feat: finalize_integers()

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: datawrap
 Title: Final Steps for Dataset Preparation
-Version: 0.0.0.9000
+Version: 0.0.0.9001
 Authors@R: 
     person("Jon", "Harmon", , "jonthegeek@gmail.com", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0003-4781-4346"))
@@ -10,17 +10,22 @@ License: MIT + file LICENSE
 URL: https://wranglezone.github.io/datawrap/,
     https://github.com/wranglezone/datawrap
 BugReports: https://github.com/wranglezone/datawrap/issues
+Depends:
+    R (>= 4.1)
 Imports:
     knitr,
     purrr,
     rlang,
+    stbl,
     stringr,
     tibble,
     vctrs
 Suggests:
     testthat (>= 3.0.0),
     usethis,
     withr
+Remotes:
+    stbl=wranglezone/stbl
 Config/testthat/edition: 3
 Encoding: UTF-8
 Language: en-US

NAMESPACE

@@ -2,4 +2,5 @@
 
 export(create_dataset_dictionary)
 export(describe_dataset)
+export(finalize_integers)
 export(write_dataset_dictionary)

NEWS.md

@@ -0,0 +1,7 @@
+# datawrap (development version)
+
+* `finalize_integers()` converts integerish columns to integer for efficient storage (@copilot & @jonthegeek, #10).
+
+# datawrap 0.0.0.9000
+
+* Initial release.

R/finalize_integers.R

@@ -0,0 +1,33 @@
+#' Downcast integerish columns to integer
+#'
+#' Iterates over all columns (or list elements) in `dataset` and converts any
+#' non-logical column whose values can all be represented as integers without
+#' losing any information to integers. Columns that contain non-integerish data
+#' are left unchanged.
+#'
+#' @param dataset (`data.frame`, `list`, or `NULL`) The dataset to process.
+#'
+#' @returns The `dataset` with all integerish columns converted to integer (or
+#'   `NULL` if `dataset` is `NULL`).
+#' @export
+#'
+#' @examples
+#' df <- data.frame(x = c(1.0, 2.0, 3.0), y = c(1.1, 2.2, 3.3))
+#' finalize_integers(df)
+finalize_integers <- function(dataset) {
+  if (!is.null(dataset) && !is.list(dataset)) {
+    stbl::pkg_abort(
+      "datawrap",
+      "{.arg dataset} must be a {.cls data.frame}, {.cls list}, or {.cls NULL}.",
+      c("invalid_dataset", "invalid_argument")
+    )
+  }
+  # NOTE: This function alone does not warrant adding dplyr to imports, but it
+  # could be refactored to use dplyr::mutate, dplyr::across, and dplyr::where
+  # if dplyr is added as a dependency elsewhere in the package.
+  int_ish_cols <- purrr::map_lgl(dataset, \(x) {
+    !is.logical(x) && stbl::is_int_ish(x)
+  })
+  dataset[int_ish_cols] <- purrr::map(dataset[int_ish_cols], stbl::to_int)
+  dataset
+}

tests/testthat/test-finalize_integers.R

@@ -0,0 +1,68 @@
+test_that("finalize_integers() converts integerish double columns to integer (#10)", {
+  df <- data.frame(x = c(1.0, 2.0, 3.0), y = c(1.1, 2.2, 3.3))
+  result <- finalize_integers(df)
+  expect_type(result$x, "integer")
+  expect_type(result$y, "double")
+})
+
+test_that("finalize_integers() converts integer-valued double elements in a list to integer (#10)", {
+  lst <- list(x = c(1.0, 2.0, 3.0), y = c(1.1, 2.2, 3.3))
+  result <- finalize_integers(lst)
+  expect_type(result$x, "integer")
+  expect_type(result$y, "double")
+})
+
+test_that("finalize_integers() converts integerish character columns to integer (#10)", {
+  df <- data.frame(x = as.character(1:3), y = as.character(c(1.1, 2.2, 3.3)))
+  result <- finalize_integers(df)
+  expect_type(result$x, "integer")
+  expect_type(result$y, "character")
+})
+
+test_that("finalize_integers() leaves non-integerish columns unchanged (#10)", {
+  df <- data.frame(
+    x = c(1.0, 2.0),
+    y = letters[1:2],
+    z = TRUE,
+    stringsAsFactors = FALSE
+  )
+  result <- finalize_integers(df)
+  expect_type(result$y, "character")
+  expect_type(result$z, "logical")
+})
+
+test_that("finalize_integers() preserves NA values in converted columns (#10)", {
+  df <- data.frame(x = c(1.0, NA, 3.0))
+  result <- finalize_integers(df)
+  expect_type(result$x, "integer")
+  expect_true(is.na(result$x[[2]]))
+})
+
+test_that("finalize_integers() preserves the class of the input (#10)", {
+  tbl <- tibble::tibble(x = c(1.0, 2.0))
+  result <- finalize_integers(tbl)
+  expect_identical(class(result), class(tbl))
+})
+
+test_that("finalize_integers() errors if dataset is not a data.frame, list, or NULL (#10)", {
+  stbl::expect_pkg_error_classes(
+    finalize_integers("not a dataset"),
+    "datawrap",
+    "invalid_dataset",
+    "invalid_argument"
+  )
+})
+
+test_that("finalize_integers() handles a dataset with no double columns (#10)", {
+  df <- data.frame(x = 1L, y = "a", stringsAsFactors = FALSE)
+  expect_identical(finalize_integers(df), df)
+})
+
+test_that("finalize_integers() handles an empty data frame (#10)", {
+  df <- data.frame()
+  expect_identical(finalize_integers(df), df)
+})
+
+test_that("finalize_integers() handles NULL (#10)", {
+  expect_null(finalize_integers(NULL))
+})