Perl_eval_pv() leaks 4 bytes every time it is called because it
[p5sagit/p5-mst-13.2.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
3818b22b 3 * Copyright (c) 1991-2000, Larry Wall
79072805 4 *
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.
7 *
a0d0e21e 8 */
9
10/*
11 * "'So that was the job I felt I had to do when I started,' thought Sam."
79072805 12 */
13
14#include "EXTERN.h"
864dbfa3 15#define PERL_IN_DOOP_C
79072805 16#include "perl.h"
17
18#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
19#include <signal.h>
20#endif
21
942e002e 22STATIC I32
cea2e8a9 23S_do_trans_CC_simple(pTHX_ SV *sv)
79072805 24{
11343788 25 dTHR;
4757a243 26 U8 *s;
27 U8 *send;
28 I32 matches = 0;
463ee0b2 29 STRLEN len;
4757a243 30 short *tbl;
31 I32 ch;
79072805 32
4757a243 33 tbl = (short*)cPVOP->op_pv;
34 if (!tbl)
cea2e8a9 35 Perl_croak(aTHX_ "panic: do_trans");
a0ed51b3 36
4757a243 37 s = (U8*)SvPV(sv, len);
38 send = s + len;
39
40 while (s < send) {
41 if ((ch = tbl[*s]) >= 0) {
42 matches++;
43 *s = ch;
79072805 44 }
4757a243 45 s++;
46 }
47 SvSETMAGIC(sv);
48
49 return matches;
50}
51
942e002e 52STATIC I32
cea2e8a9 53S_do_trans_CC_count(pTHX_ SV *sv)
4757a243 54{
55 dTHR;
56 U8 *s;
57 U8 *send;
58 I32 matches = 0;
59 STRLEN len;
60 short *tbl;
61
62 tbl = (short*)cPVOP->op_pv;
63 if (!tbl)
cea2e8a9 64 Perl_croak(aTHX_ "panic: do_trans");
4757a243 65
66 s = (U8*)SvPV(sv, len);
67 send = s + len;
68
69 while (s < send) {
70 if (tbl[*s] >= 0)
71 matches++;
72 s++;
73 }
74
75 return matches;
76}
77
942e002e 78STATIC I32
cea2e8a9 79S_do_trans_CC_complex(pTHX_ SV *sv)
4757a243 80{
81 dTHR;
82 U8 *s;
83 U8 *send;
84 U8 *d;
85 I32 matches = 0;
86 STRLEN len;
87 short *tbl;
88 I32 ch;
89
90 tbl = (short*)cPVOP->op_pv;
91 if (!tbl)
cea2e8a9 92 Perl_croak(aTHX_ "panic: do_trans");
4757a243 93
94 s = (U8*)SvPV(sv, len);
95 send = s + len;
96
97 d = s;
98 if (PL_op->op_private & OPpTRANS_SQUASH) {
99 U8* p = send;
100
101 while (s < send) {
102 if ((ch = tbl[*s]) >= 0) {
103 *d = ch;
104 matches++;
105 if (p == d - 1 && *p == *d)
106 matches--;
a0ed51b3 107 else
4757a243 108 p = d++;
a0ed51b3 109 }
4757a243 110 else if (ch == -1) /* -1 is unmapped character */
111 *d++ = *s; /* -2 is delete character */
112 s++;
a0ed51b3 113 }
4757a243 114 }
115 else {
116 while (s < send) {
117 if ((ch = tbl[*s]) >= 0) {
118 *d = ch;
119 matches++;
120 d++;
a0ed51b3 121 }
4757a243 122 else if (ch == -1) /* -1 is unmapped character */
123 *d++ = *s; /* -2 is delete character */
124 s++;
5d06d08e 125 }
4757a243 126 }
127 matches += send - d; /* account for disappeared chars */
128 *d = '\0';
129 SvCUR_set(sv, d - (U8*)SvPVX(sv));
130 SvSETMAGIC(sv);
131
132 return matches;
133}
134
942e002e 135STATIC I32
cea2e8a9 136S_do_trans_UU_simple(pTHX_ SV *sv)
4757a243 137{
138 dTHR;
139 U8 *s;
140 U8 *send;
141 U8 *d;
142 I32 matches = 0;
143 STRLEN len;
144
145 SV* rv = (SV*)cSVOP->op_sv;
146 HV* hv = (HV*)SvRV(rv);
147 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
148 UV none = svp ? SvUV(*svp) : 0x7fffffff;
149 UV extra = none + 1;
150 UV final;
151 UV uv;
152
153 s = (U8*)SvPV(sv, len);
154 send = s + len;
155
156 svp = hv_fetch(hv, "FINAL", 5, FALSE);
157 if (svp)
158 final = SvUV(*svp);
159
160 d = s;
161 while (s < send) {
162 if ((uv = swash_fetch(rv, s)) < none) {
163 s += UTF8SKIP(s);
164 matches++;
165 d = uv_to_utf8(d, uv);
166 }
167 else if (uv == none) {
168 int i;
169 for (i = UTF8SKIP(s); i; i--)
170 *d++ = *s++;
171 }
172 else if (uv == extra) {
173 s += UTF8SKIP(s);
174 matches++;
175 d = uv_to_utf8(d, final);
176 }
177 else
178 s += UTF8SKIP(s);
179 }
180 *d = '\0';
181 SvCUR_set(sv, d - (U8*)SvPVX(sv));
182 SvSETMAGIC(sv);
183
184 return matches;
185}
186
942e002e 187STATIC I32
cea2e8a9 188S_do_trans_UU_count(pTHX_ SV *sv)
4757a243 189{
190 dTHR;
191 U8 *s;
192 U8 *send;
193 I32 matches = 0;
