MdeModulePkg/Universal/Console/TerminalDxe/Vtutf8.c

   1 /** @file
   2   Implementation of translation upon VT-UTF8.
   3
   4 Copyright (c) 2006, Intel Corporation. <BR>
   5 All rights reserved. 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   @return None.
  24
  25 **/
  26 VOID
  27 VTUTF8RawDataToUnicode (
  28   IN  TERMINAL_DEV    *TerminalDevice
  29   )
  30 {
  31   UTF8_CHAR Utf8Char;
  32   UINT8     ValidBytes;
  33   UINT16    UnicodeChar;
  34
  35   ValidBytes = 0;
  36   //
  37   // pop the raw data out from the raw fifo,
  38   // and translate it into unicode, then push
  39   // the unicode into unicode fifo, until the raw fifo is empty.
  40   //
  41   while (!IsRawFiFoEmpty (TerminalDevice)) {
  42
  43     GetOneValidUtf8Char (TerminalDevice, &Utf8Char, &ValidBytes);
  44
  45     if (ValidBytes < 1 || ValidBytes > 3) {
  46       continue;
  47     }
  48
  49     Utf8ToUnicode (Utf8Char, ValidBytes, (CHAR16 *) &UnicodeChar);
  50
  51     UnicodeFiFoInsertOneKey (TerminalDevice, UnicodeChar);
  52   }
  53 }
  54
  55 /**
  56   Get one valid VT-UTF8 characters set from Raw Data FIFO.
  57
  58   @param  Utf8Device          The terminal device.
  59   @param  Utf8Char            Returned valid VT-UTF8 characters set.
  60   @param  ValidBytes          The count of returned VT-VTF8 characters.
  61                               If ValidBytes is zero, no valid VT-UTF8 returned.
  62
  63   @retval None.
  64
  65 **/
  66 VOID
  67 GetOneValidUtf8Char (
  68   IN  TERMINAL_DEV      *Utf8Device,
  69   OUT UTF8_CHAR         *Utf8Char,
  70   OUT UINT8             *ValidBytes
  71   )
  72 {
  73   UINT8   Temp;
  74   UINT8   Index;
  75   BOOLEAN FetchFlag;
  76
  77   Temp      = 0;
  78   Index     = 0;
  79   FetchFlag = TRUE;
  80
  81   //
  82   // if no valid Utf8 char is found in the RawFiFo,
  83   // then *ValidBytes will be zero.
  84   //
  85   *ValidBytes = 0;
  86
  87   while (!IsRawFiFoEmpty (Utf8Device)) {
  88
  89     RawFiFoRemoveOneKey (Utf8Device, &Temp);
  90
  91     switch (*ValidBytes) {
  92
  93     case 0:
  94       if ((Temp & 0x80) == 0) {
  95         //
  96         // one-byte utf8 char
  97         //
  98         *ValidBytes       = 1;
  99
 100         Utf8Char->Utf8_1  = Temp;
 101
 102         FetchFlag         = FALSE;
 103
 104       } else if ((Temp & 0xe0) == 0xc0) {
 105         //
 106         // two-byte utf8 char
 107         //
 108         *ValidBytes         = 2;
 109
 110         Utf8Char->Utf8_2[1] = Temp;
 111
 112       } else if ((Temp & 0xf0) == 0xe0) {
 113         //
 114         // three-byte utf8 char
 115         //
 116         *ValidBytes         = 3;
 117
 118         Utf8Char->Utf8_3[2] = Temp;
 119
 120         Index++;
 121
 122       } else {
 123         //
 124         // reset *ValidBytes to zero, let valid utf8 char search restart
 125         //
 126         *ValidBytes = 0;
 127       }
 128
 129       break;
 130
 131     case 2:
 132       if ((Temp & 0xc0) == 0x80) {
 133
 134         Utf8Char->Utf8_2[0] = Temp;
 135
 136         FetchFlag           = FALSE;
 137
 138       } else {
 139
 140         *ValidBytes = 0;
 141       }
 142       break;
 143
 144     case 3:
 145       if ((Temp & 0xc0) == 0x80) {
 146
 147         Utf8Char->Utf8_3[2 - Index] = Temp;
 148         Index++;
 149         if (Index == 3) {
 150           FetchFlag = FALSE;
 151         }
 152       } else {
 153
 154         *ValidBytes = 0;
 155         Index       = 0;
 156       }
 157       break;
 158
 159     default:
 160       break;
 161     }
 162
 163     if (!FetchFlag) {
 164       break;
 165     }
 166   }
 167
 168   return ;
 169 }
 170
 171 /**
 172   Translate VT-UTF8 characters into one Unicode character.
 173
 174   UTF8 Encoding Table
 175   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 176         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 177         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 178        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 179
 180
 181   @param  Utf8Char         VT-UTF8 character set needs translating.
 182   @param  ValidBytes       The count of valid VT-UTF8 characters.
 183   @param  UnicodeChar      Returned unicode character.
 184
 185   @return None.
 186
 187 **/
 188 VOID
 189 Utf8ToUnicode (
 190   IN  UTF8_CHAR       Utf8Char,
 191   IN  UINT8           ValidBytes,
 192   OUT CHAR16          *UnicodeChar
 193   )
 194 {
 195   UINT8 UnicodeByte0;
 196   UINT8 UnicodeByte1;
 197   UINT8 Byte0;
 198   UINT8 Byte1;
 199   UINT8 Byte2;
 200
 201   *UnicodeChar = 0;
 202
 203   //
 204   // translate utf8 code to unicode, in terminal standard,
 205   // up to 3 bytes utf8 code is supported.
 206   //
 207   switch (ValidBytes) {
 208   case 1:
 209     //
 210     // one-byte utf8 code
 211     //
 212     *UnicodeChar = (UINT16) Utf8Char.Utf8_1;
 213     break;
 214
 215   case 2:
 216     //
 217     // two-byte utf8 code
 218     //
 219     Byte0         = Utf8Char.Utf8_2[0];
 220     Byte1         = Utf8Char.Utf8_2[1];
 221
 222     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 223     UnicodeByte1  = (UINT8) ((Byte1 >> 2) & 0x07);
 224     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 225     break;
 226
 227   case 3:
 228     //
 229     // three-byte utf8 code
 230     //
 231     Byte0         = Utf8Char.Utf8_3[0];
 232     Byte1         = Utf8Char.Utf8_3[1];
 233     Byte2         = Utf8Char.Utf8_3[2];
 234
 235     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 236     UnicodeByte1  = (UINT8) ((Byte2 << 4) | ((Byte1 >> 2) & 0x0f));
 237     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 238
 239   default:
 240     break;
 241   }
 242
 243   return ;
 244 }
 245
 246 /**
 247   Translate one Unicode character into VT-UTF8 characters.
 248
 249   UTF8 Encoding Table
 250   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 251         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 252         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 253        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 254
 255
 256   @param  Unicode          Unicode character need translating.
 257   @param  Utf8Char         Return VT-UTF8 character set.
 258   @param  ValidBytes       The count of valid VT-UTF8 characters. If
 259                            ValidBytes is zero, no valid VT-UTF8 returned.
 260
 261   @return None.
 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 }