3 * Copyright (c) 1991-2000, 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
19 #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
25 S_do_trans_simple(pTHX_ SV *sv)
36 tbl = (short*)cPVOP->op_pv;
38 Perl_croak(aTHX_ "panic: do_trans_simple");
40 s = (U8*)SvPV(sv, len);
43 /* First, take care of non-UTF8 input strings, because they're easy */
46 if ((ch = tbl[*s]) >= 0) {
57 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
58 Newz(0, d, len*2+1, U8);
65 /* Need to check this, otherwise 128..255 won't match */
66 c = utf8_to_uv(s, send - s, &ulen, 0);
67 if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
75 else { /* No match -> copy */
81 sv_setpvn(sv, (char*)dstart, d - dstart);
88 S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
96 tbl = (short*)cPVOP->op_pv;
98 Perl_croak(aTHX_ "panic: do_trans_count");
100 s = (U8*)SvPV(sv, len);
112 c = utf8_to_uv(s, send - s, &ulen, 0);
113 if (c < 0x100 && tbl[c] >= 0)
122 S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
134 tbl = (short*)cPVOP->op_pv;
136 Perl_croak(aTHX_ "panic: do_trans_complex");
138 s = (U8*)SvPV(sv, len);
144 if (PL_op->op_private & OPpTRANS_SQUASH) {
147 if ((ch = tbl[*s]) >= 0) {
150 if (p != d - 1 || *p != *d)
153 else if (ch == -1) /* -1 is unmapped character */
155 else if (ch == -2) /* -2 is delete character */
162 if ((ch = tbl[*s]) >= 0) {
166 else if (ch == -1) /* -1 is unmapped character */
168 else if (ch == -2) /* -2 is delete character */
173 SvCUR_set(sv, d - dstart);
176 Newz(0, d, len*2+1, U8);
179 if (PL_op->op_private & OPpTRANS_SQUASH) {
184 UV comp = utf8_to_uv_simple(s, &len);
187 d = uv_to_utf8(d, comp); /* always unmapped */
188 else if ((ch = tbl[comp]) >= 0) {
191 d = uv_to_utf8(d, ch);
197 else if (ch == -1) /* -1 is unmapped character */
198 d = uv_to_utf8(d, comp);
199 else if (ch == -2) /* -2 is delete character */
208 UV comp = utf8_to_uv_simple(s, &len);
210 d = uv_to_utf8(d, comp); /* always unmapped */
211 else if ((ch = tbl[comp]) >= 0) {
212 d = uv_to_utf8(d, ch);
215 else if (ch == -1) { /* -1 is unmapped character */
216 d = uv_to_utf8(d, comp);
218 else if (ch == -2) /* -2 is delete character */
224 sv_setpvn(sv, (char*)dstart, d - dstart);
232 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
242 SV* rv = (SV*)cSVOP->op_sv;
243 HV* hv = (HV*)SvRV(rv);
244 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
245 UV none = svp ? SvUV(*svp) : 0x7fffffff;
252 s = (U8*)SvPV(sv, len);
255 U8 *t = s, *e = s + len;
257 if ((hibit = *t++ & 0x80))
260 s = bytes_to_utf8(s, &len);
265 svp = hv_fetch(hv, "FINAL", 5, FALSE);
269 /* d needs to be bigger than s, in case e.g. upgrading is required */
270 New(0, d, len*3+UTF8_MAXLEN, U8);
275 if ((uv = swash_fetch(rv, s)) < none) {
278 d = uv_to_utf8(d, uv);
280 else if (uv == none) {
285 else if (uv == extra) {
289 d = uv_to_utf8(d, final);
295 STRLEN clen = d - dstart;
296 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
297 Renew(dstart, nlen+UTF8_MAXLEN, U8);
299 dend = dstart + nlen;
303 sv_setpvn(sv, (char*)dstart, d - dstart);
308 if (!isutf8 && !(PL_hints & HINT_UTF8))
309 sv_utf8_downgrade(sv, TRUE);
