3 * Copyright (c) 1991-2001, 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) {
69 d = uv_to_utf8(d, ch);
72 else { /* No match -> copy */
78 sv_setpvn(sv, (char*)dstart, d - dstart);
85 S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
93 tbl = (short*)cPVOP->op_pv;
95 Perl_croak(aTHX_ "panic: do_trans_count");
97 s = (U8*)SvPV(sv, len);
109 c = utf8_to_uv(s, send - s, &ulen, 0);
110 if (c < 0x100 && tbl[c] >= 0)
119 S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
131 tbl = (short*)cPVOP->op_pv;
133 Perl_croak(aTHX_ "panic: do_trans_complex");
135 s = (U8*)SvPV(sv, len);
141 if (PL_op->op_private & OPpTRANS_SQUASH) {
144 if ((ch = tbl[*s]) >= 0) {
147 if (p != d - 1 || *p != *d)
150 else if (ch == -1) /* -1 is unmapped character */
152 else if (ch == -2) /* -2 is delete character */
159 if ((ch = tbl[*s]) >= 0) {
163 else if (ch == -1) /* -1 is unmapped character */
165 else if (ch == -2) /* -2 is delete character */
170 SvCUR_set(sv, d - dstart);
173 Newz(0, d, len*2+1, U8);
176 if (PL_op->op_private & OPpTRANS_SQUASH) {
181 UV comp = utf8_to_uv_simple(s, &len);
184 d = uv_to_utf8(d, comp); /* always unmapped */
185 else if ((ch = tbl[comp]) >= 0) {
188 d = uv_to_utf8(d, ch);
194 else if (ch == -1) /* -1 is unmapped character */
195 d = uv_to_utf8(d, comp);
196 else if (ch == -2) /* -2 is delete character */
205 UV comp = utf8_to_uv_simple(s, &len);
207 d = uv_to_utf8(d, comp); /* always unmapped */
208 else if ((ch = tbl[comp]) >= 0) {
209 d = uv_to_utf8(d, ch);
212 else if (ch == -1) { /* -1 is unmapped character */
213 d = uv_to_utf8(d, comp);
215 else if (ch == -2) /* -2 is delete character */
221 sv_setpvn(sv, (char*)dstart, d - dstart);
229 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
239 SV* rv = (SV*)cSVOP->op_sv;
240 HV* hv = (HV*)SvRV(rv);
241 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
242 UV none = svp ? SvUV(*svp) : 0x7fffffff;
249 s = (U8*)SvPV(sv, len);
252 U8 *t = s, *e = s + len;
254 if ((hibit = UTF8_IS_CONTINUED(*t++)))
257 s = bytes_to_utf8(s, &len);
262 svp = hv_fetch(hv, "FINAL", 5, FALSE);
266 /* d needs to be bigger than s, in case e.g. upgrading is required */
267 New(0, d, len*3+UTF8_MAXLEN, U8);
272 if ((uv = swash_fetch(rv, s)) < none) {
275 d = uv_to_utf8(d, uv);
277 else if (uv == none) {
282 else if (uv == extra) {
286 d = uv_to_utf8(d, final);
292 STRLEN clen = d - dstart;
293 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
294 Renew(dstart, nlen+UTF8_MAXLEN, U8);
296 dend = dstart + nlen;
300 sv_setpvn(sv, (char*)dstart, d - dstart);
305 if (!isutf8 && !(PL_hints & HINT_UTF8))
306 sv_utf8_downgrade(sv, TRUE);
312 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
319 SV* rv = (SV*)cSVOP->op_sv;
320 HV* hv = (HV*)SvRV(rv);
321 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
322 UV none = svp ? SvUV(*svp) : 0x7fffffff;
326 s = (U8*)SvPV(sv, len);
328 U8 *t = s, *e = s + len;
330 if ((hibit = !UTF8_IS_ASCII(*t++)))
