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_CC_simple(pTHX_ SV *sv)
35 tbl = (short*)cPVOP->op_pv;
37 Perl_croak(aTHX_ "panic: do_trans");
39 s = (U8*)SvPV(sv, len);
43 if ((ch = tbl[*s]) >= 0) {
55 S_do_trans_CC_count(pTHX_ SV *sv)
64 tbl = (short*)cPVOP->op_pv;
66 Perl_croak(aTHX_ "panic: do_trans");
68 s = (U8*)SvPV(sv, len);
81 S_do_trans_CC_complex(pTHX_ SV *sv)
92 tbl = (short*)cPVOP->op_pv;
94 Perl_croak(aTHX_ "panic: do_trans");
96 s = (U8*)SvPV(sv, len);
100 if (PL_op->op_private & OPpTRANS_SQUASH) {
104 if ((ch = tbl[*s]) >= 0) {
107 if (p == d - 1 && *p == *d)
112 else if (ch == -1) /* -1 is unmapped character */
113 *d++ = *s; /* -2 is delete character */
119 if ((ch = tbl[*s]) >= 0) {
124 else if (ch == -1) /* -1 is unmapped character */
125 *d++ = *s; /* -2 is delete character */
129 matches += send - d; /* account for disappeared chars */
131 SvCUR_set(sv, d - (U8*)SvPVX(sv));
138 S_do_trans_UU_simple(pTHX_ SV *sv)
147 SV* rv = (SV*)cSVOP->op_sv;
148 HV* hv = (HV*)SvRV(rv);
149 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
150 UV none = svp ? SvUV(*svp) : 0x7fffffff;
155 s = (U8*)SvPV(sv, len);
158 svp = hv_fetch(hv, "FINAL", 5, FALSE);
164 if ((uv = swash_fetch(rv, s)) < none) {
167 d = uv_to_utf8(d, uv);
169 else if (uv == none) {
171 for (i = UTF8SKIP(s); i; i--)
174 else if (uv == extra) {
177 d = uv_to_utf8(d, final);
183 SvCUR_set(sv, d - (U8*)SvPVX(sv));
190 S_do_trans_UU_count(pTHX_ SV *sv)
198 SV* rv = (SV*)cSVOP->op_sv;
199 HV* hv = (HV*)SvRV(rv);
200 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
201 UV none = svp ? SvUV(*svp) : 0x7fffffff;
204 s = (U8*)SvPV(sv, len);
208 if ((uv = swash_fetch(rv, s)) < none)
217 S_do_trans_UU_complex(pTHX_ SV *sv)
224 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
225 I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
226 I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
227 I32 del = PL_op->op_private & OPpTRANS_DELETE;
228 SV* rv = (SV*)cSVOP->op_sv;
229 HV* hv = (HV*)SvRV(rv);
230 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
231 UV none = svp ? SvUV(*svp) : 0x7fffffff;
238 s = (U8*)SvPV(sv, len);
241 svp = hv_fetch(hv, "FINAL", 5, FALSE);
245 if (PL_op->op_private & OPpTRANS_GROWS) {
248 svp = hv_fetch(hv, "BITS", 4, FALSE);
250 bits = (I32)SvIV(*svp);
252 Newz(801, d, len * (bits >> 3) + 1, U8);
264 uv = swash_fetch(rv, s);
272 tmpbuf[0] = (( uv >> 6) | 0xc0);
273 tmpbuf[1] = (( uv & 0x3f) | 0x80);
275 uv = swash_fetch(rv, tmpbuf);
280 if (uv >= 0x80 && to_utf)
281 d = uv_to_utf8(d, uv);
290 else if (uv == none) { /* "none" is unmapped character */
296 for (i = UTF8SKIP(s); i; --i)
301 *d++ = (U8)utf8_to_uv(s, &ulen);
305 else { /* must be to_utf only */
306 d = uv_to_utf8(d, s[-1]);
311 else if (uv == extra && !del) {
314 if (final >= 0x80 && to_utf)
315 d = uv_to_utf8(d, final);
324 matches++; /* "none+1" is delete character */
332 uv = swash_fetch(rv, s);
340 tmpbuf[0] = (( uv >> 6) | 0xc0);
341 tmpbuf[1] = (( uv & 0x3f) | 0x80);
343 uv = swash_fetch(rv, tmpbuf);
347 if (uv >= 0x80 && to_utf)
348 d = uv_to_utf8(d, uv);
355 else if (uv == none) { /* "none" is unmapped character */
361 for (i = UTF8SKIP(s); i; --i)
366 *d++ = (U8)utf8_to_uv(s, &ulen);
370 else { /* must be to_utf only */
371 d = uv_to_utf8(d, s[-1]);
375 else if (uv == extra && !del) {
377 if (final >= 0x80 && to_utf)
378 d = uv_to_utf8(d, final);
385 matches++; /* "none+1" is delete character */
391 sv_usepvn(sv, (char*)dst, d - dst);
394 SvCUR_set(sv, d - (U8*)SvPVX(sv));
402 Perl_do_trans(pTHX_ SV *sv)
407 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
