Allow poking holes at the UTF-8 decoding strictness.
[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 */
ba210ebe 80 c = utf8_to_uv_chk(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)
ba210ebe 128 c = utf8_to_uv_chk(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;
367 *d++ = (U8)utf8_to_uv_chk(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;
408 *d++ = (U8)utf8_to_uv_chk(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
553 if (SvUTF8(sv)) {
33b45480 554 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 555 }
556
81e118e0 557 offset *= size; /* turn into bit offset */
558 len = (offset + size + 7) / 8; /* required number of bytes */
559 if (len > srclen) {
560 if (size <= 8)
561 retnum = 0;
562 else {
563 offset >>= 3; /* turn into byte offset */
564 if (size == 16) {
565 if (offset >= srclen)
566 retnum = 0;
567 else
628e1a40 568 retnum = (UV) s[offset] << 8;
81e118e0 569 }
570 else if (size == 32) {
571 if (offset >= srclen)
572 retnum = 0;
573 else if (offset + 1 >= srclen)
574 retnum =
575 ((UV) s[offset ] << 24);
576 else if (offset + 2 >= srclen)
577 retnum =
578 ((UV) s[offset ] << 24) +
579 ((UV) s[offset + 1] << 16);
580 else
581 retnum =
582 ((UV) s[offset ] << 24) +
583 ((UV) s[offset + 1] << 16) +
584 ( s[offset + 2] << 8);
585 }
d7d93a81 586#ifdef UV_IS_QUAD
c5a0f51a 587 else if (size == 64) {
588 dTHR;
589 if (ckWARN(WARN_PORTABLE))
590 Perl_warner(aTHX_ WARN_PORTABLE,
591 "Bit vector size > 32 non-portable");
592 if (offset >= srclen)
593 retnum = 0;
594 else if (offset + 1 >= srclen)
595 retnum =
596 (UV) s[offset ] << 56;
597 else if (offset + 2 >= srclen)
598 retnum =
599 ((UV) s[offset ] << 56) +
600 ((UV) s[offset + 1] << 48);
601 else if (offset + 3 >= srclen)
602 retnum =
603 ((UV) s[offset ] << 56) +
604 ((UV) s[offset + 1] << 48) +
605 ((UV) s[offset + 2] << 40);
606 else if (offset + 4 >= srclen)
607 retnum =
608 ((UV) s[offset ] << 56) +
609 ((UV) s[offset + 1] << 48) +
610 ((UV) s[offset + 2] << 40) +
611 ((UV) s[offset + 3] << 32);
612 else if (offset + 5 >= srclen)
613 retnum =
614 ((UV) s[offset ] << 56) +
615 ((UV) s[offset + 1] << 48) +
616 ((UV) s[offset + 2] << 40) +
617 ((UV) s[offset + 3] << 32) +
618 ( s[offset + 4] << 24);
619 else if (offset + 6 >= srclen)
620 retnum =
621 ((UV) s[offset ] << 56) +
622 ((UV) s[offset + 1] << 48) +
623 ((UV) s[offset + 2] << 40) +
624 ((UV) s[offset + 3] << 32) +
625 ((UV) s[offset + 4] << 24) +
626 ((UV) s[offset + 5] << 16);
627 else
8e84507e 628 retnum =
c5a0f51a 629 ((UV) s[offset ] << 56) +
630 ((UV) s[offset + 1] << 48) +
631 ((UV) s[offset + 2] << 40) +
632 ((UV) s[offset + 3] << 32) +
633 ((UV) s[offset + 4] << 24) +
634 ((UV) s[offset + 5] << 16) +
628e1a40 635 ( s[offset + 6] << 8);
c5a0f51a 636 }
637#endif
81e118e0 638 }
639 }
640 else if (size < 8)
641 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
642 else {
643 offset >>= 3; /* turn into byte offset */
644 if (size == 8)
645 retnum = s[offset];
646 else if (size == 16)
