[regexp] Shortcut case-folding of entire non-bmp range

When the range of all non-bmp characters is passed to
AddUnicodeCaseEquivalents, icu::UnicodeSet::closeOver dutifully tries to
case-fold every single character in that range. Since we already know
this to be a nop, we can simply return instead.

This improves compilation time of /ui regexps by around 100x.

Bug: v8:6727
Change-Id: I79d73c77d6a54cbb5ad2cad0355214ed712b59b9
Reviewed-on: https://chromium-review.googlesource.com/635303
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Cr-Commit-Position: refs/heads/master@{#47636}

src/regexp/jsregexp.cc

@@ -5060,6 +5060,16 @@ RegExpNode* UnanchoredAdvance(RegExpCompiler* compiler,
 
 void AddUnicodeCaseEquivalents(ZoneList<CharacterRange>* ranges, Zone* zone) {
 #ifdef V8_INTL_SUPPORT
+  DCHECK(CharacterRange::IsCanonical(ranges));
+
+  // Micro-optimization to avoid passing large ranges to UnicodeSet::closeOver.
+  // See also https://crbug.com/v8/6727.
+  // TODO(jgruber): This only covers the special case of the {0,0x10FFFF} range,
+  // which we use frequently internally. But large ranges can also easily be
+  // created by the user. We might want to have a more general caching mechanism
+  // for such ranges.
+  if (ranges->length() == 1 && ranges->at(0).IsEverything(kNonBmpEnd)) return;
+
   // Use ICU to compute the case fold closure over the ranges.
   icu::UnicodeSet set;
   for (int i = 0; i < ranges->length(); i++) {

src/regexp/regexp-ast.h

@@ -110,7 +110,7 @@ class CharacterRange {
   uc32 to() const { return to_; }
   void set_to(uc32 value) { to_ = value; }
   bool is_valid() { return from_ <= to_; }
-  bool IsEverything(uc16 max) { return from_ == 0 && to_ >= max; }
+  bool IsEverything(uc32 max) { return from_ == 0 && to_ >= max; }
   bool IsSingleton() { return (from_ == to_); }
   static void AddCaseEquivalents(Isolate* isolate, Zone* zone,
                                  ZoneList<CharacterRange>* ranges,