408 Perl_croak(aTHX_ PL_no_modify);
414 (void)SvPV_force(sv, len);
415 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
416 (void)SvPOK_only_UTF8(sv);
418 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
420 switch (PL_op->op_private & 63) {
423 return do_trans_UU_simple(sv);
425 return do_trans_CC_simple(sv);
427 case OPpTRANS_IDENTICAL:
429 return do_trans_UU_count(sv);
431 return do_trans_CC_count(sv);
435 return do_trans_UU_complex(sv); /* could be UC or CU too */
437 return do_trans_CC_complex(sv);
442 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
445 register I32 items = sp - mark;
448 register char *delim = SvPV(del, delimlen);
452 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
453 (void)SvUPGRADE(sv, SVt_PV);
454 if (SvLEN(sv) < len + items) { /* current length is way too short */
455 while (items-- > 0) {
456 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
462 SvGROW(sv, len + 1); /* so try to pre-extend */
481 for (; items > 0; items--,mark++) {
482 sv_catpvn(sv,delim,len);
487 for (; items > 0; items--,mark++)
494 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
497 char *pat = SvPV(*sarg, patlen);
498 bool do_taint = FALSE;
500 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
506 /* XXX SvUTF8 support missing! */
508 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
511 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
516 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
517 Perl_croak(aTHX_ "Illegal number of bits in vec");
518 offset *= size; /* turn into bit offset */
519 len = (offset + size + 7) / 8; /* required number of bytes */
524 offset >>= 3; /* turn into byte offset */
526 if (offset >= srclen)
529 retnum = (UV) s[offset] << 8;
531 else if (size == 32) {
532 if (offset >= srclen)
534 else if (offset + 1 >= srclen)
536 ((UV) s[offset ] << 24);
537 else if (offset + 2 >= srclen)
539 ((UV) s[offset ] << 24) +
540 ((UV) s[offset + 1] << 16);
543 ((UV) s[offset ] << 24) +
544 ((UV) s[offset + 1] << 16) +
545 ( s[offset + 2] << 8);
548 else if (size == 64) {
550 if (ckWARN(WARN_PORTABLE))
551 Perl_warner(aTHX_ WARN_PORTABLE,
552 "Bit vector size > 32 non-portable");
553 if (offset >= srclen)
555 else if (offset + 1 >= srclen)
557 (UV) s[offset ] << 56;
558 else if (offset + 2 >= srclen)
560 ((UV) s[offset ] << 56) +
561 ((UV) s[offset + 1] << 48);
562 else if (offset + 3 >= srclen)
564 ((UV) s[offset ] << 56) +
565 ((UV) s[offset + 1] << 48) +
566 ((UV) s[offset + 2] << 40);
567 else if (offset + 4 >= srclen)
569 ((UV) s[offset ] << 56) +
570 ((UV) s[offset + 1] << 48) +
571 ((UV) s[offset + 2] << 40) +
572 ((UV) s[offset + 3] << 32);
573 else if (offset + 5 >= srclen)
575 ((UV) s[offset ] << 56) +
576 ((UV) s[offset + 1] << 48) +
577 ((UV) s[offset + 2] << 40) +
578 ((UV) s[offset + 3] << 32) +
579 ( s[offset + 4] << 24);
580 else if (offset + 6 >= srclen)
582 ((UV) s[offset ] << 56) +
583 ((UV) s[offset + 1] << 48) +
584 ((UV) s[offset + 2] << 40) +
585 ((UV) s[offset + 3] << 32) +
586 ((UV) s[offset + 4] << 24) +
587 ((UV) s[offset + 5] << 16);
590 ((UV) s[offset ] << 56) +
591 ((UV) s[offset + 1] << 48) +
592 ((UV) s[offset + 2] << 40) +
593 ((UV) s[offset + 3] << 32) +
594 ((UV) s[offset + 4] << 24) +
595 ((UV) s[offset + 5] << 16) +
596 ( s[offset + 6] << 8);
602 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
604 offset >>= 3; /* turn into byte offset */
609 ((UV) s[offset] << 8) +
613 ((UV) s[offset ] << 24) +
614 ((UV) s[offset + 1] << 16) +
615 ( s[offset + 2] << 8) +
618 else if (size == 64) {
620 if (ckWARN(WARN_PORTABLE))