647 retnum =
628e1a40 648 ((UV) s[offset] << 8) +
81e118e0 649 s[offset + 1];
650 else if (size == 32)
651 retnum =
652 ((UV) s[offset ] << 24) +
653 ((UV) s[offset + 1] << 16) +
654 ( s[offset + 2] << 8) +
655 s[offset + 3];
d7d93a81 656#ifdef UV_IS_QUAD
c5a0f51a 657 else if (size == 64) {
658 dTHR;
659 if (ckWARN(WARN_PORTABLE))
660 Perl_warner(aTHX_ WARN_PORTABLE,
661 "Bit vector size > 32 non-portable");
662 retnum =
663 ((UV) s[offset ] << 56) +
664 ((UV) s[offset + 1] << 48) +
665 ((UV) s[offset + 2] << 40) +
666 ((UV) s[offset + 3] << 32) +
667 ((UV) s[offset + 4] << 24) +
668 ((UV) s[offset + 5] << 16) +
628e1a40 669 ( s[offset + 6] << 8) +
c5a0f51a 670 s[offset + 7];
671 }
672#endif
81e118e0 673 }
674
675 return retnum;
676}
677
33b45480 678/* currently converts input to bytes if possible but doesn't sweat failures,
679 * although it does ensure that the string it clobbers is not marked as
680 * utf8-valid any more
681 */
79072805 682void
864dbfa3 683Perl_do_vecset(pTHX_ SV *sv)
79072805 684{
685 SV *targ = LvTARG(sv);
686 register I32 offset;
687 register I32 size;
8990e307 688 register unsigned char *s;
81e118e0 689 register UV lval;
79072805 690 I32 mask;
a0d0e21e 691 STRLEN targlen;
692 STRLEN len;
79072805 693
8990e307 694 if (!targ)
695 return;
a0d0e21e 696 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 697 if (SvUTF8(targ)) {
33b45480 698 /* This is handled by the SvPOK_only below...
699 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
700 SvUTF8_off(targ);
701 */
702 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53 703 }
704
4ebbc975 705 (void)SvPOK_only(targ);
81e118e0 706 lval = SvUV(sv);
79072805 707 offset = LvTARGOFF(sv);
fe58ced6 708 if (offset < 0)
709 Perl_croak(aTHX_ "Assigning to negative offset in vec");
79072805 710 size = LvTARGLEN(sv);
8e84507e 711 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 712 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 713
81e118e0 714 offset *= size; /* turn into bit offset */
715 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e 716 if (len > targlen) {
717 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 718 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e 719 SvCUR_set(targ, len);
720 }
8e84507e 721
79072805 722 if (size < 8) {
723 mask = (1 << size) - 1;
724 size = offset & 7;
725 lval &= mask;
81e118e0 726 offset >>= 3; /* turn into byte offset */
79072805 727 s[offset] &= ~(mask << size);
728 s[offset] |= lval << size;
729 }
730 else {
81e118e0 731 offset >>= 3; /* turn into byte offset */
79072805 732 if (size == 8)
c5a0f51a 733 s[offset ] = lval & 0xff;
79072805 734 else if (size == 16) {
c5a0f51a 735 s[offset ] = (lval >> 8) & 0xff;
736 s[offset+1] = lval & 0xff;
79072805 737 }
738 else if (size == 32) {
c5a0f51a 739 s[offset ] = (lval >> 24) & 0xff;
740 s[offset+1] = (lval >> 16) & 0xff;
741 s[offset+2] = (lval >> 8) & 0xff;
742 s[offset+3] = lval & 0xff;
743 }
d7d93a81 744#ifdef UV_IS_QUAD
c5a0f51a 745 else if (size == 64) {
