("https://your.server/yourpackage/large_file1.txt", "https://your.server/yourpackage/large_file2.txt"); # Remote URLs where to download files from
  md5sums = c("35261471bcd198583c3805ee2a543b1f", "85ffec2e6efb476f1ee1e3e7fddd86de");    # MD5 checksums. Optional but recommended.

  cfiles = pkgfilecache::ensure_files_available(pkg_info, local_filenames, urls, md5sums=md5sums);
  cfiles$file_status = NULL;
  return(cfiles);
}

#' @title Get file names available in package cache.
#'
#' @description Get file names of optional data files which are available in the local package cache. You can access these files with get_optional_data_file().
#'
#' @return vector of strings. The file names available, relative to the package cache.
#'
#' @export
list_optional_data <- function() {
  pkg_info = pkgfilecache::get_pkg_info("yourpackage");
  return(pkgfilecache::list_available(pkg_info));
}


#' @title Access a single file from the package cache by its file name.
#'
#' @param filename, string. The filename of the file in the package cache.
#'
#' @param mustWork, logical. Whether an error should be created if the file does not exist. If mustWork=FALSE and the file does not exist, the empty string is returned.
#'
#' @return string. The full path to the file in the package cache. Use this in your application code to open the file.
#'
#' @export
get_optional_data_filepath <- function(filename, mustWork=TRUE) {
  pkg_info = pkgfilecache::get_pkg_info("yourpackage");
  return(pkgfilecache::get_filepath(pkg_info, filename, mustWork=mustWork));
}


#' @title Delete all data in the package cache.
#'
#' @return integer. The return value of the unlink() call: 0 for success, 1 for failure. See the unlink() documentation for details.
#'
#' @export
delete_all_optional_data <- function() {
  pkg_info = pkgfilecache::get_pkg_info("yourpackage");
  return(pkgfilecache::erase_file_cache(pkg_info));
}

Limitations and Suggestions