This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Unicode::UCD: Add prop_invlist()
[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
DM
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
79072805 22#include "EXTERN.h"
864dbfa3 23#define PERL_IN_DOOP_C
79072805
LW
24#include "perl.h"
25
64ca3a65 26#ifndef PERL_MICRO
79072805
LW
27#include <signal.h>
28#endif
29
942e002e 30STATIC I32
c395bd6c 31S_do_trans_simple(pTHX_ SV * const sv)
79072805 32{
97aff369 33 dVAR;
4757a243 34 I32 matches = 0;
463ee0b2 35 STRLEN len;
9138d6ca 36 U8 *s = (U8*)SvPV_nomg(sv,len);
c395bd6c 37 U8 * const send = s+len;
c4420975 38 const short * const tbl = (short*)cPVOP->op_pv;
7918f24d
NC
39
40 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE;
41
4757a243 42 if (!tbl)
7d85a32c 43 Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
a0ed51b3 44
1e54db1a 45 /* First, take care of non-UTF-8 input strings, because they're easy */
1aa99e6b 46 if (!SvUTF8(sv)) {
01ec43d0 47 while (s < send) {
f54cb97a
AL
48 const I32 ch = tbl[*s];
49 if (ch >= 0) {
01ec43d0 50 matches++;
c4420975 51 *s = (U8)ch;
01ec43d0 52 }
c4420975 53 s++;
01ec43d0
GS
54 }
55 SvSETMAGIC(sv);
9b877dbb
IH
56 }
57 else {
c395bd6c
AL
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);
9b877dbb 95 }
4757a243
LW
96 return matches;
97}
98
942e002e 99STATIC I32
c395bd6c 100S_do_trans_count(pTHX_ SV * const sv)
4757a243 101{
97aff369 102 dVAR;
4757a243 103 STRLEN len;
9138d6ca 104 const U8 *s = (const U8*)SvPV_nomg_const(sv, len);
c395bd6c
AL
105 const U8 * const send = s + len;
106 I32 matches = 0;
f54cb97a 107 const short * const tbl = (short*)cPVOP->op_pv;
7918f24d
NC
108
109 PERL_ARGS_ASSERT_DO_TRANS_COUNT;
110
4757a243 111 if (!tbl)
7d85a32c 112 Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
4757a243 113
c395bd6c 114 if (!SvUTF8(sv)) {
1aa99e6b
IH
115 while (s < send) {
116 if (tbl[*s++] >= 0)
036b4402 117 matches++;
1aa99e6b 118 }
c395bd6c 119 }
fabdb6c0
AL
120 else {
121 const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
1aa99e6b 122 while (s < send) {
1aa99e6b 123 STRLEN ulen;
9f7f3913 124 const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT);
8973db79
JH
125 if (c < 0x100) {
126 if (tbl[c] >= 0)
127 matches++;
128 } else if (complement)
1aa99e6b
IH
129 matches++;
130 s += ulen;
131 }
fabdb6c0 132 }
4757a243
LW
133
134 return matches;
135}
136
942e002e 137STATIC I32
c395bd6c 138S_do_trans_complex(pTHX_ SV * const sv)
4757a243 139{
97aff369 140 dVAR;
c395bd6c 141 STRLEN len;
9138d6ca 142 U8 *s = (U8*)SvPV_nomg(sv, len);
c395bd6c 143 U8 * const send = s+len;
4757a243 144 I32 matches = 0;
f54cb97a 145 const short * const tbl = (short*)cPVOP->op_pv;
7918f24d
NC
146
147 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX;
148
4757a243 149 if (!tbl)
7d85a32c 150 Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
4757a243 151
c395bd6c
AL
152 if (!SvUTF8(sv)) {
153 U8 *d = s;
154 U8 * const dstart = d;
4757a243 155
1aa99e6b 156 if (PL_op->op_private & OPpTRANS_SQUASH) {
f54cb97a 157 const U8* p = send;
1aa99e6b 158 while (s < send) {
f54cb97a
AL
159 const I32 ch = tbl[*s];
160 if (ch >= 0) {
eb160463 161 *d = (U8)ch;
036b4402 162 matches++;
1aa99e6b
IH
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 }
a0ed51b3 172 }
1aa99e6b
IH
173 else {
174 while (s < send) {
f54cb97a
AL
175 const I32 ch = tbl[*s];
176 if (ch >= 0) {
1aa99e6b 177 matches++;
eb160463 178 *d++ = (U8)ch;
1aa99e6b
IH
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 }
76ef7183 187 *d = '\0';
1aa99e6b 188 SvCUR_set(sv, d - dstart);
4757a243 189 }
c395bd6c 190 else { /* is utf8 */
fabdb6c0
AL
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;
c395bd6c
AL
194 U8 *d;
195 U8 *dstart;
196 STRLEN rlen = 0;
fabdb6c0 197
9b877dbb 198 if (grows)
a02a5408 199 Newx(d, len*2+1, U8);
9b877dbb
IH
200 else
201 d = s;
1aa99e6b 202 dstart = d;
8973db79
JH
203 if (complement && !del)
204 rlen = tbl[0x100];
1aa99e6b
IH
205
206 if (PL_op->op_private & OPpTRANS_SQUASH) {
1aa99e6b
IH
207 UV pch = 0xfeedface;
208 while (s < send) {
209 STRLEN len;
9f7f3913
TS
210 const UV comp = utf8n_to_uvchr(s, send - s, &len,
211 UTF8_ALLOW_DEFAULT);
f54cb97a 212 I32 ch;
1aa99e6b 213
8973db79
JH
214 if (comp > 0xff) {
215 if (!complement) {
640af7f1 216 Move(s, d, len, U8);
8973db79
JH
217 d += len;
218 }
219 else {
220 matches++;
221 if (!del) {
bb7a0f54 222 ch = (rlen == 0) ? (I32)comp :
05d340b8 223 (comp - 0x100 < rlen) ?