194 STRLEN len;
195
196 SV* rv = (SV*)cSVOP->op_sv;
197 HV* hv = (HV*)SvRV(rv);
198 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
199 UV none = svp ? SvUV(*svp) : 0x7fffffff;
200 UV uv;
201
202 s = (U8*)SvPV(sv, len);
203 send = s + len;
204
205 while (s < send) {
834a4ddd 206 if ((uv = swash_fetch(rv, s)) < none)
4757a243 207 matches++;
834a4ddd 208 s += UTF8SKIP(s);
4757a243 209 }
210
211 return matches;
212}
213
942e002e 214STATIC I32
cea2e8a9 215S_do_trans_UC_simple(pTHX_ SV *sv)
4757a243 216{
217 dTHR;
218 U8 *s;
219 U8 *send;
220 U8 *d;
221 I32 matches = 0;
222 STRLEN len;
223
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;
228 UV extra = none + 1;
229 UV final;
230 UV uv;
231
232 s = (U8*)SvPV(sv, len);
233 send = s + len;
234
235 svp = hv_fetch(hv, "FINAL", 5, FALSE);
236 if (svp)
237 final = SvUV(*svp);
238
239 d = s;
240 while (s < send) {
241 if ((uv = swash_fetch(rv, s)) < none) {
242 s += UTF8SKIP(s);
243 matches++;
244 *d++ = (U8)uv;
245 }
246 else if (uv == none) {
247 I32 ulen;
248 uv = utf8_to_uv(s, &ulen);
249 s += ulen;
250 *d++ = (U8)uv;
251 }
252 else if (uv == extra) {
253 s += UTF8SKIP(s);
254 matches++;
255 *d++ = (U8)final;
256 }
257 else
258 s += UTF8SKIP(s);
259 }
260 *d = '\0';
261 SvCUR_set(sv, d - (U8*)SvPVX(sv));
262 SvSETMAGIC(sv);
263
264 return matches;
265}
266
942e002e 267STATIC I32
cea2e8a9 268S_do_trans_CU_simple(pTHX_ SV *sv)
4757a243 269{
270 dTHR;
271 U8 *s;
272 U8 *send;
273 U8 *d;
274 U8 *dst;
275 I32 matches = 0;
276 STRLEN len;
277
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;
282 UV extra = none + 1;
283 UV final;
284 UV uv;
806e7201 285 U8 tmpbuf[UTF8_MAXLEN];
4757a243 286 I32 bits = 16;
287
288 s = (U8*)SvPV(sv, len);
289 send = s + len;
290
291 svp = hv_fetch(hv, "BITS", 4, FALSE);
292 if (svp)
293 bits = (I32)SvIV(*svp);
294
295 svp = hv_fetch(hv, "FINAL", 5, FALSE);
296 if (svp)
297 final = SvUV(*svp);
298
299 Newz(801, d, len * (bits >> 3) + 1, U8);
300 dst = d;
301
302 while (s < send) {
303 uv = *s++;
304 if (uv < 0x80)
305 tmpbuf[0] = uv;
306 else {
307 tmpbuf[0] = (( uv >> 6) | 0xc0);
308 tmpbuf[1] = (( uv & 0x3f) | 0x80);
a0ed51b3 309 }
4757a243 310
311 if ((uv = swash_fetch(rv, tmpbuf)) < none) {
312 matches++;
313 d = uv_to_utf8(d, uv);
314 }
315 else if (uv == none)
316 d = uv_to_utf8(d, s[-1]);
317 else if (uv == extra) {
318 matches++;
319 d = uv_to_utf8(d, final);
320 }
321 }
322 *d = '\0';
323 sv_usepvn_mg(sv, (char*)dst, d - dst);
324
325 return matches;
326}
327
328/* utf-8 to latin-1 */
329
942e002e 330STATIC I32
cea2e8a9 331S_do_trans_UC_trivial(pTHX_ SV *sv)
4757a243 332{
333 dTHR;
334 U8 *s;
335 U8 *send;
336 U8 *d;
337 STRLEN len;
338
339 s = (U8*)SvPV(sv, len);
340 send = s + len;
341
342 d = s;
343 while (s < send) {
344 if (*s < 0x80)
345 *d++ = *s++;
a0ed51b3 346 else {
4757a243 347 I32 ulen;
348 UV uv = utf8_to_uv(s, &ulen);
349 s += ulen;
350 *d++ = (U8)uv;
351 }
352 }
353 *d = '\0';
354 SvCUR_set(sv, d - (U8*)SvPVX(sv));
355 SvSETMAGIC(sv);
a0ed51b3 356
4757a243 357 return SvCUR(sv);
358}
a0ed51b3 359
4757a243 360/* latin-1 to utf-8 */
a0ed51b3 361
942e002e 362STATIC I32
cea2e8a9 363S_do_trans_CU_trivial(pTHX_ SV *sv)
4757a243 364{
365 dTHR;
366 U8 *s;
367 U8 *send;
368 U8 *d;
369 U8 *dst;
370 I32 matches;
371 STRLEN len;
a0ed51b3 372
4757a243 373 s = (U8*)SvPV(sv, len);
374 send = s + len;
375
376 Newz(801, d, len * 2 + 1, U8);
377 dst = d;
378
379 matches = send - s;
380
381 while (s < send) {
382 if (*s < 0x80)
383 *d++ = *s++;
384 else {
385 UV uv = *s++;
386 *d++ = (( uv >> 6) | 0xc0);
387 *d++ = (( uv & 0x3f) | 0x80);
388 }
389 }
390 *d = '\0';
391 sv_usepvn_mg(sv, (char*)dst, d - dst);
392
393 return matches;
394}
395
942e002e 396STATIC I32
cea2e8a9 397S_do_trans_UU_complex(pTHX_ SV *sv)
4757a243 398{
399 dTHR;
400 U8 *s;
401 U8 *send;
402 U8 *d;
403 I32 matches = 0;
