This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Spread the good word of "env".
[perl5.git] / doop.c
... / ...
CommitLineData
1/* doop.c
2 *
3 * Copyright (c) 1991-2001, Larry Wall
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 *
8 */
9
10/*
11 * "'So that was the job I felt I had to do when I started,' thought Sam."
12 */
13
14#include "EXTERN.h"
15#define PERL_IN_DOOP_C
16#include "perl.h"
17
18#ifndef PERL_MICRO
19#if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX)
20#include <signal.h>
21#endif
22#endif
23
24STATIC I32
25S_do_trans_simple(pTHX_ SV *sv)
26{
27 U8 *s;
28 U8 *d;
29 U8 *send;
30 U8 *dstart;
31 I32 matches = 0;
32 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
33 STRLEN len;
34 short *tbl;
35 I32 ch;
36
37 tbl = (short*)cPVOP->op_pv;
38 if (!tbl)
39 Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__);
40
41 s = (U8*)SvPV(sv, len);
42 send = s + len;
43
44 /* First, take care of non-UTF8 input strings, because they're easy */
45 if (!SvUTF8(sv)) {
46 while (s < send) {
47 if ((ch = tbl[*s]) >= 0) {
48 matches++;
49 *s++ = ch;
50 }
51 else
52 s++;
53 }
54 SvSETMAGIC(sv);
55 return matches;
56 }
57
58 /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */
59 if (grows)
60 New(0, d, len*2+1, U8);
61 else
62 d = s;
63 dstart = d;
64 while (s < send) {
65 STRLEN ulen;
66 UV c;
67
68 /* Need to check this, otherwise 128..255 won't match */
69 c = utf8n_to_uvchr(s, send - s, &ulen, 0);
70 if (c < 0x100 && (ch = tbl[c]) >= 0) {
71 matches++;
72 d = uvchr_to_utf8(d, ch);
73 s += ulen;
74 }
75 else { /* No match -> copy */
76 Copy(s, d, ulen, U8);
77 d += ulen;
78 s += ulen;
79 }
80 }
81 if (grows) {
82 sv_setpvn(sv, (char*)dstart, d - dstart);
83 Safefree(dstart);
84 }
85 else {
86 *d = '\0';
87 SvCUR_set(sv, d - dstart);
88 }
89 SvUTF8_on(sv);
90 SvSETMAGIC(sv);
91 return matches;
92}
93
94STATIC I32
95S_do_trans_count(pTHX_ SV *sv)
96{
97 U8 *s;
98 U8 *send;
99 I32 matches = 0;
100 STRLEN len;
101 short *tbl;
102 I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
103
104 tbl = (short*)cPVOP->op_pv;
105 if (!tbl)
106 Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__);
107
108 s = (U8*)SvPV(sv, len);
109 send = s + len;
110
111 if (!SvUTF8(sv))
112 while (s < send) {
113 if (tbl[*s++] >= 0)
114 matches++;
115 }
116 else
117 while (s < send) {
118 UV c;
119 STRLEN ulen;
120 c = utf8n_to_uvchr(s, send - s, &ulen, 0);
121 if (c < 0x100) {
122 if (tbl[c] >= 0)
123 matches++;
124 } else if (complement)
125 matches++;
126 s += ulen;
127 }
128
129 return matches;
130}
131
132STATIC I32
133S_do_trans_complex(pTHX_ SV *sv)
134{
135 U8 *s;
136 U8 *send;
137 U8 *d;
138 U8 *dstart;
139 I32 isutf8;
140 I32 matches = 0;
141 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
142 I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT;
143 I32 del = PL_op->op_private & OPpTRANS_DELETE;
144 STRLEN len, rlen = 0;
145 short *tbl;
146 I32 ch;
147
148 tbl = (short*)cPVOP->op_pv;
149 if (!tbl)
150 Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__);
151
152 s = (U8*)SvPV(sv, len);
153 isutf8 = SvUTF8(sv);
154 send = s + len;
155
156 if (!isutf8) {
157 dstart = d = s;
158 if (PL_op->op_private & OPpTRANS_SQUASH) {
159 U8* p = send;
160 while (s < send) {
161 if ((ch = tbl[*s]) >= 0) {
162 *d = ch;
163 matches++;
164 if (p != d - 1 || *p != *d)
165 p = d++;
166 }
167 else if (ch == -1) /* -1 is unmapped character */
168 *d++ = *s;
169 else if (ch == -2) /* -2 is delete character */
170 matches++;
171 s++;
172 }
173 }
174 else {
175 while (s < send) {
176 if ((ch = tbl[*s]) >= 0) {
177 matches++;
178 *d++ = ch;
179 }
180 else if (ch == -1) /* -1 is unmapped character */
181 *d++ = *s;
182 else if (ch == -2) /* -2 is delete character */
183 matches++;
184 s++;
185 }
186 }
187 *d = '\0';
188 SvCUR_set(sv, d - dstart);
189 }
190 else { /* isutf8 */
191 if (grows)
192 New(0, d, len*2+1, U8);
193 else
194 d = s;
195 dstart = d;
196 if (complement && !del)
197 rlen = tbl[0x100];
198
199#ifdef MACOS_TRADITIONAL
200#define comp CoMP /* "comp" is a keyword in some compilers ... */
201#endif
202
203 if (PL_op->op_private & OPpTRANS_SQUASH) {
204 UV pch = 0xfeedface;
205 while (s < send) {
206 STRLEN len;
207 UV comp = utf8_to_uvchr(s, &len);
208
209 if (comp > 0xff) {
210 if (!complement) {
211 Copy(s, d, len, U8);
212 d += len;
213 }
214 else {
215 matches++;
216 if (!del) {
217 ch = (rlen == 0) ? comp :
218 (comp - 0x100 < rlen) ?
