/* doop.c
*
* Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
- * 2000, 2001, 2002, 2004, 2005, by Larry Wall and others
+ * 2000, 2001, 2002, 2004, 2005, 2006, by Larry Wall and others
*
* You may distribute under the terms of either the GNU General Public
* License or the Artistic License, as specified in the README file.
#endif
STATIC I32
-S_do_trans_simple(pTHX_ SV *sv)
+S_do_trans_simple(pTHX_ SV * const sv)
{
- U8 *s;
- U8 *d;
- const U8 *send;
- U8 *dstart;
+ dVAR;
I32 matches = 0;
- const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
STRLEN len;
+ U8 *s = (U8*)SvPV(sv,len);
+ U8 * const send = s+len;
- const short *tbl = (short*)cPVOP->op_pv;
+ const short * const tbl = (short*)cPVOP->op_pv;
if (!tbl)
Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
- s = (U8*)SvPV(sv, len);
- send = s + len;
-
/* First, take care of non-UTF-8 input strings, because they're easy */
if (!SvUTF8(sv)) {
while (s < send) {
const I32 ch = tbl[*s];
if (ch >= 0) {
matches++;
- *s++ = (U8)ch;
+ *s = (U8)ch;
}
- else
- s++;
+ s++;
}
SvSETMAGIC(sv);
- return matches;
- }
-
- /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
- if (grows)
- Newx(d, len*2+1, U8);
- else
- d = s;
- dstart = d;
- while (s < send) {
- STRLEN ulen;
- I32 ch;
-
- /* Need to check this, otherwise 128..255 won't match */
- const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0);
- if (c < 0x100 && (ch = tbl[c]) >= 0) {
- matches++;
- d = uvchr_to_utf8(d, ch);
- s += ulen;
- }
- else { /* No match -> copy */
- Move(s, d, ulen, U8);
- d += ulen;
- s += ulen;
- }
- }
- if (grows) {
- sv_setpvn(sv, (char*)dstart, d - dstart);
- Safefree(dstart);
}
else {
- *d = '\0';
- SvCUR_set(sv, d - dstart);
+ const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
+ U8 *d;
+ U8 *dstart;
+
+ /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
+ if (grows)
+ Newx(d, len*2+1, U8);
+ else
+ d = s;
+ dstart = d;
+ while (s < send) {
+ STRLEN ulen;
+ I32 ch;
+
+ /* Need to check this, otherwise 128..255 won't match */
+ const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
+ if (c < 0x100 && (ch = tbl[c]) >= 0) {
+ matches++;
+ d = uvchr_to_utf8(d, ch);
+ s += ulen;
+ }
+ else { /* No match -> copy */
+ Move(s, d, ulen, U8);
+ d += ulen;
+ s += ulen;
+ }
+ }
+ if (grows) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ }
+ else {
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
+ }
+ SvUTF8_on(sv);
+ SvSETMAGIC(sv);
}
- SvUTF8_on(sv);
- SvSETMAGIC(sv);
return matches;
}
STATIC I32
-S_do_trans_count(pTHX_ SV *sv)
+S_do_trans_count(pTHX_ SV * const sv)
{
- const U8 *s;
- const U8 *send;
- I32 matches = 0;
+ dVAR;
STRLEN len;
- const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
+ const U8 *s = (const U8*)SvPV_const(sv, len);
+ const U8 * const send = s + len;
+ I32 matches = 0;
const short * const tbl = (short*)cPVOP->op_pv;
if (!tbl)
Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
- s = (const U8*)SvPV_const(sv, len);
- send = s + len;
-
- if (!SvUTF8(sv))
+ if (!SvUTF8(sv)) {
while (s < send) {
if (tbl[*s++] >= 0)
matches++;
}
- else
+ }
+ else {
+ const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
while (s < send) {
STRLEN ulen;
- const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0);
+ const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
if (c < 0x100) {
if (tbl[c] >= 0)
matches++;
matches++;
s += ulen;
}
+ }
return matches;
}
STATIC I32
-S_do_trans_complex(pTHX_ SV *sv)
+S_do_trans_complex(pTHX_ SV * const sv)