333 start = s = bytes_to_utf8(s, &len);
338 if ((uv = swash_fetch(rv, s)) < none)
349 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
355 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
356 I32 del = PL_op->op_private & OPpTRANS_DELETE;
357 SV* rv = (SV*)cSVOP->op_sv;
358 HV* hv = (HV*)SvRV(rv);
359 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
360 UV none = svp ? SvUV(*svp) : 0x7fffffff;
369 s = (U8*)SvPV(sv, len);
372 U8 *t = s, *e = s + len;
374 if ((hibit = !UTF8_IS_ASCII(*t++)))
377 s = bytes_to_utf8(s, &len);
382 svp = hv_fetch(hv, "FINAL", 5, FALSE);
386 New(0, d, len*3+UTF8_MAXLEN, U8);
393 uv = swash_fetch(rv, s);
396 STRLEN clen = d - dstart, nlen = dend - dstart + len;
397 Renew(dstart, nlen+UTF8_MAXLEN, U8);
399 dend = dstart + nlen;
404 d = uv_to_utf8(d, uv);
410 else if (uv == none) { /* "none" is unmapped character */
417 else if (uv == extra && !del) {
420 d = uv_to_utf8(d, final);
426 matches++; /* "none+1" is delete character */
432 uv = swash_fetch(rv, s);
434 STRLEN clen = d - dstart, nlen = dend - dstart + len;
435 Renew(dstart, nlen+UTF8_MAXLEN, U8);
437 dend = dstart + nlen;
441 d = uv_to_utf8(d, uv);
445 else if (uv == none) { /* "none" is unmapped character */
451 else if (uv == extra && !del) {
453 d = uv_to_utf8(d, final);
457 matches++; /* "none+1" is delete character */
462 sv_setpvn(sv, (char*)dstart, d - dstart);
466 if (!isutf8 && !(PL_hints & HINT_UTF8))
467 sv_utf8_downgrade(sv, TRUE);
474 Perl_do_trans(pTHX_ SV *sv)
477 I32 hasutf = (PL_op->op_private &
478 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
480 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
481 Perl_croak(aTHX_ PL_no_modify);
487 (void)SvPV_force(sv, len);
488 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
489 (void)SvPOK_only_UTF8(sv);
491 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
493 switch (PL_op->op_private & ~hasutf & 63) {
496 return do_trans_simple_utf8(sv);
498 return do_trans_simple(sv);
500 case OPpTRANS_IDENTICAL:
502 return do_trans_count_utf8(sv);
504 return do_trans_count(sv);
508 return do_trans_complex_utf8(sv);
510 return do_trans_complex(sv);
515 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
518 register I32 items = sp - mark;
521 register char *delim = SvPV(del, delimlen);
525 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
526 (void)SvUPGRADE(sv, SVt_PV);
527 if (SvLEN(sv) < len + items) { /* current length is way too short */
528 while (items-- > 0) {
529 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
535 SvGROW(sv, len + 1); /* so try to pre-extend */
551 for (; items > 0; items--,mark++) {
557 for (; items > 0; items--,mark++)
564 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
567 char *pat = SvPV(*sarg, patlen);
568 bool do_taint = FALSE;
570 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
576 /* currently converts input to bytes if possible, but doesn't sweat failure */
578 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
