3 * Copyright (c) 1991-1999, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "'So that was the job I felt I had to do when I started,' thought Sam."
15 #define PERL_IN_DOOP_C
18 #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
23 S_do_trans_CC_simple(pTHX_ SV *sv)
33 tbl = (short*)cPVOP->op_pv;
35 Perl_croak(aTHX_ "panic: do_trans");
37 s = (U8*)SvPV(sv, len);
41 if ((ch = tbl[*s]) >= 0) {
53 S_do_trans_CC_count(pTHX_ SV *sv)
62 tbl = (short*)cPVOP->op_pv;
64 Perl_croak(aTHX_ "panic: do_trans");
66 s = (U8*)SvPV(sv, len);
79 S_do_trans_CC_complex(pTHX_ SV *sv)
90 tbl = (short*)cPVOP->op_pv;
92 Perl_croak(aTHX_ "panic: do_trans");
94 s = (U8*)SvPV(sv, len);
98 if (PL_op->op_private & OPpTRANS_SQUASH) {
102 if ((ch = tbl[*s]) >= 0) {
105 if (p == d - 1 && *p == *d)
110 else if (ch == -1) /* -1 is unmapped character */
111 *d++ = *s; /* -2 is delete character */
117 if ((ch = tbl[*s]) >= 0) {
122 else if (ch == -1) /* -1 is unmapped character */
123 *d++ = *s; /* -2 is delete character */
127 matches += send - d; /* account for disappeared chars */
129 SvCUR_set(sv, d - (U8*)SvPVX(sv));
136 S_do_trans_UU_simple(pTHX_ SV *sv)
145 SV* rv = (SV*)cSVOP->op_sv;
146 HV* hv = (HV*)SvRV(rv);
147 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
148 UV none = svp ? SvUV(*svp) : 0x7fffffff;
153 s = (U8*)SvPV(sv, len);
156 svp = hv_fetch(hv, "FINAL", 5, FALSE);
162 if ((uv = swash_fetch(rv, s)) < none) {
165 d = uv_to_utf8(d, uv);
167 else if (uv == none) {
169 for (i = UTF8SKIP(s); i; i--)
172 else if (uv == extra) {
175 d = uv_to_utf8(d, final);
181 SvCUR_set(sv, d - (U8*)SvPVX(sv));
188 S_do_trans_UU_count(pTHX_ SV *sv)
196 SV* rv = (SV*)cSVOP->op_sv;
197 HV* hv = (HV*)SvRV(rv);
198 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
199 UV none = svp ? SvUV(*svp) : 0x7fffffff;
202 s = (U8*)SvPV(sv, len);
206 if ((uv = swash_fetch(rv, s)) < none)
215 S_do_trans_UC_simple(pTHX_ SV *sv)
224 SV* rv = (SV*)cSVOP->op_sv;
225 HV* hv = (HV*)SvRV(rv);
226 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
227 UV none = svp ? SvUV(*svp) : 0x7fffffff;
232 s = (U8*)SvPV(sv, len);
235 svp = hv_fetch(hv, "FINAL", 5, FALSE);
241 if ((uv = swash_fetch(rv, s)) < none) {
246 else if (uv == none) {
248 uv = utf8_to_uv(s, &ulen);
252 else if (uv == extra) {
261 SvCUR_set(sv, d - (U8*)SvPVX(sv));
268 S_do_trans_CU_simple(pTHX_ SV *sv)
278 SV* rv = (SV*)cSVOP->op_sv;
279 HV* hv = (HV*)SvRV(rv);
280 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
281 UV none = svp ? SvUV(*svp) : 0x7fffffff;
288 s = (U8*)SvPV(sv, len);
291 svp = hv_fetch(hv, "BITS", 4, FALSE);
293 bits = (I32)SvIV(*svp);
295 svp = hv_fetch(hv, "FINAL", 5, FALSE);
299 Newz(801, d, len * (bits >> 3) + 1, U8);
307 tmpbuf[0] = (( uv >> 6) | 0xc0);
308 tmpbuf[1] = (( uv & 0x3f) | 0x80);
311 if ((uv = swash_fetch(rv, tmpbuf)) < none) {
313 d = uv_to_utf8(d, uv);
316 d = uv_to_utf8(d, s[-1]);
317 else if (uv == extra) {
319 d = uv_to_utf8(d, final);
323 sv_usepvn_mg(sv, (char*)dst, d - dst);
328 /* utf-8 to latin-1 */
331 S_do_trans_UC_trivial(pTHX_ SV *sv)
339 s = (U8*)SvPV(sv, len);
348 UV uv = utf8_to_uv(s, &ulen);
354 SvCUR_set(sv, d - (U8*)SvPVX(sv));
360 /* latin-1 to utf-8 */
363 S_do_trans_CU_trivial(pTHX_ SV *sv)
373 s = (U8*)SvPV(sv, len);
376 Newz(801, d, len * 2 + 1, U8);
386 *d++ = (( uv >> 6) | 0xc0);
387 *d++ = (( uv & 0x3f) | 0x80);
