This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Tweak perlio.c so that makedef.pl-exported perlsio_binmode()
[perl5.git] / doop.c
CommitLineData
a0d0e21e 1/* doop.c
79072805 2 *
3818b22b 3 * Copyright (c) 1991-2000, 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;
463ee0b2 32 STRLEN len;
4757a243
LW
33 short *tbl;
34 I32 ch;
79072805 35
4757a243
LW
36 tbl = (short*)cPVOP->op_pv;
37 if (!tbl)
2269b42e 38 Perl_croak(aTHX_ "panic: do_trans_simple");
a0ed51b3 39
4757a243
LW
40 s = (U8*)SvPV(sv, len);
41 send = s + len;
42
b250498f 43 /* First, take care of non-UTF8 input strings, because they're easy */
60aeb6fd 44 if (!SvUTF8(sv)) {
01ec43d0 45 while (s < send) {
036b4402 46 if ((ch = tbl[*s]) >= 0) {
01ec43d0
GS
47 matches++;
48 *s++ = ch;
49 }
50 else
51 s++;
52 }
53 SvSETMAGIC(sv);
b250498f
GS
54 return matches;
55 }
4757a243 56
b250498f
GS
57 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
58 Newz(0, d, len*2+1, U8);
59 dstart = d;
60 while (s < send) {
ba210ebe 61 STRLEN ulen;
b250498f
GS
62 short c;
63
64 ulen = 1;
65 /* Need to check this, otherwise 128..255 won't match */
dcad2880 66 c = utf8_to_uv(s, send - s, &ulen, 0);
b250498f
GS
67 if (c < 0x100 && (ch = tbl[(short)c]) >= 0) {
68 matches++;
60aeb6fd
NIS
69 if (ch < 0x80)
70 *d++ = ch;
71 else
72 d = uv_to_utf8(d,ch);
b250498f 73 s += ulen;
01ec43d0
GS
74 }
75 else { /* No match -> copy */
b250498f
GS
76 while (ulen--)
77 *d++ = *s++;
78 }
79 }
01ec43d0 80 *d = '\0';
60aeb6fd 81 sv_setpvn(sv, (char*)dstart, d - dstart);
b250498f 82 SvUTF8_on(sv);
b250498f 83 SvSETMAGIC(sv);
4757a243
LW
84 return matches;
85}
86
942e002e 87STATIC I32
036b4402 88S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */
4757a243 89{
4757a243
LW
90 U8 *s;
91 U8 *send;
92 I32 matches = 0;
93 STRLEN len;
94 short *tbl;
95
96 tbl = (short*)cPVOP->op_pv;
97 if (!tbl)
2269b42e 98 Perl_croak(aTHX_ "panic: do_trans_count");
4757a243
LW
99
100 s = (U8*)SvPV(sv, len);
101 send = s + len;
102
60aeb6fd
NIS
103 if (!SvUTF8(sv))
104 while (s < send) {
105 if (tbl[*s++] >= 0)
036b4402 106 matches++;
60aeb6fd
NIS
107 }
108 else
109 while (s < send) {
110 UV c;
111 STRLEN ulen;
112 c = utf8_to_uv(s, send - s, &ulen, 0);
113 if (c < 0x100 && tbl[c] >= 0)
114 matches++;
115 s += ulen;
116 }
4757a243
LW
117
118 return matches;
119}
120
942e002e 121STATIC I32
b250498f 122S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */
4757a243 123{
4757a243
LW
124 U8 *s;
125 U8 *send;
126 U8 *d;
5e44153e 127 U8 *dstart;
60aeb6fd 128 I32 isutf8;
4757a243
LW
129 I32 matches = 0;
130 STRLEN len;
131 short *tbl;
132 I32 ch;
133
134 tbl = (short*)cPVOP->op_pv;
135 if (!tbl)
2269b42e 136 Perl_croak(aTHX_ "panic: do_trans_complex");
4757a243
LW
137
138 s = (U8*)SvPV(sv, len);
60aeb6fd 139 isutf8 = SvUTF8(sv);
4757a243
LW
140 send = s + len;
141
60aeb6fd
NIS
142 if (!isutf8) {
143 dstart = d = s;
144 if (PL_op->op_private & OPpTRANS_SQUASH) {
145 U8* p = send;
146 while (s < send) {
147 if ((ch = tbl[*s]) >= 0) {
036b4402
GS
148 *d = ch;
149 matches++;
60aeb6fd
NIS
150 if (p != d - 1 || *p != *d)
151 p = d++;
152 }
153 else if (ch == -1) /* -1 is unmapped character */
154 *d++ = *s;
155 else if (ch == -2) /* -2 is delete character */
156 matches++;
157 s++;
158 }
a0ed51b3 159 }
60aeb6fd
NIS
160 else {
161 while (s < send) {
162 if ((ch = tbl[*s]) >= 0) {
163 matches++;
164 *d++ = ch;
165 }
166 else if (ch == -1) /* -1 is unmapped character */
167 *d++ = *s;
168 else if (ch == -2) /* -2 is delete character */
169 matches++;
170 s++;
171 }
172 }
173 SvCUR_set(sv, d - dstart);
4757a243 174 }
60aeb6fd
NIS
175 else { /* isutf8 */
176 Newz(0, d, len*2+1, U8);
177 dstart = d;
178
179 if (PL_op->op_private & OPpTRANS_SQUASH) {
180 U8* p = send;
181 UV pch = 0xfeedface;
182 while (s < send) {
183 STRLEN len;
184 UV comp = utf8_to_uv_simple(s, &len);
185
186 if (comp > 0xff)
187 d = uv_to_utf8(d, comp); /* always unmapped */
188 else if ((ch = tbl[comp]) >= 0) {
189 matches++;
190 if (ch != pch) {
191 d = uv_to_utf8(d, ch);
192 pch = ch;
193 }