581 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
586 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
587 Perl_croak(aTHX_ "Illegal number of bits in vec");
590 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
592 offset *= size; /* turn into bit offset */
593 len = (offset + size + 7) / 8; /* required number of bytes */
598 offset >>= 3; /* turn into byte offset */
600 if (offset >= srclen)
603 retnum = (UV) s[offset] << 8;
605 else if (size == 32) {
606 if (offset >= srclen)
608 else if (offset + 1 >= srclen)
610 ((UV) s[offset ] << 24);
611 else if (offset + 2 >= srclen)
613 ((UV) s[offset ] << 24) +
614 ((UV) s[offset + 1] << 16);
617 ((UV) s[offset ] << 24) +
618 ((UV) s[offset + 1] << 16) +
619 ( s[offset + 2] << 8);
622 else if (size == 64) {
623 if (ckWARN(WARN_PORTABLE))
624 Perl_warner(aTHX_ WARN_PORTABLE,
625 "Bit vector size > 32 non-portable");
626 if (offset >= srclen)
628 else if (offset + 1 >= srclen)
630 (UV) s[offset ] << 56;
631 else if (offset + 2 >= srclen)
633 ((UV) s[offset ] << 56) +
634 ((UV) s[offset + 1] << 48);
635 else if (offset + 3 >= srclen)
637 ((UV) s[offset ] << 56) +
638 ((UV) s[offset + 1] << 48) +
639 ((UV) s[offset + 2] << 40);
640 else if (offset + 4 >= srclen)
642 ((UV) s[offset ] << 56) +
643 ((UV) s[offset + 1] << 48) +
644 ((UV) s[offset + 2] << 40) +
645 ((UV) s[offset + 3] << 32);
646 else if (offset + 5 >= srclen)
648 ((UV) s[offset ] << 56) +
649 ((UV) s[offset + 1] << 48) +
650 ((UV) s[offset + 2] << 40) +
651 ((UV) s[offset + 3] << 32) +
652 ( s[offset + 4] << 24);
653 else if (offset + 6 >= srclen)
655 ((UV) s[offset ] << 56) +
656 ((UV) s[offset + 1] << 48) +
657 ((UV) s[offset + 2] << 40) +
658 ((UV) s[offset + 3] << 32) +
659 ((UV) s[offset + 4] << 24) +
660 ((UV) s[offset + 5] << 16);
663 ((UV) s[offset ] << 56) +
664 ((UV) s[offset + 1] << 48) +
665 ((UV) s[offset + 2] << 40) +
666 ((UV) s[offset + 3] << 32) +
667 ((UV) s[offset + 4] << 24) +
668 ((UV) s[offset + 5] << 16) +
669 ( s[offset + 6] << 8);
675 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
677 offset >>= 3; /* turn into byte offset */
682 ((UV) s[offset] << 8) +
686 ((UV) s[offset ] << 24) +
687 ((UV) s[offset + 1] << 16) +
688 ( s[offset + 2] << 8) +
691 else if (size == 64) {
692 if (ckWARN(WARN_PORTABLE))
693 Perl_warner(aTHX_ WARN_PORTABLE,
694 "Bit vector size > 32 non-portable");
696 ((UV) s[offset ] << 56) +
697 ((UV) s[offset + 1] << 48) +
698 ((UV) s[offset + 2] << 40) +
699 ((UV) s[offset + 3] << 32) +
700 ((UV) s[offset + 4] << 24) +
701 ((UV) s[offset + 5] << 16) +
702 ( s[offset + 6] << 8) +
711 /* currently converts input to bytes if possible but doesn't sweat failures,
712 * although it does ensure that the string it clobbers is not marked as
713 * utf8-valid any more
716 Perl_do_vecset(pTHX_ SV *sv)
718 SV *targ = LvTARG(sv);
721 register unsigned char *s;
729 s = (unsigned char*)SvPV_force(targ, targlen);
731 /* This is handled by the SvPOK_only below...
