This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Prettyprinting.
[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) && !(PL_op->op_private & OPpTRANS_IDENTICAL))
603 Perl_croak(aTHX_ PL_no_modify);
604
605 (void)SvPV(sv, len);
606 if (!len)
607 return 0;
608 if (!SvPOKp(sv))
609 (void)SvPV_force(sv, len);
610 if (!(PL_op->op_private & OPpTRANS_IDENTICAL))
611 (void)SvPOK_only_UTF8(sv);
612
613 DEBUG_t( Perl_deb(aTHX_ "2.TBL\n"));
614
615 switch (PL_op->op_private & ~hasutf & 63) {
616 case 0:
617 if (hasutf)
618 return do_trans_simple_utf8(sv);
619 else
620 return do_trans_simple(sv);
621
622 case OPpTRANS_IDENTICAL:
623 case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT:
624 if (hasutf)
625 return do_trans_count_utf8(sv);
626 else
627 return do_trans_count(sv);
628
629 default:
630 if (hasutf)
631 return do_trans_complex_utf8(sv);
632 else
633 return do_trans_complex(sv);
634 }
635}
636
637void
638Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp)
639{
640 SV **oldmark = mark;
641 register I32 items = sp - mark;
642 register STRLEN len;
643 STRLEN delimlen;
644 STRLEN tmplen;
645
646 (void) SvPV(del, delimlen); /* stringify and get the delimlen */
647 /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */
648
649 mark++;
650 len = (items > 0 ? (delimlen * (items - 1) ) : 0);
651 (void)SvUPGRADE(sv, SVt_PV);
652 if (SvLEN(sv) < len + items) { /* current length is way too short */
653 while (items-- > 0) {
654 if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) {
655 SvPV(*mark, tmplen);
656 len += tmplen;
657 }
658 mark++;
659 }
660 SvGROW(sv, len + 1); /* so try to pre-extend */
661
662 mark = oldmark;
663 items = sp - mark;
664 ++mark;
665 }
666
667 sv_setpv(sv, "");
668 if (PL_tainting && SvMAGICAL(sv))
669 SvTAINTED_off(sv);
670
671 if (items-- > 0) {
672 if (*mark)
673 sv_catsv(sv, *mark);
674 mark++;
675 }
676
677 if (delimlen) {
678 for (; items > 0; items--,mark++) {
679 sv_catsv(sv,del);
680 sv_catsv(sv,*mark);
681 }
682 }
683 else {
684 for (; items > 0; items--,mark++)
685 sv_catsv(sv,*mark);
686 }
687 SvSETMAGIC(sv);
688}
689
690void
691Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg)
692{
693 STRLEN patlen;
694 char *pat = SvPV(*sarg, patlen);
695 bool do_taint = FALSE;
696
697 sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint);
698 SvSETMAGIC(sv);
699 if (do_taint)
700 SvTAINTED_on(sv);
701}
702
703/* currently converts input to bytes if possible, but doesn't sweat failure */
704UV
705Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size)
706{
707 STRLEN srclen, len;
708 unsigned char *s = (unsigned char *) SvPV(sv, srclen);
709 UV retnum = 0;
710
711 if (offset < 0)
712 return retnum;
713 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
714 Perl_croak(aTHX_ "Illegal number of bits in vec");
715
716 if (SvUTF8(sv))
717 (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE);
718
719 offset *= size; /* turn into bit offset */
720 len = (offset + size + 7) / 8; /* required number of bytes */
721 if (len > srclen) {
722 if (size <= 8)
723 retnum = 0;
724 else {
725 offset >>= 3; /* turn into byte offset */
726 if (size == 16) {
727 if (offset >= srclen)
728 retnum = 0;
729 else
730 retnum = (UV) s[offset] << 8;
731 }