315 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
322 SV* rv = (SV*)cSVOP->op_sv;
323 HV* hv = (HV*)SvRV(rv);
324 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
325 UV none = svp ? SvUV(*svp) : 0x7fffffff;
329 s = (U8*)SvPV(sv, len);
331 U8 *t = s, *e = s + len;
333 if ((hibit = *t++ & 0x80))
336 start = s = bytes_to_utf8(s, &len);
341 if ((uv = swash_fetch(rv, s)) < none)
352 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
358 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
359 I32 del = PL_op->op_private & OPpTRANS_DELETE;
360 SV* rv = (SV*)cSVOP->op_sv;
361 HV* hv = (HV*)SvRV(rv);
362 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
363 UV none = svp ? SvUV(*svp) : 0x7fffffff;
372 s = (U8*)SvPV(sv, len);
375 U8 *t = s, *e = s + len;
377 if ((hibit = *t++ & 0x80))
380 s = bytes_to_utf8(s, &len);
385 svp = hv_fetch(hv, "FINAL", 5, FALSE);
389 New(0, d, len*3+UTF8_MAXLEN, U8);
396 uv = swash_fetch(rv, s);
399 STRLEN clen = d - dstart, nlen = dend - dstart + len;
400 Renew(dstart, nlen+UTF8_MAXLEN, U8);
402 dend = dstart + nlen;
407 d = uv_to_utf8(d, uv);
413 else if (uv == none) { /* "none" is unmapped character */
420 else if (uv == extra && !del) {
423 d = uv_to_utf8(d, final);
429 matches++; /* "none+1" is delete character */
435 uv = swash_fetch(rv, s);
437 STRLEN clen = d - dstart, nlen = dend - dstart + len;
438 Renew(dstart, nlen+UTF8_MAXLEN, U8);
440 dend = dstart + nlen;
444 d = uv_to_utf8(d, uv);
448 else if (uv == none) { /* "none" is unmapped character */
454 else if (uv == extra && !del) {
456 d = uv_to_utf8(d, final);
460 matches++; /* "none+1" is delete character */
465 sv_setpvn(sv, (char*)dstart, d - dstart);
469 if (!isutf8 && !(PL_hints & HINT_UTF8))
470 sv_utf8_downgrade(sv, TRUE);
477 Perl_do_trans(pTHX_ SV *sv)
480 I32 hasutf = (PL_op->op_private &
481 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
483 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
484 Perl_croak(aTHX_ PL_no_modify);
490 (void)SvPV_force(sv, len);
491 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
492 (void)SvPOK_only_UTF8(sv);
494 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
496 switch (PL_op->op_private & ~hasutf & 63) {
499 return do_trans_simple_utf8(sv);
501 return do_trans_simple(sv);
503 case OPpTRANS_IDENTICAL:
505 return do_trans_count_utf8(sv);
507 return do_trans_count(sv);
511 return do_trans_complex_utf8(sv);
513 return do_trans_complex(sv);
518 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
521 register I32 items = sp - mark;
524 register char *delim = SvPV(del, delimlen);
528 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
529 (void)SvUPGRADE(sv, SVt_PV);
530 if (SvLEN(sv) < len + items) { /* current length is way too short */
531 while (items-- > 0) {
532 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
538 SvGROW(sv, len + 1); /* so try to pre-extend */
554 for (; items > 0; items--,mark++) {
560 for (; items > 0; items--,mark++)
567 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
570 char *pat = SvPV(*sarg, patlen);
571 bool do_taint = FALSE;
573 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
