/* int utf8( in, out[6] );
*
* in is character code 0x0 -> 0x7FFFFFFF
* int is number of characters for output
*
*/
int
utf8_encode( unsigned int value, unsigned char out[6] )
{
int i;
for ( i=1; i<6; ++i ) out[i] = 0x80; /* 10xxxxxx */
if ( value < 0x80 ) {
out[0] = 0x0; /* 0xxxxxxx */
utf8_build( value, out, 25, 1 );
return 1;
} else if ( value < 0x800 ) {
out[0] = 0xC0; /* 110xxxxx */
utf8_build( value, out, 21, 3 );
return 2;
} else if ( value < 0x10000 ) {
out[0] = 0xE0; /* 1110xxxx */
utf8_build( value, out, 16, 4 );
return 3;
} else if ( value < 0x200000 ) {
out[0] = 0xF0; /* 11110xxx */
utf8_build( value, out, 11, 5 );
return 4;
} else if ( value < 0x4000000 ) {
out[0] = 0xF8; /* 111110xx */
utf8_build( value, out, 6, 6 );
return 5;
} else if ( value < (unsigned int ) 0x80000000 ) {
out[0] = 0xFC; /* 1111110x */
utf8_build( value, out, 1, 7 );
return 6;
} else {
/* error, above 2^31 bits encodable by UTF-8 */
return 0;
}
}
/* Lookup width of UTF-8 data in tree. */
u_int
utf8_width(u_int uc)
{
struct utf8_width_entry *item;
if (utf8_width_root == NULL)
utf8_build();
item = utf8_width_root;
while (item != NULL) {
if (uc < item->first)
item = item->left;
else if (uc > item->last)
item = item->right;
else
return (item->width);
}
return (1);
}