219 tbl[comp+1] : tbl[0x100+rlen];
220 if (ch != pch) {
221 d = uvchr_to_utf8(d, ch);
222 pch = ch;
223 }
224 s += len;
225 continue;
226 }
227 }
228 }
229 else if ((ch = tbl[comp]) >= 0) {
230 matches++;
231 if (ch != pch) {
232 d = uvchr_to_utf8(d, ch);
233 pch = ch;
234 }
235 s += len;
236 continue;
237 }
238 else if (ch == -1) { /* -1 is unmapped character */
239 Copy(s, d, len, U8);
240 d += len;
241 }
242 else if (ch == -2) /* -2 is delete character */
243 matches++;
244 s += len;
245 pch = 0xfeedface;
246 }
247 }
248 else {
249 while (s < send) {
250 STRLEN len;
251 UV comp = utf8_to_uvchr(s, &len);
252 if (comp > 0xff) {
253 if (!complement) {
254 Copy(s, d, len, U8);
255 d += len;
256 }
257 else {
258 matches++;
259 if (!del) {
260 if (comp - 0x100 < rlen)
261 d = uvchr_to_utf8(d, tbl[comp+1]);
262 else
263 d = uvchr_to_utf8(d, tbl[0x100+rlen]);
264 }
265 }
266 }
267 else if ((ch = tbl[comp]) >= 0) {
268 d = uvchr_to_utf8(d, ch);
269 matches++;
270 }
271 else if (ch == -1) { /* -1 is unmapped character */
272 Copy(s, d, len, U8);
273 d += len;
274 }
275 else if (ch == -2) /* -2 is delete character */
276 matches++;
277 s += len;
278 }
279 }
280 if (grows) {
281 sv_setpvn(sv, (char*)dstart, d - dstart);
282 Safefree(dstart);
283 }
284 else {
285 *d = '\0';
286 SvCUR_set(sv, d - dstart);
287 }
288 SvUTF8_on(sv);
289 }
290 SvSETMAGIC(sv);
291 return matches;
292}
293
294STATIC I32
295S_do_trans_simple_utf8(pTHX_ SV *sv)
296{
297 U8 *s;
298 U8 *send;
299 U8 *d;
300 U8 *start;
301 U8 *dstart, *dend;
302 I32 matches = 0;
303 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
304 STRLEN len;
305
306 SV* rv = (SV*)cSVOP->op_sv;
307 HV* hv = (HV*)SvRV(rv);
308 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
309 UV none = svp ? SvUV(*svp) : 0x7fffffff;
310 UV extra = none + 1;
311 UV final = 0;
312 UV uv;
313 I32 isutf8;
314 U8 hibit = 0;
315
316 s = (U8*)SvPV(sv, len);
317 isutf8 = SvUTF8(sv);
318 if (!isutf8) {
319 U8 *t = s, *e = s + len;
320 while (t < e) {
321 U8 ch = *t++;
322 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
323 break;
324 }
325 if (hibit)
326 s = bytes_to_utf8(s, &len);
327 }
328 send = s + len;
329 start = s;
330
331 svp = hv_fetch(hv, "FINAL", 5, FALSE);
332 if (svp)
333 final = SvUV(*svp);
334
335 if (grows) {
336 /* d needs to be bigger than s, in case e.g. upgrading is required */
337 New(0, d, len*3+UTF8_MAXLEN, U8);
338 dend = d + len * 3;
339 dstart = d;
340 }
341 else {
342 dstart = d = s;
343 dend = d + len;
344 }
345
346 while (s < send) {
347 if ((uv = swash_fetch(rv, s, TRUE)) < none) {
348 s += UTF8SKIP(s);
349 matches++;
350 d = uvuni_to_utf8(d, uv);
351 }
352 else if (uv == none) {
353 int i = UTF8SKIP(s);
354 Copy(s, d, i, U8);
355 d += i;
356 s += i;
357 }
358 else if (uv == extra) {
359 int i = UTF8SKIP(s);
360 s += i;
361 matches++;
362 d = uvuni_to_utf8(d, final);
363 }
364 else
365 s += UTF8SKIP(s);
366
367 if (d > dend) {
368 STRLEN clen = d - dstart;
369 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
370 if (!grows)
371 Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__);
372 Renew(dstart, nlen+UTF8_MAXLEN, U8);
373 d = dstart + clen;
374 dend = dstart + nlen;
375 }
376 }
377 if (grows || hibit) {
378 sv_setpvn(sv, (char*)dstart, d - dstart);
379 Safefree(dstart);
380 if (grows && hibit)
381 Safefree(start);
382 }
383 else {
384 *d = '\0';
385 SvCUR_set(sv, d - dstart);
386 }
387 SvSETMAGIC(sv);
388 SvUTF8_on(sv);
389
390 return matches;
391}
392
393STATIC I32
394S_do_trans_count_utf8(pTHX_ SV *sv)
395{
396 U8 *s;
397 U8 *start = 0, *send;
398 I32 matches = 0;
