This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
add op_sibling_splice() fn and make core use it
[perl5.git] / doop.c
... / ...
CommitLineData
1/* doop.c
2 *
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2009 by Larry Wall and others
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
11/*
12 * 'So that was the job I felt I had to do when I started,' thought Sam.
13 *
14 * [p.934 of _The Lord of the Rings_, VI/iii: "Mount Doom"]
15 */
16
17/* This file contains some common functions needed to carry out certain
18 * ops. For example, both pp_sprintf() and pp_prtf() call the function
19 * do_printf() found in this file.
20 */
21
22#include "EXTERN.h"
23#define PERL_IN_DOOP_C
24#include "perl.h"
25
26#ifndef PERL_MICRO
27#include <signal.h>
28#endif
29
30STATIC I32
31S_do_trans_simple(pTHX_ SV * const sv)
32{
33 I32 matches = 0;
34 STRLEN len;
35 U8 *s = (U8*)SvPV_nomg(sv,len);
36 U8 * const send = s+len;
37 const short * const tbl = (short*)cPVOP->op_pv;
38
39 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE;
40
41 if (!tbl)
42 Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
43
44 /* First, take care of non-UTF-8 input strings, because they're easy */
45 if (!SvUTF8(sv)) {
46 while (s < send) {
47 const I32 ch = tbl[*s];
48 if (ch >= 0) {
49 matches++;
50 *s = (U8)ch;
51 }
52 s++;
53 }
54 SvSETMAGIC(sv);
55 }
56 else {
57 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
58 U8 *d;
59 U8 *dstart;
60
61 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
62 if (grows)
63 Newx(d, len*2+1, U8);
64 else
65 d = s;
66 dstart = d;
67 while (s < send) {
68 STRLEN ulen;
69 I32 ch;
70
71 /* Need to check this, otherwise 128..255 won't match */
72 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
73 if (c < 0x100 && (ch = tbl[c]) >= 0) {
74 matches++;
75 d = uvchr_to_utf8(d, ch);
76 s += ulen;
77 }
78 else { /* No match -> copy */
79 Move(s, d, ulen, U8);
80 d += ulen;
81 s += ulen;
82 }
83 }
84 if (grows) {
85 sv_setpvn(sv, (char*)dstart, d - dstart);
86 Safefree(dstart);
87 }
88 else {
89 *d = '\0';
90 SvCUR_set(sv, d - dstart);
91 }
92 SvUTF8_on(sv);
93 SvSETMAGIC(sv);
94 }
95 return matches;
96}
97
98STATIC I32
99S_do_trans_count(pTHX_ SV * const sv)
100{
101 STRLEN len;
102 const U8 *s = (const U8*)SvPV_nomg_const(sv, len);
103 const U8 * const send = s + len;
104 I32 matches = 0;
105 const short * const tbl = (short*)cPVOP->op_pv;
106
107 PERL_ARGS_ASSERT_DO_TRANS_COUNT;
108
109 if (!tbl)
110 Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
111
112 if (!SvUTF8(sv)) {
113 while (s < send) {
114 if (tbl[*s++] >= 0)
115 matches++;
116 }
117 }
118 else {
119 const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
120 while (s < send) {
121 STRLEN ulen;
122 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
123 if (c < 0x100) {
124 if (tbl[c] >= 0)
125 matches++;
126 } else if (complement)
127 matches++;
128 s += ulen;
129 }
130 }
131
132 return matches;
133}
134
135STATIC I32
136S_do_trans_complex(pTHX_ SV * const sv)
137{
138 STRLEN len;
139 U8 *s = (U8*)SvPV_nomg(sv, len);
140 U8 * const send = s+len;
141 I32 matches = 0;
142 const short * const tbl = (short*)cPVOP->op_pv;
143
144 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX;
145
146 if (!tbl)
147 Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
148
149 if (!SvUTF8(sv)) {
150 U8 *d = s;
151 U8 * const dstart = d;
152
153 if (PL_op->op_private & OPpTRANS_SQUASH) {
154 const U8* p = send;
155 while (s < send) {
156 const I32 ch = tbl[*s];
157 if (ch >= 0) {
158 *d = (U8)ch;
159 matches++;
160 if (p != d - 1 || *p != *d)
161 p = d++;
162 }
163 else if (ch == -1) /* -1 is unmapped character */
164 *d++ = *s;
165 else if (ch == -2) /* -2 is delete character */
166 matches++;
167 s++;
168 }
169 }
170 else {
171 while (s < send) {
172 const I32 ch = tbl[*s];
173 if (ch >= 0) {
174 matches++;
175 *d++ = (U8)ch;
176 }
177 else if (ch == -1) /* -1 is unmapped character */
178 *d++ = *s;
179 else if (ch == -2) /* -2 is delete character */
180 matches++;
181 s++;
182 }
183 }
184 *d = '\0';
185 SvCUR_set(sv, d - dstart);
186 }
187 else { /* is utf8 */
188 const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
189 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
190 const I32 del = PL_op->op_private & OPpTRANS_DELETE;
191 U8 *d;
192 U8 *dstart;
193 STRLEN rlen = 0;
194
195 if (grows)
196 Newx(d, len*2+1, U8);
197 else
198 d = s;
199 dstart = d;
200 if (complement && !del)
201 rlen = tbl[0x100];
202
203 if (PL_op->op_private & OPpTRANS_SQUASH) {
204 UV pch = 0xfeedface;
205 while (s < send) {
206 STRLEN len;
207 const UV comp = utf8n_to_uvchr(s, send - s, &len,
208 UTF8_ALLOW_DEFAULT);
209 I32 ch;
210
211 if (comp > 0xff) {
212 if (!complement) {
213 Move(s, d, len, U8);
214 d += len;
215 }
216 else {
217 matches++;
218 if (!del) {
219 ch = (rlen == 0) ? (I32)comp :
220 (comp - 0x100 < rlen) ?
