This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Clean up -Dy debugging
[perl5.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
1129b882 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
f4b6e4b3 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2009 by Larry Wall and others
79072805
LW
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 *
a0d0e21e
LW
9 */
10
11/*
4ac71550
TC
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"]
79072805
LW
15 */
16
166f8a29 17/* This file contains some common functions needed to carry out certain
347f3823 18 * ops. For example, both pp_sprintf() and pp_prtf() call the function
fc7d284e 19 * do_sprintf() found in this file.
166f8a29
DM
20 */
21
79072805 22#include "EXTERN.h"
864dbfa3 23#define PERL_IN_DOOP_C
79072805 24#include "perl.h"
f34acfec 25#include "invlist_inline.h"
79072805 26
64ca3a65 27#ifndef PERL_MICRO
79072805
LW
28#include <signal.h>
29#endif
30
334c6444
DM
31
32/* Helper function for do_trans().
b8e8e0fc
KW
33 * Handles cases where the search and replacement charlists aren't UTF-8,
34 * aren't identical, and neither the /d nor /s flag is present.
35 *
36 * sv may or may not be utf8. Note that no code point above 255 can possibly
37 * be in the to-translate set
334c6444
DM
38 */
39
f0fd0980 40STATIC Size_t
f534d546 41S_do_trans_simple(pTHX_ SV * const sv, const OPtrans_map * const tbl)
79072805 42{
f0fd0980 43 Size_t matches = 0;
463ee0b2 44 STRLEN len;
9138d6ca 45 U8 *s = (U8*)SvPV_nomg(sv,len);
c395bd6c 46 U8 * const send = s+len;
7918f24d
NC
47
48 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE;
5d7580af
KW
49 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_simple:"
50 " input sv:\n",
51 __FILE__, __LINE__));
52 DEBUG_y(sv_dump(sv));
7918f24d 53
1e54db1a 54 /* First, take care of non-UTF-8 input strings, because they're easy */
1aa99e6b 55 if (!SvUTF8(sv)) {
01ec43d0 56 while (s < send) {
e4ee4c1b 57 const short ch = tbl->map[*s];
f54cb97a 58 if (ch >= 0) {
01ec43d0 59 matches++;
c4420975 60 *s = (U8)ch;
01ec43d0 61 }
c4420975 62 s++;
01ec43d0
GS
63 }
64 SvSETMAGIC(sv);
9b877dbb
IH
65 }
66 else {
e4ee4c1b 67 const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS);
c395bd6c
AL
68 U8 *d;
69 U8 *dstart;
70
b8e8e0fc
KW
71 /* Allow for worst-case expansion: Each input byte can become 2. For a
72 * given input character, this happens when it occupies a single byte
73 * under UTF-8, but is to be translated to something that occupies two:
74 * $_="a".chr(400); tr/a/\xFE/, FE needs encoding. */
c395bd6c
AL
75 if (grows)
76 Newx(d, len*2+1, U8);
77 else
78 d = s;
79 dstart = d;
80 while (s < send) {
81 STRLEN ulen;
e4ee4c1b 82 short ch;
c395bd6c
AL
83
84 /* Need to check this, otherwise 128..255 won't match */
85 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
c1048fcf 86 if (c < 0x100 && (ch = tbl->map[c]) >= 0) {
c395bd6c 87 matches++;
e4ee4c1b 88 d = uvchr_to_utf8(d, (UV)ch);
c395bd6c
AL
89 s += ulen;
90 }
91 else { /* No match -> copy */
92 Move(s, d, ulen, U8);
93 d += ulen;
94 s += ulen;
95 }
96 }
97 if (grows) {
98 sv_setpvn(sv, (char*)dstart, d - dstart);
99 Safefree(dstart);
100 }
101 else {
102 *d = '\0';
103 SvCUR_set(sv, d - dstart);
104 }
105 SvUTF8_on(sv);
106 SvSETMAGIC(sv);
9b877dbb 107 }
f1d561b5 108 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n",
5d7580af
KW
109 __FILE__, __LINE__, matches));
110 DEBUG_y(sv_dump(sv));
4757a243
LW
111 return matches;
112}
113
334c6444
DM
114
115/* Helper function for do_trans().
b8e8e0fc
KW
116 * Handles cases where the search and replacement charlists are identical and
117 * non-utf8: so the string isn't modified, and only a count of modifiable
334c6444 118 * chars is needed.
b8e8e0fc
KW
119 *
120 * Note that it doesn't handle /d or /s, since these modify the string even if
121 * the replacement list is empty.
122 *
123 * sv may or may not be utf8. Note that no code point above 255 can possibly
124 * be in the to-translate set
334c6444
DM
125 */
126
f0fd0980 127STATIC Size_t
f534d546 128S_do_trans_count(pTHX_ SV * const sv, const OPtrans_map * const tbl)
4757a243 129{
4757a243 130 STRLEN len;
9138d6ca 131 const U8 *s = (const U8*)SvPV_nomg_const(sv, len);
c395bd6c 132 const U8 * const send = s + len;
f0fd0980 133 Size_t matches = 0;
7918f24d
NC
134
135 PERL_ARGS_ASSERT_DO_TRANS_COUNT;
136
5d7580af
KW
137 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_count:"
138 " input sv:\n",
139 __FILE__, __LINE__));
140 DEBUG_y(sv_dump(sv));
141
c395bd6c 142 if (!SvUTF8(sv)) {
1aa99e6b 143 while (s < send) {
c1048fcf 144 if (tbl->map[*s++] >= 0)
036b4402 145 matches++;
1aa99e6b 146 }
c395bd6c 147 }
fabdb6c0 148 else {
e4ee4c1b 149 const bool complement = cBOOL(PL_op->op_private & OPpTRANS_COMPLEMENT);
1aa99e6b 150 while (s < send) {
1aa99e6b 151 STRLEN ulen;
9f7f3913 152 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
8973db79 153 if (c < 0x100) {
c1048fcf 154 if (tbl->map[c] >= 0)
8973db79
JH
155 matches++;
156 } else if (complement)
1aa99e6b
IH
157 matches++;
158 s += ulen;
159 }
fabdb6c0 160 }
4757a243 161
f1d561b5 162 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: count returning %zu\n",
5d7580af 163 __FILE__, __LINE__, matches));
4757a243
LW
164 return matches;
165}
166
334c6444
DM
167
168/* Helper function for do_trans().
b8e8e0fc
KW
169 * Handles cases where the search and replacement charlists aren't identical
170 * and both are non-utf8, and one or both of /d, /s is specified.
