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, send - s, &ulen, 0);
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, send - s, &ulen, 0);
131 if (c < 0x100 && tbl[c] >= 0)
141 S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
148 I32 hasutf = SvUTF8(sv);
154 tbl = (short*)cPVOP->op_pv;
156 Perl_croak(aTHX_ "panic: do_trans");
158 s = (U8*)SvPV(sv, len);
161 Newz(0, d, len*2+1, U8);
164 if (PL_op->op_private & OPpTRANS_SQUASH) {
168 if (hasutf && *s & 0x80)
171 if ((ch = tbl[*s]) >= 0) {
174 if (p != d - 1 || *p != *d)
177 else if (ch == -1) /* -1 is unmapped character */
178 *d++ = *s; /* -2 is delete character */
186 if (hasutf && *s & 0x80)
187 comp = utf8_to_uv_simple(s, NULL);
193 if (ch == -1) { /* -1 is unmapped character */
200 d = uv_to_utf8(d, ch);
206 s += hasutf && *s & 0x80 ? UNISKIP(*s) : 1;
213 sv_setpvn(sv, (const char*)dstart, d - dstart);
223 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
234 SV* rv = (SV*)cSVOP->op_sv;
235 HV* hv = (HV*)SvRV(rv);
236 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
237 UV none = svp ? SvUV(*svp) : 0x7fffffff;
245 s = (U8*)SvPV(sv, len);
249 svp = hv_fetch(hv, "FINAL", 5, FALSE);
253 /* d needs to be bigger than s, in case e.g. upgrading is required */
254 Newz(0, d, len*2+1, U8);
257 if ((uv = swash_fetch(rv, s)) < none) {
260 if ((uv & 0x80) && !isutf++)
261 HALF_UTF8_UPGRADE(dstart,d);
262 d = uv_to_utf8(d, uv);
264 else if (uv == none) {
267 if (i > 1 && !isutf++)
268 HALF_UTF8_UPGRADE(dstart,d);
272 else if (uv == extra) {
277 if (i > 1 && !isutf++)
278 HALF_UTF8_UPGRADE(dstart,d);
279 d = uv_to_utf8(d, final);
285 sv_setpvn(sv, (const char*)dstart, d - dstart);
294 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
302 SV* rv = (SV*)cSVOP->op_sv;
303 HV* hv = (HV*)SvRV(rv);
304 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
305 UV none = svp ? SvUV(*svp) : 0x7fffffff;
308 s = (U8*)SvPV(sv, len);
310 s = bytes_to_utf8(s, &len);
314 if ((uv = swash_fetch(rv, s)) < none)
323 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
330 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
331 I32 del = PL_op->op_private & OPpTRANS_DELETE;
332 SV* rv = (SV*)cSVOP->op_sv;
333 HV* hv = (HV*)SvRV(rv);
334 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
335 UV none = svp ? SvUV(*svp) : 0x7fffffff;
341 I32 isutf = SvUTF8(sv);
343 s = (U8*)SvPV(sv, len);
346 svp = hv_fetch(hv, "FINAL", 5, FALSE);
350 Newz(0, d, len*2+1, U8);
357 uv = swash_fetch(rv, s);
364 tmpbuf[0] = (( uv >> 6) | 0xc0);
365 tmpbuf[1] = (( uv & 0x3f) | 0x80);
367 uv = swash_fetch(rv, tmpbuf);
373 if ((uv & 0x80) && !isutf++)
374 HALF_UTF8_UPGRADE(dst,d);
375 d = uv_to_utf8(d, uv);
381 else if (uv == none) { /* "none" is unmapped character */
383 *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
388 else if (uv == extra && !del) {
391 d = uv_to_utf8(d, final);
397 matches++; /* "none+1" is delete character */
404 uv = swash_fetch(rv, s);
411 tmpbuf[0] = (( uv >> 6) | 0xc0);
412 tmpbuf[1] = (( uv & 0x3f) | 0x80);
414 uv = swash_fetch(rv, tmpbuf);
418 d = uv_to_utf8(d, uv);
422 else if (uv == none) { /* "none" is unmapped character */
424 *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
428 else if (uv == extra && !del) {
430 d = uv_to_utf8(d, final);
434 matches++; /* "none+1" is delete character */
439 sv_usepvn(sv, (char*)dst, d - dst);
442 SvCUR_set(sv, d - (U8*)SvPVX(sv));
450 Perl_do_trans(pTHX_ SV *sv)
454 I32 hasutf = (PL_op->op_private &
455 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
457 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
458 Perl_croak(aTHX_ PL_no_modify);
464 (void)SvPV_force(sv, len);
465 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
466 (void)SvPOK_only_UTF8(sv);
468 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
470 switch (PL_op->op_private & ~hasutf & 63) {
473 return do_trans_simple_utf8(sv);
475 return do_trans_simple(sv);
477 case OPpTRANS_IDENTICAL:
