Couple of tests from #7660 salvaged.
[p5sagit/p5-mst-13.2.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
3818b22b 3 * Copyright (c) 1991-2000, Larry Wall
79072805 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 8 */
9
10/*
11 * "'So that was the job I felt I had to do when I started,' thought Sam."
79072805 12 */
13
14#include "EXTERN.h"
864dbfa3 15#define PERL_IN_DOOP_C
79072805 16#include "perl.h"
17
64ca3a65 18#ifndef PERL_MICRO
79072805 19#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
20#include <signal.h>
21#endif
64ca3a65 22#endif
79072805 23
075321c3 24#define HALF_UTF8_UPGRADE(start,end) \
01ec43d0 25 STMT_START { \
96c7109f 26 if ((start)<(end)) { \
01ec43d0 27 U8* NeWsTr; \
48b1b78b 28 STRLEN LeN = (end) - (start); \
01ec43d0 29 NeWsTr = bytes_to_utf8(start, &LeN); \
075321c3 30 Safefree(start); \
31 (start) = NeWsTr; \
32 (end) = (start) + LeN; \
96c7109f 33 } \
01ec43d0 34 } STMT_END
b250498f 35
942e002e 36STATIC I32
b250498f 37S_do_trans_simple(pTHX_ SV *sv)
79072805 38{
11343788 39 dTHR;
4757a243 40 U8 *s;
b250498f 41 U8 *d;
4757a243 42 U8 *send;
b250498f 43 U8 *dstart;
4757a243 44 I32 matches = 0;
b250498f 45 I32 sutf = SvUTF8(sv);
463ee0b2 46 STRLEN len;
4757a243 47 short *tbl;
48 I32 ch;
79072805 49
4757a243 50 tbl = (short*)cPVOP->op_pv;
51 if (!tbl)
cea2e8a9 52 Perl_croak(aTHX_ "panic: do_trans");
a0ed51b3 53
4757a243 54 s = (U8*)SvPV(sv, len);
55 send = s + len;
56
b250498f 57 /* First, take care of non-UTF8 input strings, because they're easy */
58 if (!sutf) {
01ec43d0 59 while (s < send) {
036b4402 60 if ((ch = tbl[*s]) >= 0) {
01ec43d0 61 matches++;
62 *s++ = ch;
63 }
64 else
65 s++;
66 }
67 SvSETMAGIC(sv);
b250498f 68 return matches;
69 }
4757a243 70
b250498f 71 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
72 Newz(0, d, len*2+1, U8);
73 dstart = d;
74 while (s < send) {
ba210ebe 75 STRLEN ulen;
b250498f 76 short c;
77
78 ulen = 1;
79 /* Need to check this, otherwise 128..255 won't match */
dcad2880 80 c = utf8_to_uv(s, send - s, &ulen, 0);
b250498f 81 if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
82 matches++;
8e84507e 83 if (ch < 0x80)
b250498f 84 *d++ = ch;
8e84507e 85 else
b250498f 86 d = uv_to_utf8(d,ch);
87 s += ulen;
01ec43d0 88 }
89 else { /* No match -> copy */
b250498f 90 while (ulen--)
91 *d++ = *s++;
92 }
93 }
01ec43d0 94 *d = '\0';
be3174d2 95 sv_setpvn(sv, (const char*)dstart, d - dstart);
fdbb8cbd 96 Safefree(dstart);
b250498f 97 SvUTF8_on(sv);
b250498f 98 SvSETMAGIC(sv);
4757a243 99 return matches;
100}
101
942e002e 102STATIC I32
036b4402 103S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
4757a243 104{
105 dTHR;
106 U8 *s;
107 U8 *send;
108 I32 matches = 0;
036b4402 109 I32 hasutf = SvUTF8(sv);
4757a243 110 STRLEN len;
111 short *tbl;
112
113 tbl = (short*)cPVOP->op_pv;
114 if (!tbl)
cea2e8a9 115 Perl_croak(aTHX_ "panic: do_trans");
4757a243 116
117 s = (U8*)SvPV(sv, len);
118 send = s + len;
119
120 while (s < send) {
036b4402 121 if (hasutf && *s & 0x80)
01ec43d0 122 s += UTF8SKIP(s);
036b4402 123 else {
b250498f 124 UV c;
ba210ebe 125 STRLEN ulen;
b250498f 126 ulen = 1;
127 if (hasutf)
dcad2880 128 c = utf8_to_uv(s, send - s, &ulen, 0);
b250498f 129 else
130 c = *s;
131 if (c < 0x100 && tbl[c] >= 0)
036b4402 132 matches++;
01ec43d0 133 s += ulen;
036b4402 134 }
4757a243 135 }
136
137 return matches;
138}
139
942e002e 140STATIC I32
b250498f 141S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
4757a243 142{
143 dTHR;
144 U8 *s;
145 U8 *send;
146 U8 *d;
036b4402 147 I32 hasutf = SvUTF8(sv);
4757a243 148 I32 matches = 0;
149 STRLEN len;
150 short *tbl;
151 I32 ch;
152
153 tbl = (short*)cPVOP->op_pv;
154 if (!tbl)
cea2e8a9 155 Perl_croak(aTHX_ "panic: do_trans");
4757a243 156
157 s = (U8*)SvPV(sv, len);
158 send = s + len;
159
160 d = s;
161 if (PL_op->op_private & OPpTRANS_SQUASH) {
162 U8* p = send;
163
164 while (s < send) {
