X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/8973db79328a885c91b9dfdcafdb28dbe9e65a88..3351db0262a00327940d490d1d2727273740be19:/doop.c diff --git a/doop.c b/doop.c index 7e2b52f..20379a9 100644 --- a/doop.c +++ b/doop.c @@ -1,6 +1,6 @@ /* doop.c * - * Copyright (c) 1991-2001, Larry Wall + * Copyright (c) 1991-2002, 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. @@ -36,7 +36,7 @@ S_do_trans_simple(pTHX_ SV *sv) tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "panic: do_trans_simple"); + Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); s = (U8*)SvPV(sv, len); send = s + len; @@ -66,10 +66,10 @@ S_do_trans_simple(pTHX_ SV *sv) UV c; /* Need to check this, otherwise 128..255 won't match */ - c = utf8_to_uv(s, send - s, &ulen, 0); + c = utf8n_to_uvchr(s, send - s, &ulen, 0); if (c < 0x100 && (ch = tbl[c]) >= 0) { matches++; - d = uv_to_utf8(d, ch); + d = uvchr_to_utf8(d, ch); s += ulen; } else { /* No match -> copy */ @@ -92,7 +92,7 @@ S_do_trans_simple(pTHX_ SV *sv) } STATIC I32 -S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ +S_do_trans_count(pTHX_ SV *sv) { U8 *s; U8 *send; @@ -103,7 +103,7 @@ S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "panic: do_trans_count"); + Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); s = (U8*)SvPV(sv, len); send = s + len; @@ -117,7 +117,7 @@ S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ while (s < send) { UV c; STRLEN ulen; - c = utf8_to_uv(s, send - s, &ulen, 0); + c = utf8n_to_uvchr(s, send - s, &ulen, 0); if (c < 0x100) { if (tbl[c] >= 0) matches++; @@ -130,7 +130,7 @@ S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ } STATIC I32 -S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ +S_do_trans_complex(pTHX_ SV *sv) { U8 *s; U8 *send; @@ -141,13 +141,13 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ I32 grows = PL_op->op_private & OPpTRANS_GROWS; I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; I32 del = PL_op->op_private & OPpTRANS_DELETE; - STRLEN len, rlen; + STRLEN len, rlen = 0; short *tbl; I32 ch; tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "panic: do_trans_complex"); + Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); s = (U8*)SvPV(sv, len); isutf8 = SvUTF8(sv); @@ -184,6 +184,7 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ s++; } } + *d = '\0'; SvCUR_set(sv, d - dstart); } else { /* isutf8 */ @@ -203,7 +204,7 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ UV pch = 0xfeedface; while (s < send) { STRLEN len; - UV comp = utf8_to_uv_simple(s, &len); + UV comp = utf8_to_uvchr(s, &len); if (comp > 0xff) { if (!complement) { @@ -213,10 +214,11 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ else { matches++; if (!del) { - ch = (comp - 0x100 < rlen) ? + ch = (rlen == 0) ? comp : + (comp - 0x100 < rlen) ? tbl[comp+1] : tbl[0x100+rlen]; if (ch != pch) { - d = uv_to_utf8(d, ch); + d = uvchr_to_utf8(d, ch); pch = ch; } s += len; @@ -227,7 +229,7 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ else if ((ch = tbl[comp]) >= 0) { matches++; if (ch != pch) { - d = uv_to_utf8(d, ch); + d = uvchr_to_utf8(d, ch); pch = ch; } s += len; @@ -246,7 +248,7 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ else { while (s < send) { STRLEN len; - UV comp = utf8_to_uv_simple(s, &len); + UV comp = utf8_to_uvchr(s, &len); if (comp > 0xff) { if (!complement) { Copy(s, d, len, U8); @@ -255,15 +257,15 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ else { matches++; if (!del) { - if (comp - 0x100 < rlen) - d = uv_to_utf8(d, tbl[comp+1]); + if (comp - 0x100 < rlen) + d = uvchr_to_utf8(d, tbl[comp+1]); else - d = uv_to_utf8(d, tbl[0x100+rlen]); + d = uvchr_to_utf8(d, tbl[0x100+rlen]); } } } else if ((ch = tbl[comp]) >= 0) { - d = uv_to_utf8(d, ch); + d = uvchr_to_utf8(d, ch); matches++; } else if (ch == -1) { /* -1 is unmapped character */ @@ -290,7 +292,7 @@ S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ } STATIC I32 -S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ +S_do_trans_simple_utf8(pTHX_ SV *sv) { U8 *s; U8 *send; @@ -306,7 +308,7 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ SV** svp = hv_fetch(hv, "NONE", 4, FALSE); UV none = svp ? SvUV(*svp) : 0x7fffffff; UV extra = none + 1; - UV final; + UV final = 0; UV uv; I32 isutf8; U8 hibit = 0; @@ -315,9 +317,11 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ isutf8 = SvUTF8(sv); if (!isutf8) { U8 *t = s, *e = s + len; - while (t < e) - if ((hibit = UTF8_IS_CONTINUED(*t++))) + while (t < e) { + U8 ch = *t++; + if ((hibit = !NATIVE_IS_INVARIANT(ch))) break; + } if (hibit) s = bytes_to_utf8(s, &len); } @@ -340,10 +344,10 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ } while (s < send) { - if ((uv = swash_fetch(rv, s)) < none) { + if ((uv = swash_fetch(rv, s, TRUE)) < none) { s += UTF8SKIP(s); matches++; - d = uv_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); } else if (uv == none) { int i = UTF8SKIP(s); @@ -355,7 +359,7 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ int i = UTF8SKIP(s); s += i; matches++; - d = uv_to_utf8(d, final); + d = uvuni_to_utf8(d, final); } else s += UTF8SKIP(s); @@ -364,7 +368,7 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ STRLEN clen = d - dstart; STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; if (!grows) - Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); + Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); Renew(dstart, nlen+UTF8_MAXLEN, U8); d = dstart + clen; dend = dstart + nlen; @@ -382,17 +386,15 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ } SvSETMAGIC(sv); SvUTF8_on(sv); - if (!isutf8 && !(PL_hints & HINT_UTF8)) - sv_utf8_downgrade(sv, TRUE); return matches; } STATIC I32 -S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ +S_do_trans_count_utf8(pTHX_ SV *sv) { U8 *s; - U8 *start, *send; + U8 *start = 0, *send; I32 matches = 0; STRLEN len; @@ -400,22 +402,25 @@ S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ HV* hv = (HV*)SvRV(rv); SV** svp = hv_fetch(hv, "NONE", 4, FALSE); UV none = svp ? SvUV(*svp) : 0x7fffffff; + UV extra = none + 1; UV uv; U8 hibit = 0; s = (U8*)SvPV(sv, len); if (!SvUTF8(sv)) { U8 *t = s, *e = s + len; - while (t < e) - if ((hibit = !UTF8_IS_ASCII(*t++))) + while (t < e) { + U8 ch = *t++; + if ((hibit = !NATIVE_IS_INVARIANT(ch))) break; + } if (hibit) start = s = bytes_to_utf8(s, &len); } send = s + len; while (s < send) { - if ((uv = swash_fetch(rv, s)) < none) + if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) matches++; s += UTF8SKIP(s); } @@ -426,7 +431,7 @@ S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ } STATIC I32 -S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ +S_do_trans_complex_utf8(pTHX_ SV *sv) { U8 *s; U8 *start, *send; @@ -440,7 +445,8 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ SV** svp = hv_fetch(hv, "NONE", 4, FALSE); UV none = svp ? SvUV(*svp) : 0x7fffffff; UV extra = none + 1; - UV final; + UV final = 0; + bool havefinal = FALSE; UV uv; STRLEN len; U8 *dstart, *dend; @@ -451,9 +457,11 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ isutf8 = SvUTF8(sv); if (!isutf8) { U8 *t = s, *e = s + len; - while (t < e) - if ((hibit = !UTF8_IS_ASCII(*t++))) + while (t < e) { + U8 ch = *t++; + if ((hibit = !NATIVE_IS_INVARIANT(ch))) break; + } if (hibit) s = bytes_to_utf8(s, &len); } @@ -461,8 +469,10 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ start = s; svp = hv_fetch(hv, "FINAL", 5, FALSE); - if (svp) + if (svp) { final = SvUV(*svp); + havefinal = TRUE; + } if (grows) { /* d needs to be bigger than s, in case e.g. upgrading is required */ @@ -478,24 +488,24 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ if (squash) { UV puv = 0xfeedface; while (s < send) { - uv = swash_fetch(rv, s); - + uv = swash_fetch(rv, s, TRUE); + if (d > dend) { STRLEN clen = d - dstart; STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; if (!grows) - Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); + Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); Renew(dstart, nlen+UTF8_MAXLEN, U8); d = dstart + clen; dend = dstart + nlen; } if (uv < none) { matches++; + s += UTF8SKIP(s); if (uv != puv) { - d = uv_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); puv = uv; } - s += UTF8SKIP(s); continue; } else