8973db79 224 tbl[comp+1] : tbl[0x100+rlen];
eb160463 225 if ((UV)ch != pch) {
9041c2e3 226 d = uvchr_to_utf8(d, ch);
eb160463 227 pch = (UV)ch;
8973db79
JH
228 }
229 s += len;
230 continue;
231 }
232 }
9b877dbb 233 }
1aa99e6b
IH
234 else if ((ch = tbl[comp]) >= 0) {
235 matches++;
eb160463 236 if ((UV)ch != pch) {
9041c2e3 237 d = uvchr_to_utf8(d, ch);
eb160463 238 pch = (UV)ch;
1aa99e6b
IH
239 }
240 s += len;
241 continue;
242 }
9b877dbb 243 else if (ch == -1) { /* -1 is unmapped character */
640af7f1 244 Move(s, d, len, U8);
9b877dbb
IH
245 d += len;
246 }
1aa99e6b
IH
247 else if (ch == -2) /* -2 is delete character */
248 matches++;
249 s += len;
250 pch = 0xfeedface;
251 }
5d06d08e 252 }
1aa99e6b
IH
253 else {
254 while (s < send) {
255 STRLEN len;
9f7f3913
TS
256 const UV comp = utf8n_to_uvchr(s, send - s, &len,
257 UTF8_ALLOW_DEFAULT);
f54cb97a 258 I32 ch;
8973db79
JH
259 if (comp > 0xff) {
260 if (!complement) {
0376ff32 261 Move(s, d, len, U8);
8973db79
JH
262 d += len;
263 }
264 else {
265 matches++;
266 if (!del) {
9041c2e3
NIS
267 if (comp - 0x100 < rlen)
268 d = uvchr_to_utf8(d, tbl[comp+1]);
8973db79 269 else
9041c2e3 270 d = uvchr_to_utf8(d, tbl[0x100+rlen]);
8973db79
JH
271 }
272 }
9b877dbb 273 }
1aa99e6b 274 else if ((ch = tbl[comp]) >= 0) {
9041c2e3 275 d = uvchr_to_utf8(d, ch);
1aa99e6b
IH
276 matches++;
277 }
278 else if (ch == -1) { /* -1 is unmapped character */
640af7f1 279 Move(s, d, len, U8);
9b877dbb 280 d += len;
1aa99e6b
IH
281 }
282 else if (ch == -2) /* -2 is delete character */
283 matches++;
284 s += len;
285 }
286 }
9b877dbb
IH
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 }
1aa99e6b 295 SvUTF8_on(sv);
4757a243 296 }
5e44153e 297 SvSETMAGIC(sv);
4757a243
LW
298 return matches;
299}
300
942e002e 301STATIC I32
c395bd6c 302S_do_trans_simple_utf8(pTHX_ SV * const sv)
4757a243 303{
97aff369 304 dVAR;
4757a243
LW
305 U8 *s;
306 U8 *send;
307 U8 *d;
036b4402 308 U8 *start;
1aa99e6b 309 U8 *dstart, *dend;
4757a243 310 I32 matches = 0;
f54cb97a 311 const I32 grows = PL_op->op_private & OPpTRANS_GROWS;
4757a243 312 STRLEN len;
043e41b8
DM
313 SV* const rv =
314#ifdef USE_ITHREADS
315 PAD_SVl(cPADOP->op_padix);
316#else
ad64d0ec 317 MUTABLE_SV(cSVOP->op_sv);
043e41b8 318#endif
85fbaab2 319 HV* const hv = MUTABLE_HV(SvRV(rv));
a4fc7abc 320 SV* const * svp = hv_fetchs(hv, "NONE", FALSE);
f54cb97a
AL
321 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
322 const UV extra = none + 1;
b7953727 323 UV final = 0;
1aa99e6b 324 U8 hibit = 0;
4757a243 325
7918f24d
NC
326 PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8;
327
9138d6ca 328 s = (U8*)SvPV_nomg(sv, len);
c395bd6c 329 if (!SvUTF8(sv)) {
c4420975
AL
330 const U8 *t = s;
331 const U8 * const e = s + len;
c4d5f83a 332 while (t < e) {
f54cb97a 333 const U8 ch = *t++;
fabdb6c0
AL
334 hibit = !NATIVE_IS_INVARIANT(ch);
335 if (hibit) {
336 s = bytes_to_utf8(s, &len);
1aa99e6b 337 break;
fabdb6c0 338 }
c4d5f83a 339 }
1aa99e6b 340 }
4757a243 341 send = s + len;
036b4402 342 start = s;
4757a243 343
a4fc7abc 344 svp = hv_fetchs(hv, "FINAL", FALSE);
4757a243
LW
345 if (svp)
346 final = SvUV(*svp);
347
9b877dbb
IH
348 if (grows) {
349 /* d needs to be bigger than s, in case e.g. upgrading is required */
a02a5408 350 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
9b877dbb
IH
351 dend = d + len * 3;
352 dstart = d;
353 }
354 else {
355 dstart = d = s;
356 dend = d + len;
357 }
1aa99e6b 358
4757a243 359 while (s < send) {
0bcc34c2
AL
360 const UV uv = swash_fetch(rv, s, TRUE);
361 if (uv < none) {
4757a243
LW
362 s += UTF8SKIP(s);
363 matches++;
7d85a32c 364 d = uvuni_to_utf8(d, uv);
4757a243
LW
365 }
366 else if (uv == none) {
4373e329 367 const int i = UTF8SKIP(s);
0376ff32 368 Move(s, d, i, U8);
9b877dbb
IH
369 d += i;
370 s += i;
4757a243
LW
371 }
372 else if (uv == extra) {
4373e329 373 s += UTF8SKIP(s);
4757a243 374 matches++;
7d85a32c 375 d = uvuni_to_utf8(d, final);
4757a243
LW
376 }
377 else
378 s += UTF8SKIP(s);
1aa99e6b 379
9b877dbb 380 if (d > dend) {
4373e329
AL
381 const STRLEN clen = d - dstart;
382 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
9b877dbb 383 if (!grows)
7d85a32c 384 Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
89ebb4a3 385 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
1aa99e6b
IH
386 d = dstart + clen;
387 dend = dstart + nlen;
388 }
4757a243 389 }
16ec844d 390 if (grows || hibit) {
9b877dbb
IH
391 sv_setpvn(sv, (char*)dstart, d - dstart);
392 Safefree(dstart);
16ec844d
IH
393 if (grows && hibit)
394 Safefree(start);
9b877dbb
IH
395 }
396 else {
397 *d = '\0';
398 SvCUR_set(sv, d - dstart);
399 }
4757a243 400 SvSETMAGIC(sv);
