[regexp] Handle empty nested classes correctly

With the recent introduction of unicode sets (v-flag), nested character
classes are allowed in regular expressions.
We always expect a nested class to be of type
`RegExpClassSetExpression`, but the empty nested class was not handled
correctly.

Bug: v8:11935, chromium:1412942
Change-Id: I3b644c8627d8fc6b320a419216372810e8003983
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/4224311
Reviewed-by: Jakob Linke <jgruber@chromium.org>
Commit-Queue: Patrick Thier <pthier@chromium.org>
Cr-Commit-Position: refs/heads/main@{#85680}

src/regexp/regexp-ast.cc

@@ -200,11 +200,13 @@ void* RegExpUnparser::VisitClassSetOperand(RegExpClassSetOperand* that,
     if (i > 0) os_ << " ";
     VisitCharacterRange(that->ranges()->at(i));
   }
+  if (that->has_strings()) {
     for (auto iter : *that->strings()) {
       os_ << " '";
       os_ << std::string(iter.first.begin(), iter.first.end());
       os_ << "'";
     }
+  }
   os_ << "]";
   return nullptr;
 }
@@ -382,17 +384,18 @@ RegExpClassSetOperand::RegExpClassSetOperand(ZoneList<CharacterRange>* ranges,
                                              CharacterClassStrings* strings)
     : ranges_(ranges), strings_(strings) {
   DCHECK_NOT_NULL(ranges);
-  DCHECK_NOT_NULL(strings);
   min_match_ = 0;
   max_match_ = 0;
   if (!ranges->is_empty()) {
     min_match_ = 1;
     max_match_ = 2;
   }
+  if (has_strings()) {
     for (auto string : *strings) {
       min_match_ = std::min(min_match_, string.second->min_match());
       max_match_ = std::max(max_match_, string.second->max_match());
     }
+  }
 }
 
 RegExpClassSetExpression::RegExpClassSetExpression(
@@ -410,5 +413,20 @@ RegExpClassSetExpression::RegExpClassSetExpression(
   }
 }
 
+// static
+RegExpClassSetExpression* RegExpClassSetExpression::Empty(Zone* zone,
+                                                          bool is_negated) {
+  ZoneList<CharacterRange>* ranges =
+      zone->template New<ZoneList<CharacterRange>>(0, zone);
+  RegExpClassSetOperand* op =
+      zone->template New<RegExpClassSetOperand>(ranges, nullptr);
+  ZoneList<RegExpTree*>* operands =
+      zone->template New<ZoneList<RegExpTree*>>(1, zone);
+  operands->Add(op, zone);
+  return zone->template New<RegExpClassSetExpression>(
+      RegExpClassSetExpression::OperationType::kUnion, is_negated, false,
+      operands);
+}
+
 }  // namespace internal
 }  // namespace v8

src/regexp/regexp-ast.h

@@ -413,9 +413,12 @@ class RegExpClassSetOperand final : public RegExpTree {
   void Subtract(RegExpClassSetOperand* other,
                 ZoneList<CharacterRange>* temp_ranges, Zone* zone);
 
-  bool has_strings() const { return !strings_->empty(); }
+  bool has_strings() const { return strings_ != nullptr && !strings_->empty(); }
   ZoneList<CharacterRange>* ranges() { return ranges_; }
-  CharacterClassStrings* strings() { return strings_; }
+  CharacterClassStrings* strings() {
+    DCHECK_NOT_NULL(strings_);
+    return strings_;
+  }
 
  private:
   ZoneList<CharacterRange>* ranges_;
@@ -434,6 +437,10 @@ class RegExpClassSetExpression final : public RegExpTree {
 
   DECL_BOILERPLATE(ClassSetExpression);
 
