Merge branch 'f-#2-to-utf8'. Fixes #2. Fixes #5.

- New `to_utf8()` performs deep conversion to UTF-8, including names and other attributes (#2, #5).

R/to-utf8.R

@@ -0,0 +1,50 @@
+#' Deep conversion to UTF-8
+#'
+#' Converts all characters directly or indirectly contained in an object to
+#' UTF-8.
+to_utf8 <- function(x, ...) UseMethod("to_utf8", x)
+
+to_utf8.utf8 <- function(x, ...) {
+  attrib_to_utf8(x)
+}
+
+to_utf8.list <- function(x, ...) {
+  x[] <- lapply(x, to_utf8)
+  names(x) <- to_utf8(names(x))
+  attrib_to_utf8(x)
+}
+
+to_utf8.data.frame <- to_utf8.list
+
+to_utf8.character <- function(x, ..., use_class = TRUE) {
+  if (use_class)
+    x <- as.utf8(x)
+  else
+    x <- enc2utf8(x)
+  attrib_to_utf8(x)
+}
+
+to_utf8.default <- function(x, ...) {
+  attrib_to_utf8(x)
+}
+
+to_utf8.NULL <- function(x, ...) {
+  NULL
+}
+
+attrib_to_utf8 <- function(x) {
+  mostattributes(x) <- named_to_utf8_except_class(attributes(x))
+  x
+}
+
+named_to_utf8_except_class <- function(attrib) {
+  is_class <- which(names(attrib) == "class")
+  if (length(is_class) > 0) {
+    attrib[-is_class] <- to_utf8(unname(attrib)[-is_class])
+    attrib[[is_class]] <- to_utf8(unname(attrib)[[is_class]], use_class = FALSE)
+  } else {
+    attrib <- to_utf8(unname(attrib))
+  }
+
+  attrib
+}

src/encoding.c

@@ -1,5 +1,4 @@
 #include <sys/time.h>
-#include <sys/resource.h>
 
 #define USE_RINTERNALS
 #include <R.h>

tests/testthat/test-to-utf8.R

@@ -0,0 +1,31 @@
+context("to-utf8")
+
+test_that("character vectors", {
+  expect_is(to_utf8(letters), "utf8")
+  expect_false(inherits(class(to_utf8(letters)), "utf8"))
+})
+
+test_that("iris", {
+  iris_utf8 <- to_utf8(iris)
+  expect_is(colnames(iris_utf8), "utf8")
+  expect_true(all_utf8(names(attributes(iris_utf8))))
+  expect_is(levels(iris_utf8$Species), "utf8")
+})
+
+test_that("mtcars", {
+  mtcars_utf8 <- to_utf8(mtcars)
+  expect_is(colnames(mtcars_utf8), "utf8")
+  expect_true(all_utf8(names(attributes(mtcars_utf8))))
+  expect_true(all_utf8(rownames(mtcars_utf8)))
+})
+
+test_that("umlauts", {
+  data <- data.frame(a = I(c("o", "u")))
+  colnames(data) <- enc2native("\u00e4")
+  data[[1]] <- enc2native(c("\u00f6", "\u00fc"))
+
+  data_utf8 <- to_utf8(data)
+  expect_is(colnames(data_utf8), "utf8")
+  expect_true(all_utf8(names(attributes(data_utf8))))
+  expect_true(all_utf8(rownames(data_utf8)))
+})