/* doop.c
*
- * Copyright (c) 1991-2000, Larry Wall
+ * Copyright (c) 1991-2001, Larry Wall
*
* You may distribute under the terms of either the GNU General Public
* License or the Artistic License, as specified in the README file.
U8 *send;
U8 *dstart;
I32 matches = 0;
+ I32 grows = PL_op->op_private & OPpTRANS_GROWS;
STRLEN len;
short *tbl;
I32 ch;
}
/* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
- Newz(0, d, len*2+1, U8);
+ if (grows)
+ New(0, d, len*2+1, U8);
+ else
+ d = s;
dstart = d;
while (s < send) {
STRLEN ulen;
- short c;
+ UV c;
- ulen = 1;
/* Need to check this, otherwise 128..255 won't match */
- c = utf8_to_uv(s, send - s, &ulen, 0);
- if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
+ c = utf8n_to_uvchr(s, send - s, &ulen, 0);
+ if (c < 0x100 && (ch = tbl[c]) >= 0) {
matches++;
- if (ch < 0x80)
- *d++ = ch;
- else
- d = uv_to_utf8(d,ch);
+ d = uvchr_to_utf8(d, ch);
s += ulen;
}
else { /* No match -> copy */
- while (ulen--)
- *d++ = *s++;
+ Copy(s, d, ulen, U8);
+ d += ulen;
+ s += ulen;
}
}
- *d = '\0';
- sv_setpvn(sv, (char*)dstart, d - dstart);
+ if (grows) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ }
+ else {
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
+ }
SvUTF8_on(sv);
SvSETMAGIC(sv);
return matches;
I32 matches = 0;
STRLEN len;
short *tbl;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
tbl = (short*)cPVOP->op_pv;
if (!tbl)
while (s < send) {
UV c;
STRLEN ulen;
- c = utf8_to_uv(s, send - s, &ulen, 0);
- if (c < 0x100 && tbl[c] >= 0)
+ c = utf8n_to_uvchr(s, send - s, &ulen, 0);
+ if (c < 0x100) {
+ if (tbl[c] >= 0)
+ matches++;
+ } else if (complement)
matches++;
s += ulen;
}
U8 *dstart;
I32 isutf8;
I32 matches = 0;
- STRLEN len;
+ I32 grows = PL_op->op_private & OPpTRANS_GROWS;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
+ I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ STRLEN len, rlen;
short *tbl;
I32 ch;
SvCUR_set(sv, d - dstart);
}
else { /* isutf8 */
- Newz(0, d, len*2+1, U8);
+ if (grows)
+ New(0, d, len*2+1, U8);
+ else
+ d = s;
dstart = d;
+ if (complement && !del)
+ rlen = tbl[0x100];
+
+#ifdef MACOS_TRADITIONAL
+#define comp CoMP /* "comp" is a keyword in some compilers ... */
+#endif
if (PL_op->op_private & OPpTRANS_SQUASH) {
- U8* p = send;
UV pch = 0xfeedface;
while (s < send) {
STRLEN len;
- UV comp = utf8_to_uv_simple(s, &len);
+ UV comp = utf8_to_uvchr(s, &len);
- if (comp > 0xff)
- d = uv_to_utf8(d, comp); /* always unmapped */
+ if (comp > 0xff) {
+ if (!complement) {
+ Copy(s, d, len, U8);
+ d += len;
+ }
+ else {
+ matches++;
+ if (!del) {
+ ch = (comp - 0x100 < rlen) ?
