/* doop.c
*
- * Copyright (c) 1991-1997, Larry Wall
+ * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+ * 2000, 2001, 2002, 2004, 2005, 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.
* "'So that was the job I felt I had to do when I started,' thought Sam."
*/
+/* This file contains some common functions needed to carry out certain
+ * ops. For example both pp_schomp() and pp_chomp() - scalar and array
+ * chomp operations - call the function do_chomp() found in this file.
+ */
+
#include "EXTERN.h"
+#define PERL_IN_DOOP_C
#include "perl.h"
-#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
+#ifndef PERL_MICRO
#include <signal.h>
#endif
-static I32
-do_trans_CC_simple(SV *sv)
+STATIC I32
+S_do_trans_simple(pTHX_ SV *sv)
{
- dTHR;
U8 *s;
- U8 *send;
+ U8 *d;
+ const U8 *send;
+ U8 *dstart;
I32 matches = 0;
+ const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
STRLEN len;
- short *tbl;
- I32 ch;
- tbl = (short*)cPVOP->op_pv;
+ const short * const tbl = (short*)cPVOP->op_pv;
if (!tbl)
- croak("panic: do_trans");
+ Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
s = (U8*)SvPV(sv, len);
send = s + len;
- while (s < send) {
- if ((ch = tbl[*s]) >= 0) {
- matches++;
- *s = ch;
+ /* 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++;
}
- s++;
+ SvSETMAGIC(sv);
+ return matches;
}
- SvSETMAGIC(sv);
-
- return matches;
-}
-
-static I32
-do_trans_CC_count(SV *sv)
-{
- dTHR;
- U8 *s;
- U8 *send;
- I32 matches = 0;
- STRLEN len;
- short *tbl;
-
- tbl = (short*)cPVOP->op_pv;
- if (!tbl)
- croak("panic: do_trans");
-
- s = (U8*)SvPV(sv, len);
- send = s + len;
+ /* 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) {
- if (tbl[*s] >= 0)
- matches++;
- s++;
+ 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);
+ }
+ SvUTF8_on(sv);
+ SvSETMAGIC(sv);
return matches;
}
-static I32
-do_trans_CC_complex(SV *sv)
+STATIC I32
+S_do_trans_count(pTHX_ SV *sv)
{
- dTHR;
- U8 *s;
- U8 *send;
- U8 *d;
+ const U8 *s;
+ const U8 *send;
I32 matches = 0;
STRLEN len;
- short *tbl;
- I32 ch;
+ const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
- tbl = (short*)cPVOP->op_pv;
+ const short * const tbl = (short*)cPVOP->op_pv;
if (!tbl)
- croak("panic: do_trans");
+ Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
- s = (U8*)SvPV(sv, len);
+ s = (const U8*)SvPV_const(sv, len);
send = s + len;
- d = s;
- if (PL_op->op_private & OPpTRANS_SQUASH) {
- U8* p = send;
-
+ if (!SvUTF8(sv))
while (s < send) {
- 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++;
+ if (tbl[*s++] >= 0)
+ matches++;
}
- }
- else {
+ else
while (s < send) {
- if ((ch = tbl[*s]) >= 0) {
- *d = ch;
+ STRLEN ulen;
+ const UV c = utf8n_to_uvchr(s, send - s, &ulen, 0);
+ if (c < 0x100) {
+ if (tbl[c] >= 0)
+ matches++;
+ } else if (complement)
matches++;
- d++;
- }
- else if (ch == -1) /* -1 is unmapped character */
- *d++ = *s; /* -2 is delete character */
- s++;
+ s += ulen;
}
- }
- matches += send - d; /* account for disappeared chars */
- *d = '\0';
- SvCUR_set(sv, d - (U8*)SvPVX(sv));
- SvSETMAGIC(sv);
return matches;
}
-static I32
-do_trans_UU_simple(SV *sv)
+STATIC I32
+S_do_trans_complex(pTHX_ SV *sv)
{
- dTHR;
U8 *s;
U8 *send;
U8 *d;
+ U8 *dstart;
+ I32 isutf8;
I32 matches = 0;
- STRLEN len;
+ 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;
- 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;
+ 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;
- svp = hv_fetch(hv, "FINAL", 5, FALSE);
- if (svp)
- final = SvUV(*svp);
-
- d = s;
- while (s < send) {
- if ((uv = swash_fetch(rv, s)) < none) {
- s += UTF8SKIP(s);
- matches++;
- d = uv_to_utf8(d, uv);
- }
- else if (uv == none) {
- int i;
- for (i = UTF8SKIP(s); i; i--)
- *d++ = *s++;
+ if (!isutf8) {
+ dstart = d = s;
+ if (PL_op->op_private & OPpTRANS_SQUASH) {
+ const U8* p = send;
+ while (s < send) {