036b4402 165 if (hasutf && *s & 0x80)
01ec43d0 166 s += UTF8SKIP(s);
036b4402 167 else {
168 if ((ch = tbl[*s]) >= 0) {
169 *d = ch;
170 matches++;
171 if (p == d - 1 && *p == *d)
172 matches--;
173 else
174 p = d++;
175 }
01ec43d0 176 else if (ch == -1) /* -1 is unmapped character */
036b4402 177 *d++ = *s; /* -2 is delete character */
178 s++;
179 }
a0ed51b3 180 }
4757a243 181 }
182 else {
183 while (s < send) {
036b4402 184 if (hasutf && *s & 0x80)
01ec43d0 185 s += UTF8SKIP(s);
036b4402 186 else {
187 if ((ch = tbl[*s]) >= 0) {
188 *d = ch;
189 matches++;
190 d++;
191 }
01ec43d0 192 else if (ch == -1) /* -1 is unmapped character */
036b4402 193 *d++ = *s; /* -2 is delete character */
194 s++;
195 }
5d06d08e 196 }
4757a243 197 }
01ec43d0 198 matches += send - d; /* account for disappeared chars */
4757a243 199 *d = '\0';
200 SvCUR_set(sv, d - (U8*)SvPVX(sv));
201 SvSETMAGIC(sv);
202
203 return matches;
204}
205
942e002e 206STATIC I32
036b4402 207S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
4757a243 208{
209 dTHR;
210 U8 *s;
211 U8 *send;
212 U8 *d;
036b4402 213 U8 *start;
214 U8 *dstart;
4757a243 215 I32 matches = 0;
216 STRLEN len;
217
218 SV* rv = (SV*)cSVOP->op_sv;
219 HV* hv = (HV*)SvRV(rv);
220 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
221 UV none = svp ? SvUV(*svp) : 0x7fffffff;
222 UV extra = none + 1;
223 UV final;
224 UV uv;
8e84507e 225 I32 isutf;
036b4402 226 I32 howmany;
4757a243 227
036b4402 228 isutf = SvUTF8(sv);
4757a243 229 s = (U8*)SvPV(sv, len);
230 send = s + len;
036b4402 231 start = s;
4757a243 232
233 svp = hv_fetch(hv, "FINAL", 5, FALSE);
234 if (svp)
235 final = SvUV(*svp);
236
036b4402 237 /* d needs to be bigger than s, in case e.g. upgrading is required */
238 Newz(0, d, len*2+1, U8);
239 dstart = d;
4757a243 240 while (s < send) {
241 if ((uv = swash_fetch(rv, s)) < none) {
242 s += UTF8SKIP(s);
243 matches++;
01ec43d0 244 if ((uv & 0x80) && !isutf++)
075321c3 245 HALF_UTF8_UPGRADE(dstart,d);
4757a243 246 d = uv_to_utf8(d, uv);
247 }
248 else if (uv == none) {
249 int i;
01ec43d0 250 i = UTF8SKIP(s);
b250498f 251 if (i > 1 && !isutf++)
075321c3 252 HALF_UTF8_UPGRADE(dstart,d);
036b4402 253 while(i--)
075321c3 254 *d++ = *s++;
4757a243 255 }
256 else if (uv == extra) {
036b4402 257 int i;
01ec43d0 258 i = UTF8SKIP(s);
036b4402 259 s += i;
4757a243 260 matches++;
8e84507e 261 if (i > 1 && !isutf++)
075321c3 262 HALF_UTF8_UPGRADE(dstart,d);
4757a243 263 d = uv_to_utf8(d, final);
264 }
265 else
266 s += UTF8SKIP(s);
267 }
268 *d = '\0';
be3174d2 269 sv_setpvn(sv, (const char*)dstart, d - dstart);
4757a243 270 SvSETMAGIC(sv);
036b4402 271 if (isutf)
272 SvUTF8_on(sv);
4757a243 273
274 return matches;
275}
276
942e002e 277STATIC I32
036b4402 278S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
4757a243 279{
280 dTHR;
281 U8 *s;
282 U8 *send;
283 I32 matches = 0;
284 STRLEN len;
285
286 SV* rv = (SV*)cSVOP->op_sv;
287 HV* hv = (HV*)SvRV(rv);
288 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
289 UV none = svp ? SvUV(*svp) : 0x7fffffff;
290 UV uv;
291
292 s = (U8*)SvPV(sv, len);
036b4402 293 if (!SvUTF8(sv))
294 s = bytes_to_utf8(s, &len);
4757a243 295 send = s + len;
296
297 while (s < send) {
834a4ddd 298 if ((uv = swash_fetch(rv, s)) < none)
4757a243 299 matches++;
834a4ddd 300 s += UTF8SKIP(s);
4757a243 301 }
302
303 return matches;
304}
305
942e002e 306STATIC I32
036b4402 307S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
4757a243 308{
309 dTHR;
310 U8 *s;
311 U8 *send;
312 U8 *d;
313 I32 matches = 0;
314 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
4757a243 315 I32 del = PL_op->op_private & OPpTRANS_DELETE;
316 SV* rv = (SV*)cSVOP->op_sv;
317 HV* hv = (HV*)SvRV(rv);
318 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
319 UV none = svp ? SvUV(*svp) : 0x7fffffff;
320 UV extra = none + 1;
321 UV final;
322 UV uv;
