/* doop.c
*
- * Copyright (c) 1991-1997, Larry Wall
+ * Copyright (c) 1991-2000, 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.
*/
#include "EXTERN.h"
+#define PERL_IN_DOOP_C
#include "perl.h"
+#ifndef PERL_MICRO
#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
#include <signal.h>
#endif
+#endif
-I32
-do_trans(sv,arg)
-SV *sv;
-OP *arg;
+#define HALF_UPGRADE(start,end) \
+ STMT_START { \
+ U8* NeWsTr; \
+ STRLEN LeN = (end) - (start); \
+ NeWsTr = bytes_to_utf8(start, &LeN); \
+ Copy(NeWsTr,start,LeN,U8*); \
+ end = (start) + LeN; \
+ } STMT_END
+
+
+STATIC I32
+S_do_trans_simple(pTHX_ SV *sv)
{
dTHR;
- register short *tbl;
- register U8 *s;
- register U8 *send;
- register U8 *d;
- register I32 ch;
- register I32 matches = 0;
- register I32 squash = op->op_private & OPpTRANS_SQUASH;
+ U8 *s;
+ U8 *d;
+ U8 *send;
+ U8 *dstart;
+ I32 matches = 0;
+ I32 sutf = SvUTF8(sv);
STRLEN len;
+ short *tbl;
+ I32 ch;
- if (SvREADONLY(sv))
- croak(no_modify);
tbl = (short*)cPVOP->op_pv;
+ if (!tbl)
+ Perl_croak(aTHX_ "panic: do_trans");
+
s = (U8*)SvPV(sv, len);
- if (!len)
- return 0;
- if (!SvPOKp(sv))
- s = (U8*)SvPV_force(sv, len);
- (void)SvPOK_only(sv);
send = s + len;
- if (!tbl || !s)
- croak("panic: do_trans");
- DEBUG_t( deb("2.TBL\n"));
- if (!op->op_private) {
+
+ /* First, take care of non-UTF8 input strings, because they're easy */
+ if (!sutf) {
while (s < send) {
if ((ch = tbl[*s]) >= 0) {
matches++;
- *s = ch;
+ *s++ = ch;
}
- s++;
+ else
+ s++;
+ }
+ SvSETMAGIC(sv);
+ return matches;
+ }
+
+ /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
+ Newz(0, d, len*2+1, U8);
+ dstart = d;
+ while (s < send) {
+ I32 ulen;
+ short c;
+
+ ulen = 1;
+ /* Need to check this, otherwise 128..255 won't match */
+ c = utf8_to_uv(s, &ulen);
+ if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
+ matches++;
+ if (ch < 0x80)
+ *d++ = ch;
+ else
+ d = uv_to_utf8(d,ch);
+ s += ulen;
+ }
+ else { /* No match -> copy */
+ while (ulen--)
+ *d++ = *s++;
+ }
+ }
+ *d = '\0';
+ sv_setpvn(sv, (const char*)dstart, d - dstart);
+ SvUTF8_on(sv);
+ SvLEN_set(sv, 2*len+1);
+ SvSETMAGIC(sv);
+ return matches;
+}
+
+STATIC I32
+S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
+{
+ dTHR;
+ U8 *s;
+ U8 *send;
+ I32 matches = 0;
+ I32 hasutf = SvUTF8(sv);
+ STRLEN len;
+ short *tbl;
+
+ tbl = (short*)cPVOP->op_pv;
+ if (!tbl)
+ Perl_croak(aTHX_ "panic: do_trans");
+
+ s = (U8*)SvPV(sv, len);
+ send = s + len;
+
+ while (s < send) {
+ if (hasutf && *s & 0x80)
+ s += UTF8SKIP(s);
+ else {
+ UV c;
+ I32 ulen;
+ ulen = 1;
+ if (hasutf)
+ c = utf8_to_uv(s,&ulen);
+ else
+ c = *s;
+ if (c < 0x100 && tbl[c] >= 0)
+ matches++;
+ s += ulen;
+ }
+ }
+
+ return matches;
+}
+
+STATIC I32
+S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
+{
+ dTHR;
+ U8 *s;
+ U8 *send;
+ U8 *d;
+ I32 hasutf = SvUTF8(sv);
+ I32 matches = 0;
