Use finalized version of string translation helpers (#7744)

* Use finalized version of string translation helpers

* Use exact version of vctrs utilities

Author: DavisVaughan

src/mask.cpp

@@ -2,12 +2,28 @@
 #include "utils.h"
 
 R_xlen_t find_first(SEXP haystack, SEXP needle) {
-  SEXP needle_utf8 = PROTECT(str_as_utf8(needle));
-  R_xlen_t n = XLENGTH(haystack);
+  if (!string_is_ascii_or_utf8(needle)) {
+    needle = string_as_utf8(needle);
+  }
+  PROTECT(needle);
+
+  const R_xlen_t n = XLENGTH(haystack);
   R_xlen_t i_name = 0;
-  for (; i_name < n; i_name++) {
-    if (needle_utf8 == str_as_utf8(STRING_ELT(haystack, i_name))) break;
+
+  for (; i_name < n; ++i_name) {
+    SEXP haystack_elt = STRING_ELT(haystack, i_name);
+
+    if (!string_is_ascii_or_utf8(haystack_elt)) {
+      // No need to `PROTECT()`, we do a pointer comparison
+      // and then throw it away
+      haystack_elt = string_as_utf8(haystack_elt);
+    }
+
+    if (needle == haystack_elt) {
+      break;
+    }
   }
+
   UNPROTECT(1);
   return i_name;
 }

src/utils.h

@@ -6,25 +6,21 @@
 #include <Rinternals.h>
 #include <Rversion.h>
 
-static inline
-bool str_is_utf8(SEXP x) {
+// String encoding normalization
+// From https://github.com/r-lib/vctrs/pull/2085
+static inline bool string_is_ascii_or_utf8(SEXP x) {
 #if (R_VERSION >= R_Version(4, 5, 0))
-  return Rf_charIsUTF8(x);
+  return Rf_charIsASCII(x) || (Rf_getCharCE(x) == CE_UTF8) || (x == NA_STRING);
 #else
   const int mask_ascii = 8;
   const int mask_utf8 = 64;
   const int levels = LEVELS(x);
-  return (levels & mask_ascii) || (levels & mask_utf8);
+  return (levels & mask_ascii) || (levels & mask_utf8) || (x == NA_STRING);
 #endif
 }
 
-static inline
-SEXP str_as_utf8(SEXP x) {
-  if (str_is_utf8(x)) {
-    return x;
-  } else {
-    return Rf_mkCharCE(Rf_translateCharUTF8(x), CE_UTF8);
-  }
+static inline SEXP string_as_utf8(SEXP x) {
+  return Rf_mkCharCE(Rf_translateCharUTF8(x), CE_UTF8);
 }
 
 #endif