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;
412 UV extra = none + 1;
413 UV final;
414 UV uv;
415 STRLEN len;
416 U8 *dst;
417
418 s = (U8*)SvPV(sv, len);
419 send = s + len;
420
421 svp = hv_fetch(hv, "FINAL", 5, FALSE);
422 if (svp)
423 final = SvUV(*svp);
424
425 if (PL_op->op_private & OPpTRANS_GROWS) {
426 I32 bits = 16;
427
428 svp = hv_fetch(hv, "BITS", 4, FALSE);
429 if (svp)
430 bits = (I32)SvIV(*svp);
431
432 Newz(801, d, len * (bits >> 3) + 1, U8);
433 dst = d;
434 }
435 else {
436 d = s;
437 dst = 0;
438 }
439
440 if (squash) {
441 UV puv = 0xfeedface;
442 while (s < send) {
443 if (from_utf) {
444 uv = swash_fetch(rv, s);
a0ed51b3 445 }
446 else {
4757a243 447 U8 tmpbuf[2];
448 uv = *s++;
449 if (uv < 0x80)
450 tmpbuf[0] = uv;
451 else {
452 tmpbuf[0] = (( uv >> 6) | 0xc0);
453 tmpbuf[1] = (( uv & 0x3f) | 0x80);
454 }
455 uv = swash_fetch(rv, tmpbuf);
456 }
457 if (uv < none) {
458 matches++;
459 if (uv != puv) {
460 if (uv >= 0x80 && to_utf)
461 d = uv_to_utf8(d, uv);
462 else
463 *d++ = (U8)uv;
464 puv = uv;
465 }
466 if (from_utf)
467 s += UTF8SKIP(s);
468 continue;
469 }
470 else if (uv == none) { /* "none" is unmapped character */
471 if (from_utf) {
472 if (*s < 0x80)
473 *d++ = *s++;
474 else if (to_utf) {
a0ed51b3 475 int i;
4757a243 476 for (i = UTF8SKIP(s); i; --i)
477 *d++ = *s++;
a0ed51b3 478 }
4757a243 479 else {
480 I32 ulen;
481 *d++ = (U8)utf8_to_uv(s, &ulen);
482 s += ulen;
a0ed51b3 483 }
a0ed51b3 484 }
4757a243 485 else { /* must be to_utf only */
486 d = uv_to_utf8(d, s[-1]);
487 }
488 puv = 0xfeedface;
489 continue;
a0ed51b3 490 }
4757a243 491 else if (uv == extra && !del) {
492 matches++;
493 if (uv != puv) {
494 if (final >= 0x80 && to_utf)
495 d = uv_to_utf8(d, final);
496 else
497 *d++ = (U8)final;
498 puv = final;
499 }
500 if (from_utf)
501 s += UTF8SKIP(s);
502 continue;
503 }
504 matches++; /* "none+1" is delete character */
505 if (from_utf)
506 s += UTF8SKIP(s);
a0ed51b3 507 }
79072805 508 }
509 else {
4757a243 510 while (s < send) {
511 if (from_utf) {
512 uv = swash_fetch(rv, s);
513 }
514 else {
515 U8 tmpbuf[2];
516 uv = *s++;
517 if (uv < 0x80)
518 tmpbuf[0] = uv;
519 else {
520 tmpbuf[0] = (( uv >> 6) | 0xc0);
521 tmpbuf[1] = (( uv & 0x3f) | 0x80);
a0ed51b3 522 }
4757a243 523 uv = swash_fetch(rv, tmpbuf);
a0ed51b3 524 }
4757a243 525 if (uv < none) {
526 matches++;
527 if (uv >= 0x80 && to_utf)
528 d = uv_to_utf8(d, uv);
529 else
530 *d++ = (U8)uv;
531 if (from_utf)
532 s += UTF8SKIP(s);
533 continue;
a0ed51b3 534 }
4757a243 535 else if (uv == none) { /* "none" is unmapped character */
536 if (from_utf) {
537 if (*s < 0x80)
538 *d++ = *s++;
539 else if (to_utf) {
540 int i;
541 for (i = UTF8SKIP(s); i; --i)
542 *d++ = *s++;
543 }
544 else {
545 I32 ulen;
546 *d++ = (U8)utf8_to_uv(s, &ulen);
547 s += ulen;
a0ed51b3 548 }
79072805 549 }
4757a243 550 else { /* must be to_utf only */
551 d = uv_to_utf8(d, s[-1]);
552 }
553 continue;
79072805 554 }
4757a243 555 else if (uv == extra && !del) {
556 matches++;
557 if (final >= 0x80 && to_utf)
558 d = uv_to_utf8(d, final);
559 else
560 *d++ = (U8)final;
561 if (from_utf)
562 s += UTF8SKIP(s);
563 continue;
564 }
565 matches++; /* "none+1" is delete character */
566 if (from_utf)
567 s += UTF8SKIP(s);
79072805 568 }
4757a243 569 }
570 if (dst)
571 sv_usepvn(sv, (char*)dst, d - dst);
572 else {
573 *d = '\0';
574 SvCUR_set(sv, d - (U8*)SvPVX(sv));
575 }
576 SvSETMAGIC(sv);
577
578 return matches;
579}
580
581I32
864dbfa3 582Perl_do_trans(pTHX_ SV *sv)
4757a243 583{
46124e9e 584 dTHR;
4757a243 585 STRLEN len;
586
587 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
cea2e8a9 588 Perl_croak(aTHX_ PL_no_modify);
4757a243 589
590 (void)SvPV(sv, len);
591 if (!len)
592 return 0;
593 if (!SvPOKp(sv))
594 (void)SvPV_force(sv, len);
2de7b02f 595 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
596 (void)SvPOK_only_UTF8(sv);