732 else if (size == 32) {
733 if (offset >= srclen)
734 retnum = 0;
735 else if (offset + 1 >= srclen)
736 retnum =
737 ((UV) s[offset ] << 24);
738 else if (offset + 2 >= srclen)
739 retnum =
740 ((UV) s[offset ] << 24) +
741 ((UV) s[offset + 1] << 16);
742 else
743 retnum =
744 ((UV) s[offset ] << 24) +
745 ((UV) s[offset + 1] << 16) +
746 ( s[offset + 2] << 8);
747 }
748#ifdef UV_IS_QUAD
749 else if (size == 64) {
750 if (ckWARN(WARN_PORTABLE))
751 Perl_warner(aTHX_ WARN_PORTABLE,
752 "Bit vector size > 32 non-portable");
753 if (offset >= srclen)
754 retnum = 0;
755 else if (offset + 1 >= srclen)
756 retnum =
757 (UV) s[offset ] << 56;
758 else if (offset + 2 >= srclen)
759 retnum =
760 ((UV) s[offset ] << 56) +
761 ((UV) s[offset + 1] << 48);
762 else if (offset + 3 >= srclen)
763 retnum =
764 ((UV) s[offset ] << 56) +
765 ((UV) s[offset + 1] << 48) +
766 ((UV) s[offset + 2] << 40);
767 else if (offset + 4 >= srclen)
768 retnum =
769 ((UV) s[offset ] << 56) +
770 ((UV) s[offset + 1] << 48) +
771 ((UV) s[offset + 2] << 40) +
772 ((UV) s[offset + 3] << 32);
773 else if (offset + 5 >= srclen)
774 retnum =
775 ((UV) s[offset ] << 56) +
776 ((UV) s[offset + 1] << 48) +
777 ((UV) s[offset + 2] << 40) +
778 ((UV) s[offset + 3] << 32) +
779 ( s[offset + 4] << 24);
780 else if (offset + 6 >= srclen)
781 retnum =
782 ((UV) s[offset ] << 56) +
783 ((UV) s[offset + 1] << 48) +
784 ((UV) s[offset + 2] << 40) +
785 ((UV) s[offset + 3] << 32) +
786 ((UV) s[offset + 4] << 24) +
787 ((UV) s[offset + 5] << 16);
788 else
789 retnum =
790 ((UV) s[offset ] << 56) +
791 ((UV) s[offset + 1] << 48) +
792 ((UV) s[offset + 2] << 40) +
793 ((UV) s[offset + 3] << 32) +
794 ((UV) s[offset + 4] << 24) +
795 ((UV) s[offset + 5] << 16) +
796 ( s[offset + 6] << 8);
797 }
798#endif
799 }
800 }
801 else if (size < 8)
802 retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1);
803 else {
804 offset >>= 3; /* turn into byte offset */
805 if (size == 8)
806 retnum = s[offset];
807 else if (size == 16)
808 retnum =
809 ((UV) s[offset] << 8) +
810 s[offset + 1];
811 else if (size == 32)
812 retnum =
813 ((UV) s[offset ] << 24) +
814 ((UV) s[offset + 1] << 16) +
815 ( s[offset + 2] << 8) +
816 s[offset + 3];
817#ifdef UV_IS_QUAD
818 else if (size == 64) {
819 if (ckWARN(WARN_PORTABLE))
820 Perl_warner(aTHX_ WARN_PORTABLE,
821 "Bit vector size > 32 non-portable");
822 retnum =
823 ((UV) s[offset ] << 56) +
824 ((UV) s[offset + 1] << 48) +
825 ((UV) s[offset + 2] << 40) +
826 ((UV) s[offset + 3] << 32) +
827 ((UV) s[offset + 4] << 24) +
828 ((UV) s[offset + 5] << 16) +
829 ( s[offset + 6] << 8) +
830 s[offset + 7];
831 }
832#endif
833 }
834
835 return retnum;
836}
837
838/* currently converts input to bytes if possible but doesn't sweat failures,
839 * although it does ensure that the string it clobbers is not marked as
840 * utf8-valid any more
841 */
842void
843Perl_do_vecset(pTHX_ SV *sv)
844{
845 SV *targ = LvTARG(sv);
846 register I32 offset;
847 register I32 size;
848 register unsigned char *s;
849 register UV lval;
850 I32 mask;
851 STRLEN targlen;
852 STRLEN len;
853
854 if (!targ)
855 return;
856 s = (unsigned char*)SvPV_force(targ, targlen);
857 if (SvUTF8(targ)) {
858 /* This is handled by the SvPOK_only below...