1aa99e6b 401 SvUTF8_on(sv);
4757a243
LW
402
403 return matches;
404}
405
942e002e 406STATIC I32
c395bd6c 407S_do_trans_count_utf8(pTHX_ SV * const sv)
4757a243 408{
97aff369 409 dVAR;
93524f2b 410 const U8 *s;
cbbf8932
AL
411 const U8 *start = NULL;
412 const U8 *send;
4757a243
LW
413 I32 matches = 0;
414 STRLEN len;
043e41b8
DM
415 SV* const rv =
416#ifdef USE_ITHREADS
417 PAD_SVl(cPADOP->op_padix);
418#else
ad64d0ec 419 MUTABLE_SV(cSVOP->op_sv);
043e41b8 420#endif
85fbaab2 421 HV* const hv = MUTABLE_HV(SvRV(rv));
a4fc7abc 422 SV* const * const svp = hv_fetchs(hv, "NONE", FALSE);
f54cb97a
AL
423 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
424 const UV extra = none + 1;
1aa99e6b 425 U8 hibit = 0;
4757a243 426
7918f24d
NC
427 PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8;
428
9138d6ca 429 s = (const U8*)SvPV_nomg_const(sv, len);
1aa99e6b 430 if (!SvUTF8(sv)) {
4373e329 431 const U8 *t = s;
c4420975 432 const U8 * const e = s + len;
c4d5f83a 433 while (t < e) {
4373e329 434 const U8 ch = *t++;
fabdb6c0
AL
435 hibit = !NATIVE_IS_INVARIANT(ch);
436 if (hibit) {
437 start = s = bytes_to_utf8(s, &len);
1aa99e6b 438 break;
fabdb6c0 439 }
c4d5f83a 440 }
1aa99e6b 441 }
4757a243
LW
442 send = s + len;
443
444 while (s < send) {
c4420975
AL
445 const UV uv = swash_fetch(rv, s, TRUE);
446 if (uv < none || uv == extra)
4757a243 447 matches++;
834a4ddd 448 s += UTF8SKIP(s);
4757a243 449 }
1aa99e6b
IH
450 if (hibit)
451 Safefree(start);
4757a243
LW
452
453 return matches;
454}
455
942e002e 456STATIC I32
c395bd6c 457S_do_trans_complex_utf8(pTHX_ SV * const sv)
4757a243 458{
97aff369 459 dVAR;
1aa99e6b 460 U8 *start, *send;
4757a243
LW
461 U8 *d;
462 I32 matches = 0;
4373e329
AL
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;
043e41b8
DM
466 SV* const rv =
467#ifdef USE_ITHREADS
468 PAD_SVl(cPADOP->op_padix);
469#else
ad64d0ec 470 MUTABLE_SV(cSVOP->op_sv);
043e41b8 471#endif
85fbaab2 472 HV * const hv = MUTABLE_HV(SvRV(rv));
a4fc7abc 473 SV * const *svp = hv_fetchs(hv, "NONE", FALSE);
f54cb97a
AL
474 const UV none = svp ? SvUV(*svp) : 0x7fffffff;
475 const UV extra = none + 1;
b7953727 476 UV final = 0;
45005bfb 477 bool havefinal = FALSE;
4757a243 478 STRLEN len;
1aa99e6b 479 U8 *dstart, *dend;
1aa99e6b 480 U8 hibit = 0;
9138d6ca 481 U8 *s = (U8*)SvPV_nomg(sv, len);
7918f24d
NC
482
483 PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8;
484
c395bd6c 485 if (!SvUTF8(sv)) {
53c1dcc0
AL
486 const U8 *t = s;
487 const U8 * const e = s + len;
c4d5f83a 488 while (t < e) {
f54cb97a 489 const U8 ch = *t++;
fabdb6c0
AL
490 hibit = !NATIVE_IS_INVARIANT(ch);
491 if (hibit) {
492 s = bytes_to_utf8(s, &len);
1aa99e6b 493 break;
fabdb6c0 494 }
c4d5f83a 495 }
1aa99e6b 496 }
4757a243 497 send = s + len;
1aa99e6b 498 start = s;
4757a243 499
a4fc7abc 500 svp = hv_fetchs(hv, "FINAL", FALSE);
45005bfb 501 if (svp) {
4757a243 502 final = SvUV(*svp);
45005bfb
JH
503 havefinal = TRUE;
504 }
4757a243 505
9b877dbb
IH
506 if (grows) {
507 /* d needs to be bigger than s, in case e.g. upgrading is required */
a02a5408 508 Newx(d, len * 3 + UTF8_MAXBYTES, U8);
9b877dbb
IH
509 dend = d + len * 3;
510 dstart = d;
511 }
512 else {
513 dstart = d = s;
514 dend = d + len;
515 }
4757a243
LW
516
517 if (squash) {
518 UV puv = 0xfeedface;
519 while (s < send) {
f54cb97a 520 UV uv = swash_fetch(rv, s, TRUE);
9041c2e3 521
9b877dbb 522 if (d > dend) {
4373e329
AL
523 const STRLEN clen = d - dstart;
524 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
9b877dbb 525 if (!grows)
7d85a32c 526 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
89ebb4a3 527 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
1aa99e6b
IH
528 d = dstart + clen;
529 dend = dstart + nlen;
4757a243
LW
530 }
531 if (uv < none) {
532 matches++;
94472101 533 s += UTF8SKIP(s);
4757a243 534 if (uv != puv) {
7d85a32c 535 d = uvuni_to_utf8(d, uv);
4757a243
LW
536 puv = uv;
537 }
4757a243
LW
538 continue;
539 }
540 else if (uv == none) { /* "none" is unmapped character */
4373e329 541 const int i = UTF8SKIP(s);
0376ff32 542 Move(s, d, i, U8);
9b877dbb
IH
543 d += i;
544 s += i;
4757a243
LW
545 puv = 0xfeedface;
546 continue;
a0ed51b3 547 }
4757a243
LW
548 else if (uv == extra && !del) {
549 matches++;
45005bfb
JH
550 if (havefinal) {
551 s += UTF8SKIP(s);
552 if (puv != final) {
7d85a32c 553 d = uvuni_to_utf8(d, final);
45005bfb
JH
554 puv = final;
555 }
556 }
557 else {
558 STRLEN len;
9f7f3913 559 uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT);
45005bfb 560 if (uv != puv) {
0376ff32 561 Move(s, d, len, U8);
45005bfb
JH
562 d += len;
563 puv = uv;
564 }
565 s += len;
4757a243 566 }
4757a243
LW
567 continue;
568 }
01ec43d0
GS
