michael@0: // Copyright 2013 Google Inc. All Rights Reserved.
michael@0: //
michael@0: // Licensed under the Apache License, Version 2.0 (the "License");
michael@0: // you may not use this file except in compliance with the License.
michael@0: // You may obtain a copy of the License at
michael@0: //
michael@0: //     http://www.apache.org/licenses/LICENSE-2.0
michael@0: //
michael@0: // Unless required by applicable law or agreed to in writing, software
michael@0: // distributed under the License is distributed on an "AS IS" BASIS,
michael@0: // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
michael@0: // See the License for the specific language governing permissions and
michael@0: // limitations under the License.
michael@0: 
michael@0: //
michael@0: // Routine that maps a Unicode code point to an interchange-valid one
michael@0: //
michael@0: 
michael@0: #include "fixunicodevalue.h"
michael@0: #include "integral_types.h"
michael@0: 
michael@0: namespace CLD2 {
michael@0: 
michael@0: // Guarantees that the resulting output value is interchange valid
michael@0: //  00-FF; map to spaces or MS CP1252
michael@0: //  D800-DFFF; surrogates
michael@0: //  FDD0-FDEF; non-characters
michael@0: //  xxFFFE-xxFFFF; non-characters
michael@0: char32 FixUnicodeValue(char32 uv) {
michael@0:   uint32 uuv = static_cast<uint32>(uv);
michael@0:   if (uuv < 0x0100) {
michael@0:     return kMapFullMicrosoft1252OrSpace[uuv];
michael@0:   }
michael@0:   if (uuv < 0xD800) {
michael@0:     return uv;
michael@0:   }
michael@0:   if ((uuv & ~0x0F) == 0xFDD0) {              // non-characters
michael@0:     return 0xFFFD;
michael@0:   }
michael@0:   if ((uuv & ~0x0F) == 0xFDE0) {              // non-characters
michael@0:     return 0xFFFD;
michael@0:   }
michael@0:   if ((uuv & 0x00FFFE) == 0xFFFE) {           // non-characters
michael@0:     return 0xFFFD;
michael@0:   }
michael@0:   if ((0xE000 <= uuv) && (uuv <= 0x10FFFF))  {
michael@0:     return uv;
michael@0:   }
michael@0:   // surrogates and negative and > 0x10FFFF all land here
michael@0:   return 0xFFFD;
michael@0: }
michael@0: 
michael@0: }       // End namespace CLD2
michael@0: