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)
24 #define HALF_UTF8_UPGRADE(start,end) \
26 if ((start)<(end)) { \
28 STRLEN LeN = (end) - (start); \
29 NeWsTr = bytes_to_utf8(start, &LeN); \
32 (end) = (start) + LeN; \
37 S_do_trans_simple(pTHX_ SV *sv)
45 I32 sutf = SvUTF8(sv);
50 tbl = (short*)cPVOP->op_pv;
52 Perl_croak(aTHX_ "panic: do_trans");
54 s = (U8*)SvPV(sv, len);
57 /* First, take care of non-UTF8 input strings, because they're easy */
60 if ((ch = tbl[*s]) >= 0) {
71 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
72 Newz(0, d, len*2+1, U8);
79 /* Need to check this, otherwise 128..255 won't match */
80 c = utf8_to_uv(s, &ulen);
81 if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
89 else { /* No match -> copy */
95 sv_setpvn(sv, (const char*)dstart, d - dstart);
103 S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
109 I32 hasutf = SvUTF8(sv);
113 tbl = (short*)cPVOP->op_pv;
115 Perl_croak(aTHX_ "panic: do_trans");
117 s = (U8*)SvPV(sv, len);
121 if (hasutf && *s & 0x80)
128 c = utf8_to_uv(s,&ulen);
131 if (c < 0x100 && tbl[c] >= 0)
141 S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
147 I32 hasutf = SvUTF8(sv);
153 tbl = (short*)cPVOP->op_pv;
155 Perl_croak(aTHX_ "panic: do_trans");
157 s = (U8*)SvPV(sv, len);
161 if (PL_op->op_private & OPpTRANS_SQUASH) {
165 if (hasutf && *s & 0x80)
168 if ((ch = tbl[*s]) >= 0) {
171 if (p == d - 1 && *p == *d)
176 else if (ch == -1) /* -1 is unmapped character */
177 *d++ = *s; /* -2 is delete character */
184 if (hasutf && *s & 0x80)
187 if ((ch = tbl[*s]) >= 0) {
192 else if (ch == -1) /* -1 is unmapped character */
193 *d++ = *s; /* -2 is delete character */
198 matches += send - d; /* account for disappeared chars */
200 SvCUR_set(sv, d - (U8*)SvPVX(sv));
207 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
218 SV* rv = (SV*)cSVOP->op_sv;
219 HV* hv = (HV*)SvRV(rv);
220 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
221 UV none = svp ? SvUV(*svp) : 0x7fffffff;
229 s = (U8*)SvPV(sv, len);
233 svp = hv_fetch(hv, "FINAL", 5, FALSE);
237 /* d needs to be bigger than s, in case e.g. upgrading is required */
238 Newz(0, d, len*2+1, U8);
241 if ((uv = swash_fetch(rv, s)) < none) {
244 if ((uv & 0x80) && !isutf++)
245 HALF_UTF8_UPGRADE(dstart,d);
246 d = uv_to_utf8(d, uv);
248 else if (uv == none) {
251 if (i > 1 && !isutf++)
252 HALF_UTF8_UPGRADE(dstart,d);
256 else if (uv == extra) {
261 if (i > 1 && !isutf++)
262 HALF_UTF8_UPGRADE(dstart,d);
263 d = uv_to_utf8(d, final);
269 sv_setpvn(sv, (const char*)dstart, d - dstart);
278 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
286 SV* rv = (SV*)cSVOP->op_sv;
287 HV* hv = (HV*)SvRV(rv);
288 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
289 UV none = svp ? SvUV(*svp) : 0x7fffffff;
292 s = (U8*)SvPV(sv, len);
294 s = bytes_to_utf8(s, &len);
298 if ((uv = swash_fetch(rv, s)) < none)
307 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
314 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
315 I32 del = PL_op->op_private & OPpTRANS_DELETE;
316 SV* rv = (SV*)cSVOP->op_sv;
317 HV* hv = (HV*)SvRV(rv);
318 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
319 UV none = svp ? SvUV(*svp) : 0x7fffffff;
325 I32 isutf = SvUTF8(sv);
327 s = (U8*)SvPV(sv, len);
330 svp = hv_fetch(hv, "FINAL", 5, FALSE);
334 Newz(0, d, len*2+1, U8);
341 uv = swash_fetch(rv, s);
348 tmpbuf[0] = (( uv >> 6) | 0xc0);
349 tmpbuf[1] = (( uv & 0x3f) | 0x80);
351 uv = swash_fetch(rv, tmpbuf);
357 if ((uv & 0x80) && !isutf++)
358 HALF_UTF8_UPGRADE(dst,d);
359 d = uv_to_utf8(d, uv);
365 else if (uv == none) { /* "none" is unmapped character */
367 *d++ = (U8)utf8_to_uv(s, &ulen);
372 else if (uv == extra && !del) {
375 d = uv_to_utf8(d, final);
381 matches++; /* "none+1" is delete character */
