Any-Latin; Latin-ASCII in replace_non_ascii()

R/replace_non_ascii.R

@@ -35,7 +35,7 @@ replace_non_ascii <- function (x, replacement = '',
     remove.nonconverted = TRUE, ...) {
 
     x <- replace_curly_quote(x)
-    x <- stringi::stri_trans_general(x, "latin-ascii")
+    x <- stringi::stri_trans_general(x, "Any-Latin; Latin-ASCII")
     x <- iconv(as.character(x), "", "ASCII", "byte")
     Encoding(x) <- "latin1"
     x <- mgsub(x, ser, reps)

tests/testthat/test-replace_non_ascii.R

@@ -0,0 +1,23 @@
+context("Checking replace_non_ascii")
+
+test_that("replace_non_ascii transliterates Latin and non-Latin scripts to ASCII", {
+    x <- c("heiß", "brûlée", "Дорога", "キャンパス", "भोजन")
+    Encoding(x) <- "UTF-8"
+    expect_equal(replace_non_ascii(x), c("heiss", "brulee", "Doroga", "kyanpasu", "bhojana"))
+})
+
+test_that("replace_non_ascii with remove.nonconverted = FALSE preserves unmapped characters", {
+    x <- "hello"
+    expect_equal(replace_non_ascii(x, remove.nonconverted = FALSE), "hello")
+})
+
+test_that("replace_non_ascii2 replaces non-ASCII with regex", {
+    x <- "hello world"
+    expect_equal(replace_non_ascii2(x), "hello world")
+})
+
+test_that("replace_curly_quote replaces curly quotes", {
+    z <- '\x93Hello\x94'
+    Encoding(z) <- "latin1"
+    expect_equal(replace_curly_quote(z), '"Hello"')
+})