tbl = (short*)cPVOP->op_pv;
if (!tbl)
- Perl_croak(aTHX_ "panic: do_trans_simple");
+ Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
s = (U8*)SvPV(sv, len);
send = s + len;
UV c;
/* Need to check this, otherwise 128..255 won't match */
- c = utf8_to_uv(s, send - s, &ulen, 0);
+ c = utf8n_to_uvchr(s, send - s, &ulen, 0);
if (c < 0x100 && (ch = tbl[c]) >= 0) {
matches++;
- d = uv_to_utf8(d, ch);
+ d = uvchr_to_utf8(d, ch);
s += ulen;
}
else { /* No match -> copy */
}
STATIC I32
-S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
+S_do_trans_count(pTHX_ SV *sv)
{
U8 *s;
U8 *send;
I32 matches = 0;
STRLEN len;
short *tbl;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
tbl = (short*)cPVOP->op_pv;
if (!tbl)
- Perl_croak(aTHX_ "panic: do_trans_count");
+ Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
s = (U8*)SvPV(sv, len);
send = s + len;
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;
}
}
STATIC I32
-S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
+S_do_trans_complex(pTHX_ SV *sv)
{
U8 *s;
U8 *send;
I32 isutf8;
I32 matches = 0;
I32 grows = PL_op->op_private & OPpTRANS_GROWS;
- STRLEN len;
+ I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
+ I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ STRLEN len, rlen = 0;
short *tbl;
I32 ch;
tbl = (short*)cPVOP->op_pv;
if (!tbl)
- Perl_croak(aTHX_ "panic: do_trans_complex");
+ Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
s = (U8*)SvPV(sv, len);
isutf8 = SvUTF8(sv);
s++;
}
}
+ *d = '\0';
SvCUR_set(sv, d - dstart);
}
else { /* isutf8 */
else
d = s;
dstart = d;
+ if (complement && !del)
+ rlen = tbl[0x100];
#ifdef MACOS_TRADITIONAL
#define comp CoMP /* "comp" is a keyword in some compilers ... */
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) { /* always unmapped */
- Copy(s, d, len, U8);
- d += len;
+ if (comp > 0xff) {
+ if (!complement) {
+ Copy(s, d, len, U8);
+ d += len;
+ }
+ else {
+ matches++;
+ if (!del) {
+ ch = (rlen == 0) ? comp :
+ (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;
else {
while (s < send) {
STRLEN len;
- UV comp = utf8_to_uv_simple(s, &len);
- if (comp > 0xff) { /* always unmapped */
- Copy(s, d, len, U8);
- d += len;
+ 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 */
}
STATIC I32
-S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
+S_do_trans_simple_utf8(pTHX_ SV *sv)
{
U8 *s;
U8 *send;
SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
UV none = svp ? SvUV(*svp) : 0x7fffffff;
UV extra = none + 1;
- UV final;
+ UV final = 0;
UV uv;
I32 isutf8;
U8 hibit = 0;
isutf8 = SvUTF8(sv);
if (!isutf8) {
U8 *t = s, *e = s + len;
- while (t < e)
- if ((hibit = UTF8_IS_CONTINUED(*t++)))
+ while (t < e) {
+ U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
break;
+ }
if (hibit)
s = bytes_to_utf8(s, &len);
}
}
while (s < send) {
- if ((uv = swash_fetch(rv, s)) < none) {
+ if ((uv = swash_fetch(rv, s, TRUE)) < none) {
s += UTF8SKIP(s);
matches++;
- d = uv_to_utf8(d, uv);
+ d = uvuni_to_utf8(d, uv);
}
else if (uv == none) {
int i = UTF8SKIP(s);
int i = UTF8SKIP(s);
s += i;
matches++;
- d = uv_to_utf8(d, final);
+ d = uvuni_to_utf8(d, final);
}
else
s += UTF8SKIP(s);
STRLEN clen = d - dstart;
STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
if (!grows)
- Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
+ Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
Renew(dstart, nlen+UTF8_MAXLEN, U8);
d = dstart + clen;
dend = dstart + nlen;
}
SvSETMAGIC(sv);
SvUTF8_on(sv);
- if (!isutf8 && !(PL_hints & HINT_UTF8))
- sv_utf8_downgrade(sv, TRUE);
return matches;
}
STATIC I32
-S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
+S_do_trans_count_utf8(pTHX_ SV *sv)
{
U8 *s;
- U8 *start, *send;
+ U8 *start = 0, *send;
I32 matches = 0;
STRLEN len;
HV* hv = (HV*)SvRV(rv);
SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ UV extra = none + 1;
UV uv;
U8 hibit = 0;
s = (U8*)SvPV(sv, len);
if (!SvUTF8(sv)) {
U8 *t = s, *e = s + len;
- while (t < e)
- if ((hibit = !UTF8_IS_ASCII(*t++)))
+ while (t < e) {
+ U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
break;
+ }
if (hibit)
start = s = bytes_to_utf8(s, &len);
}
send = s + len;
while (s < send) {
- if ((uv = swash_fetch(rv, s)) < none)
+ if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra)
matches++;
s += UTF8SKIP(s);
}
}
STATIC I32
