X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/9f7f391326e967b539b86ed051c163bbf8f6e7de..c6eb6e8a70b63dbff935e4cbb2083cc52bcf7605:/doop.c diff --git a/doop.c b/doop.c index 3e60665..8bd7c0f 100644 --- a/doop.c +++ b/doop.c @@ -1,7 +1,7 @@ /* doop.c * * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, - * 2000, 2001, 2002, 2004, 2005, 2006, by Larry Wall and others + * 2000, 2001, 2002, 2004, 2005, 2006, 2007, 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. @@ -26,24 +26,20 @@ #endif STATIC I32 -S_do_trans_simple(pTHX_ SV *sv) +S_do_trans_simple(pTHX_ SV * const sv) { dVAR; - U8 *s; - U8 *d; - const U8 *send; - U8 *dstart; 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 * const tbl = (short*)cPVOP->op_pv; + + PERL_ARGS_ASSERT_DO_TRANS_SIMPLE; + 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) { @@ -55,66 +51,70 @@ S_do_trans_simple(pTHX_ SV *sv) 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, 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); + 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) { dVAR; - const U8 *s; - const U8 *send; - I32 matches = 0; STRLEN len; - + 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; + + PERL_ARGS_ASSERT_DO_TRANS_COUNT; + 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 { const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; while (s < send) { @@ -133,27 +133,24 @@ S_do_trans_count(pTHX_ SV *sv) } STATIC I32 -S_do_trans_complex(pTHX_ SV *sv) +S_do_trans_complex(pTHX_ SV * const sv) { dVAR; - U8 *s; - U8 *send; - U8 *d; - U8 *dstart; - I32 isutf8; + STRLEN len; + U8 *s = (U8*)SvPV(sv, len); + U8 * const send = s+len; I32 matches = 0; - STRLEN len, rlen = 0; - const short * const tbl = (short*)cPVOP->op_pv; + + PERL_ARGS_ASSERT_DO_TRANS_COMPLEX; + 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) { @@ -188,10 +185,13 @@ S_do_trans_complex(pTHX_ SV *sv) *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); @@ -215,13 +215,13 @@ S_do_trans_complex(pTHX_ SV *sv) if (comp > 0xff) { if (!complement) { - Copy(s, d, len, U8); + Move(s, d, len, U8); d += len; } else { matches++; if (!del) { - ch = (rlen == 0) ? comp : + ch = (rlen == 0) ? (I32)comp : (comp - 0x100 < rlen) ? tbl[comp+1] : tbl[0x100+rlen]; if ((UV)ch != pch) { @@ -243,7 +243,7 @@ S_do_trans_complex(pTHX_ SV *sv) 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 */ @@ -278,7 +278,7 @@ S_do_trans_complex(pTHX_ SV *sv) 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 */ @@ -301,7 +301,7 @@ S_do_trans_complex(pTHX_ SV *sv) } STATIC I32 -S_do_trans_simple_utf8(pTHX_ SV *sv) +S_do_trans_simple_utf8(pTHX_ SV * const sv) { dVAR; U8 *s; @@ -312,19 +312,23 @@ S_do_trans_simple_utf8(pTHX_ SV *sv) I32 matches = 0; const I32 grows = PL_op->op_private & OPpTRANS_GROWS; STRLEN len; - - SV* const rv = (SV*)cSVOP->op_sv; + SV* const rv = +#ifdef USE_ITHREADS + PAD_SVl(cPADOP->op_padix); +#else + (SV*)cSVOP->op_sv; +#endif HV* const hv = (HV*)SvRV(rv); SV* const * svp = hv_fetchs(hv, "NONE", FALSE); const UV none = svp ? SvUV(*svp) : 0x7fffffff; const UV extra = none + 1; UV final = 0; - I32 isutf8; U8 hibit = 0; + PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8; + s = (U8*)SvPV(sv, len); - isutf8 = SvUTF8(sv); - if (!isutf8) { + if (!SvUTF8(sv)) { const U8 *t = s; const U8 * const e = s + len; while (t < e) { @@ -402,7 +406,7 @@ S_do_trans_simple_utf8(pTHX_ SV *sv) } STATIC I32 -S_do_trans_count_utf8(pTHX_ SV *sv) +S_do_trans_count_utf8(pTHX_ SV * const sv) { dVAR; const U8 *s; @@ -410,14 +414,20 @@ S_do_trans_count_utf8(pTHX_ SV *sv) const U8 *send; I32 matches = 0; STRLEN len; - - SV* const rv = (SV*)cSVOP->op_sv; + SV* const rv = +#ifdef USE_ITHREADS + PAD_SVl(cPADOP->op_padix); +#else + (SV*)cSVOP->op_sv; +#endif HV* const hv = (HV*)SvRV(rv); SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); const UV none = svp ? SvUV(*svp) : 0x7fffffff; const UV extra = none + 1; U8 hibit = 0; + PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8; + s = (const U8*)SvPV_const(sv, len); if (!SvUTF8(sv)) { const U8 *t = s; @@ -446,7 +456,7 @@ S_do_trans_count_utf8(pTHX_ SV *sv) } STATIC I32 -S_do_trans_complex_utf8(pTHX_ SV *sv) +S_do_trans_complex_utf8(pTHX_ SV * const sv) { dVAR; U8 *start, *send; @@ -455,7 +465,12 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) 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; + SV* const rv = +#ifdef USE_ITHREADS + PAD_SVl(cPADOP->op_padix); +#else + (SV*)cSVOP->op_sv; +#endif HV * const hv = (HV*)SvRV(rv); SV * const *svp = hv_fetchs(hv, "NONE", FALSE); const UV none = svp ? SvUV(*svp) : 0x7fffffff; @@ -465,10 +480,11 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) STRLEN len; U8 *dstart, *dend; U8 hibit = 0; - U8 *s = (U8*)SvPV(sv, len); - const I32 isutf8 = SvUTF8(sv); - if (!isutf8) { + + PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8; + + if (!SvUTF8(sv)) { const U8 *t = s; const U8 * const e = s + len; while (t < e) { @@ -615,10 +631,12 @@ Perl_do_trans(pTHX_ SV *sv) const I32 hasutf = (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); - if (SvREADONLY(sv)) { + PERL_ARGS_ASSERT_DO_TRANS; + + if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) { if (SvIsCOW(sv)) sv_force_normal_flags(sv, 0); - if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) + if (SvREADONLY(sv)) Perl_croak(aTHX_ PL_no_modify); } (void)SvPV_const(sv, len); @@ -657,7 +675,7 @@ Perl_do_trans(pTHX_ SV *sv) } void -Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) +Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp) { dVAR; SV ** const oldmark = mark; @@ -665,7 +683,9 @@ Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **s register STRLEN len; STRLEN delimlen; - (void) SvPV_const(del, delimlen); /* stringify and get the delimlen */ + PERL_ARGS_ASSERT_DO_JOIN; + + (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */ /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ mark++; @@ -702,7 +722,7 @@ Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **s if (delimlen) { for (; items > 0; items--,mark++) { - sv_catsv(sv,del); + sv_catsv(sv,delim); sv_catsv(sv,*mark); } } @@ -721,6 +741,8 @@ Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) const char * const pat = SvPV_const(*sarg, patlen); bool do_taint = FALSE; + PERL_ARGS_ASSERT_DO_SPRINTF; + SvUTF8_off(sv); if (DO_UTF8(*sarg)) SvUTF8_on(sv); @@ -735,130 +757,138 @@ UV Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) { dVAR; - STRLEN srclen, len; + STRLEN srclen, len, uoffset, bitoffs = 0; const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); UV retnum = 0; + PERL_ARGS_ASSERT_DO_VECGET; + if (offset < 0) - return retnum; + return 0; 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 (size < 8) { + bitoffs = ((offset%8)*size)%8; + uoffset = offset/(8/size); + } + else if (size > 8) + uoffset = offset*(size/8); + else + uoffset = offset; + + len = uoffset + (bitoffs + 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) + if (uoffset >= srclen) retnum = 0; else - retnum = (UV) s[offset] << 8; + retnum = (UV) s[uoffset] << 8; } else if (size == 32) { - if ((STRLEN)offset >= srclen) + if (uoffset >= srclen) retnum = 0; - else if ((STRLEN)(offset + 1) >= srclen) + else if (uoffset + 1 >= srclen) retnum = - ((UV) s[offset ] << 24); - else if ((STRLEN)(offset + 2) >= srclen) + ((UV) s[uoffset ] << 24); + else if (uoffset + 2 >= srclen) retnum = - ((UV) s[offset ] << 24) + - ((UV) s[offset + 1] << 16); + ((UV) s[uoffset ] << 24) + + ((UV) s[uoffset + 1] << 16); else retnum = - ((UV) s[offset ] << 24) + - ((UV) s[offset + 1] << 16) + - ( s[offset + 2] << 8); + ((UV) s[uoffset ] << 24) + + ((UV) s[uoffset + 1] << 16) + + ( s[uoffset + 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) + if (uoffset >= srclen) retnum = 0; - else if (offset + 1 >= srclen) + else if (uoffset + 1 >= srclen) retnum = - (UV) s[offset ] << 56; - else if (offset + 2 >= srclen) + (UV) s[uoffset ] << 56; + else if (uoffset + 2 >= srclen) retnum = - ((UV) s[offset ] << 56) + - ((UV) s[offset + 1] << 48); - else if (offset + 3 >= srclen) + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48); + else if (uoffset + 3 >= srclen) retnum = - ((UV) s[offset ] << 56) + - ((UV) s[offset + 1] << 48) + - ((UV) s[offset + 2] << 40); - else if (offset + 4 >= srclen) + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40); + else if (uoffset + 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) + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40) + + ((UV) s[uoffset + 3] << 32); + else if (uoffset + 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) + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40) + + ((UV) s[uoffset + 3] << 32) + + ( s[uoffset + 4] << 24); + else if (uoffset + 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); + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40) + + ((UV) s[uoffset + 3] << 32) + + ((UV) s[uoffset + 4] << 24) + + ((UV) s[uoffset + 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); + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40) + + ((UV) s[uoffset + 3] << 32) + + ((UV) s[uoffset + 4] << 24) + + ((UV) s[uoffset + 5] << 16) + + ( s[uoffset + 6] << 8); } #endif } } else if (size < 8) - retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); + retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1); else { - offset >>= 3; /* turn into byte offset */ if (size == 8) - retnum = s[offset]; + retnum = s[uoffset]; else if (size == 16) retnum = - ((UV) s[offset] << 8) + - s[offset + 1]; + ((UV) s[uoffset] << 8) + + s[uoffset + 1]; else if (size == 32) retnum = - ((UV) s[offset ] << 24) + - ((UV) s[offset + 1] << 16) + - ( s[offset + 2] << 8) + - s[offset + 3]; + ((UV) s[uoffset ] << 24) + + ((UV) s[uoffset + 1] << 16) + + ( s[uoffset + 2] << 8) + + s[uoffset + 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]; + ((UV) s[uoffset ] << 56) + + ((UV) s[uoffset + 1] << 48) + + ((UV) s[uoffset + 2] << 40) + + ((UV) s[uoffset + 3] << 32) + + ((UV) s[uoffset + 4] << 24) + + ((UV) s[uoffset + 5] << 16) + + ( s[uoffset + 6] << 8) + + s[uoffset + 7]; } #endif } @@ -874,7 +904,7 @@ void Perl_do_vecset(pTHX_ SV *sv) { dVAR; - register I32 offset; + register I32 offset, bitoffs = 0; register I32 size; register unsigned char *s; register UV lval; @@ -883,6 +913,8 @@ Perl_do_vecset(pTHX_ SV *sv) STRLEN len; SV * const targ = LvTARG(sv); + PERL_ARGS_ASSERT_DO_VECSET; + if (!targ) return; s = (unsigned char*)SvPV_force(targ, targlen); @@ -903,8 +935,14 @@ Perl_do_vecset(pTHX_ SV *sv) 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 (size < 8) { + bitoffs = ((offset%8)*size)%8; + offset /= 8/size; + } + else if (size > 8) + offset *= size/8; + + len = offset + (bitoffs + size + 7)/8; /* required number of bytes */ if (len > targlen) { s = (unsigned char*)SvGROW(targ, len + 1); (void)memzero((char *)(s + targlen), len - targlen + 1); @@ -913,14 +951,11 @@ Perl_do_vecset(pTHX_ SV *sv) if (size < 8) { mask = (1 << size) - 1; - size = offset & 7; lval &= mask; - offset >>= 3; /* turn into byte offset */ - s[offset] &= ~(mask << size); - s[offset] |= lval << size; + s[offset] &= ~(mask << bitoffs); + s[offset] |= lval << bitoffs; } else { - offset >>= 3; /* turn into byte offset */ if (size == 8) s[offset ] = (U8)( lval & 0xff); else if (size == 16) { @@ -959,6 +994,8 @@ Perl_do_chop(pTHX_ register SV *astr, register SV *sv) STRLEN len; char *s; + PERL_ARGS_ASSERT_DO_CHOP; + if (SvTYPE(sv) == SVt_PVAV) { register I32 i; AV* const av = (AV*)sv; @@ -1038,6 +1075,8 @@ Perl_do_chomp(pTHX_ register SV *sv) char *temp_buffer = NULL; SV* svrecode = NULL; + PERL_ARGS_ASSERT_DO_CHOMP; + if (RsSNARF(PL_rs)) return 0; if (RsRECORD(PL_rs)) @@ -1178,50 +1217,71 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) STRLEN rightlen; register const char *lc; register const char *rc; - register I32 len; - I32 lensave; + register STRLEN len; + STRLEN lensave; const char *lsave; const char *rsave; - const bool left_utf = DO_UTF8(left); - const bool right_utf = DO_UTF8(right); - I32 needlen = 0; + bool left_utf; + bool right_utf; + STRLEN needlen = 0; - if (left_utf && !right_utf) - sv_utf8_upgrade(right); - else if (!left_utf && right_utf) - sv_utf8_upgrade(left); + PERL_ARGS_ASSERT_DO_VOP; if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ lsave = lc = SvPV_nomg_const(left, leftlen); rsave = rc = SvPV_nomg_const(right, rightlen); + + /* This need to come after SvPV to ensure that string overloading has + fired off. */ + + left_utf = DO_UTF8(left); + right_utf = DO_UTF8(right); + + if (left_utf && !right_utf) { + /* Avoid triggering overloading again by using temporaries. + Maybe there should be a variant of sv_utf8_upgrade that takes pvn + */ + right = newSVpvn_flags(rsave, rightlen, SVs_TEMP); + sv_utf8_upgrade(right); + rsave = rc = SvPV_nomg_const(right, rightlen); + right_utf = TRUE; + } + else if (!left_utf && right_utf) { + left = newSVpvn_flags(lsave, leftlen, SVs_TEMP); + sv_utf8_upgrade(left); + lsave = lc = SvPV_nomg_const(left, leftlen); + left_utf = TRUE; + } + 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) { dc = SvPV_force_nomg_nolen(sv); - if (SvLEN(sv) < (STRLEN)(len + 1)) { - dc = SvGROW(sv, (STRLEN)(len + 1)); + if (SvLEN(sv) < len + 1) { + 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 { - needlen = ((optype == OP_BIT_AND) - ? len : (leftlen > rightlen ? leftlen : rightlen)); + 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 * const dcsave = dc; + char *dcorig = dc; + char *dcsave = NULL; STRLEN lulen = leftlen; STRLEN rulen = rightlen; STRLEN ulen; @@ -1239,8 +1299,8 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) 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) { @@ -1266,16 +1326,26 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) 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; @@ -1283,11 +1353,11 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) 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... */ + !((unsigned long)dc % sizeof(long)) && + !((unsigned long)lc % sizeof(long)) && + !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ { - const I32 remainder = len % (sizeof(long)*4); + const STRLEN remainder = len % (sizeof(long)*4); len /= (sizeof(long)*4); dl = (long*)dc; @@ -1343,7 +1413,7 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) *dc++ = *lc++ | *rc++; mop_up: len = lensave; - if (rightlen > (STRLEN)len) + if (rightlen > len) sv_catpvn(sv, rsave + len, rightlen - len); else if (leftlen > (STRLEN)len) sv_catpvn(sv, lsave + len, leftlen - len); @@ -1404,8 +1474,10 @@ Perl_do_kv(pTHX) RETURN; } - if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) + if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied) ) + { i = HvKEYS(keys); + } else { i = 0; while (hv_iternext(keys)) i++;