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 #define HALF_UPGRADE(start,end) { \
29 newstr = bytes_to_utf8(start, &len); \
30 Copy(newstr,start,len,U8*); \
36 S_do_trans_simple(pTHX_ SV *sv)
44 I32 sutf = SvUTF8(sv);
49 tbl = (short*)cPVOP->op_pv;
51 Perl_croak(aTHX_ "panic: do_trans");
53 s = (U8*)SvPV(sv, len);
56 /* First, take care of non-UTF8 input strings, because they're easy */
59 if ((ch = tbl[*s]) >= 0) {
69 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
70 Newz(0, d, len*2+1, U8);
77 /* Need to check this, otherwise 128..255 won't match */
78 c = utf8_to_uv(s, &ulen);
79 if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
86 } else { /* No match -> copy */
92 sv_setpvn(sv, (const char*)dstart, d - dstart);
94 SvLEN_set(sv, 2*len+1);
100 S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
106 I32 hasutf = SvUTF8(sv);
110 tbl = (short*)cPVOP->op_pv;
112 Perl_croak(aTHX_ "panic: do_trans");
114 s = (U8*)SvPV(sv, len);
118 if (hasutf && *s & 0x80)
125 c = utf8_to_uv(s,&ulen);
128 if (c < 0x100 && tbl[c] >= 0)
138 S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
144 I32 hasutf = SvUTF8(sv);
150 tbl = (short*)cPVOP->op_pv;
152 Perl_croak(aTHX_ "panic: do_trans");
154 s = (U8*)SvPV(sv, len);
158 if (PL_op->op_private & OPpTRANS_SQUASH) {
162 if (hasutf && *s & 0x80)
165 if ((ch = tbl[*s]) >= 0) {
168 if (p == d - 1 && *p == *d)
173 else if (ch == -1) /* -1 is unmapped character */
174 *d++ = *s; /* -2 is delete character */
181 if (hasutf && *s & 0x80)
184 if ((ch = tbl[*s]) >= 0) {
189 else if (ch == -1) /* -1 is unmapped character */
190 *d++ = *s; /* -2 is delete character */
195 matches += send - d; /* account for disappeared chars */
197 SvCUR_set(sv, d - (U8*)SvPVX(sv));
204 S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
215 SV* rv = (SV*)cSVOP->op_sv;
216 HV* hv = (HV*)SvRV(rv);
217 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
218 UV none = svp ? SvUV(*svp) : 0x7fffffff;
226 s = (U8*)SvPV(sv, len);
230 svp = hv_fetch(hv, "FINAL", 5, FALSE);
234 /* d needs to be bigger than s, in case e.g. upgrading is required */
235 Newz(0, d, len*2+1, U8);
238 if ((uv = swash_fetch(rv, s)) < none) {
241 if (uv & 0x80 && !isutf++)
242 HALF_UPGRADE(dstart,d);
243 d = uv_to_utf8(d, uv);
245 else if (uv == none) {
248 if (i > 1 && !isutf++)
249 HALF_UPGRADE(dstart,d);
253 else if (uv == extra) {
258 if (i > 1 && !isutf++)
259 HALF_UPGRADE(dstart,d);
260 d = uv_to_utf8(d, final);
266 sv_setpvn(sv, (const char*)dstart, d - dstart);
275 S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
283 SV* rv = (SV*)cSVOP->op_sv;
284 HV* hv = (HV*)SvRV(rv);
285 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
286 UV none = svp ? SvUV(*svp) : 0x7fffffff;
289 s = (U8*)SvPV(sv, len);
291 s = bytes_to_utf8(s, &len);
295 if ((uv = swash_fetch(rv, s)) < none)
304 S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
311 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
312 I32 del = PL_op->op_private & OPpTRANS_DELETE;
313 SV* rv = (SV*)cSVOP->op_sv;
314 HV* hv = (HV*)SvRV(rv);
315 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
316 UV none = svp ? SvUV(*svp) : 0x7fffffff;
322 I32 isutf = SvUTF8(sv);
324 s = (U8*)SvPV(sv, len);
327 svp = hv_fetch(hv, "FINAL", 5, FALSE);
331 Newz(0, d, len*2+1, U8);
338 uv = swash_fetch(rv, s);
345 tmpbuf[0] = (( uv >> 6) | 0xc0);
346 tmpbuf[1] = (( uv & 0x3f) | 0x80);
348 uv = swash_fetch(rv, tmpbuf);
354 if (uv & 0x80 && !isutf++)
356 d = uv_to_utf8(d, uv);
362 else if (uv == none) { /* "none" is unmapped character */
364 *d++ = (U8)utf8_to_uv(s, &ulen);
369 else if (uv == extra && !del) {
372 d = uv_to_utf8(d, final);
378 matches++; /* "none+1" is delete character */
385 uv = swash_fetch(rv, s);