732 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
735 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
738 (void)SvPOK_only(targ);
740 offset = LvTARGOFF(sv);
742 Perl_croak(aTHX_ "Assigning to negative offset in vec");
743 size = LvTARGLEN(sv);
744 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
745 Perl_croak(aTHX_ "Illegal number of bits in vec");
747 offset *= size; /* turn into bit offset */
748 len = (offset + size + 7) / 8; /* required number of bytes */
750 s = (unsigned char*)SvGROW(targ, len + 1);
751 (void)memzero((char *)(s + targlen), len - targlen + 1);
752 SvCUR_set(targ, len);
756 mask = (1 << size) - 1;
759 offset >>= 3; /* turn into byte offset */
760 s[offset] &= ~(mask << size);
761 s[offset] |= lval << size;
764 offset >>= 3; /* turn into byte offset */
766 s[offset ] = lval & 0xff;
767 else if (size == 16) {
768 s[offset ] = (lval >> 8) & 0xff;
769 s[offset+1] = lval & 0xff;
771 else if (size == 32) {
772 s[offset ] = (lval >> 24) & 0xff;
773 s[offset+1] = (lval >> 16) & 0xff;
774 s[offset+2] = (lval >> 8) & 0xff;
775 s[offset+3] = lval & 0xff;
778 else if (size == 64) {
779 if (ckWARN(WARN_PORTABLE))
780 Perl_warner(aTHX_ WARN_PORTABLE,
781 "Bit vector size > 32 non-portable");
782 s[offset ] = (lval >> 56) & 0xff;
783 s[offset+1] = (lval >> 48) & 0xff;
784 s[offset+2] = (lval >> 40) & 0xff;
785 s[offset+3] = (lval >> 32) & 0xff;
786 s[offset+4] = (lval >> 24) & 0xff;
787 s[offset+5] = (lval >> 16) & 0xff;
788 s[offset+6] = (lval >> 8) & 0xff;
789 s[offset+7] = lval & 0xff;
797 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
802 if (SvTYPE(sv) == SVt_PVAV) {
807 for (i = 0; i <= max; i++) {
808 sv = (SV*)av_fetch(av, i, FALSE);
809 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
814 else if (SvTYPE(sv) == SVt_PVHV) {
817 (void)hv_iterinit(hv);
819 while ((entry = hv_iternext(hv)))
820 do_chop(astr,hv_iterval(hv,entry));
823 else if (SvREADONLY(sv))
824 Perl_croak(aTHX_ PL_no_modify);
826 if (len && !SvPOK(sv))
827 s = SvPV_force(sv, len);
830 char *send = s + len;
833 while (s > start && UTF8_IS_CONTINUATION(*s))
835 if (utf8_to_uv_simple((U8*)s, 0)) {
836 sv_setpvn(astr, s, send - s);
838 SvCUR_set(sv, s - start);
844 sv_setpvn(astr, "", 0);
848 sv_setpvn(astr, s, 1);
855 sv_setpvn(astr, "", 0);
860 Perl_do_chomp(pTHX_ register SV *sv)
871 if (SvTYPE(sv) == SVt_PVAV) {
876 for (i = 0; i <= max; i++) {
877 sv = (SV*)av_fetch(av, i, FALSE);
878 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
879 count += do_chomp(sv);
883 else if (SvTYPE(sv) == SVt_PVHV) {
886 (void)hv_iterinit(hv);
888 while ((entry = hv_iternext(hv)))
889 count += do_chomp(hv_iterval(hv,entry));
892 else if (SvREADONLY(sv))
893 Perl_croak(aTHX_ PL_no_modify);
895 if (len && !SvPOKp(sv))
896 s = SvPV_force(sv, len);
903 while (len && s[-1] == '\n') {
911 char *rsptr = SvPV(PL_rs, rslen);
922 if (memNE(s, rsptr, rslen))
937 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
953 bool left_utf = DO_UTF8(left);
954 bool right_utf = DO_UTF8(right);
957 if (left_utf && !right_utf)
958 sv_utf8_upgrade(right);
959 else if (!left_utf && right_utf)
960 sv_utf8_upgrade(left);
962 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
963 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
964 lsave = lc = SvPV(left, leftlen);
965 rsave = rc = SvPV(right, rightlen);
966 len = leftlen < rightlen ? leftlen : rightlen;
968 if ((left_utf || right_utf) && (sv == left || sv == right)) {
969 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
970 Newz(801, dc, needlen + 1, char);
972 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
