This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
AUTHORS updates.
[perl5.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
bc89e66f 3 * Copyright (c) 1991-2001, 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
GS
48 matches++;
49 *s++ = ch;
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
036b4402 95S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
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
b250498f 133S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
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;
144 STRLEN len, rlen;
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) {
036b4402
GS
162 *d = ch;
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++;
178 *d++ = ch;
179 }
180 else if (ch == -1) /* -1 is unmapped character */
181 *d++ = *s;
182 else if (ch == -2) /* -2 is delete character */
183 matches++;
184 s++;
185 }
186 }
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
JH
219 tbl[comp+1] : tbl[0x100+rlen];
220 if (ch != pch) {
9041c2e3 221 d = uvchr_to_utf8(d, ch);
8973db79
JH
222 pch = ch;
223 }
224 s += len;
225 continue;
226 }
227 }
9b877dbb 228 }
1aa99e6b
IH
229 else if ((ch = tbl[comp]) >= 0) {
230 matches++;
231 if (ch != pch) {
9041c2e3 232 d = uvchr_to_utf8(d, ch);
1aa99e6b
IH
233 pch = ch;
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
036b4402 295S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
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;
311 UV final;
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);
9041c2e3 389 /* Downgrading just 'cos it will is suspect - NI-S */
1aa99e6b
IH
390 if (!isutf8 && !(PL_hints & HINT_UTF8))
391 sv_utf8_downgrade(sv, TRUE);
4757a243
LW
392
393 return matches;
394}
395
942e002e 396STATIC I32
036b4402 397S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
4757a243 398{
4757a243 399 U8 *s;
1aa99e6b 400 U8 *start, *send;
4757a243
LW
401 I32 matches = 0;
402 STRLEN len;
403
404 SV* rv = (SV*)cSVOP->op_sv;
405 HV* hv = (HV*)SvRV(rv);
406 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
407 UV none = svp ? SvUV(*svp) : 0x7fffffff;
45005bfb 408 UV extra = none + 1;
4757a243 409 UV uv;
1aa99e6b 410 U8 hibit = 0;
4757a243
LW
411
412 s = (U8*)SvPV(sv, len);
1aa99e6b
IH
413 if (!SvUTF8(sv)) {
414 U8 *t = s, *e = s + len;
c4d5f83a
NIS
415 while (t < e) {
416 U8 ch = *t++;
417 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
1aa99e6b 418 break;
c4d5f83a 419 }
1aa99e6b
IH
420 if (hibit)
421 start = s = bytes_to_utf8(s, &len);
422 }
4757a243
LW
423 send = s + len;
424
425 while (s < send) {
3568d838 426 if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra)
4757a243 427 matches++;
834a4ddd 428 s += UTF8SKIP(s);
4757a243 429 }
1aa99e6b
IH
430 if (hibit)
431 Safefree(start);
4757a243
LW
432
433 return matches;
434}
435
942e002e 436STATIC I32
036b4402 437S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
4757a243 438{
4757a243 439 U8 *s;
1aa99e6b 440 U8 *start, *send;
4757a243
LW
441 U8 *d;
442 I32 matches = 0;
443 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
4757a243 444 I32 del = PL_op->op_private & OPpTRANS_DELETE;
9b877dbb 445 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
4757a243
LW
446 SV* rv = (SV*)cSVOP->op_sv;
447 HV* hv = (HV*)SvRV(rv);
448 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
449 UV none = svp ? SvUV(*svp) : 0x7fffffff;
450 UV extra = none + 1;
451 UV final;
45005bfb 452 bool havefinal = FALSE;
4757a243
LW
453 UV uv;
454 STRLEN len;
1aa99e6b
IH
455 U8 *dstart, *dend;
456 I32 isutf8;
457 U8 hibit = 0;
4757a243
LW
458