323 STRLEN len;
324 U8 *dst;
b250498f 325 I32 isutf = SvUTF8(sv);
4757a243 326
327 s = (U8*)SvPV(sv, len);
328 send = s + len;
329
330 svp = hv_fetch(hv, "FINAL", 5, FALSE);
331 if (svp)
332 final = SvUV(*svp);
333
b250498f 334 Newz(0, d, len*2+1, U8);
4757a243 335 dst = d;
4757a243 336
337 if (squash) {
338 UV puv = 0xfeedface;
339 while (s < send) {
8e84507e 340 if (SvUTF8(sv))
4757a243 341 uv = swash_fetch(rv, s);
a0ed51b3 342 else {
4757a243 343 U8 tmpbuf[2];
344 uv = *s++;
345 if (uv < 0x80)
346 tmpbuf[0] = uv;
347 else {
348 tmpbuf[0] = (( uv >> 6) | 0xc0);
349 tmpbuf[1] = (( uv & 0x3f) | 0x80);
350 }
351 uv = swash_fetch(rv, tmpbuf);
352 }
b250498f 353
4757a243 354 if (uv < none) {
355 matches++;
356 if (uv != puv) {
8e84507e 357 if ((uv & 0x80) && !isutf++)
075321c3 358 HALF_UTF8_UPGRADE(dst,d);
01ec43d0 359 d = uv_to_utf8(d, uv);
4757a243 360 puv = uv;
361 }
075321c3 362 s += UTF8SKIP(s);
4757a243 363 continue;
364 }
365 else if (uv == none) { /* "none" is unmapped character */
ba210ebe 366 STRLEN ulen;
dcad2880 367 *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
01ec43d0 368 s += ulen;
4757a243 369 puv = 0xfeedface;
370 continue;
a0ed51b3 371 }
4757a243 372 else if (uv == extra && !del) {
373 matches++;
374 if (uv != puv) {
01ec43d0 375 d = uv_to_utf8(d, final);
4757a243 376 puv = final;
377 }
01ec43d0 378 s += UTF8SKIP(s);
4757a243 379 continue;
380 }
01ec43d0 381 matches++; /* "none+1" is delete character */
382 s += UTF8SKIP(s);
a0ed51b3 383 }
79072805 384 }
385 else {
4757a243 386 while (s < send) {
8e84507e 387 if (SvUTF8(sv))
4757a243 388 uv = swash_fetch(rv, s);
4757a243 389 else {
390 U8 tmpbuf[2];
391 uv = *s++;
392 if (uv < 0x80)
393 tmpbuf[0] = uv;
394 else {
395 tmpbuf[0] = (( uv >> 6) | 0xc0);
396 tmpbuf[1] = (( uv & 0x3f) | 0x80);
a0ed51b3 397 }
4757a243 398 uv = swash_fetch(rv, tmpbuf);
a0ed51b3 399 }
4757a243 400 if (uv < none) {
401 matches++;
01ec43d0 402 d = uv_to_utf8(d, uv);
403 s += UTF8SKIP(s);
4757a243 404 continue;
a0ed51b3 405 }
4757a243 406 else if (uv == none) { /* "none" is unmapped character */
ba210ebe 407 STRLEN ulen;
dcad2880 408 *d++ = (U8)utf8_to_uv(s, send - s, &ulen, 0);
01ec43d0 409 s += ulen;
4757a243 410 continue;
79072805 411 }
4757a243 412 else if (uv == extra && !del) {
413 matches++;
01ec43d0 414 d = uv_to_utf8(d, final);
415 s += UTF8SKIP(s);
4757a243 416 continue;
417 }
01ec43d0 418 matches++; /* "none+1" is delete character */
419 s += UTF8SKIP(s);
79072805 420 }
4757a243 421 }
422 if (dst)
423 sv_usepvn(sv, (char*)dst, d - dst);
424 else {
425 *d = '\0';
426 SvCUR_set(sv, d - (U8*)SvPVX(sv));
427 }
428 SvSETMAGIC(sv);
429
430 return matches;
431}
432
433I32
864dbfa3 434Perl_do_trans(pTHX_ SV *sv)
4757a243 435{
46124e9e 436 dTHR;
4757a243 437 STRLEN len;
8e84507e 438 I32 hasutf = (PL_op->op_private &
036b4402 439 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
4757a243 440
441 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
cea2e8a9 442 Perl_croak(aTHX_ PL_no_modify);
4757a243 443
444 (void)SvPV(sv, len);
445 if (!len)
446 return 0;
447 if (!SvPOKp(sv))
448 (void)SvPV_force(sv, len);
2de7b02f 449 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
450 (void)SvPOK_only_UTF8(sv);
4757a243 451
cea2e8a9 452 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 453
036b4402 454 switch (PL_op->op_private & ~hasutf & 63) {
4757a243 455 case 0:
01ec43d0 456 if (hasutf)
457 return do_trans_simple_utf8(sv);
458 else
459 return do_trans_simple(sv);
4757a243 460
461 case OPpTRANS_IDENTICAL:
01ec43d0 462 if (hasutf)
463 return do_trans_count_utf8(sv);
464 else
465 return do_trans_count(sv);
4757a243 466
467 default:
01ec43d0 468 if (hasutf)
036b4402 469 return do_trans_complex_utf8(sv);
4757a243 470 else
036b4402 471 return do_trans_complex(sv);
79072805 472 }
79072805 473}
474
475void
864dbfa3 476Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