569 matches++; /* "none+1" is delete character */
570 s += UTF8SKIP(s);
a0ed51b3 571 }
79072805
LW
572 }
573 else {
4757a243 574 while (s < send) {
f54cb97a 575 const UV uv = swash_fetch(rv, s, TRUE);
9b877dbb 576 if (d > dend) {
f54cb97a
AL
577 const STRLEN clen = d - dstart;
578 const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES;
9b877dbb 579 if (!grows)
7d85a32c 580 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
89ebb4a3 581 Renew(dstart, nlen + UTF8_MAXBYTES, U8);
1aa99e6b
IH
582 d = dstart + clen;
583 dend = dstart + nlen;
a0ed51b3 584 }
4757a243
LW
585 if (uv < none) {
586 matches++;
01ec43d0 587 s += UTF8SKIP(s);
7d85a32c 588 d = uvuni_to_utf8(d, uv);
4757a243 589 continue;
a0ed51b3 590 }
4757a243 591 else if (uv == none) { /* "none" is unmapped character */
f54cb97a 592 const int i = UTF8SKIP(s);
0376ff32 593 Move(s, d, i, U8);
9b877dbb
IH
594 d += i;
595 s += i;
4757a243 596 continue;
79072805 597 }
4757a243
LW
598 else if (uv == extra && !del) {
599 matches++;
01ec43d0 600 s += UTF8SKIP(s);
7d85a32c 601 d = uvuni_to_utf8(d, final);
4757a243
LW
602 continue;
603 }
01ec43d0
GS
604 matches++; /* "none+1" is delete character */
605 s += UTF8SKIP(s);
79072805 606 }
4757a243 607 }
16ec844d 608 if (grows || hibit) {
9b877dbb
IH
609 sv_setpvn(sv, (char*)dstart, d - dstart);
610 Safefree(dstart);
16ec844d
IH
611 if (grows && hibit)
612 Safefree(start);
9b877dbb
IH
613 }
614 else {
615 *d = '\0';
616 SvCUR_set(sv, d - dstart);
617 }
1aa99e6b 618 SvUTF8_on(sv);
4757a243
LW
619 SvSETMAGIC(sv);
620
621 return matches;
622}
623
624I32
864dbfa3 625Perl_do_trans(pTHX_ SV *sv)
4757a243 626{
97aff369 627 dVAR;
4757a243 628 STRLEN len;
4373e329 629 const I32 hasutf = (PL_op->op_private &
036b4402 630 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
4757a243 631
7918f24d
NC
632 PERL_ARGS_ASSERT_DO_TRANS;
633
3788ef8f 634 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) {
765f542d
NC
635 if (SvIsCOW(sv))
636 sv_force_normal_flags(sv, 0);
3788ef8f 637 if (SvREADONLY(sv))
6ad8f254 638 Perl_croak_no_modify(aTHX);
2233f375 639 }
10516c54 640 (void)SvPV_const(sv, len);
4757a243
LW
641 if (!len)
642 return 0;
d59e14db
RGS
643 if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) {
644 if (!SvPOKp(sv))
645 (void)SvPV_force(sv, len);
2de7b02f 646 (void)SvPOK_only_UTF8(sv);
d59e14db 647 }
4757a243 648
cea2e8a9 649 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 650
a77e643a
RGS
651 switch (PL_op->op_private & ~hasutf & (
652 OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL|
653 OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) {
4757a243 654 case 0:
01ec43d0
GS
655 if (hasutf)
656 return do_trans_simple_utf8(sv);
657 else
658 return do_trans_simple(sv);
4757a243
LW
659
660 case OPpTRANS_IDENTICAL:
05d340b8 661 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
01ec43d0
GS
662 if (hasutf)
663 return do_trans_count_utf8(sv);
664 else
665 return do_trans_count(sv);
4757a243
LW
666
667 default:
01ec43d0 668 if (hasutf)
036b4402 669 return do_trans_complex_utf8(sv);
4757a243 670 else
036b4402 671 return do_trans_complex(sv);
79072805 672 }
79072805
LW
673}
674
675void
b9ac451d 676Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp)
79072805 677{
97aff369 678 dVAR;
53c1dcc0 679 SV ** const oldmark = mark;
79072805 680 register I32 items = sp - mark;
79072805 681 register STRLEN len;
463ee0b2 682 STRLEN delimlen;
79072805 683
7918f24d
NC
684 PERL_ARGS_ASSERT_DO_JOIN;
685
b9ac451d 686 (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */
5ba99574 687 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
516a5887 688
79072805
LW
689 mark++;
690 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
862a34c6 691 SvUPGRADE(sv, SVt_PV);
79072805
LW
692 if (SvLEN(sv) < len + items) { /* current length is way too short */
693 while (items-- > 0) {
1426bbf4 694 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
f54cb97a 695 STRLEN tmplen;
5c144d81 696 SvPV_const(*mark, tmplen);
463ee0b2 697 len += tmplen;
79072805
LW
698 }
699 mark++;
700 }
701 SvGROW(sv, len + 1); /* so try to pre-extend */
702
703 mark = oldmark;
db7c17d7 704 items = sp - mark;
79072805
LW
705 ++mark;
706 }
707
76f68e9b 708 sv_setpvs(sv, "");
e4803c42 709 /* sv_setpv retains old UTF8ness [perl #24846] */
fb622db0 710 SvUTF8_off(sv);
e4803c42 711
8d6d96c1
HS
712 if (PL_tainting && SvMAGICAL(sv))
713 SvTAINTED_off(sv);
714
463ee0b2 715 if (items-- > 0) {
92d29cee
JH
716 if (*mark)
717 sv_catsv(sv, *mark);
463ee0b2
LW
718 mark++;
719 }
8d6d96c1 720
c512ce4f 721 if (delimlen) {
79072805 722 for (; items > 0; items--,mark++) {
b9ac451d 723 sv_catsv(sv,delim);
79072805
LW