459 s = (U8*)SvPV(sv, len);
1aa99e6b
IH
460 isutf8 = SvUTF8(sv);
461 if (!isutf8) {
462 U8 *t = s, *e = s + len;
c4d5f83a
NIS
463 while (t < e) {
464 U8 ch = *t++;
465 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
1aa99e6b 466 break;
c4d5f83a 467 }
1aa99e6b
IH
468 if (hibit)
469 s = bytes_to_utf8(s, &len);
470 }
4757a243 471 send = s + len;
1aa99e6b 472 start = s;
4757a243
LW
473
474 svp = hv_fetch(hv, "FINAL", 5, FALSE);
45005bfb 475 if (svp) {
4757a243 476 final = SvUV(*svp);
45005bfb
JH
477 havefinal = TRUE;
478 }
4757a243 479
9b877dbb
IH
480 if (grows) {
481 /* d needs to be bigger than s, in case e.g. upgrading is required */
482 New(0, d, len*3+UTF8_MAXLEN, U8);
483 dend = d + len * 3;
484 dstart = d;
485 }
486 else {
487 dstart = d = s;
488 dend = d + len;
489 }
4757a243
LW
490
491 if (squash) {
492 UV puv = 0xfeedface;
493 while (s < send) {
3568d838 494 uv = swash_fetch(rv, s, TRUE);
9041c2e3 495
9b877dbb
IH
496 if (d > dend) {
497 STRLEN clen = d - dstart;
498 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
499 if (!grows)
7d85a32c 500 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
1aa99e6b
IH
501 Renew(dstart, nlen+UTF8_MAXLEN, U8);
502 d = dstart + clen;
503 dend = dstart + nlen;
4757a243
LW
504 }
505 if (uv < none) {
506 matches++;
94472101 507 s += UTF8SKIP(s);
4757a243 508 if (uv != puv) {
7d85a32c 509 d = uvuni_to_utf8(d, uv);
4757a243
LW
510 puv = uv;
511 }
4757a243
LW
512 continue;
513 }
514 else if (uv == none) { /* "none" is unmapped character */
1aa99e6b 515 int i = UTF8SKIP(s);
9b877dbb
IH
516 Copy(s, d, i, U8);
517 d += i;
518 s += i;
4757a243
LW
519 puv = 0xfeedface;
520 continue;
a0ed51b3 521 }
4757a243
LW
522 else if (uv == extra && !del) {
523 matches++;
45005bfb
JH
524 if (havefinal) {
525 s += UTF8SKIP(s);
526 if (puv != final) {
7d85a32c 527 d = uvuni_to_utf8(d, final);
45005bfb
JH
528 puv = final;
529 }
530 }
531 else {
532 STRLEN len;
7d85a32c 533 uv = utf8_to_uvuni(s, &len);
45005bfb
JH
534 if (uv != puv) {
535 Copy(s, d, len, U8);
536 d += len;
537 puv = uv;
538 }
539 s += len;
4757a243 540 }
4757a243
LW
541 continue;
542 }
01ec43d0
GS
543 matches++; /* "none+1" is delete character */
544 s += UTF8SKIP(s);
a0ed51b3 545 }
79072805
LW
546 }
547 else {
4757a243 548 while (s < send) {
3568d838 549 uv = swash_fetch(rv, s, TRUE);
9b877dbb
IH
550 if (d > dend) {
551 STRLEN clen = d - dstart;
552 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
553 if (!grows)
7d85a32c 554 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
1aa99e6b
IH
555 Renew(dstart, nlen+UTF8_MAXLEN, U8);
556 d = dstart + clen;
557 dend = dstart + nlen;
a0ed51b3 558 }
4757a243
LW
559 if (uv < none) {
560 matches++;
01ec43d0 561 s += UTF8SKIP(s);
7d85a32c 562 d = uvuni_to_utf8(d, uv);
4757a243 563 continue;
a0ed51b3 564 }
4757a243 565 else if (uv == none) { /* "none" is unmapped character */
1aa99e6b 566 int i = UTF8SKIP(s);
9b877dbb
IH
567 Copy(s, d, i, U8);
568 d += i;
569 s += i;
4757a243 570 continue;
79072805 571 }
4757a243
LW
572 else if (uv == extra && !del) {
573 matches++;
01ec43d0 574 s += UTF8SKIP(s);
7d85a32c 575 d = uvuni_to_utf8(d, final);
4757a243
LW
576 continue;
577 }
01ec43d0
GS
578 matches++; /* "none+1" is delete character */
579 s += UTF8SKIP(s);
79072805 580 }
4757a243 581 }
16ec844d 582 if (grows || hibit) {
9b877dbb
IH
583 sv_setpvn(sv, (char*)dstart, d - dstart);
584 Safefree(dstart);
16ec844d
IH
585 if (grows && hibit)
586 Safefree(start);
9b877dbb
IH
587 }
588 else {
589 *d = '\0';
590 SvCUR_set(sv, d - dstart);
591 }
1aa99e6b 592 SvUTF8_on(sv);
1aa99e6b
IH