{
- U8 *s;
- U8 *send;
- U8 *d;
- U8 *dstart;
- I32 isutf8;
+ dVAR;
+ STRLEN len;
+ U8 *s = (U8*)SvPV(sv, len);
+ U8 * const send = s+len;
I32 matches = 0;
- const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
- const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
- const I32 del = PL_op->op_private & OPpTRANS_DELETE;
- STRLEN len, rlen = 0;
const short * const tbl = (short*)cPVOP->op_pv;
if (!tbl)
Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
- s = (U8*)SvPV(sv, len);
- isutf8 = SvUTF8(sv);
- send = s + len;
+ if (!SvUTF8(sv)) {
+ U8 *d = s;
+ U8 * const dstart = d;
- if (!isutf8) {
- dstart = d = s;
if (PL_op->op_private & OPpTRANS_SQUASH) {
const U8* p = send;
while (s < send) {
*d = '\0';
SvCUR_set(sv, d - dstart);
}
- else { /* isutf8 */
+ else { /* is utf8 */
+ const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
+ const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
+ const I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ U8 *d;
+ U8 *dstart;
+ STRLEN rlen = 0;
+
if (grows)
Newx(d, len*2+1, U8);
else
UV pch = 0xfeedface;
while (s < send) {
STRLEN len;
- const UV comp = utf8_to_uvchr(s, &len);
+ const UV comp = utf8n_to_uvchr(s, send - s, &len,
+ UTF8_ALLOW_DEFAULT);
I32 ch;
if (comp > 0xff) {
if (!complement) {
- Copy(s, d, len, U8);
+ Move(s, d, len, U8);
d += len;
}
else {
continue;
}
else if (ch == -1) { /* -1 is unmapped character */
- Copy(s, d, len, U8);
+ Move(s, d, len, U8);
d += len;
}
else if (ch == -2) /* -2 is delete character */
else {
while (s < send) {
STRLEN len;
- const UV comp = utf8_to_uvchr(s, &len);
+ const UV comp = utf8n_to_uvchr(s, send - s, &len,
+ UTF8_ALLOW_DEFAULT);
I32 ch;
if (comp > 0xff) {
if (!complement) {
matches++;
}
else if (ch == -1) { /* -1 is unmapped character */
- Copy(s, d, len, U8);
+ Move(s, d, len, U8);
d += len;
}
else if (ch == -2) /* -2 is delete character */
}
STATIC I32
-S_do_trans_simple_utf8(pTHX_ SV *sv)
+S_do_trans_simple_utf8(pTHX_ SV * const sv)
{
+ dVAR;
U8 *s;
U8 *send;
U8 *d;
SV* const rv = (SV*)cSVOP->op_sv;
HV* const hv = (HV*)SvRV(rv);
- SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
+ SV* const * svp = hv_fetchs(hv, "NONE", FALSE);
const UV none = svp ? SvUV(*svp) : 0x7fffffff;
const UV extra = none + 1;
UV final = 0;
- UV uv;
- I32 isutf8;
U8 hibit = 0;
s = (U8*)SvPV(sv, len);
- isutf8 = SvUTF8(sv);
- if (!isutf8) {
- const U8 *t = s, *e = s + len;
+ if (!SvUTF8(sv)) {
+ const U8 *t = s;
+ const U8 * const e = s + len;
while (t < e) {
const U8 ch = *t++;
- if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ hibit = !NATIVE_IS_INVARIANT(ch);
+ if (hibit) {
+ s = bytes_to_utf8(s, &len);
break;
+ }
}
- if (hibit)
- s = bytes_to_utf8(s, &len);
}
send = s + len;
start = s;
- svp = hv_fetch(hv, "FINAL", 5, FALSE);
+ svp = hv_fetchs(hv, "FINAL", FALSE);
if (svp)
final = SvUV(*svp);
}
while (s < send) {
- if ((uv = swash_fetch(rv, s, TRUE)) < none) {
+ const UV uv = swash_fetch(rv, s, TRUE);
+ if (uv < none) {
s += UTF8SKIP(s);
matches++;
d = uvuni_to_utf8(d, uv);
}
STATIC I32
-S_do_trans_count_utf8(pTHX_ SV *sv)
+S_do_trans_count_utf8(pTHX_ SV * const sv)
{
+ dVAR;
const U8 *s;
- const U8 *start = 0, *send;
+ const U8 *start = NULL;
+ const U8 *send;
I32 matches = 0;
STRLEN len;
SV* const rv = (SV*)cSVOP->op_sv;
HV* const hv = (HV*)SvRV(rv);
- SV** const svp = hv_fetch(hv, "NONE", 4, FALSE);
+ SV* const * const svp = hv_fetchs(hv, "NONE", FALSE);
const UV none = svp ? SvUV(*svp) : 0x7fffffff;