392 tmpbuf[0] = (( uv >> 6) | 0xc0);
393 tmpbuf[1] = (( uv & 0x3f) | 0x80);
395 uv = swash_fetch(rv, tmpbuf);
399 d = uv_to_utf8(d, uv);
403 else if (uv == none) { /* "none" is unmapped character */
405 *d++ = (U8)utf8_to_uv(s, &ulen);
409 else if (uv == extra && !del) {
411 d = uv_to_utf8(d, final);
415 matches++; /* "none+1" is delete character */
420 sv_usepvn(sv, (char*)dst, d - dst);
423 SvCUR_set(sv, d - (U8*)SvPVX(sv));
431 Perl_do_trans(pTHX_ SV *sv)
435 I32 hasutf = (PL_op->op_private &
436 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
438 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
439 Perl_croak(aTHX_ PL_no_modify);
445 (void)SvPV_force(sv, len);
446 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
447 (void)SvPOK_only_UTF8(sv);
449 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
451 switch (PL_op->op_private & ~hasutf & 63) {
454 return do_trans_simple_utf8(sv);
456 return do_trans_simple(sv);
458 case OPpTRANS_IDENTICAL:
460 return do_trans_count_utf8(sv);
462 return do_trans_count(sv);
466 return do_trans_complex_utf8(sv);
468 return do_trans_complex(sv);
473 Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
476 register I32 items = sp - mark;
479 register char *delim = SvPV(del, delimlen);
483 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
484 (void)SvUPGRADE(sv, SVt_PV);
485 if (SvLEN(sv) < len + items) { /* current length is way too short */
486 while (items-- > 0) {
487 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
493 SvGROW(sv, len + 1); /* so try to pre-extend */
512 for (; items > 0; items--,mark++) {
513 sv_catpvn(sv,delim,len);
518 for (; items > 0; items--,mark++)
525 Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
528 char *pat = SvPV(*sarg, patlen);
529 bool do_taint = FALSE;
531 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
537 /* XXX SvUTF8 support missing! */
539 Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
542 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
547 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
548 Perl_croak(aTHX_ "Illegal number of bits in vec");
549 offset *= size; /* turn into bit offset */
550 len = (offset + size + 7) / 8; /* required number of bytes */
555 offset >>= 3; /* turn into byte offset */
557 if (offset >= srclen)
560 retnum = (UV) s[offset] << 8;
562 else if (size == 32) {
563 if (offset >= srclen)
565 else if (offset + 1 >= srclen)
567 ((UV) s[offset ] << 24);
568 else if (offset + 2 >= srclen)
570 ((UV) s[offset ] << 24) +
571 ((UV) s[offset + 1] << 16);
574 ((UV) s[offset ] << 24) +
575 ((UV) s[offset + 1] << 16) +
576 ( s[offset + 2] << 8);
579 else if (size == 64) {
581 if (ckWARN(WARN_PORTABLE))
582 Perl_warner(aTHX_ WARN_PORTABLE,
583 "Bit vector size > 32 non-portable");
584 if (offset >= srclen)
586 else if (offset + 1 >= srclen)
588 (UV) s[offset ] << 56;
589 else if (offset + 2 >= srclen)
591 ((UV) s[offset ] << 56) +
592 ((UV) s[offset + 1] << 48);
593 else if (offset + 3 >= srclen)
595 ((UV) s[offset ] << 56) +
596 ((UV) s[offset + 1] << 48) +
597 ((UV) s[offset + 2] << 40);
598 else if (offset + 4 >= srclen)
600 ((UV) s[offset ] << 56) +
601 ((UV) s[offset + 1] << 48) +
602 ((UV) s[offset + 2] << 40) +
603 ((UV) s[offset + 3] << 32);
604 else if (offset + 5 >= srclen)
606 ((UV) s[offset ] << 56) +
607 ((UV) s[offset + 1] << 48) +
608 ((UV) s[offset + 2] << 40) +
609 ((UV) s[offset + 3] << 32) +
610 ( s[offset + 4] << 24);
611 else if (offset + 6 >= srclen)
613 ((UV) s[offset ] << 56) +
614 ((UV) s[offset + 1] << 48) +
615 ((UV) s[offset + 2] << 40) +
616 ((UV) s[offset + 3] << 32) +
617 ((UV) s[offset + 4] << 24) +
618 ((UV) s[offset + 5] << 16);
621 ((UV) s[offset ] << 56) +
622 ((UV) s[offset + 1] << 48) +
