3 * Copyright (c) 1991-1997, 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."
17 #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
22 do_trans_CC_simple(SV *sv)
32 tbl = (short*)cPVOP->op_pv;
34 croak("panic: do_trans");
36 s = (U8*)SvPV(sv, len);
40 if ((ch = tbl[*s]) >= 0) {
52 do_trans_CC_count(SV *sv)
61 tbl = (short*)cPVOP->op_pv;
63 croak("panic: do_trans");
65 s = (U8*)SvPV(sv, len);
78 do_trans_CC_complex(SV *sv)
89 tbl = (short*)cPVOP->op_pv;
91 croak("panic: do_trans");
93 s = (U8*)SvPV(sv, len);
97 if (PL_op->op_private & OPpTRANS_SQUASH) {
101 if ((ch = tbl[*s]) >= 0) {
104 if (p == d - 1 && *p == *d)
109 else if (ch == -1) /* -1 is unmapped character */
110 *d++ = *s; /* -2 is delete character */
116 if ((ch = tbl[*s]) >= 0) {
121 else if (ch == -1) /* -1 is unmapped character */
122 *d++ = *s; /* -2 is delete character */
126 matches += send - d; /* account for disappeared chars */
128 SvCUR_set(sv, d - (U8*)SvPVX(sv));
135 do_trans_UU_simple(SV *sv)
144 SV* rv = (SV*)cSVOP->op_sv;
145 HV* hv = (HV*)SvRV(rv);
146 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
147 UV none = svp ? SvUV(*svp) : 0x7fffffff;
152 s = (U8*)SvPV(sv, len);
155 svp = hv_fetch(hv, "FINAL", 5, FALSE);
161 if ((uv = swash_fetch(rv, s)) < none) {
164 d = uv_to_utf8(d, uv);
166 else if (uv == none) {
168 for (i = UTF8SKIP(s); i; i--)
171 else if (uv == extra) {
174 d = uv_to_utf8(d, final);
180 SvCUR_set(sv, d - (U8*)SvPVX(sv));
187 do_trans_UU_count(SV *sv)
195 SV* rv = (SV*)cSVOP->op_sv;
196 HV* hv = (HV*)SvRV(rv);
197 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
198 UV none = svp ? SvUV(*svp) : 0x7fffffff;
201 s = (U8*)SvPV(sv, len);
205 if ((uv = swash_fetch(rv, s)) < none) {
215 do_trans_UC_simple(SV *sv)
224 SV* rv = (SV*)cSVOP->op_sv;
225 HV* hv = (HV*)SvRV(rv);
226 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
227 UV none = svp ? SvUV(*svp) : 0x7fffffff;
232 s = (U8*)SvPV(sv, len);
235 svp = hv_fetch(hv, "FINAL", 5, FALSE);
241 if ((uv = swash_fetch(rv, s)) < none) {
246 else if (uv == none) {
248 uv = utf8_to_uv(s, &ulen);
252 else if (uv == extra) {
261 SvCUR_set(sv, d - (U8*)SvPVX(sv));
268 do_trans_CU_simple(SV *sv)
278 SV* rv = (SV*)cSVOP->op_sv;
279 HV* hv = (HV*)SvRV(rv);
280 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
281 UV none = svp ? SvUV(*svp) : 0x7fffffff;
288 s = (U8*)SvPV(sv, len);
291 svp = hv_fetch(hv, "BITS", 4, FALSE);
293 bits = (I32)SvIV(*svp);
295 svp = hv_fetch(hv, "FINAL", 5, FALSE);
299 Newz(801, d, len * (bits >> 3) + 1, U8);
307 tmpbuf[0] = (( uv >> 6) | 0xc0);
308 tmpbuf[1] = (( uv & 0x3f) | 0x80);
311 if ((uv = swash_fetch(rv, tmpbuf)) < none) {
313 d = uv_to_utf8(d, uv);
316 d = uv_to_utf8(d, s[-1]);
317 else if (uv == extra) {
319 d = uv_to_utf8(d, final);
323 sv_usepvn_mg(sv, (char*)dst, d - dst);
328 /* utf-8 to latin-1 */
331 do_trans_UC_trivial(SV *sv)
339 s = (U8*)SvPV(sv, len);
348 UV uv = utf8_to_uv(s, &ulen);
