This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp: Add/subtract consts to match embed.fnc
[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_schomp() and pp_chomp() - scalar and array
19 * chomp operations - call the function do_chomp() 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 dVAR;
34 I32 matches = 0;
35 STRLEN len;
36 U8 *s = (U8*)SvPV_nomg(sv,len);
37 U8 * const send = s+len;
38 const short * const tbl = (short*)cPVOP->op_pv;
39
40 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE;
41
42 if (!tbl)
43 Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
44
45 /* First, take care of non-UTF-8 input strings, because they're easy */
46 if (!SvUTF8(sv)) {
47 while (s < send) {
48 const I32 ch = tbl[*s];
49 if (ch >= 0) {
50 matches++;
51 *s = (U8)ch;
52 }
53 s++;
54 }
55 SvSETMAGIC(sv);
56 }
57 else {
58 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
59 U8 *d;
60 U8 *dstart;
61
62 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
63 if (grows)
64 Newx(d, len*2+1, U8);
65 else
66 d = s;
67 dstart = d;
68 while (s < send) {
69 STRLEN ulen;
70 I32 ch;
71
72 /* Need to check this, otherwise 128..255 won't match */
73 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
74 if (c < 0x100 && (ch = tbl[c]) >= 0) {
75 matches++;
76 d = uvchr_to_utf8(d, ch);
77 s += ulen;
78 }
79 else { /* No match -> copy */
80 Move(s, d, ulen, U8);
81 d += ulen;
82 s += ulen;
83 }
84 }
85 if (grows) {
86 sv_setpvn(sv, (char*)dstart, d - dstart);
87 Safefree(dstart);
88 }
89 else {
90 *d = '\0';
91 SvCUR_set(sv, d - dstart);
92 }
93 SvUTF8_on(sv);
94 SvSETMAGIC(sv);
95 }
96 return matches;
97}
98
99STATIC I32
100S_do_trans_count(pTHX_ SV * const sv)
101{
102 dVAR;
103 STRLEN len;
104 const U8 *s = (const U8*)SvPV_nomg_const(sv, len);
105 const U8 * const send = s + len;
106 I32 matches = 0;
107 const short * const tbl = (short*)cPVOP->op_pv;
108
109 PERL_ARGS_ASSERT_DO_TRANS_COUNT;
110
111 if (!tbl)
112 Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
113
114 if (!SvUTF8(sv)) {
115 while (s < send) {
116 if (tbl[*s++] >= 0)
117 matches++;
118 }
119 }
120 else {
121 const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
122 while (s < send) {
123 STRLEN ulen;
124 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
125 if (c < 0x100) {
126 if (tbl[c] >= 0)
127 matches++;
128 } else if (complement)
129 matches++;
130 s += ulen;
131 }
132 }
133
134 return matches;
135}
136
137STATIC I32
138S_do_trans_complex(pTHX_ SV * const sv)
139{
140 dVAR;
141 STRLEN len;
142 U8 *s = (U8*)SvPV_nomg(sv, len);
143 U8 * const send = s+len;
144 I32 matches = 0;
145 const short * const tbl = (short*)cPVOP->op_pv;
146
147 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX;
148
149 if (!tbl)
150 Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
151
152 if (!SvUTF8(sv)) {
153 U8 *d = s;
154 U8 * const dstart = d;
155
156 if (PL_op->op_private & OPpTRANS_SQUASH) {
157 const U8* p = send;
158 while (s < send) {
159 const I32 ch = tbl[*s];
160 if (ch >= 0) {
161 *d = (U8)ch;
162 matches++;
163 if (p != d - 1 || *p != *d)
164 p = d++;
165 }
166 else if (ch == -1) /* -1 is unmapped character */
167 *d++ = *s;
168 else if (ch == -2) /* -2 is delete character */
169 matches++;
170 s++;
171 }
172 }
173 else {
174 while (s < send) {
175 const I32 ch = tbl[*s];
176 if (ch >= 0) {
177 matches++;
178 *d++ = (U8)ch;
179 }
180 else if (ch == -1) /* -1 is unmapped character */
181 *d++ = *s;
182 else if (ch == -2) /* -2 is delete character */
183 matches++;
184 s++;
185 }
186 }
187 *d = '\0';
188 SvCUR_set(sv, d - dstart);
189 }
190 else { /* is utf8 */
191 const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
192 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
193 const I32 del = PL_op->op_private & OPpTRANS_DELETE;
194 U8 *d;
195 U8 *dstart;
196 STRLEN rlen = 0;
197
198 if (grows)
199 Newx(d, len*2+1, U8);
200 else
201 d = s;
202 dstart = d;
203 if (complement && !del)
204 rlen = tbl[0x100];
205
206 if (PL_op->op_private & OPpTRANS_SQUASH) {
207 UV pch = 0xfeedface;
208 while (s < send) {
209 STRLEN len;
210 const UV comp = utf8n_to_uvchr(s, send - s, &len,
211 UTF8_ALLOW_DEFAULT);
212 I32 ch;
213
214 if (comp > 0xff) {
215 if (!complement) {
216 Move(s, d, len, U8);
217 d += len;
218 }
219 else {
220 matches++;
221 if (!del) {
222 ch = (rlen == 0) ? (I32)comp :
223 (comp - 0x100 < rlen) ?