+ STRLEN len;
+ short *tbl;
+ I32 ch;
+
+ tbl = (short*)cPVOP->op_pv;
+ if (!tbl)
+ Perl_croak(aTHX_ "panic: do_trans");
+
+ s = (U8*)SvPV(sv, len);
+ send = s + len;
+
+ d = s;
+ if (PL_op->op_private & OPpTRANS_SQUASH) {
+ U8* p = send;
+
+ while (s < send) {
+ if (hasutf && *s & 0x80)
+ s += UTF8SKIP(s);
+ else {
+ if ((ch = tbl[*s]) >= 0) {
+ *d = ch;
+ matches++;
+ if (p == d - 1 && *p == *d)
+ matches--;
+ else
+ p = d++;
+ }
+ else if (ch == -1) /* -1 is unmapped character */
+ *d++ = *s; /* -2 is delete character */
+ s++;
+ }
}
}
else {
- d = s;
while (s < send) {
- if ((ch = tbl[*s]) >= 0) {
- *d = ch;
- if (matches++ && squash) {
- if (d[-1] == *d)
- matches--;
- else
- d++;
- }
- else
+ if (hasutf && *s & 0x80)
+ s += UTF8SKIP(s);
+ else {
+ if ((ch = tbl[*s]) >= 0) {
+ *d = ch;
+ matches++;
d++;
+ }
+ else if (ch == -1) /* -1 is unmapped character */
+ *d++ = *s; /* -2 is delete character */
+ s++;
+ }
+ }
+ }
+ matches += send - d; /* account for disappeared chars */
+ *d = '\0';
+ SvCUR_set(sv, d - (U8*)SvPVX(sv));
+ SvSETMAGIC(sv);
+
+ return matches;
+}
+
+STATIC I32
+S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
+{
+ dTHR;
+ U8 *s;
+ U8 *send;
+ U8 *d;
+ U8 *start;
+ U8 *dstart;
+ I32 matches = 0;
+ STRLEN len;
+
+ SV* rv = (SV*)cSVOP->op_sv;
+ HV* hv = (HV*)SvRV(rv);
+ SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
+ UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ UV extra = none + 1;
+ UV final;
+ UV uv;
+ I32 isutf;
+ I32 howmany;
+
+ isutf = SvUTF8(sv);
+ s = (U8*)SvPV(sv, len);
+ send = s + len;
+ start = s;
+
+ svp = hv_fetch(hv, "FINAL", 5, FALSE);
+ if (svp)
+ final = SvUV(*svp);
+
+ /* d needs to be bigger than s, in case e.g. upgrading is required */
+ Newz(0, d, len*2+1, U8);
+ dstart = d;
+ while (s < send) {
+ if ((uv = swash_fetch(rv, s)) < none) {
+ s += UTF8SKIP(s);
+ matches++;
+ if ((uv & 0x80) && !isutf++)
+ HALF_UPGRADE(dstart,d);
+ d = uv_to_utf8(d, uv);
+ }
+ else if (uv == none) {
+ int i;
+ i = UTF8SKIP(s);
+ if (i > 1 && !isutf++)
+ HALF_UPGRADE(dstart,d);
+ while(i--)
+ *d++ = *s++;
+ }
+ else if (uv == extra) {
+ int i;
+ i = UTF8SKIP(s);
+ s += i;
+ matches++;
+ if (i > 1 && !isutf++)
+ HALF_UPGRADE(dstart,d);
+ d = uv_to_utf8(d, final);
+ }
+ else
+ s += UTF8SKIP(s);
+ }
+ *d = '\0';
+ sv_setpvn(sv, (const char*)dstart, d - dstart);
+ SvSETMAGIC(sv);
+ if (isutf)
+ SvUTF8_on(sv);
+
+ return matches;
+}
+
+STATIC I32
+S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
+{
+ dTHR;
+ U8 *s;
+ U8 *send;
+ I32 matches = 0;
+ STRLEN len;
+
+ SV* rv = (SV*)cSVOP->op_sv;
+ HV* hv = (HV*)SvRV(rv);
+ SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
+ UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ UV uv;
+
+ s = (U8*)SvPV(sv, len);
+ if (!SvUTF8(sv))
+ s = bytes_to_utf8(s, &len);
+ send = s + len;
+
+ while (s < send) {
+ if ((uv = swash_fetch(rv, s)) < none)
