This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Integrate change #18420 from maint-5.8:
[perl5.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
be3c0a43 3 * Copyright (c) 1991-2002, Larry Wall
79072805
LW
4 *
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
7 *
a0d0e21e
LW
8 */
9
10/*
11 * "'So that was the job I felt I had to do when I started,' thought Sam."
79072805
LW
12 */
13
14#include "EXTERN.h"
864dbfa3 15#define PERL_IN_DOOP_C
79072805
LW
16#include "perl.h"
17
64ca3a65 18#ifndef PERL_MICRO
79072805
LW
19#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
20#include <signal.h>
21#endif
64ca3a65 22#endif
79072805 23
942e002e 24STATIC I32
b250498f 25S_do_trans_simple(pTHX_ SV *sv)
79072805 26{
4757a243 27 U8 *s;
b250498f 28 U8 *d;
4757a243 29 U8 *send;
b250498f 30 U8 *dstart;
4757a243 31 I32 matches = 0;
9b877dbb 32 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
463ee0b2 33 STRLEN len;
4757a243
LW
34 short *tbl;
35 I32 ch;
79072805 36
4757a243
LW
37 tbl = (short*)cPVOP->op_pv;
38 if (!tbl)
7d85a32c 39 Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
a0ed51b3 40
4757a243
LW
41 s = (U8*)SvPV(sv, len);
42 send = s + len;
43
b250498f 44 /* First, take care of non-UTF8 input strings, because they're easy */
1aa99e6b 45 if (!SvUTF8(sv)) {
01ec43d0 46 while (s < send) {
036b4402 47 if ((ch = tbl[*s]) >= 0) {
01ec43d0 48 matches++;
eb160463 49 *s++ = (U8)ch;
01ec43d0
GS
50 }
51 else
52 s++;
53 }
54 SvSETMAGIC(sv);
b250498f
GS
55 return matches;
56 }
4757a243 57
b250498f 58 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
9b877dbb
IH
59 if (grows)
60 New(0, d, len*2+1, U8);
61 else
62 d = s;
b250498f
GS
63 dstart = d;
64 while (s < send) {
ba210ebe 65 STRLEN ulen;
9b877dbb 66 UV c;
b250498f 67
b250498f 68 /* Need to check this, otherwise 128..255 won't match */
9041c2e3 69 c = utf8n_to_uvchr(s, send - s, &ulen, 0);
9b877dbb 70 if (c < 0x100 && (ch = tbl[c]) >= 0) {
b250498f 71 matches++;
9041c2e3 72 d = uvchr_to_utf8(d, ch);
b250498f 73 s += ulen;
01ec43d0
GS
74 }
75 else { /* No match -> copy */
9b877dbb
IH
76 Copy(s, d, ulen, U8);
77 d += ulen;
78 s += ulen;
b250498f
GS
79 }
80 }
9b877dbb
IH
81 if (grows) {
82 sv_setpvn(sv, (char*)dstart, d - dstart);
83 Safefree(dstart);
84 }
85 else {
86 *d = '\0';
87 SvCUR_set(sv, d - dstart);
88 }
b250498f 89 SvUTF8_on(sv);
b250498f 90 SvSETMAGIC(sv);
4757a243
LW
91 return matches;
92}
93
942e002e 94STATIC I32
21304a3b 95S_do_trans_count(pTHX_ SV *sv)
4757a243 96{
4757a243
LW
97 U8 *s;
98 U8 *send;
99 I32 matches = 0;
100 STRLEN len;
101 short *tbl;
8973db79 102 I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
4757a243
LW
103
104 tbl = (short*)cPVOP->op_pv;
105 if (!tbl)
7d85a32c 106 Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
4757a243
LW
107
108 s = (U8*)SvPV(sv, len);
109 send = s + len;
110
1aa99e6b
IH
111 if (!SvUTF8(sv))
112 while (s < send) {
113 if (tbl[*s++] >= 0)
036b4402 114 matches++;
1aa99e6b
IH
115 }
116 else
117 while (s < send) {
118 UV c;
119 STRLEN ulen;
9041c2e3 120 c = utf8n_to_uvchr(s, send - s, &ulen, 0);
8973db79
JH
121 if (c < 0x100) {
122 if (tbl[c] >= 0)
123 matches++;
124 } else if (complement)
1aa99e6b
IH
125 matches++;
126 s += ulen;
127 }
4757a243
LW
128
129 return matches;
130}
131
942e002e 132STATIC I32
21304a3b 133S_do_trans_complex(pTHX_ SV *sv)
4757a243 134{
4757a243
LW
135 U8 *s;
136 U8 *send;
137 U8 *d;
5e44153e 138 U8 *dstart;
1aa99e6b 139 I32 isutf8;
4757a243 140 I32 matches = 0;
9b877dbb 141 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
8973db79
JH
142 I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
143 I32 del = PL_op->op_private & OPpTRANS_DELETE;
b7953727 144 STRLEN len, rlen = 0;
4757a243
LW
145 short *tbl;
146 I32 ch;
147
148 tbl = (short*)cPVOP->op_pv;
149 if (!tbl)
7d85a32c 150 Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
4757a243
LW
151
152 s = (U8*)SvPV(sv, len);
1aa99e6b 153 isutf8 = SvUTF8(sv);
4757a243
LW
154 send = s + len;
155
1aa99e6b
IH
156 if (!isutf8) {
157 dstart = d = s;
158 if (PL_op->op_private & OPpTRANS_SQUASH) {
159 U8* p = send;
160 while (s < send) {
161 if ((ch = tbl[*s]) >= 0) {
eb160463 162 *d = (U8)ch;
036b4402 163 matches++;
1aa99e6b
IH
164 if (p != d - 1 || *p != *d)
165 p = d++;
166 }
167 else if (ch == -1) /* -1 is unmapped character */
168 *d++ = *s;
169 else if (ch == -2) /* -2 is delete character */
170 matches++;
171 s++;
172 }
a0ed51b3 173 }
1aa99e6b
IH
174 else {
175 while (s < send) {
176 if ((ch = tbl[*s]) >= 0) {
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 }
1aa99e6b 190 else { /* isutf8 */
9b877dbb
IH
191 if (grows)
192 New(0, d, len*2+1, U8);
193 else
194 d = s;
1aa99e6b 195 dstart = d;
8973db79
JH
196 if (complement && !del)
197 rlen = tbl[0x100];
1aa99e6b 198
4aecb5b5
JH
199#ifdef MACOS_TRADITIONAL
200#define comp CoMP /* "comp" is a keyword in some compilers ... */
201#endif
202
1aa99e6b 203 if (PL_op->op_private & OPpTRANS_SQUASH) {
1aa99e6b
IH
204 UV pch = 0xfeedface;
205 while (s < send) {
206 STRLEN len;
9041c2e3 207 UV comp = utf8_to_uvchr(s, &len);
1aa99e6b 208
8973db79
JH
209 if (comp > 0xff) {
210 if (!complement) {
211 Copy(s, d, len, U8);
212 d += len;
213 }
214 else {
215 matches++;
216 if (!del) {
05d340b8
JH
217 ch = (rlen == 0) ? comp :
218 (comp - 0x100 < rlen) ?
