scintilla/UniConversion.cxx

   1 // Scintilla source code edit control
   2 /** @file UniConversion.cxx
   3  ** Functions to handle UFT-8 and UCS-2 strings.
   4  **/
   5 // Copyright 1998-2001 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #include <stdlib.h>
   9
  10 #include "UniConversion.h"
  11
  12 unsigned int UTF8Length(const wchar_t *uptr, unsigned int tlen) {
  13         unsigned int len = 0;
  14         for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
  15                 unsigned int uch = uptr[i];
  16                 if (uch < 0x80)
  17                         len++;
  18                 else if (uch < 0x800)
  19                         len += 2;
  20                 else
  21                         len +=3;
  22         }
  23         return len;
  24 }
  25
  26 void UTF8FromUCS2(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned int len) {
  27         int k = 0;
  28         for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
  29                 unsigned int uch = uptr[i];
  30                 if (uch < 0x80) {
  31                         putf[k++] = static_cast<char>(uch);
  32                 } else if (uch < 0x800) {
  33                         putf[k++] = static_cast<char>(0xC0 | (uch >> 6));
  34                         putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
  35                 } else {
  36                         putf[k++] = static_cast<char>(0xE0 | (uch >> 12));
  37                         putf[k++] = static_cast<char>(0x80 | ((uch >> 6) & 0x3f));
  38                         putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
  39                 }
  40         }
  41         putf[len] = '\0';
  42 }
  43
  44 unsigned int UCS2Length(const char *s, unsigned int len) {
  45         unsigned int ulen = 0;
  46         for (unsigned int i=0;i<len;i++) {
  47                 unsigned char ch = static_cast<unsigned char>(s[i]);
  48                 if ((ch < 0x80) || (ch > (0x80 + 0x40)))
  49                         ulen++;
  50         }
  51         return ulen;
  52 }
  53
  54 unsigned int UCS2FromUTF8(const char *s, unsigned int len, wchar_t *tbuf, unsigned int tlen) {
  55         unsigned int ui=0;
  56         const unsigned char *us = reinterpret_cast<const unsigned char *>(s);
  57         unsigned int i=0;
  58         while ((i<len) && (ui<tlen)) {
  59                 unsigned char ch = us[i++];
  60                 if (ch < 0x80) {
  61                         tbuf[ui] = ch;
  62                 } else if (ch < 0x80 + 0x40 + 0x20) {
  63                         tbuf[ui] = static_cast<wchar_t>((ch & 0x1F) << 6);
  64                         ch = us[i++];
  65                         tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + (ch & 0x7F));
  66                 } else {
  67                         tbuf[ui] = static_cast<wchar_t>((ch & 0xF) << 12);
  68                         ch = us[i++];
  69                         tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + ((ch & 0x7F) << 6));
  70                         ch = us[i++];
  71                         tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + (ch & 0x7F));
  72                 }
  73                 ui++;
  74         }
  75         return ui;
  76 }