221 tbl[comp+1] : tbl[0x100+rlen];
222 if ((UV)ch != pch) {
223 d = uvchr_to_utf8(d, ch);
224 pch = (UV)ch;
225 }
226 s += len;
227 continue;
228 }
229 }
230 }
231 else if ((ch = tbl[comp]) >= 0) {
232 matches++;
233 if ((UV)ch != pch) {
234 d = uvchr_to_utf8(d, ch);
235 pch = (UV)ch;
236 }
237 s += len;
238 continue;
239 }
240 else if (ch == -1) { /* -1 is unmapped character */
241 Move(s, d, len, U8);
242 d += len;
243 }
244 else if (ch == -2) /* -2 is delete character */
245 matches++;
246 s += len;
247 pch = 0xfeedface;
248 }
249 }
250 else {
251 while (s < send) {
252 STRLEN len;
253 const UV comp = utf8n_to_uvchr(s, send - s, &len,
254 UTF8_ALLOW_DEFAULT);
255 I32 ch;
256 if (comp > 0xff) {
257 if (!complement) {
258 Move(s, d, len, U8);
259 d += len;
260 }
261 else {
262 matches++;
263 if (!del) {
264 if (comp - 0x100 < rlen)
265 d = uvchr_to_utf8(d, tbl[comp+1]);
266 else
267 d = uvchr_to_utf8(d, tbl[0x100+rlen]);
268 }
269 }
270 }
271 else if ((ch = tbl[comp]) >= 0) {
272 d = uvchr_to_utf8(d, ch);
273 matches++;
274 }
275 else if (ch == -1) { /* -1 is unmapped character */
276 Move(s, d, len, U8);
277 d += len;
278 }
279 else if (ch == -2) /* -2 is delete character */
280 matches++;
281 s += len;
282 }
283 }
284 if (grows) {
285 sv_setpvn(sv, (char*)dstart, d - dstart);
286 Safefree(dstart);
287 }
288 else {
289 *d = '\0';
290 SvCUR_set(sv, d - dstart);
291 }
292 SvUTF8_on(sv);
293 }
294 SvSETMAGIC(sv);
295 return matches;
296}
297
298STATIC I32
299S_do_trans_simple_utf8(pTHX_ SV * const sv)
300{
301 U8 *s;
302 U8 *send;
303 U8 *d;
304 U8 *start;
305 U8 *dstart, *dend;
306 I32 matches = 0;
307 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
308 STRLEN len;
309 SV* const rv =
310#ifdef USE_ITHREADS
311 PAD_SVl(cPADOP->op_padix);
312#else
313 MUTABLE_SV(cSVOP->op_sv);
314#endif
315 HV* const hv = MUTABLE_HV(SvRV(rv));
316 SV* const * svp = hv_fetchs(hv, "NONE", FALSE);
317 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
318 const UV extra = none + 1;
319 UV final = 0;
320 U8 hibit = 0;
321
322 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8;
323
324 s = (U8*)SvPV_nomg(sv, len);
325 if (!SvUTF8(sv)) {
326 const U8 *t = s;
327 const U8 * const e = s + len;
328 while (t < e) {
329 const U8 ch = *t++;
330 hibit = !NATIVE_BYTE_IS_INVARIANT(ch);
331 if (hibit) {
332 s = bytes_to_utf8(s, &len);
333 break;
334 }
335 }
336 }
337 send = s + len;
338 start = s;
339
340 svp = hv_fetchs(hv, "FINAL", FALSE);
341 if (svp)
342 final = SvUV(*svp);
343
344 if (grows) {
345 /* d needs to be bigger than s, in case e.g. upgrading is required */
346 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
347 dend = d + len * 3;
348 dstart = d;
349 }
350 else {
351 dstart = d = s;
352 dend = d + len;
353 }
354
355 while (s < send) {
356 const UV uv = swash_fetch(rv, s, TRUE);
357 if (uv < none) {
358 s += UTF8SKIP(s);
359 matches++;
360 d = uvchr_to_utf8(d, uv);
361 }
362 else if (uv == none) {
363 const int i = UTF8SKIP(s);
364 Move(s, d, i, U8);
365 d += i;
366 s += i;
367 }
368 else if (uv == extra) {
369 s += UTF8SKIP(s);
370 matches++;
371 d = uvchr_to_utf8(d, final);
372 }
373 else
374 s += UTF8SKIP(s);
375
376 if (d > dend) {
377 const STRLEN clen = d - dstart;
378 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
379 if (!grows)
380 Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
381 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
382 d = dstart + clen;
383 dend = dstart + nlen;
384 }
385 }
386 if (grows || hibit) {
387 sv_setpvn(sv, (char*)dstart, d - dstart);
388 Safefree(dstart);
389 if (grows && hibit)
390 Safefree(start);
391 }
392 else {
393 *d = '\0';
394 SvCUR_set(sv, d - dstart);
395 }
396 SvSETMAGIC(sv);
397 SvUTF8_on(sv);
398
399 return matches;
400}
401
402STATIC I32
403S_do_trans_count_utf8(pTHX_ SV * const sv)
404{
405 const U8 *s;
406 const U8 *start = NULL;
