Merge regexp2000 back into bleeding_edge

src/unicode.h

Index: src/unicode.h
===================================================================
--- src/unicode.h	(revision 830)
+++ src/unicode.h	(working copy)
@@ -44,7 +44,7 @@
  * The max length of the result of converting the case of a single
  * character.
  */
-static const int kMaxCaseConvertedSize = 3;
+static const int kMaxMappingSize = 4;
 
 template <class T, int size = 256>
 class Predicate {
@@ -80,12 +80,13 @@
   friend class Test;
   int CalculateValue(uchar c, uchar n, uchar* result);
   struct CacheEntry {
-    inline CacheEntry() : code_point_(0), offset_(0) { }
+    inline CacheEntry() : code_point_(kNoChar), offset_(0) { }
     inline CacheEntry(uchar code_point, signed offset)
       : code_point_(code_point),
         offset_(offset) { }
-    uchar code_point_ : 21;
-    signed offset_ : 11;
+    uchar code_point_;
+    signed offset_;
+    static const int kNoChar = (1 << 21) - 1;
   };
   static const int kSize = size;
   static const int kMask = kSize - 1;
@@ -222,45 +223,15 @@
 struct Space {
   static bool Is(uchar c);
 };
-struct Titlecase {
-  static bool Is(uchar c);
-};
 struct Number {
   static bool Is(uchar c);
 };
-struct DecimalDigit {
-  static bool Is(uchar c);
-};
-struct Ideographic {
-  static bool Is(uchar c);
-};
 struct WhiteSpace {
   static bool Is(uchar c);
 };
-struct HexDigit {
-  static bool Is(uchar c);
-};
-struct AsciiHexDigit {
-  static bool Is(uchar c);
-};
-struct BidiControl {
-  static bool Is(uchar c);
-};
-struct JoinControl {
-  static bool Is(uchar c);
-};
-struct Dash {
-  static bool Is(uchar c);
-};
-struct Hyphen {
-  static bool Is(uchar c);
-};
 struct LineTerminator {
   static bool Is(uchar c);
 };
-struct RegExpSpecialChar {
-  static bool Is(uchar c);
-};
 struct CombiningMark {
   static bool Is(uchar c);
 };
@@ -268,17 +239,40 @@
   static bool Is(uchar c);
 };
 struct ToLowercase {
+  static const int kMaxWidth = 3;
   static int Convert(uchar c,
                      uchar n,
                      uchar* result,
                      bool* allow_caching_ptr);
 };
 struct ToUppercase {
+  static const int kMaxWidth = 3;
   static int Convert(uchar c,
                      uchar n,
                      uchar* result,
                      bool* allow_caching_ptr);
 };
+struct Ecma262Canonicalize {
+  static const int kMaxWidth = 1;
+  static int Convert(uchar c,
+                     uchar n,
+                     uchar* result,
+                     bool* allow_caching_ptr);
+};
+struct Ecma262UnCanonicalize {
+  static const int kMaxWidth = 4;
+  static int Convert(uchar c,
+                     uchar n,
+                     uchar* result,
+                     bool* allow_caching_ptr);
+};
+struct CanonicalizationRange {
+  static const int kMaxWidth = 1;
+  static int Convert(uchar c,
+                     uchar n,
+                     uchar* result,
+                     bool* allow_caching_ptr);
+};
 
 }  // namespace unibrow