-S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
+S_do_trans_complex_utf8(pTHX_ SV *sv)
{
U8 *s;
U8 *start, *send;
SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
UV none = svp ? SvUV(*svp) : 0x7fffffff;
UV extra = none + 1;
- UV final;
+ UV final = 0;
+ bool havefinal = FALSE;
UV uv;
STRLEN len;
U8 *dstart, *dend;
isutf8 = SvUTF8(sv);
if (!isutf8) {
U8 *t = s, *e = s + len;
- while (t < e)
- if ((hibit = !UTF8_IS_ASCII(*t++)))
+ while (t < e) {
+ U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
break;
+ }
if (hibit)
s = bytes_to_utf8(s, &len);
}
start = s;
svp = hv_fetch(hv, "FINAL", 5, FALSE);
- if (svp)
+ if (svp) {
final = SvUV(*svp);
+ havefinal = TRUE;
+ }
if (grows) {
/* d needs to be bigger than s, in case e.g. upgrading is required */
if (squash) {
UV puv = 0xfeedface;
while (s < send) {
- uv = swash_fetch(rv, s);
-
+ uv = swash_fetch(rv, s, TRUE);
+
if (d > dend) {
STRLEN clen = d - dstart;
STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
if (!grows)
- Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
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 = uvuni_to_utf8(d, uv);
puv = uv;
}
- s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
}
else if (uv == extra && !del) {
matches++;
- if (uv != puv) {
- d = uv_to_utf8(d, final);
- puv = final;
+ if (havefinal) {
+ s += UTF8SKIP(s);
+ if (puv != final) {
+ d = uvuni_to_utf8(d, final);
+ puv = final;
+ }
+ }
+ else {
+ STRLEN len;
+ uv = utf8_to_uvuni(s, &len);
+ if (uv != puv) {
+ Copy(s, d, len, U8);
+ d += len;
+ puv = uv;
+ }
+ s += len;
}
- s += UTF8SKIP(s);
continue;
}
matches++; /* "none+1" is delete character */
}
else {
while (s < send) {
- uv = swash_fetch(rv, s);
+ uv = swash_fetch(rv, s, TRUE);
if (d > dend) {
STRLEN clen = d - dstart;
STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
if (!grows)
- Perl_croak(aTHX_ "panic: do_trans_complex_utf8");
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
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 = uvuni_to_utf8(d, uv);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
}
else if (uv == extra && !del) {
matches++;
- d = uv_to_utf8(d, final);
s += UTF8SKIP(s);
+ d = uvuni_to_utf8(d, final);
continue;
}
matches++; /* "none+1" is delete character */
SvCUR_set(sv, d - dstart);
}
SvUTF8_on(sv);
- if (!isutf8 && !(PL_hints & HINT_UTF8))
- sv_utf8_downgrade(sv, TRUE);
SvSETMAGIC(sv);
return matches;
return do_trans_simple(sv);
case OPpTRANS_IDENTICAL:
+ case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
if (hasutf)
return do_trans_count_utf8(sv);
else
register I32 items = sp - mark;
register STRLEN len;
STRLEN delimlen;
- register char *delim = SvPV(del, delimlen);
STRLEN tmplen;
+ (void) SvPV(del, delimlen); /* stringify and get the delimlen */
+ /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
+
mark++;
len = (items > 0 ? (delimlen * (items - 1) ) : 0);
(void)SvUPGRADE(sv, SVt_PV);
++mark;
}
+ sv_setpv(sv, "");
+ if (PL_tainting && SvMAGICAL(sv))
+ SvTAINTED_off(sv);
+
if (items-- > 0) {
- sv_setpv(sv, "");
if (*mark)
sv_catsv(sv, *mark);
mark++;
}
- else
- sv_setpv(sv,"");
+
if (delimlen) {
for (; items > 0; items--,mark++) {
sv_catsv(sv,del);
s = send - 1;
while (s > start && UTF8_IS_CONTINUATION(*s))
s--;
- if (utf8_to_uv_simple((U8*)s, 0)) {
+ if (utf8_to_uvchr((U8*)s, 0)) {
sv_setpvn(astr, s, send - s);
*s = '\0';
SvCUR_set(sv, s - start);
{
register I32 count;
STRLEN len;
+ STRLEN n_a;
char *s;
if (RsSNARF(PL_rs))
else if (SvREADONLY(sv))
Perl_croak(aTHX_ PL_no_modify);
s = SvPV(sv, len);
- if (len && !SvPOKp(sv))
- s = SvPV_force(sv, len);
if (s && len) {
s += --len;
if (RsPARA(PL_rs)) {
count += rslen;
}
}
- *s = '\0';
+ s = SvPV_force(sv, n_a);
SvCUR_set(sv, len);
+ *SvEND(sv) = '\0';
SvNIOK_off(sv);
+ SvSETMAGIC(sv);
}
nope:
- SvSETMAGIC(sv);
return count;
}
char *rsave;
bool left_utf = DO_UTF8(left);
bool right_utf = DO_UTF8(right);
- I32 needlen;
+ I32 needlen = 0;
if (left_utf && !right_utf)
sv_utf8_upgrade(right);
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)
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);
+ sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
}
LvTYPE(TARG) = 'k';
if (LvTARG(TARG) != (SV*)keys) {
RETURN;
}
- if (! SvTIED_mg((SV*)keys, 'P'))
+ if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied))
i = HvKEYS(keys);
else {
i = 0;