const UV extra = none + 1;
U8 hibit = 0;
s = (const U8*)SvPV_const(sv, len);
if (!SvUTF8(sv)) {
const U8 *t = s;
- const U8 *e = s + len;
+ const U8 * const e = s + len;
while (t < e) {
const U8 ch = *t++;
- if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ hibit = !NATIVE_IS_INVARIANT(ch);
+ if (hibit) {
+ start = s = bytes_to_utf8(s, &len);
break;
+ }
}
- if (hibit)
- start = s = bytes_to_utf8(s, &len);
}
send = s + len;
while (s < send) {
- UV uv;
- if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra)
+ const UV uv = swash_fetch(rv, s, TRUE);
+ if (uv < none || uv == extra)
matches++;
s += UTF8SKIP(s);
}
}
STATIC I32
-S_do_trans_complex_utf8(pTHX_ SV *sv)
+S_do_trans_complex_utf8(pTHX_ SV * const sv)
{
+ dVAR;
U8 *start, *send;
U8 *d;
I32 matches = 0;
const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
SV * const rv = (SV*)cSVOP->op_sv;
HV * const hv = (HV*)SvRV(rv);
- SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
+ SV * const *svp = hv_fetchs(hv, "NONE", FALSE);
const UV none = svp ? SvUV(*svp) : 0x7fffffff;
const UV extra = none + 1;
UV final = 0;
U8 hibit = 0;
U8 *s = (U8*)SvPV(sv, len);
- const I32 isutf8 = SvUTF8(sv);
- if (!isutf8) {
+ if (!SvUTF8(sv)) {
const U8 *t = s;
const U8 * const e = s + len;
while (t < e) {
const U8 ch = *t++;
- if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ hibit = !NATIVE_IS_INVARIANT(ch);
+ if (hibit) {
+ s = bytes_to_utf8(s, &len);
break;
+ }
}
- if (hibit)
- s = bytes_to_utf8(s, &len);
}
send = s + len;
start = s;
- svp = hv_fetch(hv, "FINAL", 5, FALSE);
+ svp = hv_fetchs(hv, "FINAL", FALSE);
if (svp) {
final = SvUV(*svp);
havefinal = TRUE;
}
else {
STRLEN len;
- uv = utf8_to_uvuni(s, &len);
+ uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT);
if (uv != puv) {
Move(s, d, len, U8);
d += len;
I32
Perl_do_trans(pTHX_ SV *sv)
{
+ dVAR;
STRLEN len;
const I32 hasutf = (PL_op->op_private &
(OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
void
Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
{
+ dVAR;
SV ** const oldmark = mark;
register I32 items = sp - mark;
register STRLEN len;
void
Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
{
+ dVAR;
STRLEN patlen;
const char * const pat = SvPV_const(*sarg, patlen);
bool do_taint = FALSE;
SvUTF8_off(sv);
if (DO_UTF8(*sarg))
SvUTF8_on(sv);
- sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
+ sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint);
SvSETMAGIC(sv);
if (do_taint)
SvTAINTED_on(sv);
UV
Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
{
+ dVAR;
STRLEN srclen, len;
const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen);
UV retnum = 0;
void
Perl_do_vecset(pTHX_ SV *sv)
{
- SV *targ = LvTARG(sv);
+ dVAR;
register I32 offset;
register I32 size;
register unsigned char *s;
I32 mask;
STRLEN targlen;
STRLEN len;
+ SV * const targ = LvTARG(sv);
if (!targ)
return;
void
Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
{
+ dVAR;
STRLEN len;
char *s;
if (SvTYPE(sv) == SVt_PVAV) {
register I32 i;
- AV* av = (AV*)sv;
+ AV* const av = (AV*)sv;
const I32 max = AvFILL(av);
for (i = 0; i <= max; i++) {
return;
}
else if (SvTYPE(sv) == SVt_PVHV) {
- HV* hv = (HV*)sv;
+ HV* const hv = (HV*)sv;
HE* entry;
(void)hv_iterinit(hv);
while ((entry = hv_iternext(hv)))
if (SvREADONLY(sv))
Perl_croak(aTHX_ PL_no_modify);
}
+
+ if (PL_encoding && !SvUTF8(sv)) {
+ /* like in do_chomp(), utf8-ize the sv as a side-effect
+ * if we're using encoding. */
+ sv_recode_to_utf8(sv, PL_encoding);
