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 /* currently converts input to bytes if needed and croaks if a character
541 > 255 is encountered */
543 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
546 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
551 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
552 Perl_croak(aTHX_ "Illegal number of bits in vec");
555 if (Perl_utf8_to_bytes(aTHX_ (U8*) s, &srclen)) {
557 SvCUR_set(sv, srclen);
560 Perl_croak(aTHX_ "Character > 255 in vec()");
563 offset *= size; /* turn into bit offset */
564 len = (offset + size + 7) / 8; /* required number of bytes */
569 offset >>= 3; /* turn into byte offset */
571 if (offset >= srclen)
574 retnum = (UV) s[offset] << 8;
576 else if (size == 32) {
577 if (offset >= srclen)
579 else if (offset + 1 >= srclen)
581 ((UV) s[offset ] << 24);
582 else if (offset + 2 >= srclen)
584 ((UV) s[offset ] << 24) +
585 ((UV) s[offset + 1] << 16);
588 ((UV) s[offset ] << 24) +
589 ((UV) s[offset + 1] << 16) +
590 ( s[offset + 2] << 8);
593 else if (size == 64) {
595 if (ckWARN(WARN_PORTABLE))
596 Perl_warner(aTHX_ WARN_PORTABLE,
597 "Bit vector size > 32 non-portable");
598 if (offset >= srclen)
600 else if (offset + 1 >= srclen)
602 (UV) s[offset ] << 56;
603 else if (offset + 2 >= srclen)
605 ((UV) s[offset ] << 56) +
606 ((UV) s[offset + 1] << 48);
607 else if (offset + 3 >= srclen)
609 ((UV) s[offset ] << 56) +
610 ((UV) s[offset + 1] << 48) +
611 ((UV) s[offset + 2] << 40);
612 else if (offset + 4 >= srclen)
614 ((UV) s[offset ] << 56) +
615 ((UV) s[offset + 1] << 48) +
616 ((UV) s[offset + 2] << 40) +
617 ((UV) s[offset + 3] << 32);
618 else if (offset + 5 >= srclen)
620 ((UV) s[offset ] << 56) +
621 ((UV) s[offset + 1] << 48) +
622 ((UV) s[offset + 2] << 40) +
623 ((UV) s[offset + 3] << 32) +
624 ( s[offset + 4] << 24);
625 else if (offset + 6 >= srclen)
627 ((UV) s[offset ] << 56) +
628 ((UV) s[offset + 1] << 48) +
629 ((UV) s[offset + 2] << 40) +
630 ((UV) s[offset + 3] << 32) +
631 ((UV) s[offset + 4] << 24) +
632 ((UV) s[offset + 5] << 16);
635 ((UV) s[offset ] << 56) +
636 ((UV) s[offset + 1] << 48) +
637 ((UV) s[offset + 2] << 40) +
638 ((UV) s[offset + 3] << 32) +
639 ((UV) s[offset + 4] << 24) +
640 ((UV) s[offset + 5] << 16) +
641 ( s[offset + 6] << 8);
647 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
649 offset >>= 3; /* turn into byte offset */
654 ((UV) s[offset] << 8) +
658 ((UV) s[offset ] << 24) +
659 ((UV) s[offset + 1] << 16) +
660 ( s[offset + 2] << 8) +
663 else if (size == 64) {
665 if (ckWARN(WARN_PORTABLE))
666 Perl_warner(aTHX_ WARN_PORTABLE,
667 "Bit vector size > 32 non-portable");
669 ((UV) s[offset ] << 56) +
670 ((UV) s[offset + 1] << 48) +
671 ((UV) s[offset + 2] << 40) +
672 ((UV) s[offset + 3] << 32) +
673 ((UV) s[offset + 4] << 24) +
674 ((UV) s[offset + 5] << 16) +
675 ( s[offset + 6] << 8) +
684 /* currently converts input to bytes if needed and croaks if a character
685 > 255 is encountered */
687 Perl_do_vecset(pTHX_ SV *sv)
689 SV *targ = LvTARG(sv);
692 register unsigned char *s;
700 s = (unsigned char*)SvPV_force(targ, targlen);
702 if (Perl_utf8_to_bytes(aTHX_ (U8*) s, &targlen)) {
703 /* SvUTF8_off(targ); SvPOK_only below ensures this */
704 SvCUR_set(targ, targlen);
707 Perl_croak(aTHX_ "Character > 255 in vec()");
710 (void)SvPOK_only(targ);
712 offset = LvTARGOFF(sv);
714 Perl_croak(aTHX_ "Assigning to negative offset in vec");
715 size = LvTARGLEN(sv);