593 if (!isutf8 && !(PL_hints & HINT_UTF8))
594 sv_utf8_downgrade(sv, TRUE);
4757a243
LW
595 SvSETMAGIC(sv);
596
597 return matches;
598}
599
600I32
864dbfa3 601Perl_do_trans(pTHX_ SV *sv)
4757a243
LW
602{
603 STRLEN len;
8e84507e 604 I32 hasutf = (PL_op->op_private &
036b4402 605 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
4757a243
LW
606
607 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
cea2e8a9 608 Perl_croak(aTHX_ PL_no_modify);
4757a243
LW
609
610 (void)SvPV(sv, len);
611 if (!len)
612 return 0;
613 if (!SvPOKp(sv))
614 (void)SvPV_force(sv, len);
2de7b02f
GS
615 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
616 (void)SvPOK_only_UTF8(sv);
4757a243 617
cea2e8a9 618 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 619
036b4402 620 switch (PL_op->op_private & ~hasutf & 63) {
4757a243 621 case 0:
01ec43d0
GS
622 if (hasutf)
623 return do_trans_simple_utf8(sv);
624 else
625 return do_trans_simple(sv);
4757a243
LW
626
627 case OPpTRANS_IDENTICAL:
05d340b8 628 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
01ec43d0
GS
629 if (hasutf)
630 return do_trans_count_utf8(sv);
631 else
632 return do_trans_count(sv);
4757a243
LW
633
634 default:
01ec43d0 635 if (hasutf)
036b4402 636 return do_trans_complex_utf8(sv);
4757a243 637 else
036b4402 638 return do_trans_complex(sv);
79072805 639 }
79072805
LW
640}
641
642void
864dbfa3 643Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
79072805
LW
644{
645 SV **oldmark = mark;
646 register I32 items = sp - mark;
79072805 647 register STRLEN len;
463ee0b2
LW
648 STRLEN delimlen;
649 register char *delim = SvPV(del, delimlen);
650 STRLEN tmplen;
79072805
LW
651
652 mark++;
653 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
07f14f54 654 (void)SvUPGRADE(sv, SVt_PV);
79072805
LW
655 if (SvLEN(sv) < len + items) { /* current length is way too short */
656 while (items-- > 0) {
1426bbf4 657 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
463ee0b2
LW
658 SvPV(*mark, tmplen);
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
463ee0b2 670 if (items-- > 0) {
92d29cee
JH
671 sv_setpv(sv, "");
672 if (*mark)
673 sv_catsv(sv, *mark);
463ee0b2
LW
674 mark++;
675 }
79072805
LW
676 else
677 sv_setpv(sv,"");
c512ce4f 678 if (delimlen) {
79072805 679 for (; items > 0; items--,mark++) {
c512ce4f 680 sv_catsv(sv,del);
79072805
LW
681 sv_catsv(sv,*mark);
682 }
683 }
684 else {
685 for (; items > 0; items--,mark++)
686 sv_catsv(sv,*mark);
687 }
688 SvSETMAGIC(sv);
689}
690
691void
864dbfa3 692Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 693{
46fc3d4c 694 STRLEN patlen;
695 char *pat = SvPV(*sarg, patlen);
696 bool do_taint = FALSE;
697
698 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
79072805 699 SvSETMAGIC(sv);
46fc3d4c 700 if (do_taint)
701 SvTAINTED_on(sv);
79072805
LW
702}
703
33b45480 704/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0
JH
705UV
706Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
707{
708 STRLEN srclen, len;
709 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
710 UV retnum = 0;
711
a50d7633 712 if (offset < 0)
81e118e0 713 return retnum;
8e84507e 714 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 715 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 716
dcad2880 717 if (SvUTF8(sv))
33b45480 718 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 719
81e118e0
JH
720 offset *= size; /* turn into bit offset */
721 len = (offset + size + 7) / 8; /* required number of bytes */
722 if (len > srclen) {
723 if (size <= 8)
724 retnum = 0;
725 else {
726 offset >>= 3; /* turn into byte offset */