407 const U8 *send;
408 I32 matches = 0;
409 STRLEN len;
410 SV* const rv =
411#ifdef USE_ITHREADS
412 PAD_SVl(cPADOP->op_padix);
413#else
414 MUTABLE_SV(cSVOP->op_sv);
415#endif
416 HV* const hv = MUTABLE_HV(SvRV(rv));
417 SV* const * const svp = hv_fetchs(hv, "NONE", FALSE);
418 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
419 const UV extra = none + 1;
420 U8 hibit = 0;
421
422 PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8;
423
424 s = (const U8*)SvPV_nomg_const(sv, len);
425 if (!SvUTF8(sv)) {
426 const U8 *t = s;
427 const U8 * const e = s + len;
428 while (t < e) {
429 const U8 ch = *t++;
430 hibit = !NATIVE_BYTE_IS_INVARIANT(ch);
431 if (hibit) {
432 start = s = bytes_to_utf8(s, &len);
433 break;
434 }
435 }
436 }
437 send = s + len;
438
439 while (s < send) {
440 const UV uv = swash_fetch(rv, s, TRUE);
441 if (uv < none || uv == extra)
442 matches++;
443 s += UTF8SKIP(s);
444 }
445 if (hibit)
446 Safefree(start);
447
448 return matches;
449}
450
451STATIC I32
452S_do_trans_complex_utf8(pTHX_ SV * const sv)
453{
454 U8 *start, *send;
455 U8 *d;
456 I32 matches = 0;
457 const I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
458 const I32 del = PL_op->op_private & OPpTRANS_DELETE;
459 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
460 SV* const rv =
461#ifdef USE_ITHREADS
462 PAD_SVl(cPADOP->op_padix);
463#else
464 MUTABLE_SV(cSVOP->op_sv);
465#endif
466 HV * const hv = MUTABLE_HV(SvRV(rv));
467 SV * const *svp = hv_fetchs(hv, "NONE", FALSE);
468 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
469 const UV extra = none + 1;
470 UV final = 0;
471 bool havefinal = FALSE;
472 STRLEN len;
473 U8 *dstart, *dend;
474 U8 hibit = 0;
475 U8 *s = (U8*)SvPV_nomg(sv, len);
476
477 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8;
478
479 if (!SvUTF8(sv)) {
480 const U8 *t = s;
481 const U8 * const e = s + len;
482 while (t < e) {
483 const U8 ch = *t++;
484 hibit = !NATIVE_BYTE_IS_INVARIANT(ch);
485 if (hibit) {
486 s = bytes_to_utf8(s, &len);
487 break;
488 }
489 }
490 }
491 send = s + len;
492 start = s;
493
494 svp = hv_fetchs(hv, "FINAL", FALSE);
495 if (svp) {
496 final = SvUV(*svp);
497 havefinal = TRUE;
498 }
499
500 if (grows) {
501 /* d needs to be bigger than s, in case e.g. upgrading is required */
502 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
503 dend = d + len * 3;
504 dstart = d;
505 }
506 else {
507 dstart = d = s;
508 dend = d + len;
509 }
510
511 if (squash) {
512 UV puv = 0xfeedface;
513 while (s < send) {
514 UV uv = swash_fetch(rv, s, TRUE);
515
516 if (d > dend) {
517 const STRLEN clen = d - dstart;
518 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
519 if (!grows)
520 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
521 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
522 d = dstart + clen;
523 dend = dstart + nlen;
524 }
525 if (uv < none) {
526 matches++;
527 s += UTF8SKIP(s);
528 if (uv != puv) {
529 d = uvchr_to_utf8(d, uv);
530 puv = uv;
531 }
532 continue;
533 }
534 else if (uv == none) { /* "none" is unmapped character */
535 const int i = UTF8SKIP(s);
536 Move(s, d, i, U8);
537 d += i;
538 s += i;
539 puv = 0xfeedface;
540 continue;
541 }
542 else if (uv == extra && !del) {
543 matches++;
544 if (havefinal) {
545 s += UTF8SKIP(s);
546 if (puv != final) {
547 d = uvchr_to_utf8(d, final);
548 puv = final;
549 }
550 }
551 else {
552 STRLEN len;
553 uv = utf8n_to_uvchr(s, send - s, &len, UTF8_ALLOW_DEFAULT);
554 if (uv != puv) {
555 Move(s, d, len, U8);
556 d += len;
557 puv = uv;
558 }
559 s += len;
560 }
561 continue;
562 }
563 matches++; /* "none+1" is delete character */
564 s += UTF8SKIP(s);
565 }
566 }
567 else {
568 while (s < send) {
569 const UV uv = swash_fetch(rv, s, TRUE);
570 if (d > dend) {
571 const STRLEN clen = d - dstart;
572 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
573 if (!grows)
574 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
575 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
576 d = dstart + clen;
577 dend = dstart + nlen;
578 }
579 if (uv < none) {
580 matches++;
581 s += UTF8SKIP(s);
582 d = uvchr_to_utf8(d, uv);
583 continue;
584 }
585 else if (uv == none) { /* "none" is unmapped character */
586 const int i = UTF8SKIP(s);
587 Move(s, d, i, U8);
588 d += i;
589 s += i;
590 continue;
591 }
592 else if (uv == extra && !del) {
593 matches++;
594 s += UTF8SKIP(s);
595 d = uvchr_to_utf8(d, final);
596 continue;
597 }
598 matches++; /* "none+1" is delete character */
599 s += UTF8SKIP(s);
600 }
601 }
602 if (grows || hibit) {
603 sv_setpvn(sv, (char*)dstart, d - dstart);
604 Safefree(dstart);
605 if (grows && hibit)
606 Safefree(start);
607 }
608 else {
609 *d = '\0';
610 SvCUR_set(sv, d - dstart);
611 }
612 SvUTF8_on(sv);
613 SvSETMAGIC(sv);
614
615 return matches;
616}
617
618I32
619Perl_do_trans(pTHX_ SV *sv)
620{
621 STRLEN len;
622 const I32 hasutf = (PL_op->op_private &
623 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
624
625 PERL_ARGS_ASSERT_DO_TRANS;
626
627 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) {
628 Perl_croak_no_modify();
629 }
630 (void)SvPV_const(sv, len);
631 if (!len)
632 return 0;
633 if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) {
634 if (!SvPOKp(sv) || SvTHINKFIRST(sv))
635 (void)SvPV_force_nomg(sv, len);
636 (void)SvPOK_only_UTF8(sv);
637 }
638
639 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
640
641 switch (PL_op->op_private & ~hasutf & (
642 OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL|
643 OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) {
644 case 0:
645 if (hasutf)
646 return do_trans_simple_utf8(sv);
647 else
648 return do_trans_simple(sv);
649
650 case OPpTRANS_IDENTICAL:
651 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
652 if (hasutf)
653 return do_trans_count_utf8(sv);
654 else
655 return do_trans_count(sv);
656
657 default:
658 if (hasutf)
659 return do_trans_complex_utf8(sv);
660 else
661 return do_trans_complex(sv);
662 }
663}
664
665void
666Perl_do_join(pTHX_ SV *sv, SV *delim, SV **mark, SV **sp)
667{
668 SV ** const oldmark = mark;
669 I32 items = sp - mark;
670 STRLEN len;
671 STRLEN delimlen;
672
673 PERL_ARGS_ASSERT_DO_JOIN;
674
675 (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */
676 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
677
678 mark++;
679 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
680 SvUPGRADE(sv, SVt_PV);
681 if (SvLEN(sv) < len + items) { /* current length is way too short */
682 while (items-- > 0) {
683 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
684 STRLEN tmplen;
685 SvPV_const(*mark, tmplen);
686 len += tmplen;
687 }
688 mark++;
689 }
690 SvGROW(sv, len + 1); /* so try to pre-extend */
691
692 mark = oldmark;
693 items = sp - mark;
694 ++mark;
695 }
696
697 sv_setpvs(sv, "");
698 /* sv_setpv retains old UTF8ness [perl #24846] */
699 SvUTF8_off(sv);
700
701 if (TAINTING_get && SvMAGICAL(sv))
702 SvTAINTED_off(sv);
703
704 if (items-- > 0) {
705 if (*mark)
706 sv_catsv(sv, *mark);
707 mark++;
708 }
709
710 if (delimlen) {
711 for (; items > 0; items--,mark++) {
712 sv_catsv_nomg(sv,delim);
713 sv_catsv(sv,*mark);
714 }
715 }
716 else {
717 for (; items > 0; items--,mark++)
718 sv_catsv(sv,*mark);
719 }
720 SvSETMAGIC(sv);
721}
722
723void
724Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
725{
726 STRLEN patlen;
727 const char * const pat = SvPV_const(*sarg, patlen);
728 bool do_taint = FALSE;
729
730 PERL_ARGS_ASSERT_DO_SPRINTF;
731
732 if (SvTAINTED(*sarg))
733 TAINT_PROPER(
734 (PL_op && PL_op->op_type < OP_max)
735 ? (PL_op->op_type == OP_PRTF)
736 ? "printf"
737 : PL_op_name[PL_op->op_type]
738 : "(unknown)"
739 );
740 SvUTF8_off(sv);
741 if (DO_UTF8(*sarg))