859 if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE))
860 SvUTF8_off(targ);
861 */
862 (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE);
863 }
864
865 (void)SvPOK_only(targ);
866 lval = SvUV(sv);
867 offset = LvTARGOFF(sv);
868 if (offset < 0)
869 Perl_croak(aTHX_ "Assigning to negative offset in vec");
870 size = LvTARGLEN(sv);
871 if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */
872 Perl_croak(aTHX_ "Illegal number of bits in vec");
873
874 offset *= size; /* turn into bit offset */
875 len = (offset + size + 7) / 8; /* required number of bytes */
876 if (len > targlen) {
877 s = (unsigned char*)SvGROW(targ, len + 1);
878 (void)memzero((char *)(s + targlen), len - targlen + 1);
879 SvCUR_set(targ, len);
880 }
881
882 if (size < 8) {
883 mask = (1 << size) - 1;
884 size = offset & 7;
885 lval &= mask;
886 offset >>= 3; /* turn into byte offset */
887 s[offset] &= ~(mask << size);
888 s[offset] |= lval << size;
889 }
890 else {
891 offset >>= 3; /* turn into byte offset */
892 if (size == 8)
893 s[offset ] = lval & 0xff;
894 else if (size == 16) {
895 s[offset ] = (lval >> 8) & 0xff;
896 s[offset+1] = lval & 0xff;
897 }
898 else if (size == 32) {
899 s[offset ] = (lval >> 24) & 0xff;
900 s[offset+1] = (lval >> 16) & 0xff;
901 s[offset+2] = (lval >> 8) & 0xff;
902 s[offset+3] = lval & 0xff;
903 }
904#ifdef UV_IS_QUAD
905 else if (size == 64) {
906 if (ckWARN(WARN_PORTABLE))
907 Perl_warner(aTHX_ WARN_PORTABLE,
908 "Bit vector size > 32 non-portable");
909 s[offset ] = (lval >> 56) & 0xff;
910 s[offset+1] = (lval >> 48) & 0xff;
911 s[offset+2] = (lval >> 40) & 0xff;
912 s[offset+3] = (lval >> 32) & 0xff;
913 s[offset+4] = (lval >> 24) & 0xff;
914 s[offset+5] = (lval >> 16) & 0xff;
915 s[offset+6] = (lval >> 8) & 0xff;
916 s[offset+7] = lval & 0xff;
917 }
918#endif
919 }
920 SvSETMAGIC(targ);
921}
922
923void
924Perl_do_chop(pTHX_ register SV *astr, register SV *sv)
925{
926 STRLEN len;
927 char *s;
928
929 if (SvTYPE(sv) == SVt_PVAV) {
930 register I32 i;
931 I32 max;
932 AV* av = (AV*)sv;
933 max = AvFILL(av);
934 for (i = 0; i <= max; i++) {
935 sv = (SV*)av_fetch(av, i, FALSE);
936 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
937 do_chop(astr, sv);
938 }
939 return;
940 }
941 else if (SvTYPE(sv) == SVt_PVHV) {
942 HV* hv = (HV*)sv;
943 HE* entry;
944 (void)hv_iterinit(hv);
945 /*SUPPRESS 560*/
946 while ((entry = hv_iternext(hv)))
947 do_chop(astr,hv_iterval(hv,entry));
948 return;
949 }
950 else if (SvREADONLY(sv))
951 Perl_croak(aTHX_ PL_no_modify);
952 s = SvPV(sv, len);
953 if (len && !SvPOK(sv))
954 s = SvPV_force(sv, len);
955 if (DO_UTF8(sv)) {
956 if (s && len) {
957 char *send = s + len;
958 char *start = s;
959 s = send - 1;
960 while (s > start && UTF8_IS_CONTINUATION(*s))