727 if (size == 16) {
728 if (offset >= srclen)
729 retnum = 0;
730 else
628e1a40 731 retnum = (UV) s[offset] << 8;
81e118e0
JH
732 }
733 else if (size == 32) {
734 if (offset >= srclen)
735 retnum = 0;
736 else if (offset + 1 >= srclen)
737 retnum =
738 ((UV) s[offset ] << 24);
739 else if (offset + 2 >= srclen)
740 retnum =
741 ((UV) s[offset ] << 24) +
742 ((UV) s[offset + 1] << 16);
743 else
744 retnum =
745 ((UV) s[offset ] << 24) +
746 ((UV) s[offset + 1] << 16) +
747 ( s[offset + 2] << 8);
748 }
d7d93a81 749#ifdef UV_IS_QUAD
c5a0f51a 750 else if (size == 64) {
c5a0f51a
JH
751 if (ckWARN(WARN_PORTABLE))
752 Perl_warner(aTHX_ WARN_PORTABLE,
753 "Bit vector size > 32 non-portable");
754 if (offset >= srclen)
755 retnum = 0;
756 else if (offset + 1 >= srclen)
757 retnum =
758 (UV) s[offset ] << 56;
759 else if (offset + 2 >= srclen)
760 retnum =
761 ((UV) s[offset ] << 56) +
762 ((UV) s[offset + 1] << 48);
763 else if (offset + 3 >= srclen)
764 retnum =
765 ((UV) s[offset ] << 56) +
766 ((UV) s[offset + 1] << 48) +
767 ((UV) s[offset + 2] << 40);
768 else if (offset + 4 >= srclen)
769 retnum =
770 ((UV) s[offset ] << 56) +
771 ((UV) s[offset + 1] << 48) +
772 ((UV) s[offset + 2] << 40) +
773 ((UV) s[offset + 3] << 32);
774 else if (offset + 5 >= 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 ( s[offset + 4] << 24);
781 else if (offset + 6 >= srclen)
782 retnum =
783 ((UV) s[offset ] << 56) +
784 ((UV) s[offset + 1] << 48) +
785 ((UV) s[offset + 2] << 40) +
786 ((UV) s[offset + 3] << 32) +
787 ((UV) s[offset + 4] << 24) +
788 ((UV) s[offset + 5] << 16);
789 else
8e84507e 790 retnum =
c5a0f51a
JH
791 ((UV) s[offset ] << 56) +
792 ((UV) s[offset + 1] << 48) +
793 ((UV) s[offset + 2] << 40) +
794 ((UV) s[offset + 3] << 32) +
795 ((UV) s[offset + 4] << 24) +
796 ((UV) s[offset + 5] << 16) +
628e1a40 797 ( s[offset + 6] << 8);
c5a0f51a
JH
798 }
799#endif
81e118e0
JH
800 }
801 }
802 else if (size < 8)
803 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
804 else {
805 offset >>= 3; /* turn into byte offset */
806 if (size == 8)
807 retnum = s[offset];
808 else if (size == 16)
809 retnum =
628e1a40 810 ((UV) s[offset] << 8) +
81e118e0
JH
811 s[offset + 1];
812 else if (size == 32)
813 retnum =
814 ((UV) s[offset ] << 24) +
815 ((UV) s[offset + 1] << 16) +
816 ( s[offset + 2] << 8) +
817 s[offset + 3];
d7d93a81 818#ifdef UV_IS_QUAD
c5a0f51a 819 else if (size == 64) {
c5a0f51a
JH
820 if (ckWARN(WARN_PORTABLE))
821 Perl_warner(aTHX_ WARN_PORTABLE,
822 "Bit vector size > 32 non-portable");
823 retnum =
824 ((UV) s[offset ] << 56) +
825 ((UV) s[offset + 1] << 48) +
826 ((UV) s[offset + 2] << 40) +
827 ((UV) s[offset + 3] << 32) +
828 ((UV) s[offset + 4] << 24) +
829 ((UV) s[offset + 5] << 16) +
628e1a40 830 ( s[offset + 6] << 8) +
c5a0f51a
JH
831 s[offset + 7];
832 }
833#endif
81e118e0
JH
834 }
835
836 return retnum;
837}
838
33b45480
SB
839/* currently converts input to bytes if possible but doesn't sweat failures,
840 * although it does ensure that the string it clobbers is not marked as
841 * utf8-valid any more
842 */
79072805 843void
864dbfa3 844Perl_do_vecset(pTHX_ SV *sv)
79072805
LW
845{
846 SV *targ = LvTARG(sv);
847 register I32 offset;
848 register I32 size;
8990e307 849 register unsigned char *s;
81e118e0 850 register UV lval;
79072805 851 I32 mask;
a0d0e21e
LW
852 STRLEN targlen;
853 STRLEN len;
79072805 854
8990e307
LW
855 if (!targ)
856 return;
a0d0e21e 857 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 858 if (SvUTF8(targ)) {
33b45480
SB
859 /* This is handled by the SvPOK_only below...