742 SvUTF8_on(sv);
743 sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint);
744 SvSETMAGIC(sv);
745 if (do_taint)
746 SvTAINTED_on(sv);
747}
748
749/* currently converts input to bytes if possible, but doesn't sweat failure */
750UV
751Perl_do_vecget(pTHX_ SV *sv, SSize_t offset, int size)
752{
753 STRLEN srclen, len, uoffset, bitoffs = 0;
754 const I32 svpv_flags = ((PL_op->op_flags & OPf_MOD || LVRET)
755 ? SV_UNDEF_RETURNS_NULL : 0);
756 unsigned char *s = (unsigned char *)
757 SvPV_flags(sv, srclen, (svpv_flags|SV_GMAGIC));
758 UV retnum = 0;
759
760 if (!s) {
761 s = (unsigned char *)"";
762 }
763
764 PERL_ARGS_ASSERT_DO_VECGET;
765
766 if (offset < 0)
767 return 0;
768 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
769 Perl_croak(aTHX_ "Illegal number of bits in vec");
770
771 if (SvUTF8(sv)) {
772 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
773 /* PVX may have changed */
774 s = (unsigned char *) SvPV_flags(sv, srclen, svpv_flags);
775 }
776
777 if (size < 8) {
778 bitoffs = ((offset%8)*size)%8;
779 uoffset = offset/(8/size);
780 }
781 else if (size > 8)
782 uoffset = offset*(size/8);
783 else
784 uoffset = offset;
785
786 len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */
787 if (len > srclen) {
788 if (size <= 8)
789 retnum = 0;
790 else {
791 if (size == 16) {
792 if (uoffset >= srclen)
793 retnum = 0;
794 else
795 retnum = (UV) s[uoffset] << 8;
796 }
797 else if (size == 32) {
798 if (uoffset >= srclen)
799 retnum = 0;
800 else if (uoffset + 1 >= srclen)
801 retnum =
802 ((UV) s[uoffset ] << 24);
803 else if (uoffset + 2 >= srclen)
804 retnum =
805 ((UV) s[uoffset ] << 24) +
806 ((UV) s[uoffset + 1] << 16);
807 else
808 retnum =
809 ((UV) s[uoffset ] << 24) +
810 ((UV) s[uoffset + 1] << 16) +
811 ( s[uoffset + 2] << 8);
812 }
813#ifdef UV_IS_QUAD
814 else if (size == 64) {
815 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
816 "Bit vector size > 32 non-portable");
817 if (uoffset >= srclen)
818 retnum = 0;
819 else if (uoffset + 1 >= srclen)
820 retnum =
821 (UV) s[uoffset ] << 56;
822 else if (uoffset + 2 >= srclen)
823 retnum =
824 ((UV) s[uoffset ] << 56) +
825 ((UV) s[uoffset + 1] << 48);
826 else if (uoffset + 3 >= srclen)
827 retnum =
828 ((UV) s[uoffset ] << 56) +
829 ((UV) s[uoffset + 1] << 48) +
830 ((UV) s[uoffset + 2] << 40);
831 else if (uoffset + 4 >= srclen)
832 retnum =
833 ((UV) s[uoffset ] << 56) +
834 ((UV) s[uoffset + 1] << 48) +
835 ((UV) s[uoffset + 2] << 40) +
836 ((UV) s[uoffset + 3] << 32);
837 else if (uoffset + 5 >= srclen)
838 retnum =
839 ((UV) s[uoffset ] << 56) +
840 ((UV) s[uoffset + 1] << 48) +
841 ((UV) s[uoffset + 2] << 40) +
842 ((UV) s[uoffset + 3] << 32) +
843 ((UV) s[uoffset + 4] << 24);
844 else if (uoffset + 6 >= srclen)
845 retnum =
846 ((UV) s[uoffset ] << 56) +
847 ((UV) s[uoffset + 1] << 48) +
848 ((UV) s[uoffset + 2] << 40) +
849 ((UV) s[uoffset + 3] << 32) +
850 ((UV) s[uoffset + 4] << 24) +
851 ((UV) s[uoffset + 5] << 16);
852 else
853 retnum =
854 ((UV) s[uoffset ] << 56) +
855 ((UV) s[uoffset + 1] << 48) +
856 ((UV) s[uoffset + 2] << 40) +
857 ((UV) s[uoffset + 3] << 32) +
858 ((UV) s[uoffset + 4] << 24) +
859 ((UV) s[uoffset + 5] << 16) +
860 ((UV) s[uoffset + 6] << 8);
861 }
862#endif
863 }
864 }
865 else if (size < 8)
866 retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1);
867 else {
868 if (size == 8)
869 retnum = s[uoffset];
870 else if (size == 16)
871 retnum =
872 ((UV) s[uoffset] << 8) +
873 s[uoffset + 1];
874 else if (size == 32)
875 retnum =
876 ((UV) s[uoffset ] << 24) +
877 ((UV) s[uoffset + 1] << 16) +
878 ( s[uoffset + 2] << 8) +
879 s[uoffset + 3];
880#ifdef UV_IS_QUAD
881 else if (size == 64) {
882 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
883 "Bit vector size > 32 non-portable");
884 retnum =
885 ((UV) s[uoffset ] << 56) +
886 ((UV) s[uoffset + 1] << 48) +