8973db79 219 tbl[comp+1] : tbl[0x100+rlen];
eb160463 220 if ((UV)ch != pch) {
9041c2e3 221 d = uvchr_to_utf8(d, ch);
eb160463 222 pch = (UV)ch;
8973db79
JH
223 }
224 s += len;
225 continue;
226 }
227 }
9b877dbb 228 }
1aa99e6b
IH
229 else if ((ch = tbl[comp]) >= 0) {
230 matches++;
eb160463 231 if ((UV)ch != pch) {
9041c2e3 232 d = uvchr_to_utf8(d, ch);
eb160463 233 pch = (UV)ch;
1aa99e6b
IH
234 }
235 s += len;
236 continue;
237 }
9b877dbb
IH
238 else if (ch == -1) { /* -1 is unmapped character */
239 Copy(s, d, len, U8);
240 d += len;
241 }
1aa99e6b
IH
242 else if (ch == -2) /* -2 is delete character */
243 matches++;
244 s += len;
245 pch = 0xfeedface;
246 }
5d06d08e 247 }
1aa99e6b
IH
248 else {
249 while (s < send) {
250 STRLEN len;
9041c2e3 251 UV comp = utf8_to_uvchr(s, &len);
8973db79
JH
252 if (comp > 0xff) {
253 if (!complement) {
254 Copy(s, d, len, U8);
255 d += len;
256 }
257 else {
258 matches++;
259 if (!del) {
9041c2e3
NIS
260 if (comp - 0x100 < rlen)
261 d = uvchr_to_utf8(d, tbl[comp+1]);
8973db79 262 else
9041c2e3 263 d = uvchr_to_utf8(d, tbl[0x100+rlen]);
8973db79
JH
264 }
265 }
9b877dbb 266 }
1aa99e6b 267 else if ((ch = tbl[comp]) >= 0) {
9041c2e3 268 d = uvchr_to_utf8(d, ch);
1aa99e6b
IH
269 matches++;
270 }
271 else if (ch == -1) { /* -1 is unmapped character */
9b877dbb
IH
272 Copy(s, d, len, U8);
273 d += len;
1aa99e6b
IH
274 }
275 else if (ch == -2) /* -2 is delete character */
276 matches++;
277 s += len;
278 }
279 }
9b877dbb
IH
280 if (grows) {
281 sv_setpvn(sv, (char*)dstart, d - dstart);
282 Safefree(dstart);
283 }
284 else {
285 *d = '\0';
286 SvCUR_set(sv, d - dstart);
287 }
1aa99e6b 288 SvUTF8_on(sv);
4757a243 289 }
5e44153e 290 SvSETMAGIC(sv);
4757a243
LW
291 return matches;
292}
293
942e002e 294STATIC I32
21304a3b 295S_do_trans_simple_utf8(pTHX_ SV *sv)
4757a243 296{
4757a243
LW
297 U8 *s;
298 U8 *send;
299 U8 *d;
036b4402 300 U8 *start;
1aa99e6b 301 U8 *dstart, *dend;
4757a243 302 I32 matches = 0;
9b877dbb 303 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
4757a243
LW
304 STRLEN len;
305
306 SV* rv = (SV*)cSVOP->op_sv;
307 HV* hv = (HV*)SvRV(rv);
308 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
309 UV none = svp ? SvUV(*svp) : 0x7fffffff;
310 UV extra = none + 1;
b7953727 311 UV final = 0;
4757a243 312 UV uv;
1aa99e6b
IH
313 I32 isutf8;
314 U8 hibit = 0;
4757a243
LW
315
316 s = (U8*)SvPV(sv, len);
1aa99e6b
IH
317 isutf8 = SvUTF8(sv);
318 if (!isutf8) {
319 U8 *t = s, *e = s + len;
c4d5f83a
NIS
320 while (t < e) {
321 U8 ch = *t++;
322 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
1aa99e6b 323 break;
c4d5f83a 324 }
1aa99e6b
IH
325 if (hibit)
326 s = bytes_to_utf8(s, &len);
327 }
4757a243 328 send = s + len;
036b4402 329 start = s;
4757a243
LW
330
331 svp = hv_fetch(hv, "FINAL", 5, FALSE);
332 if (svp)
333 final = SvUV(*svp);
334
9b877dbb
IH
335 if (grows) {
336 /* d needs to be bigger than s, in case e.g. upgrading is required */
337 New(0, d, len*3+UTF8_MAXLEN, U8);
338 dend = d + len * 3;
339 dstart = d;
340 }
341 else {
342 dstart = d = s;
343 dend = d + len;
344 }
1aa99e6b 345
4757a243 346 while (s < send) {
3568d838 347 if ((uv = swash_fetch(rv, s, TRUE)) < none) {
4757a243
LW
348 s += UTF8SKIP(s);
349 matches++;
7d85a32c 350 d = uvuni_to_utf8(d, uv);
4757a243
LW
351 }
352 else if (uv == none) {
1aa99e6b 353 int i = UTF8SKIP(s);
9b877dbb
IH
354 Copy(s, d, i, U8);
355 d += i;
356 s += i;
4757a243
LW
357 }
358 else if (uv == extra) {
1aa99e6b 359 int i = UTF8SKIP(s);
036b4402 360 s += i;
4757a243 361 matches++;
7d85a32c 362 d = uvuni_to_utf8(d, final);
4757a243
LW
363 }
364 else
365 s += UTF8SKIP(s);
1aa99e6b 366
9b877dbb 367 if (d > dend) {
1aa99e6b
IH
368 STRLEN clen = d - dstart;
369 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
9b877dbb 370 if (!grows)
7d85a32c 371 Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
1aa99e6b
IH
372 Renew(dstart, nlen+UTF8_MAXLEN, U8);
373 d = dstart + clen;
374 dend = dstart + nlen;
375 }
4757a243 376 }
16ec844d 377 if (grows || hibit) {
9b877dbb
IH
378 sv_setpvn(sv, (char*)dstart, d - dstart);
379 Safefree(dstart);
16ec844d
IH
380 if (grows && hibit)
381 Safefree(start);
9b877dbb
IH
382 }
383 else {
384 *d = '\0';
385 SvCUR_set(sv, d - dstart);
386 }
4757a243 387 SvSETMAGIC(sv);
1aa99e6b 388 SvUTF8_on(sv);
4757a243
LW
389
390 return matches;
391}
392
942e002e 393STATIC I32
21304a3b 394S_do_trans_count_utf8(pTHX_ SV *sv)