399 STRLEN len;
400
401 SV* rv = (SV*)cSVOP->op_sv;
402 HV* hv = (HV*)SvRV(rv);
403 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
404 UV none = svp ? SvUV(*svp) : 0x7fffffff;
405 UV extra = none + 1;
406 UV uv;
407 U8 hibit = 0;
408
409 s = (U8*)SvPV(sv, len);
410 if (!SvUTF8(sv)) {
411 U8 *t = s, *e = s + len;
412 while (t < e) {
413 U8 ch = *t++;
414 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
415 break;
416 }
417 if (hibit)
418 start = s = bytes_to_utf8(s, &len);
419 }
420 send = s + len;
421
422 while (s < send) {
423 if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra)
424 matches++;
425 s += UTF8SKIP(s);
426 }
427 if (hibit)
428 Safefree(start);
429
430 return matches;
431}
432
433STATIC I32
434S_do_trans_complex_utf8(pTHX_ SV *sv)
435{
436 U8 *s;
437 U8 *start, *send;
438 U8 *d;
439 I32 matches = 0;
440 I32 squash = PL_op->op_private & OPpTRANS_SQUASH;
441 I32 del = PL_op->op_private & OPpTRANS_DELETE;
442 I32 grows = PL_op->op_private & OPpTRANS_GROWS;
443 SV* rv = (SV*)cSVOP->op_sv;
444 HV* hv = (HV*)SvRV(rv);
445 SV** svp = hv_fetch(hv, "NONE", 4, FALSE);
446 UV none = svp ? SvUV(*svp) : 0x7fffffff;
447 UV extra = none + 1;
448 UV final = 0;
449 bool havefinal = FALSE;
450 UV uv;
451 STRLEN len;
452 U8 *dstart, *dend;
453 I32 isutf8;
454 U8 hibit = 0;
455
456 s = (U8*)SvPV(sv, len);
457 isutf8 = SvUTF8(sv);
458 if (!isutf8) {
459 U8 *t = s, *e = s + len;
460 while (t < e) {
461 U8 ch = *t++;
462 if ((hibit = !NATIVE_IS_INVARIANT(ch)))
463 break;
464 }
465 if (hibit)
466 s = bytes_to_utf8(s, &len);
467 }
468 send = s + len;
469 start = s;
470
471 svp = hv_fetch(hv, "FINAL", 5, FALSE);
472 if (svp) {
473 final = SvUV(*svp);
474 havefinal = TRUE;
475 }
476
477 if (grows) {
478 /* d needs to be bigger than s, in case e.g. upgrading is required */
479 New(0, d, len*3+UTF8_MAXLEN, U8);
480 dend = d + len * 3;
481 dstart = d;
482 }
483 else {
484 dstart = d = s;
485 dend = d + len;
486 }
487
488 if (squash) {
489 UV puv = 0xfeedface;
490 while (s < send) {
491 uv = swash_fetch(rv, s, TRUE);
492
493 if (d > dend) {
494 STRLEN clen = d - dstart;
495 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
496 if (!grows)
497 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
498 Renew(dstart, nlen+UTF8_MAXLEN, U8);
499 d = dstart + clen;
500 dend = dstart + nlen;
501 }
502 if (uv < none) {
503 matches++;
504 s += UTF8SKIP(s);
505 if (uv != puv) {
506 d = uvuni_to_utf8(d, uv);
507 puv = uv;
508 }
509 continue;
510 }
511 else if (uv == none) { /* "none" is unmapped character */
512 int i = UTF8SKIP(s);
513 Copy(s, d, i, U8);
514 d += i;
515 s += i;
516 puv = 0xfeedface;
517 continue;
518 }
519 else if (uv == extra && !del) {
520 matches++;
521 if (havefinal) {
522 s += UTF8SKIP(s);
523 if (puv != final) {
524 d = uvuni_to_utf8(d, final);
525 puv = final;
526 }
527 }
528 else {
529 STRLEN len;
530 uv = utf8_to_uvuni(s, &len);
531 if (uv != puv) {
532 Copy(s, d, len, U8);
533 d += len;
534 puv = uv;
535 }
536 s += len;
537 }
538 continue;
539 }
540 matches++; /* "none+1" is delete character */
541 s += UTF8SKIP(s);
542 }
543 }
544 else {
545 while (s < send) {
546 uv = swash_fetch(rv, s, TRUE);
547 if (d > dend) {
548 STRLEN clen = d - dstart;
549 STRLEN nlen = dend - dstart + len + UTF8_MAXLEN;
550 if (!grows)
551 Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__);
552 Renew(dstart, nlen+UTF8_MAXLEN, U8);
553 d = dstart + clen;
554 dend = dstart + nlen;
555 }
556 if (uv < none) {
557 matches++;
558 s += UTF8SKIP(s);
559 d = uvuni_to_utf8(d, uv);
560 continue;
561 }
562 else if (uv == none) { /* "none" is unmapped character */