171 *
172 * sv may or may not be utf8. Note that no code point above 255 can possibly
173 * be in the to-translate set
334c6444
DM
174 */
175
f0fd0980 176STATIC Size_t
f534d546 177S_do_trans_complex(pTHX_ SV * const sv, const OPtrans_map * const tbl)
4757a243 178{
c395bd6c 179 STRLEN len;
9138d6ca 180 U8 *s = (U8*)SvPV_nomg(sv, len);
c395bd6c 181 U8 * const send = s+len;
f0fd0980 182 Size_t matches = 0;
00bd451d 183 const bool complement = cBOOL(PL_op->op_private & OPpTRANS_COMPLEMENT);
7918f24d
NC
184
185 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX;
186
5d7580af
KW
187 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_complex:"
188 " input sv:\n",
189 __FILE__, __LINE__));
190 DEBUG_y(sv_dump(sv));
191
c395bd6c
AL
192 if (!SvUTF8(sv)) {
193 U8 *d = s;
194 U8 * const dstart = d;
4757a243 195
1aa99e6b 196 if (PL_op->op_private & OPpTRANS_SQUASH) {
00bd451d
KW
197
198 /* What the mapping of the previous character was to. If the new
199 * character has the same mapping, it is squashed from the output
200 * (but still is included in the count) */
201 short previous_map = (short) TR_OOB;
202
1aa99e6b 203 while (s < send) {
5e874c42
KW
204 const short this_map = tbl->map[*s];
205 if (this_map >= 0) {
00bd451d
KW
206 matches++;
207 if (this_map != previous_map) {
208 *d++ = (U8)this_map;
209 previous_map = this_map;
210 }
1aa99e6b 211 }
00bd451d
KW
212 else {
213 if (this_map == (short) TR_UNMAPPED)
214 *d++ = *s;
215 else {
216 assert(this_map == (short) TR_DELETE);
217 matches++;
218 }
219 previous_map = (short) TR_OOB;
220 }
221
1aa99e6b
IH
222 s++;
223 }
a0ed51b3 224 }
b8e8e0fc 225 else { /* Not to squash */
1aa99e6b 226 while (s < send) {
5e874c42
KW
227 const short this_map = tbl->map[*s];
228 if (this_map >= 0) {
1aa99e6b 229 matches++;
5e874c42 230 *d++ = (U8)this_map;
1aa99e6b 231 }
5e874c42 232 else if (this_map == (short) TR_UNMAPPED)
1aa99e6b 233 *d++ = *s;
5e874c42 234 else if (this_map == (short) TR_DELETE)
1aa99e6b
IH
235 matches++;
236 s++;
237 }
238 }
76ef7183 239 *d = '\0';
1aa99e6b 240 SvCUR_set(sv, d - dstart);
4757a243 241 }
c395bd6c 242 else { /* is utf8 */
e4ee4c1b
DM
243 const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH);
244 const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS);
c395bd6c
AL
245 U8 *d;
246 U8 *dstart;
0b9a13c3 247 Size_t size = tbl->size;
b8e8e0fc
KW
248
249 /* What the mapping of the previous character was to. If the new
250 * character has the same mapping, it is squashed from the output (but
251 * still is included in the count) */
47279991 252 UV pch = TR_OOB;
fabdb6c0 253
9b877dbb 254 if (grows)
b8e8e0fc
KW
255 /* Allow for worst-case expansion: Each input byte can become 2.
256 * For a given input character, this happens when it occupies a
257 * single byte under UTF-8, but is to be translated to something
258 * that occupies two: */
a02a5408 259 Newx(d, len*2+1, U8);
9b877dbb
IH
260 else
261 d = s;
1aa99e6b
IH
262 dstart = d;
263
4eaf16e8
DM
264 while (s < send) {
265 STRLEN len;
266 const UV comp = utf8n_to_uvchr(s, send - s, &len,
267 UTF8_ALLOW_DEFAULT);
268 UV ch;
269 short sch;
270
00bd451d
KW
271 sch = (comp < size)
272 ? tbl->map[comp]
273 : (! complement)
274 ? (short) TR_UNMAPPED
275 : tbl->map[size];
4eaf16e8
DM
276
277 if (sch >= 0) {
278 ch = (UV)sch;
279 replace:
280 matches++;
281 if (LIKELY(!squash || ch != pch)) {
282 d = uvchr_to_utf8(d, ch);
283 pch = ch;
0b9a13c3 284 }
4eaf16e8
DM
285 s += len;
286 continue;
287 }
482bf615 288 else if (sch == (short) TR_UNMAPPED) {
4eaf16e8
DM
289 Move(s, d, len, U8);
290 d += len;
291 }
482bf615 292 else if (sch == (short) TR_DELETE)
4eaf16e8
DM
293 matches++;
294 else {
482bf615 295 assert(sch == (short) TR_R_EMPTY); /* empty replacement */
4eaf16e8
DM
296 ch = comp;
297 goto replace;
298 }
299
300 s += len;
47279991 301 pch = TR_OOB;
4eaf16e8 302 }
0b9a13c3 303
9b877dbb
IH
304 if (grows) {
305 sv_setpvn(sv, (char*)dstart, d - dstart);
306 Safefree(dstart);
307 }
308 else {
309 *d = '\0';
310 SvCUR_set(sv, d - dstart);
311 }
1aa99e6b 312 SvUTF8_on(sv);
4757a243 313 }
5e44153e 314 SvSETMAGIC(sv);
f1d561b5 315 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n",
5d7580af
KW
316 __FILE__, __LINE__, matches));
317 DEBUG_y(sv_dump(sv));
4757a243
LW
318 return matches;
319}
320
334c6444
DM
321
322/* Helper function for do_trans().
f34acfec
KW
323 * Handles cases where an inversion map implementation is to be used and the
324 * search and replacement charlists are identical: so the string isn't
325 * modified, and only a count of modifiable chars is needed.
326 *
327 * Note that it doesn't handle /d nor /s, since these modify the string
328 * even if the replacement charlist is empty.
329 *
330 * sv may or may not be utf8.