391 sv_usepvn_mg(sv, (char*)dst, d - dst);
397 S_do_trans_UU_complex(pTHX_ SV *sv)
404 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
405 I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
406 I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
407 I32 del = PL_op->op_private & OPpTRANS_DELETE;
408 SV* rv = (SV*)cSVOP->op_sv;
409 HV* hv = (HV*)SvRV(rv);
410 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
411 UV none = svp ? SvUV(*svp) : 0x7fffffff;
418 s = (U8*)SvPV(sv, len);
421 svp = hv_fetch(hv, "FINAL", 5, FALSE);
425 if (PL_op->op_private & OPpTRANS_GROWS) {
428 svp = hv_fetch(hv, "BITS", 4, FALSE);
430 bits = (I32)SvIV(*svp);
432 Newz(801, d, len * (bits >> 3) + 1, U8);
444 uv = swash_fetch(rv, s);
452 tmpbuf[0] = (( uv >> 6) | 0xc0);
453 tmpbuf[1] = (( uv & 0x3f) | 0x80);
455 uv = swash_fetch(rv, tmpbuf);
460 if (uv >= 0x80 && to_utf)
461 d = uv_to_utf8(d, uv);
470 else if (uv == none) { /* "none" is unmapped character */
476 for (i = UTF8SKIP(s); i; --i)
481 *d++ = (U8)utf8_to_uv(s, &ulen);
485 else { /* must be to_utf only */
486 d = uv_to_utf8(d, s[-1]);
491 else if (uv == extra && !del) {
494 if (final >= 0x80 && to_utf)
495 d = uv_to_utf8(d, final);
504 matches++; /* "none+1" is delete character */
512 uv = swash_fetch(rv, s);
520 tmpbuf[0] = (( uv >> 6) | 0xc0);
521 tmpbuf[1] = (( uv & 0x3f) | 0x80);
523 uv = swash_fetch(rv, tmpbuf);
527 if (uv >= 0x80 && to_utf)
528 d = uv_to_utf8(d, uv);
535 else if (uv == none) { /* "none" is unmapped character */
541 for (i = UTF8SKIP(s); i; --i)
546 *d++ = (U8)utf8_to_uv(s, &ulen);
550 else { /* must be to_utf only */
551 d = uv_to_utf8(d, s[-1]);
555 else if (uv == extra && !del) {
557 if (final >= 0x80 && to_utf)
558 d = uv_to_utf8(d, final);
565 matches++; /* "none+1" is delete character */
571 sv_usepvn(sv, (char*)dst, d - dst);
574 SvCUR_set(sv, d - (U8*)SvPVX(sv));
582 Perl_do_trans(pTHX_ SV *sv)
587 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
588 Perl_croak(aTHX_ PL_no_modify);
594 (void)SvPV_force(sv, len);
595 (void)SvPOK_only(sv);
597 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
599 switch (PL_op->op_private & 63) {
601 return do_trans_CC_simple(sv);
603 case OPpTRANS_FROM_UTF:
604 return do_trans_UC_simple(sv);
606 case OPpTRANS_TO_UTF:
607 return do_trans_CU_simple(sv);
609 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF:
610 return do_trans_UU_simple(sv);
612 case OPpTRANS_IDENTICAL:
613 return do_trans_CC_count(sv);
615 case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL:
616 return do_trans_UC_trivial(sv);
618 case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
619 return do_trans_CU_trivial(sv);
621 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
622 return do_trans_UU_count(sv);
625 if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
626 return do_trans_UU_complex(sv); /* could be UC or CU too */
628 return do_trans_CC_complex(sv);
633 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
636 register I32 items = sp - mark;
639 register char *delim = SvPV(del, delimlen);
643 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
644 (void)SvUPGRADE(sv, SVt_PV);
645 if (SvLEN(sv) < len + items) { /* current length is way too short */
646 while (items-- > 0) {
647 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
653 SvGROW(sv, len + 1); /* so try to pre-extend */