79072805 477{
478 SV **oldmark = mark;
479 register I32 items = sp - mark;
79072805 480 register STRLEN len;
463ee0b2 481 STRLEN delimlen;
482 register char *delim = SvPV(del, delimlen);
483 STRLEN tmplen;
79072805 484
485 mark++;
486 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
07f14f54 487 (void)SvUPGRADE(sv, SVt_PV);
79072805 488 if (SvLEN(sv) < len + items) { /* current length is way too short */
489 while (items-- > 0) {
1426bbf4 490 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
463ee0b2 491 SvPV(*mark, tmplen);
492 len += tmplen;
79072805 493 }
494 mark++;
495 }
496 SvGROW(sv, len + 1); /* so try to pre-extend */
497
498 mark = oldmark;
db7c17d7 499 items = sp - mark;
79072805 500 ++mark;
501 }
502
463ee0b2 503 if (items-- > 0) {
8990e307 504 char *s;
505
92d29cee 506 sv_setpv(sv, "");
507 if (*mark)
508 sv_catsv(sv, *mark);
463ee0b2 509 mark++;
510 }
79072805 511 else
512 sv_setpv(sv,"");
513 len = delimlen;
514 if (len) {
515 for (; items > 0; items--,mark++) {
516 sv_catpvn(sv,delim,len);
517 sv_catsv(sv,*mark);
518 }
519 }
520 else {
521 for (; items > 0; items--,mark++)
522 sv_catsv(sv,*mark);
523 }
524 SvSETMAGIC(sv);
525}
526
527void
864dbfa3 528Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 529{
46fc3d4c 530 STRLEN patlen;
531 char *pat = SvPV(*sarg, patlen);
532 bool do_taint = FALSE;
533
534 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
79072805 535 SvSETMAGIC(sv);
46fc3d4c 536 if (do_taint)
537 SvTAINTED_on(sv);
79072805 538}
539
33b45480 540/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0 541UV
542Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
543{
544 STRLEN srclen, len;
545 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
546 UV retnum = 0;
547
a50d7633 548 if (offset < 0)
81e118e0 549 return retnum;
8e84507e 550 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 551 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 552
dcad2880 553 if (SvUTF8(sv))
33b45480 554 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 555
81e118e0 556 offset *= size; /* turn into bit offset */
557 len = (offset + size + 7) / 8; /* required number of bytes */
558 if (len > srclen) {
559 if (size <= 8)
560 retnum = 0;
561 else {
562 offset >>= 3; /* turn into byte offset */
563 if (size == 16) {
564 if (offset >= srclen)
565 retnum = 0;
566 else
628e1a40 567 retnum = (UV) s[offset] << 8;
81e118e0 568 }
569 else if (size == 32) {
570 if (offset >= srclen)
571 retnum = 0;
572 else if (offset + 1 >= srclen)
573 retnum =
574 ((UV) s[offset ] << 24);
575 else if (offset + 2 >= srclen)
576 retnum =
577 ((UV) s[offset ] << 24) +
578 ((UV) s[offset + 1] << 16);
579 else
580 retnum =
581 ((UV) s[offset ] << 24) +
582 ((UV) s[offset + 1] << 16) +
583 ( s[offset + 2] << 8);
584 }
d7d93a81 585#ifdef UV_IS_QUAD
c5a0f51a 586 else if (size == 64) {
587 dTHR;
588 if (ckWARN(WARN_PORTABLE))
589 Perl_warner(aTHX_ WARN_PORTABLE,
590 "Bit vector size > 32 non-portable");
591 if (offset >= srclen)
592 retnum = 0;
593 else if (offset + 1 >= srclen)
594 retnum =
595 (UV) s[offset ] << 56;
596 else if (offset + 2 >= srclen)
597 retnum =
598 ((UV) s[offset ] << 56) +
599 ((UV) s[offset + 1] << 48);
600 else if (offset + 3 >= srclen)
601 retnum =
602 ((UV) s[offset ] << 56) +
603 ((UV) s[offset + 1] << 48) +
604 ((UV) s[offset + 2] << 40);
605 else if (offset + 4 >= srclen)
606 retnum =
607 ((UV) s[offset ] << 56) +
608 ((UV) s[offset + 1] << 48) +
609 ((UV) s[offset + 2] << 40) +
610 ((UV) s[offset + 3] << 32);
611 else if (offset + 5 >= srclen)
612 retnum =
613 ((UV) s[offset ] << 56) +
614 ((UV) s[offset + 1] << 48) +
615 ((UV) s[offset + 2] << 40) +
616 ((UV) s[offset + 3] << 32) +
617 ( s[offset + 4] << 24);
618 else if (offset + 6 >= srclen)
619 retnum =
620 ((UV) s[offset ] << 56) +
621 ((UV) s[offset + 1] << 48) +
622 ((UV) s[offset + 2] << 40) +
623 ((UV) s[offset + 3] << 32) +