860 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
861 SvUTF8_off(targ);
862 */
863 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53
MG
864 }
865
4ebbc975 866 (void)SvPOK_only(targ);
81e118e0 867 lval = SvUV(sv);
79072805 868 offset = LvTARGOFF(sv);
fe58ced6
MG
869 if (offset < 0)
870 Perl_croak(aTHX_ "Assigning to negative offset in vec");
79072805 871 size = LvTARGLEN(sv);
8e84507e 872 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 873 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 874
81e118e0
JH
875 offset *= size; /* turn into bit offset */
876 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e
LW
877 if (len > targlen) {
878 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 879 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e
LW
880 SvCUR_set(targ, len);
881 }
8e84507e 882
79072805
LW
883 if (size < 8) {
884 mask = (1 << size) - 1;
885 size = offset & 7;
886 lval &= mask;
81e118e0 887 offset >>= 3; /* turn into byte offset */
79072805
LW
888 s[offset] &= ~(mask << size);
889 s[offset] |= lval << size;
890 }
891 else {
81e118e0 892 offset >>= 3; /* turn into byte offset */
79072805 893 if (size == 8)
c5a0f51a 894 s[offset ] = lval & 0xff;
79072805 895 else if (size == 16) {
c5a0f51a
JH
896 s[offset ] = (lval >> 8) & 0xff;
897 s[offset+1] = lval & 0xff;
79072805
LW
898 }
899 else if (size == 32) {
c5a0f51a
JH
900 s[offset ] = (lval >> 24) & 0xff;
901 s[offset+1] = (lval >> 16) & 0xff;
902 s[offset+2] = (lval >> 8) & 0xff;
903 s[offset+3] = lval & 0xff;
904 }
d7d93a81 905#ifdef UV_IS_QUAD
c5a0f51a 906 else if (size == 64) {
c5a0f51a
JH
907 if (ckWARN(WARN_PORTABLE))
908 Perl_warner(aTHX_ WARN_PORTABLE,
909 "Bit vector size > 32 non-portable");
910 s[offset ] = (lval >> 56) & 0xff;
911 s[offset+1] = (lval >> 48) & 0xff;
912 s[offset+2] = (lval >> 40) & 0xff;
913 s[offset+3] = (lval >> 32) & 0xff;
914 s[offset+4] = (lval >> 24) & 0xff;
915 s[offset+5] = (lval >> 16) & 0xff;
916 s[offset+6] = (lval >> 8) & 0xff;
917 s[offset+7] = lval & 0xff;
79072805 918 }
dc1e3f56 919#endif
79072805 920 }
7bb043c3 921 SvSETMAGIC(targ);
79072805
LW
922}
923
924void
864dbfa3 925Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 926{
463ee0b2 927 STRLEN len;
a0d0e21e 928 char *s;
8e84507e 929
79072805 930 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e
LW
931 register I32 i;
932 I32 max;
933 AV* av = (AV*)sv;
934 max = AvFILL(av);
935 for (i = 0; i <= max; i++) {
936 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 937 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
938 do_chop(astr, sv);
939 }
940 return;
79072805 941 }
aa854799 942 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
943 HV* hv = (HV*)sv;
944 HE* entry;
945 (void)hv_iterinit(hv);
946 /*SUPPRESS 560*/
155aba94 947 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
948 do_chop(astr,hv_iterval(hv,entry));
949 return;
79072805 950 }
aa854799 951 else if (SvREADONLY(sv))
cea2e8a9 952 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 953 s = SvPV(sv, len);
748a9306 954 if (len && !SvPOK(sv))
a0d0e21e 955 s = SvPV_force(sv, len);
7e2040f0 956 if (DO_UTF8(sv)) {
a0ed51b3
LW
957 if (s && len) {
958 char *send = s + len;
959 char *start = s;
960 s = send - 1;
a0dbb045
JH
961 while (s > start && UTF8_IS_CONTINUATION(*s))
962 s--;
9041c2e3 963 if (utf8_to_uvchr((U8*)s, 0)) {
a0dbb045
JH
964 sv_setpvn(astr, s, send - s);
965 *s = '\0';
966 SvCUR_set(sv, s - start);
967 SvNIOK_off(sv);