388 uv = swash_fetch(rv, s);
395 tmpbuf[0] = (( uv >> 6) | 0xc0);
396 tmpbuf[1] = (( uv & 0x3f) | 0x80);
398 uv = swash_fetch(rv, tmpbuf);
402 d = uv_to_utf8(d, uv);
406 else if (uv == none) { /* "none" is unmapped character */
408 *d++ = (U8)utf8_to_uv(s, &ulen);
412 else if (uv == extra && !del) {
414 d = uv_to_utf8(d, final);
418 matches++; /* "none+1" is delete character */
423 sv_usepvn(sv, (char*)dst, d - dst);
426 SvCUR_set(sv, d - (U8*)SvPVX(sv));
434 Perl_do_trans(pTHX_ SV *sv)
438 I32 hasutf = (PL_op->op_private &
439 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
441 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
442 Perl_croak(aTHX_ PL_no_modify);
448 (void)SvPV_force(sv, len);
449 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
450 (void)SvPOK_only_UTF8(sv);
452 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
454 switch (PL_op->op_private & ~hasutf & 63) {
457 return do_trans_simple_utf8(sv);
459 return do_trans_simple(sv);
461 case OPpTRANS_IDENTICAL:
463 return do_trans_count_utf8(sv);
465 return do_trans_count(sv);
469 return do_trans_complex_utf8(sv);
471 return do_trans_complex(sv);
476 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
479 register I32 items = sp - mark;
482 register char *delim = SvPV(del, delimlen);
486 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
487 (void)SvUPGRADE(sv, SVt_PV);
488 if (SvLEN(sv) < len + items) { /* current length is way too short */
489 while (items-- > 0) {
490 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
496 SvGROW(sv, len + 1); /* so try to pre-extend */
515 for (; items > 0; items--,mark++) {
516 sv_catpvn(sv,delim,len);
521 for (; items > 0; items--,mark++)
528 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
531 char *pat = SvPV(*sarg, patlen);
532 bool do_taint = FALSE;
534 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
540 /* XXX SvUTF8 support missing! */
542 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
545 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
550 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
551 Perl_croak(aTHX_ "Illegal number of bits in vec");
552 offset *= size; /* turn into bit offset */
553 len = (offset + size + 7) / 8; /* required number of bytes */
558 offset >>= 3; /* turn into byte offset */
560 if (offset >= srclen)
563 retnum = (UV) s[offset] << 8;
565 else if (size == 32) {
566 if (offset >= srclen)
568 else if (offset + 1 >= srclen)
570 ((UV) s[offset ] << 24);
571 else if (offset + 2 >= srclen)
573 ((UV) s[offset ] << 24) +
574 ((UV) s[offset + 1] << 16);
577 ((UV) s[offset ] << 24) +
578 ((UV) s[offset + 1] << 16) +
579 ( s[offset + 2] << 8);
582 else if (size == 64) {
584 if (ckWARN(WARN_PORTABLE))
585 Perl_warner(aTHX_ WARN_PORTABLE,
586 "Bit vector size > 32 non-portable");
587 if (offset >= srclen)
589 else if (offset + 1 >= srclen)
591 (UV) s[offset ] << 56;
592 else if (offset + 2 >= srclen)
594 ((UV) s[offset ] << 56) +
595 ((UV) s[offset + 1] << 48);
596 else if (offset + 3 >= srclen)
598 ((UV) s[offset ] << 56) +
599 ((UV) s[offset + 1] << 48) +
600 ((UV) s[offset + 2] << 40);
601 else if (offset + 4 >= srclen)
603 ((UV) s[offset ] << 56) +
604 ((UV) s[offset + 1] << 48) +
605 ((UV) s[offset + 2] << 40) +
606 ((UV) s[offset + 3] << 32);
607 else if (offset + 5 >= srclen)
609 ((UV) s[offset ] << 56) +
610 ((UV) s[offset + 1] << 48) +
611 ((UV) s[offset + 2] << 40) +
612 ((UV) s[offset + 3] << 32) +
613 ( s[offset + 4] << 24);
614 else if (offset + 6 >= srclen)
616 ((UV) s[offset ] << 56) +
617 ((UV) s[offset + 1] << 48) +
618 ((UV) s[offset + 2] << 40) +
619 ((UV) s[offset + 3] << 32) +
620 ((UV) s[offset + 4] << 24) +
621 ((UV) s[offset + 5] << 16);
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);
