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_simple(pTHX_ SV *sv) /* SPC - OK */
31 I32 hasutf = SvUTF8(sv);
36 tbl = (short*)cPVOP->op_pv;
38 Perl_croak(aTHX_ "panic: do_trans");
40 s = (U8*)SvPV(sv, len);
44 if (hasutf && *s & 0x80)
45 s+=UTF8SKIP(s); /* Given that we're here because tbl is !UTF8...*/
47 if ((ch = tbl[*s]) >= 0) {
60 S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
66 I32 hasutf = SvUTF8(sv);
70 tbl = (short*)cPVOP->op_pv;
72 Perl_croak(aTHX_ "panic: do_trans");
74 s = (U8*)SvPV(sv, len);
78 if (hasutf && *s & 0x80)
91 S_do_trans_complex(pTHX_ SV *sv)/* SPC - OK */
97 I32 hasutf = SvUTF8(sv);
103 tbl = (short*)cPVOP->op_pv;
105 Perl_croak(aTHX_ "panic: do_trans");
107 s = (U8*)SvPV(sv, len);
111 if (PL_op->op_private & OPpTRANS_SQUASH) {
115 if (hasutf && *s & 0x80)
118 if ((ch = tbl[*s]) >= 0) {
121 if (p == d - 1 && *p == *d)
126 else if (ch == -1) /* -1 is unmapped character */
127 *d++ = *s; /* -2 is delete character */
134 if (hasutf && *s & 0x80)
137 if ((ch = tbl[*s]) >= 0) {
142 else if (ch == -1) /* -1 is unmapped character */
143 *d++ = *s; /* -2 is delete character */
148 matches += send - d; /* account for disappeared chars */
150 SvCUR_set(sv, d - (U8*)SvPVX(sv));
157 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
168 SV* rv = (SV*)cSVOP->op_sv;
169 HV* hv = (HV*)SvRV(rv);
170 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
171 UV none = svp ? SvUV(*svp) : 0x7fffffff;
179 s = (U8*)SvPV(sv, len);
183 svp = hv_fetch(hv, "FINAL", 5, FALSE);
187 /* d needs to be bigger than s, in case e.g. upgrading is required */
188 Newz(0, d, len*2+1, U8);
191 if ((uv = swash_fetch(rv, s)) < none) {
194 if (uv & 0x80 && !isutf) {
195 /* Sneaky-upgrade dstart...d */
199 new = bytes_to_utf8(dstart, &len);
200 Copy(new,dstart,len,U8*);
204 d = uv_to_utf8(d, uv);
206 else if (uv == none) {
209 if (i > 1 && !isutf) {
213 new = bytes_to_utf8(dstart, &len);
214 Copy(new,dstart,len,U8*);
221 else if (uv == extra) {
226 if (i > 1 && !isutf) {
230 new = bytes_to_utf8(dstart, &len);
231 Copy(new,dstart,len,U8*);
235 d = uv_to_utf8(d, final);
241 SvPV_set(sv, dstart);
242 SvCUR_set(sv, d - dstart);
251 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
259 SV* rv = (SV*)cSVOP->op_sv;
260 HV* hv = (HV*)SvRV(rv);
261 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
262 UV none = svp ? SvUV(*svp) : 0x7fffffff;
265 s = (U8*)SvPV(sv, len);
267 s = bytes_to_utf8(s, &len);
271 if ((uv = swash_fetch(rv, s)) < none)
280 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
287 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
288 I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
289 I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
290 I32 del = PL_op->op_private & OPpTRANS_DELETE;
291 SV* rv = (SV*)cSVOP->op_sv;
292 HV* hv = (HV*)SvRV(rv);
293 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
294 UV none = svp ? SvUV(*svp) : 0x7fffffff;
301 s = (U8*)SvPV(sv, len);
304 svp = hv_fetch(hv, "FINAL", 5, FALSE);
308 if (PL_op->op_private & OPpTRANS_GROWS) {
311 svp = hv_fetch(hv, "BITS", 4, FALSE);
313 bits = (I32)SvIV(*svp);
315 Newz(801, d, len * (bits >> 3) + 1, U8);
327 uv = swash_fetch(rv, s);
335 tmpbuf[0] = (( uv >> 6) | 0xc0);
336 tmpbuf[1] = (( uv & 0x3f) | 0x80);
338 uv = swash_fetch(rv, tmpbuf);
343 if (uv >= 0x80 && to_utf)
344 d = uv_to_utf8(d, uv);
353 else if (uv == none) { /* "none" is unmapped character */
359 for (i = UTF8SKIP(s); i; --i)
364 *d++ = (U8)utf8_to_uv(s, &ulen);
368 else { /* must be to_utf only */
369 d = uv_to_utf8(d, s[-1]);
374 else if (uv == extra && !del) {
377 if (final >= 0x80 && to_utf)
378 d = uv_to_utf8(d, final);
387 matches++; /* "none+1" is delete character */
395 uv = swash_fetch(rv, s);