968 SvUTF8_on(astr);
969 }
a0ed51b3
LW
970 }
971 else
972 sv_setpvn(astr, "", 0);
973 }
7e2040f0 974 else if (s && len) {
a0d0e21e
LW
975 s += --len;
976 sv_setpvn(astr, s, 1);
977 *s = '\0';
978 SvCUR_set(sv, len);
2c19a612 979 SvUTF8_off(sv);
a0d0e21e 980 SvNIOK_off(sv);
79072805
LW
981 }
982 else
a0d0e21e
LW
983 sv_setpvn(astr, "", 0);
984 SvSETMAGIC(sv);
7e2040f0 985}
a0d0e21e
LW
986
987I32
864dbfa3 988Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 989{
c07a80fd 990 register I32 count;
a0d0e21e
LW
991 STRLEN len;
992 char *s;
c07a80fd 993
3280af22 994 if (RsSNARF(PL_rs))
c07a80fd 995 return 0;
4c5a6083
GS
996 if (RsRECORD(PL_rs))
997 return 0;
c07a80fd 998 count = 0;
a0d0e21e
LW
999 if (SvTYPE(sv) == SVt_PVAV) {
1000 register I32 i;
1001 I32 max;
1002 AV* av = (AV*)sv;
1003 max = AvFILL(av);
1004 for (i = 0; i <= max; i++) {
1005 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 1006 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
1007 count += do_chomp(sv);
1008 }
1009 return count;
1010 }
aa854799 1011 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
1012 HV* hv = (HV*)sv;
1013 HE* entry;
1014 (void)hv_iterinit(hv);
1015 /*SUPPRESS 560*/
155aba94 1016 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
1017 count += do_chomp(hv_iterval(hv,entry));
1018 return count;
1019 }
aa854799 1020 else if (SvREADONLY(sv))
cea2e8a9 1021 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e
LW
1022 s = SvPV(sv, len);
1023 if (len && !SvPOKp(sv))
1024 s = SvPV_force(sv, len);
1025 if (s && len) {
1026 s += --len;
3280af22 1027 if (RsPARA(PL_rs)) {
a0d0e21e
LW
1028 if (*s != '\n')
1029 goto nope;
1030 ++count;
1031 while (len && s[-1] == '\n') {
1032 --len;
1033 --s;
1034 ++count;
1035 }
1036 }
a0d0e21e 1037 else {
c07a80fd 1038 STRLEN rslen;
3280af22 1039 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 1040 if (rslen == 1) {
1041 if (*s != *rsptr)
1042 goto nope;
1043 ++count;
1044 }
1045 else {
8c2cee6f 1046 if (len < rslen - 1)
c07a80fd 1047 goto nope;
1048 len -= rslen - 1;
1049 s -= rslen - 1;
36477c24 1050 if (memNE(s, rsptr, rslen))
c07a80fd 1051 goto nope;
1052 count += rslen;
1053 }
a0d0e21e 1054 }
a0d0e21e
LW
1055 *s = '\0';
1056 SvCUR_set(sv, len);
1057 SvNIOK_off(sv);
1058 }
1059 nope:
1060 SvSETMAGIC(sv);
1061 return count;
8e84507e 1062}
79072805
LW
1063
1064void
864dbfa3 1065Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805
LW
1066{
1067#ifdef LIBERAL
1068 register long *dl;
1069 register long *ll;
1070 register long *rl;
1071#endif
1072 register char *dc;
463ee0b2
LW
1073 STRLEN leftlen;
1074 STRLEN rightlen;
7a4c00b4 1075 register char *lc;
1076 register char *rc;
79072805 1077 register I32 len;
a0d0e21e 1078 I32 lensave;
7a4c00b4 1079 char *lsave;
1080 char *rsave;
0c57e439
GS
1081 bool left_utf = DO_UTF8(left);
1082 bool right_utf = DO_UTF8(right);
c9b3c8d0 1083 I32 needlen;
0c57e439
GS
1084
1085 if (left_utf && !right_utf)
1086 sv_utf8_upgrade(right);
a1ca4561 1087 else if (!left_utf && right_utf)
0c57e439 1088 sv_utf8_upgrade(left);
79072805 1089
1fbd88dc
CS
1090 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1091 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 1092 lsave = lc = SvPV(left, leftlen);
1093 rsave = rc = SvPV(right, rightlen);
93a17b20 1094 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 1095 lensave = len;
c9b3c8d0
JH
1096 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1097 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1098 Newz(801, dc, needlen + 1, char);