636 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
638 offset >>= 3; /* turn into byte offset */
643 ((UV) s[offset] << 8) +
647 ((UV) s[offset ] << 24) +
648 ((UV) s[offset + 1] << 16) +
649 ( s[offset + 2] << 8) +
652 else if (size == 64) {
654 if (ckWARN(WARN_PORTABLE))
655 Perl_warner(aTHX_ WARN_PORTABLE,
656 "Bit vector size > 32 non-portable");
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) +
664 ( s[offset + 6] << 8) +
673 /* XXX SvUTF8 support missing! */
675 Perl_do_vecset(pTHX_ SV *sv)
677 SV *targ = LvTARG(sv);
680 register unsigned char *s;
688 s = (unsigned char*)SvPV_force(targ, targlen);
689 (void)SvPOK_only(targ);
691 offset = LvTARGOFF(sv);
693 Perl_croak(aTHX_ "Assigning to negative offset in vec");
694 size = LvTARGLEN(sv);
695 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
696 Perl_croak(aTHX_ "Illegal number of bits in vec");
698 offset *= size; /* turn into bit offset */
699 len = (offset + size + 7) / 8; /* required number of bytes */
701 s = (unsigned char*)SvGROW(targ, len + 1);
702 (void)memzero((char *)(s + targlen), len - targlen + 1);
703 SvCUR_set(targ, len);
707 mask = (1 << size) - 1;
710 offset >>= 3; /* turn into byte offset */
711 s[offset] &= ~(mask << size);
712 s[offset] |= lval << size;
715 offset >>= 3; /* turn into byte offset */
717 s[offset ] = lval & 0xff;
718 else if (size == 16) {
719 s[offset ] = (lval >> 8) & 0xff;
720 s[offset+1] = lval & 0xff;
722 else if (size == 32) {
723 s[offset ] = (lval >> 24) & 0xff;
724 s[offset+1] = (lval >> 16) & 0xff;
725 s[offset+2] = (lval >> 8) & 0xff;
726 s[offset+3] = lval & 0xff;
729 else if (size == 64) {
731 if (ckWARN(WARN_PORTABLE))
732 Perl_warner(aTHX_ WARN_PORTABLE,
733 "Bit vector size > 32 non-portable");
734 s[offset ] = (lval >> 56) & 0xff;
735 s[offset+1] = (lval >> 48) & 0xff;
736 s[offset+2] = (lval >> 40) & 0xff;
737 s[offset+3] = (lval >> 32) & 0xff;
738 s[offset+4] = (lval >> 24) & 0xff;
739 s[offset+5] = (lval >> 16) & 0xff;
740 s[offset+6] = (lval >> 8) & 0xff;
741 s[offset+7] = lval & 0xff;
749 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
755 if (SvTYPE(sv) == SVt_PVAV) {
760 for (i = 0; i <= max; i++) {
761 sv = (SV*)av_fetch(av, i, FALSE);
762 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
767 else if (SvTYPE(sv) == SVt_PVHV) {
770 (void)hv_iterinit(hv);
772 while ((entry = hv_iternext(hv)))
773 do_chop(astr,hv_iterval(hv,entry));
776 else if (SvREADONLY(sv))
777 Perl_croak(aTHX_ PL_no_modify);
779 if (len && !SvPOK(sv))
780 s = SvPV_force(sv, len);
783 char *send = s + len;
786 while ((*s & 0xc0) == 0x80)
788 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
789 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
790 sv_setpvn(astr, s, send - s);
792 SvCUR_set(sv, s - start);
797 sv_setpvn(astr, "", 0);
801 sv_setpvn(astr, s, 1);
808 sv_setpvn(astr, "", 0);
813 Perl_do_chomp(pTHX_ 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))
833 count += do_chomp(sv);
837 else if (SvTYPE(sv) == SVt_PVHV) {
840 (void)hv_iterinit(hv);
842 while ((entry = hv_iternext(hv)))
843 count += do_chomp(hv_iterval(hv,entry));
846 else if (SvREADONLY(sv))
847 Perl_croak(aTHX_ PL_no_modify);
849 if (len && !SvPOKp(sv))
850 s = SvPV_force(sv, len);
857 while (len && s[-1] == '\n') {
865 char *rsptr = SvPV(PL_rs, rslen);
876 if (memNE(s, rsptr, rslen))
891 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
893 dTHR; /* just for taint */
908 bool left_utf = DO_UTF8(left);
909 bool right_utf = DO_UTF8(right);
912 if (left_utf && !right_utf)
913 sv_utf8_upgrade(right);
914 if (!left_utf && right_utf)
915 sv_utf8_upgrade(left);
917 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
918 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
919 lsave = lc = SvPV(left, leftlen);
920 rsave = rc = SvPV(right, rightlen);