224 tbl[comp+1] : tbl[0x100+rlen];
225 if ((UV)ch != pch) {
226 d = uvchr_to_utf8(d, ch);
227 pch = (UV)ch;
228 }
229 s += len;
230 continue;
231 }
232 }
233 }
234 else if ((ch = tbl[comp]) >= 0) {
235 matches++;
236 if ((UV)ch != pch) {
237 d = uvchr_to_utf8(d, ch);
238 pch = (UV)ch;
239 }
240 s += len;
241 continue;
242 }
243 else if (ch == -1) { /* -1 is unmapped character */
244 Move(s, d, len, U8);
245 d += len;
246 }
247 else if (ch == -2) /* -2 is delete character */
248 matches++;
249 s += len;
250 pch = 0xfeedface;
251 }
252 }
253 else {
254 while (s < send) {
255 STRLEN len;
256 const UV comp = utf8n_to_uvchr(s, send - s, &len,
257 UTF8_ALLOW_DEFAULT);
258 I32 ch;
259 if (comp > 0xff) {
260 if (!complement) {
261 Move(s, d, len, U8);
262 d += len;
263 }
264 else {
265 matches++;
266 if (!del) {
267 if (comp - 0x100 < rlen)
268 d = uvchr_to_utf8(d, tbl[comp+1]);
269 else
270 d = uvchr_to_utf8(d, tbl[0x100+rlen]);
271 }
272 }
273 }
274 else if ((ch = tbl[comp]) >= 0) {
275 d = uvchr_to_utf8(d, ch);
276 matches++;
277 }
278 else if (ch == -1) { /* -1 is unmapped character */
279 Move(s, d, len, U8);
280 d += len;
281 }
282 else if (ch == -2) /* -2 is delete character */
283 matches++;
284 s += len;
285 }
286 }
287 if (grows) {
288 sv_setpvn(sv, (char*)dstart, d - dstart);
289 Safefree(dstart);
290 }
291 else {
292 *d = '\0';
293 SvCUR_set(sv, d - dstart);
294 }
295 SvUTF8_on(sv);
296 }
297 SvSETMAGIC(sv);
298 return matches;
299}
300
301STATIC I32
302S_do_trans_simple_utf8(pTHX_ SV * const sv)
303{
304 dVAR;
305 U8 *s;
306 U8 *send;
307 U8 *d;
308 U8 *start;
309 U8 *dstart, *dend;
310 I32 matches = 0;
311 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
312 STRLEN len;
313 SV* const rv =
314#ifdef USE_ITHREADS
315 PAD_SVl(cPADOP->op_padix);
316#else
317 MUTABLE_SV(cSVOP->op_sv);
318#endif
319 HV* const hv = MUTABLE_HV(SvRV(rv));
320 SV* const * svp = hv_fetchs(hv, "NONE", FALSE);
321 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
322 const UV extra = none + 1;
323 UV final = 0;
324 U8 hibit = 0;
325
326 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8;
327
328 s = (U8*)SvPV_nomg(sv, len);
329 if (!SvUTF8(sv)) {
330 const U8 *t = s;
331 const U8 * const e = s + len;
332 while (t < e) {
333 const U8 ch = *t++;
334 hibit = !NATIVE_IS_INVARIANT(ch);
335 if (hibit) {
336 s = bytes_to_utf8(s, &len);
337 break;
338 }
339 }
340 }
341 send = s + len;
342 start = s;
343
344 svp = hv_fetchs(hv, "FINAL", FALSE);
345 if (svp)
346 final = SvUV(*svp);
347
348 if (grows) {
349 /* d needs to be bigger than s, in case e.g. upgrading is required */
350 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
351 dend = d + len * 3;
352 dstart = d;
353 }
354 else {
355 dstart = d = s;
356 dend = d + len;
357 }
358
359 while (s < send) {
360 const UV uv = swash_fetch(rv, s, TRUE);
361 if (uv < none) {
362 s += UTF8SKIP(s);
363 matches++;
364 d = uvuni_to_utf8(d, uv);
365 }
366 else if (uv == none) {
367 const int i = UTF8SKIP(s);
368 Move(s, d, i, U8);
369 d += i;
370 s += i;
371 }
372 else if (uv == extra) {
373 s += UTF8SKIP(s);
374 matches++;
375 d = uvuni_to_utf8(d, final);
376 }
377 else
378 s += UTF8SKIP(s);
379
380 if (d > dend) {
381 const STRLEN clen = d - dstart;
382 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
383 if (!grows)
384 Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
385 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
386 d = dstart + clen;
387 dend = dstart + nlen;
388 }
389 }
390 if (grows || hibit) {
391 sv_setpvn(sv, (char*)dstart, d - dstart);
392 Safefree(dstart);
393 if (grows && hibit)
394 Safefree(start);
395 }
396 else {
397 *d = '\0';
398 SvCUR_set(sv, d - dstart);
399 }
400 SvSETMAGIC(sv);
401 SvUTF8_on(sv);
402
403 return matches;
404}
405
406STATIC I32