4757a243 597
cea2e8a9 598 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 599
600 switch (PL_op->op_private & 63) {
601 case 0:
602 return do_trans_CC_simple(sv);
603
604 case OPpTRANS_FROM_UTF:
605 return do_trans_UC_simple(sv);
606
607 case OPpTRANS_TO_UTF:
608 return do_trans_CU_simple(sv);
609
610 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF:
611 return do_trans_UU_simple(sv);
612
613 case OPpTRANS_IDENTICAL:
614 return do_trans_CC_count(sv);
615
616 case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL:
617 return do_trans_UC_trivial(sv);
618
619 case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
620 return do_trans_CU_trivial(sv);
621
622 case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL:
623 return do_trans_UU_count(sv);
624
625 default:
626 if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
627 return do_trans_UU_complex(sv); /* could be UC or CU too */
628 else
629 return do_trans_CC_complex(sv);
79072805 630 }
79072805 631}
632
633void
864dbfa3 634Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
79072805 635{
636 SV **oldmark = mark;
637 register I32 items = sp - mark;
79072805 638 register STRLEN len;
463ee0b2 639 STRLEN delimlen;
640 register char *delim = SvPV(del, delimlen);
641 STRLEN tmplen;
79072805 642
643 mark++;
644 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
07f14f54 645 (void)SvUPGRADE(sv, SVt_PV);
79072805 646 if (SvLEN(sv) < len + items) { /* current length is way too short */
647 while (items-- > 0) {
48c036b1 648 if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) {
463ee0b2 649 SvPV(*mark, tmplen);
650 len += tmplen;
79072805 651 }
652 mark++;
653 }
654 SvGROW(sv, len + 1); /* so try to pre-extend */
655
656 mark = oldmark;
db7c17d7 657 items = sp - mark;
79072805 658 ++mark;
659 }
660
463ee0b2 661 if (items-- > 0) {
8990e307 662 char *s;
663
8dbfaa5d 664 sv_setpv(sv, "");
665 if (*mark)
666 sv_catsv(sv, *mark);
463ee0b2 667 mark++;
668 }
79072805 669 else
670 sv_setpv(sv,"");
671 len = delimlen;
672 if (len) {
673 for (; items > 0; items--,mark++) {
674 sv_catpvn(sv,delim,len);
675 sv_catsv(sv,*mark);
676 }
677 }
678 else {
679 for (; items > 0; items--,mark++)
680 sv_catsv(sv,*mark);
681 }
682 SvSETMAGIC(sv);
683}
684
685void
864dbfa3 686Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 687{
46fc3d4c 688 STRLEN patlen;
689 char *pat = SvPV(*sarg, patlen);
690 bool do_taint = FALSE;
691
692 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
79072805 693 SvSETMAGIC(sv);
46fc3d4c 694 if (do_taint)
695 SvTAINTED_on(sv);
79072805 696}
697
4ebbc975 698/* XXX SvUTF8 support missing! */
81e118e0 699UV
700Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
701{
702 STRLEN srclen, len;
703 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
704 UV retnum = 0;
705
a50d7633 706 if (offset < 0)
81e118e0 707 return retnum;
a50d7633 708 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
709 Perl_croak(aTHX_ "Illegal number of bits in vec");
81e118e0 710 offset *= size; /* turn into bit offset */
711 len = (offset + size + 7) / 8; /* required number of bytes */
712 if (len > srclen) {
713 if (size <= 8)
714 retnum = 0;
715 else {
716 offset >>= 3; /* turn into byte offset */
717 if (size == 16) {
718 if (offset >= srclen)
719 retnum = 0;
720 else
628e1a40 721 retnum = (UV) s[offset] << 8;
81e118e0 722 }
723 else if (size == 32) {
724 if (offset >= srclen)
725 retnum = 0;
726 else if (offset + 1 >= srclen)
727 retnum =
728 ((UV) s[offset ] << 24);
729 else if (offset + 2 >= srclen)
730 retnum =
731 ((UV) s[offset ] << 24) +
732 ((UV) s[offset + 1] << 16);
733 else
734 retnum =
735 ((UV) s[offset ] << 24) +
736 ((UV) s[offset + 1] << 16) +
737 ( s[offset + 2] << 8);
738 }
d7d93a81 739#ifdef UV_IS_QUAD
c5a0f51a 740 else if (size == 64) {
741 dTHR;
742 if (ckWARN(WARN_PORTABLE))
743 Perl_warner(aTHX_ WARN_PORTABLE,