887 ((UV) s[uoffset + 2] << 40) +
888 ((UV) s[uoffset + 3] << 32) +
889 ((UV) s[uoffset + 4] << 24) +
890 ((UV) s[uoffset + 5] << 16) +
891 ( s[uoffset + 6] << 8) +
892 s[uoffset + 7];
893 }
894#endif
895 }
896
897 return retnum;
898}
899
900/* currently converts input to bytes if possible but doesn't sweat failures,
901 * although it does ensure that the string it clobbers is not marked as
902 * utf8-valid any more
903 */
904void
905Perl_do_vecset(pTHX_ SV *sv)
906{
907 SSize_t offset, bitoffs = 0;
908 int size;
909 unsigned char *s;
910 UV lval;
911 I32 mask;
912 STRLEN targlen;
913 STRLEN len;
914 SV * const targ = LvTARG(sv);
915
916 PERL_ARGS_ASSERT_DO_VECSET;
917
918 if (!targ)
919 return;
920 s = (unsigned char*)SvPV_force_flags(targ, targlen,
921 SV_GMAGIC | SV_UNDEF_RETURNS_NULL);
922 if (SvUTF8(targ)) {
923 /* This is handled by the SvPOK_only below...
924 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
925 SvUTF8_off(targ);
926 */
927 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
928 }
929
930 (void)SvPOK_only(targ);
931 lval = SvUV(sv);
932 offset = LvTARGOFF(sv);
933 if (offset < 0)
934 Perl_croak(aTHX_ "Negative offset to vec in lvalue context");
935 size = LvTARGLEN(sv);
936 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
937 Perl_croak(aTHX_ "Illegal number of bits in vec");
938
939 if (size < 8) {
940 bitoffs = ((offset%8)*size)%8;
941 offset /= 8/size;
942 }
943 else if (size > 8)
944 offset *= size/8;
945
946 len = offset + (bitoffs + size + 7)/8; /* required number of bytes */
947 if (len > targlen) {
948 s = (unsigned char*)SvGROW(targ, len + 1);
949 (void)memzero((char *)(s + targlen), len - targlen + 1);
950 SvCUR_set(targ, len);
951 }
952
953 if (size < 8) {
954 mask = (1 << size) - 1;
955 lval &= mask;
956 s[offset] &= ~(mask << bitoffs);
957 s[offset] |= lval << bitoffs;
958 }
959 else {
960 if (size == 8)
961 s[offset ] = (U8)( lval & 0xff);
962 else if (size == 16) {
963 s[offset ] = (U8)((lval >> 8) & 0xff);
964 s[offset+1] = (U8)( lval & 0xff);
965 }
966 else if (size == 32) {
967 s[offset ] = (U8)((lval >> 24) & 0xff);
968 s[offset+1] = (U8)((lval >> 16) & 0xff);
969 s[offset+2] = (U8)((lval >> 8) & 0xff);
970 s[offset+3] = (U8)( lval & 0xff);
971 }
972#ifdef UV_IS_QUAD
973 else if (size == 64) {
974 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
975 "Bit vector size > 32 non-portable");
976 s[offset ] = (U8)((lval >> 56) & 0xff);
977 s[offset+1] = (U8)((lval >> 48) & 0xff);
978 s[offset+2] = (U8)((lval >> 40) & 0xff);
979 s[offset+3] = (U8)((lval >> 32) & 0xff);
980 s[offset+4] = (U8)((lval >> 24) & 0xff);
981 s[offset+5] = (U8)((lval >> 16) & 0xff);
982 s[offset+6] = (U8)((lval >> 8) & 0xff);
983 s[offset+7] = (U8)( lval & 0xff);
984 }
985#endif
986 }
987 SvSETMAGIC(targ);
988}
989
990void
991Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
992{
993#ifdef LIBERAL
994 long *dl;
995 long *ll;
996 long *rl;
997#endif
998 char *dc;
999 STRLEN leftlen;
1000 STRLEN rightlen;
1001 const char *lc;
1002 const char *rc;
1003 STRLEN len;
1004 STRLEN lensave;
1005 const char *lsave;
1006 const char *rsave;
1007 bool left_utf;
1008 bool right_utf;
1009 STRLEN needlen = 0;
1010
1011 PERL_ARGS_ASSERT_DO_VOP;
1012
1013 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv)))
1014 sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */
1015 if (sv == left) {
1016 lsave = lc = SvPV_force_nomg(left, leftlen);
1017 }
1018 else {
1019 lsave = lc = SvPV_nomg_const(left, leftlen);
1020 SvPV_force_nomg_nolen(sv);
1021 }
1022 rsave = rc = SvPV_nomg_const(right, rightlen);
1023
1024 /* This need to come after SvPV to ensure that string overloading has
1025 fired off. */
1026
1027 left_utf = DO_UTF8(left);
1028 right_utf = DO_UTF8(right);
1029
1030 if (left_utf && !right_utf) {
1031 /* Avoid triggering overloading again by using temporaries.