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
864dbfa3 735Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 736{
97aff369 737 dVAR;
46fc3d4c 738 STRLEN patlen;
53c1dcc0 739 const char * const pat = SvPV_const(*sarg, patlen);
46fc3d4c 740 bool do_taint = FALSE;
741
7918f24d
NC
742 PERL_ARGS_ASSERT_DO_SPRINTF;
743
e06d98fb
DM
744 if (SvTAINTED(*sarg))
745 TAINT_PROPER(
746 (PL_op && PL_op->op_type < OP_max)
747 ? (PL_op->op_type == OP_PRTF)
748 ? "printf"
749 : PL_op_name[PL_op->op_type]
750 : "(unknown)"
751 );
5b781b5b 752 SvUTF8_off(sv);
2cf2cfc6
A
753 if (DO_UTF8(*sarg))
754 SvUTF8_on(sv);
4608196e 755 sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint);
79072805 756 SvSETMAGIC(sv);
46fc3d4c 757 if (do_taint)
758 SvTAINTED_on(sv);
79072805
LW
759}
760
33b45480 761/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0
JH
762UV
763Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
764{
97aff369 765 dVAR;
bbb8a7e0 766 STRLEN srclen, len, uoffset, bitoffs = 0;
10516c54 767 const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen);
81e118e0
JH
768 UV retnum = 0;
769
7918f24d
NC
770 PERL_ARGS_ASSERT_DO_VECGET;
771
a50d7633 772 if (offset < 0)
61f9802b 773 return 0;
8e84507e 774 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 775 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 776
dcad2880 777 if (SvUTF8(sv))
33b45480 778 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 779
bbb8a7e0
MHM
780 if (size < 8) {
781 bitoffs = ((offset%8)*size)%8;
782 uoffset = offset/(8/size);
783 }
784 else if (size > 8)
785 uoffset = offset*(size/8);
786 else
787 uoffset = offset;
788
789 len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */
81e118e0
JH
790 if (len > srclen) {
791 if (size <= 8)
792 retnum = 0;
793 else {
81e118e0 794 if (size == 16) {
bb7a0f54 795 if (uoffset >= srclen)
81e118e0
JH
796 retnum = 0;
797 else
bb7a0f54 798 retnum = (UV) s[uoffset] << 8;
81e118e0
JH
799 }
800 else if (size == 32) {
bb7a0f54 801 if (uoffset >= srclen)
81e118e0 802 retnum = 0;
bb7a0f54 803 else if (uoffset + 1 >= srclen)
81e118e0 804 retnum =
bb7a0f54
MHM
805 ((UV) s[uoffset ] << 24);
806 else if (uoffset + 2 >= srclen)
81e118e0 807 retnum =
bb7a0f54
MHM
808 ((UV) s[uoffset ] << 24) +
809 ((UV) s[uoffset + 1] << 16);
81e118e0
JH
810 else
811 retnum =
bb7a0f54
MHM
812 ((UV) s[uoffset ] << 24) +
813 ((UV) s[uoffset + 1] << 16) +
814 ( s[uoffset + 2] << 8);
81e118e0 815 }
d7d93a81 816#ifdef UV_IS_QUAD
c5a0f51a 817 else if (size == 64) {
a2a5de95
NC
818 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
819 "Bit vector size > 32 non-portable");
bb7a0f54 820 if (uoffset >= srclen)
c5a0f51a 821 retnum = 0;
bb7a0f54 822 else if (uoffset + 1 >= srclen)
c5a0f51a 823 retnum =
bb7a0f54
MHM
824 (UV) s[uoffset ] << 56;
825 else if (uoffset + 2 >= srclen)
c5a0f51a 826 retnum =
bb7a0f54
MHM
827 ((UV) s[uoffset ] << 56) +
828 ((UV) s[uoffset + 1] << 48);
829 else if (uoffset + 3 >= srclen)
c5a0f51a 830 retnum =
bb7a0f54
MHM
831 ((UV) s[uoffset ] << 56) +
832 ((UV) s[uoffset + 1] << 48) +
833 ((UV) s[uoffset + 2] << 40);
834 else if (uoffset + 4 >= srclen)
c5a0f51a 835 retnum =
bb7a0f54
MHM
836 ((UV) s[uoffset ] << 56) +
837 ((UV) s[uoffset + 1] << 48) +
838 ((UV) s[uoffset + 2] << 40) +
839 ((UV) s[uoffset + 3] << 32);
840 else if (uoffset + 5 >= srclen)
c5a0f51a 841 retnum =
bb7a0f54
MHM
842 ((UV) s[uoffset ] << 56) +
843 ((UV) s[uoffset + 1] << 48) +
844 ((UV) s[uoffset + 2] << 40) +
845 ((UV) s[uoffset + 3] << 32) +
846 ( s[uoffset + 4] << 24);
847 else if (uoffset + 6 >= srclen)
c5a0f51a 848 retnum =
bb7a0f54
MHM
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);
c5a0f51a 855 else
8e84507e 856 retnum =
bb7a0f54
MHM
857 ((UV) s[uoffset ] << 56) +
858 ((UV) s[uoffset + 1] << 48) +
859 ((UV) s[uoffset + 2] << 40) +
860 ((UV) s[uoffset + 3] << 32) +
861 ((UV) s[uoffset + 4] << 24) +
862 ((UV) s[uoffset + 5] << 16) +
863 ( s[uoffset + 6] << 8);
c5a0f51a
JH
864 }
865#endif
81e118e0
JH
866 }
867 }
868 else if (size < 8)
bbb8a7e0 869 retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1);
81e118e0 870 else {
81e118e0 871 if (size == 8)
bb7a0f54 872 retnum = s[uoffset];
81e118e0
JH
873 else if (size == 16)
874 retnum =
bb7a0f54
MHM
875 ((UV) s[uoffset] << 8) +
876 s[uoffset + 1];
81e118e0
JH
877 else if (size == 32)
878 retnum =
bb7a0f54
MHM
879 ((UV) s[uoffset ] << 24) +
880 ((UV) s[uoffset + 1] << 16) +
881 ( s[uoffset + 2] << 8) +
882 s[uoffset + 3];
d7d93a81 883#ifdef UV_IS_QUAD
c5a0f51a 884 else if (size == 64) {
a2a5de95
NC
885 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
886 "Bit vector size > 32 non-portable");
c5a0f51a 887 retnum =
bb7a0f54
MHM
888 ((UV) s[uoffset ] << 56) +
