contrib/src/stc/scintilla/src/UniConversion.cxx

   1 // UniConversion.h - functions to handle UFT-8 and UCS-2 strings
   2 // Copyright 1998-2000 by Neil Hodgson <neilh@scintilla.org>
   3 // The License.txt file describes the conditions under which this software may be distributed.
   4
   5 #include <stdlib.h>
   6
   7 #include "UniConversion.h"
   8
   9 unsigned int UTF8Length(const wchar_t *uptr, unsigned int tlen) {
  10         unsigned int len = 0;
  11         for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
  12                 unsigned int uch = uptr[i];
  13                 if (uch < 0x80)
  14                         len++;
  15                 else if (uch < 0x800)
  16                         len+=2;
  17                 else
  18                         len +=3;
  19         }
  20         return len;
  21 }
  22
  23 void UTF8FromUCS2(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned int len) {
  24         int k = 0;
  25         for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
  26                 unsigned int uch = uptr[i];
  27                 if (uch < 0x80) {
  28                         putf[k++] = static_cast<char>(uch);
  29                 } else if (uch < 0x800) {
  30                         putf[k++] = static_cast<char>(0xC0 | (uch >> 6));
  31                         putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
  32                 } else {
  33                         putf[k++] = static_cast<char>(0xE0 | (uch >> 12));
  34                         putf[k++] = static_cast<char>(0x80 | ((uch >> 6) & 0x3f));
  35                         putf[k++] = static_cast<char>(0x80 | (uch & 0x3f));
  36                 }
  37         }
  38         putf[len] = '\0';
  39 }
  40
  41 unsigned int UCS2Length(const char *s, unsigned int len) {
  42         unsigned int ulen = 0;
  43         for (unsigned int i=0;i<len;i++) {
  44                 unsigned char ch = static_cast<unsigned char>(s[i]);
  45                 if ((ch < 0x80) || (ch > (0x80 + 0x40)))
  46                         ulen++;
  47         }
  48         return ulen;
  49 }
  50
  51 unsigned int UCS2FromUTF8(const char *s, unsigned int len, wchar_t *tbuf, unsigned int tlen) {
  52 #ifdef USE_API
  53         return ::MultiByteToWideChar(CP_UTF8, 0, s, len, tbuf, tlen);
  54 #else
  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 #endif
  77 }