563 int i = UTF8SKIP(s);
564 Copy(s, d, i, U8);
565 d += i;
566 s += i;
567 continue;
568 }
569 else if (uv == extra && !del) {
570 matches++;
571 s += UTF8SKIP(s);
572 d = uvuni_to_utf8(d, final);
573 continue;
574 }
575 matches++; /* "none+1" is delete character */
576 s += UTF8SKIP(s);
577 }
578 }
579 if (grows || hibit) {
580 sv_setpvn(sv, (char*)dstart, d - dstart);
581 Safefree(dstart);
582 if (grows && hibit)
583 Safefree(start);
584 }
585 else {
586 *d = '\0';
587 SvCUR_set(sv, d - dstart);
588 }
589 SvUTF8_on(sv);
590 SvSETMAGIC(sv);
591
592 return matches;
593}
594
595I32
596Perl_do_trans(pTHX_ SV *sv)
597{
598 STRLEN len;
599 I32 hasutf = (PL_op->op_private &
600 (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF));
601
602 if (SvREADONLY(sv)) {
603 if (SvFAKE(sv))
604 sv_force_normal(sv);
605 if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
606 Perl_croak(aTHX_ PL_no_modify);
607 }
608 (void)SvPV(sv, len);
609 if (!len)
610 return 0;
611 if (!SvPOKp(sv))
612 (void)SvPV_force(sv, len);
613 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
614 (void)SvPOK_only_UTF8(sv);
615
616 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
617
618 switch (PL_op->op_private & ~hasutf & 63) {
619 case 0:
620 if (hasutf)
621 return do_trans_simple_utf8(sv);
622 else
623 return do_trans_simple(sv);
624
625 case OPpTRANS_IDENTICAL:
626 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
627 if (hasutf)
628 return do_trans_count_utf8(sv);
629 else
630 return do_trans_count(sv);
631
632 default:
633 if (hasutf)
634 return do_trans_complex_utf8(sv);
635 else
636 return do_trans_complex(sv);
637 }
638}
639
640void
641Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
642{
643 SV **oldmark = mark;
644 register I32 items = sp - mark;
645 register STRLEN len;
646 STRLEN delimlen;
647 STRLEN tmplen;
648
649 (void) SvPV(del, delimlen); /* stringify and get the delimlen */
650 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
651
652 mark++;
653 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
654 (void)SvUPGRADE(sv, SVt_PV);
655 if (SvLEN(sv) < len + items) { /* current length is way too short */
656 while (items-- > 0) {
657 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
658 SvPV(*mark, tmplen);
659 len += tmplen;
660 }
661 mark++;
662 }
663 SvGROW(sv, len + 1); /* so try to pre-extend */
664
665 mark = oldmark;
666 items = sp - mark;
667 ++mark;
668 }
669
670 sv_setpv(sv, "");
671 if (PL_tainting && SvMAGICAL(sv))
672 SvTAINTED_off(sv);
673
674 if (items-- > 0) {
675 if (*mark)
676 sv_catsv(sv, *mark);
677 mark++;
678 }
679
680 if (delimlen) {
681 for (; items > 0; items--,mark++) {
682 sv_catsv(sv,del);
683 sv_catsv(sv,*mark);
684 }
685 }
686 else {
687 for (; items > 0; items--,mark++)
688 sv_catsv(sv,*mark);
689 }
690 SvSETMAGIC(sv);
691}
692
693void
694Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
695{
696 STRLEN patlen;
697 char *pat = SvPV(*sarg, patlen);
698 bool do_taint = FALSE;
699
700 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
701 SvSETMAGIC(sv);
702 if (do_taint)
703 SvTAINTED_on(sv);
704}
705
706/* currently converts input to bytes if possible, but doesn't sweat failure */
707UV
708Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
709{
710 STRLEN srclen, len;
711 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
712 UV retnum = 0;
713
714 if (offset < 0)
715 return retnum;
716 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
717 Perl_croak(aTHX_ "Illegal number of bits in vec");
718
719 if (SvUTF8(sv))
720 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
