S_sublex_done(pTHX)
{
if (!PL_lex_starts++) {
+ SV *sv = newSVpvn("",0);
+ if (SvUTF8(PL_linestr))
+ SvUTF8_on(sv);
PL_expect = XOPERATOR;
- yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
+ yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
return THING;
}
register char *d = SvPVX(sv); /* destination for copies */
bool dorange = FALSE; /* are we in a translit range? */
bool didrange = FALSE; /* did we just finish a range? */
- bool has_utf8 = FALSE; /* embedded \x{} */
+ bool has_utf8 = (PL_linestr && SvUTF8(PL_linestr));
+ /* the constant is UTF8 */
UV uv;
I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
/* backslashes */
if (*s == '\\' && s+1 < send) {
- bool to_be_utf8 = FALSE;
-
s++;
/* some backslashes we leave behind */
else {
STRLEN len = 1; /* allow underscores */
uv = (UV)scan_hex(s + 1, e - s - 1, &len);
- if (PL_hints & HINT_UTF8)
- to_be_utf8 = TRUE;
}
s = e + 1;
}
* repertoire. --jhi */
if (uv > 127) {
- if (!has_utf8 && (to_be_utf8 || uv > 255)) {
+ if (!has_utf8 && uv > 255) {
/* Might need to recode whatever we have
* accumulated so far if it contains any
* hibit chars.
}
}
- if (to_be_utf8 || has_utf8 || uv > 255) {
+ if (has_utf8 || uv > 255) {
d = (char*)uv_to_utf8((U8*)d, uv);
has_utf8 = TRUE;
}
if (ftst) {
PL_last_lop_op = ftst;
DEBUG_T( { PerlIO_printf(Perl_debug_log,
- "### Saw file test %c\n", ftst);
+ "### Saw file test %c\n", (int)ftst);
} )
FTST(ftst);
}
/* Assume it was a minus followed by a one-letter named
* subroutine call (or a -bareword), then. */
DEBUG_T( { PerlIO_printf(Perl_debug_log,
- "### %c looked like a file test but was not\n", ftst);
+ "### %c looked like a file test but was not\n",
+ (int)ftst);
} )
s -= 2;
}
TOKEN('(');
case KEY_qq:
+ case KEY_qu:
s = scan_str(s,FALSE,FALSE);
+ if (tmp == KEY_qu && is_utf8_string((U8*)s, SvCUR(PL_lex_stuff)))
+ SvUTF8_on(PL_lex_stuff);
if (!s)
missingterm((char*)0);
yylval.ival = OP_STRINGIFY;
if (strEQ(d,"q")) return KEY_q;
if (strEQ(d,"qr")) return KEY_qr;
if (strEQ(d,"qq")) return KEY_qq;
+ if (strEQ(d,"qu")) return KEY_qu;
if (strEQ(d,"qw")) return KEY_qw;
if (strEQ(d,"qx")) return KEY_qx;
}
while (isDIGIT(*pos) || *pos == '_')
pos++;
if (!isALPHA(*pos)) {
- UV rev;
+ UV rev, revmax = 0;
U8 tmpbuf[UTF8_MAXLEN+1];
U8 *tmpend;
- bool utf8 = FALSE;
s++; /* get past 'v' */
sv = NEWSV(92,5);
}
}
tmpend = uv_to_utf8(tmpbuf, rev);
- utf8 = utf8 || rev > 127;
+ if (rev > revmax)
+ revmax = rev;
sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
if (*pos == '.' && isDIGIT(pos[1]))
s = ++pos;
SvPOK_on(sv);
SvREADONLY_on(sv);
- if (utf8) {
+ if (revmax > 127) {
SvUTF8_on(sv);
- if (!UTF||IN_BYTE)
+ if (revmax < 256)
sv_utf8_downgrade(sv, TRUE);
}
}