746 dTHR;
747 if (ckWARN(WARN_PORTABLE))
748 Perl_warner(aTHX_ WARN_PORTABLE,
749 "Bit vector size > 32 non-portable");
750 s[offset ] = (lval >> 56) & 0xff;
751 s[offset+1] = (lval >> 48) & 0xff;
752 s[offset+2] = (lval >> 40) & 0xff;
753 s[offset+3] = (lval >> 32) & 0xff;
754 s[offset+4] = (lval >> 24) & 0xff;
755 s[offset+5] = (lval >> 16) & 0xff;
756 s[offset+6] = (lval >> 8) & 0xff;
757 s[offset+7] = lval & 0xff;
79072805 758 }
dc1e3f56 759#endif
79072805 760 }
7bb043c3 761 SvSETMAGIC(targ);
79072805 762}
763
764void
864dbfa3 765Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 766{
463ee0b2 767 STRLEN len;
a0d0e21e 768 char *s;
c485e607 769 dTHR;
8e84507e 770
79072805 771 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e 772 register I32 i;
773 I32 max;
774 AV* av = (AV*)sv;
775 max = AvFILL(av);
776 for (i = 0; i <= max; i++) {
777 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 778 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 779 do_chop(astr, sv);
780 }
781 return;
79072805 782 }
aa854799 783 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 784 HV* hv = (HV*)sv;
785 HE* entry;
786 (void)hv_iterinit(hv);
787 /*SUPPRESS 560*/
155aba94 788 while ((entry = hv_iternext(hv)))
a0d0e21e 789 do_chop(astr,hv_iterval(hv,entry));
790 return;
79072805 791 }
aa854799 792 else if (SvREADONLY(sv))
cea2e8a9 793 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 794 s = SvPV(sv, len);
748a9306 795 if (len && !SvPOK(sv))
a0d0e21e 796 s = SvPV_force(sv, len);
7e2040f0 797 if (DO_UTF8(sv)) {
a0ed51b3 798 if (s && len) {
799 char *send = s + len;
800 char *start = s;
801 s = send - 1;
802 while ((*s & 0xc0) == 0x80)
803 --s;
0453d815 804 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
805 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
a0ed51b3 806 sv_setpvn(astr, s, send - s);
807 *s = '\0';
808 SvCUR_set(sv, s - start);
809 SvNIOK_off(sv);
7e2040f0 810 SvUTF8_on(astr);
a0ed51b3 811 }
812 else
813 sv_setpvn(astr, "", 0);
814 }
7e2040f0 815 else if (s && len) {
a0d0e21e 816 s += --len;
817 sv_setpvn(astr, s, 1);
818 *s = '\0';
819 SvCUR_set(sv, len);
2c19a612 820 SvUTF8_off(sv);
a0d0e21e 821 SvNIOK_off(sv);
79072805 822 }
823 else
a0d0e21e 824 sv_setpvn(astr, "", 0);
825 SvSETMAGIC(sv);
7e2040f0 826}
a0d0e21e 827
828I32
864dbfa3 829Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 830{
aeea060c 831 dTHR;
c07a80fd 832 register I32 count;
a0d0e21e 833 STRLEN len;
834 char *s;
c07a80fd 835
3280af22 836 if (RsSNARF(PL_rs))
c07a80fd 837 return 0;
4c5a6083 838 if (RsRECORD(PL_rs))
839 return 0;
c07a80fd 840 count = 0;
a0d0e21e 841 if (SvTYPE(sv) == SVt_PVAV) {
842 register I32 i;
843 I32 max;
844 AV* av = (AV*)sv;
845 max = AvFILL(av);
846 for (i = 0; i <= max; i++) {
847 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 848 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e 849 count += do_chomp(sv);
850 }
851 return count;
852 }
aa854799 853 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e 854 HV* hv = (HV*)sv;