721
722 offset *= size; /* turn into bit offset */
723 len = (offset + size + 7) / 8; /* required number of bytes */
724 if (len > srclen) {
725 if (size <= 8)
726 retnum = 0;
727 else {
728 offset >>= 3; /* turn into byte offset */
729 if (size == 16) {
730 if (offset >= srclen)
731 retnum = 0;
732 else
733 retnum = (UV) s[offset] << 8;
734 }
735 else if (size == 32) {
736 if (offset >= srclen)
737 retnum = 0;
738 else if (offset + 1 >= srclen)
739 retnum =
740 ((UV) s[offset ] << 24);
741 else if (offset + 2 >= srclen)
742 retnum =
743 ((UV) s[offset ] << 24) +
744 ((UV) s[offset + 1] << 16);
745 else
746 retnum =
747 ((UV) s[offset ] << 24) +
748 ((UV) s[offset + 1] << 16) +
749 ( s[offset + 2] << 8);
750 }
751#ifdef UV_IS_QUAD
752 else if (size == 64) {
753 if (ckWARN(WARN_PORTABLE))
754 Perl_warner(aTHX_ WARN_PORTABLE,
755 "Bit vector size > 32 non-portable");
756 if (offset >= srclen)
757 retnum = 0;
758 else if (offset + 1 >= srclen)
759 retnum =
760 (UV) s[offset ] << 56;
761 else if (offset + 2 >= srclen)
762 retnum =
763 ((UV) s[offset ] << 56) +
764 ((UV) s[offset + 1] << 48);
765 else if (offset + 3 >= srclen)
766 retnum =
767 ((UV) s[offset ] << 56) +
768 ((UV) s[offset + 1] << 48) +
769 ((UV) s[offset + 2] << 40);
770 else if (offset + 4 >= srclen)
771 retnum =
772 ((UV) s[offset ] << 56) +
773 ((UV) s[offset + 1] << 48) +
774 ((UV) s[offset + 2] << 40) +
775 ((UV) s[offset + 3] << 32);
776 else if (offset + 5 >= srclen)
777 retnum =
778 ((UV) s[offset ] << 56) +
779 ((UV) s[offset + 1] << 48) +
780 ((UV) s[offset + 2] << 40) +
781 ((UV) s[offset + 3] << 32) +
782 ( s[offset + 4] << 24);
783 else if (offset + 6 >= srclen)
784 retnum =
785 ((UV) s[offset ] << 56) +
786 ((UV) s[offset + 1] << 48) +
787 ((UV) s[offset + 2] << 40) +
788 ((UV) s[offset + 3] << 32) +
789 ((UV) s[offset + 4] << 24) +
790 ((UV) s[offset + 5] << 16);
791 else
792 retnum =
793 ((UV) s[offset ] << 56) +
794 ((UV) s[offset + 1] << 48) +
795 ((UV) s[offset + 2] << 40) +
796 ((UV) s[offset + 3] << 32) +
797 ((UV) s[offset + 4] << 24) +
798 ((UV) s[offset + 5] << 16) +
799 ( s[offset + 6] << 8);
800 }
801#endif
802 }
803 }
804 else if (size < 8)
805 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
806 else {
807 offset >>= 3; /* turn into byte offset */
808 if (size == 8)
809 retnum = s[offset];
810 else if (size == 16)
811 retnum =
812 ((UV) s[offset] << 8) +
813 s[offset + 1];
814 else if (size == 32)
815 retnum =
816 ((UV) s[offset ] << 24) +
817 ((UV) s[offset + 1] << 16) +
818 ( s[offset + 2] << 8) +
819 s[offset + 3];
820#ifdef UV_IS_QUAD
821 else if (size == 64) {
822 if (ckWARN(WARN_PORTABLE))
823 Perl_warner(aTHX_ WARN_PORTABLE,
824 "Bit vector size > 32 non-portable");
825 retnum =
826 ((UV) s[offset ] << 56) +
827 ((UV) s[offset + 1] << 48) +
828 ((UV) s[offset + 2] << 40) +
829 ((UV) s[offset + 3] << 32) +
830 ((UV) s[offset + 4] << 24) +
831 ((UV) s[offset + 5] << 16) +
832 ( s[offset + 6] << 8) +
833 s[offset + 7];
834 }
835#endif
836 }
837
838 return retnum;
839}
840
841/* currently converts input to bytes if possible but doesn't sweat failures,
842 * although it does ensure that the string it clobbers is not marked as
843 * utf8-valid any more
844 */
845void
846Perl_do_vecset(pTHX_ SV *sv)
847{
848 SV *targ = LvTARG(sv);
849 register I32 offset;
850 register I32 size;
851 register unsigned char *s;
852 register UV lval;
853 I32 mask;
854 STRLEN targlen;
855 STRLEN len;
856
857 if (!targ)
858 return;
859 s = (unsigned char*)SvPV_force(targ, targlen);
860 if (SvUTF8(targ)) {
861 /* This is handled by the SvPOK_only below...