334c6444
DM
331 */
332
f0fd0980 333STATIC Size_t
f34acfec 334S_do_trans_count_invmap(pTHX_ SV * const sv, AV * const invmap)
4757a243 335{
4757a243
LW
336 U8 *s;
337 U8 *send;
f0fd0980 338 Size_t matches = 0;
4757a243 339 STRLEN len;
f34acfec
KW
340 SV** const from_invlist_ptr = av_fetch(invmap, 0, TRUE);
341 SV** const to_invmap_ptr = av_fetch(invmap, 1, TRUE);
342 SV* from_invlist = *from_invlist_ptr;
343 SV* to_invmap_sv = *to_invmap_ptr;
344 UV* map = (UV *) SvPVX(to_invmap_sv);
4757a243 345
f34acfec 346 PERL_ARGS_ASSERT_DO_TRANS_COUNT_INVMAP;
7918f24d 347
5d7580af
KW
348 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d:"
349 "entering do_trans_count_invmap:"
350 " input sv:\n",
351 __FILE__, __LINE__));
352 DEBUG_y(sv_dump(sv));
353 DEBUG_y(PerlIO_printf(Perl_debug_log, "mapping:\n"));
354 DEBUG_y(invmap_dump(from_invlist, (UV *) SvPVX(to_invmap_sv)));
355
9138d6ca 356 s = (U8*)SvPV_nomg(sv, len);
f34acfec 357
4757a243
LW
358 send = s + len;
359
f34acfec
KW
360 while (s < send) {
361 UV from;
362 SSize_t i;
363 STRLEN s_len;
364
365 /* Get the code point of the next character in the string */
366 if (! SvUTF8(sv) || UTF8_IS_INVARIANT(*s)) {
367 from = *s;
368 s_len = 1;
369 }
370 else {
371 from = utf8_to_uvchr_buf(s, send, &s_len);
372 if (from == 0 && *s != '\0') {
373 _force_out_malformed_utf8_message(s, send, 0, /*die*/TRUE);
374 }
375 }
4757a243 376
f34acfec
KW
377 /* Look the code point up in the data structure for this tr/// to get
378 * what it maps to */
379 i = _invlist_search(from_invlist, from);
380 assert(i >= 0);
1aa99e6b 381
f34acfec
KW
382 if (map[i] != (UV) TR_UNLISTED) {
383 matches++;
384 }
385
386 s += s_len;
9b877dbb 387 }
4757a243 388
f1d561b5 389 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n",
5d7580af 390 __FILE__, __LINE__, matches));
4757a243
LW
391 return matches;
392}
393
334c6444 394/* Helper function for do_trans().
f34acfec
KW
395 * Handles cases where an inversion map implementation is to be used and the
396 * search and replacement charlists are either not identical or flags are
397 * present.
398 *
399 * sv may or may not be utf8.
334c6444
DM
400 */
401
f0fd0980 402STATIC Size_t
f34acfec 403S_do_trans_invmap(pTHX_ SV * const sv, AV * const invmap)
4757a243 404{
f34acfec
KW
405 U8 *s;
406 U8 *send;
407 U8 *d;
408 U8 *s0;
409 U8 *d0;
f0fd0980 410 Size_t matches = 0;
4757a243 411 STRLEN len;
f34acfec
KW
412 SV** const from_invlist_ptr = av_fetch(invmap, 0, TRUE);
413 SV** const to_invmap_ptr = av_fetch(invmap, 1, TRUE);
414 SV** const to_expansion_ptr = av_fetch(invmap, 2, TRUE);
415 NV max_expansion = SvNV(*to_expansion_ptr);
416 SV* from_invlist = *from_invlist_ptr;
417 SV* to_invmap_sv = *to_invmap_ptr;
418 UV* map = (UV *) SvPVX(to_invmap_sv);
419 UV previous_map = TR_OOB;
420 const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH);
421 const bool delete_unfound = cBOOL(PL_op->op_private & OPpTRANS_DELETE);
422 bool inplace = ! cBOOL(PL_op->op_private & OPpTRANS_GROWS);
423 const UV* from_array = invlist_array(from_invlist);
8b4eae17 424 UV final_map = TR_OOB;
58aa6738 425 bool out_is_utf8 = cBOOL(SvUTF8(sv));
f34acfec
KW
426 STRLEN s_len;
427
428 PERL_ARGS_ASSERT_DO_TRANS_INVMAP;
429
430 /* A third element in the array indicates that the replacement list was
431 * shorter than the search list, and this element contains the value to use
432 * for the items that don't correspond */
433 if (av_top_index(invmap) >= 3) {
434 SV** const final_map_ptr = av_fetch(invmap, 3, TRUE);
435 SV* const final_map_sv = *final_map_ptr;
436 final_map = SvUV(final_map_sv);
437 }
7918f24d 438
f34acfec
KW
439 /* If there is something in the transliteration that could force the input
440 * to be changed to UTF-8, we don't know if we can do it in place, so
441 * assume cannot */
442 if (! out_is_utf8 && (PL_op->op_private & OPpTRANS_CAN_FORCE_UTF8)) {
443 inplace = FALSE;
444 if (max_expansion < 2) {
445 max_expansion = 2;
446 }
1aa99e6b 447 }
f34acfec
KW
448
449 s = (U8*)SvPV_nomg(sv, len);
5d7580af
KW
450 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: entering do_trans_invmap:"
451 " input sv:\n",
452 __FILE__, __LINE__));
453 DEBUG_y(sv_dump(sv));
454 DEBUG_y(PerlIO_printf(Perl_debug_log, "mapping:\n"));
455 DEBUG_y(invmap_dump(from_invlist, map));
456
4757a243 457 send = s + len;
f34acfec 458 s0 = s;
4757a243 459
f34acfec
KW
460 /* We know by now if there are some possible input strings whose
461 * transliterations are longer than the input. If none can, we just edit
462 * in place. */
463 if (inplace) {
464 d0 = d = s;
465 }
466 else {
467 /* Here, we can't edit in place. We have no idea how much, if any,
468 * this particular input string will grow. However, the compilation
469 * calculated the maximum expansion possible. Use that to allocale
470 * based on the worst case scenario. */
2077aa94 471 Newx(d, (STRLEN) (len * max_expansion + 1 + 1), U8);
f34acfec 472 d0 = d;
4757a243
LW
473 }
474
f34acfec 475 restart:
4757a243 476
f34acfec
KW
477 /* Do the actual transliteration */
478 while (s < send) {
479 UV from;
480 UV to;