579 /* currently converts input to bytes if possible, but doesn't sweat failure */
581 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
584 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
589 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
590 Perl_croak(aTHX_ "Illegal number of bits in vec");
593 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
595 offset *= size; /* turn into bit offset */
596 len = (offset + size + 7) / 8; /* required number of bytes */
601 offset >>= 3; /* turn into byte offset */
603 if (offset >= srclen)
606 retnum = (UV) s[offset] << 8;
608 else if (size == 32) {
609 if (offset >= srclen)
611 else if (offset + 1 >= srclen)
613 ((UV) s[offset ] << 24);
614 else if (offset + 2 >= srclen)
616 ((UV) s[offset ] << 24) +
617 ((UV) s[offset + 1] << 16);
620 ((UV) s[offset ] << 24) +
621 ((UV) s[offset + 1] << 16) +
622 ( s[offset + 2] << 8);
625 else if (size == 64) {
626 if (ckWARN(WARN_PORTABLE))
627 Perl_warner(aTHX_ WARN_PORTABLE,
628 "Bit vector size > 32 non-portable");
629 if (offset >= srclen)
631 else if (offset + 1 >= srclen)
633 (UV) s[offset ] << 56;
634 else if (offset + 2 >= srclen)
636 ((UV) s[offset ] << 56) +
637 ((UV) s[offset + 1] << 48);
638 else if (offset + 3 >= srclen)
640 ((UV) s[offset ] << 56) +
641 ((UV) s[offset + 1] << 48) +
642 ((UV) s[offset + 2] << 40);
643 else if (offset + 4 >= srclen)
645 ((UV) s[offset ] << 56) +
646 ((UV) s[offset + 1] << 48) +
647 ((UV) s[offset + 2] << 40) +
648 ((UV) s[offset + 3] << 32);
649 else if (offset + 5 >= srclen)
651 ((UV) s[offset ] << 56) +
652 ((UV) s[offset + 1] << 48) +
653 ((UV) s[offset + 2] << 40) +
654 ((UV) s[offset + 3] << 32) +
655 ( s[offset + 4] << 24);
656 else if (offset + 6 >= srclen)
658 ((UV) s[offset ] << 56) +
659 ((UV) s[offset + 1] << 48) +
660 ((UV) s[offset + 2] << 40) +
661 ((UV) s[offset + 3] << 32) +
662 ((UV) s[offset + 4] << 24) +
663 ((UV) s[offset + 5] << 16);
666 ((UV) s[offset ] << 56) +
667 ((UV) s[offset + 1] << 48) +
668 ((UV) s[offset + 2] << 40) +
669 ((UV) s[offset + 3] << 32) +
670 ((UV) s[offset + 4] << 24) +
671 ((UV) s[offset + 5] << 16) +
672 ( s[offset + 6] << 8);
678 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
680 offset >>= 3; /* turn into byte offset */
685 ((UV) s[offset] << 8) +
689 ((UV) s[offset ] << 24) +
690 ((UV) s[offset + 1] << 16) +
691 ( s[offset + 2] << 8) +
694 else if (size == 64) {
695 if (ckWARN(WARN_PORTABLE))
696 Perl_warner(aTHX_ WARN_PORTABLE,
697 "Bit vector size > 32 non-portable");
699 ((UV) s[offset ] << 56) +
700 ((UV) s[offset + 1] << 48) +
701 ((UV) s[offset + 2] << 40) +
702 ((UV) s[offset + 3] << 32) +
703 ((UV) s[offset + 4] << 24) +
704 ((UV) s[offset + 5] << 16) +
705 ( s[offset + 6] << 8) +
714 /* currently converts input to bytes if possible but doesn't sweat failures,
715 * although it does ensure that the string it clobbers is not marked as
716 * utf8-valid any more
719 Perl_do_vecset(pTHX_ SV *sv)
721 SV *targ = LvTARG(sv);
724 register unsigned char *s;
732 s = (unsigned char*)SvPV_force(targ, targlen);
734 /* This is handled by the SvPOK_only below...