621 Perl_warner(aTHX_ WARN_PORTABLE,
622 "Bit vector size > 32 non-portable");
624 ((UV) s[offset ] << 56) +
625 ((UV) s[offset + 1] << 48) +
626 ((UV) s[offset + 2] << 40) +
627 ((UV) s[offset + 3] << 32) +
628 ((UV) s[offset + 4] << 24) +
629 ((UV) s[offset + 5] << 16) +
630 ( s[offset + 6] << 8) +
639 /* XXX SvUTF8 support missing! */
641 Perl_do_vecset(pTHX_ SV *sv)
643 SV *targ = LvTARG(sv);
646 register unsigned char *s;
654 s = (unsigned char*)SvPV_force(targ, targlen);
655 (void)SvPOK_only(targ);
657 offset = LvTARGOFF(sv);
658 size = LvTARGLEN(sv);
659 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
660 Perl_croak(aTHX_ "Illegal number of bits in vec");
662 offset *= size; /* turn into bit offset */
663 len = (offset + size + 7) / 8; /* required number of bytes */
665 s = (unsigned char*)SvGROW(targ, len + 1);
666 (void)memzero((char *)(s + targlen), len - targlen + 1);
667 SvCUR_set(targ, len);
671 mask = (1 << size) - 1;
674 offset >>= 3; /* turn into byte offset */
675 s[offset] &= ~(mask << size);
676 s[offset] |= lval << size;
679 offset >>= 3; /* turn into byte offset */
681 s[offset ] = lval & 0xff;
682 else if (size == 16) {
683 s[offset ] = (lval >> 8) & 0xff;
684 s[offset+1] = lval & 0xff;
686 else if (size == 32) {
687 s[offset ] = (lval >> 24) & 0xff;
688 s[offset+1] = (lval >> 16) & 0xff;
689 s[offset+2] = (lval >> 8) & 0xff;
690 s[offset+3] = lval & 0xff;
693 else if (size == 64) {
695 if (ckWARN(WARN_PORTABLE))
696 Perl_warner(aTHX_ WARN_PORTABLE,
697 "Bit vector size > 32 non-portable");
698 s[offset ] = (lval >> 56) & 0xff;
699 s[offset+1] = (lval >> 48) & 0xff;
700 s[offset+2] = (lval >> 40) & 0xff;
701 s[offset+3] = (lval >> 32) & 0xff;
702 s[offset+4] = (lval >> 24) & 0xff;
703 s[offset+5] = (lval >> 16) & 0xff;
704 s[offset+6] = (lval >> 8) & 0xff;
705 s[offset+7] = lval & 0xff;
713 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
719 if (SvTYPE(sv) == SVt_PVAV) {
724 for (i = 0; i <= max; i++) {
725 sv = (SV*)av_fetch(av, i, FALSE);
726 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
731 else if (SvTYPE(sv) == SVt_PVHV) {
734 (void)hv_iterinit(hv);
736 while ((entry = hv_iternext(hv)))
737 do_chop(astr,hv_iterval(hv,entry));
740 else if (SvREADONLY(sv))
741 Perl_croak(aTHX_ PL_no_modify);
743 if (len && !SvPOK(sv))
744 s = SvPV_force(sv, len);
747 char *send = s + len;
750 while ((*s & 0xc0) == 0x80)
752 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
753 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
754 sv_setpvn(astr, s, send - s);
756 SvCUR_set(sv, s - start);
761 sv_setpvn(astr, "", 0);
765 sv_setpvn(astr, s, 1);
772 sv_setpvn(astr, "", 0);
777 Perl_do_chomp(pTHX_ register SV *sv)
789 if (SvTYPE(sv) == SVt_PVAV) {
794 for (i = 0; i <= max; i++) {
795 sv = (SV*)av_fetch(av, i, FALSE);
796 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
797 count += do_chomp(sv);
801 else if (SvTYPE(sv) == SVt_PVHV) {
804 (void)hv_iterinit(hv);
806 while ((entry = hv_iternext(hv)))
807 count += do_chomp(hv_iterval(hv,entry));
810 else if (SvREADONLY(sv))
811 Perl_croak(aTHX_ PL_no_modify);
813 if (len && !SvPOKp(sv))
814 s = SvPV_force(sv, len);
821 while (len && s[-1] == '\n') {
829 char *rsptr = SvPV(PL_rs, rslen);
840 if (memNE(s, rsptr, rslen))
855 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
857 dTHR; /* just for taint */
872 bool left_utf = DO_UTF8(left);
873 bool right_utf = DO_UTF8(right);
875 if (left_utf && !right_utf)
876 sv_utf8_upgrade(right);
877 if (!left_utf && right_utf)
878 sv_utf8_upgrade(left);
880 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