if (uv == none) { /* "none" is unmapped character */ @@ -508,11 +518,23 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ } else if (uv == extra && !del) { matches++; - if (uv != puv) { - d = uv_to_utf8(d, final); - puv = final; + if (havefinal) { + s += UTF8SKIP(s); + if (puv != final) { + d = uvuni_to_utf8(d, final); + puv = final; + } + } + else { + STRLEN len; + uv = utf8_to_uvuni(s, &len); + if (uv != puv) { + Copy(s, d, len, U8); + d += len; + puv = uv; + } + s += len; } - s += UTF8SKIP(s); continue; } matches++; /* "none+1" is delete character */ @@ -521,12 +543,12 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ } else { while (s < send) { - uv = swash_fetch(rv, s); + uv = swash_fetch(rv, s, TRUE); if (d > dend) { STRLEN clen = d - dstart; STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; if (!grows) - Perl_croak(aTHX_ "panic: do_trans_complex_utf8"); + Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); Renew(dstart, nlen+UTF8_MAXLEN, U8); d = dstart + clen; dend = dstart + nlen; @@ -534,7 +556,7 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ if (uv < none) { matches++; s += UTF8SKIP(s); - d = uv_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); continue; } else if (uv == none) { /* "none" is unmapped character */ @@ -546,8 +568,8 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ } else if (uv == extra && !del) { matches++; - d = uv_to_utf8(d, final); s += UTF8SKIP(s); + d = uvuni_to_utf8(d, final); continue; } matches++; /* "none+1" is delete character */ @@ -565,8 +587,6 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ SvCUR_set(sv, d - dstart); } SvUTF8_on(sv); - if (!isutf8 && !(PL_hints & HINT_UTF8)) - sv_utf8_downgrade(sv, TRUE); SvSETMAGIC(sv); return matches; @@ -579,9 +599,12 @@ Perl_do_trans(pTHX_ SV *sv) 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); - + if (SvREADONLY(sv)) { + if (SvFAKE(sv)) + sv_force_normal(sv); + if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) + Perl_croak(aTHX_ PL_no_modify); + } (void)SvPV(sv, len); if (!len) return 0; @@ -600,6 +623,7 @@ Perl_do_trans(pTHX_ SV *sv) return do_trans_simple(sv); case OPpTRANS_IDENTICAL: + case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: if (hasutf) return do_trans_count_utf8(sv); else @@ -620,9 +644,11 @@ Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **s register I32 items = sp - mark; register STRLEN len; STRLEN delimlen; - register char *delim = SvPV(del, delimlen); STRLEN tmplen; + (void) SvPV(del, delimlen); /* stringify and get the delimlen */ + /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ + mark++; len = (items > 0 ? (delimlen * (items - 1) ) : 0); (void)SvUPGRADE(sv, SVt_PV); @@ -641,14 +667,16 @@ Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **s ++mark; } + sv_setpv(sv, ""); + if (PL_tainting && SvMAGICAL(sv)) + SvTAINTED_off(sv); + if (items-- > 0) { - sv_setpv(sv, ""); if (*mark) sv_catsv(sv, *mark); mark++; } - else - sv_setpv(sv,""); + if (delimlen) { for (; items > 0; items--,mark++) { sv_catsv(sv,del); @@ -669,6 +697,9 @@ Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) char *pat = SvPV(*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) @@ -723,7 +754,7 @@ Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) #ifdef UV_IS_QUAD else if (size == 64) { if (ckWARN(WARN_PORTABLE)) - Perl_warner(aTHX_ WARN_PORTABLE, + Perl_warner(aTHX_ packWARN(WARN_PORTABLE), "Bit vector size > 32 non-portable"); if (offset >= srclen) retnum = 0; @@ -792,7 +823,7 @@ Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) #ifdef UV_IS_QUAD else if (size == 64) { if (ckWARN(WARN_PORTABLE)) - Perl_warner(aTHX_ WARN_PORTABLE, + Perl_warner(aTHX_ packWARN(WARN_PORTABLE), "Bit vector size > 32 non-portable"); retnum = ((UV) s[offset ] << 56) + @@ -841,7 +872,7 @@ Perl_do_vecset(pTHX_ SV *sv) lval = SvUV(sv); offset = LvTARGOFF(sv); if (offset < 0) - Perl_croak(aTHX_ "Assigning to negative offset in vec"); + Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); 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"); @@ -879,7 +910,7 @@ Perl_do_vecset(pTHX_ SV *sv) #ifdef UV_IS_QUAD else if (size == 64) { if (ckWARN(WARN_PORTABLE)) - Perl_warner(aTHX_ WARN_PORTABLE, + Perl_warner(aTHX_ packWARN(WARN_PORTABLE), "Bit vector size > 32 non-portable"); s[offset ] = (lval >> 56) & 0xff; s[offset+1] = (lval >> 48) & 0xff; @@ -922,8 +953,14 @@ Perl_do_chop(pTHX_ register SV *astr, register SV *sv) do_chop(astr,hv_iterval(hv,entry)); return; } - else if (SvREADONLY(sv)) - Perl_croak(aTHX_ PL_no_modify); + 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); @@ -934,7 +971,7 @@ Perl_do_chop(pTHX_ register SV *astr, register SV *sv) s = send - 1; while (s > start && UTF8_IS_CONTINUATION(*s)) s--; - if (utf8_to_uv_simple((U8*)s, 0)) { + if (utf8_to_uvchr((U8*)s, 0)) { sv_setpvn(astr, s, send - s); *s = '\0'; SvCUR_set(sv, s - start); @@ -963,6 +1000,7 @@ Perl_do_chomp(pTHX_ register SV *sv) { register I32 count; STRLEN len; + STRLEN n_a; char *s; if (RsSNARF(PL_rs)) @@ -991,11 +1029,15 @@ Perl_do_chomp(pTHX_ register SV *sv) count += do_chomp(hv_iterval(hv,entry)); return count; } - else if (SvREADONLY(sv)) - Perl_croak(aTHX_ PL_no_modify); + 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 && !SvPOKp(sv)) - s = SvPV_force(sv, len); if (s && len) { s += --len; if (RsPARA(PL_rs)) { @@ -1026,12 +1068,13 @@ Perl_do_chomp(pTHX_ register SV *sv) count += rslen; } } - *s = '\0'; + s = SvPV_force(sv, n_a); SvCUR_set(sv, len); + *SvEND(sv) = '\0'; SvNIOK_off(sv); + SvSETMAGIC(sv); } nope: - SvSETMAGIC(sv); return count; } @@ -1054,7 +1097,7 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) char *rsave; bool left_utf = DO_UTF8(left); bool right_utf = DO_UTF8(right); - I32 needlen; + I32 needlen = 0; if (left_utf && !right_utf) sv_utf8_upgrade(right); @@ -1100,14 +1143,14 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) switch (optype) { case OP_BIT_AND: while (lulen && rulen) { - luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); + luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); lc += ulen; lulen -= ulen; - ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); + ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); rc += ulen; rulen -= ulen; duc = luc & ruc; - dc = (char*)uv_to_utf8((U8*)dc, duc); + dc = (char*)uvchr_to_utf8((U8*)dc, duc); } if (sv == left || sv == right) (void)sv_usepvn(sv, dcsave, needlen); @@ -1115,26 +1158,26 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) break; case OP_BIT_XOR: while (lulen && rulen) { - luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); + luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); lc += ulen; lulen -= ulen; - ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); + ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); rc += ulen; rulen -= ulen; duc = luc ^ ruc; - dc = (char*)uv_to_utf8((U8*)dc, duc); + dc = (char*)uvchr_to_utf8((U8*)dc, duc); } goto mop_up_utf; case OP_BIT_OR: while (lulen && rulen) { - luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); + luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); lc += ulen; lulen -= ulen; - ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); + ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); rc += ulen; rulen -= ulen; duc = luc | ruc; - dc = (char*)uv_to_utf8((U8*)dc, duc); + dc = (char*)uvchr_to_utf8((U8*)dc, duc); } mop_up_utf: if (sv == left || sv == right) @@ -1265,7 +1308,7 @@ Perl_do_kv(pTHX) 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) { @@ -1277,7 +1320,7 @@ Perl_do_kv(pTHX) RETURN; } - if (! SvTIED_mg((SV*)keys, 'P')) + if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) i = HvKEYS(keys); else { i = 0; @@ -1293,8 +1336,19 @@ Perl_do_kv(pTHX) PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ while ((entry = hv_iternext(keys))) { SPAGAIN; - if (dokeys) - XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ + if (dokeys) { + SV* sv = hv_iterkeysv(entry); + if (HvUTF8KEYS((SV*)hv) && !DO_UTF8(sv)) { + STRLEN len, i; + char* s = SvPV(sv, len); + for (i = 0; i < len && NATIVE_IS_INVARIANT(s[i]); i++); + if (i < len) { + sv = newSVsv(sv); + sv_utf8_upgrade(sv); + } + } + XPUSHs(sv); /* won't clobber stack_sp */ + } if (dovalues) { PUTBACK; tmpstr = realhv ?