4757a243 395{
4757a243 396 U8 *s;
b7953727 397 U8 *start = 0, *send;
4757a243
LW
398 I32 matches = 0;
399 STRLEN len;
400
401 SV* rv = (SV*)cSVOP->op_sv;
402 HV* hv = (HV*)SvRV(rv);
403 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
404 UV none = svp ? SvUV(*svp) : 0x7fffffff;
45005bfb 405 UV extra = none + 1;
4757a243 406 UV uv;
1aa99e6b 407 U8 hibit = 0;
4757a243
LW
408
409 s = (U8*)SvPV(sv, len);
1aa99e6b
IH
410 if (!SvUTF8(sv)) {
411 U8 *t = s, *e = s + len;
c4d5f83a
NIS
412 while (t < e) {
413 U8 ch = *t++;
414 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
1aa99e6b 415 break;
c4d5f83a 416 }
1aa99e6b
IH
417 if (hibit)
418 start = s = bytes_to_utf8(s, &len);
419 }
4757a243
LW
420 send = s + len;
421
422 while (s < send) {
3568d838 423 if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra)
4757a243 424 matches++;
834a4ddd 425 s += UTF8SKIP(s);
4757a243 426 }
1aa99e6b
IH
427 if (hibit)
428 Safefree(start);
4757a243
LW
429
430 return matches;
431}
432
942e002e 433STATIC I32
21304a3b 434S_do_trans_complex_utf8(pTHX_ SV *sv)
4757a243 435{
4757a243 436 U8 *s;
1aa99e6b 437 U8 *start, *send;
4757a243
LW
438 U8 *d;
439 I32 matches = 0;
440 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
4757a243 441 I32 del = PL_op->op_private & OPpTRANS_DELETE;
9b877dbb 442 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
4757a243
LW
443 SV* rv = (SV*)cSVOP->op_sv;
444 HV* hv = (HV*)SvRV(rv);
445 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
446 UV none = svp ? SvUV(*svp) : 0x7fffffff;
447 UV extra = none + 1;
b7953727 448 UV final = 0;
45005bfb 449 bool havefinal = FALSE;
4757a243
LW
450 UV uv;
451 STRLEN len;
1aa99e6b
IH
452 U8 *dstart, *dend;
453 I32 isutf8;
454 U8 hibit = 0;
4757a243
LW
455
456 s = (U8*)SvPV(sv, len);
1aa99e6b
IH
457 isutf8 = SvUTF8(sv);
458 if (!isutf8) {
459 U8 *t = s, *e = s + len;
c4d5f83a
NIS
460 while (t < e) {
461 U8 ch = *t++;
462 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
1aa99e6b 463 break;
c4d5f83a 464 }
1aa99e6b
IH
465 if (hibit)
466 s = bytes_to_utf8(s, &len);
467 }
4757a243 468 send = s + len;
1aa99e6b 469 start = s;
4757a243
LW
470
471 svp = hv_fetch(hv, "FINAL", 5, FALSE);
45005bfb 472 if (svp) {
4757a243 473 final = SvUV(*svp);
45005bfb
JH
474 havefinal = TRUE;
475 }
4757a243 476
9b877dbb
IH
477 if (grows) {
478 /* d needs to be bigger than s, in case e.g. upgrading is required */
479 New(0, d, len*3+UTF8_MAXLEN, U8);
480 dend = d + len * 3;
481 dstart = d;
482 }
483 else {
484 dstart = d = s;
485 dend = d + len;
486 }
4757a243
LW
487
488 if (squash) {
489 UV puv = 0xfeedface;
490 while (s < send) {
3568d838 491 uv = swash_fetch(rv, s, TRUE);
9041c2e3 492
9b877dbb
IH
493 if (d > dend) {
494 STRLEN clen = d - dstart;
495 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
496 if (!grows)
7d85a32c 497 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
1aa99e6b
IH
498 Renew(dstart, nlen+UTF8_MAXLEN, U8);
499 d = dstart + clen;
500 dend = dstart + nlen;
4757a243
LW
501 }
502 if (uv < none) {
503 matches++;
94472101 504 s += UTF8SKIP(s);
4757a243 505 if (uv != puv) {
7d85a32c 506 d = uvuni_to_utf8(d, uv);
4757a243
LW
507 puv = uv;
508 }
4757a243
LW
509 continue;
510 }
511 else if (uv == none) { /* "none" is unmapped character */
1aa99e6b 512 int i = UTF8SKIP(s);
9b877dbb
IH
513 Copy(s, d, i, U8);
514 d += i;
515 s += i;
4757a243
LW
516 puv = 0xfeedface;
517 continue;
a0ed51b3 518 }
4757a243
LW
519 else if (uv == extra && !del) {
520 matches++;
45005bfb
JH
521 if (havefinal) {
522 s += UTF8SKIP(s);
523 if (puv != final) {
7d85a32c 524 d = uvuni_to_utf8(d, final);
45005bfb
JH
525 puv = final;
526 }
527 }
528 else {
529 STRLEN len;
7d85a32c 530 uv = utf8_to_uvuni(s, &len);
45005bfb
JH
531 if (uv != puv) {
532 Copy(s, d, len, U8);
533 d += len;
534 puv = uv;
535 }
536 s += len;
4757a243 537 }
4757a243
LW
538 continue;
539 }
01ec43d0
GS
540 matches++; /* "none+1" is delete character */
541 s += UTF8SKIP(s);
a0ed51b3 542 }
79072805
LW
543 }
544 else {
4757a243 545 while (s < send) {
3568d838 546 uv = swash_fetch(rv, s, TRUE);
9b877dbb
IH
547 if (d > dend) {
548 STRLEN clen = d - dstart;
549 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
550 if (!grows)
7d85a32c 551 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
1aa99e6b
IH
552 Renew(dstart, nlen+UTF8_MAXLEN, U8);
553 d = dstart + clen;
554 dend = dstart + nlen;
a0ed51b3 555 }
4757a243
LW
556 if (uv < none) {