623 ((UV) s[offset + 2] << 40) +
624 ((UV) s[offset + 3] << 32) +
625 ((UV) s[offset + 4] << 24) +
626 ((UV) s[offset + 5] << 16) +
627 ( s[offset + 6] << 8);
633 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
635 offset >>= 3; /* turn into byte offset */
640 ((UV) s[offset] << 8) +
644 ((UV) s[offset ] << 24) +
645 ((UV) s[offset + 1] << 16) +
646 ( s[offset + 2] << 8) +
649 else if (size == 64) {
651 if (ckWARN(WARN_PORTABLE))
652 Perl_warner(aTHX_ WARN_PORTABLE,
653 "Bit vector size > 32 non-portable");
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) +
670 /* XXX SvUTF8 support missing! */
672 Perl_do_vecset(pTHX_ SV *sv)
674 SV *targ = LvTARG(sv);
677 register unsigned char *s;
685 s = (unsigned char*)SvPV_force(targ, targlen);
686 (void)SvPOK_only(targ);
688 offset = LvTARGOFF(sv);
689 size = LvTARGLEN(sv);
690 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
691 Perl_croak(aTHX_ "Illegal number of bits in vec");
693 offset *= size; /* turn into bit offset */
694 len = (offset + size + 7) / 8; /* required number of bytes */
696 s = (unsigned char*)SvGROW(targ, len + 1);
697 (void)memzero((char *)(s + targlen), len - targlen + 1);
698 SvCUR_set(targ, len);
702 mask = (1 << size) - 1;
705 offset >>= 3; /* turn into byte offset */
706 s[offset] &= ~(mask << size);
707 s[offset] |= lval << size;
710 offset >>= 3; /* turn into byte offset */
712 s[offset ] = lval & 0xff;
713 else if (size == 16) {
714 s[offset ] = (lval >> 8) & 0xff;
715 s[offset+1] = lval & 0xff;
717 else if (size == 32) {
718 s[offset ] = (lval >> 24) & 0xff;
719 s[offset+1] = (lval >> 16) & 0xff;
720 s[offset+2] = (lval >> 8) & 0xff;
721 s[offset+3] = lval & 0xff;
724 else if (size == 64) {
726 if (ckWARN(WARN_PORTABLE))
727 Perl_warner(aTHX_ WARN_PORTABLE,
728 "Bit vector size > 32 non-portable");
729 s[offset ] = (lval >> 56) & 0xff;
730 s[offset+1] = (lval >> 48) & 0xff;
731 s[offset+2] = (lval >> 40) & 0xff;
732 s[offset+3] = (lval >> 32) & 0xff;
733 s[offset+4] = (lval >> 24) & 0xff;
734 s[offset+5] = (lval >> 16) & 0xff;
735 s[offset+6] = (lval >> 8) & 0xff;
736 s[offset+7] = lval & 0xff;
744 Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
750 if (SvTYPE(sv) == SVt_PVAV) {
755 for (i = 0; i <= max; i++) {
756 sv = (SV*)av_fetch(av, i, FALSE);
757 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
762 else if (SvTYPE(sv) == SVt_PVHV) {
765 (void)hv_iterinit(hv);
767 while ((entry = hv_iternext(hv)))
768 do_chop(astr,hv_iterval(hv,entry));
771 else if (SvREADONLY(sv))
772 Perl_croak(aTHX_ PL_no_modify);
774 if (len && !SvPOK(sv))
775 s = SvPV_force(sv, len);
778 char *send = s + len;
781 while ((*s & 0xc0) == 0x80)
783 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
784 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
785 sv_setpvn(astr, s, send - s);
787 SvCUR_set(sv, s - start);
792 sv_setpvn(astr, "", 0);
796 sv_setpvn(astr, s, 1);
803 sv_setpvn(astr, "", 0);
808 Perl_do_chomp(pTHX_ register SV *sv)
820 if (SvTYPE(sv) == SVt_PVAV) {
825 for (i = 0; i <= max; i++) {
826 sv = (SV*)av_fetch(av, i, FALSE);
827 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
828 count += do_chomp(sv);
832 else if (SvTYPE(sv) == SVt_PVHV) {
835 (void)hv_iterinit(hv);
837 while ((entry = hv_iternext(hv)))
838 count += do_chomp(hv_iterval(hv,entry));
841 else if (SvREADONLY(sv))
842 Perl_croak(aTHX_ PL_no_modify);
844 if (len && !SvPOKp(sv))
845 s = SvPV_force(sv, len);
852 while (len && s[-1] == '\n') {
860 char *rsptr = SvPV(PL_rs, rslen);
871 if (memNE(s, rsptr, rslen))
886 Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
888 dTHR; /* just for taint */
903 bool left_utf = DO_UTF8(left);
904 bool right_utf = DO_UTF8(right);
906 if (left_utf && !right_utf)
907 sv_utf8_upgrade(right);