194 s += len;
195 continue;
196 }
197 else if (ch == -1) /* -1 is unmapped character */
198 d = uv_to_utf8(d, comp);
199 else if (ch == -2) /* -2 is delete character */
200 matches++;
201 s += len;
202 pch = 0xfeedface;
203 }
5d06d08e 204 }
60aeb6fd
NIS
205 else {
206 while (s < send) {
207 STRLEN len;
208 UV comp = utf8_to_uv_simple(s, &len);
209 if (comp > 0xff)
210 d = uv_to_utf8(d, comp); /* always unmapped */
211 else if ((ch = tbl[comp]) >= 0) {
212 d = uv_to_utf8(d, ch);
213 matches++;
214 }
215 else if (ch == -1) { /* -1 is unmapped character */
216 d = uv_to_utf8(d, comp);
217 }
218 else if (ch == -2) /* -2 is delete character */
219 matches++;
220 s += len;
221 }
222 }
223 *d = '\0';
224 sv_setpvn(sv, (char*)dstart, d - dstart);
225 SvUTF8_on(sv);
4757a243 226 }
5e44153e 227 SvSETMAGIC(sv);
4757a243
LW
228 return matches;
229}
230
942e002e 231STATIC I32
036b4402 232S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */
4757a243 233{
4757a243
LW
234 U8 *s;
235 U8 *send;
236 U8 *d;
036b4402 237 U8 *start;
60aeb6fd 238 U8 *dstart, *dend;
4757a243
LW
239 I32 matches = 0;
240 STRLEN len;
241
242 SV* rv = (SV*)cSVOP->op_sv;
243 HV* hv = (HV*)SvRV(rv);
244 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
245 UV none = svp ? SvUV(*svp) : 0x7fffffff;
246 UV extra = none + 1;
247 UV final;
248 UV uv;
60aeb6fd
NIS
249 I32 isutf8;
250 U8 hibit = 0;
4757a243
LW
251
252 s = (U8*)SvPV(sv, len);
60aeb6fd
NIS
253 isutf8 = SvUTF8(sv);
254 if (!isutf8) {
255 U8 *t = s, *e = s + len;
256 while (t < e)
257 if ((hibit = *t++ & 0x80))
258 break;
259 if (hibit)
260 s = bytes_to_utf8(s, &len);
261 }
4757a243 262 send = s + len;
036b4402 263 start = s;
4757a243
LW
264
265 svp = hv_fetch(hv, "FINAL", 5, FALSE);
266 if (svp)
267 final = SvUV(*svp);
268
036b4402 269 /* d needs to be bigger than s, in case e.g. upgrading is required */
60aeb6fd
NIS
270 New(0, d, len*3+UTF8_MAXLEN, U8);
271 dend = d + len * 3;
036b4402 272 dstart = d;
60aeb6fd 273
4757a243
LW
274 while (s < send) {
275 if ((uv = swash_fetch(rv, s)) < none) {
276 s += UTF8SKIP(s);
277 matches++;
278 d = uv_to_utf8(d, uv);
279 }
280 else if (uv == none) {
60aeb6fd 281 int i = UTF8SKIP(s);
036b4402 282 while(i--)
075321c3 283 *d++ = *s++;
4757a243
LW
284 }
285 else if (uv == extra) {
60aeb6fd 286 int i = UTF8SKIP(s);
036b4402 287 s += i;
4757a243
LW
288 matches++;
289 d = uv_to_utf8(d, final);
290 }
291 else
292 s += UTF8SKIP(s);
60aeb6fd
NIS
293
294 if (d >= dend) {
295 STRLEN clen = d - dstart;
296 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
297 Renew(dstart, nlen+UTF8_MAXLEN, U8);
298 d = dstart + clen;
299 dend = dstart + nlen;
300 }
4757a243
LW
301 }
302 *d = '\0';
60aeb6fd 303 sv_setpvn(sv, (char*)dstart, d - dstart);
4757a243 304 SvSETMAGIC(sv);
60aeb6fd
NIS
305 SvUTF8_on(sv);
306 if (hibit)
307 Safefree(start);
308 if (!isutf8 && !(PL_hints & HINT_UTF8))
309 sv_utf8_downgrade(sv, TRUE);
4757a243
LW
310
311 return matches;
312}
313
942e002e 314STATIC I32
036b4402 315S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */
4757a243 316{
4757a243 317 U8 *s;
60aeb6fd 318 U8 *start, *send;
4757a243
LW
319 I32 matches = 0;
320 STRLEN len;
321
322 SV* rv = (SV*)cSVOP->op_sv;
323 HV* hv = (HV*)SvRV(rv);
324 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
325 UV none = svp ? SvUV(*svp) : 0x7fffffff;
326 UV uv;
60aeb6fd 327 U8 hibit = 0;
4757a243
LW
328
329 s = (U8*)SvPV(sv, len);
60aeb6fd
NIS
330 if (!SvUTF8(sv)) {
331 U8 *t = s, *e = s + len;
332 while (t < e)
333 if ((hibit = *t++ & 0x80))
334 break;
335 if (hibit)
336 start = s = bytes_to_utf8(s, &len);
337 }
4757a243
LW
338 send = s + len;
339
340 while (s < send) {
834a4ddd 341 if ((uv = swash_fetch(rv, s)) < none)
4757a243 342 matches++;
834a4ddd 343 s += UTF8SKIP(s);
4757a243 344 }
60aeb6fd
NIS
345 if (hibit)
346 Safefree(start);
4757a243
LW
347
348 return matches;
349}
350
942e002e 351STATIC I32
036b4402 352S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */
4757a243 353{
4757a243 354 U8 *s;
60aeb6fd 355 U8 *start, *send;
4757a243