+ matches++;
+ s += UTF8SKIP(s);
+ }
+
+ return matches;
+}
+
+STATIC I32
+S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
+{
+ dTHR;
+ U8 *s;
+ U8 *send;
+ U8 *d;
+ I32 matches = 0;
+ I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
+ I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ SV* rv = (SV*)cSVOP->op_sv;
+ HV* hv = (HV*)SvRV(rv);
+ SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
+ UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ UV extra = none + 1;
+ UV final;
+ UV uv;
+ STRLEN len;
+ U8 *dst;
+ I32 isutf = SvUTF8(sv);
+
+ s = (U8*)SvPV(sv, len);
+ send = s + len;
+
+ svp = hv_fetch(hv, "FINAL", 5, FALSE);
+ if (svp)
+ final = SvUV(*svp);
+
+ Newz(0, d, len*2+1, U8);
+ dst = d;
+
+ if (squash) {
+ UV puv = 0xfeedface;
+ while (s < send) {
+ if (SvUTF8(sv))
+ uv = swash_fetch(rv, s);
+ else {
+ U8 tmpbuf[2];
+ uv = *s++;
+ if (uv < 0x80)
+ tmpbuf[0] = uv;
+ else {
+ tmpbuf[0] = (( uv >> 6) | 0xc0);
+ tmpbuf[1] = (( uv & 0x3f) | 0x80);
+ }
+ uv = swash_fetch(rv, tmpbuf);
}
- else if (ch == -1) /* -1 is unmapped character */
- *d++ = *s; /* -2 is delete character */
- s++;
+
+ if (uv < none) {
+ matches++;
+ if (uv != puv) {
+ if ((uv & 0x80) && !isutf++)
+ HALF_UPGRADE(dst,d);
+ d = uv_to_utf8(d, uv);
+ puv = uv;
+ }
+ s += UTF8SKIP(s);
+ continue;
+ }
+ else if (uv == none) { /* "none" is unmapped character */
+ I32 ulen;
+ *d++ = (U8)utf8_to_uv(s, &ulen);
+ s += ulen;
+ puv = 0xfeedface;
+ continue;
+ }
+ else if (uv == extra && !del) {
+ matches++;
+ if (uv != puv) {
+ d = uv_to_utf8(d, final);
+ puv = final;
+ }
+ s += UTF8SKIP(s);
+ continue;
+ }
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
}
- matches += send - d; /* account for disappeared chars */
+ }
+ else {
+ while (s < send) {
+ if (SvUTF8(sv))
+ uv = swash_fetch(rv, s);
+ else {
+ U8 tmpbuf[2];
+ uv = *s++;
+ if (uv < 0x80)
+ tmpbuf[0] = uv;
+ else {
+ tmpbuf[0] = (( uv >> 6) | 0xc0);
+ tmpbuf[1] = (( uv & 0x3f) | 0x80);
+ }
+ uv = swash_fetch(rv, tmpbuf);
+ }
+ if (uv < none) {
+ matches++;
+ d = uv_to_utf8(d, uv);
+ s += UTF8SKIP(s);
+ continue;
+ }
+ else if (uv == none) { /* "none" is unmapped character */
+ I32 ulen;
+ *d++ = (U8)utf8_to_uv(s, &ulen);
+ s += ulen;
+ continue;
+ }
+ else if (uv == extra && !del) {
+ matches++;
+ d = uv_to_utf8(d, final);
+ s += UTF8SKIP(s);
+ continue;
+ }
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
+ }
+ }
+ if (dst)
+ sv_usepvn(sv, (char*)dst, d - dst);
+ else {
*d = '\0';
SvCUR_set(sv, d - (U8*)SvPVX(sv));
}
SvSETMAGIC(sv);
+
return matches;
}
+I32
+Perl_do_trans(pTHX_ SV *sv)
+{
+ dTHR;
+ STRLEN len;
+ I32 hasutf = (PL_op->op_private &
+ (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
+
+ if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
+ Perl_croak(aTHX_ PL_no_modify);
+
+ (void)SvPV(sv, len);
+ if (!len)
+ return 0;
+ if (!SvPOKp(sv))
