X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/0453d815b8a74697ff1e5451c27aba2fe537b8e0..bc769488d48dfc3f9427d7cf0b8a8f5098cc9d84:/doop.c diff --git a/doop.c b/doop.c index e31af50..560dbe2 100644 --- a/doop.c +++ b/doop.c @@ -1,6 +1,6 @@ /* doop.c * - * Copyright (c) 1991-1999, Larry Wall + * Copyright (c) 1991-2001, 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. @@ -15,210 +15,292 @@ #define PERL_IN_DOOP_C #include "perl.h" +#ifndef PERL_MICRO #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) #include #endif +#endif STATIC I32 -S_do_trans_CC_simple(pTHX_ SV *sv) +S_do_trans_simple(pTHX_ SV *sv) { - dTHR; U8 *s; + U8 *d; U8 *send; + U8 *dstart; I32 matches = 0; + I32 grows = PL_op->op_private & OPpTRANS_GROWS; STRLEN len; short *tbl; I32 ch; tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "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-UTF8 input strings, because they're easy */ + if (!SvUTF8(sv)) { + while (s < send) { + if ((ch = tbl[*s]) >= 0) { + matches++; + *s++ = ch; + } + else + s++; } - s++; + SvSETMAGIC(sv); + return matches; } - SvSETMAGIC(sv); + /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ + if (grows) + New(0, d, len*2+1, U8); + else + d = s; + dstart = d; + while (s < send) { + STRLEN ulen; + UV c; + + /* Need to check this, otherwise 128..255 won't match */ + 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 */ + Copy(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 -S_do_trans_CC_count(pTHX_ SV *sv) +S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ { - dTHR; U8 *s; U8 *send; I32 matches = 0; STRLEN len; short *tbl; + I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "panic: do_trans"); + Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); s = (U8*)SvPV(sv, len); send = s + len; - while (s < send) { - if (tbl[*s] >= 0) - matches++; - s++; - } + if (!SvUTF8(sv)) + while (s < send) { + if (tbl[*s++] >= 0) + matches++; + } + else + while (s < send) { + UV c; + STRLEN ulen; + c = utf8n_to_uvchr(s, send - s, &ulen, 0); + if (c < 0x100) { + if (tbl[c] >= 0) + matches++; + } else if (complement) + matches++; + s += ulen; + } return matches; } STATIC I32 -S_do_trans_CC_complex(pTHX_ SV *sv) +S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ { - dTHR; U8 *s; U8 *send; U8 *d; + U8 *dstart; + I32 isutf8; I32 matches = 0; - STRLEN len; + 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 = 0; short *tbl; I32 ch; tbl = (short*)cPVOP->op_pv; if (!tbl) - Perl_croak(aTHX_ "panic: do_trans"); + Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); s = (U8*)SvPV(sv, len); + isutf8 = SvUTF8(sv); send = s + len; - d = s; - if (PL_op->op_private & OPpTRANS_SQUASH) { - U8* p = send; - - while (s < send) { - if ((ch = tbl[*s]) >= 0) { - *d = ch; - matches++; - if (p == d - 1 && *p == *d) - matches--; - else - p = d++; + if (!isutf8) { + dstart = d = s; + if (PL_op->op_private & OPpTRANS_SQUASH) { + U8* p = send; + while (s < send) { + if ((ch = tbl[*s]) >= 0) { + *d = 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 (ch == -1) /* -1 is unmapped character */ - *d++ = *s; /* -2 is delete character */ - s++; } - } - else { - while (s < send) { - if ((ch = tbl[*s]) >= 0) { - *d = ch; - matches++; - d++; + else { + while (s < send) { + if ((ch = tbl[*s]) >= 0) { + matches++; + *d++ = ch; + } + else if (ch == -1) /* -1 is unmapped character */ + *d++ = *s; + else if (ch == -2) /* -2 is delete character */ + matches++; + s++; } - else if (ch == -1) /* -1 is unmapped character */ - *d++ = *s; /* -2 is delete character */ - s++; } + *d = '\0'; + SvCUR_set(sv, d - dstart); } - 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_UU_simple(pTHX_ SV *sv) -{ - dTHR; - U8 *s; - U8 *send; - U8 *d; - 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; + else { /* isutf8 */ + if (grows) + New(0, 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 - svp = hv_fetch(hv, "FINAL", 5, FALSE); - if (svp) - final = SvUV(*svp); + if (PL_op->op_private & OPpTRANS_SQUASH) { + UV pch = 0xfeedface; + while (s < send) { + STRLEN len; + UV comp = utf8_to_uvchr(s, &len); - d = s; - while (s < send) { - if ((uv = swash_fetch(rv, s)) < none) { - s += UTF8SKIP(s); - matches++; - d = uv_to_utf8(d, uv); + 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 (ch != pch) { + d = uvchr_to_utf8(d, ch); + pch = ch; + } + s += len; + continue; + } + } + } + else if ((ch = tbl[comp]) >= 0) { + matches++; + if (ch != pch) { + d = uvchr_to_utf8(d, ch); + pch = 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 if (uv == none) { - int i; - for (i = UTF8SKIP(s); i; i--) - *d++ = *s++; + else { + while (s < send) { + STRLEN len; + UV comp = utf8_to_uvchr(s, &len); + if (comp > 0xff) { + if (!complement) { + Copy(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; + } } - else if (uv == extra) { - s += UTF8SKIP(s); - matches++; - d = uv_to_utf8(d, final); + if (grows) { + sv_setpvn(sv, (char*)dstart, d - dstart); + Safefree(dstart); } - else - s += UTF8SKIP(s); + else { + *d = '\0'; + SvCUR_set(sv, d - dstart); + } + SvUTF8_on(sv); } - *d = '\0'; - SvCUR_set(sv, d - (U8*)SvPVX(sv)); SvSETMAGIC(sv); - - return matches; -} - -STATIC I32 -S_do_trans_UU_count(pTHX_ 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; - - s = (U8*)SvPV(sv, 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_UC_simple(pTHX_ SV *sv) +S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ { - dTHR; U8 *s; U8 *send; U8 *d; + U8 *start; + U8 *dstart, *dend; I32 matches = 0; + I32 grows = PL_op->op_private & OPpTRANS_GROWS; STRLEN len; SV* rv = (SV*)cSVOP->op_sv; @@ -226,52 +308,93 @@ S_do_trans_UC_simple(pTHX_ SV *sv) 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; s = (U8*)SvPV(sv, len); + isutf8 = SvUTF8(sv); + if (!isutf8) { + U8 *t = s, *e = s + len; + while (t < e) { + 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 */ + New(0, d, len*3+UTF8_MAXLEN, U8); + dend = d + len * 3; + dstart = d; + } + else { + dstart = d = s; + dend = d + len; + } + while (s < send) { - if ((uv = swash_fetch(rv, s)) < none) { + if ((uv = swash_fetch(rv, s, TRUE)) < 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; + int i = UTF8SKIP(s); + Copy(s, d, i, U8); + d += i; + s += i; } else if (uv == extra) { - s += UTF8SKIP(s); + int i = UTF8SKIP(s); + s += i; matches++; - *d++ = (U8)final; + d = uvuni_to_utf8(d, final); } else s += UTF8SKIP(s); + + if (d > dend) { + STRLEN clen = d - dstart; + STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; + if (!grows) + Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); + Renew(dstart, nlen+UTF8_MAXLEN, U8); + d = dstart + clen; + dend = dstart + nlen; + } + } + 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); } - *d = '\0'; - SvCUR_set(sv, d - (U8*)SvPVX(sv)); SvSETMAGIC(sv); + SvUTF8_on(sv); return matches; } STATIC I32 -S_do_trans_CU_simple(pTHX_ SV *sv) +S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ { - dTHR; U8 *s; - U8 *send; - U8 *d; - U8 *dst; + U8 *start = 0, *send; I32 matches = 0; STRLEN len; @@ -280,299 +403,190 @@ S_do_trans_CU_simple(pTHX_ SV *sv) 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); - } - } - *d = '\0'; - sv_usepvn_mg(sv, (char*)dst, d - dst); - - return matches; -} - -/* utf-8 to latin-1 */ - -STATIC I32 -S_do_trans_UC_trivial(pTHX_ SV *sv) -{ - dTHR; - U8 *s; - U8 *send; - U8 *d; - STRLEN len; + U8 hibit = 0; 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; + if (!SvUTF8(sv)) { + U8 *t = s, *e = s + len; + while (t < e) { + 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 -S_do_trans_CU_trivial(pTHX_ 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); - } + if ((uv = swash_fetch(rv, s, TRUE)) < 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 -S_do_trans_UU_complex(pTHX_ SV *sv) +S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ { - 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; + I32 grows = PL_op->op_private & OPpTRANS_GROWS; 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 final = 0; + bool havefinal = FALSE; UV uv; STRLEN len; - U8 *dst; + U8 *dstart, *dend; + I32 isutf8; + U8 hibit = 0; s = (U8*)SvPV(sv, len); + isutf8 = SvUTF8(sv); + if (!isutf8) { + U8 *t = s, *e = s + len; + while (t < e) { + 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 */ + New(0, d, len*3+UTF8_MAXLEN, 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 = 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 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) { - 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]); - } + int i = UTF8SKIP(s); + Copy(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) { + Copy(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); + 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 line %d",__LINE__); + Renew(dstart, nlen+UTF8_MAXLEN, 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]); - } + int i = UTF8SKIP(s); + Copy(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; @@ -581,8 +595,9 @@ S_do_trans_UU_complex(pTHX_ SV *sv) 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); @@ -592,40 +607,30 @@ Perl_do_trans(pTHX_ SV *sv) return 0; if (!SvPOKp(sv)) (void)SvPV_force(sv, len); - (void)SvPOK_only(sv); + 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 & 63) { + switch (PL_op->op_private & ~hasutf & 63) { 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); } } @@ -636,15 +641,17 @@ 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); if (SvLEN(sv) < len + items) { /* current length is way too short */ while (items-- > 0) { - if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) { + if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { SvPV(*mark, tmplen); len += tmplen; } @@ -657,23 +664,19 @@ Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **s ++mark; } - if (items-- > 0) { - char *s; + sv_setpv(sv, ""); + if (PL_tainting && SvMAGICAL(sv)) + SvTAINTED_off(sv); - if (*mark) { - s = SvPV(*mark, tmplen); - sv_setpvn(sv, s, tmplen); - } - else - sv_setpv(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); } } @@ -697,6 +700,145 @@ Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) 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; + 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"); + + 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 (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) { + 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) { + 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; +} + +/* 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 Perl_do_vecset(pTHX_ SV *sv) { @@ -704,7 +846,7 @@ Perl_do_vecset(pTHX_ SV *sv) register I32 offset; register I32 size; register unsigned char *s; - register unsigned long lval; + register UV lval; I32 mask; STRLEN targlen; STRLEN len; @@ -712,40 +854,70 @@ Perl_do_vecset(pTHX_ SV *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_ "Assigning to negative offset in vec"); 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 ] = 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) { + 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 @@ -753,8 +925,7 @@ 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; @@ -772,7 +943,7 @@ Perl_do_chop(pTHX_ register SV *astr, register SV *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; } @@ -781,42 +952,43 @@ Perl_do_chop(pTHX_ register SV *astr, register SV *sv) 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; 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); + while (s > start && UTF8_IS_CONTINUATION(*s)) + s--; + if (utf8_to_uvchr((U8*)s, 0)) { + 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 Perl_do_chomp(pTHX_ register SV *sv) { - dTHR; register I32 count; STRLEN len; + STRLEN n_a; char *s; if (RsSNARF(PL_rs)) @@ -841,15 +1013,13 @@ Perl_do_chomp(pTHX_ register SV *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(PL_rs)) { @@ -880,19 +1050,19 @@ 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; -} +} void Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) { - dTHR; /* just for taint */ #ifdef LIBERAL register long *dl; register long *ll; @@ -907,6 +1077,14 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) I32 lensave; char *lsave; char *rsave; + bool left_utf = DO_UTF8(left); + 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 ^= */ @@ -914,23 +1092,91 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) rsave = rc = SvPV(right, rightlen); len = leftlen < rightlen ? leftlen : rightlen; lensave = len; - if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { + 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; + 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)) && @@ -1001,13 +1247,14 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) break; } } +finish: SvTAINT(sv); } OP * Perl_do_kv(pTHX) { - djSP; + dSP; HV *hv = (HV*)POPs; HV *keys; register HE *entry; @@ -1016,12 +1263,12 @@ Perl_do_kv(pTHX) 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) + + if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) dokeys = dovalues = TRUE; 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; @@ -1040,10 +1287,10 @@ Perl_do_kv(pTHX) 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) { @@ -1055,7 +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; @@ -1069,7 +1316,7 @@ Perl_do_kv(pTHX) 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 */