1032 Maybe there should be a variant of sv_utf8_upgrade that takes pvn
1033 */
1034 right = newSVpvn_flags(rsave, rightlen, SVs_TEMP);
1035 sv_utf8_upgrade(right);
1036 rsave = rc = SvPV_nomg_const(right, rightlen);
1037 right_utf = TRUE;
1038 }
1039 else if (!left_utf && right_utf) {
1040 left = newSVpvn_flags(lsave, leftlen, SVs_TEMP);
1041 sv_utf8_upgrade(left);
1042 lsave = lc = SvPV_nomg_const(left, leftlen);
1043 left_utf = TRUE;
1044 }
1045
1046 len = leftlen < rightlen ? leftlen : rightlen;
1047 lensave = len;
1048 SvCUR_set(sv, len);
1049 (void)SvPOK_only(sv);
1050 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1051 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1052 Newxz(dc, needlen + 1, char);
1053 }
1054 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
1055 dc = SvPV_force_nomg_nolen(sv);
1056 if (SvLEN(sv) < len + 1) {
1057 dc = SvGROW(sv, len + 1);
1058 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1059 }
1060 if (optype != OP_BIT_AND && (left_utf || right_utf))
1061 dc = SvGROW(sv, leftlen + rightlen + 1);
1062 }
1063 else {
1064 needlen = optype == OP_BIT_AND
1065 ? len : (leftlen > rightlen ? leftlen : rightlen);
1066 Newxz(dc, needlen + 1, char);
1067 sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
1068 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
1069 }
1070 if (left_utf || right_utf) {
1071 UV duc, luc, ruc;
1072 char *dcorig = dc;
1073 char *dcsave = NULL;
1074 STRLEN lulen = leftlen;
1075 STRLEN rulen = rightlen;
1076 STRLEN ulen;
1077
1078 switch (optype) {
1079 case OP_BIT_AND:
1080 while (lulen && rulen) {
1081 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1082 lc += ulen;
1083 lulen -= ulen;
1084 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1085 rc += ulen;
1086 rulen -= ulen;
1087 duc = luc & ruc;
1088 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1089 }
1090 if (sv == left || sv == right)
1091 (void)sv_usepvn(sv, dcorig, needlen);
1092 SvCUR_set(sv, dc - dcorig);
1093 break;
1094 case OP_BIT_XOR:
1095 while (lulen && rulen) {
1096 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1097 lc += ulen;
1098 lulen -= ulen;
1099 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1100 rc += ulen;
1101 rulen -= ulen;
1102 duc = luc ^ ruc;
1103 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1104 }
1105 goto mop_up_utf;
1106 case OP_BIT_OR:
1107 while (lulen && rulen) {
1108 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1109 lc += ulen;
1110 lulen -= ulen;
1111 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1112 rc += ulen;
1113 rulen -= ulen;
1114 duc = luc | ruc;
1115 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1116 }
1117 mop_up_utf:
1118 if (rulen)
1119 dcsave = savepvn(rc, rulen);
1120 else if (lulen)
1121 dcsave = savepvn(lc, lulen);
1122 if (sv == left || sv == right)
1123 (void)sv_usepvn(sv, dcorig, needlen); /* uses Renew(); defaults to nomg */
1124 SvCUR_set(sv, dc - dcorig);
1125 if (rulen)
1126 sv_catpvn_nomg(sv, dcsave, rulen);
1127 else if (lulen)
1128 sv_catpvn_nomg(sv, dcsave, lulen);
1129 else
1130 *SvEND(sv) = '\0';
1131 Safefree(dcsave);
1132 break;
1133 default:
1134 if (sv == left || sv == right)
1135 Safefree(dcorig);
1136 Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)",
1137 (unsigned)optype, PL_op_name[optype]);
1138 }
1139 SvUTF8_on(sv);
1140 goto finish;
1141 }
1142 else
1143#ifdef LIBERAL
1144 if (len >= sizeof(long)*4 &&
1145 !((unsigned long)dc % sizeof(long)) &&