481 SSize_t i;
482 STRLEN s_len;
483
484 /* Get the code point of the next character in the string */
485 if (! SvUTF8(sv) || UTF8_IS_INVARIANT(*s)) {
486 from = *s;
487 s_len = 1;
488 }
489 else {
490 from = utf8_to_uvchr_buf(s, send, &s_len);
491 if (from == 0 && *s != '\0') {
492 _force_out_malformed_utf8_message(s, send, 0, /*die*/TRUE);
493 }
494 }
334c6444 495
f34acfec
KW
496 /* Look the code point up in the data structure for this tr/// to get
497 * what it maps to */
498 i = _invlist_search(from_invlist, from);
499 assert(i >= 0);
334c6444 500
f34acfec 501 to = map[i];
7918f24d 502
f34acfec
KW
503 if (to == (UV) TR_UNLISTED) { /* Just copy the unreplaced character */
504 if (UVCHR_IS_INVARIANT(from) || ! out_is_utf8) {
2077aa94 505 *d++ = (U8) from;
f34acfec
KW
506 }
507 else if (SvUTF8(sv)) {
508 Move(s, d, s_len, U8);
509 d += s_len;
510 }
511 else { /* Convert to UTF-8 */
512 append_utf8_from_native_byte(*s, &d);
513 }
7918f24d 514
f34acfec
KW
515 previous_map = to;
516 s += s_len;
517 continue;
c4d5f83a 518 }
4757a243 519
f34acfec
KW
520 /* Everything else is counted as a match */
521 matches++;
4757a243 522
f34acfec
KW
523 if (to == (UV) TR_SPECIAL_HANDLING) {
524 if (delete_unfound) {
525 previous_map = to;
526 s += s_len;
527 continue;
528 }
4757a243 529
f34acfec
KW
530 /* Use the final character in the replacement list */
531 to = final_map;
532 }
533 else { /* Here the input code point is to be remapped. The actual
534 value is offset from the base of this entry */
535 to += from - from_array[i];
536 }
537
538 /* If copying all occurrences, or this is the first occurrence, copy it
539 * to the output */
540 if (! squash || to != previous_map) {
541 if (out_is_utf8) {
542 d = uvchr_to_utf8(d, to);
543 }
544 else {
545 if (to >= 256) { /* If need to convert to UTF-8, restart */
546 out_is_utf8 = TRUE;
547 s = s0;
548 d = d0;
549 matches = 0;
550 goto restart;
551 }
2077aa94 552 *d++ = (U8) to;
f34acfec
KW
553 }
554 }
555
556 previous_map = to;
557 s += s_len;
4757a243 558 }
f34acfec
KW
559
560 s_len = 0;
561 s += s_len;
562 if (! inplace) {
563 sv_setpvn(sv, (char*)d0, d - d0);
ecfd8588 564 Safefree(d0);
9b877dbb
IH
565 }
566 else {
567 *d = '\0';
f34acfec
KW
568 SvCUR_set(sv, d - d0);
569 }
570
571 if (! SvUTF8(sv) && out_is_utf8) {
572 SvUTF8_on(sv);
9b877dbb 573 }
4757a243
LW
574 SvSETMAGIC(sv);
575
f1d561b5 576 DEBUG_y(PerlIO_printf(Perl_debug_log, "%s: %d: returning %zu\n",
5d7580af
KW
577 __FILE__, __LINE__, matches));
578 DEBUG_y(sv_dump(sv));
4757a243
LW
579 return matches;
580}
581
334c6444
DM
582/* Execute a tr//. sv is the value to be translated, while PL_op
583 * should be an OP_TRANS or OP_TRANSR op, whose op_pv field contains a
f34acfec
KW
584 * translation table or whose op_sv field contains an inversion map.
585 *
334c6444
DM
586 * Returns a count of number of characters translated
587 */
588
f0fd0980 589Size_t
864dbfa3 590Perl_do_trans(pTHX_ SV *sv)
4757a243
LW
591{
592 STRLEN len;
e4ee4c1b 593 const U8 flags = PL_op->op_private;
f34acfec
KW
594 bool use_utf8_fcns = cBOOL(flags & OPpTRANS_USE_SVOP);
595 bool identical = cBOOL(flags & OPpTRANS_IDENTICAL);
4757a243 596
7918f24d
NC
597 PERL_ARGS_ASSERT_DO_TRANS;
598
f34acfec 599 if (SvREADONLY(sv) && ! identical) {
a53bfdae 600 Perl_croak_no_modify();
2233f375 601 }
10516c54 602 (void)SvPV_const(sv, len);
4757a243
LW
603 if (!len)
604 return 0;
f34acfec 605 if (! identical) {
4499db73 606 if (!SvPOKp(sv) || SvTHINKFIRST(sv))
b4cc4d79 607 (void)SvPV_force_nomg(sv, len);
2de7b02f 608 (void)SvPOK_only_UTF8(sv);
d59e14db 609 }
4757a243 610
f34acfec
KW
611 if (use_utf8_fcns) {
612 SV* const map =
613#ifdef USE_ITHREADS
614 PAD_SVl(cPADOP->op_padix);
615#else
616 MUTABLE_SV(cSVOP->op_sv);
617#endif
618
619 if (identical) {
620 return do_trans_count_invmap(sv, (AV *) map);
621 }
622 else {
623 return do_trans_invmap(sv, (AV *) map);
624 }
625 }
626 else {
627 const OPtrans_map * const map = (OPtrans_map*)cPVOP->op_pv;
628
629 if (identical) {
630 return do_trans_count(sv, map);
631 }
632 else if (flags & (OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) {
633 return do_trans_complex(sv, map);
634 }
635 else
636 return do_trans_simple(sv, map);
79072805 637 }
79072805
LW
638}
639
640void
5aaab254 641Perl_do_join(pTHX_ SV *sv, SV *delim, SV **mark, SV **sp)
79072805 642{
53c1dcc0 643 SV ** const oldmark = mark;
eb578fdb
KW
644 I32 items = sp - mark;
645 STRLEN len;
463ee0b2 646 STRLEN delimlen;
810a07df 647 const char * const delims = SvPV_const(delim, delimlen);
79072805 648
7918f24d
NC
649 PERL_ARGS_ASSERT_DO_JOIN;
650
79072805
LW
651 mark++;
652 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
862a34c6 653 SvUPGRADE(sv, SVt_PV);
79072805
LW
654 if (SvLEN(sv) < len + items) { /* current length is way too short */
655 while (items-- > 0) {
1426bbf4 656 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
f54cb97a 657 STRLEN tmplen;
5c144d81 658 SvPV_const(*mark, tmplen);
463ee0b2 659 len += tmplen;
79072805
LW
660 }
661 mark++;
662 }
663 SvGROW(sv, len + 1); /* so try to pre-extend */
664