744 "Bit vector size > 32 non-portable");
745 if (offset >= srclen)
746 retnum = 0;
747 else if (offset + 1 >= srclen)
748 retnum =
749 (UV) s[offset ] << 56;
750 else if (offset + 2 >= srclen)
751 retnum =
752 ((UV) s[offset ] << 56) +
753 ((UV) s[offset + 1] << 48);
754 else if (offset + 3 >= srclen)
755 retnum =
756 ((UV) s[offset ] << 56) +
757 ((UV) s[offset + 1] << 48) +
758 ((UV) s[offset + 2] << 40);
759 else if (offset + 4 >= srclen)
760 retnum =
761 ((UV) s[offset ] << 56) +
762 ((UV) s[offset + 1] << 48) +
763 ((UV) s[offset + 2] << 40) +
764 ((UV) s[offset + 3] << 32);
765 else if (offset + 5 >= srclen)
766 retnum =
767 ((UV) s[offset ] << 56) +
768 ((UV) s[offset + 1] << 48) +
769 ((UV) s[offset + 2] << 40) +
770 ((UV) s[offset + 3] << 32) +
771 ( s[offset + 4] << 24);
772 else if (offset + 6 >= srclen)
773 retnum =
774 ((UV) s[offset ] << 56) +
775 ((UV) s[offset + 1] << 48) +
776 ((UV) s[offset + 2] << 40) +
777 ((UV) s[offset + 3] << 32) +
778 ((UV) s[offset + 4] << 24) +
779 ((UV) s[offset + 5] << 16);
780 else
781 retnum =
782 ((UV) s[offset ] << 56) +
783 ((UV) s[offset + 1] << 48) +
784 ((UV) s[offset + 2] << 40) +
785 ((UV) s[offset + 3] << 32) +
786 ((UV) s[offset + 4] << 24) +
787 ((UV) s[offset + 5] << 16) +
628e1a40 788 ( s[offset + 6] << 8);
c5a0f51a 789 }
790#endif
81e118e0 791 }
792 }
793 else if (size < 8)
794 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
795 else {
796 offset >>= 3; /* turn into byte offset */
797 if (size == 8)
798 retnum = s[offset];
799 else if (size == 16)
800 retnum =
628e1a40 801 ((UV) s[offset] << 8) +
81e118e0 802 s[offset + 1];
803 else if (size == 32)
804 retnum =
805 ((UV) s[offset ] << 24) +
806 ((UV) s[offset + 1] << 16) +
807 ( s[offset + 2] << 8) +
808 s[offset + 3];
d7d93a81 809#ifdef UV_IS_QUAD
c5a0f51a 810 else if (size == 64) {
811 dTHR;
812 if (ckWARN(WARN_PORTABLE))
813 Perl_warner(aTHX_ WARN_PORTABLE,
814 "Bit vector size > 32 non-portable");
815 retnum =
816 ((UV) s[offset ] << 56) +
817 ((UV) s[offset + 1] << 48) +
818 ((UV) s[offset + 2] << 40) +
819 ((UV) s[offset + 3] << 32) +
820 ((UV) s[offset + 4] << 24) +
821 ((UV) s[offset + 5] << 16) +
628e1a40 822 ( s[offset + 6] << 8) +
c5a0f51a 823 s[offset + 7];
824 }
825#endif
81e118e0 826 }
827
828 return retnum;
829}
830
4ebbc975 831/* XXX SvUTF8 support missing! */
79072805 832void
864dbfa3 833Perl_do_vecset(pTHX_ SV *sv)
79072805 834{
835 SV *targ = LvTARG(sv);
836 register I32 offset;
837 register I32 size;
8990e307 838 register unsigned char *s;
81e118e0 839 register UV lval;
79072805 840 I32 mask;
a0d0e21e 841 STRLEN targlen;
842 STRLEN len;
79072805 843
8990e307 844 if (!targ)
845 return;
a0d0e21e 846 s = (unsigned char*)SvPV_force(targ, targlen);
4ebbc975 847 (void)SvPOK_only(targ);
81e118e0 848 lval = SvUV(sv);
79072805 849 offset = LvTARGOFF(sv);
850 size = LvTARGLEN(sv);
a50d7633 851 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
852 Perl_croak(aTHX_ "Illegal number of bits in vec");
a0d0e21e 853
81e118e0 854 offset *= size; /* turn into bit offset */
855 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e 856 if (len > targlen) {
857 s = (unsigned char*)SvGROW(targ, len + 1);
858 (void)memzero(s + targlen, len - targlen + 1);
859 SvCUR_set(targ, len);
860 }
861
79072805 862 if (size < 8) {
863 mask = (1 << size) - 1;
864 size = offset & 7;
865 lval &= mask;
81e118e0 866 offset >>= 3; /* turn into byte offset */
79072805 867 s[offset] &= ~(mask << size);
868 s[offset] |= lval << size;
869 }
870 else {
81e118e0 871 offset >>= 3; /* turn into byte offset */
79072805 872 if (size == 8)
c5a0f51a 873 s[offset ] = lval & 0xff;
79072805 874 else if (size == 16) {
c5a0f51a 875 s[offset ] = (lval >> 8) & 0xff;
876 s[offset+1] = lval & 0xff;
79072805 877 }
878 else if (size == 32) {
c5a0f51a 879 s[offset ] = (lval >> 24) & 0xff;
