MdeModulePkg/Universal/Console/TerminalDxe/Vtutf8.c

   1 /** @file
   2   Implementation of translation upon VT-UTF8.
   3
   4 Copyright (c) 2006 - 2009, Intel Corporation. All rights reserved.<BR>
   5 This program and the accompanying materials
   6 are licensed and made available under the terms and conditions of the BSD License
   7 which accompanies this distribution.  The full text of the license may be found at
   8 http://opensource.org/licenses/bsd-license.php
   9
  10 THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
  11 WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
  12
  13 **/
  14
  15 #include "Terminal.h"
  16
  17 /**
  18   Translate all VT-UTF8 characters in the Raw FIFI into unicode characters,
  19   and insert them into Unicode FIFO.
  20
  21   @param TerminalDevice          The terminal device.
  22
  23 **/
  24 VOID
  25 VTUTF8RawDataToUnicode (
  26   IN  TERMINAL_DEV    *TerminalDevice
  27   )
  28 {
  29   UTF8_CHAR Utf8Char;
  30   UINT8     ValidBytes;
  31   UINT16    UnicodeChar;
  32
  33   ValidBytes = 0;
  34   //
  35   // pop the raw data out from the raw fifo,
  36   // and translate it into unicode, then push
  37   // the unicode into unicode fifo, until the raw fifo is empty.
  38   //
  39   while (!IsRawFiFoEmpty (TerminalDevice)) {
  40
  41     GetOneValidUtf8Char (TerminalDevice, &Utf8Char, &ValidBytes);
  42
  43     if (ValidBytes < 1 || ValidBytes > 3) {
  44       continue;
  45     }
  46
  47     Utf8ToUnicode (Utf8Char, ValidBytes, (CHAR16 *) &UnicodeChar);
  48
  49     UnicodeFiFoInsertOneKey (TerminalDevice, UnicodeChar);
  50   }
  51 }
  52
  53 /**
  54   Get one valid VT-UTF8 characters set from Raw Data FIFO.
  55
  56   @param  Utf8Device          The terminal device.
  57   @param  Utf8Char            Returned valid VT-UTF8 characters set.
  58   @param  ValidBytes          The count of returned VT-VTF8 characters.
  59                               If ValidBytes is zero, no valid VT-UTF8 returned.
  60
  61 **/
  62 VOID
  63 GetOneValidUtf8Char (
  64   IN  TERMINAL_DEV      *Utf8Device,
  65   OUT UTF8_CHAR         *Utf8Char,
  66   OUT UINT8             *ValidBytes
  67   )
  68 {
  69   UINT8   Temp;
  70   UINT8   Index;
  71   BOOLEAN FetchFlag;
  72
  73   Temp      = 0;
  74   Index     = 0;
  75   FetchFlag = TRUE;
  76
  77   //
  78   // if no valid Utf8 char is found in the RawFiFo,
  79   // then *ValidBytes will be zero.
  80   //
  81   *ValidBytes = 0;
  82
  83   while (!IsRawFiFoEmpty (Utf8Device)) {
  84
  85     RawFiFoRemoveOneKey (Utf8Device, &Temp);
  86
  87     switch (*ValidBytes) {
  88
  89     case 0:
  90       if ((Temp & 0x80) == 0) {
  91         //
  92         // one-byte utf8 char
  93         //
  94         *ValidBytes       = 1;
  95
  96         Utf8Char->Utf8_1  = Temp;
  97
  98         FetchFlag         = FALSE;
  99
 100       } else if ((Temp & 0xe0) == 0xc0) {
 101         //
 102         // two-byte utf8 char
 103         //
 104         *ValidBytes         = 2;
 105
 106         Utf8Char->Utf8_2[1] = Temp;
 107
 108       } else if ((Temp & 0xf0) == 0xe0) {
 109         //
 110         // three-byte utf8 char
 111         //
 112         *ValidBytes         = 3;
 113
 114         Utf8Char->Utf8_3[2] = Temp;
 115
 116         Index++;
 117
 118       } else {
 119         //
 120         // reset *ValidBytes to zero, let valid utf8 char search restart
 121         //
 122         *ValidBytes = 0;
 123       }
 124
 125       break;
 126
 127     case 2:
 128       //
 129       // two-byte utf8 char go on
 130       //
 131       if ((Temp & 0xc0) == 0x80) {
 132
 133         Utf8Char->Utf8_2[0] = Temp;
 134
 135         FetchFlag           = FALSE;
 136
 137       } else {
 138
 139         *ValidBytes = 0;
 140       }
 141       break;
 142
 143     case 3:
 144       //
 145       // three-byte utf8 char go on
 146       //
 147       if ((Temp & 0xc0) == 0x80) {
 148
 149         Utf8Char->Utf8_3[2 - Index] = Temp;
 150         Index++;
 151         if (Index > 2) {
 152           FetchFlag = FALSE;
 153         }
 154       } else {
 155         //
 156         // reset *ValidBytes and Index to zero, let valid utf8 char search restart
 157         //
 158         *ValidBytes = 0;
 159         Index       = 0;
 160       }
 161       break;
 162
 163     default:
 164       break;
 165     }
 166
 167     if (!FetchFlag) {
 168       break;
 169     }
 170   }
 171
 172   return ;
 173 }
 174
 175 /**
 176   Translate VT-UTF8 characters into one Unicode character.
 177
 178   UTF8 Encoding Table
 179   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 180         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 181         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 182        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 183
 184
 185   @param  Utf8Char         VT-UTF8 character set needs translating.