+ const I32 ch = tbl[*s];
+ if (ch >= 0) {
+ *d = (U8)ch;
+ matches++;
+ if (p != d - 1 || *p != *d)
+ p = d++;
+ }
+ else if (ch == -1) /* -1 is unmapped character */
+ *d++ = *s;
+ else if (ch == -2) /* -2 is delete character */
+ matches++;
+ s++;
+ }
}
- else if (uv == extra) {
- s += UTF8SKIP(s);
- matches++;
- d = uv_to_utf8(d, final);
+ else {
+ while (s < send) {
+ const I32 ch = tbl[*s];
+ if (ch >= 0) {
+ matches++;
+ *d++ = (U8)ch;
+ }
+ else if (ch == -1) /* -1 is unmapped character */
+ *d++ = *s;
+ else if (ch == -2) /* -2 is delete character */
+ matches++;
+ s++;
+ }
}
- else
- s += UTF8SKIP(s);
+ *d = '\0';
+ SvCUR_set(sv, d - dstart);
}
- *d = '\0';
- SvCUR_set(sv, d - (U8*)SvPVX(sv));
- SvSETMAGIC(sv);
-
- return matches;
-}
-
-static I32
-do_trans_UU_count(SV *sv)
-{
- 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;
+ else { /* isutf8 */
+ if (grows)
+ Newx(d, len*2+1, U8);
+ else
+ d = s;
+ dstart = d;
+ if (complement && !del)
+ rlen = tbl[0x100];
- s = (U8*)SvPV(sv, len);
- send = s + len;
+#ifdef MACOS_TRADITIONAL
+#define comp CoMP /* "comp" is a keyword in some compilers ... */
+#endif
- while (s < send) {
- if ((uv = swash_fetch(rv, s)) < none) {
- s += UTF8SKIP(s);
- matches++;
+ if (PL_op->op_private & OPpTRANS_SQUASH) {
+ UV pch = 0xfeedface;
+ while (s < send) {
+ STRLEN len;
+ const UV comp = utf8_to_uvchr(s, &len);
+ I32 ch;
+
+ 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 ((UV)ch != pch) {
+ d = uvchr_to_utf8(d, ch);
+ pch = (UV)ch;
+ }
+ s += len;
+ continue;
+ }
+ }
+ }
+ else if ((ch = tbl[comp]) >= 0) {
+ matches++;
+ if ((UV)ch != pch) {
+ d = uvchr_to_utf8(d, ch);
+ pch = (UV)ch;
+ }
+ s += len;
+ continue;
+ }
+ 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;
+ pch = 0xfeedface;
+ }
}
+ else {
+ while (s < send) {
+ STRLEN len;
+ const UV comp = utf8_to_uvchr(s, &len);
+ I32 ch;
+ if (comp > 0xff) {
+ if (!complement) {
+ Move(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 = uvchr_to_utf8(d, ch);
+ matches++;
+ }
+ 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;
+ }
+ }
+ 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;
}
-static I32
-do_trans_UC_simple(SV *sv)
+STATIC I32
+S_do_trans_simple_utf8(pTHX_ SV *sv)
{
- dTHR;
U8 *s;
U8 *send;
U8 *d;
+ U8 *start;
+ U8 *dstart, *dend;
I32 matches = 0;
+ const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
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;
+ SV* const rv = (SV*)cSVOP->op_sv;
+ HV* const hv = (HV*)SvRV(rv);
+ SV* const * svp = hv_fetch(hv, "NONE", 4, FALSE);
+ const UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ const UV extra = none + 1;
+ UV final = 0;
+ I32 isutf8;
+ U8 hibit = 0;
s = (U8*)SvPV(sv, len);
+ isutf8 = SvUTF8(sv);
+ if (!isutf8) {
+ const U8 *t = s;
+ const U8 * const e = s + len;
+ while (t < e) {
+ const U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ break;
+ }
+ if (hibit)
+ s = bytes_to_utf8(s, &len);
+ }
send = s + len;
+ start = s;
svp = hv_fetch(hv, "FINAL", 5, FALSE);
if (svp)
final = SvUV(*svp);
- d = s;
+ if (grows) {
+ /* d needs to be bigger than s, in case e.g. upgrading is required */
+ Newx(d, len * 3 + UTF8_MAXBYTES, U8);
+ dend = d + len * 3;
+ dstart = d;
+ }
+ else {
+ dstart = d = s;
+ dend = d + len;
+ }
+
while (s < send) {
- if ((uv = swash_fetch(rv, s)) < none) {
+ const UV uv = swash_fetch(rv, s, TRUE);
+ if (uv < none) {
s += UTF8SKIP(s);
matches++;
- *d++ = (U8)uv;
+ d = uvuni_to_utf8(d, uv);
}
else if (uv == none) {
- I32 ulen;
- uv = utf8_to_uv(s, &ulen);
- s += ulen;
- *d++ = (U8)uv;
+ const int i = UTF8SKIP(s);
+ Move(s, d, i, U8);
+ d += i;
+ s += i;
}
else if (uv == extra) {
s += UTF8SKIP(s);
matches++;
- *d++ = (U8)final;
+ d = uvuni_to_utf8(d, final);
}
else
s += UTF8SKIP(s);
- }
- *d = '\0';
- SvCUR_set(sv, d - (U8*)SvPVX(sv));
