* Fixed bug in handling of quantified look-aheads.

src/parser.cc

 // Copyright 2006-2008 the V8 project authors. All rights reserved.
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions are
 // met:
 //
 //     * Redistributions of source code must retain the above copyright
 //       notice, this list of conditions and the following disclaimer.
 //     * Redistributions in binary form must reproduce the above
 //       copyright notice, this list of conditions and the following
 //       disclaimer in the documentation and/or other materials provided
@@ skipping 509 matching lines @@
   // it fails it will push back the characters read so the same characters
   // can be reparsed.
   bool ParseBackReferenceIndex(int* index_out);
 
   CharacterRange ParseClassAtom(uc16* char_class);
   RegExpTree* ReportError(Vector<const char> message);
   void Advance();
   void Advance(int dist);
   void Reset(int pos);
 
-  bool HasCharacterEscapes();
+  bool IsNonSimple();

[Christian Plesner Hansen, 2008/12/12 09:52:01]

You should generally avoid negative method names.  How about IsSimple()?

 
   int captures_started() { return captures_ == NULL ? 0 : captures_->length(); }
   int position() { return next_pos_ - 1; }
   bool failed() { return failed_; }
 
   static const uc32 kEndMarker = (1 << 21);
  private:
   uc32 current() { return current_; }
   bool has_more() { return has_more_; }
   bool has_next() { return next_pos_ < in()->length(); }
   uc32 Next();
   FlatStringReader* in() { return in_; }
   void ScanForCaptures();
   bool CaptureAvailable(int index);
   uc32 current_;
   bool has_more_;
   bool multiline_;
   int next_pos_;
   FlatStringReader* in_;
   Handle<String>* error_;
-  bool has_character_escapes_;
+  bool non_simple_;

[Christian Plesner Hansen, 2008/12/12 09:52:01]

Ditto.

   ZoneList<RegExpCapture*>* captures_;
   bool is_scanned_for_captures_;
   // The capture count is only valid after we have scanned for captures.
   int capture_count_;
   bool failed_;
 };
 
 
 // A temporary scope stores information during parsing, just like
 // a plain scope.  However, temporary scopes are not kept around
@@ skipping 2933 matching lines @@
 
 RegExpParser::RegExpParser(FlatStringReader* in,
                            Handle<String>* error,
                            bool multiline)
   : current_(kEndMarker),
     has_more_(true),
     multiline_(multiline),
     next_pos_(0),
     in_(in),
     error_(error),
-    has_character_escapes_(false),
+    non_simple_(false),
     captures_(NULL),
     is_scanned_for_captures_(false),
     capture_count_(0),
     failed_(false) {
   Advance(1);
 }
 
 
 uc32 RegExpParser::Next() {
   if (has_next()) {
@@ skipping 27 matching lines @@
   Advance();
 }
 
 
 void RegExpParser::Advance(int dist) {
   for (int i = 0; i < dist; i++)
     Advance();
 }
 
 
-// Reports whether the parsed string atoms contain any characters that were
-// escaped in the original pattern. If not, all atoms are proper substrings
-// of the original pattern.
-bool RegExpParser::HasCharacterEscapes() {
-  return has_character_escapes_;
+// Reports whether the pattern might be used as a literal search string.
+// Only use if the result of the parse is a single atom node.
+bool RegExpParser::IsNonSimple() {

[Christian Plesner Hansen, 2008/12/12 09:52:01]

This should be a simple accessor, not a full camel case method.

+  return non_simple_;
 }
 
 RegExpTree* RegExpParser::ReportError(Vector<const char> message) {
   failed_ = true;
   *error_ = Factory::NewStringFromAscii(message, NOT_TENURED);
   // Zip to the end to make sure the no more input is read.
   current_ = kEndMarker;
   next_pos_ = in()->length();
   return NULL;
 }
@@ skipping 194 matching lines @@
           builder.AddCharacter('u');
         }
         break;
       }
       default:
         // Identity escape.
         builder.AddCharacter(Next());
         Advance(2);
         break;
       }
-      has_character_escapes_ = true;
+      non_simple_ = true;
       break;
     case '{': {
       int dummy;
       if (ParseIntervalQuantifier(&dummy, &dummy)) {
         ReportError(CStrVector("Nothing to repeat") CHECK_FAILED);
       }
       // fallthrough
     }
     default:
       builder.AddCharacter(current());
@@ skipping 32 matching lines @@
         continue;
       }
     default:
       continue;
     }
     bool is_greedy = true;
     if (current() == '?') {
       is_greedy = false;
       Advance();
     }
+    non_simple_ = true;  // Adding quantifier might *remove* look-ahead.
     builder.AddQuantifierToAtom(min, max, is_greedy);
   }
 }
 
 class SourceCharacter {
  public:
   static bool Is(uc32 c) {
     switch (c) {
       // case ']': case '}':
       // In spidermonkey and jsc these are treated as source characters
@@ skipping 472 matching lines @@
   // Make sure we have a stack guard.
   StackGuard guard;
   RegExpParser parser(input, &result->error, multiline);
   result->tree = parser.ParsePattern();
   if (parser.failed()) {
     ASSERT(result->tree == NULL);
     ASSERT(!result->error.is_null());
   } else {
     ASSERT(result->tree != NULL);
     ASSERT(result->error.is_null());
-    result->has_character_escapes = parser.HasCharacterEscapes();
+    result->non_simple = parser.IsNonSimple();
     result->capture_count = parser.captures_started();
   }
   return !parser.failed();
 }
 
 
 FunctionLiteral* MakeAST(bool compile_in_global_context,
                          Handle<Script> script,
                          v8::Extension* extension,
                          ScriptDataImpl* pre_data) {
@@ skipping 38 matching lines @@
                        start_position,
                        is_expression);
   return result;
 }
 
 
 #undef NEW
 
 
 } }  // namespace v8::internal