1
// Scintilla source code edit control
2
/** @file UniConversion.cxx
3
** Functions to handle UFT-8 and UCS-2 strings.
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.
10
#include "UniConversion.h"
12
unsigned int UTF8Length(const wchar_t *uptr, unsigned int tlen) {
14
for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
15
unsigned int uch = uptr[i];
26
void UTF8FromUCS2(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned int len) {
28
for (unsigned int i = 0; i < tlen && uptr[i]; i++) {
29
unsigned int uch = uptr[i];
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));
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));
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)))
54
unsigned int UCS2FromUTF8(const char *s, unsigned int len, wchar_t *tbuf, unsigned int tlen) {
56
const unsigned char *us = reinterpret_cast<const unsigned char *>(s);
58
while ((i<len) && (ui<tlen)) {
59
unsigned char ch = us[i++];
62
} else if (ch < 0x80 + 0x40 + 0x20) {
63
tbuf[ui] = static_cast<wchar_t>((ch & 0x1F) << 6);
65
tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + (ch & 0x7F));
67
tbuf[ui] = static_cast<wchar_t>((ch & 0xF) << 12);
69
tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + ((ch & 0x7F) << 6));
71
tbuf[ui] = static_cast<wchar_t>(tbuf[ui] + (ch & 0x7F));