1146 !((unsigned long)lc % sizeof(long)) &&
1147 !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */
1148 {
1149 const STRLEN remainder = len % (sizeof(long)*4);
1150 len /= (sizeof(long)*4);
1151
1152 dl = (long*)dc;
1153 ll = (long*)lc;
1154 rl = (long*)rc;
1155
1156 switch (optype) {
1157 case OP_BIT_AND:
1158 while (len--) {
1159 *dl++ = *ll++ & *rl++;
1160 *dl++ = *ll++ & *rl++;
1161 *dl++ = *ll++ & *rl++;
1162 *dl++ = *ll++ & *rl++;
1163 }
1164 break;
1165 case OP_BIT_XOR:
1166 while (len--) {
1167 *dl++ = *ll++ ^ *rl++;
1168 *dl++ = *ll++ ^ *rl++;
1169 *dl++ = *ll++ ^ *rl++;
1170 *dl++ = *ll++ ^ *rl++;
1171 }
1172 break;
1173 case OP_BIT_OR:
1174 while (len--) {
1175 *dl++ = *ll++ | *rl++;
1176 *dl++ = *ll++ | *rl++;
1177 *dl++ = *ll++ | *rl++;
1178 *dl++ = *ll++ | *rl++;
1179 }
1180 }
1181
1182 dc = (char*)dl;
1183 lc = (char*)ll;
1184 rc = (char*)rl;
1185
1186 len = remainder;
1187 }
1188#endif
1189 {
1190 switch (optype) {
1191 case OP_BIT_AND:
1192 while (len--)
1193 *dc++ = *lc++ & *rc++;
1194 *dc = '\0';
1195 break;
1196 case OP_BIT_XOR:
1197 while (len--)
1198 *dc++ = *lc++ ^ *rc++;
1199 goto mop_up;
1200 case OP_BIT_OR:
1201 while (len--)
1202 *dc++ = *lc++ | *rc++;
1203 mop_up:
1204 len = lensave;
1205 if (rightlen > len)
1206 sv_catpvn_nomg(sv, rsave + len, rightlen - len);
1207 else if (leftlen > (STRLEN)len)
1208 sv_catpvn_nomg(sv, lsave + len, leftlen - len);
1209 else
1210 *SvEND(sv) = '\0';
1211 break;
1212 }
1213 }
1214finish:
1215 SvTAINT(sv);
1216}
1217
1218OP *
1219Perl_do_kv(pTHX)
1220{
1221 dSP;
1222 HV * const keys = MUTABLE_HV(POPs);
1223 HE *entry;
1224 const I32 gimme = GIMME_V;
1225 const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV);
1226 /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */
1227 const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS || PL_op->op_type == OP_RKEYS);
1228 const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES || PL_op->op_type == OP_RVALUES);
1229
1230 (void)hv_iterinit(keys); /* always reset iterator regardless */
1231
1232 if (gimme == G_VOID)
1233 RETURN;
1234
1235 if (gimme == G_SCALAR) {
1236 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1237 SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */
1238 sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0);
1239 LvTYPE(ret) = 'k';
1240 LvTARG(ret) = SvREFCNT_inc_simple(keys);
1241 PUSHs(ret);
1242 }
1243 else {
1244 IV i;
1245 dTARGET;
1246
1247 if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) {
1248 i = HvUSEDKEYS(keys);
1249 }
1250 else {
1251 i = 0;
1252 while (hv_iternext(keys)) i++;
1253 }
1254 PUSHi( i );
1255 }
1256 RETURN;
1257 }
1258
1259 EXTEND(SP, HvUSEDKEYS(keys) * (dokeys + dovalues));
1260
1261 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1262 while ((entry = hv_iternext(keys))) {
1263 SPAGAIN;
1264 if (dokeys) {
1265 SV* const sv = hv_iterkeysv(entry);
1266 XPUSHs(sv); /* won't clobber stack_sp */
1267 }
1268 if (dovalues) {
1269 SV *tmpstr;
1270 PUTBACK;
1271 tmpstr = hv_iterval(keys,entry);
1272 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1273 (unsigned long)HeHASH(entry),
1274 (int)HvMAX(keys)+1,
1275 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
1276 SPAGAIN;
1277 XPUSHs(tmpstr);
1278 }
1279 PUTBACK;
1280 }
1281 return NORMAL;
1282}
1283
1284/*
1285 * Local variables:
1286 * c-indentation-style: bsd
1287 * c-basic-offset: 4
1288 * indent-tabs-mode: nil
1289 * End:
1290 *
1291 * ex: set ts=8 sts=4 sw=4 et:
1292 */