Merge changes in Rc/v2.1.0 to main branch

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: dv.loader
 Type: Package
 Title: Data loading module 
-Version: 2.0.0
+Version: 2.1.0
 Authors@R: c(
         person( "Boehringer-Ingelheim Pharma GmbH & Co.KG", role = c("cph", "fnd")),
         person( given = "Ming", family = "Yang", role = c("aut", "cre"), email = "[email protected]"),
@@ -13,10 +13,13 @@ License: Apache License (>= 2)
 Encoding: UTF-8
 LazyData: true
 Depends: R (>= 3.5.0)
-Imports: haven
+Imports:
+    checkmate,
+    haven
 Suggests: 
-    testthat,
+    testthat (>= 3.0.0),
     knitr,
     rmarkdown
 RoxygenNote: 7.3.0
 VignetteBuilder: knitr
+Config/testthat/edition: 3

NAMESPACE

@@ -1,5 +1,7 @@
 # Generated by roxygen2: do not edit by hand
 
 export(get_cre_path)
+export(get_file_paths)
 export(get_nfs_path)
 export(load_data)
+export(load_data_files)

NEWS.md

@@ -1,3 +1,11 @@
+# dv.loader 2.1.0
+
+- Refactored code to improve readability and maintainability.
+
+- Fixed issue of partial matching when the `file_names` argument contains no file extensions.
+
+- Added arguments `env_var` and `print_file_paths` in `load_data()` function to provide more flexibility and control.
+
 # dv.loader 2.0.0
 
 - GitHub release with QC report

R/dvloader.R

@@ -1,61 +1,138 @@
-#' gets the NFS base path from an env var
-#' It assumes there is an env var
-#' called RXD_DATA which holds the path suffix.
-#' @return the NFS base path
-#' @export
-get_nfs_path <- function() {
-  base_path <- Sys.getenv("RXD_DATA")
-  # check that RXD_DATA is set
-  if (base_path == "") {
-    stop("Usage: get_nfs_path: RXD_DATA must be set")
+#' Get Base Directory Path
+#'
+#' This function retrieves the base directory path from a specified environment variable.
+#' It checks if the environment variable is set and if the directory exists.
+#'
+#' @param env_var [character(1)] The name of the environment variable containing the base directory path.
+#'
+#' @return [character(1)] The normalized path to the base directory.
+#'
+#' @examples
+#' # Create a temporary directory
+#' temp_dir <- tempdir()
+#'
+#' # Set the BASE_DIR environment variable
+#' Sys.setenv(BASE_DIR = temp_dir)
+#'
+#' # Get the base directory path
+#' dv.loader:::get_base_dir("BASE_DIR")
+#'
+#' @keywords internal
+get_base_dir <- function(env_var) {
+  # Ensure env_var is a single character string
+  checkmate::assert_character(env_var, len = 1)
+
+  # Get the value of the environment variable
+  base_dir <- Sys.getenv(env_var)
+
+  # Stop if the environment variable is not set
+  if (base_dir == "") {
+    stop("Environment variable ", env_var, " is not set")
   }
-  return(base_path)
+
+  # Ensure the directory exists
+  checkmate::assert_directory_exists(base_dir)
+
+  # Return the normalized path
+  return(normalizePath(base_dir))
 }
 
