[regexp] implement character classes for unicode regexps.

src/regexp/regexp-ast.h

 // Copyright 2016 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #ifndef V8_REGEXP_REGEXP_AST_H_
 #define V8_REGEXP_REGEXP_AST_H_
 
+#include "src/objects.h"
 #include "src/utils.h"
 #include "src/zone.h"
 
 namespace v8 {
 namespace internal {
 
 #define FOR_EACH_REG_EXP_TREE_TYPE(VISIT) \
   VISIT(Disjunction)                      \
   VISIT(Alternative)                      \
   VISIT(Assertion)                        \
@@ skipping 52 matching lines @@
 };
 
 
 // Represents code units in the range from from_ to to_, both ends are
 // inclusive.
 class CharacterRange {
  public:
   CharacterRange() : from_(0), to_(0) {}
   // For compatibility with the CHECK_OK macro
   CharacterRange(void* null) { DCHECK_NULL(null); }  // NOLINT
-  CharacterRange(uc16 from, uc16 to) : from_(from), to_(to) {}
+  CharacterRange(uc32 from, uc32 to) : from_(from), to_(to) {}
   static void AddClassEscape(uc16 type, ZoneList<CharacterRange>* ranges,
                              Zone* zone);
   static Vector<const int> GetWordBounds();
-  static inline CharacterRange Singleton(uc16 value) {
+  static inline CharacterRange Singleton(uc32 value) {
     return CharacterRange(value, value);
   }
-  static inline CharacterRange Range(uc16 from, uc16 to) {
-    DCHECK(from <= to);
+  static inline CharacterRange Range(uc32 from, uc32 to) {
+    DCHECK(0 <= from && to <= String::kMaxCodePoint);
+    DCHECK(static_cast<uint32_t>(from) <= static_cast<uint32_t>(to));
     return CharacterRange(from, to);
   }
   static inline CharacterRange Everything() {
-    return CharacterRange(0, 0xFFFF);
+    return CharacterRange(0, String::kMaxCodePoint);
   }
-  bool Contains(uc16 i) { return from_ <= i && i <= to_; }
-  uc16 from() const { return from_; }
-  void set_from(uc16 value) { from_ = value; }
-  uc16 to() const { return to_; }
-  void set_to(uc16 value) { to_ = value; }
+  static inline ZoneList<CharacterRange>* List(Zone* zone,
+                                               CharacterRange range) {
+    ZoneList<CharacterRange>* list =
+        new (zone) ZoneList<CharacterRange>(1, zone);
+    list->Add(range, zone);
+    return list;
+  }
+  bool Contains(uc32 i) { return from_ <= i && i <= to_; }
+  uc32 from() const { return from_; }
+  void set_from(uc32 value) { from_ = value; }
+  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 IsSingleton() { return (from_ == to_); }
   void AddCaseEquivalents(Isolate* isolate, Zone* zone,
                           ZoneList<CharacterRange>* ranges, bool is_one_byte);
-  static void Split(ZoneList<CharacterRange>* base, Vector<const int> overlay,
-                    ZoneList<CharacterRange>** included,
-                    ZoneList<CharacterRange>** excluded, Zone* zone);
   // Whether a range list is in canonical form: Ranges ordered by from value,
   // and ranges non-overlapping and non-adjacent.
   static bool IsCanonical(ZoneList<CharacterRange>* ranges);
   // Convert range list to canonical form. The characters covered by the ranges
   // will still be the same, but no character is in more than one range, and
   // adjacent ranges are merged. The resulting list may be shorter than the
   // original, but cannot be longer.
   static void Canonicalize(ZoneList<CharacterRange>* ranges);
   // Negate the contents of a character range in canonical form.
   static void Negate(ZoneList<CharacterRange>* src,
                      ZoneList<CharacterRange>* dst, Zone* zone);
   static const int kStartMarker = (1 << 24);
   static const int kPayloadMask = (1 << 24) - 1;
 
  private:
-  uc16 from_;
-  uc16 to_;
+  uc32 from_;
+  uc32 to_;
 };
 
 
 class CharacterSet final BASE_EMBEDDED {
  public:
   explicit CharacterSet(uc16 standard_set_type)
       : ranges_(NULL), standard_set_type_(standard_set_type) {}
   explicit CharacterSet(ZoneList<CharacterRange>* ranges)
       : ranges_(ranges), standard_set_type_(0) {}
   ZoneList<CharacterRange>* ranges(Zone* zone);
@@ skipping 146 matching lines @@
 class RegExpCharacterClass final : public RegExpTree {
  public:
   RegExpCharacterClass(ZoneList<CharacterRange>* ranges, bool is_negated)
       : set_(ranges), is_negated_(is_negated) {}
   explicit RegExpCharacterClass(uc16 type) : set_(type), is_negated_(false) {}
   void* Accept(RegExpVisitor* visitor, void* data) override;
   RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
   RegExpCharacterClass* AsCharacterClass() override;
   bool IsCharacterClass() override;
   bool IsTextElement() override { return true; }
+  bool NeedsDesugaringForUnicode(Zone* zone);
   int min_match() override { return 1; }
   int max_match() override { return 1; }
   void AppendToText(RegExpText* text, Zone* zone) override;
   CharacterSet character_set() { return set_; }
   // TODO(lrn): Remove need for complex version if is_standard that
   // recognizes a mangled standard set and just do { return set_.is_special(); }
   bool is_standard(Zone* zone);
   // Returns a value representing the standard character set if is_standard()
   // returns true.
   // Currently used values are:
@@ skipping 144 matching lines @@
   bool IsLookaround() override;
   bool IsAnchoredAtStart() override;
   int min_match() override { return 0; }
   int max_match() override { return 0; }
   RegExpTree* body() { return body_; }
   bool is_positive() { return is_positive_; }
   int capture_count() { return capture_count_; }
   int capture_from() { return capture_from_; }
   Type type() { return type_; }
 
+  class Builder {
+   public:
+    Builder(bool is_positive, RegExpNode* on_success,
+            int stack_pointer_register, int position_register,
+            int capture_register_count = 0, int capture_register_start = 0);
+    RegExpNode* on_match_success() { return on_match_success_; }
+    RegExpNode* ForMatch(RegExpNode* match);
+
+   private:
+    bool is_positive_;
+    RegExpNode* on_match_success_;
+    RegExpNode* on_success_;
+    int stack_pointer_register_;
+    int position_register_;
+  };
+
  private:
   RegExpTree* body_;
   bool is_positive_;
   int capture_count_;
   int capture_from_;
   Type type_;
 };
 
 
 class RegExpBackReference final : public RegExpTree {
@@ skipping 23 matching lines @@
   RegExpEmpty* AsEmpty() override;
   bool IsEmpty() override;
   int min_match() override { return 0; }
   int max_match() override { return 0; }
 };
 
 }  // namespace internal
 }  // namespace v8
 
 #endif  // V8_REGEXP_REGEXP_AST_H_