403 tmpbuf[0] = (( uv >> 6) | 0xc0);
404 tmpbuf[1] = (( uv & 0x3f) | 0x80);
406 uv = swash_fetch(rv, tmpbuf);
410 if (uv >= 0x80 && to_utf)
411 d = uv_to_utf8(d, uv);
418 else if (uv == none) { /* "none" is unmapped character */
424 for (i = UTF8SKIP(s); i; --i)
429 *d++ = (U8)utf8_to_uv(s, &ulen);
433 else { /* must be to_utf only */
434 d = uv_to_utf8(d, s[-1]);
438 else if (uv == extra && !del) {
440 if (final >= 0x80 && to_utf)
441 d = uv_to_utf8(d, final);
448 matches++; /* "none+1" is delete character */
454 sv_usepvn(sv, (char*)dst, d - dst);
457 SvCUR_set(sv, d - (U8*)SvPVX(sv));
465 Perl_do_trans(pTHX_ SV *sv)
469 I32 hasutf = (PL_op->op_private &
470 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
472 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
473 Perl_croak(aTHX_ PL_no_modify);
479 (void)SvPV_force(sv, len);
480 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
481 (void)SvPOK_only_UTF8(sv);
483 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
485 switch (PL_op->op_private & ~hasutf & 63) {
488 return do_trans_simple_utf8(sv);
490 return do_trans_simple(sv);
492 case OPpTRANS_IDENTICAL:
494 return do_trans_count_utf8(sv);
496 return do_trans_count(sv);
500 return do_trans_complex_utf8(sv);
502 return do_trans_complex(sv);
507 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
510 register I32 items = sp - mark;
513 register char *delim = SvPV(del, delimlen);
517 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
518 (void)SvUPGRADE(sv, SVt_PV);
519 if (SvLEN(sv) < len + items) { /* current length is way too short */
520 while (items-- > 0) {
521 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
527 SvGROW(sv, len + 1); /* so try to pre-extend */
546 for (; items > 0; items--,mark++) {
547 sv_catpvn(sv,delim,len);
552 for (; items > 0; items--,mark++)
559 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
562 char *pat = SvPV(*sarg, patlen);
563 bool do_taint = FALSE;
565 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
571 /* XXX SvUTF8 support missing! */
573 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
576 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
581 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
582 Perl_croak(aTHX_ "Illegal number of bits in vec");
583 offset *= size; /* turn into bit offset */
584 len = (offset + size + 7) / 8; /* required number of bytes */
589 offset >>= 3; /* turn into byte offset */
591 if (offset >= srclen)
594 retnum = (UV) s[offset] << 8;
596 else if (size == 32) {
597 if (offset >= srclen)
599 else if (offset + 1 >= srclen)
601 ((UV) s[offset ] << 24);
602 else if (offset + 2 >= srclen)
604 ((UV) s[offset ] << 24) +
605 ((UV) s[offset + 1] << 16);
608 ((UV) s[offset ] << 24) +
609 ((UV) s[offset + 1] << 16) +
610 ( s[offset + 2] << 8);
613 else if (size == 64) {
615 if (ckWARN(WARN_PORTABLE))
616 Perl_warner(aTHX_ WARN_PORTABLE,
617 "Bit vector size > 32 non-portable");
618 if (offset >= srclen)
620 else if (offset + 1 >= srclen)
622 (UV) s[offset ] << 56;
623 else if (offset + 2 >= srclen)
625 ((UV) s[offset ] << 56) +
626 ((UV) s[offset + 1] << 48);
627 else if (offset + 3 >= srclen)
629 ((UV) s[offset ] << 56) +
630 ((UV) s[offset + 1] << 48) +
631 ((UV) s[offset + 2] << 40);
632 else if (offset + 4 >= srclen)
634 ((UV) s[offset ] << 56) +
635 ((UV) s[offset + 1] << 48) +
636 ((UV) s[offset + 2] << 40) +
637 ((UV) s[offset + 3] << 32);
638 else if (offset + 5 >= srclen)
640 ((UV) s[offset ] << 56) +
641 ((UV) s[offset + 1] << 48) +
642 ((UV) s[offset + 2] << 40) +
643 ((UV) s[offset + 3] << 32) +
644 ( s[offset + 4] << 24);
645 else if (offset + 6 >= srclen)
647 ((UV) s[offset ] << 56) +
648 ((UV) s[offset + 1] << 48) +
649 ((UV) s[offset + 2] << 40) +