1099 }
1100 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d
GS
1101 STRLEN n_a;
1102 dc = SvPV_force(sv, n_a);
ff68c719 1103 if (SvCUR(sv) < len) {
1104 dc = SvGROW(sv, len + 1);
1105 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1106 }
c9b3c8d0
JH
1107 if (optype != OP_BIT_AND && (left_utf || right_utf))
1108 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 1109 }
1110 else {
c9b3c8d0
JH
1111 needlen = ((optype == OP_BIT_AND)
1112 ? len : (leftlen > rightlen ? leftlen : rightlen));
ff68c719 1113 Newz(801, dc, needlen + 1, char);
1114 (void)sv_usepvn(sv, dc, needlen);
1115 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 1116 }
a0d0e21e
LW
1117 SvCUR_set(sv, len);
1118 (void)SvPOK_only(sv);
0c57e439
GS
1119 if (left_utf || right_utf) {
1120 UV duc, luc, ruc;
c9b3c8d0 1121 char *dcsave = dc;
0c57e439
GS
1122 STRLEN lulen = leftlen;
1123 STRLEN rulen = rightlen;
ba210ebe 1124 STRLEN ulen;
0c57e439
GS
1125
1126 switch (optype) {
1127 case OP_BIT_AND:
1128 while (lulen && rulen) {
9041c2e3 1129 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1130 lc += ulen;
1131 lulen -= ulen;
9041c2e3 1132 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1133 rc += ulen;
1134 rulen -= ulen;
1135 duc = luc & ruc;
9041c2e3 1136 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439 1137 }
c9b3c8d0
JH
1138 if (sv == left || sv == right)
1139 (void)sv_usepvn(sv, dcsave, needlen);
1140 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1141 break;
1142 case OP_BIT_XOR:
1143 while (lulen && rulen) {
9041c2e3 1144 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1145 lc += ulen;
1146 lulen -= ulen;
9041c2e3 1147 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1148 rc += ulen;
1149 rulen -= ulen;
1150 duc = luc ^ ruc;
9041c2e3 1151 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1152 }
1153 goto mop_up_utf;
1154 case OP_BIT_OR:
1155 while (lulen && rulen) {
9041c2e3 1156 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1157 lc += ulen;
1158 lulen -= ulen;
9041c2e3 1159 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1160 rc += ulen;
1161 rulen -= ulen;
1162 duc = luc | ruc;
9041c2e3 1163 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
0c57e439
GS
1164 }
1165 mop_up_utf:
c9b3c8d0
JH
1166 if (sv == left || sv == right)
1167 (void)sv_usepvn(sv, dcsave, needlen);
1168 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1169 if (rulen)
1170 sv_catpvn(sv, rc, rulen);
1171 else if (lulen)
1172 sv_catpvn(sv, lc, lulen);
1173 else
1174 *SvEND(sv) = '\0';
1175 break;
1176 }
1177 SvUTF8_on(sv);
1178 goto finish;
1179 }
1180 else
79072805
LW
1181#ifdef LIBERAL
1182 if (len >= sizeof(long)*4 &&
1183 !((long)dc % sizeof(long)) &&
1184 !((long)lc % sizeof(long)) &&
1185 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1186 {
1187 I32 remainder = len % (sizeof(long)*4);
1188 len /= (sizeof(long)*4);
1189
1190 dl = (long*)dc;
1191 ll = (long*)lc;
1192 rl = (long*)rc;
1193
1194 switch (optype) {
1195 case OP_BIT_AND:
1196 while (len--) {
1197 *dl++ = *ll++ & *rl++;
1198 *dl++ = *ll++ & *rl++;
1199 *dl++ = *ll++ & *rl++;
1200 *dl++ = *ll++ & *rl++;
1201 }
1202 break;
a0d0e21e 1203 case OP_BIT_XOR:
79072805
LW
1204 while (len--) {
1205 *dl++ = *ll++ ^ *rl++;
1206 *dl++ = *ll++ ^ *rl++;
1207 *dl++ = *ll++ ^ *rl++;
1208 *dl++ = *ll++ ^ *rl++;
1209 }
1210 break;
1211 case OP_BIT_OR:
1212 while (len--) {
1213 *dl++ = *ll++ | *rl++;