407S_do_trans_count_utf8(pTHX_ SV * const sv)
408{
409 dVAR;
410 const U8 *s;
411 const U8 *start = NULL;
412 const U8 *send;
413 I32 matches = 0;
414 STRLEN len;
415 SV* const rv =
416#ifdef USE_ITHREADS
417 PAD_SVl(cPADOP->op_padix);
418#else
419 MUTABLE_SV(cSVOP->op_sv);
420#endif
421 HV* const hv = MUTABLE_HV(SvRV(rv));
422 SV* const * const svp = hv_fetchs(hv, "NONE", FALSE);
423 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
424 const UV extra = none + 1;
425 U8 hibit = 0;
426
427 PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8;
428
429 s = (const U8*)SvPV_nomg_const(sv, len);
430 if (!SvUTF8(sv)) {
431 const U8 *t = s;
432 const U8 * const e = s + len;
433 while (t < e) {
434 const U8 ch = *t++;
435 hibit = !NATIVE_IS_INVARIANT(ch);
436 if (hibit) {
437 start = s = bytes_to_utf8(s, &len);
438 break;
439 }
440 }
441 }
442 send = s + len;
443
444 while (s < send) {
445 const UV uv = swash_fetch(rv, s, TRUE);
446 if (uv < none || uv == extra)
447 matches++;
448 s += UTF8SKIP(s);
449 }
450 if (hibit)
451 Safefree(start);
452
453 return matches;
454}
455
456STATIC I32
457S_do_trans_complex_utf8(pTHX_ SV * const sv)
458{
459 dVAR;
460 U8 *start, *send;
461 U8 *d;
462 I32 matches = 0;
463 const I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
464 const I32 del = PL_op->op_private & OPpTRANS_DELETE;
465 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
466 SV* const rv =
467#ifdef USE_ITHREADS
468 PAD_SVl(cPADOP->op_padix);
469#else
470 MUTABLE_SV(cSVOP->op_sv);
471#endif
472 HV * const hv = MUTABLE_HV(SvRV(rv));
473 SV * const *svp = hv_fetchs(hv, "NONE", FALSE);
474 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
475 const UV extra = none + 1;
476 UV final = 0;
477 bool havefinal = FALSE;
478 STRLEN len;
479 U8 *dstart, *dend;
480 U8 hibit = 0;
481 U8 *s = (U8*)SvPV_nomg(sv, len);
482
483 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8;
484
485 if (!SvUTF8(sv)) {
486 const U8 *t = s;
487 const U8 * const e = s + len;
488 while (t < e) {
489 const U8 ch = *t++;
490 hibit = !NATIVE_IS_INVARIANT(ch);
491 if (hibit) {
492 s = bytes_to_utf8(s, &len);
493 break;
494 }
495 }
496 }
497 send = s + len;
498 start = s;
499
500 svp = hv_fetchs(hv, "FINAL", FALSE);
501 if (svp) {
502 final = SvUV(*svp);
503 havefinal = TRUE;
504 }
505
506 if (grows) {
507 /* d needs to be bigger than s, in case e.g. upgrading is required */
508 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
509 dend = d + len * 3;
510 dstart = d;
511 }
512 else {
513 dstart = d = s;
514 dend = d + len;
515 }
516
517 if (squash) {
518 UV puv = 0xfeedface;
519 while (s < send) {
520 UV uv = swash_fetch(rv, s, TRUE);
521
522 if (d > dend) {
523 const STRLEN clen = d - dstart;
524 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
525 if (!grows)
526 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
527 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
528 d = dstart + clen;
529 dend = dstart + nlen;
530 }
531 if (uv < none) {
532 matches++;
533 s += UTF8SKIP(s);
534 if (uv != puv) {
535 d = uvuni_to_utf8(d, uv);
536 puv = uv;
537 }
538 continue;
539 }
540 else if (uv == none) { /* "none" is unmapped character */
541 const int i = UTF8SKIP(s);
542 Move(s, d, i, U8);
543 d += i;
544 s += i;
545 puv = 0xfeedface;
546 continue;
547 }
548 else if (uv == extra && !del) {
549 matches++;
550 if (havefinal) {
551 s += UTF8SKIP(s);
552 if (puv != final) {
553 d = uvuni_to_utf8(d, final);
554 puv = final;
555 }
556 }
557 else {
558 STRLEN len;
559 uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT);
560 if (uv != puv) {
561 Move(s, d, len, U8);
562 d += len;
563 puv = uv;
564 }
565 s += len;
566 }
567 continue;
568 }
569 matches++; /* "none+1" is delete character */
570 s += UTF8SKIP(s);
571 }
572 }
573 else {
574 while (s < send) {
