int
sms_utf8_from_message_str( const char* str, int strlen, unsigned char* utf8, int utf8len )
{
cbytes_t p = (cbytes_t)str;
cbytes_t end = p + strlen;
int count = 0;
int escaped = 0;
while (p < end)
{
int c = p[0];
/* read the value from the string */
p += 1;
if (c >= 128) {
if ((c & 0xe0) == 0xc0)
c &= 0x1f;
else if ((c & 0xf0) == 0xe0)
c &= 0x0f;
else
c &= 0x07;
p++;
while (p < end && (p[0] & 0xc0) == 0x80) {
c = (c << 6) | (p[0] & 0x3f);
p++;
}
}
if (escaped) {
switch (c) {
case '\\':
break;
case 'n': /* \n is line feed */
c = 10;
break;
case 'x': /* \xNN, where NN is a 2-digit hexadecimal value */
if (p+2 > end)
return -1;
c = gsm_hex2_to_byte( (const char*)p );
if (c < 0)
return -1;
p += 2;
break;
case 'u': /* \uNNNN where NNNN is a 4-digiti hexadecimal value */
if (p + 4 > end)
return -1;
c = gsm_hex4_to_short( (const char*)p );
if (c < 0)
return -1;
p += 4;
break;
default: /* invalid escape, return -1 */
return -1;
}
escaped = 0;
}
else if (c == '\\')
{
escaped = 1;
continue;
}
/* now, try to write it to the destination */
if (c < 128) {
if (count < utf8len)
utf8[count] = (byte_t) c;
count += 1;
}
else if (c < 0x800) {
if (count < utf8len)
utf8[count] = (byte_t)(0xc0 | ((c >> 6) & 0x1f));
if (count+1 < utf8len)
utf8[count+1] = (byte_t)(0x80 | (c & 0x3f));
count += 2;
}
else {
if (count < utf8len)
utf8[count] = (byte_t)(0xc0 | ((c >> 12) & 0xf));
if (count+1 < utf8len)
utf8[count+1] = (byte_t)(0x80 | ((c >> 6) & 0x3f));
if (count+2 < utf8len)
utf8[count+2] = (byte_t)(0x80 | (c & 0x3f));
count += 3;
}
}
int
sms_utf8_from_message_str( const char* str, int strlen, unsigned char* utf8, int utf8len )
{
cbytes_t p = (cbytes_t)str;
cbytes_t end = p + strlen;
int count = 0;
int escaped = 0;
while (p < end)
{
int c = p[0];
p += 1;
if (c >= 128) {
if ((c & 0xe0) == 0xc0)
c &= 0x1f;
else if ((c & 0xf0) == 0xe0)
c &= 0x0f;
else
c &= 0x07;
p++;
while (p < end && (p[0] & 0xc0) == 0x80) {
c = (c << 6) | (p[0] & 0x3f);
p++;
}
}
if (escaped) {
switch (c) {
case '\\':
break;
case 'n':
c = 10;
break;
case 'x':
if (p+2 > end)
return -1;
c = gsm_hex2_to_byte( (const char*)p );
if (c < 0)
return -1;
p += 2;
break;
case 'u':
if (p + 4 > end)
return -1;
c = gsm_hex4_to_short( (const char*)p );
if (c < 0)
return -1;
p += 4;
break;
default:
return -1;
}
escaped = 0;
}
else if (c == '\\')
{
escaped = 1;
continue;
}
if (c < 128) {
if (count < utf8len)
utf8[count] = (byte_t) c;
count += 1;
}
else if (c < 0x800) {
if (count < utf8len)
utf8[count] = (byte_t)(0xc0 | ((c >> 6) & 0x1f));
if (count+1 < utf8len)
utf8[count+1] = (byte_t)(0x80 | (c & 0x3f));
count += 2;
}
else {
if (count < utf8len)
utf8[count] = (byte_t)(0xc0 | ((c >> 12) & 0xf));
if (count+1 < utf8len)
utf8[count+1] = (byte_t)(0x80 | ((c >> 6) & 0x3f));
if (count+2 < utf8len)
utf8[count+2] = (byte_t)(0x80 | (c & 0x3f));
count += 3;
}
}