624 ((UV) s[offset + 4] << 24) +
625 ((UV) s[offset + 5] << 16);
626 else
8e84507e 627 retnum =
c5a0f51a 628 ((UV) s[offset ] << 56) +
629 ((UV) s[offset + 1] << 48) +
630 ((UV) s[offset + 2] << 40) +
631 ((UV) s[offset + 3] << 32) +
632 ((UV) s[offset + 4] << 24) +
633 ((UV) s[offset + 5] << 16) +
628e1a40 634 ( s[offset + 6] << 8);
c5a0f51a 635 }
636#endif
81e118e0 637 }
638 }
639 else if (size < 8)
640 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
641 else {
642 offset >>= 3; /* turn into byte offset */
643 if (size == 8)
644 retnum = s[offset];
645 else if (size == 16)
646 retnum =
628e1a40 647 ((UV) s[offset] << 8) +
81e118e0 648 s[offset + 1];
649 else if (size == 32)
650 retnum =
651 ((UV) s[offset ] << 24) +
652 ((UV) s[offset + 1] << 16) +
653 ( s[offset + 2] << 8) +
654 s[offset + 3];
d7d93a81 655#ifdef UV_IS_QUAD
c5a0f51a 656 else if (size == 64) {
657 dTHR;
658 if (ckWARN(WARN_PORTABLE))
659 Perl_warner(aTHX_ WARN_PORTABLE,
660 "Bit vector size > 32 non-portable");
661 retnum =
662 ((UV) s[offset ] << 56) +
663 ((UV) s[offset + 1] << 48) +
664 ((UV) s[offset + 2] << 40) +
665 ((UV) s[offset + 3] << 32) +
666 ((UV) s[offset + 4] << 24) +
667 ((UV) s[offset + 5] << 16) +
628e1a40 668 ( s[offset + 6] << 8) +
c5a0f51a 669 s[offset + 7];
670 }
671#endif
81e118e0 672 }
673
674 return retnum;
675}
676
33b45480 677/* currently converts input to bytes if possible but doesn't sweat failures,
678 * although it does ensure that the string it clobbers is not marked as
679 * utf8-valid any more
680 */
79072805 681void
864dbfa3 682Perl_do_vecset(pTHX_ SV *sv)
79072805 683{
684 SV *targ = LvTARG(sv);
685 register I32 offset;
686 register I32 size;
8990e307 687 register unsigned char *s;
81e118e0 688 register UV lval;
79072805 689 I32 mask;
a0d0e21e 690 STRLEN targlen;
691 STRLEN len;
79072805 692
8990e307 693 if (!targ)
694 return;
a0d0e21e 695 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 696 if (SvUTF8(targ)) {
33b45480 697 /* This is handled by the SvPOK_only below...
698 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
699 SvUTF8_off(targ);
700 */
701 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53 702 }
703
4ebbc975 704 (void)SvPOK_only(targ);
81e118e0 705 lval = SvUV(sv);
79072805 706 offset = LvTARGOFF(sv);
fe58ced6 707 if (offset < 0)
708 Perl_croak(aTHX_ "Assigning to negative offset in vec");
79072805 709 size = LvTARGLEN(sv);
8e84507e 710 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 711 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 712
81e118e0 713 offset *= size; /* turn into bit offset */
714 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e 715 if (len > targlen) {
716 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 717 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e 718 SvCUR_set(targ, len);
719 }
8e84507e 720
79072805 721 if (size < 8) {
722 mask = (1 << size) - 1;
723 size = offset & 7;
724 lval &= mask;
81e118e0 725 offset >>= 3; /* turn into byte offset */
79072805 726 s[offset] &= ~(mask << size);
727 s[offset] |= lval << size;
728 }
729 else {
81e118e0 730 offset >>= 3; /* turn into byte offset */
79072805 731 if (size == 8)
c5a0f51a 732 s[offset ] = lval & 0xff;
79072805 733 else if (size == 16) {
c5a0f51a 734 s[offset ] = (lval >> 8) & 0xff;
735 s[offset+1] = lval & 0xff;
79072805 736 }
737 else if (size == 32) {
c5a0f51a 738 s[offset ] = (lval >> 24) & 0xff;
739 s[offset+1] = (lval >> 16) & 0xff;
740 s[offset+2] = (lval >> 8) & 0xff;
741 s[offset+3] = lval & 0xff;
742 }
d7d93a81 743#ifdef UV_IS_QUAD
c5a0f51a 744 else if (size == 64) {
745 dTHR;
746 if (ckWARN(WARN_PORTABLE))
747 Perl_warner(aTHX_ WARN_PORTABLE,
748 "Bit vector size > 32 non-portable");
749 s[offset ] = (lval >> 56) & 0xff;
750 s[offset+1] = (lval >> 48) & 0xff;
751 s[offset+2] = (lval >> 40) & 0xff;
752 s[offset+3] = (lval >> 32) & 0xff;