881 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
882 lsave = lc = SvPV(left, leftlen);
883 rsave = rc = SvPV(right, rightlen);
884 len = leftlen < rightlen ? leftlen : rightlen;
886 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
888 dc = SvPV_force(sv, n_a);
889 if (SvCUR(sv) < len) {
890 dc = SvGROW(sv, len + 1);
891 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
895 I32 needlen = ((optype == OP_BIT_AND)
896 ? len : (leftlen > rightlen ? leftlen : rightlen));
897 Newz(801, dc, needlen + 1, char);
898 (void)sv_usepvn(sv, dc, needlen);
899 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
902 (void)SvPOK_only(sv);
903 if (left_utf || right_utf) {
905 STRLEN lulen = leftlen;
906 STRLEN rulen = rightlen;
910 if (optype != OP_BIT_AND)
911 dc = SvGROW(sv, leftlen+rightlen+1);
915 while (lulen && rulen) {
916 luc = utf8_to_uv((U8*)lc, &ulen);
919 ruc = utf8_to_uv((U8*)rc, &ulen);
923 dc = (char*)uv_to_utf8((U8*)dc, duc);
925 dulen = dc - SvPVX(sv);
926 SvCUR_set(sv, dulen);
929 while (lulen && rulen) {
930 luc = utf8_to_uv((U8*)lc, &ulen);
933 ruc = utf8_to_uv((U8*)rc, &ulen);
937 dc = (char*)uv_to_utf8((U8*)dc, duc);
941 while (lulen && rulen) {
942 luc = utf8_to_uv((U8*)lc, &ulen);
945 ruc = utf8_to_uv((U8*)rc, &ulen);
949 dc = (char*)uv_to_utf8((U8*)dc, duc);
952 dulen = dc - SvPVX(sv);
953 SvCUR_set(sv, dulen);
955 sv_catpvn(sv, rc, rulen);
957 sv_catpvn(sv, lc, lulen);
967 if (len >= sizeof(long)*4 &&
968 !((long)dc % sizeof(long)) &&
969 !((long)lc % sizeof(long)) &&
970 !((long)rc % sizeof(long))) /* It's almost always aligned... */
972 I32 remainder = len % (sizeof(long)*4);
973 len /= (sizeof(long)*4);
982 *dl++ = *ll++ & *rl++;
983 *dl++ = *ll++ & *rl++;
984 *dl++ = *ll++ & *rl++;
985 *dl++ = *ll++ & *rl++;
990 *dl++ = *ll++ ^ *rl++;
991 *dl++ = *ll++ ^ *rl++;
992 *dl++ = *ll++ ^ *rl++;
993 *dl++ = *ll++ ^ *rl++;
998 *dl++ = *ll++ | *rl++;
999 *dl++ = *ll++ | *rl++;
1000 *dl++ = *ll++ | *rl++;
1001 *dl++ = *ll++ | *rl++;
1016 *dc++ = *lc++ & *rc++;
1020 *dc++ = *lc++ ^ *rc++;
1024 *dc++ = *lc++ | *rc++;
1028 sv_catpvn(sv, rsave + len, rightlen - len);
1029 else if (leftlen > len)
1030 sv_catpvn(sv, lsave + len, leftlen - len);
1048 I32 gimme = GIMME_V;
1049 I32 dokeys = (PL_op->op_type == OP_KEYS);
1050 I32 dovalues = (PL_op->op_type == OP_VALUES);
1051 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1053 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1054 dokeys = dovalues = TRUE;
1057 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1058 dTARGET; /* make sure to clear its target here */
1059 if (SvTYPE(TARG) == SVt_PVLV)
1060 LvTARG(TARG) = Nullsv;
1066 keys = realhv ? hv : avhv_keys((AV*)hv);
1067 (void)hv_iterinit(keys); /* always reset iterator regardless */
1069 if (gimme == G_VOID)
1072 if (gimme == G_SCALAR) {
1076 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1077 if (SvTYPE(TARG) < SVt_PVLV) {
1078 sv_upgrade(TARG, SVt_PVLV);
1079 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1082 if (LvTARG(TARG) != (SV*)keys) {
1084 SvREFCNT_dec(LvTARG(TARG));
1085 LvTARG(TARG) = SvREFCNT_inc(keys);
1091 if (! SvTIED_mg((SV*)keys, 'P'))
1096 while (hv_iternext(keys)) i++;
1102 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1104 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1105 while ((entry = hv_iternext(keys))) {
1108 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1112 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1113 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1114 (unsigned long)HeHASH(entry),
1116 (unsigned long)(HeHASH(entry) & HvMAX(keys))));