- SvSETMAGIC(sv);
- return matches;
-}
-
-static I32
-do_trans_CU_simple(SV *sv)
-{
- dTHR;
- U8 *s;
- U8 *send;
- U8 *d;
- U8 *dst;
- 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;
- U8 tmpbuf[10];
- I32 bits = 16;
-
- s = (U8*)SvPV(sv, len);
- send = s + len;
-
- svp = hv_fetch(hv, "BITS", 4, FALSE);
- if (svp)
- bits = (I32)SvIV(*svp);
-
- svp = hv_fetch(hv, "FINAL", 5, FALSE);
- if (svp)
- final = SvUV(*svp);
-
- Newz(801, d, len * (bits >> 3) + 1, U8);
- dst = d;
-
- while (s < send) {
- uv = *s++;
- if (uv < 0x80)
- tmpbuf[0] = uv;
- else {
- tmpbuf[0] = (( uv >> 6) | 0xc0);
- tmpbuf[1] = (( uv & 0x3f) | 0x80);
- }
-
- if ((uv = swash_fetch(rv, tmpbuf)) < none) {
- matches++;
- d = uv_to_utf8(d, uv);
- }
- else if (uv == none)
- d = uv_to_utf8(d, s[-1]);
- else if (uv == extra) {
- matches++;
- d = uv_to_utf8(d, final);
+ if (d > dend) {
+ const STRLEN clen = d - dstart;
+ const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
+ Renew(dstart, nlen + UTF8_MAXBYTES, U8);
+ d = dstart + clen;
+ dend = dstart + nlen;
}
}
- *d = '\0';
- sv_usepvn_mg(sv, (char*)dst, d - dst);
+ 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);
return matches;
}
-/* utf-8 to latin-1 */
-
-static I32
-do_trans_UC_trivial(SV *sv)
+STATIC I32
+S_do_trans_count_utf8(pTHX_ SV *sv)
{
- dTHR;
- U8 *s;
- U8 *send;
- U8 *d;
+ const U8 *s;
+ const U8 *start = 0, *send;
+ I32 matches = 0;
STRLEN len;
- s = (U8*)SvPV(sv, len);
- send = s + len;
-
- d = s;
- while (s < send) {
- if (*s < 0x80)
- *d++ = *s++;
- else {
- I32 ulen;
- UV uv = utf8_to_uv(s, &ulen);
- s += ulen;
- *d++ = (U8)uv;
+ SV* const rv = (SV*)cSVOP->op_sv;
+ HV* const hv = (HV*)SvRV(rv);
+ SV* const * const svp = hv_fetch(hv, "NONE", 4, 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 * const e = s + len;
+ while (t < e) {
+ const U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ break;
}
+ if (hibit)
+ start = s = bytes_to_utf8(s, &len);
}
- *d = '\0';
- SvCUR_set(sv, d - (U8*)SvPVX(sv));
- SvSETMAGIC(sv);
-
- return SvCUR(sv);
-}
-
-/* latin-1 to utf-8 */
-
-static I32
-do_trans_CU_trivial(SV *sv)
-{
- dTHR;
- U8 *s;
- U8 *send;
- U8 *d;
- U8 *dst;
- I32 matches;
- STRLEN len;
-
- s = (U8*)SvPV(sv, len);
send = s + len;
- Newz(801, d, len * 2 + 1, U8);
- dst = d;
-
- matches = send - s;
-
while (s < send) {
- if (*s < 0x80)
- *d++ = *s++;
- else {
- UV uv = *s++;
- *d++ = (( uv >> 6) | 0xc0);
- *d++ = (( uv & 0x3f) | 0x80);
- }
+ const UV uv = swash_fetch(rv, s, TRUE);
+ if (uv < none || uv == extra)
+ matches++;
+ s += UTF8SKIP(s);
}
- *d = '\0';
- sv_usepvn_mg(sv, (char*)dst, d - dst);
+ if (hibit)
+ Safefree(start);
return matches;
}
-static I32
-do_trans_UU_complex(SV *sv)
+STATIC I32
+S_do_trans_complex_utf8(pTHX_ SV *sv)
{
- dTHR;
- U8 *s;
- U8 *send;
+ U8 *start, *send;
U8 *d;
I32 matches = 0;
- I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
- I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
- I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
- 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;
+ const I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
+ const I32 del = PL_op->op_private & OPpTRANS_DELETE;
+ const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
+ SV * const rv = (SV*)cSVOP->op_sv;
+ HV * const hv = (HV*)SvRV(rv);
+ SV * const *svp = hv_fetch(hv, "NONE", 4, FALSE);
+ const UV none = svp ? SvUV(*svp) : 0x7fffffff;
+ const UV extra = none + 1;
+ UV final = 0;
+ bool havefinal = FALSE;
STRLEN len;
- U8 *dst;
-
- s = (U8*)SvPV(sv, len);
+ U8 *dstart, *dend;
+ U8 hibit = 0;
+
+ U8 *s = (U8*)SvPV(sv, len);
+ const I32 isutf8 = SvUTF8(sv);
+ if (!isutf8) {
+ const U8 *t = s;
+ const U8 * const e = s + len;
+ while (t < e) {
+ const U8 ch = *t++;
+ if ((hibit = !NATIVE_IS_INVARIANT(ch)))
+ break;
+ }
+ if (hibit)
+ s = bytes_to_utf8(s, &len);