LW
356 U8 *d;
357 I32 matches = 0;
358 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
4757a243
LW
359 I32 del = PL_op->op_private & OPpTRANS_DELETE;
360 SV* rv = (SV*)cSVOP->op_sv;
361 HV* hv = (HV*)SvRV(rv);
362 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
363 UV none = svp ? SvUV(*svp) : 0x7fffffff;
364 UV extra = none + 1;
365 UV final;
366 UV uv;
367 STRLEN len;
60aeb6fd
NIS
368 U8 *dstart, *dend;
369 I32 isutf8;
370 U8 hibit = 0;
4757a243
LW
371
372 s = (U8*)SvPV(sv, len);
60aeb6fd
NIS
373 isutf8 = SvUTF8(sv);
374 if (!isutf8) {
375 U8 *t = s, *e = s + len;
376 while (t < e)
377 if ((hibit = *t++ & 0x80))
378 break;
379 if (hibit)
380 s = bytes_to_utf8(s, &len);
381 }
4757a243 382 send = s + len;
60aeb6fd 383 start = s;
4757a243
LW
384
385 svp = hv_fetch(hv, "FINAL", 5, FALSE);
386 if (svp)
387 final = SvUV(*svp);
388
60aeb6fd
NIS
389 New(0, d, len*3+UTF8_MAXLEN, U8);
390 dend = d + len * 3;
391 dstart = d;
4757a243
LW
392
393 if (squash) {
394 UV puv = 0xfeedface;
395 while (s < send) {
60aeb6fd
NIS
396 uv = swash_fetch(rv, s);
397
398 if (d >= dend) {
399 STRLEN clen = d - dstart, nlen = dend - dstart + len;
400 Renew(dstart, nlen+UTF8_MAXLEN, U8);
401 d = dstart + clen;
402 dend = dstart + nlen;
4757a243
LW
403 }
404 if (uv < none) {
405 matches++;
406 if (uv != puv) {
01ec43d0 407 d = uv_to_utf8(d, uv);
4757a243
LW
408 puv = uv;
409 }
075321c3 410 s += UTF8SKIP(s);
4757a243
LW
411 continue;
412 }
413 else if (uv == none) { /* "none" is unmapped character */
60aeb6fd
NIS
414 int i = UTF8SKIP(s);
415 while(i--)
416 *d++ = *s++;
4757a243
LW
417 puv = 0xfeedface;
418 continue;
a0ed51b3 419 }
4757a243
LW
420 else if (uv == extra && !del) {
421 matches++;
422 if (uv != puv) {
01ec43d0 423 d = uv_to_utf8(d, final);
4757a243
LW
424 puv = final;
425 }
01ec43d0 426 s += UTF8SKIP(s);
4757a243
LW
427 continue;
428 }
01ec43d0
GS
429 matches++; /* "none+1" is delete character */
430 s += UTF8SKIP(s);
a0ed51b3 431 }
79072805
LW
432 }
433 else {
4757a243 434 while (s < send) {
60aeb6fd
NIS
435 uv = swash_fetch(rv, s);
436 if (d >= dend) {
437 STRLEN clen = d - dstart, nlen = dend - dstart + len;
438 Renew(dstart, nlen+UTF8_MAXLEN, U8);
439 d = dstart + clen;
440 dend = dstart + nlen;
a0ed51b3 441 }
4757a243
LW
442 if (uv < none) {
443 matches++;
01ec43d0
GS
444 d = uv_to_utf8(d, uv);
445 s += UTF8SKIP(s);
4757a243 446 continue;
a0ed51b3 447 }
4757a243 448 else if (uv == none) { /* "none" is unmapped character */
60aeb6fd
NIS
449 int i = UTF8SKIP(s);
450 while(i--)
451 *d++ = *s++;
4757a243 452 continue;
79072805 453 }
4757a243
LW
454 else if (uv == extra && !del) {
455 matches++;
01ec43d0
GS
456 d = uv_to_utf8(d, final);
457 s += UTF8SKIP(s);
4757a243
LW
458 continue;
459 }
01ec43d0
GS
460 matches++; /* "none+1" is delete character */
461 s += UTF8SKIP(s);
79072805 462 }
4757a243 463 }
60aeb6fd
NIS
464 *d = '\0';
465 sv_setpvn(sv, (char*)dstart, d - dstart);
466 SvUTF8_on(sv);
467 if (hibit)
468 Safefree(start);
469 if (!isutf8 && !(PL_hints & HINT_UTF8))
470 sv_utf8_downgrade(sv, TRUE);
4757a243
LW
471 SvSETMAGIC(sv);
472
473 return matches;
474}
475
476I32
864dbfa3 477Perl_do_trans(pTHX_ SV *sv)
4757a243
LW
478{
479 STRLEN len;
8e84507e 480 I32 hasutf = (PL_op->op_private &
036b4402 481 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
4757a243
LW
482
483 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
cea2e8a9 484 Perl_croak(aTHX_ PL_no_modify);
4757a243
LW
485
486 (void)SvPV(sv, len);
487 if (!len)
488 return 0;
489 if (!SvPOKp(sv))
490 (void)SvPV_force(sv, len);
2de7b02f
GS
491 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
492 (void)SvPOK_only_UTF8(sv);
4757a243 493
cea2e8a9 494 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
4757a243 495
036b4402 496 switch (PL_op->op_private & ~hasutf & 63) {
4757a243 497 case 0:
01ec43d0
GS
498 if (hasutf)
499 return do_trans_simple_utf8(sv);
500 else
501 return do_trans_simple(sv);
4757a243
LW
502
503 case OPpTRANS_IDENTICAL:
01ec43d0
GS
504 if (hasutf)
505 return do_trans_count_utf8(sv);