716 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
717 Perl_croak(aTHX_ "Illegal number of bits in vec");
719 offset *= size; /* turn into bit offset */
720 len = (offset + size + 7) / 8; /* required number of bytes */
722 s = (unsigned char*)SvGROW(targ, len + 1);
723 (void)memzero((char *)(s + targlen), len - targlen + 1);
724 SvCUR_set(targ, len);
728 mask = (1 << size) - 1;
731 offset >>= 3; /* turn into byte offset */
732 s[offset] &= ~(mask << size);
733 s[offset] |= lval << size;
736 offset >>= 3; /* turn into byte offset */
738 s[offset ] = lval & 0xff;
739 else if (size == 16) {
740 s[offset ] = (lval >> 8) & 0xff;
741 s[offset+1] = lval & 0xff;
743 else if (size == 32) {
744 s[offset ] = (lval >> 24) & 0xff;
745 s[offset+1] = (lval >> 16) & 0xff;
746 s[offset+2] = (lval >> 8) & 0xff;
747 s[offset+3] = lval & 0xff;
750 else if (size == 64) {
752 if (ckWARN(WARN_PORTABLE))
753 Perl_warner(aTHX_ WARN_PORTABLE,
754 "Bit vector size > 32 non-portable");
755 s[offset ] = (lval >> 56) & 0xff;
756 s[offset+1] = (lval >> 48) & 0xff;
757 s[offset+2] = (lval >> 40) & 0xff;
758 s[offset+3] = (lval >> 32) & 0xff;
759 s[offset+4] = (lval >> 24) & 0xff;
760 s[offset+5] = (lval >> 16) & 0xff;
761 s[offset+6] = (lval >> 8) & 0xff;
762 s[offset+7] = lval & 0xff;
770 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
776 if (SvTYPE(sv) == SVt_PVAV) {
781 for (i = 0; i <= max; i++) {
782 sv = (SV*)av_fetch(av, i, FALSE);
783 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
788 else if (SvTYPE(sv) == SVt_PVHV) {
791 (void)hv_iterinit(hv);
793 while ((entry = hv_iternext(hv)))
794 do_chop(astr,hv_iterval(hv,entry));
797 else if (SvREADONLY(sv))
798 Perl_croak(aTHX_ PL_no_modify);
800 if (len && !SvPOK(sv))
801 s = SvPV_force(sv, len);
804 char *send = s + len;
807 while ((*s & 0xc0) == 0x80)
809 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
810 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
811 sv_setpvn(astr, s, send - s);
813 SvCUR_set(sv, s - start);
818 sv_setpvn(astr, "", 0);
822 sv_setpvn(astr, s, 1);
829 sv_setpvn(astr, "", 0);
834 Perl_do_chomp(pTHX_ register SV *sv)
846 if (SvTYPE(sv) == SVt_PVAV) {
851 for (i = 0; i <= max; i++) {
852 sv = (SV*)av_fetch(av, i, FALSE);
853 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
854 count += do_chomp(sv);
858 else if (SvTYPE(sv) == SVt_PVHV) {
861 (void)hv_iterinit(hv);
863 while ((entry = hv_iternext(hv)))
864 count += do_chomp(hv_iterval(hv,entry));
867 else if (SvREADONLY(sv))
868 Perl_croak(aTHX_ PL_no_modify);
870 if (len && !SvPOKp(sv))
871 s = SvPV_force(sv, len);
878 while (len && s[-1] == '\n') {
886 char *rsptr = SvPV(PL_rs, rslen);
897 if (memNE(s, rsptr, rslen))
912 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
914 dTHR; /* just for taint */
929 bool left_utf = DO_UTF8(left);
930 bool right_utf = DO_UTF8(right);
933 if (left_utf && !right_utf)
934 sv_utf8_upgrade(right);
935 if (!left_utf && right_utf)
936 sv_utf8_upgrade(left);
938 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
939 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
940 lsave = lc = SvPV(left, leftlen);
941 rsave = rc = SvPV(right, rightlen);
942 len = leftlen < rightlen ? leftlen : rightlen;
944 if ((left_utf || right_utf) && (sv == left || sv == right)) {
945 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
946 Newz(801, dc, needlen + 1, char);
948 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
950 dc = SvPV_force(sv, n_a);