+ }
send = s + len;
+ start = s;
svp = hv_fetch(hv, "FINAL", 5, FALSE);
- if (svp)
+ if (svp) {
final = SvUV(*svp);
+ havefinal = TRUE;
+ }
- if (PL_op->op_private & OPpTRANS_GROWS) {
- I32 bits = 16;
-
- svp = hv_fetch(hv, "BITS", 4, FALSE);
- if (svp)
- bits = (I32)SvIV(*svp);
-
- Newz(801, d, len * (bits >> 3) + 1, U8);
- dst = d;
+ if (grows) {
+ /* d needs to be bigger than s, in case e.g. upgrading is required */
+ Newx(d, len * 3 + UTF8_MAXBYTES, U8);
+ dend = d + len * 3;
+ dstart = d;
}
else {
- d = s;
- dst = 0;
+ dstart = d = s;
+ dend = d + len;
}
if (squash) {
UV puv = 0xfeedface;
while (s < send) {
- if (from_utf) {
- 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);
+ UV uv = swash_fetch(rv, s, TRUE);
+
+ if (d > dend) {
+ const STRLEN clen = d - dstart;
+ const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
+ Renew(dstart, nlen + UTF8_MAXBYTES, U8);
+ d = dstart + clen;
+ dend = dstart + nlen;
}
if (uv < none) {
matches++;
+ s += UTF8SKIP(s);
if (uv != puv) {
- if (uv >= 0x80 && to_utf)
- d = uv_to_utf8(d, uv);
- else
- *d++ = (U8)uv;
+ d = uvuni_to_utf8(d, uv);
puv = uv;
}
- if (from_utf)
- s += UTF8SKIP(s);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
- if (from_utf) {
- if (*s < 0x80)
- *d++ = *s++;
- else if (to_utf) {
- int i;
- for (i = UTF8SKIP(s); i; --i)
- *d++ = *s++;
- }
- else {
- I32 ulen;
- *d++ = (U8)utf8_to_uv(s, &ulen);
- s += ulen;
- }
- }
- else { /* must be to_utf only */
- d = uv_to_utf8(d, s[-1]);
- }
+ const int i = UTF8SKIP(s);
+ Move(s, d, i, U8);
+ d += i;
+ s += i;
puv = 0xfeedface;
continue;
}
else if (uv == extra && !del) {
matches++;
- if (uv != puv) {
- if (final >= 0x80 && to_utf)
- d = uv_to_utf8(d, final);
- else
- *d++ = (U8)final;
- puv = final;
- }
- if (from_utf)
+ 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) {
+ Move(s, d, len, U8);
+ d += len;
+ puv = uv;
+ }
+ s += len;
+ }
continue;
}
- matches++; /* "none+1" is delete character */
- if (from_utf)
- s += UTF8SKIP(s);
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
}
}
else {
while (s < send) {
- if (from_utf) {
- 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);
+ const UV uv = swash_fetch(rv, s, TRUE);
+ if (d > dend) {
+ const STRLEN clen = d - dstart;
+ const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
+ if (!grows)
+ Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
+ Renew(dstart, nlen + UTF8_MAXBYTES, U8);
+ d = dstart + clen;
+ dend = dstart + nlen;
}
if (uv < none) {
matches++;
- if (uv >= 0x80 && to_utf)
- d = uv_to_utf8(d, uv);
- else
- *d++ = (U8)uv;
- if (from_utf)
- s += UTF8SKIP(s);
+ s += UTF8SKIP(s);
+ d = uvuni_to_utf8(d, uv);
continue;
}
else if (uv == none) { /* "none" is unmapped character */
- if (from_utf) {
- if (*s < 0x80)
- *d++ = *s++;
- else if (to_utf) {
- int i;
- for (i = UTF8SKIP(s); i; --i)
- *d++ = *s++;
- }
- else {
- I32 ulen;
- *d++ = (U8)utf8_to_uv(s, &ulen);
- s += ulen;
- }
- }
- else { /* must be to_utf only */
- d = uv_to_utf8(d, s[-1]);
- }
+ const int i = UTF8SKIP(s);
+ Move(s, d, i, U8);
+ d += i;
+ s += i;
continue;
}
else if (uv == extra && !del) {
matches++;
- if (final >= 0x80 && to_utf)
- d = uv_to_utf8(d, final);
- else
- *d++ = (U8)final;
- if (from_utf)
- s += UTF8SKIP(s);
+ s += UTF8SKIP(s);
+ d = uvuni_to_utf8(d, final);
continue;
}
- matches++; /* "none+1" is delete character */
- if (from_utf)
- s += UTF8SKIP(s);
+ matches++; /* "none+1" is delete character */
+ s += UTF8SKIP(s);
}
}
- if (dst)
- sv_usepvn(sv, (char*)dst, d - dst);
+ if (grows || hibit) {
+ sv_setpvn(sv, (char*)dstart, d - dstart);
+ Safefree(dstart);
+ if (grows && hibit)
+ Safefree(start);
+ }
else {
*d = '\0';
- SvCUR_set(sv, d - (U8*)SvPVX(sv));
+ SvCUR_set(sv, d - dstart);
}
+ SvUTF8_on(sv);
SvSETMAGIC(sv);
return matches;
}
I32
-do_trans(SV *sv)
+Perl_do_trans(pTHX_ SV *sv)