479 return do_trans_count_utf8(sv);
481 return do_trans_count(sv);
485 return do_trans_complex_utf8(sv);
487 return do_trans_complex(sv);
492 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
495 register I32 items = sp - mark;
498 register char *delim = SvPV(del, delimlen);
502 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
503 (void)SvUPGRADE(sv, SVt_PV);
504 if (SvLEN(sv) < len + items) { /* current length is way too short */
505 while (items-- > 0) {
506 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
512 SvGROW(sv, len + 1); /* so try to pre-extend */
531 for (; items > 0; items--,mark++) {
532 sv_catpvn(sv,delim,len);
537 for (; items > 0; items--,mark++)
544 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
547 char *pat = SvPV(*sarg, patlen);
548 bool do_taint = FALSE;
550 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
556 /* currently converts input to bytes if possible, but doesn't sweat failure */
558 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
561 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
566 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
567 Perl_croak(aTHX_ "Illegal number of bits in vec");
570 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
572 offset *= size; /* turn into bit offset */
573 len = (offset + size + 7) / 8; /* required number of bytes */
578 offset >>= 3; /* turn into byte offset */
580 if (offset >= srclen)
583 retnum = (UV) s[offset] << 8;
585 else if (size == 32) {
586 if (offset >= srclen)
588 else if (offset + 1 >= srclen)
590 ((UV) s[offset ] << 24);
591 else if (offset + 2 >= srclen)
593 ((UV) s[offset ] << 24) +
594 ((UV) s[offset + 1] << 16);
597 ((UV) s[offset ] << 24) +
598 ((UV) s[offset + 1] << 16) +
599 ( s[offset + 2] << 8);
602 else if (size == 64) {
604 if (ckWARN(WARN_PORTABLE))
605 Perl_warner(aTHX_ WARN_PORTABLE,
606 "Bit vector size > 32 non-portable");
607 if (offset >= srclen)
609 else if (offset + 1 >= srclen)
611 (UV) s[offset ] << 56;
612 else if (offset + 2 >= srclen)
614 ((UV) s[offset ] << 56) +
615 ((UV) s[offset + 1] << 48);
616 else if (offset + 3 >= srclen)
618 ((UV) s[offset ] << 56) +
619 ((UV) s[offset + 1] << 48) +
620 ((UV) s[offset + 2] << 40);
621 else if (offset + 4 >= srclen)
623 ((UV) s[offset ] << 56) +
624 ((UV) s[offset + 1] << 48) +
625 ((UV) s[offset + 2] << 40) +
626 ((UV) s[offset + 3] << 32);
627 else if (offset + 5 >= srclen)
629 ((UV) s[offset ] << 56) +
630 ((UV) s[offset + 1] << 48) +
631 ((UV) s[offset + 2] << 40) +
632 ((UV) s[offset + 3] << 32) +
633 ( s[offset + 4] << 24);
634 else if (offset + 6 >= srclen)
636 ((UV) s[offset ] << 56) +
637 ((UV) s[offset + 1] << 48) +
638 ((UV) s[offset + 2] << 40) +
639 ((UV) s[offset + 3] << 32) +
640 ((UV) s[offset + 4] << 24) +
641 ((UV) s[offset + 5] << 16);
644 ((UV) s[offset ] << 56) +
645 ((UV) s[offset + 1] << 48) +
646 ((UV) s[offset + 2] << 40) +
647 ((UV) s[offset + 3] << 32) +
648 ((UV) s[offset + 4] << 24) +
649 ((UV) s[offset + 5] << 16) +
650 ( s[offset + 6] << 8);
656 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
658 offset >>= 3; /* turn into byte offset */
663 ((UV) s[offset] << 8) +
667 ((UV) s[offset ] << 24) +
668 ((UV) s[offset + 1] << 16) +
669 ( s[offset + 2] << 8) +
672 else if (size == 64) {
674 if (ckWARN(WARN_PORTABLE))
675 Perl_warner(aTHX_ WARN_PORTABLE,
676 "Bit vector size > 32 non-portable");
678 ((UV) s[offset ] << 56) +
679 ((UV) s[offset + 1] << 48) +
680 ((UV) s[offset + 2] << 40) +
681 ((UV) s[offset + 3] << 32) +
682 ((UV) s[offset + 4] << 24) +
683 ((UV) s[offset + 5] << 16) +
684 ( s[offset + 6] << 8) +
693 /* currently converts input to bytes if possible but doesn't sweat failures,
694 * although it does ensure that the string it clobbers is not marked as
695 * utf8-valid any more
698 Perl_do_vecset(pTHX_ SV *sv)
700 SV *targ = LvTARG(sv);
703 register unsigned char *s;
711 s = (unsigned char*)SvPV_force(targ, targlen);
713 /* This is handled by the SvPOK_only below...
