X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=doop.c;h=8600b7c93e5334827fb717e7496411aa228aec5a;hb=bbd5c0f5ad81733b079008f34cd05cd9aef7d917;hp=e4a516a1a353d4f52f06f1141685cbd0bc3b5ebc;hpb=63cd067459124b5dc1d2ae98453df4ffdff11607;p=p5sagit%2Fp5-mst-13.2.git diff --git a/doop.c b/doop.c index e4a516a..8600b7c 100644 --- a/doop.c +++ b/doop.c @@ -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; @@ -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; @@ -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 */ @@ -291,7 +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; @@ -307,7 +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; @@ -316,9 +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_INVARIANT(*t++))) + while (t < e) { + U8 ch = *t++; + if ((hibit = !NATIVE_IS_INVARIANT(ch))) break; + } if (hibit) s = bytes_to_utf8(s, &len); } @@ -341,10 +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 = uvchr_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); } else if (uv == none) { int i = UTF8SKIP(s); @@ -356,7 +359,7 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ int i = UTF8SKIP(s); s += i; matches++; - d = uvchr_to_utf8(d, final); + d = uvuni_to_utf8(d, final); } else s += UTF8SKIP(s); @@ -365,7 +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; @@ -383,18 +386,15 @@ S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ } SvSETMAGIC(sv); SvUTF8_on(sv); - /* Downgrading just 'cos it will is suspect - NI-S */ - 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; @@ -402,22 +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_INVARIANT(*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); } @@ -428,7 +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; @@ -442,7 +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; @@ -453,9 +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_INVARIANT(*t++))) + while (t < e) { + U8 ch = *t++; + if ((hibit = !NATIVE_IS_INVARIANT(ch))) break; + } if (hibit) s = bytes_to_utf8(s, &len); } @@ -463,8 +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 */ @@ -480,13 +488,13 @@ 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; @@ -495,7 +503,7 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ matches++; s += UTF8SKIP(s); if (uv != puv) { - d = uvchr_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); puv = uv; } continue; @@ -510,10 +518,22 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ } else if (uv == extra && !del) { matches++; - s += UTF8SKIP(s); - if (uv != puv) { - d = uvchr_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; } continue; } @@ -523,12 +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; @@ -536,7 +556,7 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ if (uv < none) { matches++; s += UTF8SKIP(s); - d = uvchr_to_utf8(d, uv); + d = uvuni_to_utf8(d, uv); continue; } else if (uv == none) { /* "none" is unmapped character */ @@ -549,7 +569,7 @@ S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ else if (uv == extra && !del) { matches++; s += UTF8SKIP(s); - d = uvchr_to_utf8(d, final); + d = uvuni_to_utf8(d, final); continue; } matches++; /* "none+1" is delete character */ @@ -567,8 +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; @@ -623,9 +641,11 @@ 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); @@ -644,14 +664,16 @@ 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); @@ -966,6 +988,7 @@ Perl_do_chomp(pTHX_ register SV *sv) { register I32 count; STRLEN len; + STRLEN n_a; char *s; if (RsSNARF(PL_rs)) @@ -997,8 +1020,6 @@ Perl_do_chomp(pTHX_ register SV *sv) 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(PL_rs)) { @@ -1029,12 +1050,13 @@ 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; } @@ -1057,7 +1079,7 @@ 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); @@ -1268,7 +1290,7 @@ 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) { @@ -1280,7 +1302,7 @@ 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;