889 ((UV) s[uoffset + 1] << 48) +
890 ((UV) s[uoffset + 2] << 40) +
891 ((UV) s[uoffset + 3] << 32) +
892 ((UV) s[uoffset + 4] << 24) +
893 ((UV) s[uoffset + 5] << 16) +
894 ( s[uoffset + 6] << 8) +
895 s[uoffset + 7];
c5a0f51a
JH
896 }
897#endif
81e118e0
JH
898 }
899
900 return retnum;
901}
902
33b45480
SB
903/* currently converts input to bytes if possible but doesn't sweat failures,
904 * although it does ensure that the string it clobbers is not marked as
905 * utf8-valid any more
906 */
79072805 907void
864dbfa3 908Perl_do_vecset(pTHX_ SV *sv)
79072805 909{
97aff369 910 dVAR;
bbb8a7e0 911 register I32 offset, bitoffs = 0;
79072805 912 register I32 size;
8990e307 913 register unsigned char *s;
81e118e0 914 register UV lval;
79072805 915 I32 mask;
a0d0e21e
LW
916 STRLEN targlen;
917 STRLEN len;
c4420975 918 SV * const targ = LvTARG(sv);
79072805 919
7918f24d
NC
920 PERL_ARGS_ASSERT_DO_VECSET;
921
8990e307
LW
922 if (!targ)
923 return;
a0d0e21e 924 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 925 if (SvUTF8(targ)) {
33b45480
SB
926 /* This is handled by the SvPOK_only below...
927 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
928 SvUTF8_off(targ);
929 */
930 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53
MG
931 }
932
4ebbc975 933 (void)SvPOK_only(targ);
81e118e0 934 lval = SvUV(sv);
79072805 935 offset = LvTARGOFF(sv);
fe58ced6 936 if (offset < 0)
ed9aa3b7 937 Perl_croak(aTHX_ "Negative offset to vec in lvalue context");
79072805 938 size = LvTARGLEN(sv);
8e84507e 939 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 940 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 941
bbb8a7e0
MHM
942 if (size < 8) {
943 bitoffs = ((offset%8)*size)%8;
944 offset /= 8/size;
945 }
946 else if (size > 8)
947 offset *= size/8;
948
949 len = offset + (bitoffs + size + 7)/8; /* required number of bytes */
a0d0e21e
LW
950 if (len > targlen) {
951 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 952 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e
LW
953 SvCUR_set(targ, len);
954 }
8e84507e 955
79072805
LW
956 if (size < 8) {
957 mask = (1 << size) - 1;
79072805 958 lval &= mask;
bbb8a7e0
MHM
959 s[offset] &= ~(mask << bitoffs);
960 s[offset] |= lval << bitoffs;
79072805
LW
961 }
962 else {
963 if (size == 8)
eb160463 964 s[offset ] = (U8)( lval & 0xff);
79072805 965 else if (size == 16) {
eb160463
GS
966 s[offset ] = (U8)((lval >> 8) & 0xff);
967 s[offset+1] = (U8)( lval & 0xff);
79072805
LW
968 }
969 else if (size == 32) {
eb160463
GS
970 s[offset ] = (U8)((lval >> 24) & 0xff);
971 s[offset+1] = (U8)((lval >> 16) & 0xff);
972 s[offset+2] = (U8)((lval >> 8) & 0xff);
973 s[offset+3] = (U8)( lval & 0xff);
c5a0f51a 974 }
d7d93a81 975#ifdef UV_IS_QUAD
c5a0f51a 976 else if (size == 64) {
a2a5de95
NC
977 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
978 "Bit vector size > 32 non-portable");
eb160463
GS
979 s[offset ] = (U8)((lval >> 56) & 0xff);
980 s[offset+1] = (U8)((lval >> 48) & 0xff);
981 s[offset+2] = (U8)((lval >> 40) & 0xff);
982 s[offset+3] = (U8)((lval >> 32) & 0xff);
983 s[offset+4] = (U8)((lval >> 24) & 0xff);
984 s[offset+5] = (U8)((lval >> 16) & 0xff);
985 s[offset+6] = (U8)((lval >> 8) & 0xff);
986 s[offset+7] = (U8)( lval & 0xff);
79072805 987 }
dc1e3f56 988#endif
79072805 989 }
7bb043c3 990 SvSETMAGIC(targ);
79072805
LW
991}
992
993void
864dbfa3 994Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805 995{
97aff369 996 dVAR;
79072805
LW
997#ifdef LIBERAL
998 register long *dl;
999 register long *ll;
1000 register long *rl;
1001#endif
1002 register char *dc;
463ee0b2
LW
1003 STRLEN leftlen;
1004 STRLEN rightlen;
e62f0680
NC
1005 register const char *lc;
1006 register const char *rc;
bb7a0f54
MHM
1007 register STRLEN len;
1008 STRLEN lensave;
e62f0680
NC
1009 const char *lsave;
1010 const char *rsave;
12abf4f0
NC
1011 bool left_utf;
1012 bool right_utf;
bb7a0f54 1013 STRLEN needlen = 0;
0c57e439 1014
7918f24d 1015 PERL_ARGS_ASSERT_DO_VOP;
79072805 1016
1fbd88dc 1017 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
76f68e9b 1018 sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */
8c8eee82
BM
1019 if (sv == left) {
1020 lsave = lc = SvPV_force_nomg(left, leftlen);
1021 }
1022 else {
1023 lsave = lc = SvPV_nomg_const(left, leftlen);
1024 SvPV_force_nomg_nolen(sv);
1025 }
e62f0680 1026 rsave = rc = SvPV_nomg_const(right, rightlen);
12abf4f0
NC
1027
1028 /* This need to come after SvPV to ensure that string overloading has
1029 fired off. */
1030
1031 left_utf = DO_UTF8(left);
1032 right_utf = DO_UTF8(right);
1033
1034 if (left_utf && !right_utf) {
1035 /* Avoid triggering overloading again by using temporaries.