+ }
+
s = SvPV(sv, len);
if (len && !SvPOK(sv))
s = SvPV_force(sv, len);
if (DO_UTF8(sv)) {
if (s && len) {
- char *send = s + len;
- char *start = s;
+ char * const send = s + len;
+ char * const start = s;
s = send - 1;
while (s > start && UTF8_IS_CONTINUATION(*s))
s--;
I32
Perl_do_chomp(pTHX_ register SV *sv)
{
+ dVAR;
register I32 count;
STRLEN len;
char *s;
char *temp_buffer = NULL;
- SV* svrecode = Nullsv;
+ SV* svrecode = NULL;
if (RsSNARF(PL_rs))
return 0;
void
Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
{
+ dVAR;
#ifdef LIBERAL
register long *dl;
register long *ll;
rsave = rc = SvPV_nomg_const(right, rightlen);
len = leftlen < rightlen ? leftlen : rightlen;
lensave = len;
+ SvCUR_set(sv, len);
+ (void)SvPOK_only(sv);
if ((left_utf || right_utf) && (sv == left || sv == right)) {
needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
Newxz(dc, needlen + 1, char);
}
else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
- STRLEN n_a;
- dc = SvPV_force_nomg(sv, n_a);
- if (SvCUR(sv) < (STRLEN)len) {
+ dc = SvPV_force_nomg_nolen(sv);
+ if (SvLEN(sv) < (STRLEN)(len + 1)) {
dc = SvGROW(sv, (STRLEN)(len + 1));
(void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
}
needlen = ((optype == OP_BIT_AND)
? len : (leftlen > rightlen ? leftlen : rightlen));
Newxz(dc, needlen + 1, char);
- (void)sv_usepvn(sv, dc, needlen);
+ sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
}
- SvCUR_set(sv, len);
- (void)SvPOK_only(sv);
if (left_utf || right_utf) {
UV duc, luc, ruc;
- char *dcsave = dc;
+ char *dcorig = dc;
+ char *dcsave = NULL;
STRLEN lulen = leftlen;
STRLEN rulen = rightlen;
STRLEN ulen;
dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
if (sv == left || sv == right)
- (void)sv_usepvn(sv, dcsave, needlen);
- SvCUR_set(sv, dc - dcsave);
+ (void)sv_usepvn(sv, dcorig, needlen);
+ SvCUR_set(sv, dc - dcorig);
break;
case OP_BIT_XOR:
while (lulen && rulen) {
dc = (char*)uvchr_to_utf8((U8*)dc, duc);
}
mop_up_utf:
+ if (rulen)
+ dcsave = savepvn(rc, rulen);
+ else if (lulen)
+ dcsave = savepvn(lc, lulen);
if (sv == left || sv == right)
- (void)sv_usepvn(sv, dcsave, needlen);
- SvCUR_set(sv, dc - dcsave);
+ (void)sv_usepvn(sv, dcorig, needlen); /* Uses Renew(). */
+ SvCUR_set(sv, dc - dcorig);
if (rulen)
- sv_catpvn(sv, rc, rulen);
+ sv_catpvn(sv, dcsave, rulen);
else if (lulen)
- sv_catpvn(sv, lc, lulen);
+ sv_catpvn(sv, dcsave, lulen);
else
*SvEND(sv) = '\0';
+ Safefree(dcsave);
break;
+ default:
+ if (sv == left || sv == right)
+ Safefree(dcorig);
+ Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)",
+ (unsigned)optype, PL_op_name[optype]);
}
SvUTF8_on(sv);
goto finish;
case OP_BIT_AND:
while (len--)
*dc++ = *lc++ & *rc++;
+ *dc = '\0';
break;
case OP_BIT_XOR:
while (len--)
OP *
Perl_do_kv(pTHX)
{
+ dVAR;
dSP;
- HV *hv = (HV*)POPs;
+ HV * const hv = (HV*)POPs;
HV *keys;
register HE *entry;
const I32 gimme = GIMME_V;
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;
+ LvTARG(TARG) = NULL;
PUSHs(TARG);
}
RETURN;
if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
if (SvTYPE(TARG) < SVt_PVLV) {
sv_upgrade(TARG, SVt_PVLV);
- sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
+ sv_magic(TARG, NULL, PERL_MAGIC_nkeys, NULL, 0);
}
LvTYPE(TARG) = 'k';
if (LvTARG(TARG) != (SV*)keys) {
if (LvTARG(TARG))
SvREFCNT_dec(LvTARG(TARG));
- LvTARG(TARG) = SvREFCNT_inc(keys);
+ LvTARG(TARG) = SvREFCNT_inc_simple(keys);
}
PUSHs(TARG);
RETURN;
https://perl5.git.perl.org / perl5.git / blobdiff