557 matches++;
01ec43d0 558 s += UTF8SKIP(s);
7d85a32c 559 d = uvuni_to_utf8(d, uv);
4757a243 560 continue;
a0ed51b3 561 }
4757a243 562 else if (uv == none) { /* "none" is unmapped character */
1aa99e6b 563 int i = UTF8SKIP(s);
9b877dbb
IH
564 Copy(s, d, i, U8);
565 d += i;
566 s += i;
4757a243 567 continue;
79072805 568 }
4757a243
LW
569 else if (uv == extra && !del) {
570 matches++;
01ec43d0 571 s += UTF8SKIP(s);
7d85a32c 572 d = uvuni_to_utf8(d, final);
4757a243
LW
573 continue;
574 }
01ec43d0
GS
575 matches++; /* "none+1" is delete character */
576 s += UTF8SKIP(s);
79072805 577 }
4757a243 578 }
16ec844d 579 if (grows || hibit) {
9b877dbb
IH
580 sv_setpvn(sv, (char*)dstart, d - dstart);
581 Safefree(dstart);
16ec844d
IH
582 if (grows && hibit)
583 Safefree(start);
9b877dbb
IH
584 }
585 else {
586 *d = '\0';
587 SvCUR_set(sv, d - dstart);
588 }
1aa99e6b 589 SvUTF8_on(sv);
4757a243
LW
590 SvSETMAGIC(sv);
591
592 return matches;
593}
594
595I32
864dbfa3 596Perl_do_trans(pTHX_ SV *sv)
4757a243
LW
597{
598 STRLEN len;
8e84507e 599 I32 hasutf = (PL_op->op_private &
036b4402 600 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
4757a243 601
2233f375 602 if (SvREADONLY(sv)) {
765f542d
NC
603 if (SvIsCOW(sv))
604 sv_force_normal_flags(sv, 0);
2233f375
NC
605 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
606 Perl_croak(aTHX_ PL_no_modify);
607 }
4757a243
LW
608 (void)SvPV(sv, len);
609 if (!len)
610 return 0;
d59e14db
RGS
611 if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) {
612 if (!SvPOKp(sv))
613 (void)SvPV_force(sv, len);
2de7b02f 614 (void)SvPOK_only_UTF8(sv);
d59e14db 615 }
4757a243 616
cea2e8a9 617 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 618
036b4402 619 switch (PL_op->op_private & ~hasutf & 63) {
4757a243 620 case 0:
01ec43d0
GS
621 if (hasutf)
622 return do_trans_simple_utf8(sv);
623 else
624 return do_trans_simple(sv);
4757a243
LW
625
626 case OPpTRANS_IDENTICAL:
05d340b8 627 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
01ec43d0
GS
628 if (hasutf)
629 return do_trans_count_utf8(sv);
630 else
631 return do_trans_count(sv);
4757a243
LW
632
633 default:
01ec43d0 634 if (hasutf)
036b4402 635 return do_trans_complex_utf8(sv);
4757a243 636 else
036b4402 637 return do_trans_complex(sv);
79072805 638 }
79072805
LW
639}
640
641void
864dbfa3 642Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
79072805
LW
643{
644 SV **oldmark = mark;
645 register I32 items = sp - mark;
79072805 646 register STRLEN len;
463ee0b2 647 STRLEN delimlen;
463ee0b2 648 STRLEN tmplen;
79072805 649
5ba99574
NC
650 (void) SvPV(del, delimlen); /* stringify and get the delimlen */
651 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
516a5887 652
79072805
LW
653 mark++;
654 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
07f14f54 655 (void)SvUPGRADE(sv, SVt_PV);
79072805
LW
656 if (SvLEN(sv) < len + items) { /* current length is way too short */
657 while (items-- > 0) {
1426bbf4 658 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
463ee0b2
LW
659 SvPV(*mark, tmplen);
660 len += tmplen;
79072805
LW
661 }
662 mark++;
663 }
664 SvGROW(sv, len + 1); /* so try to pre-extend */
665
666 mark = oldmark;
db7c17d7 667 items = sp - mark;
79072805
LW
668 ++mark;
669 }
670
8d6d96c1
HS
671 sv_setpv(sv, "");
672 if (PL_tainting && SvMAGICAL(sv))
673 SvTAINTED_off(sv);
674
463ee0b2 675 if (items-- > 0) {
92d29cee
JH
676 if (*mark)
677 sv_catsv(sv, *mark);
463ee0b2
LW
678 mark++;
679 }
8d6d96c1 680
c512ce4f 681 if (delimlen) {
79072805 682 for (; items > 0; items--,mark++) {
c512ce4f 683 sv_catsv(sv,del);
79072805
LW
684 sv_catsv(sv,*mark);
685 }
686 }
687 else {
688 for (; items > 0; items--,mark++)
689 sv_catsv(sv,*mark);
690 }
691 SvSETMAGIC(sv);
692}
693
694void
864dbfa3 695Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 696{
46fc3d4c 697 STRLEN patlen;
698 char *pat = SvPV(*sarg, patlen);
699 bool do_taint = FALSE;
700
5b781b5b 701 SvUTF8_off(sv);
2cf2cfc6
A
702 if (DO_UTF8(*sarg))
703 SvUTF8_on(sv);
46fc3d4c 704 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
79072805 705 SvSETMAGIC(sv);
46fc3d4c 706 if (do_taint)
707 SvTAINTED_on(sv);
79072805
LW
708}
709
33b45480 710/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0
JH
711UV
712Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
713{
714 STRLEN srclen, len;
715 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
716 UV retnum = 0;