506 else
507 return do_trans_count(sv);
4757a243
LW
508
509 default:
01ec43d0 510 if (hasutf)
036b4402 511 return do_trans_complex_utf8(sv);
4757a243 512 else
036b4402 513 return do_trans_complex(sv);
79072805 514 }
79072805
LW
515}
516
517void
864dbfa3 518Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
79072805
LW
519{
520 SV **oldmark = mark;
521 register I32 items = sp - mark;
79072805 522 register STRLEN len;
463ee0b2
LW
523 STRLEN delimlen;
524 register char *delim = SvPV(del, delimlen);
525 STRLEN tmplen;
79072805
LW
526
527 mark++;
528 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
07f14f54 529 (void)SvUPGRADE(sv, SVt_PV);
79072805
LW
530 if (SvLEN(sv) < len + items) { /* current length is way too short */
531 while (items-- > 0) {
1426bbf4 532 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
463ee0b2
LW
533 SvPV(*mark, tmplen);
534 len += tmplen;
79072805
LW
535 }
536 mark++;
537 }
538 SvGROW(sv, len + 1); /* so try to pre-extend */
539
540 mark = oldmark;
db7c17d7 541 items = sp - mark;
79072805
LW
542 ++mark;
543 }
544
463ee0b2 545 if (items-- > 0) {
92d29cee
JH
546 sv_setpv(sv, "");
547 if (*mark)
548 sv_catsv(sv, *mark);
463ee0b2
LW
549 mark++;
550 }
79072805
LW
551 else
552 sv_setpv(sv,"");
c512ce4f 553 if (delimlen) {
79072805 554 for (; items > 0; items--,mark++) {
c512ce4f 555 sv_catsv(sv,del);
79072805
LW
556 sv_catsv(sv,*mark);
557 }
558 }
559 else {
560 for (; items > 0; items--,mark++)
561 sv_catsv(sv,*mark);
562 }
563 SvSETMAGIC(sv);
564}
565
566void
864dbfa3 567Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
79072805 568{
46fc3d4c 569 STRLEN patlen;
570 char *pat = SvPV(*sarg, patlen);
571 bool do_taint = FALSE;
572
573 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
79072805 574 SvSETMAGIC(sv);
46fc3d4c 575 if (do_taint)
576 SvTAINTED_on(sv);
79072805
LW
577}
578
33b45480 579/* currently converts input to bytes if possible, but doesn't sweat failure */
81e118e0
JH
580UV
581Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
582{
583 STRLEN srclen, len;
584 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
585 UV retnum = 0;
586
a50d7633 587 if (offset < 0)
81e118e0 588 return retnum;
8e84507e 589 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 590 Perl_croak(aTHX_ "Illegal number of bits in vec");
246fae53 591
dcad2880 592 if (SvUTF8(sv))
33b45480 593 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
246fae53 594
81e118e0
JH
595 offset *= size; /* turn into bit offset */
596 len = (offset + size + 7) / 8; /* required number of bytes */
597 if (len > srclen) {
598 if (size <= 8)
599 retnum = 0;
600 else {
601 offset >>= 3; /* turn into byte offset */
602 if (size == 16) {
603 if (offset >= srclen)
604 retnum = 0;
605 else
628e1a40 606 retnum = (UV) s[offset] << 8;
81e118e0
JH
607 }
608 else if (size == 32) {
609 if (offset >= srclen)
610 retnum = 0;
611 else if (offset + 1 >= srclen)
612 retnum =
613 ((UV) s[offset ] << 24);
614 else if (offset + 2 >= srclen)
615 retnum =
616 ((UV) s[offset ] << 24) +
617 ((UV) s[offset + 1] << 16);
618 else
619 retnum =
620 ((UV) s[offset ] << 24) +
621 ((UV) s[offset + 1] << 16) +
622 ( s[offset + 2] << 8);
623 }
d7d93a81 624#ifdef UV_IS_QUAD
c5a0f51a 625 else if (size == 64) {
c5a0f51a
JH
626 if (ckWARN(WARN_PORTABLE))
627 Perl_warner(aTHX_ WARN_PORTABLE,
628 "Bit vector size > 32 non-portable");
629 if (offset >= srclen)
630 retnum = 0;
631 else if (offset + 1 >= srclen)
632 retnum =
633 (UV) s[offset ] << 56;
634 else if (offset + 2 >= srclen)
635 retnum =
636 ((UV) s[offset ] << 56) +
637 ((UV) s[offset + 1] << 48);
638 else if (offset + 3 >= srclen)
639 retnum =
640 ((UV) s[offset ] << 56) +
641 ((UV) s[offset + 1] << 48) +
642 ((UV) s[offset + 2] << 40);
643 else if (offset + 4 >= srclen)
644 retnum =
645 ((UV) s[offset ] << 56) +
646 ((UV) s[offset + 1] << 48) +
647 ((UV) s[offset + 2] << 40) +
648 ((UV) s[offset + 3] << 32);
649 else if (offset + 5 >= srclen)
650 retnum =
651 ((UV) s[offset ] << 56) +
652 ((UV) s[offset + 1] << 48) +