664 s = SvPV(*mark, tmplen);
665 sv_setpvn(sv, s, tmplen);
675 for (; items > 0; items--,mark++) {
676 sv_catpvn(sv,delim,len);
681 for (; items > 0; items--,mark++)
688 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
691 char *pat = SvPV(*sarg, patlen);
692 bool do_taint = FALSE;
694 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
701 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
704 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
709 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
710 Perl_croak(aTHX_ "Illegal number of bits in vec");
711 offset *= size; /* turn into bit offset */
712 len = (offset + size + 7) / 8; /* required number of bytes */
717 offset >>= 3; /* turn into byte offset */
719 if (offset >= srclen)
722 retnum = (UV) s[offset] << 8;
724 else if (size == 32) {
725 if (offset >= srclen)
727 else if (offset + 1 >= srclen)
729 ((UV) s[offset ] << 24);
730 else if (offset + 2 >= srclen)
732 ((UV) s[offset ] << 24) +
733 ((UV) s[offset + 1] << 16);
736 ((UV) s[offset ] << 24) +
737 ((UV) s[offset + 1] << 16) +
738 ( s[offset + 2] << 8);
743 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
745 offset >>= 3; /* turn into byte offset */
750 ((UV) s[offset] << 8) +
754 ((UV) s[offset ] << 24) +
755 ((UV) s[offset + 1] << 16) +
756 ( s[offset + 2] << 8) +
764 Perl_do_vecset(pTHX_ SV *sv)
766 SV *targ = LvTARG(sv);
769 register unsigned char *s;
777 s = (unsigned char*)SvPV_force(targ, targlen);
779 offset = LvTARGOFF(sv);
780 size = LvTARGLEN(sv);
781 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
782 Perl_croak(aTHX_ "Illegal number of bits in vec");
784 offset *= size; /* turn into bit offset */
785 len = (offset + size + 7) / 8; /* required number of bytes */
787 s = (unsigned char*)SvGROW(targ, len + 1);
788 (void)memzero(s + targlen, len - targlen + 1);
789 SvCUR_set(targ, len);
793 mask = (1 << size) - 1;
796 offset >>= 3; /* turn into byte offset */
797 s[offset] &= ~(mask << size);
798 s[offset] |= lval << size;
801 offset >>= 3; /* turn into byte offset */
803 s[offset] = lval & 255;
804 else if (size == 16) {
805 s[offset] = (lval >> 8) & 255;
806 s[offset+1] = lval & 255;
808 else if (size == 32) {
809 s[offset] = (lval >> 24) & 255;
810 s[offset+1] = (lval >> 16) & 255;
811 s[offset+2] = (lval >> 8) & 255;
812 s[offset+3] = lval & 255;
819 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
825 if (SvTYPE(sv) == SVt_PVAV) {
830 for (i = 0; i <= max; i++) {
831 sv = (SV*)av_fetch(av, i, FALSE);
832 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
837 else if (SvTYPE(sv) == SVt_PVHV) {
840 (void)hv_iterinit(hv);
842 while (entry = hv_iternext(hv))
843 do_chop(astr,hv_iterval(hv,entry));
846 else if (SvREADONLY(sv))
847 Perl_croak(aTHX_ PL_no_modify);
849 if (len && !SvPOK(sv))
850 s = SvPV_force(sv, len);
853 char *send = s + len;
856 while ((*s & 0xc0) == 0x80)
858 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
859 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
860 sv_setpvn(astr, s, send - s);
862 SvCUR_set(sv, s - start);
866 sv_setpvn(astr, "", 0);
871 sv_setpvn(astr, s, 1);
877 sv_setpvn(astr, "", 0);
882 Perl_do_chomp(pTHX_ register SV *sv)
894 if (SvTYPE(sv) == SVt_PVAV) {
899 for (i = 0; i <= max; i++) {
900 sv = (SV*)av_fetch(av, i, FALSE);
901 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
902 count += do_chomp(sv);