961 s--;
962 if (utf8_to_uvchr((U8*)s, 0)) {
963 sv_setpvn(astr, s, send - s);
964 *s = '\0';
965 SvCUR_set(sv, s - start);
966 SvNIOK_off(sv);
967 SvUTF8_on(astr);
968 }
969 }
970 else
971 sv_setpvn(astr, "", 0);
972 }
973 else if (s && len) {
974 s += --len;
975 sv_setpvn(astr, s, 1);
976 *s = '\0';
977 SvCUR_set(sv, len);
978 SvUTF8_off(sv);
979 SvNIOK_off(sv);
980 }
981 else
982 sv_setpvn(astr, "", 0);
983 SvSETMAGIC(sv);
984}
985
986I32
987Perl_do_chomp(pTHX_ register SV *sv)
988{
989 register I32 count;
990 STRLEN len;
991 STRLEN n_a;
992 char *s;
993
994 if (RsSNARF(PL_rs))
995 return 0;
996 if (RsRECORD(PL_rs))
997 return 0;
998 count = 0;
999 if (SvTYPE(sv) == SVt_PVAV) {
1000 register I32 i;
1001 I32 max;
1002 AV* av = (AV*)sv;
1003 max = AvFILL(av);
1004 for (i = 0; i <= max; i++) {
1005 sv = (SV*)av_fetch(av, i, FALSE);
1006 if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef))
1007 count += do_chomp(sv);
1008 }
1009 return count;
1010 }
1011 else if (SvTYPE(sv) == SVt_PVHV) {
1012 HV* hv = (HV*)sv;
1013 HE* entry;
1014 (void)hv_iterinit(hv);
1015 /*SUPPRESS 560*/
1016 while ((entry = hv_iternext(hv)))
1017 count += do_chomp(hv_iterval(hv,entry));
1018 return count;
1019 }
1020 else if (SvREADONLY(sv))
1021 Perl_croak(aTHX_ PL_no_modify);
1022 s = SvPV(sv, len);
1023 if (s && len) {
1024 s += --len;
1025 if (RsPARA(PL_rs)) {
1026 if (*s != '\n')
1027 goto nope;
1028 ++count;
1029 while (len && s[-1] == '\n') {
1030 --len;
1031 --s;
1032 ++count;
1033 }
1034 }
1035 else {
1036 STRLEN rslen;
1037 char *rsptr = SvPV(PL_rs, rslen);
1038 if (rslen == 1) {
1039 if (*s != *rsptr)
1040 goto nope;
1041 ++count;
1042 }
1043 else {
1044 if (len < rslen - 1)
1045 goto nope;
1046 len -= rslen - 1;
1047 s -= rslen - 1;
1048 if (memNE(s, rsptr, rslen))
1049 goto nope;
1050 count += rslen;
1051 }
1052 }
1053 s = SvPV_force(sv, n_a);
1054 SvCUR_set(sv, len);
1055 *SvEND(sv) = '\0';
1056 SvNIOK_off(sv);
1057 SvSETMAGIC(sv);
1058 }
1059 nope:
1060 return count;
1061}
1062
1063void
1064Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
1065{
1066#ifdef LIBERAL
1067 register long *dl;
1068 register long *ll;
1069 register long *rl;
1070#endif
1071 register char *dc;
1072 STRLEN leftlen;
1073 STRLEN rightlen;
1074 register char *lc;
1075 register char *rc;
1076 register I32 len;
1077 I32 lensave;
1078 char *lsave;
1079 char *rsave;
1080 bool left_utf = DO_UTF8(left);
1081 bool right_utf = DO_UTF8(right);
1082 I32 needlen = 0;
1083
1084 if (left_utf && !right_utf)
1085 sv_utf8_upgrade(right);
1086 else if (!left_utf && right_utf)
1087 sv_utf8_upgrade(left);
1088
1089 if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))
1090 sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */
1091 lsave = lc = SvPV(left, leftlen);