855 HE* entry;
856 (void)hv_iterinit(hv);
857 /*SUPPRESS 560*/
155aba94 858 while ((entry = hv_iternext(hv)))
a0d0e21e 859 count += do_chomp(hv_iterval(hv,entry));
860 return count;
861 }
aa854799 862 else if (SvREADONLY(sv))
cea2e8a9 863 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 864 s = SvPV(sv, len);
865 if (len && !SvPOKp(sv))
866 s = SvPV_force(sv, len);
867 if (s && len) {
868 s += --len;
3280af22 869 if (RsPARA(PL_rs)) {
a0d0e21e 870 if (*s != '\n')
871 goto nope;
872 ++count;
873 while (len && s[-1] == '\n') {
874 --len;
875 --s;
876 ++count;
877 }
878 }
a0d0e21e 879 else {
c07a80fd 880 STRLEN rslen;
3280af22 881 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 882 if (rslen == 1) {
883 if (*s != *rsptr)
884 goto nope;
885 ++count;
886 }
887 else {
8c2cee6f 888 if (len < rslen - 1)
c07a80fd 889 goto nope;
890 len -= rslen - 1;
891 s -= rslen - 1;
36477c24 892 if (memNE(s, rsptr, rslen))
c07a80fd 893 goto nope;
894 count += rslen;
895 }
a0d0e21e 896 }
a0d0e21e 897 *s = '\0';
898 SvCUR_set(sv, len);
899 SvNIOK_off(sv);
900 }
901 nope:
902 SvSETMAGIC(sv);
903 return count;
8e84507e 904}
79072805 905
906void
864dbfa3 907Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805 908{
aeea060c 909 dTHR; /* just for taint */
79072805 910#ifdef LIBERAL
911 register long *dl;
912 register long *ll;
913 register long *rl;
914#endif
915 register char *dc;
463ee0b2 916 STRLEN leftlen;
917 STRLEN rightlen;
7a4c00b4 918 register char *lc;
919 register char *rc;
79072805 920 register I32 len;
a0d0e21e 921 I32 lensave;
7a4c00b4 922 char *lsave;
923 char *rsave;
0c57e439 924 bool left_utf = DO_UTF8(left);
925 bool right_utf = DO_UTF8(right);
c9b3c8d0 926 I32 needlen;
0c57e439 927
928 if (left_utf && !right_utf)
929 sv_utf8_upgrade(right);
930 if (!left_utf && right_utf)
931 sv_utf8_upgrade(left);
79072805 932
1fbd88dc 933 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
934 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 935 lsave = lc = SvPV(left, leftlen);
936 rsave = rc = SvPV(right, rightlen);
93a17b20 937 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 938 lensave = len;
c9b3c8d0 939 if ((left_utf || right_utf) && (sv == left || sv == right)) {
940 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
941 Newz(801, dc, needlen + 1, char);
942 }
943 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d 944 STRLEN n_a;
945 dc = SvPV_force(sv, n_a);
ff68c719 946 if (SvCUR(sv) < len) {
947 dc = SvGROW(sv, len + 1);
948 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
949 }
c9b3c8d0 950 if (optype != OP_BIT_AND && (left_utf || right_utf))
951 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 952 }
953 else {
c9b3c8d0 954 needlen = ((optype == OP_BIT_AND)
955 ? len : (leftlen > rightlen ? leftlen : rightlen));
ff68c719 956 Newz(801, dc, needlen + 1, char);
957 (void)sv_usepvn(sv, dc, needlen);
958 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 959 }