862 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
863 SvUTF8_off(targ);
864 */
865 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
866 }
867
868 (void)SvPOK_only(targ);
869 lval = SvUV(sv);
870 offset = LvTARGOFF(sv);
871 if (offset < 0)
872 Perl_croak(aTHX_ "Assigning to negative offset in vec");
873 size = LvTARGLEN(sv);
874 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
875 Perl_croak(aTHX_ "Illegal number of bits in vec");
876
877 offset *= size; /* turn into bit offset */
878 len = (offset + size + 7) / 8; /* required number of bytes */
879 if (len > targlen) {
880 s = (unsigned char*)SvGROW(targ, len + 1);
881 (void)memzero((char *)(s + targlen), len - targlen + 1);
882 SvCUR_set(targ, len);
883 }
884
885 if (size < 8) {
886 mask = (1 << size) - 1;
887 size = offset & 7;
888 lval &= mask;
889 offset >>= 3; /* turn into byte offset */
890 s[offset] &= ~(mask << size);
891 s[offset] |= lval << size;
892 }
893 else {
894 offset >>= 3; /* turn into byte offset */
895 if (size == 8)
896 s[offset ] = lval & 0xff;
897 else if (size == 16) {
898 s[offset ] = (lval >> 8) & 0xff;
899 s[offset+1] = lval & 0xff;
900 }
901 else if (size == 32) {
902 s[offset ] = (lval >> 24) & 0xff;
903 s[offset+1] = (lval >> 16) & 0xff;
904 s[offset+2] = (lval >> 8) & 0xff;
905 s[offset+3] = lval & 0xff;
906 }
907#ifdef UV_IS_QUAD
908 else if (size == 64) {
909 if (ckWARN(WARN_PORTABLE))
910 Perl_warner(aTHX_ WARN_PORTABLE,
911 "Bit vector size > 32 non-portable");
912 s[offset ] = (lval >> 56) & 0xff;
913 s[offset+1] = (lval >> 48) & 0xff;
914 s[offset+2] = (lval >> 40) & 0xff;
915 s[offset+3] = (lval >> 32) & 0xff;
916 s[offset+4] = (lval >> 24) & 0xff;
917 s[offset+5] = (lval >> 16) & 0xff;
918 s[offset+6] = (lval >> 8) & 0xff;
919 s[offset+7] = lval & 0xff;
920 }
921#endif
922 }
923 SvSETMAGIC(targ);
924}
925
926void
927Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
928{
929 STRLEN len;
930 char *s;
931
932 if (SvTYPE(sv) == SVt_PVAV) {
933 register I32 i;
934 I32 max;
935 AV* av = (AV*)sv;
936 max = AvFILL(av);
937 for (i = 0; i <= max; i++) {
938 sv = (SV*)av_fetch(av, i, FALSE);
939 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
940 do_chop(astr, sv);
941 }
942 return;
943 }
944 else if (SvTYPE(sv) == SVt_PVHV) {
945 HV* hv = (HV*)sv;
946 HE* entry;
947 (void)hv_iterinit(hv);
948 /*SUPPRESS 560*/
949 while ((entry = hv_iternext(hv)))
950 do_chop(astr,hv_iterval(hv,entry));
951 return;
952 }
953 else if (SvREADONLY(sv)) {
954 if (SvFAKE(sv)) {
955 /* SV is copy-on-write */
956 sv_force_normal_flags(sv, 0);
957 }
958 if (SvREADONLY(sv))
959 Perl_croak(aTHX_ PL_no_modify);
960 }
961 s = SvPV(sv, len);
962 if (len && !SvPOK(sv))
963 s = SvPV_force(sv, len);
964 if (DO_UTF8(sv)) {
965 if (s && len) {
966 char *send = s + len;
967 char *start = s;
968 s = send - 1;
969 while (s > start && UTF8_IS_CONTINUATION(*s))
970 s--;
971 if (utf8_to_uvchr((U8*)s, 0)) {
972 sv_setpvn(astr, s, send - s);
973 *s = '\0';
974 SvCUR_set(sv, s - start);
975 SvNIOK_off(sv);
976 SvUTF8_on(astr);
977 }
978 }
979 else
980 sv_setpvn(astr, "", 0);
981 }
982 else if (s && len) {
983 s += --len;
984 sv_setpvn(astr, s, 1);
985 *s = '\0';
986 SvCUR_set(sv, len);
987 SvUTF8_off(sv);
988 SvNIOK_off(sv);
989 }
990 else
991 sv_setpvn(astr, "", 0);
992 SvSETMAGIC(sv);
993}
994
995I32
996Perl_do_chomp(pTHX_ register SV *sv)
997{
998 register I32 count;
999 STRLEN len;
1000 STRLEN n_a;
1001 char *s;
1002
1003 if (RsSNARF(PL_rs))
1004 return 0;
1005 if (RsRECORD(PL_rs))
1006 return 0;
1007 count = 0;
1008 if (SvTYPE(sv) == SVt_PVAV) {
1009 register I32 i;
1010 I32 max;
1011 AV* av = (AV*)sv;
1012 max = AvFILL(av);
1013 for (i = 0; i <= max; i++) {
1014 sv = (SV*)av_fetch(av, i, FALSE);
1015 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
1016 count += do_chomp(sv);
1017 }
1018 return count;
1019 }
1020 else if (SvTYPE(sv) == SVt_PVHV) {
1021 HV* hv = (HV*)sv;
1022 HE* entry;
1023 (void)hv_iterinit(hv);
1024 /*SUPPRESS 560*/
1025 while ((entry = hv_iternext(hv)))