880 s[offset+1] = (lval >> 16) & 0xff;
881 s[offset+2] = (lval >> 8) & 0xff;
882 s[offset+3] = lval & 0xff;
883 }
d7d93a81 884#ifdef UV_IS_QUAD
c5a0f51a 885 else if (size == 64) {
886 dTHR;
887 if (ckWARN(WARN_PORTABLE))
888 Perl_warner(aTHX_ WARN_PORTABLE,
889 "Bit vector size > 32 non-portable");
890 s[offset ] = (lval >> 56) & 0xff;
891 s[offset+1] = (lval >> 48) & 0xff;
892 s[offset+2] = (lval >> 40) & 0xff;
893 s[offset+3] = (lval >> 32) & 0xff;
894 s[offset+4] = (lval >> 24) & 0xff;
895 s[offset+5] = (lval >> 16) & 0xff;
896 s[offset+6] = (lval >> 8) & 0xff;
897 s[offset+7] = lval & 0xff;
79072805 898 }
dc1e3f56 899#endif
79072805 900 }
7bb043c3 901 SvSETMAGIC(targ);
79072805 902}
903
904void
864dbfa3 905Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 906{
463ee0b2 907 STRLEN len;
a0d0e21e 908 char *s;
c485e607 909 dTHR;
a0d0e21e 910
79072805 911 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e 912 register I32 i;
913 I32 max;
914 AV* av = (AV*)sv;
915 max = AvFILL(av);
916 for (i = 0; i <= max; i++) {
917 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 918 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 919 do_chop(astr, sv);
920 }
921 return;
79072805 922 }
aa854799 923 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 924 HV* hv = (HV*)sv;
925 HE* entry;
926 (void)hv_iterinit(hv);
927 /*SUPPRESS 560*/
155aba94 928 while ((entry = hv_iternext(hv)))
a0d0e21e 929 do_chop(astr,hv_iterval(hv,entry));
930 return;
79072805 931 }
aa854799 932 else if (SvREADONLY(sv))
cea2e8a9 933 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 934 s = SvPV(sv, len);
748a9306 935 if (len && !SvPOK(sv))
a0d0e21e 936 s = SvPV_force(sv, len);
7e2040f0 937 if (DO_UTF8(sv)) {
a0ed51b3 938 if (s && len) {
939 char *send = s + len;
940 char *start = s;
941 s = send - 1;
942 while ((*s & 0xc0) == 0x80)
943 --s;
0453d815 944 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
945 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
a0ed51b3 946 sv_setpvn(astr, s, send - s);
947 *s = '\0';
948 SvCUR_set(sv, s - start);
949 SvNIOK_off(sv);
7e2040f0 950 SvUTF8_on(astr);
a0ed51b3 951 }
952 else
953 sv_setpvn(astr, "", 0);
954 }
7e2040f0 955 else if (s && len) {
a0d0e21e 956 s += --len;
957 sv_setpvn(astr, s, 1);
958 *s = '\0';
959 SvCUR_set(sv, len);
2c19a612 960 SvUTF8_off(sv);
a0d0e21e 961 SvNIOK_off(sv);
79072805 962 }
963 else
a0d0e21e 964 sv_setpvn(astr, "", 0);
965 SvSETMAGIC(sv);
7e2040f0 966}
a0d0e21e 967
968I32
864dbfa3 969Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 970{
aeea060c 971 dTHR;
c07a80fd 972 register I32 count;
a0d0e21e 973 STRLEN len;
974 char *s;
c07a80fd 975
3280af22 976 if (RsSNARF(PL_rs))
c07a80fd 977 return 0;
4c5a6083 978 if (RsRECORD(PL_rs))
979 return 0;
c07a80fd 980 count = 0;
a0d0e21e 981 if (SvTYPE(sv) == SVt_PVAV) {
982 register I32 i;
983 I32 max;
984 AV* av = (AV*)sv;
985 max = AvFILL(av);
986 for (i = 0; i <= max; i++) {
987 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 988 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 989 count += do_chomp(sv);
990 }
991 return count;
992 }
aa854799 993 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 994 HV* hv = (HV*)sv;
995 HE* entry;
996 (void)hv_iterinit(hv);
997 /*SUPPRESS 560*/
155aba94 998 while ((entry = hv_iternext(hv)))
a0d0e21e 999 count += do_chomp(hv_iterval(hv,entry));
1000 return count;
1001 }
aa854799 1002 else if (SvREADONLY(sv))
cea2e8a9 1003 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 1004 s = SvPV(sv, len);
1005 if (len && !SvPOKp(sv))
1006 s = SvPV_force(sv, len);
1007 if (s && len) {
1008 s += --len;
3280af22 1009 if (RsPARA(PL_rs)) {
a0d0e21e 1010 if (*s != '\n')
1011 goto nope;
1012 ++count;
1013 while (len && s[-1] == '\n') {
1014 --len;
1015 --s;
1016 ++count;
1017 }
1018 }
a0d0e21e 1019 else {
c07a80fd 1020 STRLEN rslen;