a0d0e21e 960 SvCUR_set(sv, len);
961 (void)SvPOK_only(sv);
0c57e439 962 if (left_utf || right_utf) {
963 UV duc, luc, ruc;
c9b3c8d0 964 char *dcsave = dc;
0c57e439 965 STRLEN lulen = leftlen;
966 STRLEN rulen = rightlen;
ba210ebe 967 STRLEN ulen;
0c57e439 968
969 switch (optype) {
970 case OP_BIT_AND:
971 while (lulen && rulen) {
ba210ebe 972 luc = utf8_to_uv_chk((U8*)lc, lulen, &ulen, 0);
0c57e439 973 lc += ulen;
974 lulen -= ulen;
ba210ebe 975 ruc = utf8_to_uv_chk((U8*)rc, rulen, &ulen, 0);
0c57e439 976 rc += ulen;
977 rulen -= ulen;
978 duc = luc & ruc;
979 dc = (char*)uv_to_utf8((U8*)dc, duc);
980 }
c9b3c8d0 981 if (sv == left || sv == right)
982 (void)sv_usepvn(sv, dcsave, needlen);
983 SvCUR_set(sv, dc - dcsave);
0c57e439 984 break;
985 case OP_BIT_XOR:
986 while (lulen && rulen) {
ba210ebe 987 luc = utf8_to_uv_chk((U8*)lc, lulen, &ulen, 0);
0c57e439 988 lc += ulen;
989 lulen -= ulen;
ba210ebe 990 ruc = utf8_to_uv_chk((U8*)rc, rulen, &ulen, 0);
0c57e439 991 rc += ulen;
992 rulen -= ulen;
993 duc = luc ^ ruc;
994 dc = (char*)uv_to_utf8((U8*)dc, duc);
995 }
996 goto mop_up_utf;
997 case OP_BIT_OR:
998 while (lulen && rulen) {
ba210ebe 999 luc = utf8_to_uv_chk((U8*)lc, lulen, &ulen, 0);
0c57e439 1000 lc += ulen;
1001 lulen -= ulen;
ba210ebe 1002 ruc = utf8_to_uv_chk((U8*)rc, rulen, &ulen, 0);
0c57e439 1003 rc += ulen;
1004 rulen -= ulen;
1005 duc = luc | ruc;
1006 dc = (char*)uv_to_utf8((U8*)dc, duc);
1007 }
1008 mop_up_utf:
c9b3c8d0 1009 if (sv == left || sv == right)
1010 (void)sv_usepvn(sv, dcsave, needlen);
1011 SvCUR_set(sv, dc - dcsave);
0c57e439 1012 if (rulen)
1013 sv_catpvn(sv, rc, rulen);
1014 else if (lulen)
1015 sv_catpvn(sv, lc, lulen);
1016 else
1017 *SvEND(sv) = '\0';
1018 break;
1019 }
1020 SvUTF8_on(sv);
1021 goto finish;
1022 }
1023 else
79072805 1024#ifdef LIBERAL
1025 if (len >= sizeof(long)*4 &&
1026 !((long)dc % sizeof(long)) &&
1027 !((long)lc % sizeof(long)) &&
1028 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1029 {
1030 I32 remainder = len % (sizeof(long)*4);
1031 len /= (sizeof(long)*4);
1032
1033 dl = (long*)dc;
1034 ll = (long*)lc;
1035 rl = (long*)rc;
1036
1037 switch (optype) {
1038 case OP_BIT_AND:
1039 while (len--) {
1040 *dl++ = *ll++ & *rl++;
1041 *dl++ = *ll++ & *rl++;
1042 *dl++ = *ll++ & *rl++;
1043 *dl++ = *ll++ & *rl++;
1044 }
1045 break;
a0d0e21e 1046 case OP_BIT_XOR:
79072805 1047 while (len--) {
1048 *dl++ = *ll++ ^ *rl++;
1049 *dl++ = *ll++ ^ *rl++;
1050 *dl++ = *ll++ ^ *rl++;
1051 *dl++ = *ll++ ^ *rl++;
1052 }
1053 break;
1054 case OP_BIT_OR:
1055 while (len--) {
1056 *dl++ = *ll++ | *rl++;
1057 *dl++ = *ll++ | *rl++;
1058 *dl++ = *ll++ | *rl++;
1059 *dl++ = *ll++ | *rl++;
1060 }
1061 }
1062
1063 dc = (char*)dl;
1064 lc = (char*)ll;
1065 rc = (char*)rl;
1066
1067 len = remainder;
1068 }
1069#endif
a0d0e21e 1070 {
a0d0e21e 1071 switch (optype) {
1072 case OP_BIT_AND:
1073 while (len--)
1074 *dc++ = *lc++ & *rc++;
1075 break;
1076 case OP_BIT_XOR:
1077 while (len--)
1078 *dc++ = *lc++ ^ *rc++;
