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 **/
  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       if ((Temp & 0xc0) == 0x80) {
 129
 130         Utf8Char->Utf8_2[0] = Temp;
 131
 132         FetchFlag           = FALSE;
 133
 134       } else {
 135
 136         *ValidBytes = 0;
 137       }
 138       break;
 139
 140     case 3:
 141       if ((Temp & 0xc0) == 0x80) {
 142
 143         Utf8Char->Utf8_3[2 - Index] = Temp;
 144         Index++;
 145         if (Index == 3) {
 146           FetchFlag = FALSE;
 147         }
 148       } else {
 149
 150         *ValidBytes = 0;
 151         Index       = 0;
 152       }
 153       break;
 154
 155     default:
 156       break;
 157     }
 158
 159     if (!FetchFlag) {
 160       break;
 161     }
 162   }
 163
 164   return ;
 165 }
 166
 167 /**
 168   Translate VT-UTF8 characters into one Unicode character.
 169
 170   UTF8 Encoding Table
 171   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 172         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 173         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 174        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 175
 176
 177   @param  Utf8Char         VT-UTF8 character set needs translating.
 178   @param  ValidBytes       The count of valid VT-UTF8 characters.
 179   @param  UnicodeChar      Returned unicode character.
 180
 181 **/
 182 VOID
 183 Utf8ToUnicode (
 184   IN  UTF8_CHAR       Utf8Char,
 185   IN  UINT8           ValidBytes,
 186   OUT CHAR16          *UnicodeChar
 187   )
 188 {
 189   UINT8 UnicodeByte0;
 190   UINT8 UnicodeByte1;
 191   UINT8 Byte0;
 192   UINT8 Byte1;
 193   UINT8 Byte2;
 194
 195   *UnicodeChar = 0;
 196
 197   //
 198   // translate utf8 code to unicode, in terminal standard,
 199   // up to 3 bytes utf8 code is supported.
 200   //
 201   switch (ValidBytes) {
 202   case 1:
 203     //
 204     // one-byte utf8 code
 205     //
 206     *UnicodeChar = (UINT16) Utf8Char.Utf8_1;
 207     break;
 208
 209   case 2:
 210     //
 211     // two-byte utf8 code
 212     //
 213     Byte0         = Utf8Char.Utf8_2[0];
 214     Byte1         = Utf8Char.Utf8_2[1];
 215
 216     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 217     UnicodeByte1  = (UINT8) ((Byte1 >> 2) & 0x07);
 218     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 219     break;
 220
 221   case 3:
 222     //
 223     // three-byte utf8 code
 224     //
 225     Byte0         = Utf8Char.Utf8_3[0];
 226     Byte1         = Utf8Char.Utf8_3[1];
 227     Byte2         = Utf8Char.Utf8_3[2];
 228
 229     UnicodeByte0  = (UINT8) ((Byte1 << 6) | (Byte0 & 0x3f));
 230     UnicodeByte1  = (UINT8) ((Byte2 << 4) | ((Byte1 >> 2) & 0x0f));
 231     *UnicodeChar  = (UINT16) (UnicodeByte0 | (UnicodeByte1 << 8));
 232
 233   default:
 234     break;
 235   }
 236
 237   return ;
 238 }
 239
 240 /**
 241   Translate one Unicode character into VT-UTF8 characters.
 242
 243   UTF8 Encoding Table
 244   Bits per Character | Unicode Character Range | Unicode Binary  Encoding |     UTF8 Binary Encoding
 245         0-7              |     0x0000 - 0x007F     |     00000000 0xxxxxxx        |   0xxxxxxx
 246         8-11           |     0x0080 - 0x07FF       |     00000xxx xxxxxxxx        |   110xxxxx 10xxxxxx
 247        12-16           |     0x0800 - 0xFFFF       |     xxxxxxxx xxxxxxxx        |   1110xxxx 10xxxxxx 10xxxxxx
 248
 249
 250   @param  Unicode          Unicode character need translating.
 251   @param  Utf8Char         Return VT-UTF8 character set.
 252   @param  ValidBytes       The count of valid VT-UTF8 characters. If
 253                            ValidBytes is zero, no valid VT-UTF8 returned.
 254
 255 **/
 256 VOID
 257 UnicodeToUtf8 (
 258   IN  CHAR16      Unicode,
 259   OUT UTF8_CHAR   *Utf8Char,
 260   OUT UINT8       *ValidBytes
 261   )
 262 {
 263   UINT8 UnicodeByte0;
 264   UINT8 UnicodeByte1;
 265   //
 266   // translate unicode to utf8 code
 267   //
 268   UnicodeByte0  = (UINT8) Unicode;
 269   UnicodeByte1  = (UINT8) (Unicode >> 8);
 270
 271   if (Unicode < 0x0080) {
 272
 273     Utf8Char->Utf8_1  = (UINT8) (UnicodeByte0 & 0x7f);
 274     *ValidBytes       = 1;
 275
 276   } else if (Unicode < 0x0800) {
 277     //
 278     // byte sequence: high -> low
 279     //                Utf8_2[0], Utf8_2[1]
 280     //
 281     Utf8Char->Utf8_2[1] = (UINT8) ((UnicodeByte0 & 0x3f) + 0x80);
 282     Utf8Char->Utf8_2[0] = (UINT8) ((((UnicodeByte1 << 2) + (UnicodeByte0 >> 6)) & 0x1f) + 0xc0);
 283
 284     *ValidBytes         = 2;
 285
 286   } else {
 287     //
 288     // byte sequence: high -> low
 289     //                Utf8_3[0], Utf8_3[1], Utf8_3[2]
 290     //
 291     Utf8Char->Utf8_3[2] = (UINT8) ((UnicodeByte0 & 0x3f) + 0x80);
 292     Utf8Char->Utf8_3[1] = (UINT8) ((((UnicodeByte1 << 2) + (UnicodeByte0 >> 6)) & 0x3f) + 0x80);
 293     Utf8Char->Utf8_3[0] = (UINT8) (((UnicodeByte1 >> 4) & 0x0f) + 0xe0);
 294
 295     *ValidBytes         = 3;
 296   }
 297 }
 298
 299
 300 /**
 301   Check if input string is valid VT-UTF8 string.
 302
 303   @param  TerminalDevice          The terminal device.
 304   @param  WString                 The input string.
 305
 306   @retval EFI_SUCCESS             If all input characters are valid.
 307
 308 **/
 309 EFI_STATUS
 310 VTUTF8TestString (
 311   IN  TERMINAL_DEV    *TerminalDevice,
 312   IN  CHAR16          *WString
 313   )
 314 {
 315   //
 316   // to utf8, all kind of characters are supported.
 317   //
 318   return EFI_SUCCESS;
 319 }