+ (void)SvPV_force(sv, len);
+ if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
+ (void)SvPOK_only_UTF8(sv);
+
+ DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
+
+ switch (PL_op->op_private & ~hasutf & 63) {
+ case 0:
+ if (hasutf)
+ return do_trans_simple_utf8(sv);
+ else
+ return do_trans_simple(sv);
+
+ case OPpTRANS_IDENTICAL:
+ if (hasutf)
+ return do_trans_count_utf8(sv);
+ else
+ return do_trans_count(sv);
+
+ default:
+ if (hasutf)
+ return do_trans_complex_utf8(sv);
+ else
+ return do_trans_complex(sv);
+ }
+}
+
void
-do_join(sv,del,mark,sp)
-register SV *sv;
-SV *del;
-register SV **mark;
-register SV **sp;
+Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
{
SV **oldmark = mark;
register I32 items = sp - mark;
mark++;
len = (items > 0 ? (delimlen * (items - 1) ) : 0);
- if (SvTYPE(sv) < SVt_PV)
- sv_upgrade(sv, SVt_PV);
+ (void)SvUPGRADE(sv, SVt_PV);
if (SvLEN(sv) < len + items) { /* current length is way too short */
while (items-- > 0) {
- if (*mark) {
+ if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
SvPV(*mark, tmplen);
len += tmplen;
}
SvGROW(sv, len + 1); /* so try to pre-extend */
mark = oldmark;
- items = sp - mark;;
+ items = sp - mark;
++mark;
}
if (items-- > 0) {
char *s;
- if (*mark) {
- s = SvPV(*mark, tmplen);
- sv_setpvn(sv, s, tmplen);
- }
- else
- sv_setpv(sv, "");
+ sv_setpv(sv, "");
+ if (*mark)
+ sv_catsv(sv, *mark);
mark++;
}
else
}
void
-do_sprintf(sv,len,sarg)
-SV *sv;
-I32 len;
-SV **sarg;
+Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
{
STRLEN patlen;
char *pat = SvPV(*sarg, patlen);
SvTAINTED_on(sv);
}
+/* XXX SvUTF8 support missing! */
+UV
+Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
+{
+ STRLEN srclen, len;
+ unsigned char *s = (unsigned char *) SvPV(sv, srclen);
+ UV retnum = 0;
+
+ if (offset < 0)
+ return retnum;
+ if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
+ Perl_croak(aTHX_ "Illegal number of bits in vec");
+ offset *= size; /* turn into bit offset */
+ len = (offset + size + 7) / 8; /* required number of bytes */
+ if (len > srclen) {
+ if (size <= 8)
+ retnum = 0;
+ else {
+ offset >>= 3; /* turn into byte offset */
+ if (size == 16) {
+ if (offset >= srclen)
+ retnum = 0;
+ else
+ retnum = (UV) s[offset] << 8;
+ }
+ else if (size == 32) {
+ if (offset >= srclen)
+ retnum = 0;
+ else if (offset + 1 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 24);
+ else if (offset + 2 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 24) +
+ ((UV) s[offset + 1] << 16);
+ else
+ retnum =
+ ((UV) s[offset ] << 24) +
+ ((UV) s[offset + 1] << 16) +
+ ( s[offset + 2] << 8);
+ }
+#ifdef UV_IS_QUAD
+ else if (size == 64) {
+ dTHR;
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ WARN_PORTABLE,
+ "Bit vector size > 32 non-portable");
+ if (offset >= srclen)
+ retnum = 0;
+ else if (offset + 1 >= srclen)
+ retnum =