665 mark = oldmark;
db7c17d7 666 items = sp - mark;
79072805
LW
667 ++mark;
668 }
669
f607343f 670 SvPVCLEAR(sv);
e4803c42 671 /* sv_setpv retains old UTF8ness [perl #24846] */
fb622db0 672 SvUTF8_off(sv);
e4803c42 673
284167a5 674 if (TAINTING_get && SvMAGICAL(sv))
8d6d96c1
HS
675 SvTAINTED_off(sv);
676
463ee0b2 677 if (items-- > 0) {
92d29cee
JH
678 if (*mark)
679 sv_catsv(sv, *mark);
463ee0b2
LW
680 mark++;
681 }
8d6d96c1 682
c512ce4f 683 if (delimlen) {
810a07df 684 const U32 delimflag = DO_UTF8(delim) ? SV_CATUTF8 : SV_CATBYTES;
79072805 685 for (; items > 0; items--,mark++) {
f0ee3863
FC
686 STRLEN len;
687 const char *s;
810a07df 688 sv_catpvn_flags(sv,delims,delimlen,delimflag);
f0ee3863
FC
689 s = SvPV_const(*mark,len);
690 sv_catpvn_flags(sv,s,len,
691 DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES);
79072805
LW
692 }
693 }
694 else {
695 for (; items > 0; items--,mark++)
f0ee3863
FC
696 {
697 STRLEN len;
698 const char *s = SvPV_const(*mark,len);
699 sv_catpvn_flags(sv,s,len,
700 DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES);
701 }
79072805
LW
702 }
703 SvSETMAGIC(sv);
704}
705
706void
03a22d83 707Perl_do_sprintf(pTHX_ SV *sv, SSize_t len, SV **sarg)
79072805 708{
46fc3d4c 709 STRLEN patlen;
53c1dcc0 710 const char * const pat = SvPV_const(*sarg, patlen);
46fc3d4c
PP
711 bool do_taint = FALSE;
712
7918f24d 713 PERL_ARGS_ASSERT_DO_SPRINTF;
03a22d83 714 assert(len >= 1);
7918f24d 715
e06d98fb
DM
716 if (SvTAINTED(*sarg))
717 TAINT_PROPER(
718 (PL_op && PL_op->op_type < OP_max)
719 ? (PL_op->op_type == OP_PRTF)
720 ? "printf"
721 : PL_op_name[PL_op->op_type]
722 : "(unknown)"
723 );
5b781b5b 724 SvUTF8_off(sv);
2cf2cfc6
A
725 if (DO_UTF8(*sarg))
726 SvUTF8_on(sv);
03a22d83 727 sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, (Size_t)(len - 1), &do_taint);
79072805 728 SvSETMAGIC(sv);
46fc3d4c
PP
729 if (do_taint)
730 SvTAINTED_on(sv);
79072805
LW
731}
732
33b45480 733/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0 734UV
d69c4304 735Perl_do_vecget(pTHX_ SV *sv, STRLEN offset, int size)
81e118e0 736{
67dd6f35 737 STRLEN srclen, len, avail, uoffset, bitoffs = 0;
fc9668ae
DM
738 const I32 svpv_flags = ((PL_op->op_flags & OPf_MOD || LVRET)
739 ? SV_UNDEF_RETURNS_NULL : 0);
740 unsigned char *s = (unsigned char *)
741 SvPV_flags(sv, srclen, (svpv_flags|SV_GMAGIC));
81e118e0
JH
742 UV retnum = 0;
743
032061d2 744 if (!s) {
fc9668ae 745 s = (unsigned char *)"";
032061d2 746 }
2f96a1b4 747
7918f24d
NC
748 PERL_ARGS_ASSERT_DO_VECGET;
749
8e84507e 750 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 751 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 752
fc9668ae 753 if (SvUTF8(sv)) {
27c41eac 754 if (Perl_sv_utf8_downgrade_flags(aTHX_ sv, TRUE, 0)) {
315f3fc1
KW
755 /* PVX may have changed */
756 s = (unsigned char *) SvPV_flags(sv, srclen, svpv_flags);
757 }
758 else {
da5a0da2 759 Perl_croak(aTHX_ "Use of strings with code points over 0xFF as arguments to vec is forbidden");
315f3fc1 760 }
fc9668ae 761 }
246fae53 762
bbb8a7e0
MHM
763 if (size < 8) {
764 bitoffs = ((offset%8)*size)%8;
765 uoffset = offset/(8/size);
766 }
67dd6f35
DM
767 else if (size > 8) {
768 int n = size/8;
769 if (offset > Size_t_MAX / n - 1) /* would overflow */
770 return 0;
771 uoffset = offset*n;
772 }
bbb8a7e0
MHM
773 else
774 uoffset = offset;
775
67dd6f35
DM
776 if (uoffset >= srclen)
777 return 0;
778
779 len = (bitoffs + size + 7)/8; /* required number of bytes */
780 avail = srclen - uoffset; /* available number of bytes */
781
782 /* Does the byte range overlap the end of the string? If so,
783 * handle specially. */
784 if (avail < len) {
81e118e0
JH
785 if (size <= 8)
786 retnum = 0;
787 else {
81e118e0 788 if (size == 16) {
67dd6f35
DM
789 assert(avail == 1);
790 retnum = (UV) s[uoffset] << 8;
81e118e0
JH
791 }
792 else if (size == 32) {
67dd6f35
DM
793 assert(avail >= 1 && avail <= 3);
794 if (avail == 1)
81e118e0 795 retnum =
bb7a0f54 796 ((UV) s[uoffset ] << 24);
67dd6f35 797 else if (avail == 2)
81e118e0 798 retnum =
bb7a0f54
MHM
799 ((UV) s[uoffset ] << 24) +
800 ((UV) s[uoffset + 1] << 16);
81e118e0
JH
801 else
802 retnum =
bb7a0f54
MHM
803 ((UV) s[uoffset ] << 24) +
804 ((UV) s[uoffset + 1] << 16) +
805 ( s[uoffset + 2] << 8);
81e118e0 806 }
d7d93a81 807#ifdef UV_IS_QUAD
c5a0f51a 808 else if (size == 64) {
a2a5de95
NC
809 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
810 "Bit vector size > 32 non-portable");
67dd6f35
DM
811 assert(avail >= 1 && avail <= 7);
812 if (avail == 1)
c5a0f51a 813 retnum =
bb7a0f54 814 (UV) s[uoffset ] << 56;
67dd6f35 815 else if (avail == 2)
c5a0f51a 816 retnum =
bb7a0f54
MHM
817 ((UV) s[uoffset ] << 56) +
818 ((UV) s[uoffset + 1] << 48);
67dd6f35 819 else if (avail == 3)
c5a0f51a 820 retnum =
bb7a0f54
MHM
821 ((UV) s[uoffset ] << 56) +
822 ((UV) s[uoffset + 1] << 48) +
823 ((UV) s[uoffset + 2] << 40);
67dd6f35 824 else if (avail == 4)
c5a0f51a 825 retnum =
bb7a0f54
MHM
826 ((UV) s[uoffset ] << 56) +
827 ((UV) s[uoffset + 1] << 48) +
828 ((UV) s[uoffset + 2] << 40) +
829 ((UV) s[uoffset + 3] << 32);