974 dc = SvPV_force(sv, n_a);
975 if (SvCUR(sv) < len) {
976 dc = SvGROW(sv, len + 1);
977 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
979 if (optype != OP_BIT_AND && (left_utf || right_utf))
980 dc = SvGROW(sv, leftlen + rightlen + 1);
983 needlen = ((optype == OP_BIT_AND)
984 ? len : (leftlen > rightlen ? leftlen : rightlen));
985 Newz(801, dc, needlen + 1, char);
986 (void)sv_usepvn(sv, dc, needlen);
987 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
990 (void)SvPOK_only(sv);
991 if (left_utf || right_utf) {
994 STRLEN lulen = leftlen;
995 STRLEN rulen = rightlen;
1000 while (lulen && rulen) {
1001 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1004 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1008 dc = (char*)uv_to_utf8((U8*)dc, duc);
1010 if (sv == left || sv == right)
1011 (void)sv_usepvn(sv, dcsave, needlen);
1012 SvCUR_set(sv, dc - dcsave);
1015 while (lulen && rulen) {
1016 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1019 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1023 dc = (char*)uv_to_utf8((U8*)dc, duc);
1027 while (lulen && rulen) {
1028 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1031 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1035 dc = (char*)uv_to_utf8((U8*)dc, duc);
1038 if (sv == left || sv == right)
1039 (void)sv_usepvn(sv, dcsave, needlen);
1040 SvCUR_set(sv, dc - dcsave);
1042 sv_catpvn(sv, rc, rulen);
1044 sv_catpvn(sv, lc, lulen);
1054 if (len >= sizeof(long)*4 &&
1055 !((long)dc % sizeof(long)) &&
1056 !((long)lc % sizeof(long)) &&
1057 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1059 I32 remainder = len % (sizeof(long)*4);
1060 len /= (sizeof(long)*4);
1069 *dl++ = *ll++ & *rl++;
1070 *dl++ = *ll++ & *rl++;
1071 *dl++ = *ll++ & *rl++;
1072 *dl++ = *ll++ & *rl++;
1077 *dl++ = *ll++ ^ *rl++;
1078 *dl++ = *ll++ ^ *rl++;
1079 *dl++ = *ll++ ^ *rl++;
1080 *dl++ = *ll++ ^ *rl++;
1085 *dl++ = *ll++ | *rl++;
1086 *dl++ = *ll++ | *rl++;
1087 *dl++ = *ll++ | *rl++;
1088 *dl++ = *ll++ | *rl++;
1103 *dc++ = *lc++ & *rc++;
1107 *dc++ = *lc++ ^ *rc++;
1111 *dc++ = *lc++ | *rc++;
1115 sv_catpvn(sv, rsave + len, rightlen - len);
1116 else if (leftlen > len)
1117 sv_catpvn(sv, lsave + len, leftlen - len);
1135 I32 gimme = GIMME_V;
1136 I32 dokeys = (PL_op->op_type == OP_KEYS);
1137 I32 dovalues = (PL_op->op_type == OP_VALUES);
1138 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1140 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1141 dokeys = dovalues = TRUE;
1144 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1145 dTARGET; /* make sure to clear its target here */
1146 if (SvTYPE(TARG) == SVt_PVLV)
1147 LvTARG(TARG) = Nullsv;
1153 keys = realhv ? hv : avhv_keys((AV*)hv);
1154 (void)hv_iterinit(keys); /* always reset iterator regardless */
1156 if (gimme == G_VOID)
1159 if (gimme == G_SCALAR) {
1163 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1164 if (SvTYPE(TARG) < SVt_PVLV) {
1165 sv_upgrade(TARG, SVt_PVLV);
1166 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1169 if (LvTARG(TARG) != (SV*)keys) {
1171 SvREFCNT_dec(LvTARG(TARG));
1172 LvTARG(TARG) = SvREFCNT_inc(keys);
1178 if (! SvTIED_mg((SV*)keys, 'P'))
1183 while (hv_iternext(keys)) i++;
1189 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1191 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1192 while ((entry = hv_iternext(keys))) {
1195 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1199 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1200 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1201 (unsigned long)HeHASH(entry),
1203 (unsigned long)(HeHASH(entry) & HvMAX(keys))));