575 const UV uv = swash_fetch(rv, s, TRUE);
576 if (d > dend) {
577 const STRLEN clen = d - dstart;
578 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
579 if (!grows)
580 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
581 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
582 d = dstart + clen;
583 dend = dstart + nlen;
584 }
585 if (uv < none) {
586 matches++;
587 s += UTF8SKIP(s);
588 d = uvuni_to_utf8(d, uv);
589 continue;
590 }
591 else if (uv == none) { /* "none" is unmapped character */
592 const int i = UTF8SKIP(s);
593 Move(s, d, i, U8);
594 d += i;
595 s += i;
596 continue;
597 }
598 else if (uv == extra && !del) {
599 matches++;
600 s += UTF8SKIP(s);
601 d = uvuni_to_utf8(d, final);
602 continue;
603 }
604 matches++; /* "none+1" is delete character */
605 s += UTF8SKIP(s);
606 }
607 }
608 if (grows || hibit) {
609 sv_setpvn(sv, (char*)dstart, d - dstart);
610 Safefree(dstart);
611 if (grows && hibit)
612 Safefree(start);
613 }
614 else {
615 *d = '\0';
616 SvCUR_set(sv, d - dstart);
617 }
618 SvUTF8_on(sv);
619 SvSETMAGIC(sv);
620
621 return matches;
622}
623
624I32
625Perl_do_trans(pTHX_ SV *sv)
626{
627 dVAR;
628 STRLEN len;
629 const I32 hasutf = (PL_op->op_private &
630 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
631
632 PERL_ARGS_ASSERT_DO_TRANS;
633
634 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) {
635 if (SvIsCOW(sv))
636 sv_force_normal_flags(sv, 0);
637 if (SvREADONLY(sv))
638 Perl_croak_no_modify(aTHX);
639 }
640 (void)SvPV_const(sv, len);
641 if (!len)
642 return 0;
643 if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) {
644 if (!SvPOKp(sv))
645 (void)SvPV_force(sv, len);
646 (void)SvPOK_only_UTF8(sv);
647 }
648
649 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
650
651 switch (PL_op->op_private & ~hasutf & (
652 OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL|
653 OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) {
654 case 0:
655 if (hasutf)
656 return do_trans_simple_utf8(sv);
657 else
658 return do_trans_simple(sv);
659
660 case OPpTRANS_IDENTICAL:
661 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
662 if (hasutf)
663 return do_trans_count_utf8(sv);
664 else
665 return do_trans_count(sv);
666
667 default:
668 if (hasutf)
669 return do_trans_complex_utf8(sv);
670 else
671 return do_trans_complex(sv);
672 }
673}
674
675void
676Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp)
677{
678 dVAR;
679 SV ** const oldmark = mark;
680 register I32 items = sp - mark;
681 register STRLEN len;
682 STRLEN delimlen;
683
684 PERL_ARGS_ASSERT_DO_JOIN;
685
686 (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */
687 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
688
689 mark++;
690 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
691 SvUPGRADE(sv, SVt_PV);
692 if (SvLEN(sv) < len + items) { /* current length is way too short */
693 while (items-- > 0) {
694 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
695 STRLEN tmplen;
696 SvPV_const(*mark, tmplen);
697 len += tmplen;
698 }
699 mark++;
700 }
701 SvGROW(sv, len + 1); /* so try to pre-extend */
702
703 mark = oldmark;
704 items = sp - mark;
705 ++mark;
706 }
707
708 sv_setpvs(sv, "");
709 /* sv_setpv retains old UTF8ness [perl #24846] */
710 SvUTF8_off(sv);
711
712 if (PL_tainting && SvMAGICAL(sv))
713 SvTAINTED_off(sv);
714
715 if (items-- > 0) {
716 if (*mark)
717 sv_catsv(sv, *mark);
718 mark++;
719 }
720
721 if (delimlen) {
722 for (; items > 0; items--,mark++) {
723 sv_catsv(sv,delim);
724 sv_catsv(sv,*mark);
725 }
726 }
727 else {
728 for (; items > 0; items--,mark++)
729 sv_catsv(sv,*mark);
730 }
731 SvSETMAGIC(sv);
732}
733
734void
735Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
736{
737 dVAR;
738 STRLEN patlen;
739 const char * const pat = SvPV_const(*sarg, patlen);