1036 Maybe there should be a variant of sv_utf8_upgrade that takes pvn
1037 */
59cd0e26 1038 right = newSVpvn_flags(rsave, rightlen, SVs_TEMP);
12abf4f0
NC
1039 sv_utf8_upgrade(right);
1040 rsave = rc = SvPV_nomg_const(right, rightlen);
1041 right_utf = TRUE;
1042 }
1043 else if (!left_utf && right_utf) {
59cd0e26 1044 left = newSVpvn_flags(lsave, leftlen, SVs_TEMP);
12abf4f0
NC
1045 sv_utf8_upgrade(left);
1046 lsave = lc = SvPV_nomg_const(left, leftlen);
1047 left_utf = TRUE;
1048 }
1049
93a17b20 1050 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 1051 lensave = len;
9fdd7463
JH
1052 SvCUR_set(sv, len);
1053 (void)SvPOK_only(sv);
c9b3c8d0
JH
1054 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1055 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
a02a5408 1056 Newxz(dc, needlen + 1, char);
c9b3c8d0
JH
1057 }
1058 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2596d9fe 1059 dc = SvPV_force_nomg_nolen(sv);
bb7a0f54
MHM
1060 if (SvLEN(sv) < len + 1) {
1061 dc = SvGROW(sv, len + 1);
ff68c719 1062 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1063 }
c9b3c8d0
JH
1064 if (optype != OP_BIT_AND && (left_utf || right_utf))
1065 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 1066 }
1067 else {
bb7a0f54
MHM
1068 needlen = optype == OP_BIT_AND
1069 ? len : (leftlen > rightlen ? leftlen : rightlen);
a02a5408 1070 Newxz(dc, needlen + 1, char);
aa0a69cb 1071 sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL);
ff68c719 1072 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 1073 }
0c57e439
GS
1074 if (left_utf || right_utf) {
1075 UV duc, luc, ruc;
794a0d33
JH
1076 char *dcorig = dc;
1077 char *dcsave = NULL;
0c57e439
GS
1078 STRLEN lulen = leftlen;
1079 STRLEN rulen = rightlen;
ba210ebe 1080 STRLEN ulen;
0c57e439
GS
1081
1082 switch (optype) {
1083 case OP_BIT_AND:
1084 while (lulen && rulen) {
9041c2e3 1085 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1086 lc += ulen;
1087 lulen -= ulen;
9041c2e3 1088 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1089 rc += ulen;
1090 rulen -= ulen;
1091 duc = luc & ruc;
9041c2e3 1092 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439 1093 }
3ce4c532
NC
1094 if (sv == left || sv == right)
1095 (void)sv_usepvn(sv, dcorig, needlen);
794a0d33 1096 SvCUR_set(sv, dc - dcorig);
0c57e439
GS
1097 break;
1098 case OP_BIT_XOR:
1099 while (lulen && rulen) {
9041c2e3 1100 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1101 lc += ulen;
1102 lulen -= ulen;
9041c2e3 1103 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1104 rc += ulen;
1105 rulen -= ulen;
1106 duc = luc ^ ruc;
9041c2e3 1107 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1108 }
1109 goto mop_up_utf;
1110 case OP_BIT_OR:
1111 while (lulen && rulen) {
9041c2e3 1112 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1113 lc += ulen;
1114 lulen -= ulen;
9041c2e3 1115 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1116 rc += ulen;
1117 rulen -= ulen;
1118 duc = luc | ruc;
9041c2e3 1119 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1120 }
1121 mop_up_utf:
794a0d33
JH
1122 if (rulen)
1123 dcsave = savepvn(rc, rulen);
1124 else if (lulen)
1125 dcsave = savepvn(lc, lulen);
3ce4c532 1126 if (sv == left || sv == right)
794a0d33
JH
1127 (void)sv_usepvn(sv, dcorig, needlen); /* Uses Renew(). */
1128 SvCUR_set(sv, dc - dcorig);
0c57e439 1129 if (rulen)
794a0d33 1130 sv_catpvn(sv, dcsave, rulen);
0c57e439 1131 else if (lulen)
794a0d33 1132 sv_catpvn(sv, dcsave, lulen);
0c57e439
GS
1133 else
1134 *SvEND(sv) = '\0';
794a0d33 1135 Safefree(dcsave);
0c57e439 1136 break;
41547412
NC
1137 default:
1138 if (sv == left || sv == right)
1139 Safefree(dcorig);
c395bd6c
AL
1140 Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)",
1141 (unsigned)optype, PL_op_name[optype]);
0c57e439
GS
1142 }
1143 SvUTF8_on(sv);
1144 goto finish;
1145 }
1146 else
79072805
LW
1147#ifdef LIBERAL
1148 if (len >= sizeof(long)*4 &&
bb7a0f54
MHM
1149 !((unsigned long)dc % sizeof(long)) &&
1150 !((unsigned long)lc % sizeof(long)) &&
