Add support for surrogates when serializing and deserializing for native ports

runtime/vm/unicode.h

-// Copyright (c) 2011, the Dart project authors.  Please see the AUTHORS file
+// Copyright (c) 2012, the Dart project authors.  Please see the AUTHORS file
 // for details. All rights reserved. Use of this source code is governed by a
 // BSD-style license that can be found in the LICENSE file.
 
 #ifndef VM_UNICODE_H_
 #define VM_UNICODE_H_
 
 #include "vm/allocation.h"
 #include "vm/globals.h"
 
 namespace dart {
@@ skipping 13 matching lines @@
   static const intptr_t kMaxThreeByteChar = 0xFFFF;
   static const intptr_t kMaxFourByteChar  = 0x10FFFF;
 
   static intptr_t CodePointCount(const uint8_t* utf8_array,
                                  intptr_t array_len,
                                  Type* type);
 
   // Returns true if 'utf8_array' is a valid UTF-8 string.
   static bool IsValid(const uint8_t* utf8_array, intptr_t array_len);
 
+  // Returns true if 'utf8_array' is a valid UTF-8 string allowing
+  // UTF-8 encoded UTF-16 surrogate code units.
+  static bool IsValidAllowSurrogates(
+      const uint8_t* utf8_array, intptr_t array_len);
+
   static intptr_t Length(int32_t ch);
   static intptr_t Length(const String& str);
 
   static intptr_t Encode(int32_t ch, char* dst);
   static intptr_t Encode(const String& src, char* dst, intptr_t len);
 
   static intptr_t Decode(const uint8_t* utf8_array,
                          intptr_t array_len,
                          int32_t* ch);
+  static intptr_t DecodeAllowSurrogates(const uint8_t* utf8_array,
+                                        intptr_t array_len,
+                                        int32_t* ch);
 
   static bool DecodeToLatin1(const uint8_t* utf8_array,
                              intptr_t array_len,
                              uint8_t* dst,
                              intptr_t len);
   static bool DecodeToUTF16(const uint8_t* utf8_array,
                             intptr_t array_len,
                             uint16_t* dst,
                             intptr_t len);
+  static bool DecodeToUTF16AllowSurrogates(const uint8_t* utf8_array,
+                                           intptr_t array_len,
+                                           uint16_t* dst,
+                                           intptr_t len);
   static bool DecodeToUTF32(const uint8_t* utf8_array,
                             intptr_t array_len,
                             int32_t* dst,
                             intptr_t len);
   static bool DecodeCStringToUTF32(const char* str,
                                    int32_t* dst,
                                    intptr_t len) {
     ASSERT(str != NULL);
     intptr_t array_len = strlen(str);
     const uint8_t* utf8_array = reinterpret_cast<const uint8_t*>(str);
     return DecodeToUTF32(utf8_array, array_len, dst, len);
   }
 };
 
 
 class Utf16 : AllStatic {
  public:
   static const int32_t kMaxBmpCodepoint  = 0xFFFF;
 
   static const int32_t kLeadSurrogateOffset = (0xD800 - (0x10000 >> 10));
 
   static const int32_t kSurrogateOffset = (0x10000 - (0xD800 << 10) - 0xDC00);
 
+  class CodePointIterator {
+   public:
+    explicit CodePointIterator(const uint16_t* utf16_array,
+                               intptr_t array_len)

[siva, 2012/11/27 03:00:25]

Has two parameters in the constructor why is explicit needed?

[Søren Gjesse, 2012/11/27 11:35:54]

Not needed - removed.

+        : utf16_array_(utf16_array),
+          array_len_(array_len),
+          index_(-1),
+          ch_(-1) {
+    }
+
+    int32_t Current() {

[siva, 2012/11/27 03:00:25]

int32_t Current() const {

[Søren Gjesse, 2012/11/27 11:35:54]

Done (added for String::CodePointIterator as well).

+      ASSERT(index_ >= 0);
+      ASSERT(index_ < array_len_);
+      return ch_;
+    }
+
+    bool Next();
+
+   private:
+    const uint16_t* utf16_array_;
+    intptr_t array_len_;
+    intptr_t index_;
+    int32_t ch_;

[siva, 2012/11/27 03:00:25]

Missing DISALLOW stuff?

[Søren Gjesse, 2012/11/27 11:35:54]

Done.

+  };
+
   // Returns the length of the code point in UTF-16 code units.
   static intptr_t Length(int32_t ch) {
     return (ch <= kMaxBmpCodepoint) ? 1 : 2;
   }
 
   // Returns true if ch is a lead or trail surrogate.
   static bool IsSurrogate(int32_t ch) {
     return (ch & 0xFFFFF800) == 0xD800;
   }
 
@@ skipping 76 matching lines @@
   // Data for small code points with one mapping
   static const int16_t stage2_[];
 
   // Data for large code points or code points with both mappings.
   static const int32_t stage2_exception_[][2];
 };
 
 }  // namespace dart
 
 #endif  // VM_UNICODE_H_