951 if (SvCUR(sv) < len) {
952 dc = SvGROW(sv, len + 1);
953 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
955 if (optype != OP_BIT_AND && (left_utf || right_utf))
956 dc = SvGROW(sv, leftlen + rightlen + 1);
959 needlen = ((optype == OP_BIT_AND)
960 ? len : (leftlen > rightlen ? leftlen : rightlen));
961 Newz(801, dc, needlen + 1, char);
962 (void)sv_usepvn(sv, dc, needlen);
963 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
966 (void)SvPOK_only(sv);
967 if (left_utf || right_utf) {
970 STRLEN lulen = leftlen;
971 STRLEN rulen = rightlen;
976 while (lulen && rulen) {
977 luc = utf8_to_uv((U8*)lc, &ulen);
980 ruc = utf8_to_uv((U8*)rc, &ulen);
984 dc = (char*)uv_to_utf8((U8*)dc, duc);
986 if (sv == left || sv == right)
987 (void)sv_usepvn(sv, dcsave, needlen);
988 SvCUR_set(sv, dc - dcsave);
991 while (lulen && rulen) {
992 luc = utf8_to_uv((U8*)lc, &ulen);
995 ruc = utf8_to_uv((U8*)rc, &ulen);
999 dc = (char*)uv_to_utf8((U8*)dc, duc);
1003 while (lulen && rulen) {
1004 luc = utf8_to_uv((U8*)lc, &ulen);
1007 ruc = utf8_to_uv((U8*)rc, &ulen);
1011 dc = (char*)uv_to_utf8((U8*)dc, duc);
1014 if (sv == left || sv == right)
1015 (void)sv_usepvn(sv, dcsave, needlen);
1016 SvCUR_set(sv, dc - dcsave);
1018 sv_catpvn(sv, rc, rulen);
1020 sv_catpvn(sv, lc, lulen);
1030 if (len >= sizeof(long)*4 &&
1031 !((long)dc % sizeof(long)) &&
1032 !((long)lc % sizeof(long)) &&
1033 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1035 I32 remainder = len % (sizeof(long)*4);
1036 len /= (sizeof(long)*4);
1045 *dl++ = *ll++ & *rl++;
1046 *dl++ = *ll++ & *rl++;
1047 *dl++ = *ll++ & *rl++;
1048 *dl++ = *ll++ & *rl++;
1053 *dl++ = *ll++ ^ *rl++;
1054 *dl++ = *ll++ ^ *rl++;
1055 *dl++ = *ll++ ^ *rl++;
1056 *dl++ = *ll++ ^ *rl++;
1061 *dl++ = *ll++ | *rl++;
1062 *dl++ = *ll++ | *rl++;
1063 *dl++ = *ll++ | *rl++;
1064 *dl++ = *ll++ | *rl++;
1079 *dc++ = *lc++ & *rc++;
1083 *dc++ = *lc++ ^ *rc++;
1087 *dc++ = *lc++ | *rc++;
1091 sv_catpvn(sv, rsave + len, rightlen - len);
1092 else if (leftlen > len)
1093 sv_catpvn(sv, lsave + len, leftlen - len);
1111 I32 gimme = GIMME_V;
1112 I32 dokeys = (PL_op->op_type == OP_KEYS);
1113 I32 dovalues = (PL_op->op_type == OP_VALUES);
1114 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1116 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1117 dokeys = dovalues = TRUE;
1120 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1121 dTARGET; /* make sure to clear its target here */
1122 if (SvTYPE(TARG) == SVt_PVLV)
1123 LvTARG(TARG) = Nullsv;
1129 keys = realhv ? hv : avhv_keys((AV*)hv);
1130 (void)hv_iterinit(keys); /* always reset iterator regardless */
1132 if (gimme == G_VOID)
1135 if (gimme == G_SCALAR) {
1139 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1140 if (SvTYPE(TARG) < SVt_PVLV) {
1141 sv_upgrade(TARG, SVt_PVLV);
1142 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1145 if (LvTARG(TARG) != (SV*)keys) {
1147 SvREFCNT_dec(LvTARG(TARG));
1148 LvTARG(TARG) = SvREFCNT_inc(keys);
1154 if (! SvTIED_mg((SV*)keys, 'P'))
1159 while (hv_iternext(keys)) i++;
1165 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1167 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1168 while ((entry = hv_iternext(keys))) {
1171 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1175 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1176 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1177 (unsigned long)HeHASH(entry),
1179 (unsigned long)(HeHASH(entry) & HvMAX(keys))));