67dd6f35 830 else if (avail == 5)
c5a0f51a 831 retnum =
bb7a0f54
MHM
832 ((UV) s[uoffset ] << 56) +
833 ((UV) s[uoffset + 1] << 48) +
834 ((UV) s[uoffset + 2] << 40) +
835 ((UV) s[uoffset + 3] << 32) +
e7aca353 836 ((UV) s[uoffset + 4] << 24);
67dd6f35 837 else if (avail == 6)
c5a0f51a 838 retnum =
bb7a0f54
MHM
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 ((UV) s[uoffset + 5] << 16);
c5a0f51a 845 else
8e84507e 846 retnum =
bb7a0f54
MHM
847 ((UV) s[uoffset ] << 56) +
848 ((UV) s[uoffset + 1] << 48) +
849 ((UV) s[uoffset + 2] << 40) +
850 ((UV) s[uoffset + 3] << 32) +
851 ((UV) s[uoffset + 4] << 24) +
852 ((UV) s[uoffset + 5] << 16) +
e7aca353 853 ((UV) s[uoffset + 6] << 8);
c5a0f51a
JH
854 }
855#endif
81e118e0
JH
856 }
857 }
858 else if (size < 8)
bbb8a7e0 859 retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1);
81e118e0 860 else {
81e118e0 861 if (size == 8)
bb7a0f54 862 retnum = s[uoffset];
81e118e0
JH
863 else if (size == 16)
864 retnum =
bb7a0f54
MHM
865 ((UV) s[uoffset] << 8) +
866 s[uoffset + 1];
81e118e0
JH
867 else if (size == 32)
868 retnum =
bb7a0f54
MHM
869 ((UV) s[uoffset ] << 24) +
870 ((UV) s[uoffset + 1] << 16) +
871 ( s[uoffset + 2] << 8) +
872 s[uoffset + 3];
d7d93a81 873#ifdef UV_IS_QUAD
c5a0f51a 874 else if (size == 64) {
a2a5de95
NC
875 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
876 "Bit vector size > 32 non-portable");
c5a0f51a 877 retnum =
bb7a0f54
MHM
878 ((UV) s[uoffset ] << 56) +
879 ((UV) s[uoffset + 1] << 48) +
880 ((UV) s[uoffset + 2] << 40) +
881 ((UV) s[uoffset + 3] << 32) +
882 ((UV) s[uoffset + 4] << 24) +
883 ((UV) s[uoffset + 5] << 16) +
884 ( s[uoffset + 6] << 8) +
885 s[uoffset + 7];
c5a0f51a
JH
886 }
887#endif
81e118e0
JH
888 }
889
890 return retnum;
891}
892
33b45480
SB
893/* currently converts input to bytes if possible but doesn't sweat failures,
894 * although it does ensure that the string it clobbers is not marked as
895 * utf8-valid any more
896 */
79072805 897void
864dbfa3 898Perl_do_vecset(pTHX_ SV *sv)
79072805 899{
67dd6f35 900 STRLEN offset, bitoffs = 0;
eb578fdb
KW
901 int size;
902 unsigned char *s;
903 UV lval;
79072805 904 I32 mask;
a0d0e21e
LW
905 STRLEN targlen;
906 STRLEN len;
c4420975 907 SV * const targ = LvTARG(sv);
1b92e694 908 char errflags = LvFLAGS(sv);
79072805 909
7918f24d
NC
910 PERL_ARGS_ASSERT_DO_VECSET;
911
1b92e694
DM
912 /* some out-of-range errors have been deferred if/until the LV is
913 * actually written to: f(vec($s,-1,8)) is not always fatal */
914 if (errflags) {
b063b0a8
DIM
915 assert(!(errflags & ~(LVf_NEG_OFF|LVf_OUT_OF_RANGE)));
916 if (errflags & LVf_NEG_OFF)
1b92e694
DM
917 Perl_croak_nocontext("Negative offset to vec in lvalue context");
918 Perl_croak_nocontext("Out of memory!");
919 }
920
8990e307
LW
921 if (!targ)
922 return;
032061d2
BF
923 s = (unsigned char*)SvPV_force_flags(targ, targlen,
924 SV_GMAGIC | SV_UNDEF_RETURNS_NULL);
246fae53 925 if (SvUTF8(targ)) {
33b45480 926 /* This is handled by the SvPOK_only below...
27c41eac 927 if (!Perl_sv_utf8_downgrade_flags(aTHX_ targ, TRUE, 0))
33b45480
SB
928 SvUTF8_off(targ);
929 */
27c41eac 930 (void) Perl_sv_utf8_downgrade_flags(aTHX_ targ, TRUE, 0);
246fae53
MG
931 }
932
4ebbc975 933 (void)SvPOK_only(targ);
81e118e0 934 lval = SvUV(sv);
79072805
LW
935 offset = LvTARGOFF(sv);
936 size = LvTARGLEN(sv);
67dd6f35 937
8e84507e 938 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 939 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 940
bbb8a7e0
MHM
941 if (size < 8) {
942 bitoffs = ((offset%8)*size)%8;
943 offset /= 8/size;
944 }
67dd6f35
DM
945 else if (size > 8) {
946 int n = size/8;
947 if (offset > Size_t_MAX / n - 1) /* would overflow */
948 Perl_croak_nocontext("Out of memory!");
949 offset *= n;
950 }
bbb8a7e0 951
67dd6f35
DM
952 len = (bitoffs + size + 7)/8; /* required number of bytes */
953 if (targlen < offset || targlen - offset < len) {
954 STRLEN newlen = offset > Size_t_MAX - len - 1 ? /* avoid overflow */
955 Size_t_MAX : offset + len + 1;
956 s = (unsigned char*)SvGROW(targ, newlen);
957 (void)memzero((char *)(s + targlen), newlen - targlen);
958 SvCUR_set(targ, newlen - 1);
a0d0e21e 959 }
8e84507e 960
79072805
LW
961 if (size < 8) {
962 mask = (1 << size) - 1;
79072805 963 lval &= mask;
bbb8a7e0
MHM
964 s[offset] &= ~(mask << bitoffs);
965 s[offset] |= lval << bitoffs;
79072805
LW
966 }
967 else {
968 if (size == 8)
eb160463 969 s[offset ] = (U8)( lval & 0xff);
79072805 970 else if (size == 16) {
eb160463
GS
971 s[offset ] = (U8)((lval >> 8) & 0xff);
972 s[offset+1] = (U8)( lval & 0xff);
79072805
LW
973 }
974 else if (size == 32) {
eb160463
GS
975 s[offset ] = (U8)((lval >> 24) & 0xff);
976 s[offset+1] = (U8)((lval >> 16) & 0xff);
977 s[offset+2] = (U8)((lval >> 8) & 0xff);
978 s[offset+3] = (U8)( lval & 0xff);
c5a0f51a 979 }
d7d93a81 980#ifdef UV_IS_QUAD
c5a0f51a 981 else if (size == 64) {
a2a5de95
NC
982 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
983 "Bit vector size > 32 non-portable");
eb160463
GS
984 s[offset ] = (U8)((lval >> 56) & 0xff);
985 s[offset+1] = (U8)((lval >> 48) & 0xff);