653 ((UV) s[offset + 2] << 40) +
654 ((UV) s[offset + 3] << 32) +
655 ( s[offset + 4] << 24);
656 else if (offset + 6 >= srclen)
657 retnum =
658 ((UV) s[offset ] << 56) +
659 ((UV) s[offset + 1] << 48) +
660 ((UV) s[offset + 2] << 40) +
661 ((UV) s[offset + 3] << 32) +
662 ((UV) s[offset + 4] << 24) +
663 ((UV) s[offset + 5] << 16);
664 else
8e84507e 665 retnum =
c5a0f51a
JH
666 ((UV) s[offset ] << 56) +
667 ((UV) s[offset + 1] << 48) +
668 ((UV) s[offset + 2] << 40) +
669 ((UV) s[offset + 3] << 32) +
670 ((UV) s[offset + 4] << 24) +
671 ((UV) s[offset + 5] << 16) +
628e1a40 672 ( s[offset + 6] << 8);
c5a0f51a
JH
673 }
674#endif
81e118e0
JH
675 }
676 }
677 else if (size < 8)
678 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
679 else {
680 offset >>= 3; /* turn into byte offset */
681 if (size == 8)
682 retnum = s[offset];
683 else if (size == 16)
684 retnum =
628e1a40 685 ((UV) s[offset] << 8) +
81e118e0
JH
686 s[offset + 1];
687 else if (size == 32)
688 retnum =
689 ((UV) s[offset ] << 24) +
690 ((UV) s[offset + 1] << 16) +
691 ( s[offset + 2] << 8) +
692 s[offset + 3];
d7d93a81 693#ifdef UV_IS_QUAD
c5a0f51a 694 else if (size == 64) {
c5a0f51a
JH
695 if (ckWARN(WARN_PORTABLE))
696 Perl_warner(aTHX_ WARN_PORTABLE,
697 "Bit vector size > 32 non-portable");
698 retnum =
699 ((UV) s[offset ] << 56) +
700 ((UV) s[offset + 1] << 48) +
701 ((UV) s[offset + 2] << 40) +
702 ((UV) s[offset + 3] << 32) +
703 ((UV) s[offset + 4] << 24) +
704 ((UV) s[offset + 5] << 16) +
628e1a40 705 ( s[offset + 6] << 8) +
c5a0f51a
JH
706 s[offset + 7];
707 }
708#endif
81e118e0
JH
709 }
710
711 return retnum;
712}
713
33b45480
SB
714/* currently converts input to bytes if possible but doesn't sweat failures,
715 * although it does ensure that the string it clobbers is not marked as
716 * utf8-valid any more
717 */
79072805 718void
864dbfa3 719Perl_do_vecset(pTHX_ SV *sv)
79072805
LW
720{
721 SV *targ = LvTARG(sv);
722 register I32 offset;
723 register I32 size;
8990e307 724 register unsigned char *s;
81e118e0 725 register UV lval;
79072805 726 I32 mask;
a0d0e21e
LW
727 STRLEN targlen;
728 STRLEN len;
79072805 729
8990e307
LW
730 if (!targ)
731 return;
a0d0e21e 732 s = (unsigned char*)SvPV_force(targ, targlen);
246fae53 733 if (SvUTF8(targ)) {
33b45480
SB
734 /* This is handled by the SvPOK_only below...
735 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
736 SvUTF8_off(targ);
737 */
738 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
246fae53
MG
739 }
740
4ebbc975 741 (void)SvPOK_only(targ);
81e118e0 742 lval = SvUV(sv);
79072805 743 offset = LvTARGOFF(sv);
fe58ced6
MG
744 if (offset < 0)
745 Perl_croak(aTHX_ "Assigning to negative offset in vec");
79072805 746 size = LvTARGLEN(sv);
8e84507e 747 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
a50d7633 748 Perl_croak(aTHX_ "Illegal number of bits in vec");
8e84507e 749
81e118e0
JH
750 offset *= size; /* turn into bit offset */
751 len = (offset + size + 7) / 8; /* required number of bytes */
a0d0e21e
LW
752 if (len > targlen) {
753 s = (unsigned char*)SvGROW(targ, len + 1);
12ae5dfc 754 (void)memzero((char *)(s + targlen), len - targlen + 1);
a0d0e21e
LW
755 SvCUR_set(targ, len);
756 }
8e84507e 757
79072805
LW
758 if (size < 8) {
759 mask = (1 << size) - 1;
760 size = offset & 7;
761 lval &= mask;
81e118e0 762 offset >>= 3; /* turn into byte offset */
79072805
LW
763 s[offset] &= ~(mask << size);
764 s[offset] |= lval << size;
765 }
766 else {
81e118e0 767 offset >>= 3; /* turn into byte offset */
79072805 768 if (size == 8)
c5a0f51a 769 s[offset ] = lval & 0xff;
79072805 770 else if (size == 16) {
c5a0f51a
JH
771 s[offset ] = (lval >> 8) & 0xff;
772 s[offset+1] = lval & 0xff;
79072805
LW
773 }
774 else if (size == 32) {
c5a0f51a
JH
775 s[offset ] = (lval >> 24) & 0xff;
776 s[offset+1] = (lval >> 16) & 0xff;
777 s[offset+2] = (lval >> 8) & 0xff;
778 s[offset+3] = lval & 0xff;
779 }
d7d93a81 780#ifdef UV_IS_QUAD
c5a0f51a 781 else if (size == 64) {
c5a0f51a
JH
782 if (ckWARN(WARN_PORTABLE))
783 Perl_warner(aTHX_ WARN_PORTABLE,