+ tbl[comp+1] : tbl[0x100+rlen];
+ if (ch != pch) {
+ d = uvchr_to_utf8(d, ch);
+ pch = ch;
+ }
+ s += len;
+ continue;
+ }
+ }
+ }
else if ((ch = tbl[comp]) >= 0) {
matches++;
if (ch != pch) {
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
pch = ch;
}
s += len;
continue;
}
- else if (ch == -1) /* -1 is unmapped character */
- d = uv_to_utf8(d, comp);
+ else if (ch == -1) { /* -1 is unmapped character */
+ Copy(s, d, len, U8);
+ d += len;
+ }
else if (ch == -2) /* -2 is delete character */
matches++;
s += len;
else {
while (s < send) {
STRLEN len;
- UV comp = utf8_to_uv_simple(s, &len);
- if (comp > 0xff)
- d = uv_to_utf8(d, comp); /* always unmapped */
+ UV comp = utf8_to_uvchr(s, &len);
+ if (comp > 0xff) {
+ if (!complement) {
+ Copy(s, d, len, U8);
+ d += len;
+ }
+ else {
+ matches++;
+ if (!del) {
+ if (comp - 0x100 < rlen)
+ d = uvchr_to_utf8(d, tbl[comp+1]);
+ else
+ d = uvchr_to_utf8(d, tbl[0x100+rlen]);
+ }
+ }
+ }
else if ((ch = tbl[comp]) >= 0) {
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
matches++;
}
else if (ch == -1) { /* -1 is unmapped character */
- d = uv_to_utf8(d, comp);
+ Copy(s, d, len, U8);
+ d += len;
}
else if (ch == -2) /* -2 is delete character */
matches++;
s += len;
}
}
- *d = '\0';
- sv_setpvn(sv, (char*)dstart, d - dstart);
+ if (grows) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ }
+ else {
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
+ }
SvUTF8_on(sv);
}
SvSETMAGIC(sv);
U8 *start;
U8 *dstart, *dend;
I32 matches = 0;
+ I32 grows = PL_op->op_private & OPpTRANS_GROWS;
STRLEN len;
SV* rv = (SV*)cSVOP->op_sv;
if (!isutf8) {
U8 *t = s, *e = s + len;
while (t < e)
- if ((hibit = *t++ & 0x80))
+ if ((hibit = UTF8_IS_CONTINUED(*t++)))
break;
if (hibit)
s = bytes_to_utf8(s, &len);
if (svp)
final = SvUV(*svp);
- /* d needs to be bigger than s, in case e.g. upgrading is required */
- New(0, d, len*3+UTF8_MAXLEN, U8);
- dend = d + len * 3;
- dstart = d;
+ if (grows) {
+ /* d needs to be bigger than s, in case e.g. upgrading is required */
+ New(0, d, len*3+UTF8_MAXLEN, U8);
+ dend = d + len * 3;
+ dstart = d;
+ }
+ else {
+ dstart = d = s;
+ dend = d + len;
+ }
while (s < send) {
if ((uv = swash_fetch(rv, s)) < none) {
s += UTF8SKIP(s);
matches++;
- d = uv_to_utf8(d, uv);
+ d = uvchr_to_utf8(d, uv);
}
else if (uv == none) {
int i = UTF8SKIP(s);
- while(i--)
- *d++ = *s++;
+ Copy(s, d, i, U8);
+ d += i;
+ s += i;
}
else if (uv == extra) {
int i = UTF8SKIP(s);
s += i;
matches++;
- d = uv_to_utf8(d, final);
+ d = uvchr_to_utf8(d, final);
}
else
s += UTF8SKIP(s);
- if (d >= dend) {
+ if (d > dend) {
STRLEN clen = d - dstart;
STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
Renew(dstart, nlen+UTF8_MAXLEN, U8);
d = dstart + clen;
dend = dstart + nlen;
}
}
- *d = '\0';
- sv_setpvn(sv, (char*)dstart, d - dstart);
+ if (grows || hibit) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ if (grows && hibit)
+ Safefree(start);
+ }
+ else {
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
+ }
SvSETMAGIC(sv);
SvUTF8_on(sv);
- if (hibit)
- Safefree(start);
+ /* Downgrading just 'cos it will is suspect - NI-S */
if (!isutf8 && !(PL_hints & HINT_UTF8))
sv_utf8_downgrade(sv, TRUE);
if (!SvUTF8(sv)) {
U8 *t = s, *e = s + len;
while (t < e)
- if ((hibit = *t++ & 0x80))
+ if ((hibit = !UTF8_IS_ASCII(*t++)))
break;
if (hibit)
start = s = bytes_to_utf8(s, &len);
I32 matches = 0;
I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ I32 grows = PL_op->op_private & OPpTRANS_GROWS;
SV* rv = (SV*)cSVOP->op_sv;
HV* hv = (HV*)SvRV(rv);
SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
if (!isutf8) {
U8 *t = s, *e = s + len;
while (t < e)
- if ((hibit = *t++ & 0x80))
+ if ((hibit = !UTF8_IS_ASCII(*t++)))
break;
if (hibit)
s = bytes_to_utf8(s, &len);
if (svp)
final = SvUV(*svp);
- New(0, d, len*3+UTF8_MAXLEN, U8);
- dend = d + len * 3;
- dstart = d;
+ if (grows) {
+ /* d needs to be bigger than s, in case e.g. upgrading is required */