717
a50d7633 718 if (offset < 0)
81e118e0 719 return retnum;
8e84507e 720 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 721 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 722
dcad2880 723 if (SvUTF8(sv))
33b45480 724 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 725
81e118e0
JH
726 offset *= size; /* turn into bit offset */
727 len = (offset + size + 7) / 8; /* required number of bytes */
728 if (len > srclen) {
729 if (size <= 8)
730 retnum = 0;
731 else {
732 offset >>= 3; /* turn into byte offset */
733 if (size == 16) {
eb160463 734 if ((STRLEN)offset >= srclen)
81e118e0
JH
735 retnum = 0;
736 else
628e1a40 737 retnum = (UV) s[offset] << 8;
81e118e0
JH
738 }
739 else if (size == 32) {
eb160463 740 if ((STRLEN)offset >= srclen)
81e118e0 741 retnum = 0;
eb160463 742 else if ((STRLEN)(offset + 1) >= srclen)
81e118e0
JH
743 retnum =
744 ((UV) s[offset ] << 24);
eb160463 745 else if ((STRLEN)(offset + 2) >= srclen)
81e118e0
JH
746 retnum =
747 ((UV) s[offset ] << 24) +
748 ((UV) s[offset + 1] << 16);
749 else
750 retnum =
751 ((UV) s[offset ] << 24) +
752 ((UV) s[offset + 1] << 16) +
753 ( s[offset + 2] << 8);
754 }
d7d93a81 755#ifdef UV_IS_QUAD
c5a0f51a 756 else if (size == 64) {
c5a0f51a 757 if (ckWARN(WARN_PORTABLE))
9014280d 758 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
c5a0f51a
JH
759 "Bit vector size > 32 non-portable");
760 if (offset >= srclen)
761 retnum = 0;
762 else if (offset + 1 >= srclen)
763 retnum =
764 (UV) s[offset ] << 56;
765 else if (offset + 2 >= srclen)
766 retnum =
767 ((UV) s[offset ] << 56) +
768 ((UV) s[offset + 1] << 48);
769 else if (offset + 3 >= srclen)
770 retnum =
771 ((UV) s[offset ] << 56) +
772 ((UV) s[offset + 1] << 48) +
773 ((UV) s[offset + 2] << 40);
774 else if (offset + 4 >= srclen)
775 retnum =
776 ((UV) s[offset ] << 56) +
777 ((UV) s[offset + 1] << 48) +
778 ((UV) s[offset + 2] << 40) +
779 ((UV) s[offset + 3] << 32);
780 else if (offset + 5 >= srclen)
781 retnum =
782 ((UV) s[offset ] << 56) +
783 ((UV) s[offset + 1] << 48) +
784 ((UV) s[offset + 2] << 40) +
785 ((UV) s[offset + 3] << 32) +
786 ( s[offset + 4] << 24);
787 else if (offset + 6 >= srclen)
788 retnum =
789 ((UV) s[offset ] << 56) +
790 ((UV) s[offset + 1] << 48) +
791 ((UV) s[offset + 2] << 40) +
792 ((UV) s[offset + 3] << 32) +
793 ((UV) s[offset + 4] << 24) +
794 ((UV) s[offset + 5] << 16);
795 else
8e84507e 796 retnum =
c5a0f51a
JH
797 ((UV) s[offset ] << 56) +
798 ((UV) s[offset + 1] << 48) +
799 ((UV) s[offset + 2] << 40) +
800 ((UV) s[offset + 3] << 32) +
801 ((UV) s[offset + 4] << 24) +
802 ((UV) s[offset + 5] << 16) +
628e1a40 803 ( s[offset + 6] << 8);
c5a0f51a
JH
804 }
805#endif
81e118e0
JH
806 }
807 }
808 else if (size < 8)
809 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
810 else {
811 offset >>= 3; /* turn into byte offset */
812 if (size == 8)
813 retnum = s[offset];
814 else if (size == 16)
815 retnum =
628e1a40 816 ((UV) s[offset] << 8) +
81e118e0
JH
817 s[offset + 1];
818 else if (size == 32)
819 retnum =
820 ((UV) s[offset ] << 24) +
821 ((UV) s[offset + 1] << 16) +
822 ( s[offset + 2] << 8) +
823 s[offset + 3];
d7d93a81 824#ifdef UV_IS_QUAD
c5a0f51a 825 else if (size == 64) {
c5a0f51a 826 if (ckWARN(WARN_PORTABLE))
9014280d 827 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
c5a0f51a
JH
828 "Bit vector size > 32 non-portable");
829 retnum =
830 ((UV) s[offset ] << 56) +
831 ((UV) s[offset + 1] << 48) +
832 ((UV) s[offset + 2] << 40) +
833 ((UV) s[offset + 3] << 32) +
834 ((UV) s[offset + 4] << 24) +
835 ((UV) s[offset + 5] << 16) +
628e1a40 836 ( s[offset + 6] << 8) +
c5a0f51a
JH
837 s[offset + 7];
838 }
839#endif
81e118e0
JH
840 }
841
842 return retnum;
843}
844
33b45480
SB
845/* currently converts input to bytes if possible but doesn't sweat failures,
846 * although it does ensure that the string it clobbers is not marked as
847 * utf8-valid any more
848 */
79072805 849void
864dbfa3 850Perl_do_vecset(pTHX_ SV *sv)
79072805
LW
851{
852 SV *targ = LvTARG(sv);
853 register I32 offset;
854 register I32 size;
8990e307 855 register unsigned char *s;
81e118e0 856 register UV lval;
79072805 857 I32 mask;
a0d0e21e
LW
858 STRLEN targlen;
859 STRLEN len;
79072805 860
8990e307
LW
861 if (!targ)
862 return;
a0d0e21e 863 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 864 if (SvUTF8(targ)) {
33b45480
SB
865 /* This is handled by the SvPOK_only below...