1214 *dl++ = *ll++ | *rl++;
1215 *dl++ = *ll++ | *rl++;
1216 *dl++ = *ll++ | *rl++;
1217 }
1218 }
1219
1220 dc = (char*)dl;
1221 lc = (char*)ll;
1222 rc = (char*)rl;
1223
1224 len = remainder;
1225 }
1226#endif
a0d0e21e 1227 {
a0d0e21e
LW
1228 switch (optype) {
1229 case OP_BIT_AND:
1230 while (len--)
1231 *dc++ = *lc++ & *rc++;
1232 break;
1233 case OP_BIT_XOR:
1234 while (len--)
1235 *dc++ = *lc++ ^ *rc++;
1236 goto mop_up;
1237 case OP_BIT_OR:
1238 while (len--)
1239 *dc++ = *lc++ | *rc++;
1240 mop_up:
1241 len = lensave;
1242 if (rightlen > len)
1243 sv_catpvn(sv, rsave + len, rightlen - len);
1244 else if (leftlen > len)
1245 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4
LW
1246 else
1247 *SvEND(sv) = '\0';
a0d0e21e
LW
1248 break;
1249 }
79072805 1250 }
0c57e439 1251finish:
fb73857a 1252 SvTAINT(sv);
79072805 1253}
463ee0b2
LW
1254
1255OP *
cea2e8a9 1256Perl_do_kv(pTHX)
463ee0b2 1257{
39644a26 1258 dSP;
463ee0b2 1259 HV *hv = (HV*)POPs;
800e9ae0 1260 HV *keys;
463ee0b2 1261 register HE *entry;
463ee0b2 1262 SV *tmpstr;
54310121 1263 I32 gimme = GIMME_V;
533c011a
NIS
1264 I32 dokeys = (PL_op->op_type == OP_KEYS);
1265 I32 dovalues = (PL_op->op_type == OP_VALUES);
c750a3ec 1266 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
8e84507e
NIS
1267
1268 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1269 dokeys = dovalues = TRUE;
463ee0b2 1270
85581909 1271 if (!hv) {
78f9721b 1272 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
85581909
SB
1273 dTARGET; /* make sure to clear its target here */
1274 if (SvTYPE(TARG) == SVt_PVLV)
1275 LvTARG(TARG) = Nullsv;
1276 PUSHs(TARG);
1277 }
463ee0b2 1278 RETURN;
85581909 1279 }
748a9306 1280
800e9ae0
JP
1281 keys = realhv ? hv : avhv_keys((AV*)hv);
1282 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1283
54310121 1284 if (gimme == G_VOID)
aa689395 1285 RETURN;
1286
54310121 1287 if (gimme == G_SCALAR) {
6ee623d5 1288 IV i;
463ee0b2
LW
1289 dTARGET;
1290
78f9721b 1291 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
85581909
SB
1292 if (SvTYPE(TARG) < SVt_PVLV) {
1293 sv_upgrade(TARG, SVt_PVLV);
1294 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1295 }
1296 LvTYPE(TARG) = 'k';
800e9ae0 1297 if (LvTARG(TARG) != (SV*)keys) {
6ff81951
GS
1298 if (LvTARG(TARG))
1299 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1300 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1301 }
85581909
SB
1302 PUSHs(TARG);
1303 RETURN;
1304 }
1305
33c27489 1306 if (! SvTIED_mg((SV*)keys, 'P'))
800e9ae0 1307 i = HvKEYS(keys);
463ee0b2
LW
1308 else {
1309 i = 0;
463ee0b2 1310 /*SUPPRESS 560*/
800e9ae0 1311 while (hv_iternext(keys)) i++;
463ee0b2
LW
1312 }
1313 PUSHi( i );
1314 RETURN;
1315 }
1316
8ed4b672 1317 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1318
463ee0b2 1319 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1320 while ((entry = hv_iternext(keys))) {
463ee0b2 1321 SPAGAIN;
6bfc225d 1322 if (dokeys)
8c2cee6f 1323 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
463ee0b2 1324 if (dovalues) {
463ee0b2 1325 PUTBACK;
b6429b1b
GS
1326 tmpstr = realhv ?
1327 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
cea2e8a9 1328 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1329 (unsigned long)HeHASH(entry),
800e9ae0
JP
1330 HvMAX(keys)+1,
1331 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1332 SPAGAIN;
46fc3d4c 1333 XPUSHs(tmpstr);
463ee0b2
LW
1334 }
1335 PUTBACK;
1336 }
1337 return NORMAL;
1338}
4e35701f 1339