static int jl_id_start_char(uint32_t wc)
{
if ((wc >= 'A' && wc <= 'Z') || (wc >= 'a' && wc <= 'z') || wc == '_')
return 1;
if (wc < 0xA1 || wc > 0x10ffff)
return 0;
const utf8proc_property_t *prop = utf8proc_get_property(wc);
return is_wc_cat_id_start(wc, prop->category);
}
static void
test_utf8proc_get_property ( void )
{
int32_t unicode_char;
const utf8proc_property_t* properties;
for ( unicode_char = 0x0000; unicode_char <= 0x10FFFF; ++unicode_char ) {
properties = utf8proc_get_property ( unicode_char );
rtems_test_assert ( NULL != properties );
}
}
static int jl_id_char(uint32_t wc)
{
if ((wc >= 'A' && wc <= 'Z') || (wc >= 'a' && wc <= 'z') || wc == '_' ||
(wc >= '0' && wc <= '9') || wc == '!')
return 1;
if (wc < 0xA1 || wc > 0x10ffff)
return 0;
const utf8proc_property_t *prop = utf8proc_get_property(wc);
utf8proc_propval_t cat = prop->category;
if (is_wc_cat_id_start(wc, cat)) return 1;
if (cat == UTF8PROC_CATEGORY_MN || cat == UTF8PROC_CATEGORY_MC ||
cat == UTF8PROC_CATEGORY_ND || cat == UTF8PROC_CATEGORY_PC ||
cat == UTF8PROC_CATEGORY_SK || cat == UTF8PROC_CATEGORY_ME ||
cat == UTF8PROC_CATEGORY_NO ||
// primes
(wc >= 0x2032 && wc <= 0x2034) ||
// Other_ID_Continue
wc == 0x0387 || wc == 0x19da || (wc >= 0x1369 && wc <= 0x1371))
return 1;
return 0;
}
unsigned int utf8_glypheme_length(const char *p)
{
int32_t cp;
int len = utf8proc_iterate((uint8_t *) p, -1, &cp);
if (len == UTF8PROC_ERROR_INVALIDUTF8)
return 1;
// Check for combining characters
const char *p2 = p + len;
while (*p2)
{
int comb_len = utf8proc_iterate((uint8_t *) p2, -1, &cp);
if (comb_len == UTF8PROC_ERROR_INVALIDUTF8)
return 1;
const utf8proc_property_t* prop = utf8proc_get_property(cp);
if (!prop->combining_class)
break;
len += comb_len;
p2 += comb_len;
}
return len;
}