906 else if (SvTYPE(sv) == SVt_PVHV) {
909 (void)hv_iterinit(hv);
911 while (entry = hv_iternext(hv))
912 count += do_chomp(hv_iterval(hv,entry));
915 else if (SvREADONLY(sv))
916 Perl_croak(aTHX_ PL_no_modify);
918 if (len && !SvPOKp(sv))
919 s = SvPV_force(sv, len);
926 while (len && s[-1] == '\n') {
934 char *rsptr = SvPV(PL_rs, rslen);
945 if (memNE(s, rsptr, rslen))
960 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
962 dTHR; /* just for taint */
978 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
979 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
980 lsave = lc = SvPV(left, leftlen);
981 rsave = rc = SvPV(right, rightlen);
982 len = leftlen < rightlen ? leftlen : rightlen;
984 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
986 dc = SvPV_force(sv, n_a);
987 if (SvCUR(sv) < len) {
988 dc = SvGROW(sv, len + 1);
989 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
993 I32 needlen = ((optype == OP_BIT_AND)
994 ? len : (leftlen > rightlen ? leftlen : rightlen));
995 Newz(801, dc, needlen + 1, char);
996 (void)sv_usepvn(sv, dc, needlen);
997 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
1000 (void)SvPOK_only(sv);
1002 if (len >= sizeof(long)*4 &&
1003 !((long)dc % sizeof(long)) &&
1004 !((long)lc % sizeof(long)) &&
1005 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1007 I32 remainder = len % (sizeof(long)*4);
1008 len /= (sizeof(long)*4);
1017 *dl++ = *ll++ & *rl++;
1018 *dl++ = *ll++ & *rl++;
1019 *dl++ = *ll++ & *rl++;
1020 *dl++ = *ll++ & *rl++;
1025 *dl++ = *ll++ ^ *rl++;
1026 *dl++ = *ll++ ^ *rl++;
1027 *dl++ = *ll++ ^ *rl++;
1028 *dl++ = *ll++ ^ *rl++;
1033 *dl++ = *ll++ | *rl++;
1034 *dl++ = *ll++ | *rl++;
1035 *dl++ = *ll++ | *rl++;
1036 *dl++ = *ll++ | *rl++;
1051 *dc++ = *lc++ & *rc++;
1055 *dc++ = *lc++ ^ *rc++;
1059 *dc++ = *lc++ | *rc++;
1063 sv_catpvn(sv, rsave + len, rightlen - len);
1064 else if (leftlen > len)
1065 sv_catpvn(sv, lsave + len, leftlen - len);
1082 I32 gimme = GIMME_V;
1083 I32 dokeys = (PL_op->op_type == OP_KEYS);
1084 I32 dovalues = (PL_op->op_type == OP_VALUES);
1085 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1087 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1088 dokeys = dovalues = TRUE;
1091 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1092 dTARGET; /* make sure to clear its target here */
1093 if (SvTYPE(TARG) == SVt_PVLV)
1094 LvTARG(TARG) = Nullsv;
1100 keys = realhv ? hv : avhv_keys((AV*)hv);
1101 (void)hv_iterinit(keys); /* always reset iterator regardless */
1103 if (gimme == G_VOID)
1106 if (gimme == G_SCALAR) {
1110 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1111 if (SvTYPE(TARG) < SVt_PVLV) {
1112 sv_upgrade(TARG, SVt_PVLV);
1113 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1116 if (LvTARG(TARG) != (SV*)keys) {
1118 SvREFCNT_dec(LvTARG(TARG));
1119 LvTARG(TARG) = SvREFCNT_inc(keys);
1125 if (! SvTIED_mg((SV*)keys, 'P'))
1130 while (hv_iternext(keys)) i++;
1136 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1138 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1139 while (entry = hv_iternext(keys)) {
1142 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1146 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1147 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1148 (unsigned long)HeHASH(entry),
1150 (unsigned long)(HeHASH(entry) & HvMAX(keys))));