354 SvCUR_set(sv, d - (U8*)SvPVX(sv));
360 /* latin-1 to utf-8 */
363 do_trans_CU_trivial(SV *sv)
373 s = (U8*)SvPV(sv, len);
376 Newz(801, d, len * 2 + 1, U8);
386 *d++ = (( uv >> 6) | 0xc0);
387 *d++ = (( uv & 0x3f) | 0x80);
391 sv_usepvn_mg(sv, (char*)dst, d - dst);
397 do_trans_UU_complex(SV *sv)
404 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
405 I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF;
406 I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF;
407 I32 del = PL_op->op_private & OPpTRANS_DELETE;
408 SV* rv = (SV*)cSVOP->op_sv;
409 HV* hv = (HV*)SvRV(rv);
410 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
411 UV none = svp ? SvUV(*svp) : 0x7fffffff;
418 s = (U8*)SvPV(sv, len);
421 svp = hv_fetch(hv, "FINAL", 5, FALSE);
425 if (PL_op->op_private & OPpTRANS_GROWS) {
428 svp = hv_fetch(hv, "BITS", 4, FALSE);
430 bits = (I32)SvIV(*svp);
432 Newz(801, d, len * (bits >> 3) + 1, U8);
444 uv = swash_fetch(rv, s);
452 tmpbuf[0] = (( uv >> 6) | 0xc0);
453 tmpbuf[1] = (( uv & 0x3f) | 0x80);
455 uv = swash_fetch(rv, tmpbuf);
460 if (uv >= 0x80 && to_utf)
461 d = uv_to_utf8(d, uv);
470 else if (uv == none) { /* "none" is unmapped character */
476 for (i = UTF8SKIP(s); i; --i)
481 *d++ = (U8)utf8_to_uv(s, &ulen);
485 else { /* must be to_utf only */
486 d = uv_to_utf8(d, s[-1]);
491 else if (uv == extra && !del) {
494 if (final >= 0x80 && to_utf)
495 d = uv_to_utf8(d, final);
504 matches++; /* "none+1" is delete character */
512 uv = swash_fetch(rv, s);
520 tmpbuf[0] = (( uv >> 6) | 0xc0);
521 tmpbuf[1] = (( uv & 0x3f) | 0x80);
523 uv = swash_fetch(rv, tmpbuf);
527 if (uv >= 0x80 && to_utf)
528 d = uv_to_utf8(d, uv);
535 else if (uv == none) { /* "none" is unmapped character */
541 for (i = UTF8SKIP(s); i; --i)
546 *d++ = (U8)utf8_to_uv(s, &ulen);
550 else { /* must be to_utf only */
551 d = uv_to_utf8(d, s[-1]);
555 else if (uv == extra && !del) {
557 if (final >= 0x80 && to_utf)
558 d = uv_to_utf8(d, final);
565 matches++; /* "none+1" is delete character */
571 sv_usepvn(sv, (char*)dst, d - dst);
574 SvCUR_set(sv, d - (U8*)SvPVX(sv));
586 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
593 (void)SvPV_force(sv, len);
594 (void)SvPOK_only(sv);
596 DEBUG_t( deb("2.TBL\n"));
598 switch (PL_op->op_private & 63) {
600 return do_trans_CC_simple(sv);
602 case OPpTRANS_FROM_UTF:
603 return do_trans_UC_simple(sv);
605 case OPpTRANS_TO_UTF:
606 return do_trans_CU_simple(sv);
608 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF:
609 return do_trans_UU_simple(sv);
611 case OPpTRANS_IDENTICAL:
612 return do_trans_CC_count(sv);
614 case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL:
615 return do_trans_UC_trivial(sv);
617 case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
618 return do_trans_CU_trivial(sv);
620 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
621 return do_trans_UU_count(sv);
624 if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