735 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
738 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
741 (void)SvPOK_only(targ);
743 offset = LvTARGOFF(sv);
745 Perl_croak(aTHX_ "Assigning to negative offset in vec");
746 size = LvTARGLEN(sv);
747 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
748 Perl_croak(aTHX_ "Illegal number of bits in vec");
750 offset *= size; /* turn into bit offset */
751 len = (offset + size + 7) / 8; /* required number of bytes */
753 s = (unsigned char*)SvGROW(targ, len + 1);
754 (void)memzero((char *)(s + targlen), len - targlen + 1);
755 SvCUR_set(targ, len);
759 mask = (1 << size) - 1;
762 offset >>= 3; /* turn into byte offset */
763 s[offset] &= ~(mask << size);
764 s[offset] |= lval << size;
767 offset >>= 3; /* turn into byte offset */
769 s[offset ] = lval & 0xff;
770 else if (size == 16) {
771 s[offset ] = (lval >> 8) & 0xff;
772 s[offset+1] = lval & 0xff;
774 else if (size == 32) {
775 s[offset ] = (lval >> 24) & 0xff;
776 s[offset+1] = (lval >> 16) & 0xff;
777 s[offset+2] = (lval >> 8) & 0xff;
778 s[offset+3] = lval & 0xff;
781 else if (size == 64) {
782 if (ckWARN(WARN_PORTABLE))
783 Perl_warner(aTHX_ WARN_PORTABLE,
784 "Bit vector size > 32 non-portable");
785 s[offset ] = (lval >> 56) & 0xff;
786 s[offset+1] = (lval >> 48) & 0xff;
787 s[offset+2] = (lval >> 40) & 0xff;
788 s[offset+3] = (lval >> 32) & 0xff;
789 s[offset+4] = (lval >> 24) & 0xff;
790 s[offset+5] = (lval >> 16) & 0xff;
791 s[offset+6] = (lval >> 8) & 0xff;
792 s[offset+7] = lval & 0xff;
800 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
805 if (SvTYPE(sv) == SVt_PVAV) {
810 for (i = 0; i <= max; i++) {
811 sv = (SV*)av_fetch(av, i, FALSE);
812 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
817 else if (SvTYPE(sv) == SVt_PVHV) {
820 (void)hv_iterinit(hv);
822 while ((entry = hv_iternext(hv)))
823 do_chop(astr,hv_iterval(hv,entry));
826 else if (SvREADONLY(sv))
827 Perl_croak(aTHX_ PL_no_modify);
829 if (len && !SvPOK(sv))
830 s = SvPV_force(sv, len);
833 char *send = s + len;
836 while ((*s & 0xc0) == 0x80)
838 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
839 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
840 sv_setpvn(astr, s, send - s);
842 SvCUR_set(sv, s - start);
847 sv_setpvn(astr, "", 0);
851 sv_setpvn(astr, s, 1);
858 sv_setpvn(astr, "", 0);
863 Perl_do_chomp(pTHX_ register SV *sv)
874 if (SvTYPE(sv) == SVt_PVAV) {
879 for (i = 0; i <= max; i++) {
880 sv = (SV*)av_fetch(av, i, FALSE);
881 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
882 count += do_chomp(sv);
886 else if (SvTYPE(sv) == SVt_PVHV) {
889 (void)hv_iterinit(hv);
891 while ((entry = hv_iternext(hv)))
892 count += do_chomp(hv_iterval(hv,entry));
895 else if (SvREADONLY(sv))
896 Perl_croak(aTHX_ PL_no_modify);
898 if (len && !SvPOKp(sv))
899 s = SvPV_force(sv, len);
906 while (len && s[-1] == '\n') {
914 char *rsptr = SvPV(PL_rs, rslen);
925 if (memNE(s, rsptr, rslen))
940 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
956 bool left_utf = DO_UTF8(left);
957 bool right_utf = DO_UTF8(right);
960 if (left_utf && !right_utf)
961 sv_utf8_upgrade(right);
962 else if (!left_utf && right_utf)
963 sv_utf8_upgrade(left);
965 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
966 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
967 lsave = lc = SvPV(left, leftlen);
968 rsave = rc = SvPV(right, rightlen);
969 len = leftlen < rightlen ? leftlen : rightlen;
971 if ((left_utf || right_utf) && (sv == left || sv == right)) {
972 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
973 Newz(801, dc, needlen + 1, char);
975 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