753 s[offset+4] = (lval >> 24) & 0xff;
754 s[offset+5] = (lval >> 16) & 0xff;
755 s[offset+6] = (lval >> 8) & 0xff;
756 s[offset+7] = lval & 0xff;
79072805 757 }
dc1e3f56 758#endif
79072805 759 }
7bb043c3 760 SvSETMAGIC(targ);
79072805 761}
762
763void
864dbfa3 764Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 765{
463ee0b2 766 STRLEN len;
a0d0e21e 767 char *s;
c485e607 768 dTHR;
8e84507e 769
79072805 770 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e 771 register I32 i;
772 I32 max;
773 AV* av = (AV*)sv;
774 max = AvFILL(av);
775 for (i = 0; i <= max; i++) {
776 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 777 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 778 do_chop(astr, sv);
779 }
780 return;
79072805 781 }
aa854799 782 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 783 HV* hv = (HV*)sv;
784 HE* entry;
785 (void)hv_iterinit(hv);
786 /*SUPPRESS 560*/
155aba94 787 while ((entry = hv_iternext(hv)))
a0d0e21e 788 do_chop(astr,hv_iterval(hv,entry));
789 return;
79072805 790 }
aa854799 791 else if (SvREADONLY(sv))
cea2e8a9 792 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 793 s = SvPV(sv, len);
748a9306 794 if (len && !SvPOK(sv))
a0d0e21e 795 s = SvPV_force(sv, len);
7e2040f0 796 if (DO_UTF8(sv)) {
a0ed51b3 797 if (s && len) {
798 char *send = s + len;
799 char *start = s;
800 s = send - 1;
801 while ((*s & 0xc0) == 0x80)
802 --s;
0453d815 803 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
804 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
a0ed51b3 805 sv_setpvn(astr, s, send - s);
806 *s = '\0';
807 SvCUR_set(sv, s - start);
808 SvNIOK_off(sv);
7e2040f0 809 SvUTF8_on(astr);
a0ed51b3 810 }
811 else
812 sv_setpvn(astr, "", 0);
813 }
7e2040f0 814 else if (s && len) {
a0d0e21e 815 s += --len;
816 sv_setpvn(astr, s, 1);
817 *s = '\0';
818 SvCUR_set(sv, len);
2c19a612 819 SvUTF8_off(sv);
a0d0e21e 820 SvNIOK_off(sv);
79072805 821 }
822 else
a0d0e21e 823 sv_setpvn(astr, "", 0);
824 SvSETMAGIC(sv);
7e2040f0 825}
a0d0e21e 826
827I32
864dbfa3 828Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 829{
aeea060c 830 dTHR;
c07a80fd 831 register I32 count;
a0d0e21e 832 STRLEN len;
833 char *s;
c07a80fd 834
3280af22 835 if (RsSNARF(PL_rs))
c07a80fd 836 return 0;
4c5a6083 837 if (RsRECORD(PL_rs))
838 return 0;
c07a80fd 839 count = 0;
a0d0e21e 840 if (SvTYPE(sv) == SVt_PVAV) {
841 register I32 i;
842 I32 max;
843 AV* av = (AV*)sv;
844 max = AvFILL(av);
845 for (i = 0; i <= max; i++) {
846 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 847 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 848 count += do_chomp(sv);
849 }
850 return count;
851 }
aa854799 852 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 853 HV* hv = (HV*)sv;
854 HE* entry;
855 (void)hv_iterinit(hv);
856 /*SUPPRESS 560*/
155aba94 857 while ((entry = hv_iternext(hv)))
a0d0e21e 858 count += do_chomp(hv_iterval(hv,entry));
859 return count;
860 }
aa854799 861 else if (SvREADONLY(sv))
cea2e8a9 862 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 863 s = SvPV(sv, len);
864 if (len && !SvPOKp(sv))
865 s = SvPV_force(sv, len);
866 if (s && len) {
867 s += --len;
3280af22 868 if (RsPARA(PL_rs)) {
a0d0e21e 869 if (*s != '\n')
870 goto nope;
871 ++count;
872 while (len && s[-1] == '\n') {
873 --len;
874 --s;
875 ++count;
876 }
877 }
a0d0e21e 878 else {
c07a80fd 879 STRLEN rslen;
3280af22 880 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 881 if (rslen == 1) {
882 if (*s != *rsptr)
883 goto nope;
884 ++count;
885 }
886 else {
8c2cee6f 887 if (len < rslen - 1)
c07a80fd 888 goto nope;
889 len -= rslen - 1;
890 s -= rslen - 1;
36477c24 891 if (memNE(s, rsptr, rslen))
c07a80fd 892 goto nope;
893 count += rslen;
894 }
a0d0e21e 895 }
a0d0e21e 896 *s = '\0';
897 SvCUR_set(sv, len);
898 SvNIOK_off(sv);
899 }
900 nope:
901 SvSETMAGIC(sv);
902 return count;
8e84507e 903}
79072805 904
905void