740 bool do_taint = FALSE;
741
742 PERL_ARGS_ASSERT_DO_SPRINTF;
743
744 SvUTF8_off(sv);
745 if (DO_UTF8(*sarg))
746 SvUTF8_on(sv);
747 sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint);
748 SvSETMAGIC(sv);
749 if (do_taint)
750 SvTAINTED_on(sv);
751}
752
753/* currently converts input to bytes if possible, but doesn't sweat failure */
754UV
755Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
756{
757 dVAR;
758 STRLEN srclen, len, uoffset, bitoffs = 0;
759 const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen);
760 UV retnum = 0;
761
762 PERL_ARGS_ASSERT_DO_VECGET;
763
764 if (offset < 0)
765 return 0;
766 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
767 Perl_croak(aTHX_ "Illegal number of bits in vec");
768
769 if (SvUTF8(sv))
770 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
771
772 if (size < 8) {
773 bitoffs = ((offset%8)*size)%8;
774 uoffset = offset/(8/size);
775 }
776 else if (size > 8)
777 uoffset = offset*(size/8);
778 else
779 uoffset = offset;
780
781 len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */
782 if (len > srclen) {
783 if (size <= 8)
784 retnum = 0;
785 else {
786 if (size == 16) {
787 if (uoffset >= srclen)
788 retnum = 0;
789 else
790 retnum = (UV) s[uoffset] << 8;
791 }
792 else if (size == 32) {
793 if (uoffset >= srclen)
794 retnum = 0;
795 else if (uoffset + 1 >= srclen)
796 retnum =
797 ((UV) s[uoffset ] << 24);
798 else if (uoffset + 2 >= srclen)
799 retnum =
800 ((UV) s[uoffset ] << 24) +
801 ((UV) s[uoffset + 1] << 16);
802 else
803 retnum =
804 ((UV) s[uoffset ] << 24) +
805 ((UV) s[uoffset + 1] << 16) +
806 ( s[uoffset + 2] << 8);
807 }
808#ifdef UV_IS_QUAD
809 else if (size == 64) {
810 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
811 "Bit vector size > 32 non-portable");
812 if (uoffset >= srclen)
813 retnum = 0;
814 else if (uoffset + 1 >= srclen)
815 retnum =
816 (UV) s[uoffset ] << 56;
817 else if (uoffset + 2 >= srclen)
818 retnum =
819 ((UV) s[uoffset ] << 56) +
820 ((UV) s[uoffset + 1] << 48);
821 else if (uoffset + 3 >= srclen)
822 retnum =
823 ((UV) s[uoffset ] << 56) +
824 ((UV) s[uoffset + 1] << 48) +
825 ((UV) s[uoffset + 2] << 40);
826 else if (uoffset + 4 >= srclen)
827 retnum =
828 ((UV) s[uoffset ] << 56) +
829 ((UV) s[uoffset + 1] << 48) +
830 ((UV) s[uoffset + 2] << 40) +
831 ((UV) s[uoffset + 3] << 32);
832 else if (uoffset + 5 >= srclen)
833 retnum =
834 ((UV) s[uoffset ] << 56) +
835 ((UV) s[uoffset + 1] << 48) +
836 ((UV) s[uoffset + 2] << 40) +
837 ((UV) s[uoffset + 3] << 32) +
838 ( s[uoffset + 4] << 24);
839 else if (uoffset + 6 >= srclen)
840 retnum =
841 ((UV) s[uoffset ] << 56) +
842 ((UV) s[uoffset + 1] << 48) +
843 ((UV) s[uoffset + 2] << 40) +
844 ((UV) s[uoffset + 3] << 32) +
845 ((UV) s[uoffset + 4] << 24) +
846 ((UV) s[uoffset + 5] << 16);
847 else
848 retnum =
849 ((UV) s[uoffset ] << 56) +
850 ((UV) s[uoffset + 1] << 48) +
851 ((UV) s[uoffset + 2] << 40) +
852 ((UV) s[uoffset + 3] << 32) +
853 ((UV) s[uoffset + 4] << 24) +
854 ((UV) s[uoffset + 5] << 16) +
855 ( s[uoffset + 6] << 8);
856 }
857#endif
858 }
859 }
860 else if (size < 8)
861 retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1);
862 else {
863 if (size == 8)
864 retnum = s[uoffset];
865 else if (size == 16)
866 retnum =
867 ((UV) s[uoffset] << 8) +
868 s[uoffset + 1];
869 else if (size == 32)
870 retnum =
871 ((UV) s[uoffset ] << 24) +
872 ((UV) s[uoffset + 1] << 16) +
873 ( s[uoffset + 2] << 8) +
874 s[uoffset + 3];
875#ifdef UV_IS_QUAD
876 else if (size == 64) {
877 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
878 "Bit vector size > 32 non-portable");
879 retnum =
880 ((UV) s[uoffset ] << 56) +
881 ((UV) s[uoffset + 1] << 48) +
882 ((UV) s[uoffset + 2] << 40) +
883 ((UV) s[uoffset + 3] << 32) +
884 ((UV) s[uoffset + 4] << 24) +