784 "Bit vector size > 32 non-portable");
785 s[offset ] = (lval >> 56) & 0xff;
786 s[offset+1] = (lval >> 48) & 0xff;
787 s[offset+2] = (lval >> 40) & 0xff;
788 s[offset+3] = (lval >> 32) & 0xff;
789 s[offset+4] = (lval >> 24) & 0xff;
790 s[offset+5] = (lval >> 16) & 0xff;
791 s[offset+6] = (lval >> 8) & 0xff;
792 s[offset+7] = lval & 0xff;
79072805 793 }
dc1e3f56 794#endif
79072805 795 }
7bb043c3 796 SvSETMAGIC(targ);
79072805
LW
797}
798
799void
864dbfa3 800Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
79072805 801{
463ee0b2 802 STRLEN len;
a0d0e21e 803 char *s;
8e84507e 804
79072805 805 if (SvTYPE(sv) == SVt_PVAV) {
a0d0e21e
LW
806 register I32 i;
807 I32 max;
808 AV* av = (AV*)sv;
809 max = AvFILL(av);
810 for (i = 0; i <= max; i++) {
811 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 812 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
813 do_chop(astr, sv);
814 }
815 return;
79072805 816 }
aa854799 817 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
818 HV* hv = (HV*)sv;
819 HE* entry;
820 (void)hv_iterinit(hv);
821 /*SUPPRESS 560*/
155aba94 822 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
823 do_chop(astr,hv_iterval(hv,entry));
824 return;
79072805 825 }
aa854799 826 else if (SvREADONLY(sv))
cea2e8a9 827 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e 828 s = SvPV(sv, len);
748a9306 829 if (len && !SvPOK(sv))
a0d0e21e 830 s = SvPV_force(sv, len);
7e2040f0 831 if (DO_UTF8(sv)) {
a0ed51b3
LW
832 if (s && len) {
833 char *send = s + len;
834 char *start = s;
835 s = send - 1;
836 while ((*s & 0xc0) == 0x80)
837 --s;
0453d815
PM
838 if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8))
839 Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character");
a0ed51b3
LW
840 sv_setpvn(astr, s, send - s);
841 *s = '\0';
842 SvCUR_set(sv, s - start);
843 SvNIOK_off(sv);
7e2040f0 844 SvUTF8_on(astr);
a0ed51b3
LW
845 }
846 else
847 sv_setpvn(astr, "", 0);
848 }
7e2040f0 849 else if (s && len) {
a0d0e21e
LW
850 s += --len;
851 sv_setpvn(astr, s, 1);
852 *s = '\0';
853 SvCUR_set(sv, len);
2c19a612 854 SvUTF8_off(sv);
a0d0e21e 855 SvNIOK_off(sv);
79072805
LW
856 }
857 else
a0d0e21e
LW
858 sv_setpvn(astr, "", 0);
859 SvSETMAGIC(sv);
7e2040f0 860}
a0d0e21e
LW
861
862I32
864dbfa3 863Perl_do_chomp(pTHX_ register SV *sv)
a0d0e21e 864{
c07a80fd 865 register I32 count;
a0d0e21e
LW
866 STRLEN len;
867 char *s;
c07a80fd 868
3280af22 869 if (RsSNARF(PL_rs))
c07a80fd 870 return 0;
4c5a6083
GS
871 if (RsRECORD(PL_rs))
872 return 0;
c07a80fd 873 count = 0;
a0d0e21e
LW
874 if (SvTYPE(sv) == SVt_PVAV) {
875 register I32 i;
876 I32 max;
877 AV* av = (AV*)sv;
878 max = AvFILL(av);
879 for (i = 0; i <= max; i++) {
880 sv = (SV*)av_fetch(av, i, FALSE);
3280af22 881 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
a0d0e21e
LW
882 count += do_chomp(sv);
883 }
884 return count;
885 }
aa854799 886 else if (SvTYPE(sv) == SVt_PVHV) {
a0d0e21e
LW
887 HV* hv = (HV*)sv;
888 HE* entry;
889 (void)hv_iterinit(hv);
890 /*SUPPRESS 560*/
155aba94 891 while ((entry = hv_iternext(hv)))
a0d0e21e
LW
892 count += do_chomp(hv_iterval(hv,entry));
893 return count;
894 }
aa854799 895 else if (SvREADONLY(sv))
cea2e8a9 896 Perl_croak(aTHX_ PL_no_modify);
a0d0e21e
LW
897 s = SvPV(sv, len);
898 if (len && !SvPOKp(sv))
899 s = SvPV_force(sv, len);
900 if (s && len) {
901 s += --len;
3280af22 902 if (RsPARA(PL_rs)) {
a0d0e21e
LW
903 if (*s != '\n')
904 goto nope;
905 ++count;
906 while (len && s[-1] == '\n') {
907 --len;
908 --s;
909 ++count;
910 }
911 }
a0d0e21e 912 else {
c07a80fd 913 STRLEN rslen;
3280af22 914 char *rsptr = SvPV(PL_rs, rslen);
c07a80fd 915 if (rslen == 1) {
916 if (*s != *rsptr)
917 goto nope;
918 ++count;
919 }
920 else {
8c2cee6f 921 if (len < rslen - 1)
c07a80fd 922 goto nope;
923 len -= rslen - 1;
924 s -= rslen - 1;
36477c24 925 if (memNE(s, rsptr, rslen))
c07a80fd 926 goto nope;
927 count += rslen;
928 }
a0d0e21e 929 }
a0d0e21e
LW
930 *s = '\0';
931 SvCUR_set(sv, len);
932 SvNIOK_off(sv);
933 }
934 nope:
935 SvSETMAGIC(sv);