{
STRLEN len;
-
- if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
- croak(no_modify);
-
- (void)SvPV(sv, len);
+ const I32 hasutf = (PL_op->op_private &
+ (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
+
+ if (SvREADONLY(sv)) {
+ if (SvIsCOW(sv))
+ sv_force_normal_flags(sv, 0);
+ if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
+ Perl_croak(aTHX_ PL_no_modify);
+ }
+ (void)SvPV_const(sv, len);
if (!len)
return 0;
- if (!SvPOKp(sv))
- (void)SvPV_force(sv, len);
- (void)SvPOK_only(sv);
+ if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) {
+ if (!SvPOKp(sv))
+ (void)SvPV_force(sv, len);
+ (void)SvPOK_only_UTF8(sv);
+ }
- DEBUG_t( deb("2.TBL\n"));
+ DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
- switch (PL_op->op_private & 63) {
+ switch (PL_op->op_private & ~hasutf & (
+ OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL|
+ OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) {
case 0:
- return do_trans_CC_simple(sv);
-
- case OPpTRANS_FROM_UTF:
- return do_trans_UC_simple(sv);
-
- case OPpTRANS_TO_UTF:
- return do_trans_CU_simple(sv);
-
- case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF:
- return do_trans_UU_simple(sv);
+ if (hasutf)
+ return do_trans_simple_utf8(sv);
+ else
+ return do_trans_simple(sv);
case OPpTRANS_IDENTICAL:
- return do_trans_CC_count(sv);
-
- case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL:
- return do_trans_UC_trivial(sv);
-
- case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
- return do_trans_CU_trivial(sv);
-
- case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
- return do_trans_UU_count(sv);
+ case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
+ if (hasutf)
+ return do_trans_count_utf8(sv);
+ else
+ return do_trans_count(sv);
default:
- if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
- return do_trans_UU_complex(sv); /* could be UC or CU too */
+ if (hasutf)
+ return do_trans_complex_utf8(sv);
else
- return do_trans_CC_complex(sv);
+ return do_trans_complex(sv);
}
}
void
-do_join(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;
+ SV ** const oldmark = mark;
register I32 items = sp - mark;
register STRLEN len;
STRLEN delimlen;
- register char *delim = SvPV(del, delimlen);
- STRLEN tmplen;
+
+ (void) SvPV_const(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);
- if (SvTYPE(sv) < SVt_PV)
- sv_upgrade(sv, SVt_PV);
+ SvUPGRADE(sv, SVt_PV);
if (SvLEN(sv) < len + items) { /* current length is way too short */
while (items-- > 0) {
- if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
- SvPV(*mark, tmplen);
+ if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
+ STRLEN tmplen;
+ SvPV_const(*mark, tmplen);
len += tmplen;
}
mark++;
SvGROW(sv, len + 1); /* so try to pre-extend */
mark = oldmark;
- items = sp - mark;;
+ items = sp - mark;
++mark;
}
- if (items-- > 0) {
- char *s;
+ sv_setpvn(sv, "", 0);
+ /* sv_setpv retains old UTF8ness [perl #24846] */
+ SvUTF8_off(sv);
- if (*mark) {
- s = SvPV(*mark, tmplen);
- sv_setpvn(sv, s, tmplen);
- }
- else
- sv_setpv(sv, "");
+ if (PL_tainting && SvMAGICAL(sv))
+ SvTAINTED_off(sv);
+
+ if (items-- > 0) {
+ if (*mark)
+ sv_catsv(sv, *mark);
mark++;
}
- else
- sv_setpv(sv,"");
- len = delimlen;
- if (len) {
+
+ if (delimlen) {
for (; items > 0; items--,mark++) {
- sv_catpvn(sv,delim,len);
+ sv_catsv(sv,del);
sv_catsv(sv,*mark);
}
}
}
void
-do_sprintf(SV *sv, I32 len, SV **sarg)
+Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
{
STRLEN patlen;
- char *pat = SvPV(*sarg, 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);
SvSETMAGIC(sv);
if (do_taint)
SvTAINTED_on(sv);
}
+/* currently converts input to bytes if possible, but doesn't sweat failure */
+UV
+Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
+{
+ STRLEN srclen, len;