3280af22 1021 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 1022 if (rslen == 1) {
1023 if (*s != *rsptr)
1024 goto nope;
1025 ++count;
1026 }
1027 else {
8c2cee6f 1028 if (len < rslen - 1)
c07a80fd 1029 goto nope;
1030 len -= rslen - 1;
1031 s -= rslen - 1;
36477c24 1032 if (memNE(s, rsptr, rslen))
c07a80fd 1033 goto nope;
1034 count += rslen;
1035 }
a0d0e21e 1036 }
a0d0e21e 1037 *s = '\0';
1038 SvCUR_set(sv, len);
1039 SvNIOK_off(sv);
1040 }
1041 nope:
1042 SvSETMAGIC(sv);
1043 return count;
1044}
79072805 1045
1046void
864dbfa3 1047Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805 1048{
aeea060c 1049 dTHR; /* just for taint */
79072805 1050#ifdef LIBERAL
1051 register long *dl;
1052 register long *ll;
1053 register long *rl;
1054#endif
1055 register char *dc;
463ee0b2 1056 STRLEN leftlen;
1057 STRLEN rightlen;
7a4c00b4 1058 register char *lc;
1059 register char *rc;
79072805 1060 register I32 len;
a0d0e21e 1061 I32 lensave;
7a4c00b4 1062 char *lsave;
1063 char *rsave;
0c57e439 1064 bool left_utf = DO_UTF8(left);
1065 bool right_utf = DO_UTF8(right);
1066
1067 if (left_utf && !right_utf)
1068 sv_utf8_upgrade(right);
1069 if (!left_utf && right_utf)
1070 sv_utf8_upgrade(left);
79072805 1071
1fbd88dc 1072 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1073 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 1074 lsave = lc = SvPV(left, leftlen);
1075 rsave = rc = SvPV(right, rightlen);
93a17b20 1076 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 1077 lensave = len;
7a4c00b4 1078 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d 1079 STRLEN n_a;
1080 dc = SvPV_force(sv, n_a);
ff68c719 1081 if (SvCUR(sv) < len) {
1082 dc = SvGROW(sv, len + 1);
1083 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1084 }
1085 }
1086 else {
1087 I32 needlen = ((optype == OP_BIT_AND)
1088 ? len : (leftlen > rightlen ? leftlen : rightlen));
1089 Newz(801, dc, needlen + 1, char);
1090 (void)sv_usepvn(sv, dc, needlen);
1091 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 1092 }
a0d0e21e 1093 SvCUR_set(sv, len);
1094 (void)SvPOK_only(sv);
0c57e439 1095 if (left_utf || right_utf) {
1096 UV duc, luc, ruc;
1097 STRLEN lulen = leftlen;
1098 STRLEN rulen = rightlen;
1099 STRLEN dulen = 0;
1100 I32 ulen;
1101
6b7c0e6e 1102 if (optype != OP_BIT_AND)
1103 dc = SvGROW(sv, leftlen+rightlen+1);
1104
0c57e439 1105 switch (optype) {
1106 case OP_BIT_AND:
1107 while (lulen && rulen) {
1108 luc = utf8_to_uv((U8*)lc, &ulen);
1109 lc += ulen;
1110 lulen -= ulen;
1111 ruc = utf8_to_uv((U8*)rc, &ulen);
1112 rc += ulen;
1113 rulen -= ulen;
1114 duc = luc & ruc;
1115 dc = (char*)uv_to_utf8((U8*)dc, duc);
1116 }
1117 dulen = dc - SvPVX(sv);
1118 SvCUR_set(sv, dulen);
1119 break;
1120 case OP_BIT_XOR:
1121 while (lulen && rulen) {
1122 luc = utf8_to_uv((U8*)lc, &ulen);
1123 lc += ulen;
1124 lulen -= ulen;
1125 ruc = utf8_to_uv((U8*)rc, &ulen);
1126 rc += ulen;
1127 rulen -= ulen;
1128 duc = luc ^ ruc;
1129 dc = (char*)uv_to_utf8((U8*)dc, duc);
1130 }
1131 goto mop_up_utf;
1132 case OP_BIT_OR:
1133 while (lulen && rulen) {
1134 luc = utf8_to_uv((U8*)lc, &ulen);
1135 lc += ulen;
1136 lulen -= ulen;
1137 ruc = utf8_to_uv((U8*)rc, &ulen);
1138 rc += ulen;
1139 rulen -= ulen;
1140 duc = luc | ruc;
1141 dc = (char*)uv_to_utf8((U8*)dc, duc);
1142 }
1143 mop_up_utf:
1144 dulen = dc - SvPVX(sv);
1145 SvCUR_set(sv, dulen);
1146 if (rulen)
1147 sv_catpvn(sv, rc, rulen);
1148 else if (lulen)
1149 sv_catpvn(sv, lc, lulen);
1150 else
1151 *SvEND(sv) = '\0';
1152 break;
1153 }
1154 SvUTF8_on(sv);
1155 goto finish;
1156 }
1157 else
79072805 1158#ifdef LIBERAL
1159 if (len >= sizeof(long)*4 &&
1160 !((long)dc % sizeof(long)) &&
1161 !((long)lc % sizeof(long)) &&
1162 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1163 {
1164 I32 remainder = len % (sizeof(long)*4);
1165 len /= (sizeof(long)*4);
1166
1167 dl = (long*)dc;
1168 ll = (long*)lc;
