R/bcat_import_data.R

In Rbearcat: University of Cincinnati Themes and Utilities for Econometrics and Data Science

#' Import data
#'
#' Utility to import multiple data sets from a directory into a named list.
#'
#' @param location Character. Directory from which to import data.
#' @param extension Character. File extension to match (without ".").
#' @param import_function Function used to read each file (e.g., \code{readr::read_csv}).
#' @param recursive Logical. Search subdirectories? Default is \code{TRUE}.
#'
#' @return A named list of imported objects. Names are derived from filenames
#'   without their extensions, and each element contains the result returned by
#'   \code{import_function()} for one matching file.
#'
#' @author Saannidhya Rawat
#' @family utilities
#' @export
#' @importFrom stats setNames
#'
#' @examples
#' tmp_dir <- tempfile("rbearcat-data-")
#' dir.create(tmp_dir)
#' csv_path <- file.path(tmp_dir, "example.csv")
#' utils::write.csv(data.frame(x = 1:3, y = letters[1:3]), csv_path, row.names = FALSE)
#'
#' datasets <- bcat_import_data(tmp_dir, "csv", utils::read.csv)
#' names(datasets)
#' datasets$example
#'
#' unlink(tmp_dir, recursive = TRUE)
bcat_import_data <- function(location, extension, import_function, recursive = TRUE) {

  # Find matching files
  files <- list.files(location,
                      pattern = paste0("\\.", extension, "$"),
                      recursive = recursive,
                      full.names = TRUE)

  if (length(files) == 0L) {
    warning("No files with extension '", extension, "' found in '", location, "'.",
            call. = FALSE)
    return(list())
  }

  # Derive clean names (filename without extension or path)
  dataset_names <- tools::file_path_sans_ext(basename(files))

  # Check uniqueness
  if (anyDuplicated(dataset_names)) {
    stop("File names are not unique. All files must have unique names, ",
         "even if in separate subfolders. Duplicated: ",
         paste(dataset_names[duplicated(dataset_names)], collapse = ", "),
         call. = FALSE)
  }

  # Import all files
  datasets <- lapply(files, import_function)
  stats::setNames(datasets, dataset_names)
}