936 return count;
8e84507e 937}
79072805
LW
938
939void
864dbfa3 940Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
79072805
LW
941{
942#ifdef LIBERAL
943 register long *dl;
944 register long *ll;
945 register long *rl;
946#endif
947 register char *dc;
463ee0b2
LW
948 STRLEN leftlen;
949 STRLEN rightlen;
7a4c00b4 950 register char *lc;
951 register char *rc;
79072805 952 register I32 len;
a0d0e21e 953 I32 lensave;
7a4c00b4 954 char *lsave;
955 char *rsave;
0c57e439
GS
956 bool left_utf = DO_UTF8(left);
957 bool right_utf = DO_UTF8(right);
c9b3c8d0 958 I32 needlen;
0c57e439
GS
959
960 if (left_utf && !right_utf)
961 sv_utf8_upgrade(right);
a1ca4561 962 else if (!left_utf && right_utf)
0c57e439 963 sv_utf8_upgrade(left);
79072805 964
1fbd88dc
CS
965 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
966 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
7a4c00b4 967 lsave = lc = SvPV(left, leftlen);
968 rsave = rc = SvPV(right, rightlen);
93a17b20 969 len = leftlen < rightlen ? leftlen : rightlen;
a0d0e21e 970 lensave = len;
c9b3c8d0
JH
971 if ((left_utf || right_utf) && (sv == left || sv == right)) {
972 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
973 Newz(801, dc, needlen + 1, char);
974 }
975 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
2d8e6c8d
GS
976 STRLEN n_a;
977 dc = SvPV_force(sv, n_a);
ff68c719 978 if (SvCUR(sv) < len) {
979 dc = SvGROW(sv, len + 1);
980 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
981 }
c9b3c8d0
JH
982 if (optype != OP_BIT_AND && (left_utf || right_utf))
983 dc = SvGROW(sv, leftlen + rightlen + 1);
ff68c719 984 }
985 else {
c9b3c8d0
JH
986 needlen = ((optype == OP_BIT_AND)
987 ? len : (leftlen > rightlen ? leftlen : rightlen));
ff68c719 988 Newz(801, dc, needlen + 1, char);
989 (void)sv_usepvn(sv, dc, needlen);
990 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
79072805 991 }
a0d0e21e
LW
992 SvCUR_set(sv, len);
993 (void)SvPOK_only(sv);
0c57e439
GS
994 if (left_utf || right_utf) {
995 UV duc, luc, ruc;
c9b3c8d0 996 char *dcsave = dc;
0c57e439
GS
997 STRLEN lulen = leftlen;
998 STRLEN rulen = rightlen;
ba210ebe 999 STRLEN ulen;
0c57e439
GS
1000
1001 switch (optype) {
1002 case OP_BIT_AND:
1003 while (lulen && rulen) {
cc366d4b 1004 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1005 lc += ulen;
1006 lulen -= ulen;
cc366d4b 1007 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1008 rc += ulen;
1009 rulen -= ulen;
1010 duc = luc & ruc;
1011 dc = (char*)uv_to_utf8((U8*)dc, duc);
1012 }
c9b3c8d0
JH
1013 if (sv == left || sv == right)
1014 (void)sv_usepvn(sv, dcsave, needlen);
1015 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1016 break;
1017 case OP_BIT_XOR:
1018 while (lulen && rulen) {
cc366d4b 1019 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1020 lc += ulen;
1021 lulen -= ulen;
cc366d4b 1022 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1023 rc += ulen;
1024 rulen -= ulen;
1025 duc = luc ^ ruc;
1026 dc = (char*)uv_to_utf8((U8*)dc, duc);
1027 }
1028 goto mop_up_utf;
1029 case OP_BIT_OR:
1030 while (lulen && rulen) {
cc366d4b 1031 luc = utf8_to_uv((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1032 lc += ulen;
1033 lulen -= ulen;
cc366d4b 1034 ruc = utf8_to_uv((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
0c57e439
GS
1035 rc += ulen;
1036 rulen -= ulen;
1037 duc = luc | ruc;
1038 dc = (char*)uv_to_utf8((U8*)dc, duc);
1039 }
1040 mop_up_utf:
c9b3c8d0
JH
1041 if (sv == left || sv == right)
1042 (void)sv_usepvn(sv, dcsave, needlen);
1043 SvCUR_set(sv, dc - dcsave);
0c57e439
GS
1044 if (rulen)
1045 sv_catpvn(sv, rc, rulen);
1046 else if (lulen)
1047 sv_catpvn(sv, lc, lulen);
1048 else
1049 *SvEND(sv) = '\0';
1050 break;
1051 }
1052 SvUTF8_on(sv);
1053 goto finish;
1054 }
1055 else
79072805
LW
1056#ifdef LIBERAL
1057 if (len >= sizeof(long)*4 &&
1058 !((long)dc % sizeof(long)) &&
1059 !((long)lc % sizeof(long)) &&
1060 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1061 {
1062 I32 remainder = len % (sizeof(long)*4);
1063 len /= (sizeof(long)*4);