866 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
867 SvUTF8_off(targ);
868 */
869 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53
MG
870 }
871
4ebbc975 872 (void)SvPOK_only(targ);
81e118e0 873 lval = SvUV(sv);
79072805 874 offset = LvTARGOFF(sv);
fe58ced6 875 if (offset < 0)
ed9aa3b7 876 Perl_croak(aTHX_ "Negative offset to vec in lvalue context");
79072805 877 size = LvTARGLEN(sv);
8e84507e 878 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 879 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 880
81e118e0
JH
881 offset *= size; /* turn into bit offset */
882 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e
LW
883 if (len > targlen) {
884 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 885 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e
LW
886 SvCUR_set(targ, len);
887 }
8e84507e 888
79072805
LW
889 if (size < 8) {
890 mask = (1 << size) - 1;
891 size = offset & 7;
892 lval &= mask;
81e118e0 893 offset >>= 3; /* turn into byte offset */
79072805
LW
894 s[offset] &= ~(mask << size);
895 s[offset] |= lval << size;
896 }
897 else {
81e118e0 898 offset >>= 3; /* turn into byte offset */
79072805 899 if (size == 8)
eb160463 900 s[offset ] = (U8)( lval & 0xff);
79072805 901 else if (size == 16) {
eb160463
GS
902 s[offset ] = (U8)((lval >> 8) & 0xff);
903 s[offset+1] = (U8)( lval & 0xff);
79072805
LW
904 }
905 else if (size == 32) {
eb160463
GS
906 s[offset ] = (U8)((lval >> 24) & 0xff);
907 s[offset+1] = (U8)((lval >> 16) & 0xff);
908 s[offset+2] = (U8)((lval >> 8) & 0xff);
909 s[offset+3] = (U8)( lval & 0xff);
c5a0f51a 910 }
d7d93a81 911#ifdef UV_IS_QUAD
c5a0f51a 912 else if (size == 64) {
c5a0f51a 913 if (ckWARN(WARN_PORTABLE))
9014280d 914 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
c5a0f51a 915 "Bit vector size > 32 non-portable");
eb160463
GS
916 s[offset ] = (U8)((lval >> 56) & 0xff);
917 s[offset+1] = (U8)((lval >> 48) & 0xff);
918 s[offset+2] = (U8)((lval >> 40) & 0xff);
919 s[offset+3] = (U8)((lval >> 32) & 0xff);
920 s[offset+4] = (U8)((lval >> 24) & 0xff);
921 s[offset+5] = (U8)((lval >> 16) & 0xff);
922 s[offset+6] = (U8)((lval >> 8) & 0xff);
923 s[offset+7] = (U8)( lval & 0xff);
79072805 924 }
dc1e3f56 925#endif
79072805 926 }
7bb043c3 927 SvSETMAGIC(targ);
79072805
LW
928}
929
930void
864dbfa3 931Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 932{
463ee0b2 933 STRLEN len;
a0d0e21e 934 char *s;
8e84507e 935
79072805 936 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e
LW
937 register I32 i;
938 I32 max;
939 AV* av = (AV*)sv;
940 max = AvFILL(av);
941 for (i = 0; i <= max; i++) {
942 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 943 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
944 do_chop(astr, sv);
945 }
946 return;
79072805 947 }
aa854799 948 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
949 HV* hv = (HV*)sv;
950 HE* entry;
951 (void)hv_iterinit(hv);
952 /*SUPPRESS 560*/
155aba94 953 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
954 do_chop(astr,hv_iterval(hv,entry));
955 return;
79072805 956 }
991e6d41
NC
957 else if (SvREADONLY(sv)) {
958 if (SvFAKE(sv)) {
959 /* SV is copy-on-write */
960 sv_force_normal_flags(sv, 0);
961 }
962 if (SvREADONLY(sv))
963 Perl_croak(aTHX_ PL_no_modify);
964 }
a0d0e21e 965 s = SvPV(sv, len);
748a9306 966 if (len && !SvPOK(sv))
a0d0e21e 967 s = SvPV_force(sv, len);
7e2040f0 968 if (DO_UTF8(sv)) {
a0ed51b3
LW
969 if (s && len) {
970 char *send = s + len;
971 char *start = s;
972 s = send - 1;
a0dbb045
JH
973 while (s > start && UTF8_IS_CONTINUATION(*s))
974 s--;
9041c2e3 975 if (utf8_to_uvchr((U8*)s, 0)) {
a0dbb045
JH
976 sv_setpvn(astr, s, send - s);
977 *s = '\0';
978 SvCUR_set(sv, s - start);
979 SvNIOK_off(sv);
980 SvUTF8_on(astr);
981 }
a0ed51b3
LW
982 }
983 else
984 sv_setpvn(astr, "", 0);
985 }
7e2040f0 986 else if (s && len) {
a0d0e21e
LW
987 s += --len;
988 sv_setpvn(astr, s, 1);
989 *s = '\0';
990 SvCUR_set(sv, len);
2c19a612 991 SvUTF8_off(sv);
a0d0e21e 992 SvNIOK_off(sv);
79072805
LW
993 }
994 else
a0d0e21e
LW
995 sv_setpvn(astr, "", 0);
996 SvSETMAGIC(sv);
7e2040f0 997}
a0d0e21e
LW
998
999I32
864dbfa3 1000Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 1001{
c07a80fd 1002 register I32 count;
a0d0e21e 1003 STRLEN len;
9b33ce3b 1004 STRLEN n_a;
a0d0e21e 1005 char *s;
c07a80fd 1006
3280af22 1007 if (RsSNARF(PL_rs))
c07a80fd 1008 return 0;
4c5a6083
GS
1009 if (RsRECORD(PL_rs))
1010 return 0;
c07a80fd 1011 count = 0;
a0d0e21e
LW
1012 if (SvTYPE(sv) == SVt_PVAV) {
1013 register I32 i;
1014 I32 max;
1015 AV* av = (AV*)sv;
1016 max = AvFILL(av);
1017 for (i = 0; i <= max; i++) {
1018 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 1019 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
1020 count += do_chomp(sv);
1021 }
1022 return count;