+ (UV) s[offset ] << 56;
+ else if (offset + 2 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48);
+ else if (offset + 3 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40);
+ else if (offset + 4 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40) +
+ ((UV) s[offset + 3] << 32);
+ else if (offset + 5 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40) +
+ ((UV) s[offset + 3] << 32) +
+ ( s[offset + 4] << 24);
+ else if (offset + 6 >= srclen)
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40) +
+ ((UV) s[offset + 3] << 32) +
+ ((UV) s[offset + 4] << 24) +
+ ((UV) s[offset + 5] << 16);
+ else
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40) +
+ ((UV) s[offset + 3] << 32) +
+ ((UV) s[offset + 4] << 24) +
+ ((UV) s[offset + 5] << 16) +
+ ( s[offset + 6] << 8);
+ }
+#endif
+ }
+ }
+ else if (size < 8)
+ retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
+ else {
+ offset >>= 3; /* turn into byte offset */
+ if (size == 8)
+ retnum = s[offset];
+ else if (size == 16)
+ retnum =
+ ((UV) s[offset] << 8) +
+ s[offset + 1];
+ else if (size == 32)
+ retnum =
+ ((UV) s[offset ] << 24) +
+ ((UV) s[offset + 1] << 16) +
+ ( s[offset + 2] << 8) +
+ s[offset + 3];
+#ifdef UV_IS_QUAD
+ else if (size == 64) {
+ dTHR;
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ WARN_PORTABLE,
+ "Bit vector size > 32 non-portable");
+ retnum =
+ ((UV) s[offset ] << 56) +
+ ((UV) s[offset + 1] << 48) +
+ ((UV) s[offset + 2] << 40) +
+ ((UV) s[offset + 3] << 32) +
+ ((UV) s[offset + 4] << 24) +
+ ((UV) s[offset + 5] << 16) +
+ ( s[offset + 6] << 8) +
+ s[offset + 7];
+ }
+#endif
+ }
+
+ return retnum;
+}
+
+/* XXX SvUTF8 support missing! */
void
-do_vecset(sv)
-SV *sv;
+Perl_do_vecset(pTHX_ SV *sv)
{
SV *targ = LvTARG(sv);
register I32 offset;
register I32 size;
register unsigned char *s;
- register unsigned long lval;
+ register UV lval;
I32 mask;
STRLEN targlen;
STRLEN len;
if (!targ)
return;
s = (unsigned char*)SvPV_force(targ, targlen);
- lval = U_L(SvNV(sv));
+ (void)SvPOK_only(targ);
+ lval = SvUV(sv);
offset = LvTARGOFF(sv);
size = LvTARGLEN(sv);
+ if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
+ Perl_croak(aTHX_ "Illegal number of bits in vec");
- len = (offset + size + 7) / 8;
+ offset *= size; /* turn into bit offset */
+ len = (offset + size + 7) / 8; /* required number of bytes */
if (len > targlen) {
s = (unsigned char*)SvGROW(targ, len + 1);
- (void)memzero(s + targlen, len - targlen + 1);
+ (void)memzero((char *)(s + targlen), len - targlen + 1);
SvCUR_set(targ, len);
}
mask = (1 << size) - 1;
size = offset & 7;
lval &= mask;
- offset >>= 3;
+ offset >>= 3; /* turn into byte offset */
s[offset] &= ~(mask << size);
s[offset] |= lval << size;
}
else {
- offset >>= 3;
+ offset >>= 3; /* turn into byte offset */
if (size == 8)
- s[offset] = lval & 255;
+ s[offset ] = lval & 0xff;