1026 count += do_chomp(hv_iterval(hv,entry));
1027 return count;
1028 }
1029 else if (SvREADONLY(sv)) {
1030 if (SvFAKE(sv)) {
1031 /* SV is copy-on-write */
1032 sv_force_normal_flags(sv, 0);
1033 }
1034 if (SvREADONLY(sv))
1035 Perl_croak(aTHX_ PL_no_modify);
1036 }
1037 s = SvPV(sv, len);
1038 if (s && len) {
1039 s += --len;
1040 if (RsPARA(PL_rs)) {
1041 if (*s != '\n')
1042 goto nope;
1043 ++count;
1044 while (len && s[-1] == '\n') {
1045 --len;
1046 --s;
1047 ++count;
1048 }
1049 }
1050 else {
1051 STRLEN rslen;
1052 char *rsptr = SvPV(PL_rs, rslen);
1053 if (rslen == 1) {
1054 if (*s != *rsptr)
1055 goto nope;
1056 ++count;
1057 }
1058 else {
1059 if (len < rslen - 1)
1060 goto nope;
1061 len -= rslen - 1;
1062 s -= rslen - 1;
1063 if (memNE(s, rsptr, rslen))
1064 goto nope;
1065 count += rslen;
1066 }
1067 }
1068 s = SvPV_force(sv, n_a);
1069 SvCUR_set(sv, len);
1070 *SvEND(sv) = '\0';
1071 SvNIOK_off(sv);
1072 SvSETMAGIC(sv);
1073 }
1074 nope:
1075 return count;
1076}
1077
1078void
1079Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
1080{
1081#ifdef LIBERAL
1082 register long *dl;
1083 register long *ll;
1084 register long *rl;
1085#endif
1086 register char *dc;
1087 STRLEN leftlen;
1088 STRLEN rightlen;
1089 register char *lc;
1090 register char *rc;
1091 register I32 len;
1092 I32 lensave;
1093 char *lsave;
1094 char *rsave;
1095 bool left_utf = DO_UTF8(left);
1096 bool right_utf = DO_UTF8(right);
1097 I32 needlen = 0;
1098
1099 if (left_utf && !right_utf)
1100 sv_utf8_upgrade(right);
1101 else if (!left_utf && right_utf)
1102 sv_utf8_upgrade(left);
1103
1104 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1105 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
1106 lsave = lc = SvPV(left, leftlen);
1107 rsave = rc = SvPV(right, rightlen);
1108 len = leftlen < rightlen ? leftlen : rightlen;
1109 lensave = len;
1110 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1111 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1112 Newz(801, dc, needlen + 1, char);
1113 }
1114 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
1115 STRLEN n_a;
1116 dc = SvPV_force(sv, n_a);
1117 if (SvCUR(sv) < len) {
1118 dc = SvGROW(sv, len + 1);
1119 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1120 }
1121 if (optype != OP_BIT_AND && (left_utf || right_utf))
1122 dc = SvGROW(sv, leftlen + rightlen + 1);
1123 }
1124 else {
1125 needlen = ((optype == OP_BIT_AND)
1126 ? len : (leftlen > rightlen ? leftlen : rightlen));
1127 Newz(801, dc, needlen + 1, char);
1128 (void)sv_usepvn(sv, dc, needlen);
1129 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
1130 }
1131 SvCUR_set(sv, len);
1132 (void)SvPOK_only(sv);
1133 if (left_utf || right_utf) {
1134 UV duc, luc, ruc;
1135 char *dcsave = dc;
1136 STRLEN lulen = leftlen;
1137 STRLEN rulen = rightlen;
1138 STRLEN ulen;
1139
1140 switch (optype) {
1141 case OP_BIT_AND:
1142 while (lulen && rulen) {
1143 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1144 lc += ulen;
1145 lulen -= ulen;
1146 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1147 rc += ulen;
1148 rulen -= ulen;
1149 duc = luc & ruc;
1150 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1151 }
1152 if (sv == left || sv == right)
1153 (void)sv_usepvn(sv, dcsave, needlen);
1154 SvCUR_set(sv, dc - dcsave);
1155 break;
1156 case OP_BIT_XOR:
1157 while (lulen && rulen) {
1158 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1159 lc += ulen;
1160 lulen -= ulen;
1161 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1162 rc += ulen;
1163 rulen -= ulen;
1164 duc = luc ^ ruc;
1165 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1166 }
1167 goto mop_up_utf;
1168 case OP_BIT_OR:
1169 while (lulen && rulen) {
1170 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1171 lc += ulen;
1172 lulen -= ulen;
1173 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1174 rc += ulen;
1175 rulen -= ulen;
1176 duc = luc | ruc;
1177 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1178 }