625 return do_trans_UU_complex(sv); /* could be UC or CU too */
627 return do_trans_CC_complex(sv);
632 do_join(register SV *sv, SV *del, register SV **mark, register SV **sp)
635 register I32 items = sp - mark;
638 register char *delim = SvPV(del, delimlen);
642 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
643 if (SvTYPE(sv) < SVt_PV)
644 sv_upgrade(sv, SVt_PV);
645 if (SvLEN(sv) < len + items) { /* current length is way too short */
646 while (items-- > 0) {
647 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
653 SvGROW(sv, len + 1); /* so try to pre-extend */
664 s = SvPV(*mark, tmplen);
665 sv_setpvn(sv, s, tmplen);
675 for (; items > 0; items--,mark++) {
676 sv_catpvn(sv,delim,len);
681 for (; items > 0; items--,mark++)
688 do_sprintf(SV *sv, I32 len, SV **sarg)
691 char *pat = SvPV(*sarg, patlen);
692 bool do_taint = FALSE;
694 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
703 SV *targ = LvTARG(sv);
706 register unsigned char *s;
707 register unsigned long lval;
714 s = (unsigned char*)SvPV_force(targ, targlen);
715 lval = U_L(SvNV(sv));
716 offset = LvTARGOFF(sv);
717 size = LvTARGLEN(sv);
719 len = (offset + size + 7) / 8;
721 s = (unsigned char*)SvGROW(targ, len + 1);
722 (void)memzero(s + targlen, len - targlen + 1);
723 SvCUR_set(targ, len);
727 mask = (1 << size) - 1;
731 s[offset] &= ~(mask << size);
732 s[offset] |= lval << size;
737 s[offset] = lval & 255;
738 else if (size == 16) {
739 s[offset] = (lval >> 8) & 255;
740 s[offset+1] = lval & 255;
742 else if (size == 32) {
743 s[offset] = (lval >> 24) & 255;
744 s[offset+1] = (lval >> 16) & 255;
745 s[offset+2] = (lval >> 8) & 255;
746 s[offset+3] = lval & 255;
752 do_chop(register SV *astr, register SV *sv)
758 if (SvTYPE(sv) == SVt_PVAV) {
763 for (i = 0; i <= max; i++) {
764 sv = (SV*)av_fetch(av, i, FALSE);
765 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
770 if (SvTYPE(sv) == SVt_PVHV) {
773 (void)hv_iterinit(hv);
775 while (entry = hv_iternext(hv))
776 do_chop(astr,hv_iterval(hv,entry));
780 if (len && !SvPOK(sv))
781 s = SvPV_force(sv, len);
784 char *send = s + len;
787 while ((*s & 0xc0) == 0x80)
789 if (UTF8SKIP(s) != send - s)
790 warn("Malformed UTF-8 character");
791 sv_setpvn(astr, s, send - s);
793 SvCUR_set(sv, s - start);
797 sv_setpvn(astr, "", 0);
802 sv_setpvn(astr, s, 1);
808 sv_setpvn(astr, "", 0);
813 do_chomp(register SV *sv)
823 if (SvTYPE(sv) == SVt_PVAV) {
828 for (i = 0; i <= max; i++) {
829 sv = (SV*)av_fetch(av, i, FALSE);
830 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
831 count += do_chomp(sv);
835 if (SvTYPE(sv) == SVt_PVHV) {
838 (void)hv_iterinit(hv);
840 while (entry = hv_iternext(hv))
841 count += do_chomp(hv_iterval(hv,entry));
845 if (len && !SvPOKp(sv))
846 s = SvPV_force(sv, len);
853 while (len && s[-1] == '\n') {
861 char *rsptr = SvPV(PL_rs, rslen);
872 if (memNE(s, rsptr, rslen))
887 do_vop(I32 optype, SV *sv, SV *left, SV *right)
889 dTHR; /* just for taint */
