Add support for displaying the label attribute for column name.

NEWS.md

@@ -8,6 +8,8 @@ These are all minor breaking changes resulting from enhancements and are not exp
 
 ## New features
 
+* `tabyl()` now defaults to displaying the label attribute for the column name (@olivroy, #394).
+
 * A new function `paste_skip_na()` pastes without including NA values (#537).
 
 * `row_to_names()` now accepts multiple rows as input, and merges them using a new `sep` argument (#536). The default is `sep = "_"`. When handling multiple `NA` values, `row_to_names()` ignores them and only merges non-NA values for column names. When all values are `NA`, `row_to_names()` creates a column name of `"NA"`, a character, rather than `NA`.

R/tabyl.R

@@ -65,8 +65,7 @@ tabyl.default <- function(dat, show_na = TRUE, show_missing_levels = TRUE, ...)
   } else {
     var_name <- names(dat)
   }
-
-
+
   # useful error message if input vector doesn't exist
   if (is.null(dat)) {
     stop(paste0("object ", var_name, " not found"))
@@ -75,6 +74,13 @@ tabyl.default <- function(dat, show_na = TRUE, show_missing_levels = TRUE, ...)
   if (length(var_name) > 1) {
     var_name <- paste(var_name, collapse = "")
   }
+
+  # Try to retrieve label
+  if (is.data.frame(dat)) {
+    var_label <- attr(dat[, var_name], "label", exact = TRUE) %||% var_name
+  } else {
+    var_label <-  attr(dat, "label", exact = TRUE) %||% var_name
+  }
 
   # if show_na is not length-1 logical, error helpfully (#377)
   if (length(show_na) > 1 || !inherits(show_na, "logical")) {
@@ -133,8 +139,8 @@ tabyl.default <- function(dat, show_na = TRUE, show_missing_levels = TRUE, ...)
       dplyr::mutate(percent = n / sum(n, na.rm = TRUE)) # recalculate % without NAs
   }
 
-  # reassign correct variable name
-  names(result)[1] <- var_name
+  # reassign correct variable name (or label if it exists)
+  names(result)[1] <- var_label
 
   # in case input var name was "n" or "percent", call helper function to set unique names
   result <- handle_if_special_names_used(result)
@@ -169,6 +175,10 @@ tabyl.data.frame <- function(dat, var1, var2, var3, show_na = TRUE, show_missing
   }
 }
 
+get_label <- function(x) {
+  attr(x, "label", exact = TRUE)
+}
+
 # a one-way frequency table; this was called "tabyl" in janitor <= 0.3.0
 tabyl_1way <- function(dat, var1, show_na = TRUE, show_missing_levels = TRUE) {
   x <- dplyr::select(dat, !!var1)

tests/testthat/test-tabyl.R

@@ -395,6 +395,21 @@ test_that("3-way tabyl with 3rd var factor is listed in right order, #250", {
   expect_equal(names(tabyl(z, am, gear, cyl)), c("8", "6", "NA_"))
 })
 
+test_that("tabyl works with label attributes (#394)", {
+  mt_label <- mtcars
+  attr(mt_label$cyl, "label") <- "Number of cyl"
+  tab <- tabyl(mt_label, cyl)
+  expect_named(
+    tab,
+    c("Number of cyl", "n", "percent")
+  )
+  tab2 <-  tabyl(mt_label, cyl, am)
+  expect_named(
+    tab2,
+    c("Number of cyl", "0", "1")
+  )
+})
+
 test_that("tabyl works with ordered 1st variable, #386", {
   mt_ordered <- mtcars
   mt_ordered$cyl <- ordered(mt_ordered$cyl, levels = c("4", "8", "6"))