int unicode_lb_lookup(unicode_char ch)
{
return unicode_tab_lookup(ch,
unicode_indextab,
sizeof(unicode_indextab)
/ sizeof(unicode_indextab[0]),
unicode_rangetab,
unicode_classtab,
UNICODE_LB_AL /* XX, LB1 */);
}
int unicode_grapheme_break(unicode_char a, unicode_char b)
{
uint8_t ac=unicode_tab_lookup(a, unicode_indextab,
sizeof(unicode_indextab)/sizeof(unicode_indextab[0]),
unicode_rangetab,
unicode_classtab,
UNICODE_GRAPHEMEBREAK_ANY),
bc=unicode_tab_lookup(b, unicode_indextab,
sizeof(unicode_indextab)/sizeof(unicode_indextab[0]),
unicode_rangetab,
unicode_classtab,
UNICODE_GRAPHEMEBREAK_ANY);
/* GB1 and GB2 are implied */
if (ac == UNICODE_GRAPHEMEBREAK_CR && bc == UNICODE_GRAPHEMEBREAK_LF)
return 0; /* GB3 */
switch (ac) {
case UNICODE_GRAPHEMEBREAK_CR:
case UNICODE_GRAPHEMEBREAK_LF:
case UNICODE_GRAPHEMEBREAK_Control:
return 1; /* GB4 */
default:
break;
}
switch (bc) {
case UNICODE_GRAPHEMEBREAK_CR:
case UNICODE_GRAPHEMEBREAK_LF:
case UNICODE_GRAPHEMEBREAK_Control:
return 1; /* GB5 */
default:
break;
}
if (ac == UNICODE_GRAPHEMEBREAK_L)
switch (bc) {
case UNICODE_GRAPHEMEBREAK_L:
case UNICODE_GRAPHEMEBREAK_V:
case UNICODE_GRAPHEMEBREAK_LV:
case UNICODE_GRAPHEMEBREAK_LVT:
return 0; /* GB6 */
}
if ((ac == UNICODE_GRAPHEMEBREAK_LV ||
ac == UNICODE_GRAPHEMEBREAK_V) &&
(bc == UNICODE_GRAPHEMEBREAK_V ||
bc == UNICODE_GRAPHEMEBREAK_T))
return 0; /* GB7 */
if ((ac == UNICODE_GRAPHEMEBREAK_LVT ||
ac == UNICODE_GRAPHEMEBREAK_T) &&
bc == UNICODE_GRAPHEMEBREAK_T)
return 0; /* GB8 */
if (bc == UNICODE_GRAPHEMEBREAK_Extend)
return 0; /* GB9 */
if (bc == UNICODE_GRAPHEMEBREAK_SpacingMark)
return 0; /* GB9a */
if (ac == UNICODE_GRAPHEMEBREAK_Prepend)
return 0; /* GB9b */
return 1; /* GB10 */
}
