Updates to address the review comments

R/cal_min_max_date.R

@@ -1,4 +1,4 @@
-#' Calculate minimum and maximum date and time in the dataframe
+#' Calculate minimum and maximum date and time in the data frame
 #'
 #' @description This function derives the earliest/latest ISO8601 datetime
 #'
@@ -18,24 +18,24 @@
 #' @examples
 #' EX <- tibble::tribble(
 #'   ~patient_number,    ~EX_ST_DT, ~EX_ST_TM,
-#'   "001",           "26-04-2022",   "10:20",
+#'   "001",           "25-04-2022",   "10:20",
 #'   "001",           "25-04-2022",   "10:15",
 #'   "001",           "25-04-2022",   "10:19",
 #'   "002",           "26-05-2022", "UNK:UNK",
 #'   "002",           "26-05-2022",   "05:59"
 #' )
 #'
 #' min <- cal_min_max_date(EX,
-#'   "EX_ST_DT",
-#'   "EX_ST_TM",
+#'   date_variable = "EX_ST_DT",
+#'   time_variable = "EX_ST_TM",
 #'   val_type = "min",
 #'   date_format = "dd-mmm-yyyy",
 #'   time_format = "H:M"
 #' )
 #'
 #' max <- cal_min_max_date(EX,
-#'   "EX_ST_DT",
-#'   "EX_ST_TM",
+#'   date_variable = "EX_ST_DT",
+#'   time_variable = "EX_ST_TM",
 #'   val_type = "max",
 #'   date_format = "dd-mmm-yyyy",
 #'   time_format = "H:M"
@@ -47,13 +47,15 @@ cal_min_max_date <- function(raw_dataset,
                              val_type = "min",
                              date_format,
                              time_format) {
-  # Check if date is present in the raw data frame
-  date_not_in_data <- !(date_variable %in% colnames(raw_dataset))
+  # Check if date parameter is missing or date variable is present in the raw data frame
+  date_not_in_data <- is.na(date_variable) ||
+    !utils::hasName(raw_dataset, date_variable)
 
   # Check if time variable is used and present in the raw data frame
-  time_not_in_data <- !(time_variable %in% colnames(raw_dataset)) && !is.na(time_variable)
+  time_not_in_data <- !is.na(time_variable) &&
+    !utils::hasName(raw_dataset, time_variable)
 
-  # If date/time variables not present return the empty data frame
+  # If date/time variables not present return empty data frame
   if (date_not_in_data || time_not_in_data) {
     # Return Empty data frame with patient_number and datetime columns
     empty_df <- stats::setNames(
@@ -68,13 +70,14 @@ cal_min_max_date <- function(raw_dataset,
   }
 
   fin_df <- raw_dataset
-  # Time variable is not used then use only date
+
+  # Time is not used in reference date then use only date
   if (is.na(time_variable)) {
     fin_df$datetime <- create_iso8601(raw_dataset[[date_variable]],
       .format = date_format
     )
   } else {
-    # If both date and time variables are presen use both date and time
+    # If both date and time variables are present use both date and time
     raw_dataset$date_time <- paste0(
       raw_dataset[[date_variable]],
       raw_dataset[[time_variable]]

R/oak_cal_ref_dates.R

@@ -11,7 +11,7 @@
 #'        Default set to Minimum. Values should be min or max.
 #' @param ref_date_config_df Data frame which has the details of the variables to
 #'        be used for the calculation of reference dates.
-#'        Should has columns listed below:
+#'        Should have columns listed below:
 #'          dataset_name : Name of the raw dataset.
 #'          date_var : Date variable name from the raw dataset.
 #'          time_var : Time variable name from the raw dataset.
@@ -59,7 +59,7 @@
 #'
 #' dm_df <- oak_cal_ref_dates(dm,
 #'   der_var = "RFSTDTC",
-#'   min_max = "max",
+#'   min_max = "min",
 #'   ref_date_config_df = ref_date_config_df,
 #'   raw_source
 #' )
@@ -76,6 +76,8 @@ oak_cal_ref_dates <- function(ds_in = dm,
     tformat, sdtm_var_name
   ))
 
+  admiraldev::assert_list_of(raw_source, "data.frame")
+
   ds_out <- data.frame()
   for (i in seq_along(ref_date_config_df$dataset_name)) {
     raw_dataset_name <- ref_date_config_df$dataset_name[i]
@@ -86,7 +88,7 @@ oak_cal_ref_dates <- function(ds_in = dm,
     sdtm_var <- ref_date_config_df$sdtm_var_name[i]
     raw_dataset <- raw_source[[raw_dataset_name]]
 
-    if (der_var == sdtm_var) {
+    if (der_var == sdtm_var && !is.null(raw_dataset)) {
       ds_out1 <- cal_min_max_date(
         raw_dataset = raw_dataset,
         date_variable = date_variable,
@@ -96,6 +98,11 @@ oak_cal_ref_dates <- function(ds_in = dm,
         val_type = min_max
       )
       ds_out <- rbind(ds_out, ds_out1)
+    } else if (der_var == sdtm_var && is.null(raw_dataset)) {
+      warning(paste0(
+        raw_dataset_name,
+        " is not present in the source data list but referenced in ref_date_config_df"
+      ))
     }
   }
 

tests/testthat/test-cal_min_max_date.R

@@ -1,3 +1,29 @@
+test_that("Warn if date variable parameter is NULL", {
+  EX <- tibble::tribble(
+    ~patient_number,    ~EX_ST_DT,
+    "001",           "26-04-2022"
+  )
+
+  warning_msg <- "Date variable NA or Time variable NA not present in source data"
+  expect_warning(
+    observed <- cal_min_max_date(EX,
+      date_variable = NA,
+      time_variable = NA,
+      val_type = "max",
+      date_format = "dd-mmm-yyyy",
+      time_format = "H:M"
+    ),
+    regexp = warning_msg
+  )
+
+  expected <- stats::setNames(
+    data.frame(matrix(ncol = 2L, nrow = 0L)),
+    c("patient_number", "datetime")
+  )
+
+  expect_identical(observed, expected)
+})
+
 test_that("cal_min_max_date works as expected", {
   EX <- tibble::tribble(
     ~patient_number,    ~EX_ST_DT, ~EX_ST_TM,
@@ -22,8 +48,8 @@ test_that("cal_min_max_date works as expected", {
   )
 
   observed_min <- cal_min_max_date(EX,
-    "EX_ST_DT",
-    "EX_ST_TM",
+    date_variable = "EX_ST_DT",
+    time_variable = "EX_ST_TM",
     val_type = "min",
     date_format = "dd-mmm-yyyy",
     time_format = "H:M"
@@ -32,8 +58,8 @@ test_that("cal_min_max_date works as expected", {
   expect_identical(observed_min, expected_min)
 
   observed_max <- cal_min_max_date(EX,
-    "EX_ST_DT",
-    "EX_ST_TM",
+    date_variable = "EX_ST_DT",
+    time_variable = "EX_ST_TM",
     val_type = "max",
     date_format = "dd-mmm-yyyy",
     time_format = "H:M"
@@ -51,8 +77,8 @@ test_that("Warning is displayed if date or time variables parameters passed are
   warning_msg <- "Date variable EX_ST_DT or Time variable EX_ST_TM not present in source data"
   expect_warning(
     observed <- cal_min_max_date(EX,
-      "EX_ST_DT",
-      "EX_ST_TM",
+      date_variable = "EX_ST_DT",
+      time_variable = "EX_ST_TM",
       val_type = "max",
       date_format = "dd-mmm-yyyy",
       time_format = "H:M"