977 dc = SvPV_force(sv, n_a);
978 if (SvCUR(sv) < len) {
979 dc = SvGROW(sv, len + 1);
980 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
982 if (optype != OP_BIT_AND && (left_utf || right_utf))
983 dc = SvGROW(sv, leftlen + rightlen + 1);
986 needlen = ((optype == OP_BIT_AND)
987 ? len : (leftlen > rightlen ? leftlen : rightlen));
988 Newz(801, dc, needlen + 1, char);
989 (void)sv_usepvn(sv, dc, needlen);
990 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
993 (void)SvPOK_only(sv);
994 if (left_utf || right_utf) {
997 STRLEN lulen = leftlen;
998 STRLEN rulen = rightlen;
1003 while (lulen && rulen) {
1004 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1007 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1011 dc = (char*)uv_to_utf8((U8*)dc, duc);
1013 if (sv == left || sv == right)
1014 (void)sv_usepvn(sv, dcsave, needlen);
1015 SvCUR_set(sv, dc - dcsave);
1018 while (lulen && rulen) {
1019 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1022 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1026 dc = (char*)uv_to_utf8((U8*)dc, duc);
1030 while (lulen && rulen) {
1031 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1034 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1038 dc = (char*)uv_to_utf8((U8*)dc, duc);
1041 if (sv == left || sv == right)
1042 (void)sv_usepvn(sv, dcsave, needlen);
1043 SvCUR_set(sv, dc - dcsave);
1045 sv_catpvn(sv, rc, rulen);
1047 sv_catpvn(sv, lc, lulen);
1057 if (len >= sizeof(long)*4 &&
1058 !((long)dc % sizeof(long)) &&
1059 !((long)lc % sizeof(long)) &&
1060 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1062 I32 remainder = len % (sizeof(long)*4);
1063 len /= (sizeof(long)*4);
1072 *dl++ = *ll++ & *rl++;
1073 *dl++ = *ll++ & *rl++;
1074 *dl++ = *ll++ & *rl++;
1075 *dl++ = *ll++ & *rl++;
1080 *dl++ = *ll++ ^ *rl++;
1081 *dl++ = *ll++ ^ *rl++;
1082 *dl++ = *ll++ ^ *rl++;
1083 *dl++ = *ll++ ^ *rl++;
1088 *dl++ = *ll++ | *rl++;
1089 *dl++ = *ll++ | *rl++;
1090 *dl++ = *ll++ | *rl++;
1091 *dl++ = *ll++ | *rl++;
1106 *dc++ = *lc++ & *rc++;
1110 *dc++ = *lc++ ^ *rc++;
1114 *dc++ = *lc++ | *rc++;
1118 sv_catpvn(sv, rsave + len, rightlen - len);
1119 else if (leftlen > len)
1120 sv_catpvn(sv, lsave + len, leftlen - len);
1138 I32 gimme = GIMME_V;
1139 I32 dokeys = (PL_op->op_type == OP_KEYS);
1140 I32 dovalues = (PL_op->op_type == OP_VALUES);
1141 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1143 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1144 dokeys = dovalues = TRUE;
1147 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1148 dTARGET; /* make sure to clear its target here */
1149 if (SvTYPE(TARG) == SVt_PVLV)
1150 LvTARG(TARG) = Nullsv;
1156 keys = realhv ? hv : avhv_keys((AV*)hv);
1157 (void)hv_iterinit(keys); /* always reset iterator regardless */
1159 if (gimme == G_VOID)
1162 if (gimme == G_SCALAR) {
1166 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1167 if (SvTYPE(TARG) < SVt_PVLV) {
1168 sv_upgrade(TARG, SVt_PVLV);
1169 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1172 if (LvTARG(TARG) != (SV*)keys) {
1174 SvREFCNT_dec(LvTARG(TARG));
1175 LvTARG(TARG) = SvREFCNT_inc(keys);
1181 if (! SvTIED_mg((SV*)keys, 'P'))
1186 while (hv_iternext(keys)) i++;
1192 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1194 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1195 while ((entry = hv_iternext(keys))) {
1198 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1202 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1203 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1204 (unsigned long)HeHASH(entry),
1206 (unsigned long)(HeHASH(entry) & HvMAX(keys))));