else if (size == 16) {
- s[offset] = (lval >> 8) & 255;
- s[offset+1] = lval & 255;
+ s[offset ] = (lval >> 8) & 0xff;
+ s[offset+1] = lval & 0xff;
}
else if (size == 32) {
- s[offset] = (lval >> 24) & 255;
- s[offset+1] = (lval >> 16) & 255;
- s[offset+2] = (lval >> 8) & 255;
- s[offset+3] = lval & 255;
+ s[offset ] = (lval >> 24) & 0xff;
+ s[offset+1] = (lval >> 16) & 0xff;
+ s[offset+2] = (lval >> 8) & 0xff;
+ s[offset+3] = lval & 0xff;
}
+#ifdef UV_IS_QUAD
+ else if (size == 64) {
+ dTHR;
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ WARN_PORTABLE,
+ "Bit vector size > 32 non-portable");
+ s[offset ] = (lval >> 56) & 0xff;
+ s[offset+1] = (lval >> 48) & 0xff;
+ s[offset+2] = (lval >> 40) & 0xff;
+ s[offset+3] = (lval >> 32) & 0xff;
+ s[offset+4] = (lval >> 24) & 0xff;
+ s[offset+5] = (lval >> 16) & 0xff;
+ s[offset+6] = (lval >> 8) & 0xff;
+ s[offset+7] = lval & 0xff;
+ }
+#endif
}
+ SvSETMAGIC(targ);
}
void
-do_chop(astr,sv)
-register SV *astr;
-register SV *sv;
+Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
{
STRLEN len;
char *s;
+ dTHR;
if (SvTYPE(sv) == SVt_PVAV) {
register I32 i;
max = AvFILL(av);
for (i = 0; i <= max; i++) {
sv = (SV*)av_fetch(av, i, FALSE);
- if (sv && ((sv = *(SV**)sv), sv != &sv_undef))
+ if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
do_chop(astr, sv);
}
return;
}
- if (SvTYPE(sv) == SVt_PVHV) {
+ else if (SvTYPE(sv) == SVt_PVHV) {
HV* hv = (HV*)sv;
HE* entry;
(void)hv_iterinit(hv);
/*SUPPRESS 560*/
- while (entry = hv_iternext(hv))
+ while ((entry = hv_iternext(hv)))
do_chop(astr,hv_iterval(hv,entry));
return;
}
+ else if (SvREADONLY(sv))
+ Perl_croak(aTHX_ PL_no_modify);
s = SvPV(sv, len);
if (len && !SvPOK(sv))
s = SvPV_force(sv, len);
- if (s && len) {
+ if (DO_UTF8(sv)) {
+ if (s && len) {
+ 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);
+ }
+ else
+ sv_setpvn(astr, "", 0);
+ }
+ else if (s && len) {
s += --len;
sv_setpvn(astr, s, 1);
*s = '\0';
SvCUR_set(sv, len);
+ SvUTF8_off(sv);
SvNIOK_off(sv);
}
else
sv_setpvn(astr, "", 0);
SvSETMAGIC(sv);
-}
+}
I32
-do_chomp(sv)
-register SV *sv;
+Perl_do_chomp(pTHX_ register SV *sv)
{
+ dTHR;
register I32 count;
STRLEN len;
char *s;
- if (RsSNARF(rs))
+ if (RsSNARF(PL_rs))
return 0;
+ if (RsRECORD(PL_rs))
+ return 0;
count = 0;
if (SvTYPE(sv) == SVt_PVAV) {
register I32 i;
max = AvFILL(av);
for (i = 0; i <= max; i++) {
sv = (SV*)av_fetch(av, i, FALSE);
- if (sv && ((sv = *(SV**)sv), sv != &sv_undef))
+ if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
count += do_chomp(sv);
}
return count;
}
- if (SvTYPE(sv) == SVt_PVHV) {
+ else if (SvTYPE(sv) == SVt_PVHV) {
HV* hv = (HV*)sv;
HE* entry;
(void)hv_iterinit(hv);
/*SUPPRESS 560*/
- while (entry = hv_iternext(hv))
+ while ((entry = hv_iternext(hv)))
count += do_chomp(hv_iterval(hv,entry));
return count;
}
+ 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(rs)) {