714 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
717 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
720 (void)SvPOK_only(targ);
722 offset = LvTARGOFF(sv);
724 Perl_croak(aTHX_ "Assigning to negative offset in vec");
725 size = LvTARGLEN(sv);
726 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
727 Perl_croak(aTHX_ "Illegal number of bits in vec");
729 offset *= size; /* turn into bit offset */
730 len = (offset + size + 7) / 8; /* required number of bytes */
732 s = (unsigned char*)SvGROW(targ, len + 1);
733 (void)memzero((char *)(s + targlen), len - targlen + 1);
734 SvCUR_set(targ, len);
738 mask = (1 << size) - 1;
741 offset >>= 3; /* turn into byte offset */
742 s[offset] &= ~(mask << size);
743 s[offset] |= lval << size;
746 offset >>= 3; /* turn into byte offset */
748 s[offset ] = lval & 0xff;
749 else if (size == 16) {
750 s[offset ] = (lval >> 8) & 0xff;
751 s[offset+1] = lval & 0xff;
753 else if (size == 32) {
754 s[offset ] = (lval >> 24) & 0xff;
755 s[offset+1] = (lval >> 16) & 0xff;
756 s[offset+2] = (lval >> 8) & 0xff;
757 s[offset+3] = lval & 0xff;
760 else if (size == 64) {
762 if (ckWARN(WARN_PORTABLE))
763 Perl_warner(aTHX_ WARN_PORTABLE,
764 "Bit vector size > 32 non-portable");
765 s[offset ] = (lval >> 56) & 0xff;
766 s[offset+1] = (lval >> 48) & 0xff;
767 s[offset+2] = (lval >> 40) & 0xff;
768 s[offset+3] = (lval >> 32) & 0xff;
769 s[offset+4] = (lval >> 24) & 0xff;
770 s[offset+5] = (lval >> 16) & 0xff;
771 s[offset+6] = (lval >> 8) & 0xff;
772 s[offset+7] = lval & 0xff;
780 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
786 if (SvTYPE(sv) == SVt_PVAV) {
791 for (i = 0; i <= max; i++) {
792 sv = (SV*)av_fetch(av, i, FALSE);
793 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
798 else if (SvTYPE(sv) == SVt_PVHV) {
801 (void)hv_iterinit(hv);
803 while ((entry = hv_iternext(hv)))
804 do_chop(astr,hv_iterval(hv,entry));
807 else if (SvREADONLY(sv))
808 Perl_croak(aTHX_ PL_no_modify);
810 if (len && !SvPOK(sv))
811 s = SvPV_force(sv, len);
814 char *send = s + len;
817 while ((*s & 0xc0) == 0x80)
819 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
820 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
821 sv_setpvn(astr, s, send - s);
823 SvCUR_set(sv, s - start);
828 sv_setpvn(astr, "", 0);
832 sv_setpvn(astr, s, 1);
839 sv_setpvn(astr, "", 0);
844 Perl_do_chomp(pTHX_ register SV *sv)
856 if (SvTYPE(sv) == SVt_PVAV) {
861 for (i = 0; i <= max; i++) {
862 sv = (SV*)av_fetch(av, i, FALSE);
863 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
864 count += do_chomp(sv);
868 else if (SvTYPE(sv) == SVt_PVHV) {
871 (void)hv_iterinit(hv);
873 while ((entry = hv_iternext(hv)))
874 count += do_chomp(hv_iterval(hv,entry));
877 else if (SvREADONLY(sv))
878 Perl_croak(aTHX_ PL_no_modify);
880 if (len && !SvPOKp(sv))
881 s = SvPV_force(sv, len);
888 while (len && s[-1] == '\n') {
896 char *rsptr = SvPV(PL_rs, rslen);
907 if (memNE(s, rsptr, rslen))
922 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
924 dTHR; /* just for taint */
939 bool left_utf = DO_UTF8(left);
940 bool right_utf = DO_UTF8(right);
943 if (left_utf && !right_utf)
944 sv_utf8_upgrade(right);
945 else if (!left_utf && right_utf)
946 sv_utf8_upgrade(left);
948 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
949 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
950 lsave = lc = SvPV(left, leftlen);
951 rsave = rc = SvPV(right, rightlen);
952 len = leftlen < rightlen ? leftlen : rightlen;
954 if ((left_utf || right_utf) && (sv == left || sv == right)) {
955 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
956 Newz(801, dc, needlen + 1, char);