864dbfa3 906Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805 907{
aeea060c 908 dTHR; /* just for taint */
79072805 909#ifdef LIBERAL
910 register long *dl;
911 register long *ll;
912 register long *rl;
913#endif
914 register char *dc;
463ee0b2 915 STRLEN leftlen;
916 STRLEN rightlen;
7a4c00b4 917 register char *lc;
918 register char *rc;
79072805 919 register I32 len;
a0d0e21e 920 I32 lensave;
7a4c00b4 921 char *lsave;
922 char *rsave;
0c57e439 923 bool left_utf = DO_UTF8(left);
924 bool right_utf = DO_UTF8(right);
c9b3c8d0 925 I32 needlen;
0c57e439 926
927 if (left_utf && !right_utf)
928 sv_utf8_upgrade(right);
929 if (!left_utf && right_utf)
930 sv_utf8_upgrade(left);
79072805 931
1fbd88dc 932 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
933 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 934 lsave = lc = SvPV(left, leftlen);
935 rsave = rc = SvPV(right, rightlen);
93a17b20 936 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 937 lensave = len;
c9b3c8d0 938 if ((left_utf || right_utf) && (sv == left || sv == right)) {
939 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
940 Newz(801, dc, needlen + 1, char);
941 }
942 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d 943 STRLEN n_a;
944 dc = SvPV_force(sv, n_a);
ff68c719 945 if (SvCUR(sv) < len) {
946 dc = SvGROW(sv, len + 1);
947 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
948 }
c9b3c8d0 949 if (optype != OP_BIT_AND && (left_utf || right_utf))
950 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 951 }
952 else {
c9b3c8d0 953 needlen = ((optype == OP_BIT_AND)
954 ? len : (leftlen > rightlen ? leftlen : rightlen));
ff68c719 955 Newz(801, dc, needlen + 1, char);
956 (void)sv_usepvn(sv, dc, needlen);
957 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 958 }
a0d0e21e 959 SvCUR_set(sv, len);
960 (void)SvPOK_only(sv);
0c57e439 961 if (left_utf || right_utf) {
962 UV duc, luc, ruc;
c9b3c8d0 963 char *dcsave = dc;
0c57e439 964 STRLEN lulen = leftlen;
965 STRLEN rulen = rightlen;
ba210ebe 966 STRLEN ulen;
0c57e439 967
968 switch (optype) {
969 case OP_BIT_AND:
970 while (lulen && rulen) {
287eef1b 971 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 972 lc += ulen;
973 lulen -= ulen;
287eef1b 974 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 975 rc += ulen;
976 rulen -= ulen;
977 duc = luc & ruc;
978 dc = (char*)uv_to_utf8((U8*)dc, duc);
979 }
c9b3c8d0 980 if (sv == left || sv == right)
981 (void)sv_usepvn(sv, dcsave, needlen);
982 SvCUR_set(sv, dc - dcsave);
0c57e439 983 break;
984 case OP_BIT_XOR:
985 while (lulen && rulen) {
287eef1b 986 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 987 lc += ulen;
988 lulen -= ulen;
287eef1b 989 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 990 rc += ulen;
991 rulen -= ulen;
992 duc = luc ^ ruc;
993 dc = (char*)uv_to_utf8((U8*)dc, duc);
994 }
995 goto mop_up_utf;
996 case OP_BIT_OR:
997 while (lulen && rulen) {
287eef1b 998 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 999 lc += ulen;
1000 lulen -= ulen;
287eef1b 1001 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANY);
0c57e439 1002 rc += ulen;
1003 rulen -= ulen;
1004 duc = luc | ruc;
1005 dc = (char*)uv_to_utf8((U8*)dc, duc);
1006 }
1007 mop_up_utf:
c9b3c8d0 1008 if (sv == left || sv == right)
1009 (void)sv_usepvn(sv, dcsave, needlen);
1010 SvCUR_set(sv, dc - dcsave);
0c57e439 1011 if (rulen)
1012 sv_catpvn(sv, rc, rulen);
1013 else if (lulen)
1014 sv_catpvn(sv, lc, lulen);
1015 else
1016 *SvEND(sv) = '\0';
1017 break;
1018 }
1019 SvUTF8_on(sv);
1020 goto finish;
1021 }
1022 else
79072805 1023#ifdef LIBERAL
1024 if (len >= sizeof(long)*4 &&
1025 !((long)dc % sizeof(long)) &&
1026 !((long)lc % sizeof(long)) &&
1027 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1028 {
1029 I32 remainder = len % (sizeof(long)*4);
1030 len /= (sizeof(long)*4);
1031
1032 dl = (long*)dc;
1033 ll = (long*)lc;