650 ((UV) s[offset + 3] << 32) +
651 ((UV) s[offset + 4] << 24) +
652 ((UV) s[offset + 5] << 16);
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) +
661 ( s[offset + 6] << 8);
667 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
669 offset >>= 3; /* turn into byte offset */
674 ((UV) s[offset] << 8) +
678 ((UV) s[offset ] << 24) +
679 ((UV) s[offset + 1] << 16) +
680 ( s[offset + 2] << 8) +
683 else if (size == 64) {
685 if (ckWARN(WARN_PORTABLE))
686 Perl_warner(aTHX_ WARN_PORTABLE,
687 "Bit vector size > 32 non-portable");
689 ((UV) s[offset ] << 56) +
690 ((UV) s[offset + 1] << 48) +
691 ((UV) s[offset + 2] << 40) +
692 ((UV) s[offset + 3] << 32) +
693 ((UV) s[offset + 4] << 24) +
694 ((UV) s[offset + 5] << 16) +
695 ( s[offset + 6] << 8) +
704 /* XXX SvUTF8 support missing! */
706 Perl_do_vecset(pTHX_ SV *sv)
708 SV *targ = LvTARG(sv);
711 register unsigned char *s;
719 s = (unsigned char*)SvPV_force(targ, targlen);
720 (void)SvPOK_only(targ);
722 offset = LvTARGOFF(sv);
723 size = LvTARGLEN(sv);
724 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
725 Perl_croak(aTHX_ "Illegal number of bits in vec");
727 offset *= size; /* turn into bit offset */
728 len = (offset + size + 7) / 8; /* required number of bytes */
730 s = (unsigned char*)SvGROW(targ, len + 1);
731 (void)memzero((char *)(s + targlen), len - targlen + 1);
732 SvCUR_set(targ, len);
736 mask = (1 << size) - 1;
739 offset >>= 3; /* turn into byte offset */
740 s[offset] &= ~(mask << size);
741 s[offset] |= lval << size;
744 offset >>= 3; /* turn into byte offset */
746 s[offset ] = lval & 0xff;
747 else if (size == 16) {
748 s[offset ] = (lval >> 8) & 0xff;
749 s[offset+1] = lval & 0xff;
751 else if (size == 32) {
752 s[offset ] = (lval >> 24) & 0xff;
753 s[offset+1] = (lval >> 16) & 0xff;
754 s[offset+2] = (lval >> 8) & 0xff;
755 s[offset+3] = lval & 0xff;
758 else if (size == 64) {
760 if (ckWARN(WARN_PORTABLE))
761 Perl_warner(aTHX_ WARN_PORTABLE,
762 "Bit vector size > 32 non-portable");
763 s[offset ] = (lval >> 56) & 0xff;
764 s[offset+1] = (lval >> 48) & 0xff;
765 s[offset+2] = (lval >> 40) & 0xff;
766 s[offset+3] = (lval >> 32) & 0xff;
767 s[offset+4] = (lval >> 24) & 0xff;
768 s[offset+5] = (lval >> 16) & 0xff;
769 s[offset+6] = (lval >> 8) & 0xff;
770 s[offset+7] = lval & 0xff;
778 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
784 if (SvTYPE(sv) == SVt_PVAV) {
789 for (i = 0; i <= max; i++) {
790 sv = (SV*)av_fetch(av, i, FALSE);
791 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
796 else if (SvTYPE(sv) == SVt_PVHV) {
799 (void)hv_iterinit(hv);
801 while ((entry = hv_iternext(hv)))
802 do_chop(astr,hv_iterval(hv,entry));
805 else if (SvREADONLY(sv))
806 Perl_croak(aTHX_ PL_no_modify);
808 if (len && !SvPOK(sv))
809 s = SvPV_force(sv, len);
812 char *send = s + len;
815 while ((*s & 0xc0) == 0x80)
817 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
818 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
819 sv_setpvn(astr, s, send - s);
821 SvCUR_set(sv, s - start);
826 sv_setpvn(astr, "", 0);
830 sv_setpvn(astr, s, 1);
837 sv_setpvn(astr, "", 0);
842 Perl_do_chomp(pTHX_ register SV *sv)
854 if (SvTYPE(sv) == SVt_PVAV) {
859 for (i = 0; i <= max; i++) {
860 sv = (SV*)av_fetch(av, i, FALSE);
861 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
862 count += do_chomp(sv);
866 else if (SvTYPE(sv) == SVt_PVHV) {
869 (void)hv_iterinit(hv);
871 while ((entry = hv_iternext(hv)))
872 count += do_chomp(hv_iterval(hv,entry));
875 else if (SvREADONLY(sv))
876 Perl_croak(aTHX_ PL_no_modify);
878 if (len && !SvPOKp(sv))
879 s = SvPV_force(sv, len);
886 while (len && s[-1] == '\n') {
894 char *rsptr = SvPV(PL_rs, rslen);
905 if (memNE(s, rsptr, rslen))
920 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