 186   @param  ValidBytes       The count of valid VT-UTF8 characters.
 187   @param  UnicodeChar      Returned unicode character.
 188
 189 **/
 190 VOID
 191 Utf8ToUnicode (
 192   IN  UTF8_CHAR       Utf8Char,
 193   IN  UINT8           ValidBytes,
 194   OUT CHAR16          *UnicodeChar
 195   )
 196 {
 197   UINT8 UnicodeByte0;
 198   UINT8 UnicodeByte1;
 199   UINT8 Byte0;
 200   UINT8 Byte1;
 201   UINT8 Byte2;
 202
 203   *UnicodeChar = 0;
 204
 205   //
 206   // translate utf8 code to unicode, in terminal standard,
 207   // up to 3 bytes utf8 code is supported.
 208   //
 209   switch (ValidBytes) {
 210   case 1:
 211     //
 212     // one-byte utf8 code
 213     //
 214     *UnicodeChar = (UINT16) Utf8Char.Utf8_1;
 215     break;
 216
 217   case 2:
 218     //
 219     // two-byte utf8 code
 220     //
 221     Byte0         = Utf8Char.Utf8_2[0];
 222     Byte1         = Utf8Char.Utf8_2[1];
 223
 224     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 225     UnicodeByte1  = (UINT8) ((Byte1 >> 2) & 0x07);
 226     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 227     break;
 228
 229   case 3:
 230     //
 231     // three-byte utf8 code
 232     //
 233     Byte0         = Utf8Char.Utf8_3[0];
 234     Byte1         = Utf8Char.Utf8_3[1];
 235     Byte2         = Utf8Char.Utf8_3[2];
 236
 237     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 238     UnicodeByte1  = (UINT8) ((Byte2 << 4) | ((Byte1 >> 2) & 0x0f));
 239     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 240
 241   default:
 242     break;
 243   }
 244
 245   return ;
 246 }
 247
 248 /**
 249   Translate one Unicode character into VT-UTF8 characters.
 250
 251   UTF8 Encoding Table
 252   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 253         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 254         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 255        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 256
 257
 258   @param  Unicode          Unicode character need translating.
 259   @param  Utf8Char         Return VT-UTF8 character set.
 260   @param  ValidBytes       The count of valid VT-UTF8 characters. If
 261                            ValidBytes is zero, no valid VT-UTF8 returned.
 262
 263 **/
 264 VOID
 265 UnicodeToUtf8 (
 266   IN  CHAR16      Unicode,
 267   OUT UTF8_CHAR   *Utf8Char,
 268   OUT UINT8       *ValidBytes
 269   )
 270 {
 271   UINT8 UnicodeByte0;
 272   UINT8 UnicodeByte1;
 273   //
 274   // translate unicode to utf8 code
 275   //
 276   UnicodeByte0  = (UINT8) Unicode;
 277   UnicodeByte1  = (UINT8) (Unicode >> 8);
 278
 279   if (Unicode < 0x0080) {
 280
 281     Utf8Char->Utf8_1  = (UINT8) (UnicodeByte0 & 0x7f);
 282     *ValidBytes       = 1;
 283
 284   } else if (Unicode < 0x0800) {
 285     //
 286     // byte sequence: high -> low
 287     //                Utf8_2[0], Utf8_2[1]
 288     //
 289     Utf8Char->Utf8_2[1] = (UINT8) ((UnicodeByte0 & 0x3f) + 0x80);
 290     Utf8Char->Utf8_2[0] = (UINT8) ((((UnicodeByte1 << 2) + (UnicodeByte0 >> 6)) & 0x1f) + 0xc0);
 291
 292     *ValidBytes         = 2;
 293
 294   } else {
 295     //
 296     // byte sequence: high -> low
 297     //                Utf8_3[0], Utf8_3[1], Utf8_3[2]
 298     //
 299     Utf8Char->Utf8_3[2] = (UINT8) ((UnicodeByte0 & 0x3f) + 0x80);
 300     Utf8Char->Utf8_3[1] = (UINT8) ((((UnicodeByte1 << 2) + (UnicodeByte0 >> 6)) & 0x3f) + 0x80);
 301     Utf8Char->Utf8_3[0] = (UINT8) (((UnicodeByte1 >> 4) & 0x0f) + 0xe0);
 302
 303     *ValidBytes         = 3;
 304   }
 305 }
 306
 307
 308 /**
 309   Check if input string is valid VT-UTF8 string.
 310
 311   @param  TerminalDevice          The terminal device.
 312   @param  WString                 The input string.
 313
 314   @retval EFI_SUCCESS             If all input characters are valid.
 315
 316 **/
 317 EFI_STATUS
 318 VTUTF8TestString (
 319   IN  TERMINAL_DEV    *TerminalDevice,
 320   IN  CHAR16          *WString
 321   )
 322 {
 323   //
 324   // to utf8, all kind of characters are supported.
 325   //
 326   return EFI_SUCCESS;
 327 }