+  // Create an empty class set expression (matches everything if |is_negated|,
+  // nothing otherwise).
+  static RegExpClassSetExpression* Empty(Zone* zone, bool is_negated);
+
   bool IsTextElement() override { return true; }
   int min_match() override { return 0; }
   int max_match() override { return max_match_; }

src/regexp/regexp-compiler-tonode.cc

@@ -593,7 +593,12 @@ RegExpNode* RegExpClassSetExpression::ToNode(RegExpCompiler* compiler,
 
 void RegExpClassSetOperand::Union(RegExpClassSetOperand* other, Zone* zone) {
   ranges()->AddAll(*other->ranges(), zone);
+  if (other->has_strings()) {
+    if (strings_ == nullptr) {
+      strings_ = zone->template New<CharacterClassStrings>(zone);
+    }
     strings()->insert(other->strings()->begin(), other->strings()->end());
+  }
 }
 
 void RegExpClassSetOperand::Intersect(RegExpClassSetOperand* other,
@@ -602,6 +607,10 @@ void RegExpClassSetOperand::Intersect(RegExpClassSetOperand* other,
   CharacterRange::Intersect(ranges(), other->ranges(), temp_ranges, zone);
   std::swap(*ranges(), *temp_ranges);
   temp_ranges->Rewind(0);
+  if (has_strings()) {
+    if (!other->has_strings()) {
+      strings()->clear();
+    } else {
       for (auto iter = strings()->begin(); iter != strings()->end();) {
         if (other->strings()->find(iter->first) == other->strings()->end()) {
           iter = strings()->erase(iter);
@@ -609,6 +618,8 @@ void RegExpClassSetOperand::Intersect(RegExpClassSetOperand* other,
           iter++;
         }
       }
+    }
+  }
 }
 
 void RegExpClassSetOperand::Subtract(RegExpClassSetOperand* other,
@@ -617,6 +628,7 @@ void RegExpClassSetOperand::Subtract(RegExpClassSetOperand* other,
   CharacterRange::Subtract(ranges(), other->ranges(), temp_ranges, zone);
   std::swap(*ranges(), *temp_ranges);
   temp_ranges->Rewind(0);
+  if (has_strings() && other->has_strings()) {
     for (auto iter = strings()->begin(); iter != strings()->end();) {
       if (other->strings()->find(iter->first) != other->strings()->end()) {
         iter = strings()->erase(iter);
@@ -624,6 +636,7 @@ void RegExpClassSetOperand::Subtract(RegExpClassSetOperand* other,
         iter++;
       }
     }
+  }
 }
 
 // static

src/regexp/regexp-parser.cc

@@ -2913,11 +2913,15 @@ RegExpTree* RegExpParserImpl<CharT>::ParseCharacterClass(
       zone()->template New<ZoneList<CharacterRange>>(2, zone());
   if (current() == ']') {
     Advance();
+    if (unicode_sets()) {
+      return RegExpClassSetExpression::Empty(zone(), is_negated);
+    } else {
       RegExpClassRanges::ClassRangesFlags class_ranges_flags;
       if (is_negated) class_ranges_flags = RegExpClassRanges::NEGATED;
       return zone()->template New<RegExpClassRanges>(zone(), ranges,
                                                      class_ranges_flags);
     }
+  }
 
   if (!unicode_sets()) {
     bool add_unicode_case_equivalents = IsUnicodeMode() && ignore_case();

test/mjsunit/harmony/regexp-unicode-sets.js

@@ -184,6 +184,13 @@ check(
     /[\q{ĀĂĄĆ|AaAc}--\q{āăąć}]/vi, ['AaAc', 'aAaC'], ['ĀĂĄĆ', 'āăąć'],
     false);
 
+// Empty nested classes.
+check(/[a-c\q{foo|bar}[]]/v, ['a','b','c','foo','bar'], [], false);
+check(/[[a-c\q{foo|bar}]&&[]]/v, [], ['a','b','c','foo','bar'], true);
+check(/[[a-c\q{foo|bar}]--[]]/v, ['a','b','c','foo','bar'], [], false);
+check(/[[]&&[a-c\q{foo|bar}]]/v, [], ['a','b','c','foo','bar'], true);
+check(/[[]--[a-c\q{foo|bar}]]/v, [], ['a','b','c','foo','bar'], true);
+
 // Empty string disjunctions matches nothing, but succeeds.
 let res = /[\q{}]/v.exec('foo');
 assertNotNull(res);