885 ((UV) s[uoffset + 5] << 16) +
886 ( s[uoffset + 6] << 8) +
887 s[uoffset + 7];
888 }
889#endif
890 }
891
892 return retnum;
893}
894
895/* currently converts input to bytes if possible but doesn't sweat failures,
896 * although it does ensure that the string it clobbers is not marked as
897 * utf8-valid any more
898 */
899void
900Perl_do_vecset(pTHX_ SV *sv)
901{
902 dVAR;
903 register I32 offset, bitoffs = 0;
904 register I32 size;
905 register unsigned char *s;
906 register UV lval;
907 I32 mask;
908 STRLEN targlen;
909 STRLEN len;
910 SV * const targ = LvTARG(sv);
911
912 PERL_ARGS_ASSERT_DO_VECSET;
913
914 if (!targ)
915 return;
916 s = (unsigned char*)SvPV_force(targ, targlen);
917 if (SvUTF8(targ)) {
918 /* This is handled by the SvPOK_only below...
919 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
920 SvUTF8_off(targ);
921 */
922 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
923 }
924
925 (void)SvPOK_only(targ);
926 lval = SvUV(sv);
927 offset = LvTARGOFF(sv);
928 if (offset < 0)
929 Perl_croak(aTHX_ "Negative offset to vec in lvalue context");
930 size = LvTARGLEN(sv);
931 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
932 Perl_croak(aTHX_ "Illegal number of bits in vec");
933
934 if (size < 8) {
935 bitoffs = ((offset%8)*size)%8;
936 offset /= 8/size;
937 }
938 else if (size > 8)
939 offset *= size/8;
940
941 len = offset + (bitoffs + size + 7)/8; /* required number of bytes */
942 if (len > targlen) {
943 s = (unsigned char*)SvGROW(targ, len + 1);
944 (void)memzero((char *)(s + targlen), len - targlen + 1);
945 SvCUR_set(targ, len);
946 }
947
948 if (size < 8) {
949 mask = (1 << size) - 1;
950 lval &= mask;
951 s[offset] &= ~(mask << bitoffs);
952 s[offset] |= lval << bitoffs;
953 }
954 else {
955 if (size == 8)
956 s[offset ] = (U8)( lval & 0xff);
957 else if (size == 16) {
958 s[offset ] = (U8)((lval >> 8) & 0xff);
959 s[offset+1] = (U8)( lval & 0xff);
960 }
961 else if (size == 32) {
962 s[offset ] = (U8)((lval >> 24) & 0xff);
963 s[offset+1] = (U8)((lval >> 16) & 0xff);
964 s[offset+2] = (U8)((lval >> 8) & 0xff);
965 s[offset+3] = (U8)( lval & 0xff);
966 }
967#ifdef UV_IS_QUAD
968 else if (size == 64) {
969 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
970 "Bit vector size > 32 non-portable");
971 s[offset ] = (U8)((lval >> 56) & 0xff);
972 s[offset+1] = (U8)((lval >> 48) & 0xff);
973 s[offset+2] = (U8)((lval >> 40) & 0xff);
974 s[offset+3] = (U8)((lval >> 32) & 0xff);
975 s[offset+4] = (U8)((lval >> 24) & 0xff);
976 s[offset+5] = (U8)((lval >> 16) & 0xff);
977 s[offset+6] = (U8)((lval >> 8) & 0xff);
978 s[offset+7] = (U8)( lval & 0xff);
979 }
980#endif
981 }
982 SvSETMAGIC(targ);
983}
984
985void
986Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
987{
988 dVAR;
989#ifdef LIBERAL
990 register long *dl;
991 register long *ll;
992 register long *rl;
993#endif
994 register char *dc;
995 STRLEN leftlen;
996 STRLEN rightlen;
997 register const char *lc;
998 register const char *rc;
999 register STRLEN len;
1000 STRLEN lensave;
1001 const char *lsave;
1002 const char *rsave;
1003 bool left_utf;
1004 bool right_utf;
1005 STRLEN needlen = 0;
1006
1007 PERL_ARGS_ASSERT_DO_VOP;
1008
1009 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1010 sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */
1011 if (sv == left) {
1012 lsave = lc = SvPV_force_nomg(left, leftlen);
1013 }
1014 else {
1015 lsave = lc = SvPV_nomg_const(left, leftlen);
1016 SvPV_force_nomg_nolen(sv);
1017 }
1018 rsave = rc = SvPV_nomg_const(right, rightlen);
1019
1020 /* This need to come after SvPV to ensure that string overloading has
1021 fired off. */
1022
1023 left_utf = DO_UTF8(left);
1024 right_utf = DO_UTF8(right);
1025
1026 if (left_utf && !right_utf) {
1027 /* Avoid triggering overloading again by using temporaries.