1151 !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */
79072805 1152 {
bb7a0f54 1153 const STRLEN remainder = len % (sizeof(long)*4);
79072805
LW
1154 len /= (sizeof(long)*4);
1155
1156 dl = (long*)dc;
1157 ll = (long*)lc;
1158 rl = (long*)rc;
1159
1160 switch (optype) {
1161 case OP_BIT_AND:
1162 while (len--) {
1163 *dl++ = *ll++ & *rl++;
1164 *dl++ = *ll++ & *rl++;
1165 *dl++ = *ll++ & *rl++;
1166 *dl++ = *ll++ & *rl++;
1167 }
1168 break;
a0d0e21e 1169 case OP_BIT_XOR:
79072805
LW
1170 while (len--) {
1171 *dl++ = *ll++ ^ *rl++;
1172 *dl++ = *ll++ ^ *rl++;
1173 *dl++ = *ll++ ^ *rl++;
1174 *dl++ = *ll++ ^ *rl++;
1175 }
1176 break;
1177 case OP_BIT_OR:
1178 while (len--) {
1179 *dl++ = *ll++ | *rl++;
1180 *dl++ = *ll++ | *rl++;
1181 *dl++ = *ll++ | *rl++;
1182 *dl++ = *ll++ | *rl++;
1183 }
1184 }
1185
1186 dc = (char*)dl;
1187 lc = (char*)ll;
1188 rc = (char*)rl;
1189
1190 len = remainder;
1191 }
1192#endif
a0d0e21e 1193 {
a0d0e21e
LW
1194 switch (optype) {
1195 case OP_BIT_AND:
1196 while (len--)
1197 *dc++ = *lc++ & *rc++;
1a787b95 1198 *dc = '\0';
a0d0e21e
LW
1199 break;
1200 case OP_BIT_XOR:
1201 while (len--)
1202 *dc++ = *lc++ ^ *rc++;
1203 goto mop_up;
1204 case OP_BIT_OR:
1205 while (len--)
1206 *dc++ = *lc++ | *rc++;
1207 mop_up:
1208 len = lensave;
bb7a0f54 1209 if (rightlen > len)
a0d0e21e 1210 sv_catpvn(sv, rsave + len, rightlen - len);
eb160463 1211 else if (leftlen > (STRLEN)len)
a0d0e21e 1212 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4
LW
1213 else
1214 *SvEND(sv) = '\0';
a0d0e21e
LW
1215 break;
1216 }
79072805 1217 }
0c57e439 1218finish:
fb73857a 1219 SvTAINT(sv);
79072805 1220}
463ee0b2
LW
1221
1222OP *
cea2e8a9 1223Perl_do_kv(pTHX)
463ee0b2 1224{
97aff369 1225 dVAR;
39644a26 1226 dSP;
73ff03e8 1227 HV * const keys = MUTABLE_HV(POPs);
463ee0b2 1228 register HE *entry;
f54cb97a
AL
1229 const I32 gimme = GIMME_V;
1230 const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV);
cba5a3b0
DG
1231 /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */
1232 const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS || PL_op->op_type == OP_RKEYS);
1233 const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES || PL_op->op_type == OP_RVALUES);
463ee0b2 1234
800e9ae0 1235 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1236
54310121 1237 if (gimme == G_VOID)
aa689395 1238 RETURN;
1239
54310121 1240 if (gimme == G_SCALAR) {
78f9721b 1241 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
2154eca7
EB
1242 SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */
1243 sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0);
1244 LvTYPE(ret) = 'k';
1245 LvTARG(ret) = SvREFCNT_inc_simple(keys);
1246 PUSHs(ret);
81714fb9 1247 }
463ee0b2 1248 else {
2154eca7
EB
1249 IV i;
1250 dTARGET;
1251
1252 if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) {
1b95d04f 1253 i = HvUSEDKEYS(keys);
2154eca7
EB
1254 }
1255 else {
1256 i = 0;
1257 while (hv_iternext(keys)) i++;
1258 }
1259 PUSHi( i );
463ee0b2 1260 }
463ee0b2
LW
1261 RETURN;
1262 }
1263
1b95d04f 1264 EXTEND(SP, HvUSEDKEYS(keys) * (dokeys + dovalues));
463ee0b2 1265
463ee0b2 1266 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1267 while ((entry = hv_iternext(keys))) {
463ee0b2 1268 SPAGAIN;
574c8022 1269 if (dokeys) {
890ce7af 1270 SV* const sv = hv_iterkeysv(entry);
574c8022
JH
1271 XPUSHs(sv); /* won't clobber stack_sp */
1272 }
463ee0b2 1273 if (dovalues) {
f54cb97a 1274 SV *tmpstr;
463ee0b2 1275 PUTBACK;
73ff03e8 1276 tmpstr = hv_iterval(keys,entry);
cea2e8a9 1277 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1278 (unsigned long)HeHASH(entry),
cdd3ba14 1279 (int)HvMAX(keys)+1,
800e9ae0 1280 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1281 SPAGAIN;
46fc3d4c 1282 XPUSHs(tmpstr);
463ee0b2
LW
1283 }
1284 PUTBACK;
1285 }
1286 return NORMAL;
1287}
4e35701f 1288
af3babe4
NC
1289/*
1290 * Local variables:
1291 * c-indentation-style: bsd
1292 * c-basic-offset: 4
1293 * indent-tabs-mode: t
1294 * End:
1295 *
37442d52
RGS
1296 * ex: set ts=8 sts=4 sw=4 noet:
1297 */