[regexp] Support unicode capture names in non-unicode patterns

src/regexp/regexp-parser.h

Index: src/regexp/regexp-parser.h
diff --git a/src/regexp/regexp-parser.h b/src/regexp/regexp-parser.h
index a3ef22d8b7ac8f83fe0dc777bfeb49010d780574..607e61a9379a6440acee6798b3d7fdf46e09ca5e 100644
--- a/src/regexp/regexp-parser.h
+++ b/src/regexp/regexp-parser.h
@@ -184,11 +184,18 @@ class RegExpParser BASE_EMBEDDED {
   // can be reparsed.
   bool ParseBackReferenceIndex(int* index_out);
 
+  // The default behavior is to combine surrogate pairs in unicode mode and
+  // don't combine them otherwise (a quantifier after a surrogate pair would
+  // then apply only to the trailing surrogate). Forcing combination is required
+  // when parsing capture names since they can always legally contain surrogate
+  // pairs.
+  enum class ScanMode { DEFAULT, FORCE_COMBINE_SURROGATE_PAIRS };
+
   bool ParseClassProperty(ZoneList<CharacterRange>* result);
   CharacterRange ParseClassAtom(uc16* char_class);
   RegExpTree* ReportError(Vector<const char> message);
-  void Advance();
-  void Advance(int dist);
+  void Advance(ScanMode mode = ScanMode::DEFAULT);
+  void Advance(int dist, ScanMode mode = ScanMode::DEFAULT);
   void Reset(int pos);
 
   // Reports whether the pattern might be used as a literal search string.
@@ -305,7 +312,7 @@ class RegExpParser BASE_EMBEDDED {
   bool has_next() { return next_pos_ < in()->length(); }
   uc32 Next();
   template <bool update_position>
-  uc32 ReadNext();
+  uc32 ReadNext(ScanMode mode);
   FlatStringReader* in() { return in_; }
   void ScanForCaptures();