1179 mop_up_utf:
1180 if (sv == left || sv == right)
1181 (void)sv_usepvn(sv, dcsave, needlen);
1182 SvCUR_set(sv, dc - dcsave);
1183 if (rulen)
1184 sv_catpvn(sv, rc, rulen);
1185 else if (lulen)
1186 sv_catpvn(sv, lc, lulen);
1187 else
1188 *SvEND(sv) = '\0';
1189 break;
1190 }
1191 SvUTF8_on(sv);
1192 goto finish;
1193 }
1194 else
1195#ifdef LIBERAL
1196 if (len >= sizeof(long)*4 &&
1197 !((long)dc % sizeof(long)) &&
1198 !((long)lc % sizeof(long)) &&
1199 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1200 {
1201 I32 remainder = len % (sizeof(long)*4);
1202 len /= (sizeof(long)*4);
1203
1204 dl = (long*)dc;
1205 ll = (long*)lc;
1206 rl = (long*)rc;
1207
1208 switch (optype) {
1209 case OP_BIT_AND:
1210 while (len--) {
1211 *dl++ = *ll++ & *rl++;
1212 *dl++ = *ll++ & *rl++;
1213 *dl++ = *ll++ & *rl++;
1214 *dl++ = *ll++ & *rl++;
1215 }
1216 break;
1217 case OP_BIT_XOR:
1218 while (len--) {
1219 *dl++ = *ll++ ^ *rl++;
1220 *dl++ = *ll++ ^ *rl++;
1221 *dl++ = *ll++ ^ *rl++;
1222 *dl++ = *ll++ ^ *rl++;
1223 }
1224 break;
1225 case OP_BIT_OR:
1226 while (len--) {
1227 *dl++ = *ll++ | *rl++;
1228 *dl++ = *ll++ | *rl++;
1229 *dl++ = *ll++ | *rl++;
1230 *dl++ = *ll++ | *rl++;
1231 }
1232 }
1233
1234 dc = (char*)dl;
1235 lc = (char*)ll;
1236 rc = (char*)rl;
1237
1238 len = remainder;
1239 }
1240#endif
1241 {
1242 switch (optype) {
1243 case OP_BIT_AND:
1244 while (len--)
1245 *dc++ = *lc++ & *rc++;
1246 break;
1247 case OP_BIT_XOR:
1248 while (len--)
1249 *dc++ = *lc++ ^ *rc++;
1250 goto mop_up;
1251 case OP_BIT_OR:
1252 while (len--)
1253 *dc++ = *lc++ | *rc++;
1254 mop_up:
1255 len = lensave;
1256 if (rightlen > len)
1257 sv_catpvn(sv, rsave + len, rightlen - len);
1258 else if (leftlen > len)
1259 sv_catpvn(sv, lsave + len, leftlen - len);
1260 else
1261 *SvEND(sv) = '\0';
1262 break;
1263 }
1264 }
1265finish:
1266 SvTAINT(sv);
1267}
1268
1269OP *
1270Perl_do_kv(pTHX)
1271{
1272 dSP;
1273 HV *hv = (HV*)POPs;
1274 HV *keys;
1275 register HE *entry;
1276 SV *tmpstr;
1277 I32 gimme = GIMME_V;
1278 I32 dokeys = (PL_op->op_type == OP_KEYS);
1279 I32 dovalues = (PL_op->op_type == OP_VALUES);
1280 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1281
1282 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1283 dokeys = dovalues = TRUE;
1284
1285 if (!hv) {
1286 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1287 dTARGET; /* make sure to clear its target here */
1288 if (SvTYPE(TARG) == SVt_PVLV)
1289 LvTARG(TARG) = Nullsv;
1290 PUSHs(TARG);
1291 }
1292 RETURN;
1293 }
1294
1295 keys = realhv ? hv : avhv_keys((AV*)hv);
1296 (void)hv_iterinit(keys); /* always reset iterator regardless */
1297
1298 if (gimme == G_VOID)
1299 RETURN;
1300
1301 if (gimme == G_SCALAR) {
1302 IV i;
1303 dTARGET;
1304
1305 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1306 if (SvTYPE(TARG) < SVt_PVLV) {
1307 sv_upgrade(TARG, SVt_PVLV);
1308 sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
1309 }
1310 LvTYPE(TARG) = 'k';
1311 if (LvTARG(TARG) != (SV*)keys) {
1312 if (LvTARG(TARG))
1313 SvREFCNT_dec(LvTARG(TARG));
1314 LvTARG(TARG) = SvREFCNT_inc(keys);
1315 }
1316 PUSHs(TARG);
1317 RETURN;
1318 }
1319
1320 if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied))
1321 i = HvKEYS(keys);
1322 else {
1323 i = 0;
1324 /*SUPPRESS 560*/
1325 while (hv_iternext(keys)) i++;
1326 }
1327 PUSHi( i );
1328 RETURN;
1329 }
1330
1331 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1332
1333 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1334 while ((entry = hv_iternext(keys))) {
1335 SPAGAIN;
1336 if (dokeys)
1337 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1338 if (dovalues) {
1339 PUTBACK;
1340 tmpstr = realhv ?
1341 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1342 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1343 (unsigned long)HeHASH(entry),
1344 HvMAX(keys)+1,
1345 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
1346 SPAGAIN;
1347 XPUSHs(tmpstr);
1348 }
1349 PUTBACK;
1350 }
1351 return NORMAL;
1352}
1353