+ if (RsPARA(PL_rs)) {
if (*s != '\n')
goto nope;
++count;
}
else {
STRLEN rslen;
- char *rsptr = SvPV(rs, rslen);
+ char *rsptr = SvPV(PL_rs, rslen);
if (rslen == 1) {
if (*s != *rsptr)
goto nope;
}
void
-do_vop(optype,sv,left,right)
-I32 optype;
-SV *sv;
-SV *left;
-SV *right;
+Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
{
+ dTHR; /* just for taint */
#ifdef LIBERAL
register long *dl;
register long *ll;
I32 lensave;
char *lsave;
char *rsave;
+ bool left_utf = DO_UTF8(left);
+ bool right_utf = DO_UTF8(right);
+ I32 needlen;
+
+ if (left_utf && !right_utf)
+ sv_utf8_upgrade(right);
+ if (!left_utf && right_utf)
+ sv_utf8_upgrade(left);
if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
rsave = rc = SvPV(right, rightlen);
len = leftlen < rightlen ? leftlen : rightlen;
lensave = len;
- if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
- dc = SvPV_force(sv, na);
+ if ((left_utf || right_utf) && (sv == left || sv == right)) {
+ needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
+ Newz(801, dc, needlen + 1, char);
+ }
+ else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
+ STRLEN n_a;
+ dc = SvPV_force(sv, n_a);
if (SvCUR(sv) < len) {
dc = SvGROW(sv, len + 1);
(void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
}
+ if (optype != OP_BIT_AND && (left_utf || right_utf))
+ dc = SvGROW(sv, leftlen + rightlen + 1);
}
else {
- I32 needlen = ((optype == OP_BIT_AND)
- ? len : (leftlen > rightlen ? leftlen : rightlen));
+ needlen = ((optype == OP_BIT_AND)
+ ? len : (leftlen > rightlen ? leftlen : rightlen));
Newz(801, dc, needlen + 1, char);
(void)sv_usepvn(sv, dc, needlen);
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;
+ STRLEN lulen = leftlen;
+ STRLEN rulen = rightlen;
+ I32 ulen;
+
+ switch (optype) {
+ case OP_BIT_AND:
+ while (lulen && rulen) {
+ luc = utf8_to_uv((U8*)lc, &ulen);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8_to_uv((U8*)rc, &ulen);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc & ruc;
+ dc = (char*)uv_to_utf8((U8*)dc, duc);
+ }
+ if (sv == left || sv == right)
+ (void)sv_usepvn(sv, dcsave, needlen);
+ SvCUR_set(sv, dc - dcsave);
+ break;
+ case OP_BIT_XOR:
+ while (lulen && rulen) {
+ luc = utf8_to_uv((U8*)lc, &ulen);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8_to_uv((U8*)rc, &ulen);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc ^ ruc;
+ dc = (char*)uv_to_utf8((U8*)dc, duc);
+ }
+ goto mop_up_utf;
+ case OP_BIT_OR:
+ while (lulen && rulen) {
+ luc = utf8_to_uv((U8*)lc, &ulen);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8_to_uv((U8*)rc, &ulen);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc | ruc;
+ dc = (char*)uv_to_utf8((U8*)dc, duc);
+ }
+ mop_up_utf:
+ if (sv == left || sv == right)
+ (void)sv_usepvn(sv, dcsave, needlen);
+ SvCUR_set(sv, dc - dcsave);
+ if (rulen)
+ sv_catpvn(sv, rc, rulen);
+ else if (lulen)