986 s[offset+2] = (U8)((lval >> 40) & 0xff);
987 s[offset+3] = (U8)((lval >> 32) & 0xff);
988 s[offset+4] = (U8)((lval >> 24) & 0xff);
989 s[offset+5] = (U8)((lval >> 16) & 0xff);
990 s[offset+6] = (U8)((lval >> 8) & 0xff);
991 s[offset+7] = (U8)( lval & 0xff);
79072805 992 }
dc1e3f56 993#endif
79072805 994 }
7bb043c3 995 SvSETMAGIC(targ);
79072805
LW
996}
997
998void
864dbfa3 999Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805 1000{
eb578fdb
KW
1001 long *dl;
1002 long *ll;
1003 long *rl;
eb578fdb 1004 char *dc;
463ee0b2
LW
1005 STRLEN leftlen;
1006 STRLEN rightlen;
eb578fdb
KW
1007 const char *lc;
1008 const char *rc;
b404a7f5 1009 STRLEN len = 0;
bb7a0f54 1010 STRLEN lensave;
e62f0680
NC
1011 const char *lsave;
1012 const char *rsave;
bb7a0f54 1013 STRLEN needlen = 0;
08b6664b 1014 bool result_needs_to_be_utf8 = FALSE;
b50535da
KW
1015 bool left_utf8 = FALSE;
1016 bool right_utf8 = FALSE;
1017 U8 * left_non_downgraded = NULL;
1018 U8 * right_non_downgraded = NULL;
1019 Size_t left_non_downgraded_len = 0;
1020 Size_t right_non_downgraded_len = 0;
1021 char * non_downgraded = NULL;
1022 Size_t non_downgraded_len = 0;
0c57e439 1023
7918f24d 1024 PERL_ARGS_ASSERT_DO_VOP;
79072805 1025
6b349a5c 1026 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv)))
f607343f 1027 SvPVCLEAR(sv); /* avoid undef warning on |= and ^= */
8c8eee82 1028 if (sv == left) {
08b6664b 1029 lc = SvPV_force_nomg(left, leftlen);
8c8eee82
BM
1030 }
1031 else {
08b6664b 1032 lc = SvPV_nomg_const(left, leftlen);
8c8eee82
BM
1033 SvPV_force_nomg_nolen(sv);
1034 }
08b6664b 1035 rc = SvPV_nomg_const(right, rightlen);
12abf4f0 1036
0e6e7171 1037 /* This needs to come after SvPV to ensure that string overloading has
12abf4f0
NC
1038 fired off. */
1039
08b6664b
KW
1040 /* Create downgraded temporaries of any UTF-8 encoded operands */
1041 if (DO_UTF8(left)) {
b50535da 1042 const U8 * save_lc = (U8 *) lc;
08b6664b 1043
b50535da 1044 left_utf8 = TRUE;
08b6664b
KW
1045 result_needs_to_be_utf8 = TRUE;
1046
b50535da
KW
1047 left_non_downgraded_len = leftlen;
1048 lc = (char *) bytes_from_utf8_loc((const U8 *) lc, &leftlen,
1049 &left_utf8,
1050 (const U8 **) &left_non_downgraded);
1051 /* Calculate the number of trailing unconvertible bytes. This quantity
1052 * is the original length minus the length of the converted portion. */
1053 left_non_downgraded_len -= left_non_downgraded - save_lc;
1054 SAVEFREEPV(lc);
12abf4f0 1055 }
08b6664b 1056 if (DO_UTF8(right)) {
b50535da 1057 const U8 * save_rc = (U8 *) rc;
08b6664b 1058
b50535da 1059 right_utf8 = TRUE;
08b6664b
KW
1060 result_needs_to_be_utf8 = TRUE;
1061
b50535da
KW
1062 right_non_downgraded_len = rightlen;
1063 rc = (char *) bytes_from_utf8_loc((const U8 *) rc, &rightlen,
1064 &right_utf8,
1065 (const U8 **) &right_non_downgraded);
1066 right_non_downgraded_len -= right_non_downgraded - save_rc;
1067 SAVEFREEPV(rc);
1068 }
1069
1070 /* We set 'len' to the length that the operation actually operates on. The
1071 * dangling part of the longer operand doesn't actually participate in the
1072 * operation. What happens is that we pretend that the shorter operand has
1073 * been extended to the right by enough imaginary zeros to match the length
1074 * of the longer one. But we know in advance the result of the operation
1075 * on zeros without having to do it. In the case of '&', the result is
1076 * zero, and the dangling portion is simply discarded. For '|' and '^', the
1077 * result is the same as the other operand, so the dangling part is just
c8b94fe0
JK
1078 * appended to the final result, unchanged. As of perl-5.32, we no longer
1079 * accept above-FF code points in the dangling portion.
1080 */
ba52ce15 1081 if (left_utf8 || right_utf8) {
c8b94fe0 1082 Perl_croak(aTHX_ FATAL_ABOVE_FF_MSG, PL_op_desc[optype]);
ba52ce15 1083 }
b50535da 1084 else { /* Neither is UTF-8 */
78ba9007 1085 len = MIN(leftlen, rightlen);
12abf4f0
NC
1086 }
1087
b50535da 1088 lensave = len;
08b6664b
KW
1089 lsave = lc;
1090 rsave = rc;
b50535da 1091
9fdd7463
JH
1092 SvCUR_set(sv, len);
1093 (void)SvPOK_only(sv);
08b6664b 1094 if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2596d9fe 1095 dc = SvPV_force_nomg_nolen(sv);
bb7a0f54
MHM
1096 if (SvLEN(sv) < len + 1) {
1097 dc = SvGROW(sv, len + 1);
ff68c719
PP
1098 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1099 }
1100 }
1101 else {
bb7a0f54
MHM
1102 needlen = optype == OP_BIT_AND
1103 ? len : (leftlen > rightlen ? leftlen : rightlen);
a02a5408 1104 Newxz(dc, needlen + 1, char);
aa0a69cb 1105 sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
ff68c719 1106 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 1107 }
0c57e439 1108
79072805 1109 if (len >= sizeof(long)*4 &&
d398c6bf
TK
1110 !(PTR2nat(dc) % sizeof(long)) &&
1111 !(PTR2nat(lc) % sizeof(long)) &&
1112 !(PTR2nat(rc) % sizeof(long))) /* It's almost always aligned... */
79072805 1113 {
bb7a0f54 1114 const STRLEN remainder = len % (sizeof(long)*4);
79072805
LW
1115 len /= (sizeof(long)*4);
1116
1117 dl = (long*)dc;
1118 ll = (long*)lc;
1119 rl = (long*)rc;