922 dTHR; /* just for taint */
937 bool left_utf = DO_UTF8(left);
938 bool right_utf = DO_UTF8(right);
940 if (left_utf && !right_utf)
941 sv_utf8_upgrade(right);
942 if (!left_utf && right_utf)
943 sv_utf8_upgrade(left);
945 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
946 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
947 lsave = lc = SvPV(left, leftlen);
948 rsave = rc = SvPV(right, rightlen);
949 len = leftlen < rightlen ? leftlen : rightlen;
951 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
953 dc = SvPV_force(sv, n_a);
954 if (SvCUR(sv) < len) {
955 dc = SvGROW(sv, len + 1);
956 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
960 I32 needlen = ((optype == OP_BIT_AND)
961 ? len : (leftlen > rightlen ? leftlen : rightlen));
962 Newz(801, dc, needlen + 1, char);
963 (void)sv_usepvn(sv, dc, needlen);
964 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
967 (void)SvPOK_only(sv);
968 if (left_utf || right_utf) {
970 STRLEN lulen = leftlen;
971 STRLEN rulen = rightlen;
975 if (optype != OP_BIT_AND)
976 dc = SvGROW(sv, leftlen+rightlen+1);
980 while (lulen && rulen) {
981 luc = utf8_to_uv((U8*)lc, &ulen);
984 ruc = utf8_to_uv((U8*)rc, &ulen);
988 dc = (char*)uv_to_utf8((U8*)dc, duc);
990 dulen = dc - SvPVX(sv);
991 SvCUR_set(sv, dulen);
994 while (lulen && rulen) {
995 luc = utf8_to_uv((U8*)lc, &ulen);
998 ruc = utf8_to_uv((U8*)rc, &ulen);
1002 dc = (char*)uv_to_utf8((U8*)dc, duc);
1006 while (lulen && rulen) {
1007 luc = utf8_to_uv((U8*)lc, &ulen);
1010 ruc = utf8_to_uv((U8*)rc, &ulen);
1014 dc = (char*)uv_to_utf8((U8*)dc, duc);
1017 dulen = dc - SvPVX(sv);
1018 SvCUR_set(sv, dulen);
1020 sv_catpvn(sv, rc, rulen);
1022 sv_catpvn(sv, lc, lulen);
1032 if (len >= sizeof(long)*4 &&
1033 !((long)dc % sizeof(long)) &&
1034 !((long)lc % sizeof(long)) &&
1035 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1037 I32 remainder = len % (sizeof(long)*4);
1038 len /= (sizeof(long)*4);
1047 *dl++ = *ll++ & *rl++;
1048 *dl++ = *ll++ & *rl++;
1049 *dl++ = *ll++ & *rl++;
1050 *dl++ = *ll++ & *rl++;
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++;
1081 *dc++ = *lc++ & *rc++;
1085 *dc++ = *lc++ ^ *rc++;
1089 *dc++ = *lc++ | *rc++;
1093 sv_catpvn(sv, rsave + len, rightlen - len);
1094 else if (leftlen > len)
1095 sv_catpvn(sv, lsave + len, leftlen - len);
1113 I32 gimme = GIMME_V;
1114 I32 dokeys = (PL_op->op_type == OP_KEYS);
1115 I32 dovalues = (PL_op->op_type == OP_VALUES);
1116 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1118 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1119 dokeys = dovalues = TRUE;
1122 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1123 dTARGET; /* make sure to clear its target here */
1124 if (SvTYPE(TARG) == SVt_PVLV)
1125 LvTARG(TARG) = Nullsv;
1131 keys = realhv ? hv : avhv_keys((AV*)hv);
1132 (void)hv_iterinit(keys); /* always reset iterator regardless */
1134 if (gimme == G_VOID)
1137 if (gimme == G_SCALAR) {
1141 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1142 if (SvTYPE(TARG) < SVt_PVLV) {
1143 sv_upgrade(TARG, SVt_PVLV);
1144 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1147 if (LvTARG(TARG) != (SV*)keys) {
1149 SvREFCNT_dec(LvTARG(TARG));
1150 LvTARG(TARG) = SvREFCNT_inc(keys);
1156 if (! SvTIED_mg((SV*)keys, 'P'))
1161 while (hv_iternext(keys)) i++;
1167 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1169 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1170 while ((entry = hv_iternext(keys))) {
1173 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1177 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1178 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1179 (unsigned long)HeHASH(entry),
1181 (unsigned long)(HeHASH(entry) & HvMAX(keys))));