/*
* Wordch tells whether character at *wc is a word character
* i.e. an alfa, digit, or underscore.
*/
int
wordch(char *wc)
{
int c;
#ifdef MB
if (mb_cur_max > 0 && *wc & 0200) {
mbtowi(&c, wc, mb_cur_max);
if (c & INVBIT)
return 1;
} else
#endif
c = wc[0]&0377;
return (xisalnum(c) || c == '_'
#ifdef BIT8
#ifdef ISO8859_1
/*
* We consider all ISO 8859-1 characters except for
* no-break-space as word characters.
*/
|| c&0200 && (!(c"E) && (c&TRIM) != 0240)
#endif
#endif
);
}
static int
cblank(char *cp)
{
if (mb_cur_max > 1 && *cp & 0200) {
int c;
return mbtowi(&c, cp, mb_cur_max) > 0 && iswspace(c);
} else
return isspace(*cp&0377);
}
int
wsamechar(char *cp, int d)
{
int c;
if (mbtowi(&c, cp, mb_cur_max) >= 0 && c == d)
return 1;
return 0;
}
/*
* Range search for multibyte locales using the modified UNIX(R) Regular
* Expression Library DFA.
*/
static int
rc_rangew(struct iblok *ip, char *last)
{
char *p;
int n, cstat, nstat;
wint_t wc;
Dfa *dp = e0->e_exp->re_dfa;
p = ip->ib_cur;
lineno++;
cstat = dp->anybol;
if (dp->acc[cstat])
goto found;
for (;;) {
if (*p & 0200) {
if ((n = mbtowi(&wc, p, last + 1 - p)) < 0) {
n = 1;
wc = WEOF;
}
} else {
wc = *p;
n = 1;
}
if ((wc & ~(wchar_t)(NCHAR-1)) != 0 ||
(nstat = dp->trans[cstat][wc]) == 0) {
/*
* '\0' is used to indicate end-of-line. If a '\0'
* character appears in input, it matches '$' but
* the DFA remains in dead state afterwards; there
* is thus no need to handle this condition
* specially to get the same behavior as in plain
* regexec().
*/
if (wc == '\n')
wc = '\0';
if ((nstat = regtrans(dp, cstat, wc, mb_cur_max)) == 0)
goto fail;
dp->trans[cstat]['\n'] = dp->trans[cstat]['\0'];
}
if (dp->acc[cstat = nstat - 1]) {
found: for (;;) {
if (vflag == 0) {
succeed: outline(ip, last, p - ip->ib_cur);
if (qflag || lflag)
return 1;
} else {
fail: ip->ib_cur = p;
while (*ip->ib_cur++ != '\n');
}
if ((p = ip->ib_cur) > last)
return 0;
lineno++;
if (dp->acc[cstat = dp->anybol] == 0)
goto brk2;
}
}
p += n;
if (p[-n] == '\n') {
if (vflag) {
p--;
goto succeed;
}
if ((ip->ib_cur = p) > last)
return 0;
lineno++;
if (dp->acc[cstat = dp->anybol])
goto found;
}
brk2:;
}
}