+ const unsigned char *s = (const unsigned char *) SvPV_const(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");
+
+ if (SvUTF8(sv))
+ (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
+
+ 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 ((STRLEN)offset >= srclen)
+ retnum = 0;
+ else
+ retnum = (UV) s[offset] << 8;
+ }
+ else if (size == 32) {
+ if ((STRLEN)offset >= srclen)
+ retnum = 0;
+ else if ((STRLEN)(offset + 1) >= srclen)
+ retnum =
+ ((UV) s[offset ] << 24);
+ else if ((STRLEN)(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) {
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ packWARN(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) {
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ packWARN(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;
+}
+
+/* currently converts input to bytes if possible but doesn't sweat failures,
+ * although it does ensure that the string it clobbers is not marked as
+ * utf8-valid any more
+ */
void
-do_vecset(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;
+ SV * const targ = LvTARG(sv);
if (!targ)
return;
s = (unsigned char*)SvPV_force(targ, targlen);
- lval = U_L(SvNV(sv));
+ if (SvUTF8(targ)) {
+ /* This is handled by the SvPOK_only below...
+ if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
+ SvUTF8_off(targ);
+ */
+ (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
+ }
+
+ (void)SvPOK_only(targ);
+ lval = SvUV(sv);
offset = LvTARGOFF(sv);
+ if (offset < 0)
+ Perl_croak(aTHX_ "Negative offset to vec in lvalue context");
size = LvTARGLEN(sv);
-
- len = (offset + size + 7) / 8;
+ 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 > 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);
}
-
+
if (size < 8) {
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 ] = (U8)( lval & 0xff);
else if (size == 16) {
- s[offset] = (lval >> 8) & 255;
- s[offset+1] = lval & 255;
+ s[offset ] = (U8)((lval >> 8) & 0xff);
+ s[offset+1] = (U8)( 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 ] = (U8)((lval >> 24) & 0xff);
+ s[offset+1] = (U8)((lval >> 16) & 0xff);
+ s[offset+2] = (U8)((lval >> 8) & 0xff);
+ s[offset+3] = (U8)( lval & 0xff);
+ }
+#ifdef UV_IS_QUAD
+ else if (size == 64) {
+ if (ckWARN(WARN_PORTABLE))
+ Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
+ "Bit vector size > 32 non-portable");
+ s[offset ] = (U8)((lval >> 56) & 0xff);
+ s[offset+1] = (U8)((lval >> 48) & 0xff);
+ s[offset+2] = (U8)((lval >> 40) & 0xff);
+ s[offset+3] = (U8)((lval >> 32) & 0xff);
+ s[offset+4] = (U8)((lval >> 24) & 0xff);
+ s[offset+5] = (U8)((lval >> 16) & 0xff);
+ s[offset+6] = (U8)((lval >> 8) & 0xff);
+ s[offset+7] = (U8)( lval & 0xff);
}
+#endif
}
+ SvSETMAGIC(targ);
}
void
-do_chop(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;
- I32 max;
- AV* av = (AV*)sv;
- max = AvFILL(av);
- for (i = 0; i <= max; i++) {
+ AV* const av = (AV*)sv;
+ const I32 max = AvFILL(av);
+
+ for (i = 0; i <= max; i++) {
sv = (SV*)av_fetch(av, i, FALSE);
if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
do_chop(astr, sv);
}
return;
}
- if (SvTYPE(sv) == SVt_PVHV) {
- HV* hv = (HV*)sv;
+ else if (SvTYPE(sv) == SVt_PVHV) {
+ HV* const 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)) {
+ if (SvFAKE(sv)) {
+ /* SV is copy-on-write */
+ sv_force_normal_flags(sv, 0);
+ }
+ if (SvREADONLY(sv))
+ Perl_croak(aTHX_ PL_no_modify);
+ }
s = SvPV(sv, len);
if (len && !SvPOK(sv))
s = SvPV_force(sv, len);
- if (IN_UTF8) {
+ 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 & 0xc0) == 0x80)
- --s;
- if (UTF8SKIP(s) != send - s)
- warn("Malformed UTF-8 character");
- sv_setpvn(astr, s, send - s);
- *s = '\0';
- SvCUR_set(sv, s - start);
- SvNIOK_off(sv);
+ while (s > start && UTF8_IS_CONTINUATION(*s))
+ s--;
+ if (is_utf8_string((U8*)s, send - s)) {
+ 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) {