-#' gets the NFS base path from an env var
-#' alias for get_nfs_path to maintain backwards compatibility
+#' Get NFS Path
+#'
+#' This function retrieves the path to the NFS (Network File System) directory.
+#'
+#' @param env_var [character(1)] The environment variable name for the base directory. Default is "RXD_DATA".
+#'
+#' @return [character(1)] The path to the NFS directory.
+#'
 #' @export
-get_cre_path <- get_nfs_path
-
-#' Loads data into memory based on study directory and one or more file_names.
-#' @param sub_dir A relative directory/folder that will be appended to a base path defined by `Sys.getenv("RXD_DATA")`.
-#' If the argument is left as NULL, the function will load data from the working directory `getwd()`.
-#' @param file_names Study file or file_names name(s) - can be a vector of strings.
-#' This is the only required argument.
-#' @param use_wd for "use working directory" - a flag used when importing local files
-#' not on NFS - default value is FALSE
-#' @param prefer_sas if set to TRUE, imports sas7bdat files first before looking for
-#' RDS files (the opposite of default behavior)
-#' @return a list of dataframes
+get_nfs_path <- function(env_var = "RXD_DATA") {
+  get_base_dir(env_var = env_var)
+}
+
+
+#' Get CRE Path
+#'
+#' This function retrieves the path to the CRE (Clinical Research Environment) directory.
+#' It uses the "RXD_DATA" environment variable as the base directory.
+#'
+#' @return [character(1)] The path to the CRE directory.
+#'
 #' @export
+get_cre_path <- function() {
+  get_base_dir(env_var = "RXD_DATA")
+}
+
+
+#' Load Data Files
+#'
+#' This function loads data files from a specified directory or the current working directory.
+#' It supports loading both RDS and SAS7BDAT files.
+#'
+#' @param sub_dir [character(1)] Optional character string specifying a subdirectory. Default is NULL.
+#' @param file_names [character(1+)] Character vector of file names to load (without extension).
+#' @param use_wd [logical(1)] Logical indicating whether to use the current working directory. Default is FALSE.
+#' @param prefer_sas [logical(1)] Logical indicating whether to prefer SAS7BDAT files over RDS. Default is FALSE.
+#' @param env_var [character(1)] The environment variable name for the base directory. Default is "RXD_DATA".
+#' @param print_file_paths [logical(1)] Logical indicating whether to print the directory path and file names.
+#' Default is FALSE.
+#'
+#' @return A named list of data frames, where each name corresponds to a loaded file.
+#'
 #' @examples
-#' \dontrun{
-#' test_data_path <- "../inst/extdata/"
-#' data_list <- load_data(
-#'   sub_dir = test_data_path,
-#'   file_names = "dummyads2",
-#'   use_wd = TRUE
-#' )
-#' }
-load_data <- function(sub_dir = NULL, file_names, use_wd = FALSE, prefer_sas = FALSE) {
-  if (is.null(file_names)) {
-    stop("Usage: load_data: file_names: Must supply at least one file name")
+#' # Get the current value of the RXD_DATA environment variable
+#' base_dir <- Sys.getenv("RXD_DATA")
+#'
+#' # Set the RXD_DATA environment variable to the path of the haven package
+#' Sys.setenv(RXD_DATA = find.package("haven"))
+#'
+#' data_list <- load_data(sub_dir = "examples", file_names = c("iris.sas7bdat"))
+#' str(data_list)
+#'
+#' # Reset the RXD_DATA environment variable to its original value
+#' Sys.setenv(RXD_DATA = base_dir)
+#'
+#' @export
+load_data <- function(
+    sub_dir = NULL,
+    file_names,
+    use_wd = FALSE,
+    prefer_sas = FALSE,
+    env_var = "RXD_DATA",
+    print_file_paths = FALSE) {
+  # Input validation
+  checkmate::assert_character(sub_dir, len = 1, null.ok = TRUE)
+  checkmate::assert_character(file_names, min.len = 1)
+  checkmate::assert_logical(use_wd, len = 1)
+  checkmate::assert_logical(prefer_sas, len = 1)
+  checkmate::assert_character(env_var, len = 1)
+
+  # Determine the base directory
+  if (use_wd) {
+    base_dir <- getwd()
+  } else {
+    base_dir <- get_base_dir(env_var = env_var)
   }
 
-  study_path <- "" # will be built using args
+  # Construct the full directory path
+  dir_path <- if (is.null(sub_dir)) base_dir else file.path(base_dir, sub_dir)
 
-  if (is.null(sub_dir)) {
-    study_path <- getwd()
-  } else {
-    if (use_wd) {
-      study_path <- file.path(getwd(), sub_dir)
-    } else {
-      study_path <- file.path(get_cre_path(), sub_dir)
-    }
+  # Determine the file extension based on preference
+  file_ext <- if (prefer_sas) "sas7bdat" else "rds"
+
+  # Get the full file paths
+  file_paths <- get_file_paths(dir_path = dir_path, file_names = file_names, prefer_sas = prefer_sas)
+
+  # Print the directory path and file names if requested
+  if (isTRUE(print_file_paths)) {
+    cat("Loading data from", dir_path, "\n")
+    cat("Loading data file(s):", basename(file_paths), "\n")
   }
 
-  # create the output
-  data_list <- create_data_list(study_path, file_names, prefer_sas) # nolint
+  # Load the data files
+  data_list <- load_data_files(file_paths)
+
+  names(data_list) <- file_names
 
   return(data_list)
 }