1064
1065 dl = (long*)dc;
1066 ll = (long*)lc;
1067 rl = (long*)rc;
1068
1069 switch (optype) {
1070 case OP_BIT_AND:
1071 while (len--) {
1072 *dl++ = *ll++ & *rl++;
1073 *dl++ = *ll++ & *rl++;
1074 *dl++ = *ll++ & *rl++;
1075 *dl++ = *ll++ & *rl++;
1076 }
1077 break;
a0d0e21e 1078 case OP_BIT_XOR:
79072805
LW
1079 while (len--) {
1080 *dl++ = *ll++ ^ *rl++;
1081 *dl++ = *ll++ ^ *rl++;
1082 *dl++ = *ll++ ^ *rl++;
1083 *dl++ = *ll++ ^ *rl++;
1084 }
1085 break;
1086 case OP_BIT_OR:
1087 while (len--) {
1088 *dl++ = *ll++ | *rl++;
1089 *dl++ = *ll++ | *rl++;
1090 *dl++ = *ll++ | *rl++;
1091 *dl++ = *ll++ | *rl++;
1092 }
1093 }
1094
1095 dc = (char*)dl;
1096 lc = (char*)ll;
1097 rc = (char*)rl;
1098
1099 len = remainder;
1100 }
1101#endif
a0d0e21e 1102 {
a0d0e21e
LW
1103 switch (optype) {
1104 case OP_BIT_AND:
1105 while (len--)
1106 *dc++ = *lc++ & *rc++;
1107 break;
1108 case OP_BIT_XOR:
1109 while (len--)
1110 *dc++ = *lc++ ^ *rc++;
1111 goto mop_up;
1112 case OP_BIT_OR:
1113 while (len--)
1114 *dc++ = *lc++ | *rc++;
1115 mop_up:
1116 len = lensave;
1117 if (rightlen > len)
1118 sv_catpvn(sv, rsave + len, rightlen - len);
1119 else if (leftlen > len)
1120 sv_catpvn(sv, lsave + len, leftlen - len);
4633a7c4
LW
1121 else
1122 *SvEND(sv) = '\0';
a0d0e21e
LW
1123 break;
1124 }
79072805 1125 }
0c57e439 1126finish:
fb73857a 1127 SvTAINT(sv);
79072805 1128}
463ee0b2
LW
1129
1130OP *
cea2e8a9 1131Perl_do_kv(pTHX)
463ee0b2 1132{
4e35701f 1133 djSP;
463ee0b2 1134 HV *hv = (HV*)POPs;
800e9ae0 1135 HV *keys;
463ee0b2 1136 register HE *entry;
463ee0b2 1137 SV *tmpstr;
54310121 1138 I32 gimme = GIMME_V;
533c011a
NIS
1139 I32 dokeys = (PL_op->op_type == OP_KEYS);
1140 I32 dovalues = (PL_op->op_type == OP_VALUES);
c750a3ec 1141 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
8e84507e
NIS
1142
1143 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
a0d0e21e 1144 dokeys = dovalues = TRUE;
463ee0b2 1145
85581909 1146 if (!hv) {
533c011a 1147 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909
SB
1148 dTARGET; /* make sure to clear its target here */
1149 if (SvTYPE(TARG) == SVt_PVLV)
1150 LvTARG(TARG) = Nullsv;
1151 PUSHs(TARG);
1152 }
463ee0b2 1153 RETURN;
85581909 1154 }
748a9306 1155
800e9ae0
JP
1156 keys = realhv ? hv : avhv_keys((AV*)hv);
1157 (void)hv_iterinit(keys); /* always reset iterator regardless */
748a9306 1158
54310121 1159 if (gimme == G_VOID)
aa689395 1160 RETURN;
1161
54310121 1162 if (gimme == G_SCALAR) {
6ee623d5 1163 IV i;
463ee0b2
LW
1164 dTARGET;
1165
533c011a 1166 if (PL_op->op_flags & OPf_MOD) { /* lvalue */
85581909
SB
1167 if (SvTYPE(TARG) < SVt_PVLV) {
1168 sv_upgrade(TARG, SVt_PVLV);
1169 sv_magic(TARG, Nullsv, 'k', Nullch, 0);
1170 }
1171 LvTYPE(TARG) = 'k';
800e9ae0 1172 if (LvTARG(TARG) != (SV*)keys) {
6ff81951
GS
1173 if (LvTARG(TARG))
1174 SvREFCNT_dec(LvTARG(TARG));
800e9ae0 1175 LvTARG(TARG) = SvREFCNT_inc(keys);
6ff81951 1176 }
85581909
SB
1177 PUSHs(TARG);
1178 RETURN;
1179 }
1180
33c27489 1181 if (! SvTIED_mg((SV*)keys, 'P'))
800e9ae0 1182 i = HvKEYS(keys);
463ee0b2
LW
1183 else {
1184 i = 0;
463ee0b2 1185 /*SUPPRESS 560*/
800e9ae0 1186 while (hv_iternext(keys)) i++;
463ee0b2
LW
1187 }
1188 PUSHi( i );
1189 RETURN;
1190 }
1191
8ed4b672 1192 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
463ee0b2 1193
463ee0b2 1194 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
155aba94 1195 while ((entry = hv_iternext(keys))) {
463ee0b2 1196 SPAGAIN;
6bfc225d 1197 if (dokeys)
8c2cee6f 1198 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
463ee0b2 1199 if (dovalues) {
463ee0b2 1200 PUTBACK;
b6429b1b
GS
1201 tmpstr = realhv ?
1202 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
cea2e8a9 1203 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
46fc3d4c 1204 (unsigned long)HeHASH(entry),
800e9ae0
JP
1205 HvMAX(keys)+1,
1206 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
463ee0b2 1207 SPAGAIN;
46fc3d4c 1208 XPUSHs(tmpstr);
463ee0b2
LW
1209 }
1210 PUTBACK;
1211 }
1212 return NORMAL;
1213}
4e35701f 1214