Fix unicode unescaping for e.g. \\U0001F575

R/regex_gadget.R

@@ -581,7 +581,7 @@ regex_gadget <- function(text = NULL,
 
 sanitize_text_input <- function(x) {
   if (is.null(x) || !nchar(x)) return(x)
-  rx_unicode <- "\\u[0-9a-f]{4,8}"
+  rx_unicode <- "\\\\u[0-9a-f]{4,8}"
   rx_hex <- "\\\\x[0-9a-f]{2}|\\\\x\\{[0-9a-f]{1,6}\\}"
   rx_octal <- "\\\\[0][0-7]{1,3}"
   rx_escape <- paste(rx_unicode, rx_hex, rx_octal, sep = "|")

tests/testthat/test-sanitize_text_input.R

@@ -8,6 +8,7 @@ test_that("sanitizes unicode", {
   # rx_unicode <- "\\u[0-9a-f]{4,8}"
   expect_equal(sanitize_text_input("\\u2019"), "\u2019")
   expect_equal(sanitize_text_input("\\u000D"), "\r")
+  expect_equal(sanitize_text_input("\\U0001F575"), "\U001F575")
 })
 
 test_that("sanitizes hex", {