1169 rl = (long*)rc;
1170
1171 switch (optype) {
1172 case OP_BIT_AND:
1173 while (len--) {
1174 *dl++ = *ll++ & *rl++;
1175 *dl++ = *ll++ & *rl++;
1176 *dl++ = *ll++ & *rl++;
1177 *dl++ = *ll++ & *rl++;
1178 }
1179 break;
a0d0e21e 1180 case OP_BIT_XOR:
79072805 1181 while (len--) {
1182 *dl++ = *ll++ ^ *rl++;
1183 *dl++ = *ll++ ^ *rl++;
1184 *dl++ = *ll++ ^ *rl++;
1185 *dl++ = *ll++ ^ *rl++;
1186 }
1187 break;
1188 case OP_BIT_OR:
1189 while (len--) {
1190 *dl++ = *ll++ | *rl++;
1191 *dl++ = *ll++ | *rl++;
1192 *dl++ = *ll++ | *rl++;
1193 *dl++ = *ll++ | *rl++;
1194 }
1195 }
1196
1197 dc = (char*)dl;
1198 lc = (char*)ll;
1199 rc = (char*)rl;
1200
1201 len = remainder;
1202 }
1203#endif
a0d0e21e 1204 {
a0d0e21e 1205 switch (optype) {
1206 case OP_BIT_AND:
1207 while (len--)
1208 *dc++ = *lc++ & *rc++;
1209 break;
1210 case OP_BIT_XOR:
1211 while (len--)
1212 *dc++ = *lc++ ^ *rc++;
1213 goto mop_up;
1214 case OP_BIT_OR:
1215 while (len--)
1216 *dc++ = *lc++ | *rc++;
1217 mop_up:
1218 len = lensave;
1219 if (rightlen > len)
1220 sv_catpvn(sv, rsave + len, rightlen - len);
1221 else if (leftlen > len)
1222 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4 1223 else
1224 *SvEND(sv) = '\0';
a0d0e21e 1225 break;
1226 }
79072805 1227 }
0c57e439 1228finish:
fb73857a 1229 SvTAINT(sv);
79072805 1230}
463ee0b2 1231
1232OP *
cea2e8a9 1233Perl_do_kv(pTHX)
463ee0b2 1234{
4e35701f 1235 djSP;
463ee0b2 1236 HV *hv = (HV*)POPs;
800e9ae0 1237 HV *keys;
463ee0b2 1238 register HE *entry;
463ee0b2 1239 SV *tmpstr;
54310121 1240 I32 gimme = GIMME_V;
533c011a 1241 I32 dokeys = (PL_op->op_type == OP_KEYS);
1242 I32 dovalues = (PL_op->op_type == OP_VALUES);
c750a3ec 1243 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1244
533c011a 1245 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1246 dokeys = dovalues = TRUE;
463ee0b2 1247
85581909 1248 if (!hv) {
533c011a 1249 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1250 dTARGET; /* make sure to clear its target here */
1251 if (SvTYPE(TARG) == SVt_PVLV)
1252 LvTARG(TARG) = Nullsv;
1253 PUSHs(TARG);
1254 }
463ee0b2 1255 RETURN;
85581909 1256 }
748a9306 1257
800e9ae0 1258 keys = realhv ? hv : avhv_keys((AV*)hv);
1259 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1260
54310121 1261 if (gimme == G_VOID)
aa689395 1262 RETURN;
1263
54310121 1264 if (gimme == G_SCALAR) {
6ee623d5 1265 IV i;
463ee0b2 1266 dTARGET;
1267
533c011a 1268 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1269 if (SvTYPE(TARG) < SVt_PVLV) {
1270 sv_upgrade(TARG, SVt_PVLV);
1271 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1272 }
1273 LvTYPE(TARG) = 'k';
800e9ae0 1274 if (LvTARG(TARG) != (SV*)keys) {
6ff81951 1275 if (LvTARG(TARG))
1276 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1277 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1278 }
85581909 1279 PUSHs(TARG);
1280 RETURN;
1281 }
1282
33c27489 1283 if (! SvTIED_mg((SV*)keys, 'P'))
800e9ae0 1284 i = HvKEYS(keys);
463ee0b2 1285 else {
1286 i = 0;
463ee0b2 1287 /*SUPPRESS 560*/
800e9ae0 1288 while (hv_iternext(keys)) i++;
463ee0b2 1289 }
1290 PUSHi( i );
1291 RETURN;
1292 }
1293
8ed4b672 1294 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1295
463ee0b2 1296 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1297 while ((entry = hv_iternext(keys))) {
463ee0b2 1298 SPAGAIN;
8c2cee6f 1299 if (dokeys)
1300 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
463ee0b2 1301 if (dovalues) {
463ee0b2 1302 PUTBACK;
b6429b1b 1303 tmpstr = realhv ?
1304 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
cea2e8a9 1305 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1306 (unsigned long)HeHASH(entry),
800e9ae0 1307 HvMAX(keys)+1,
1308 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1309 SPAGAIN;
46fc3d4c 1310 XPUSHs(tmpstr);
463ee0b2 1311 }
1312 PUTBACK;
1313 }
1314 return NORMAL;
1315}
4e35701f 1316