958 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
960 dc = SvPV_force(sv, n_a);
961 if (SvCUR(sv) < len) {
962 dc = SvGROW(sv, len + 1);
963 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
965 if (optype != OP_BIT_AND && (left_utf || right_utf))
966 dc = SvGROW(sv, leftlen + rightlen + 1);
969 needlen = ((optype == OP_BIT_AND)
970 ? len : (leftlen > rightlen ? leftlen : rightlen));
971 Newz(801, dc, needlen + 1, char);
972 (void)sv_usepvn(sv, dc, needlen);
973 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
976 (void)SvPOK_only(sv);
977 if (left_utf || right_utf) {
980 STRLEN lulen = leftlen;
981 STRLEN rulen = rightlen;
986 while (lulen && rulen) {
987 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
990 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
994 dc = (char*)uv_to_utf8((U8*)dc, duc);
996 if (sv == left || sv == right)
997 (void)sv_usepvn(sv, dcsave, needlen);
998 SvCUR_set(sv, dc - dcsave);
1001 while (lulen && rulen) {
1002 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1005 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1009 dc = (char*)uv_to_utf8((U8*)dc, duc);
1013 while (lulen && rulen) {
1014 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1017 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1021 dc = (char*)uv_to_utf8((U8*)dc, duc);
1024 if (sv == left || sv == right)
1025 (void)sv_usepvn(sv, dcsave, needlen);
1026 SvCUR_set(sv, dc - dcsave);
1028 sv_catpvn(sv, rc, rulen);
1030 sv_catpvn(sv, lc, lulen);
1040 if (len >= sizeof(long)*4 &&
1041 !((long)dc % sizeof(long)) &&
1042 !((long)lc % sizeof(long)) &&
1043 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1045 I32 remainder = len % (sizeof(long)*4);
1046 len /= (sizeof(long)*4);
1055 *dl++ = *ll++ & *rl++;
1056 *dl++ = *ll++ & *rl++;
1057 *dl++ = *ll++ & *rl++;
1058 *dl++ = *ll++ & *rl++;
1063 *dl++ = *ll++ ^ *rl++;
1064 *dl++ = *ll++ ^ *rl++;
1065 *dl++ = *ll++ ^ *rl++;
1066 *dl++ = *ll++ ^ *rl++;
1071 *dl++ = *ll++ | *rl++;
1072 *dl++ = *ll++ | *rl++;
1073 *dl++ = *ll++ | *rl++;
1074 *dl++ = *ll++ | *rl++;
1089 *dc++ = *lc++ & *rc++;
1093 *dc++ = *lc++ ^ *rc++;
1097 *dc++ = *lc++ | *rc++;
1101 sv_catpvn(sv, rsave + len, rightlen - len);
1102 else if (leftlen > len)
1103 sv_catpvn(sv, lsave + len, leftlen - len);
1121 I32 gimme = GIMME_V;
1122 I32 dokeys = (PL_op->op_type == OP_KEYS);
1123 I32 dovalues = (PL_op->op_type == OP_VALUES);
1124 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1126 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1127 dokeys = dovalues = TRUE;
1130 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1131 dTARGET; /* make sure to clear its target here */
1132 if (SvTYPE(TARG) == SVt_PVLV)
1133 LvTARG(TARG) = Nullsv;
1139 keys = realhv ? hv : avhv_keys((AV*)hv);
1140 (void)hv_iterinit(keys); /* always reset iterator regardless */
1142 if (gimme == G_VOID)
1145 if (gimme == G_SCALAR) {
1149 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1150 if (SvTYPE(TARG) < SVt_PVLV) {
1151 sv_upgrade(TARG, SVt_PVLV);
1152 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1155 if (LvTARG(TARG) != (SV*)keys) {
1157 SvREFCNT_dec(LvTARG(TARG));
1158 LvTARG(TARG) = SvREFCNT_inc(keys);
1164 if (! SvTIED_mg((SV*)keys, 'P'))
1169 while (hv_iternext(keys)) i++;
1175 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1177 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1178 while ((entry = hv_iternext(keys))) {
1181 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1185 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1186 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1187 (unsigned long)HeHASH(entry),
1189 (unsigned long)(HeHASH(entry) & HvMAX(keys))));