908 if (!left_utf && right_utf)
909 sv_utf8_upgrade(left);
911 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
912 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
913 lsave = lc = SvPV(left, leftlen);
914 rsave = rc = SvPV(right, rightlen);
915 len = leftlen < rightlen ? leftlen : rightlen;
917 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
919 dc = SvPV_force(sv, n_a);
920 if (SvCUR(sv) < len) {
921 dc = SvGROW(sv, len + 1);
922 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
926 I32 needlen = ((optype == OP_BIT_AND)
927 ? len : (leftlen > rightlen ? leftlen : rightlen));
928 Newz(801, dc, needlen + 1, char);
929 (void)sv_usepvn(sv, dc, needlen);
930 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
933 (void)SvPOK_only(sv);
934 if (left_utf || right_utf) {
936 STRLEN lulen = leftlen;
937 STRLEN rulen = rightlen;
941 if (optype != OP_BIT_AND)
942 dc = SvGROW(sv, leftlen+rightlen+1);
946 while (lulen && rulen) {
947 luc = utf8_to_uv((U8*)lc, &ulen);
950 ruc = utf8_to_uv((U8*)rc, &ulen);
954 dc = (char*)uv_to_utf8((U8*)dc, duc);
956 dulen = dc - SvPVX(sv);
957 SvCUR_set(sv, dulen);
960 while (lulen && rulen) {
961 luc = utf8_to_uv((U8*)lc, &ulen);
964 ruc = utf8_to_uv((U8*)rc, &ulen);
968 dc = (char*)uv_to_utf8((U8*)dc, duc);
972 while (lulen && rulen) {
973 luc = utf8_to_uv((U8*)lc, &ulen);
976 ruc = utf8_to_uv((U8*)rc, &ulen);
980 dc = (char*)uv_to_utf8((U8*)dc, duc);
983 dulen = dc - SvPVX(sv);
984 SvCUR_set(sv, dulen);
986 sv_catpvn(sv, rc, rulen);
988 sv_catpvn(sv, lc, lulen);
998 if (len >= sizeof(long)*4 &&
999 !((long)dc % sizeof(long)) &&
1000 !((long)lc % sizeof(long)) &&
1001 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1003 I32 remainder = len % (sizeof(long)*4);
1004 len /= (sizeof(long)*4);
1013 *dl++ = *ll++ & *rl++;
1014 *dl++ = *ll++ & *rl++;
1015 *dl++ = *ll++ & *rl++;
1016 *dl++ = *ll++ & *rl++;
1021 *dl++ = *ll++ ^ *rl++;
1022 *dl++ = *ll++ ^ *rl++;
1023 *dl++ = *ll++ ^ *rl++;
1024 *dl++ = *ll++ ^ *rl++;
1029 *dl++ = *ll++ | *rl++;
1030 *dl++ = *ll++ | *rl++;
1031 *dl++ = *ll++ | *rl++;
1032 *dl++ = *ll++ | *rl++;
1047 *dc++ = *lc++ & *rc++;
1051 *dc++ = *lc++ ^ *rc++;
1055 *dc++ = *lc++ | *rc++;
1059 sv_catpvn(sv, rsave + len, rightlen - len);
1060 else if (leftlen > len)
1061 sv_catpvn(sv, lsave + len, leftlen - len);
1079 I32 gimme = GIMME_V;
1080 I32 dokeys = (PL_op->op_type == OP_KEYS);
1081 I32 dovalues = (PL_op->op_type == OP_VALUES);
1082 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1084 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1085 dokeys = dovalues = TRUE;
1088 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1089 dTARGET; /* make sure to clear its target here */
1090 if (SvTYPE(TARG) == SVt_PVLV)
1091 LvTARG(TARG) = Nullsv;
1097 keys = realhv ? hv : avhv_keys((AV*)hv);
1098 (void)hv_iterinit(keys); /* always reset iterator regardless */
1100 if (gimme == G_VOID)
1103 if (gimme == G_SCALAR) {
1107 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1108 if (SvTYPE(TARG) < SVt_PVLV) {
1109 sv_upgrade(TARG, SVt_PVLV);
1110 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1113 if (LvTARG(TARG) != (SV*)keys) {
1115 SvREFCNT_dec(LvTARG(TARG));
1116 LvTARG(TARG) = SvREFCNT_inc(keys);
1122 if (! SvTIED_mg((SV*)keys, 'P'))
1127 while (hv_iternext(keys)) i++;
1133 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1135 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1136 while ((entry = hv_iternext(keys))) {
1139 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1143 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1144 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1145 (unsigned long)HeHASH(entry),
1147 (unsigned long)(HeHASH(entry) & HvMAX(keys))));