1120
1121 switch (optype) {
1122 case OP_BIT_AND:
1123 while (len--) {
1124 *dl++ = *ll++ & *rl++;
1125 *dl++ = *ll++ & *rl++;
1126 *dl++ = *ll++ & *rl++;
1127 *dl++ = *ll++ & *rl++;
1128 }
1129 break;
a0d0e21e 1130 case OP_BIT_XOR:
79072805
LW
1131 while (len--) {
1132 *dl++ = *ll++ ^ *rl++;
1133 *dl++ = *ll++ ^ *rl++;
1134 *dl++ = *ll++ ^ *rl++;
1135 *dl++ = *ll++ ^ *rl++;
1136 }
1137 break;
1138 case OP_BIT_OR:
1139 while (len--) {
1140 *dl++ = *ll++ | *rl++;
1141 *dl++ = *ll++ | *rl++;
1142 *dl++ = *ll++ | *rl++;
1143 *dl++ = *ll++ | *rl++;
1144 }
1145 }
1146
1147 dc = (char*)dl;
1148 lc = (char*)ll;
1149 rc = (char*)rl;
1150
1151 len = remainder;
1152 }
17d44595 1153
27a9d47d
KW
1154 switch (optype) {
1155 case OP_BIT_AND:
1156 while (len--)
1157 *dc++ = *lc++ & *rc++;
1158 *dc = '\0';
1159 break;
1160 case OP_BIT_XOR:
1161 while (len--)
1162 *dc++ = *lc++ ^ *rc++;
1163 goto mop_up;
1164 case OP_BIT_OR:
1165 while (len--)
1166 *dc++ = *lc++ | *rc++;
1167 mop_up:
1168 len = lensave;
1169 if (rightlen > len) {
1170 if (dc == rc)
2324bdb9 1171 SvCUR_set(sv, rightlen);
27a9d47d
KW
1172 else
1173 sv_catpvn_nomg(sv, rsave + len, rightlen - len);
1174 }
1175 else if (leftlen > len) {
1176 if (dc == lc)
2324bdb9 1177 SvCUR_set(sv, leftlen);
27a9d47d
KW
1178 else
1179 sv_catpvn_nomg(sv, lsave + len, leftlen - len);
1180 }
1181 *SvEND(sv) = '\0';
392582f8 1182
b50535da
KW
1183 /* If there is trailing stuff that couldn't be converted from UTF-8, it
1184 * is appended as-is for the ^ and | operators. This preserves
1185 * backwards compatibility */
1186 if (right_non_downgraded) {
1187 non_downgraded = (char *) right_non_downgraded;
1188 non_downgraded_len = right_non_downgraded_len;
1189 }
1190 else if (left_non_downgraded) {
1191 non_downgraded = (char *) left_non_downgraded;
1192 non_downgraded_len = left_non_downgraded_len;
1193 }
1194
27a9d47d
KW
1195 break;
1196 }
08b6664b
KW
1197
1198 if (result_needs_to_be_utf8) {
b50535da
KW
1199 sv_utf8_upgrade_nomg(sv);
1200
1201 /* Append any trailing UTF-8 as-is. */
1202 if (non_downgraded) {
1203 sv_catpvn_nomg(sv, non_downgraded, non_downgraded_len);
1204 }
79072805 1205 }
08b6664b 1206
fb73857a 1207 SvTAINT(sv);
79072805 1208}
463ee0b2 1209
b1c05ba5 1210
a2232057
DM
1211/* Perl_do_kv() may be:
1212 * * called directly as the pp function for pp_keys() and pp_values();
a2232057
DM
1213 * * It may also be called directly when the op is OP_AVHVSWITCH, to
1214 * implement CORE::keys(), CORE::values().
1215 *
1216 * In all cases it expects an HV on the stack and returns a list of keys,
1217 * values, or key-value pairs, depending on PL_op.
1218 */
b1c05ba5 1219
463ee0b2 1220OP *
cea2e8a9 1221Perl_do_kv(pTHX)
463ee0b2 1222{
39644a26 1223 dSP;
73ff03e8 1224 HV * const keys = MUTABLE_HV(POPs);
1c23e2bd 1225 const U8 gimme = GIMME_V;
a2232057 1226
af3b1cba 1227 const I32 dokeys = (PL_op->op_type == OP_KEYS)
a2232057 1228 || ( PL_op->op_type == OP_AVHVSWITCH
94184451
DM
1229 && (PL_op->op_private & OPpAVHVSWITCH_MASK)
1230 + OP_EACH == OP_KEYS);
a2232057 1231
af3b1cba 1232 const I32 dovalues = (PL_op->op_type == OP_VALUES)
a2232057 1233 || ( PL_op->op_type == OP_AVHVSWITCH
94184451
DM
1234 && (PL_op->op_private & OPpAVHVSWITCH_MASK)
1235 + OP_EACH == OP_VALUES);
a2232057 1236
af3b1cba 1237 assert( PL_op->op_type == OP_KEYS
a2232057
DM
1238 || PL_op->op_type == OP_VALUES
1239 || PL_op->op_type == OP_AVHVSWITCH);
463ee0b2 1240
4fa080db
DM
1241 assert(!( PL_op->op_type == OP_VALUES
1242 && (PL_op->op_private & OPpMAYBE_LVSUB)));
1243
800e9ae0 1244 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1245
54310121 1246 if (gimme == G_VOID)
aa689395
PP
1247 RETURN;
1248
54310121 1249 if (gimme == G_SCALAR) {
78f9721b 1250 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
2154eca7
EB
1251 SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */
1252 sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0);
1253 LvTYPE(ret) = 'k';
1254 LvTARG(ret) = SvREFCNT_inc_simple(keys);
1255 PUSHs(ret);
81714fb9 1256 }
463ee0b2 1257 else {
2154eca7
EB
1258 IV i;
1259 dTARGET;
1260
af3b1cba
DM
1261 /* note that in 'scalar(keys %h)' the OP_KEYS is usually
1262 * optimised away and the action is performed directly by the
1263 * padhv or rv2hv op. We now only get here via OP_AVHVSWITCH
1264 * and \&CORE::keys
1265 */
2154eca7 1266 if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) {
1b95d04f 1267 i = HvUSEDKEYS(keys);
2154eca7
EB
1268 }
1269 else {
1270 i = 0;
1271 while (hv_iternext(keys)) i++;
1272 }
1273 PUSHi( i );
463ee0b2 1274 }
463ee0b2
LW
1275 RETURN;
1276 }
1277
a061ab0b
FC
1278 if (UNLIKELY(PL_op->op_private & OPpMAYBE_LVSUB)) {
1279 const I32 flags = is_lvalue_sub();
1280 if (flags && !(flags & OPpENTERSUB_INARGS))
1281 /* diag_listed_as: Can't modify %s in %s */
1282 Perl_croak(aTHX_ "Can't modify keys in list assignment");
1283 }
1284
8dc9003f
DM
1285 PUTBACK;
1286 hv_pushkv(keys, (dokeys | (dovalues << 1)));
1287 return NORMAL;
463ee0b2 1288}
4e35701f 1289
af3babe4 1290/*
14d04a33 1291 * ex: set ts=8 sts=4 sw=4 et:
37442d52 1292 */