1034 rl = (long*)rc;
1035
1036 switch (optype) {
1037 case OP_BIT_AND:
1038 while (len--) {
1039 *dl++ = *ll++ & *rl++;
1040 *dl++ = *ll++ & *rl++;
1041 *dl++ = *ll++ & *rl++;
1042 *dl++ = *ll++ & *rl++;
1043 }
1044 break;
a0d0e21e 1045 case OP_BIT_XOR:
79072805 1046 while (len--) {
1047 *dl++ = *ll++ ^ *rl++;
1048 *dl++ = *ll++ ^ *rl++;
1049 *dl++ = *ll++ ^ *rl++;
1050 *dl++ = *ll++ ^ *rl++;
1051 }
1052 break;
1053 case OP_BIT_OR:
1054 while (len--) {
1055 *dl++ = *ll++ | *rl++;
1056 *dl++ = *ll++ | *rl++;
1057 *dl++ = *ll++ | *rl++;
1058 *dl++ = *ll++ | *rl++;
1059 }
1060 }
1061
1062 dc = (char*)dl;
1063 lc = (char*)ll;
1064 rc = (char*)rl;
1065
1066 len = remainder;
1067 }
1068#endif
a0d0e21e 1069 {
a0d0e21e 1070 switch (optype) {
1071 case OP_BIT_AND:
1072 while (len--)
1073 *dc++ = *lc++ & *rc++;
1074 break;
1075 case OP_BIT_XOR:
1076 while (len--)
1077 *dc++ = *lc++ ^ *rc++;
1078 goto mop_up;
1079 case OP_BIT_OR:
1080 while (len--)
1081 *dc++ = *lc++ | *rc++;
1082 mop_up:
1083 len = lensave;
1084 if (rightlen > len)
1085 sv_catpvn(sv, rsave + len, rightlen - len);
1086 else if (leftlen > len)
1087 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4 1088 else
1089 *SvEND(sv) = '\0';
a0d0e21e 1090 break;
1091 }
79072805 1092 }
0c57e439 1093finish:
fb73857a 1094 SvTAINT(sv);
79072805 1095}
463ee0b2 1096
1097OP *
cea2e8a9 1098Perl_do_kv(pTHX)
463ee0b2 1099{
4e35701f 1100 djSP;
463ee0b2 1101 HV *hv = (HV*)POPs;
800e9ae0 1102 HV *keys;
463ee0b2 1103 register HE *entry;
463ee0b2 1104 SV *tmpstr;
54310121 1105 I32 gimme = GIMME_V;
533c011a 1106 I32 dokeys = (PL_op->op_type == OP_KEYS);
1107 I32 dovalues = (PL_op->op_type == OP_VALUES);
c750a3ec 1108 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
8e84507e 1109
1110 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1111 dokeys = dovalues = TRUE;
463ee0b2 1112
85581909 1113 if (!hv) {
533c011a 1114 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1115 dTARGET; /* make sure to clear its target here */
1116 if (SvTYPE(TARG) == SVt_PVLV)
1117 LvTARG(TARG) = Nullsv;
1118 PUSHs(TARG);
1119 }
463ee0b2 1120 RETURN;
85581909 1121 }
748a9306 1122
800e9ae0 1123 keys = realhv ? hv : avhv_keys((AV*)hv);
1124 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1125
54310121 1126 if (gimme == G_VOID)
aa689395 1127 RETURN;
1128
54310121 1129 if (gimme == G_SCALAR) {
6ee623d5 1130 IV i;
463ee0b2 1131 dTARGET;
1132
533c011a 1133 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1134 if (SvTYPE(TARG) < SVt_PVLV) {
1135 sv_upgrade(TARG, SVt_PVLV);
1136 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1137 }
1138 LvTYPE(TARG) = 'k';
800e9ae0 1139 if (LvTARG(TARG) != (SV*)keys) {
6ff81951 1140 if (LvTARG(TARG))
1141 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1142 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1143 }
85581909 1144 PUSHs(TARG);
1145 RETURN;
1146 }
1147
33c27489 1148 if (! SvTIED_mg((SV*)keys, 'P'))
800e9ae0 1149 i = HvKEYS(keys);
463ee0b2 1150 else {
1151 i = 0;
463ee0b2 1152 /*SUPPRESS 560*/
800e9ae0 1153 while (hv_iternext(keys)) i++;
463ee0b2 1154 }
1155 PUSHi( i );
1156 RETURN;
1157 }
1158
8ed4b672 1159 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1160
463ee0b2 1161 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1162 while ((entry = hv_iternext(keys))) {
463ee0b2 1163 SPAGAIN;
8c2cee6f 1164 if (dokeys)
1165 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
463ee0b2 1166 if (dovalues) {
463ee0b2 1167 PUTBACK;
b6429b1b 1168 tmpstr = realhv ?
1169 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
cea2e8a9 1170 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1171 (unsigned long)HeHASH(entry),
800e9ae0 1172 HvMAX(keys)+1,
1173 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1174 SPAGAIN;
46fc3d4c 1175 XPUSHs(tmpstr);
463ee0b2 1176 }
1177 PUTBACK;
1178 }
1179 return NORMAL;
1180}
4e35701f 1181