1028 Maybe there should be a variant of sv_utf8_upgrade that takes pvn
1029 */
1030 right = newSVpvn_flags(rsave, rightlen, SVs_TEMP);
1031 sv_utf8_upgrade(right);
1032 rsave = rc = SvPV_nomg_const(right, rightlen);
1033 right_utf = TRUE;
1034 }
1035 else if (!left_utf && right_utf) {
1036 left = newSVpvn_flags(lsave, leftlen, SVs_TEMP);
1037 sv_utf8_upgrade(left);
1038 lsave = lc = SvPV_nomg_const(left, leftlen);
1039 left_utf = TRUE;
1040 }
1041
1042 len = leftlen < rightlen ? leftlen : rightlen;
1043 lensave = len;
1044 SvCUR_set(sv, len);
1045 (void)SvPOK_only(sv);
1046 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1047 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1048 Newxz(dc, needlen + 1, char);
1049 }
1050 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
1051 dc = SvPV_force_nomg_nolen(sv);
1052 if (SvLEN(sv) < len + 1) {
1053 dc = SvGROW(sv, len + 1);
1054 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1055 }
1056 if (optype != OP_BIT_AND && (left_utf || right_utf))
1057 dc = SvGROW(sv, leftlen + rightlen + 1);
1058 }
1059 else {
1060 needlen = optype == OP_BIT_AND
1061 ? len : (leftlen > rightlen ? leftlen : rightlen);
1062 Newxz(dc, needlen + 1, char);
1063 sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
1064 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
1065 }
1066 if (left_utf || right_utf) {
1067 UV duc, luc, ruc;
1068 char *dcorig = dc;
1069 char *dcsave = NULL;
1070 STRLEN lulen = leftlen;
1071 STRLEN rulen = rightlen;
1072 STRLEN ulen;
1073
1074 switch (optype) {
1075 case OP_BIT_AND:
1076 while (lulen && rulen) {
1077 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1078 lc += ulen;
1079 lulen -= ulen;
1080 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1081 rc += ulen;
1082 rulen -= ulen;
1083 duc = luc & ruc;
1084 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1085 }
1086 if (sv == left || sv == right)
1087 (void)sv_usepvn(sv, dcorig, needlen);
1088 SvCUR_set(sv, dc - dcorig);
1089 break;
1090 case OP_BIT_XOR:
1091 while (lulen && rulen) {
1092 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1093 lc += ulen;
1094 lulen -= ulen;
1095 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1096 rc += ulen;
1097 rulen -= ulen;
1098 duc = luc ^ ruc;
1099 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1100 }
1101 goto mop_up_utf;
1102 case OP_BIT_OR:
1103 while (lulen && rulen) {
1104 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1105 lc += ulen;
1106 lulen -= ulen;
1107 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1108 rc += ulen;
1109 rulen -= ulen;
1110 duc = luc | ruc;
1111 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1112 }
1113 mop_up_utf:
1114 if (rulen)
1115 dcsave = savepvn(rc, rulen);
1116 else if (lulen)
1117 dcsave = savepvn(lc, lulen);
1118 if (sv == left || sv == right)
1119 (void)sv_usepvn(sv, dcorig, needlen); /* Uses Renew(). */
1120 SvCUR_set(sv, dc - dcorig);
1121 if (rulen)
1122 sv_catpvn(sv, dcsave, rulen);
1123 else if (lulen)
1124 sv_catpvn(sv, dcsave, lulen);
1125 else
1126 *SvEND(sv) = '\0';
1127 Safefree(dcsave);
1128 break;
1129 default:
1130 if (sv == left || sv == right)
1131 Safefree(dcorig);
1132 Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)",
1133 (unsigned)optype, PL_op_name[optype]);
1134 }
1135 SvUTF8_on(sv);
1136 goto finish;
1137 }
1138 else
1139#ifdef LIBERAL
1140 if (len >= sizeof(long)*4 &&
1141 !((unsigned long)dc % sizeof(long)) &&
1142 !((unsigned long)lc % sizeof(long)) &&
1143 !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */
1144 {
1145 const STRLEN remainder = len % (sizeof(long)*4);