921 len = leftlen < rightlen ? leftlen : rightlen;
923 if ((left_utf || right_utf) && (sv == left || sv == right)) {
924 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
925 Newz(801, dc, needlen + 1, char);
927 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
929 dc = SvPV_force(sv, n_a);
930 if (SvCUR(sv) < len) {
931 dc = SvGROW(sv, len + 1);
932 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
934 if (optype != OP_BIT_AND && (left_utf || right_utf))
935 dc = SvGROW(sv, leftlen + rightlen + 1);
938 needlen = ((optype == OP_BIT_AND)
939 ? len : (leftlen > rightlen ? leftlen : rightlen));
940 Newz(801, dc, needlen + 1, char);
941 (void)sv_usepvn(sv, dc, needlen);
942 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
945 (void)SvPOK_only(sv);
946 if (left_utf || right_utf) {
949 STRLEN lulen = leftlen;
950 STRLEN rulen = rightlen;
955 while (lulen && rulen) {
956 luc = utf8_to_uv((U8*)lc, &ulen);
959 ruc = utf8_to_uv((U8*)rc, &ulen);
963 dc = (char*)uv_to_utf8((U8*)dc, duc);
965 if (sv == left || sv == right)
966 (void)sv_usepvn(sv, dcsave, needlen);
967 SvCUR_set(sv, dc - dcsave);
970 while (lulen && rulen) {
971 luc = utf8_to_uv((U8*)lc, &ulen);
974 ruc = utf8_to_uv((U8*)rc, &ulen);
978 dc = (char*)uv_to_utf8((U8*)dc, duc);
982 while (lulen && rulen) {
983 luc = utf8_to_uv((U8*)lc, &ulen);
986 ruc = utf8_to_uv((U8*)rc, &ulen);
990 dc = (char*)uv_to_utf8((U8*)dc, duc);
993 if (sv == left || sv == right)
994 (void)sv_usepvn(sv, dcsave, needlen);
995 SvCUR_set(sv, dc - dcsave);
997 sv_catpvn(sv, rc, rulen);
999 sv_catpvn(sv, lc, lulen);
1009 if (len >= sizeof(long)*4 &&
1010 !((long)dc % sizeof(long)) &&
1011 !((long)lc % sizeof(long)) &&
1012 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1014 I32 remainder = len % (sizeof(long)*4);
1015 len /= (sizeof(long)*4);
1024 *dl++ = *ll++ & *rl++;
1025 *dl++ = *ll++ & *rl++;
1026 *dl++ = *ll++ & *rl++;
1027 *dl++ = *ll++ & *rl++;
1032 *dl++ = *ll++ ^ *rl++;
1033 *dl++ = *ll++ ^ *rl++;
1034 *dl++ = *ll++ ^ *rl++;
1035 *dl++ = *ll++ ^ *rl++;
1040 *dl++ = *ll++ | *rl++;
1041 *dl++ = *ll++ | *rl++;
1042 *dl++ = *ll++ | *rl++;
1043 *dl++ = *ll++ | *rl++;
1058 *dc++ = *lc++ & *rc++;
1062 *dc++ = *lc++ ^ *rc++;
1066 *dc++ = *lc++ | *rc++;
1070 sv_catpvn(sv, rsave + len, rightlen - len);
1071 else if (leftlen > len)
1072 sv_catpvn(sv, lsave + len, leftlen - len);
1090 I32 gimme = GIMME_V;
1091 I32 dokeys = (PL_op->op_type == OP_KEYS);
1092 I32 dovalues = (PL_op->op_type == OP_VALUES);
1093 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1095 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1096 dokeys = dovalues = TRUE;
1099 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1100 dTARGET; /* make sure to clear its target here */
1101 if (SvTYPE(TARG) == SVt_PVLV)
1102 LvTARG(TARG) = Nullsv;
1108 keys = realhv ? hv : avhv_keys((AV*)hv);
1109 (void)hv_iterinit(keys); /* always reset iterator regardless */
1111 if (gimme == G_VOID)
1114 if (gimme == G_SCALAR) {
1118 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1119 if (SvTYPE(TARG) < SVt_PVLV) {
1120 sv_upgrade(TARG, SVt_PVLV);
1121 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1124 if (LvTARG(TARG) != (SV*)keys) {
1126 SvREFCNT_dec(LvTARG(TARG));
1127 LvTARG(TARG) = SvREFCNT_inc(keys);
1133 if (! SvTIED_mg((SV*)keys, 'P'))
1138 while (hv_iternext(keys)) i++;
1144 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1146 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1147 while ((entry = hv_iternext(keys))) {
1150 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1154 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1155 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1156 (unsigned long)HeHASH(entry),
1158 (unsigned long)(HeHASH(entry) & HvMAX(keys))));