+ sv_catpvn(sv, lc, lulen);
+ else
+ *SvEND(sv) = '\0';
+ break;
+ }
+ SvUTF8_on(sv);
+ goto finish;
+ }
+ else
#ifdef LIBERAL
if (len >= sizeof(long)*4 &&
!((long)dc % sizeof(long)) &&
break;
}
}
+finish:
+ SvTAINT(sv);
}
OP *
-do_kv(ARGS)
-dARGS
+Perl_do_kv(pTHX)
{
- dSP;
+ djSP;
HV *hv = (HV*)POPs;
+ HV *keys;
register HE *entry;
SV *tmpstr;
I32 gimme = GIMME_V;
- I32 dokeys = (op->op_type == OP_KEYS);
- I32 dovalues = (op->op_type == OP_VALUES);
+ I32 dokeys = (PL_op->op_type == OP_KEYS);
+ I32 dovalues = (PL_op->op_type == OP_VALUES);
I32 realhv = (SvTYPE(hv) == SVt_PVHV);
- if (op->op_type == OP_RV2HV || op->op_type == OP_PADHV)
+ if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
dokeys = dovalues = TRUE;
if (!hv) {
- if (op->op_flags & OPf_MOD) { /* lvalue */
+ if (PL_op->op_flags & OPf_MOD) { /* lvalue */
dTARGET; /* make sure to clear its target here */
if (SvTYPE(TARG) == SVt_PVLV)
LvTARG(TARG) = Nullsv;
RETURN;
}
- if (realhv)
- (void)hv_iterinit(hv); /* always reset iterator regardless */
- else
- (void)avhv_iterinit((AV*)hv);
+ keys = realhv ? hv : avhv_keys((AV*)hv);
+ (void)hv_iterinit(keys); /* always reset iterator regardless */
if (gimme == G_VOID)
RETURN;
if (gimme == G_SCALAR) {
- I32 i;
+ IV i;
dTARGET;
- if (op->op_flags & OPf_MOD) { /* lvalue */
+ if (PL_op->op_flags & OPf_MOD) { /* lvalue */
if (SvTYPE(TARG) < SVt_PVLV) {
sv_upgrade(TARG, SVt_PVLV);
sv_magic(TARG, Nullsv, 'k', Nullch, 0);
}
LvTYPE(TARG) = 'k';
- LvTARG(TARG) = (SV*)hv;
+ if (LvTARG(TARG) != (SV*)keys) {
+ if (LvTARG(TARG))
+ SvREFCNT_dec(LvTARG(TARG));
+ LvTARG(TARG) = SvREFCNT_inc(keys);
+ }
PUSHs(TARG);
RETURN;
}
- if (!SvRMAGICAL(hv) || !mg_find((SV*)hv,'P'))
- i = HvKEYS(hv);
+ if (! SvTIED_mg((SV*)keys, 'P'))
+ i = HvKEYS(keys);
else {
i = 0;
/*SUPPRESS 560*/
- while (entry = realhv ? hv_iternext(hv) : avhv_iternext((AV*)hv)) {
- i++;
- }
+ while (hv_iternext(keys)) i++;
}
PUSHi( i );
RETURN;
}
- /* Guess how much room we need. hv_max may be a few too many. Oh well. */
- EXTEND(sp, HvMAX(hv) * (dokeys + dovalues));
+ EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
- while (entry = realhv ? hv_iternext(hv) : avhv_iternext((AV*)hv)) {
+ while ((entry = hv_iternext(keys))) {
SPAGAIN;
if (dokeys)
XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
if (dovalues) {
- tmpstr = sv_newmortal();
PUTBACK;
- sv_setsv(tmpstr,realhv ?
- hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry));
- DEBUG_H(sv_setpvf(tmpstr, "%lu%%%d=%lu",
+ tmpstr = realhv ?
+ hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
+ DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
(unsigned long)HeHASH(entry),
- HvMAX(hv)+1,
- (unsigned long)(HeHASH(entry) & HvMAX(hv))));
+ HvMAX(keys)+1,
+ (unsigned long)(HeHASH(entry) & HvMAX(keys))));
SPAGAIN;
XPUSHs(tmpstr);
}
}
return NORMAL;
}
+
projects / p5sagit/p5-mst-13.2.git / blobdiff