1079 goto mop_up;
1080 case OP_BIT_OR:
1081 while (len--)
1082 *dc++ = *lc++ | *rc++;
1083 mop_up:
1084 len = lensave;
1085 if (rightlen > len)
1086 sv_catpvn(sv, rsave + len, rightlen - len);
1087 else if (leftlen > len)
1088 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4 1089 else
1090 *SvEND(sv) = '\0';
a0d0e21e 1091 break;
1092 }
79072805 1093 }
0c57e439 1094finish:
fb73857a 1095 SvTAINT(sv);
79072805 1096}
463ee0b2 1097
1098OP *
cea2e8a9 1099Perl_do_kv(pTHX)
463ee0b2 1100{
4e35701f 1101 djSP;
463ee0b2 1102 HV *hv = (HV*)POPs;
800e9ae0 1103 HV *keys;
463ee0b2 1104 register HE *entry;
463ee0b2 1105 SV *tmpstr;
54310121 1106 I32 gimme = GIMME_V;
533c011a 1107 I32 dokeys = (PL_op->op_type == OP_KEYS);
1108 I32 dovalues = (PL_op->op_type == OP_VALUES);
c750a3ec 1109 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
8e84507e 1110
1111 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1112 dokeys = dovalues = TRUE;
463ee0b2 1113
85581909 1114 if (!hv) {
533c011a 1115 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1116 dTARGET; /* make sure to clear its target here */
1117 if (SvTYPE(TARG) == SVt_PVLV)
1118 LvTARG(TARG) = Nullsv;
1119 PUSHs(TARG);
1120 }
463ee0b2 1121 RETURN;
85581909 1122 }
748a9306 1123
800e9ae0 1124 keys = realhv ? hv : avhv_keys((AV*)hv);
1125 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1126
54310121 1127 if (gimme == G_VOID)
aa689395 1128 RETURN;
1129
54310121 1130 if (gimme == G_SCALAR) {
6ee623d5 1131 IV i;
463ee0b2 1132 dTARGET;
1133
533c011a 1134 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909 1135 if (SvTYPE(TARG) < SVt_PVLV) {
1136 sv_upgrade(TARG, SVt_PVLV);
1137 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1138 }
1139 LvTYPE(TARG) = 'k';
800e9ae0 1140 if (LvTARG(TARG) != (SV*)keys) {
6ff81951 1141 if (LvTARG(TARG))
1142 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1143 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1144 }
85581909 1145 PUSHs(TARG);
1146 RETURN;
1147 }
1148
33c27489 1149 if (! SvTIED_mg((SV*)keys, 'P'))
800e9ae0 1150 i = HvKEYS(keys);
463ee0b2 1151 else {
1152 i = 0;
463ee0b2 1153 /*SUPPRESS 560*/
800e9ae0 1154 while (hv_iternext(keys)) i++;
463ee0b2 1155 }
1156 PUSHi( i );
1157 RETURN;
1158 }
1159
8ed4b672 1160 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1161
463ee0b2 1162 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1163 while ((entry = hv_iternext(keys))) {
463ee0b2 1164 SPAGAIN;
8c2cee6f 1165 if (dokeys)
1166 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
463ee0b2 1167 if (dovalues) {
463ee0b2 1168 PUTBACK;
b6429b1b 1169 tmpstr = realhv ?
1170 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
cea2e8a9 1171 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1172 (unsigned long)HeHASH(entry),
800e9ae0 1173 HvMAX(keys)+1,
1174 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1175 SPAGAIN;
46fc3d4c 1176 XPUSHs(tmpstr);
463ee0b2 1177 }
1178 PUTBACK;
1179 }
1180 return NORMAL;
1181}
4e35701f 1182