1023 }
aa854799 1024 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
1025 HV* hv = (HV*)sv;
1026 HE* entry;
1027 (void)hv_iterinit(hv);
1028 /*SUPPRESS 560*/
155aba94 1029 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
1030 count += do_chomp(hv_iterval(hv,entry));
1031 return count;
1032 }
991e6d41
NC
1033 else if (SvREADONLY(sv)) {
1034 if (SvFAKE(sv)) {
1035 /* SV is copy-on-write */
1036 sv_force_normal_flags(sv, 0);
1037 }
1038 if (SvREADONLY(sv))
1039 Perl_croak(aTHX_ PL_no_modify);
1040 }
a0d0e21e 1041 s = SvPV(sv, len);
a0d0e21e
LW
1042 if (s && len) {
1043 s += --len;
3280af22 1044 if (RsPARA(PL_rs)) {
a0d0e21e
LW
1045 if (*s != '\n')
1046 goto nope;
1047 ++count;
1048 while (len && s[-1] == '\n') {
1049 --len;
1050 --s;
1051 ++count;
1052 }
1053 }
a0d0e21e 1054 else {
c07a80fd 1055 STRLEN rslen;
3280af22 1056 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 1057 if (rslen == 1) {
1058 if (*s != *rsptr)
1059 goto nope;
1060 ++count;
1061 }
1062 else {
8c2cee6f 1063 if (len < rslen - 1)
c07a80fd 1064 goto nope;
1065 len -= rslen - 1;
1066 s -= rslen - 1;
36477c24 1067 if (memNE(s, rsptr, rslen))
c07a80fd 1068 goto nope;
1069 count += rslen;
1070 }
a0d0e21e 1071 }
9b33ce3b 1072 s = SvPV_force(sv, n_a);
a0d0e21e 1073 SvCUR_set(sv, len);
9b33ce3b 1074 *SvEND(sv) = '\0';
a0d0e21e 1075 SvNIOK_off(sv);
9b33ce3b 1076 SvSETMAGIC(sv);
a0d0e21e
LW
1077 }
1078 nope:
a0d0e21e 1079 return count;
8e84507e 1080}
79072805
LW
1081
1082void
864dbfa3 1083Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805
LW
1084{
1085#ifdef LIBERAL
1086 register long *dl;
1087 register long *ll;
1088 register long *rl;
1089#endif
1090 register char *dc;
463ee0b2
LW
1091 STRLEN leftlen;
1092 STRLEN rightlen;
7a4c00b4 1093 register char *lc;
1094 register char *rc;
79072805 1095 register I32 len;
a0d0e21e 1096 I32 lensave;
7a4c00b4 1097 char *lsave;
1098 char *rsave;
0c57e439
GS
1099 bool left_utf = DO_UTF8(left);
1100 bool right_utf = DO_UTF8(right);
b7953727 1101 I32 needlen = 0;
0c57e439
GS
1102
1103 if (left_utf && !right_utf)
1104 sv_utf8_upgrade(right);
a1ca4561 1105 else if (!left_utf && right_utf)
0c57e439 1106 sv_utf8_upgrade(left);
79072805 1107
1fbd88dc
CS
1108 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1109 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 1110 lsave = lc = SvPV(left, leftlen);
1111 rsave = rc = SvPV(right, rightlen);
93a17b20 1112 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 1113 lensave = len;
c9b3c8d0
JH
1114 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1115 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1116 Newz(801, dc, needlen + 1, char);
1117 }
1118 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d
GS
1119 STRLEN n_a;
1120 dc = SvPV_force(sv, n_a);
eb160463
GS
1121 if (SvCUR(sv) < (STRLEN)len) {
1122 dc = SvGROW(sv, (STRLEN)(len + 1));
ff68c719 1123 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1124 }
c9b3c8d0
JH
1125 if (optype != OP_BIT_AND && (left_utf || right_utf))
1126 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 1127 }
1128 else {
c9b3c8d0
JH
1129 needlen = ((optype == OP_BIT_AND)
1130 ? len : (leftlen > rightlen ? leftlen : rightlen));
ff68c719 1131 Newz(801, dc, needlen + 1, char);
1132 (void)sv_usepvn(sv, dc, needlen);
1133 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 1134 }
a0d0e21e
LW
1135 SvCUR_set(sv, len);
1136 (void)SvPOK_only(sv);
0c57e439
GS
1137 if (left_utf || right_utf) {
1138 UV duc, luc, ruc;
c9b3c8d0 1139 char *dcsave = dc;
0c57e439
GS
1140 STRLEN lulen = leftlen;
1141 STRLEN rulen = rightlen;
ba210ebe 1142 STRLEN ulen;
0c57e439
GS
1143
1144 switch (optype) {
1145 case OP_BIT_AND:
1146 while (lulen && rulen) {
9041c2e3 1147 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1148 lc += ulen;
1149 lulen -= ulen;
9041c2e3 1150 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1151 rc += ulen;
1152 rulen -= ulen;
1153 duc = luc & ruc;
9041c2e3 1154 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439 1155 }
c9b3c8d0
JH
1156 if (sv == left || sv == right)
1157 (void)sv_usepvn(sv, dcsave, needlen);
1158 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1159 break;
1160 case OP_BIT_XOR:
1161 while (lulen && rulen) {
9041c2e3 1162 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1163 lc += ulen;
1164 lulen -= ulen;
9041c2e3 1165 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1166 rc += ulen;
1167 rulen -= ulen;
1168 duc = luc ^ ruc;
9041c2e3 1169 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1170 }
1171 goto mop_up_utf;
1172 case OP_BIT_OR:
1173 while (lulen && rulen) {
9041c2e3 1174 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1175 lc += ulen;
1176 lulen -= ulen;
9041c2e3 1177 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1178 rc += ulen;