1146 len /= (sizeof(long)*4);
1147
1148 dl = (long*)dc;
1149 ll = (long*)lc;
1150 rl = (long*)rc;
1151
1152 switch (optype) {
1153 case OP_BIT_AND:
1154 while (len--) {
1155 *dl++ = *ll++ & *rl++;
1156 *dl++ = *ll++ & *rl++;
1157 *dl++ = *ll++ & *rl++;
1158 *dl++ = *ll++ & *rl++;
1159 }
1160 break;
1161 case OP_BIT_XOR:
1162 while (len--) {
1163 *dl++ = *ll++ ^ *rl++;
1164 *dl++ = *ll++ ^ *rl++;
1165 *dl++ = *ll++ ^ *rl++;
1166 *dl++ = *ll++ ^ *rl++;
1167 }
1168 break;
1169 case OP_BIT_OR:
1170 while (len--) {
1171 *dl++ = *ll++ | *rl++;
1172 *dl++ = *ll++ | *rl++;
1173 *dl++ = *ll++ | *rl++;
1174 *dl++ = *ll++ | *rl++;
1175 }
1176 }
1177
1178 dc = (char*)dl;
1179 lc = (char*)ll;
1180 rc = (char*)rl;
1181
1182 len = remainder;
1183 }
1184#endif
1185 {
1186 switch (optype) {
1187 case OP_BIT_AND:
1188 while (len--)
1189 *dc++ = *lc++ & *rc++;
1190 *dc = '\0';
1191 break;
1192 case OP_BIT_XOR:
1193 while (len--)
1194 *dc++ = *lc++ ^ *rc++;
1195 goto mop_up;
1196 case OP_BIT_OR:
1197 while (len--)
1198 *dc++ = *lc++ | *rc++;
1199 mop_up:
1200 len = lensave;
1201 if (rightlen > len)
1202 sv_catpvn(sv, rsave + len, rightlen - len);
1203 else if (leftlen > (STRLEN)len)
1204 sv_catpvn(sv, lsave + len, leftlen - len);
1205 else
1206 *SvEND(sv) = '\0';
1207 break;
1208 }
1209 }
1210finish:
1211 SvTAINT(sv);
1212}
1213
1214OP *
1215Perl_do_kv(pTHX)
1216{
1217 dVAR;
1218 dSP;
1219 HV * const hv = MUTABLE_HV(POPs);
1220 HV *keys;
1221 register HE *entry;
1222 const I32 gimme = GIMME_V;
1223 const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV);
1224 /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */
1225 const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS || PL_op->op_type == OP_RKEYS);
1226 const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES || PL_op->op_type == OP_RVALUES);
1227
1228 if (!hv) {
1229 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1230 dTARGET; /* make sure to clear its target here */
1231 if (SvTYPE(TARG) == SVt_PVLV)
1232 LvTARG(TARG) = NULL;
1233 PUSHs(TARG);
1234 }
1235 RETURN;
1236 }
1237
1238 keys = hv;
1239 (void)hv_iterinit(keys); /* always reset iterator regardless */
1240
1241 if (gimme == G_VOID)
1242 RETURN;
1243
1244 if (gimme == G_SCALAR) {
1245 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1246 SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */
1247 sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0);
1248 LvTYPE(ret) = 'k';
1249 LvTARG(ret) = SvREFCNT_inc_simple(keys);
1250 PUSHs(ret);
1251 }
1252 else {
1253 IV i;
1254 dTARGET;
1255
1256 if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) {
1257 i = HvKEYS(keys);
1258 }
1259 else {
1260 i = 0;
1261 while (hv_iternext(keys)) i++;
1262 }
1263 PUSHi( i );
1264 }
1265 RETURN;
1266 }
1267
1268 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1269
1270 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1271 while ((entry = hv_iternext(keys))) {
1272 SPAGAIN;
1273 if (dokeys) {
1274 SV* const sv = hv_iterkeysv(entry);
1275 XPUSHs(sv); /* won't clobber stack_sp */
1276 }
1277 if (dovalues) {
1278 SV *tmpstr;
1279 PUTBACK;
1280 tmpstr = hv_iterval(hv,entry);
1281 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1282 (unsigned long)HeHASH(entry),
1283 (int)HvMAX(keys)+1,
1284 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
1285 SPAGAIN;
1286 XPUSHs(tmpstr);
1287 }
1288 PUTBACK;
1289 }
1290 return NORMAL;
1291}
1292
1293/*
1294 * Local variables:
1295 * c-indentation-style: bsd
1296 * c-basic-offset: 4
1297 * indent-tabs-mode: t
1298 * End:
1299 *
1300 * ex: set ts=8 sts=4 sw=4 noet:
1301 */