+ 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(register SV *sv)
+Perl_do_chomp(pTHX_ register SV *sv)
{
- dTHR;
register I32 count;
STRLEN len;
char *s;
+ char *temp_buffer = NULL;
+ SV* svrecode = Nullsv;
if (RsSNARF(PL_rs))
return 0;
+ if (RsRECORD(PL_rs))
+ return 0;
count = 0;
if (SvTYPE(sv) == SVt_PVAV) {
register I32 i;
- I32 max;
- AV* av = (AV*)sv;
- max = AvFILL(av);
- for (i = 0; i <= max; i++) {
+ AV* const av = (AV*)sv;
+ const I32 max = AvFILL(av);
+
+ for (i = 0; i <= max; i++) {
sv = (SV*)av_fetch(av, i, FALSE);
if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
count += do_chomp(sv);
}
return count;
}
- if (SvTYPE(sv) == SVt_PVHV) {
- HV* hv = (HV*)sv;
+ else if (SvTYPE(sv) == SVt_PVHV) {
+ HV* const 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)) {
+ if (SvFAKE(sv)) {
+ /* SV is copy-on-write */
+ sv_force_normal_flags(sv, 0);
+ }
+ if (SvREADONLY(sv))
+ Perl_croak(aTHX_ PL_no_modify);
+ }
+
+ if (PL_encoding) {
+ if (!SvUTF8(sv)) {
+ /* XXX, here sv is utf8-ized as a side-effect!
+ If encoding.pm is used properly, almost string-generating
+ operations, including literal strings, chr(), input data, etc.
+ should have been utf8-ized already, right?
+ */
+ sv_recode_to_utf8(sv, PL_encoding);
+ }
+ }
+
s = SvPV(sv, len);
- if (len && !SvPOKp(sv))
- s = SvPV_force(sv, len);
if (s && len) {
s += --len;
if (RsPARA(PL_rs)) {
}
}
else {
- STRLEN rslen;
- char *rsptr = SvPV(PL_rs, rslen);
+ STRLEN rslen, rs_charlen;
+ const char *rsptr = SvPV_const(PL_rs, rslen);
+
+ rs_charlen = SvUTF8(PL_rs)
+ ? sv_len_utf8(PL_rs)
+ : rslen;
+
+ if (SvUTF8(PL_rs) != SvUTF8(sv)) {
+ /* Assumption is that rs is shorter than the scalar. */
+ if (SvUTF8(PL_rs)) {
+ /* RS is utf8, scalar is 8 bit. */
+ bool is_utf8 = TRUE;
+ temp_buffer = (char*)bytes_from_utf8((U8*)rsptr,
+ &rslen, &is_utf8);
+ if (is_utf8) {
+ /* Cannot downgrade, therefore cannot possibly match
+ */
+ assert (temp_buffer == rsptr);
+ temp_buffer = NULL;
+ goto nope;
+ }
+ rsptr = temp_buffer;
+ }
+ else if (PL_encoding) {
+ /* RS is 8 bit, encoding.pm is used.
+ * Do not recode PL_rs as a side-effect. */
+ svrecode = newSVpvn(rsptr, rslen);
+ sv_recode_to_utf8(svrecode, PL_encoding);
+ rsptr = SvPV_const(svrecode, rslen);
+ rs_charlen = sv_len_utf8(svrecode);
+ }
+ else {
+ /* RS is 8 bit, scalar is utf8. */
+ temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen);
+ rsptr = temp_buffer;
+ }
+ }
if (rslen == 1) {
if (*s != *rsptr)
goto nope;
s -= rslen - 1;
if (memNE(s, rsptr, rslen))
goto nope;
- count += rslen;
+ count += rs_charlen;
}
}
- *s = '\0';
+ s = SvPV_force_nolen(sv);
SvCUR_set(sv, len);
+ *SvEND(sv) = '\0';
SvNIOK_off(sv);
+ SvSETMAGIC(sv);
}
nope:
- SvSETMAGIC(sv);
+
+ if (svrecode)
+ SvREFCNT_dec(svrecode);
+
+ Safefree(temp_buffer);
return count;
-}
+}
void
-do_vop(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;
register char *dc;
STRLEN leftlen;
STRLEN rightlen;
- register char *lc;
- register char *rc;
+ register const char *lc;
+ register const char *rc;
register I32 len;
I32 lensave;
- char *lsave;
- char *rsave;
+ const char *lsave;
+ const char *rsave;
+ const bool left_utf = DO_UTF8(left);
+ const bool right_utf = DO_UTF8(right);
+ I32 needlen = 0;
+
+ if (left_utf && !right_utf)
+ sv_utf8_upgrade(right);
+ else 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 ^= */
- lsave = lc = SvPV(left, leftlen);
- rsave = rc = SvPV(right, rightlen);
+ lsave = lc = SvPV_nomg_const(left, leftlen);
+ rsave = rc = SvPV_nomg_const(right, rightlen);
len = leftlen < rightlen ? leftlen : rightlen;
lensave = len;
- if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
- dc = SvPV_force(sv, PL_na);
- if (SvCUR(sv) < len) {
- dc = SvGROW(sv, len + 1);
+ 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) {
+ 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);