1179 rulen -= ulen;
1180 duc = luc | ruc;
9041c2e3 1181 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1182 }
1183 mop_up_utf:
c9b3c8d0
JH
1184 if (sv == left || sv == right)
1185 (void)sv_usepvn(sv, dcsave, needlen);
1186 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1187 if (rulen)
1188 sv_catpvn(sv, rc, rulen);
1189 else if (lulen)
1190 sv_catpvn(sv, lc, lulen);
1191 else
1192 *SvEND(sv) = '\0';
1193 break;
1194 }
1195 SvUTF8_on(sv);
1196 goto finish;
1197 }
1198 else
79072805
LW
1199#ifdef LIBERAL
1200 if (len >= sizeof(long)*4 &&
1201 !((long)dc % sizeof(long)) &&
1202 !((long)lc % sizeof(long)) &&
1203 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1204 {
1205 I32 remainder = len % (sizeof(long)*4);
1206 len /= (sizeof(long)*4);
1207
1208 dl = (long*)dc;
1209 ll = (long*)lc;
1210 rl = (long*)rc;
1211
1212 switch (optype) {
1213 case OP_BIT_AND:
1214 while (len--) {
1215 *dl++ = *ll++ & *rl++;
1216 *dl++ = *ll++ & *rl++;
1217 *dl++ = *ll++ & *rl++;
1218 *dl++ = *ll++ & *rl++;
1219 }
1220 break;
a0d0e21e 1221 case OP_BIT_XOR:
79072805
LW
1222 while (len--) {
1223 *dl++ = *ll++ ^ *rl++;
1224 *dl++ = *ll++ ^ *rl++;
1225 *dl++ = *ll++ ^ *rl++;
1226 *dl++ = *ll++ ^ *rl++;
1227 }
1228 break;
1229 case OP_BIT_OR:
1230 while (len--) {
1231 *dl++ = *ll++ | *rl++;
1232 *dl++ = *ll++ | *rl++;
1233 *dl++ = *ll++ | *rl++;
1234 *dl++ = *ll++ | *rl++;
1235 }
1236 }
1237
1238 dc = (char*)dl;
1239 lc = (char*)ll;
1240 rc = (char*)rl;
1241
1242 len = remainder;
1243 }
1244#endif
a0d0e21e 1245 {
a0d0e21e
LW
1246 switch (optype) {
1247 case OP_BIT_AND:
1248 while (len--)
1249 *dc++ = *lc++ & *rc++;
1250 break;
1251 case OP_BIT_XOR:
1252 while (len--)
1253 *dc++ = *lc++ ^ *rc++;
1254 goto mop_up;
1255 case OP_BIT_OR:
1256 while (len--)
1257 *dc++ = *lc++ | *rc++;
1258 mop_up:
1259 len = lensave;
eb160463 1260 if (rightlen > (STRLEN)len)
a0d0e21e 1261 sv_catpvn(sv, rsave + len, rightlen - len);
eb160463 1262 else if (leftlen > (STRLEN)len)
a0d0e21e 1263 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4
LW
1264 else
1265 *SvEND(sv) = '\0';
a0d0e21e
LW
1266 break;
1267 }
79072805 1268 }
0c57e439 1269finish:
fb73857a 1270 SvTAINT(sv);
79072805 1271}
463ee0b2
LW
1272
1273OP *
cea2e8a9 1274Perl_do_kv(pTHX)
463ee0b2 1275{
39644a26 1276 dSP;
463ee0b2 1277 HV *hv = (HV*)POPs;
800e9ae0 1278 HV *keys;
463ee0b2 1279 register HE *entry;
463ee0b2 1280 SV *tmpstr;
54310121 1281 I32 gimme = GIMME_V;
533c011a
NIS
1282 I32 dokeys = (PL_op->op_type == OP_KEYS);
1283 I32 dovalues = (PL_op->op_type == OP_VALUES);
8e84507e
NIS
1284
1285 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1286 dokeys = dovalues = TRUE;
463ee0b2 1287
85581909 1288 if (!hv) {
78f9721b 1289 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
85581909
SB
1290 dTARGET; /* make sure to clear its target here */
1291 if (SvTYPE(TARG) == SVt_PVLV)
1292 LvTARG(TARG) = Nullsv;
1293 PUSHs(TARG);
1294 }
463ee0b2 1295 RETURN;
85581909 1296 }
748a9306 1297
6d822dc4 1298 keys = hv;
800e9ae0 1299 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1300
54310121 1301 if (gimme == G_VOID)
aa689395 1302 RETURN;
1303
54310121 1304 if (gimme == G_SCALAR) {
6ee623d5 1305 IV i;
463ee0b2
LW
1306 dTARGET;
1307
78f9721b 1308 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
85581909
SB
1309 if (SvTYPE(TARG) < SVt_PVLV) {
1310 sv_upgrade(TARG, SVt_PVLV);
14befaf4 1311 sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
85581909
SB
1312 }
1313 LvTYPE(TARG) = 'k';
800e9ae0 1314 if (LvTARG(TARG) != (SV*)keys) {
6ff81951
GS
1315 if (LvTARG(TARG))
1316 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1317 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1318 }
85581909
SB
1319 PUSHs(TARG);
1320 RETURN;
1321 }
1322
14befaf4 1323 if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied))
a98face1 1324 i = HvKEYS(keys);
463ee0b2
LW
1325 else {
1326 i = 0;
463ee0b2 1327 /*SUPPRESS 560*/
800e9ae0 1328 while (hv_iternext(keys)) i++;
463ee0b2
LW
1329 }
1330 PUSHi( i );
1331 RETURN;
1332 }
1333
8ed4b672 1334 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1335
463ee0b2 1336 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1337 while ((entry = hv_iternext(keys))) {
463ee0b2 1338 SPAGAIN;
574c8022
JH
1339 if (dokeys) {
1340 SV* sv = hv_iterkeysv(entry);
574c8022
JH
1341 XPUSHs(sv); /* won't clobber stack_sp */
1342 }
463ee0b2 1343 if (dovalues) {
463ee0b2 1344 PUTBACK;
6d822dc4 1345 tmpstr = hv_iterval(hv,entry);
cea2e8a9 1346 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1347 (unsigned long)HeHASH(entry),
800e9ae0
JP
1348 HvMAX(keys)+1,
1349 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1350 SPAGAIN;
46fc3d4c 1351 XPUSHs(tmpstr);
463ee0b2
LW
1352 }
1353 PUTBACK;
1354 }
1355 return NORMAL;
1356}
4e35701f 1357