+ New(0, d, len*3+UTF8_MAXLEN, U8);
+ dend = d + len * 3;
+ dstart = d;
+ }
+ else {
+ dstart = d = s;
+ dend = d + len;
+ }
if (squash) {
UV puv = 0xfeedface;
while (s < send) {
uv = swash_fetch(rv, s);
-
- if (d >= dend) {
- STRLEN clen = d - dstart, nlen = dend - dstart + len;
+
+ if (d > dend) {
+ STRLEN clen = d - dstart;
+ STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
Renew(dstart, nlen+UTF8_MAXLEN, U8);
d = dstart + clen;
dend = dstart + nlen;
}
if (uv < none) {
matches++;
+ s += UTF8SKIP(s);
if (uv != puv) {
- d = uv_to_utf8(d, uv);
+ d = uvchr_to_utf8(d, uv);
puv = uv;
}
- s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
int i = UTF8SKIP(s);
- while(i--)
- *d++ = *s++;
+ Copy(s, d, i, U8);
+ d += i;
+ s += i;
puv = 0xfeedface;
continue;
}
else if (uv == extra && !del) {
matches++;
+ s += UTF8SKIP(s);
if (uv != puv) {
- d = uv_to_utf8(d, final);
+ d = uvchr_to_utf8(d, final);
puv = final;
}
- s += UTF8SKIP(s);
continue;
}
matches++; /* "none+1" is delete character */
else {
while (s < send) {
uv = swash_fetch(rv, s);
- if (d >= dend) {
- STRLEN clen = d - dstart, nlen = dend - dstart + len;
+ if (d > dend) {
+ STRLEN clen = d - dstart;
+ STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
Renew(dstart, nlen+UTF8_MAXLEN, U8);
d = dstart + clen;
dend = dstart + nlen;
}
if (uv < none) {
matches++;
- d = uv_to_utf8(d, uv);
s += UTF8SKIP(s);
+ d = uvchr_to_utf8(d, uv);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
int i = UTF8SKIP(s);
- while(i--)
- *d++ = *s++;
+ Copy(s, d, i, U8);
+ d += i;
+ s += i;
continue;
}
else if (uv == extra && !del) {
matches++;
- d = uv_to_utf8(d, final);
s += UTF8SKIP(s);
+ d = uvchr_to_utf8(d, final);
continue;
}
matches++; /* "none+1" is delete character */
s += UTF8SKIP(s);
}
}
- *d = '\0';
- sv_setpvn(sv, (char*)dstart, d - dstart);
+ if (grows || hibit) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ if (grows && hibit)
+ Safefree(start);
+ }
+ else {
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
+ }
SvUTF8_on(sv);
- if (hibit)
- Safefree(start);
if (!isutf8 && !(PL_hints & HINT_UTF8))
sv_utf8_downgrade(sv, TRUE);
SvSETMAGIC(sv);
char *send = s + len;
char *start = s;
s = send - 1;
- while ((*s & 0xc0) == 0x80)
- --s;
- if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
- Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
- sv_setpvn(astr, s, send - s);
- *s = '\0';
- SvCUR_set(sv, s - start);
- SvNIOK_off(sv);
- SvUTF8_on(astr);
+ while (s > start && UTF8_IS_CONTINUATION(*s))
+ s--;
+ if (utf8_to_uvchr((U8*)s, 0)) {
+ sv_setpvn(astr, s, send - s);
+ *s = '\0';
+ SvCUR_set(sv, s - start);
+ SvNIOK_off(sv);
+ SvUTF8_on(astr);
+ }
}
else
sv_setpvn(astr, "", 0);
switch (optype) {
case OP_BIT_AND:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc & ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
if (sv == left || sv == right)
(void)sv_usepvn(sv, dcsave, needlen);
break;
case OP_BIT_XOR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc ^ ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
goto mop_up_utf;
case OP_BIT_OR:
while (lulen && rulen) {
- luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
lc += ulen;
lulen -= ulen;
- ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
rc += ulen;
rulen -= ulen;
duc = luc | ruc;
- dc = (char*)uv_to_utf8((U8*)dc, duc);
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
mop_up_utf:
if (sv == left || sv == right)
OP *
Perl_do_kv(pTHX)
{
- djSP;
+ dSP;
HV *hv = (HV*)POPs;
HV *keys;
register HE *entry;
dokeys = dovalues = TRUE;
if (!hv) {
- if (PL_op->op_flags & OPf_MOD) { /* lvalue */
+ if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
dTARGET; /* make sure to clear its target here */
if (SvTYPE(TARG) == SVt_PVLV)
LvTARG(TARG) = Nullsv;
IV i;
dTARGET;
- if (PL_op->op_flags & OPf_MOD) { /* lvalue */
+ if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
if (SvTYPE(TARG) < SVt_PVLV) {
sv_upgrade(TARG, SVt_PVLV);
sv_magic(TARG, Nullsv, 'k', Nullch, 0);