}
+ 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));
- Newz(801, dc, needlen + 1, char);
+ needlen = ((optype == OP_BIT_AND)
+ ? len : (leftlen > rightlen ? leftlen : rightlen));
+ Newxz(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 * const dcsave = dc;
+ STRLEN lulen = leftlen;
+ STRLEN rulen = rightlen;
+ STRLEN ulen;
+
+ switch (optype) {
+ case OP_BIT_AND:
+ while (lulen && rulen) {
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc & ruc;
+ dc = (char*)uvchr_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 = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc ^ ruc;
+ dc = (char*)uvchr_to_utf8((U8*)dc, duc);
+ }
+ goto mop_up_utf;
+ case OP_BIT_OR:
+ while (lulen && rulen) {
+ luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
+ lc += ulen;
+ lulen -= ulen;
+ ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
+ rc += ulen;
+ rulen -= ulen;
+ duc = luc | ruc;
+ dc = (char*)uvchr_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)) &&
!((long)lc % sizeof(long)) &&
!((long)rc % sizeof(long))) /* It's almost always aligned... */
{
- I32 remainder = len % (sizeof(long)*4);
+ const I32 remainder = len % (sizeof(long)*4);
len /= (sizeof(long)*4);
dl = (long*)dc;
case OP_BIT_AND:
while (len--)
*dc++ = *lc++ & *rc++;
+ *dc = '\0';
break;
case OP_BIT_XOR:
while (len--)
*dc++ = *lc++ | *rc++;
mop_up:
len = lensave;
- if (rightlen > len)
+ if (rightlen > (STRLEN)len)
sv_catpvn(sv, rsave + len, rightlen - len);
- else if (leftlen > len)
+ else if (leftlen > (STRLEN)len)
sv_catpvn(sv, lsave + len, leftlen - len);
else
*SvEND(sv) = '\0';
break;
}
}
+finish:
SvTAINT(sv);
}
OP *
-do_kv(ARGSproto)
+Perl_do_kv(pTHX)
{
- djSP;
- HV *hv = (HV*)POPs;
+ dSP;
+ HV * const hv = (HV*)POPs;
HV *keys;
register HE *entry;
- SV *tmpstr;
- I32 gimme = GIMME_V;
- I32 dokeys = (PL_op->op_type == OP_KEYS);
- I32 dovalues = (PL_op->op_type == OP_VALUES);
- I32 realhv = (SvTYPE(hv) == SVt_PVHV);
-
- if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
- dokeys = dovalues = TRUE;
+ const I32 gimme = GIMME_V;
+ const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV);
+ const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS);
+ const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES);
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;
RETURN;
}
- keys = realhv ? hv : avhv_keys((AV*)hv);
+ keys = hv;
(void)hv_iterinit(keys); /* always reset iterator regardless */
if (gimme == G_VOID)
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);
+ sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
}
LvTYPE(TARG) = 'k';
if (LvTARG(TARG) != (SV*)keys) {
RETURN;
}
- if (!SvRMAGICAL(keys) || !mg_find((SV*)keys,'P'))
+ if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied))
i = HvKEYS(keys);
else {
i = 0;
- /*SUPPRESS 560*/
while (hv_iternext(keys)) i++;
}
PUSHi( i );
EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
- while (entry = hv_iternext(keys)) {
+ while ((entry = hv_iternext(keys))) {
SPAGAIN;
- if (dokeys)
- XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
+ if (dokeys) {
+ SV* const sv = hv_iterkeysv(entry);
+ XPUSHs(sv); /* won't clobber stack_sp */
+ }
if (dovalues) {
- tmpstr = sv_newmortal();
+ SV *tmpstr;
PUTBACK;
- sv_setsv(tmpstr,realhv ?
- hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry));
- DEBUG_H(sv_setpvf(tmpstr, "%lu%%%d=%lu",
+ tmpstr = hv_iterval(hv,entry);
+ DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
(unsigned long)HeHASH(entry),
- HvMAX(keys)+1,
+ (int)HvMAX(keys)+1,
(unsigned long)(HeHASH(entry) & HvMAX(keys))));
SPAGAIN;
XPUSHs(tmpstr);
return NORMAL;
}
+/*
+ * Local variables:
+ * c-indentation-style: bsd
+ * c-basic-offset: 4
+ * indent-tabs-mode: t
+ * End:
+ *
+ * ex: set ts=8 sts=4 sw=4 noet:
+ */
https://perl5.git.perl.org / perl5.git / blobdiff