905 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
906 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
907 lsave = lc = SvPV(left, leftlen);
908 rsave = rc = SvPV(right, rightlen);
909 len = leftlen < rightlen ? leftlen : rightlen;
911 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
912 dc = SvPV_force(sv, PL_na);
913 if (SvCUR(sv) < len) {
914 dc = SvGROW(sv, len + 1);
915 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
919 I32 needlen = ((optype == OP_BIT_AND)
920 ? len : (leftlen > rightlen ? leftlen : rightlen));
921 Newz(801, dc, needlen + 1, char);
922 (void)sv_usepvn(sv, dc, needlen);
923 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
926 (void)SvPOK_only(sv);
928 if (len >= sizeof(long)*4 &&
929 !((long)dc % sizeof(long)) &&
930 !((long)lc % sizeof(long)) &&
931 !((long)rc % sizeof(long))) /* It's almost always aligned... */
933 I32 remainder = len % (sizeof(long)*4);
934 len /= (sizeof(long)*4);
943 *dl++ = *ll++ & *rl++;
944 *dl++ = *ll++ & *rl++;
945 *dl++ = *ll++ & *rl++;
946 *dl++ = *ll++ & *rl++;
951 *dl++ = *ll++ ^ *rl++;
952 *dl++ = *ll++ ^ *rl++;
953 *dl++ = *ll++ ^ *rl++;
954 *dl++ = *ll++ ^ *rl++;
959 *dl++ = *ll++ | *rl++;
960 *dl++ = *ll++ | *rl++;
961 *dl++ = *ll++ | *rl++;
962 *dl++ = *ll++ | *rl++;
977 *dc++ = *lc++ & *rc++;
981 *dc++ = *lc++ ^ *rc++;
985 *dc++ = *lc++ | *rc++;
989 sv_catpvn(sv, rsave + len, rightlen - len);
990 else if (leftlen > len)
991 sv_catpvn(sv, lsave + len, leftlen - len);
1008 I32 gimme = GIMME_V;
1009 I32 dokeys = (PL_op->op_type == OP_KEYS);
1010 I32 dovalues = (PL_op->op_type == OP_VALUES);
1011 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1013 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1014 dokeys = dovalues = TRUE;
1017 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1018 dTARGET; /* make sure to clear its target here */
1019 if (SvTYPE(TARG) == SVt_PVLV)
1020 LvTARG(TARG) = Nullsv;
1026 keys = realhv ? hv : avhv_keys((AV*)hv);
1027 (void)hv_iterinit(keys); /* always reset iterator regardless */
1029 if (gimme == G_VOID)
1032 if (gimme == G_SCALAR) {
1036 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
1037 if (SvTYPE(TARG) < SVt_PVLV) {
1038 sv_upgrade(TARG, SVt_PVLV);
1039 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1042 if (LvTARG(TARG) != (SV*)keys) {
1044 SvREFCNT_dec(LvTARG(TARG));
1045 LvTARG(TARG) = SvREFCNT_inc(keys);
1051 if (!SvRMAGICAL(keys) || !mg_find((SV*)keys,'P'))
1056 while (hv_iternext(keys)) i++;
1062 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1064 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1065 while (entry = hv_iternext(keys)) {
1068 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1070 tmpstr = sv_newmortal();
1072 sv_setsv(tmpstr,realhv ?
1073 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry));
1074 DEBUG_H(sv_setpvf(tmpstr, "%lu%%%d=%lu",
1075 (unsigned long)HeHASH(entry),
1077 (unsigned long)(HeHASH(entry) & HvMAX(keys))));