1092 rsave = rc = SvPV(right, rightlen);
1093 len = leftlen < rightlen ? leftlen : rightlen;
1094 lensave = len;
1095 if ((left_utf || right_utf) && (sv == left || sv == right)) {
1096 needlen = optype == OP_BIT_AND ? len : leftlen + rightlen;
1097 Newz(801, dc, needlen + 1, char);
1098 }
1099 else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) {
1100 STRLEN n_a;
1101 dc = SvPV_force(sv, n_a);
1102 if (SvCUR(sv) < len) {
1103 dc = SvGROW(sv, len + 1);
1104 (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1);
1105 }
1106 if (optype != OP_BIT_AND && (left_utf || right_utf))
1107 dc = SvGROW(sv, leftlen + rightlen + 1);
1108 }
1109 else {
1110 needlen = ((optype == OP_BIT_AND)
1111 ? len : (leftlen > rightlen ? leftlen : rightlen));
1112 Newz(801, dc, needlen + 1, char);
1113 (void)sv_usepvn(sv, dc, needlen);
1114 dc = SvPVX(sv); /* sv_usepvn() calls Renew() */
1115 }
1116 SvCUR_set(sv, len);
1117 (void)SvPOK_only(sv);
1118 if (left_utf || right_utf) {
1119 UV duc, luc, ruc;
1120 char *dcsave = dc;
1121 STRLEN lulen = leftlen;
1122 STRLEN rulen = rightlen;
1123 STRLEN ulen;
1124
1125 switch (optype) {
1126 case OP_BIT_AND:
1127 while (lulen && rulen) {
1128 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1129 lc += ulen;
1130 lulen -= ulen;
1131 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1132 rc += ulen;
1133 rulen -= ulen;
1134 duc = luc & ruc;
1135 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1136 }
1137 if (sv == left || sv == right)
1138 (void)sv_usepvn(sv, dcsave, needlen);
1139 SvCUR_set(sv, dc - dcsave);
1140 break;
1141 case OP_BIT_XOR:
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 goto mop_up_utf;
1153 case OP_BIT_OR:
1154 while (lulen && rulen) {
1155 luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV);
1156 lc += ulen;
1157 lulen -= ulen;
1158 ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV);
1159 rc += ulen;
1160 rulen -= ulen;
1161 duc = luc | ruc;
1162 dc = (char*)uvchr_to_utf8((U8*)dc, duc);
1163 }
1164 mop_up_utf:
1165 if (sv == left || sv == right)
1166 (void)sv_usepvn(sv, dcsave, needlen);
1167 SvCUR_set(sv, dc - dcsave);
1168 if (rulen)
1169 sv_catpvn(sv, rc, rulen);
1170 else if (lulen)
1171 sv_catpvn(sv, lc, lulen);
1172 else
1173 *SvEND(sv) = '\0';
1174 break;
1175 }
1176 SvUTF8_on(sv);
1177 goto finish;
1178 }
1179 else
1180#ifdef LIBERAL
1181 if (len >= sizeof(long)*4 &&
1182 !((long)dc % sizeof(long)) &&
1183 !((long)lc % sizeof(long)) &&
1184 !((long)rc % sizeof(long))) /* It's almost always aligned... */
1185 {
1186 I32 remainder = len % (sizeof(long)*4);
1187 len /= (sizeof(long)*4);
1188
1189 dl = (long*)dc;
1190 ll = (long*)lc;
1191 rl = (long*)rc;
1192
1193 switch (optype) {
1194 case OP_BIT_AND:
1195 while (len--) {
1196 *dl++ = *ll++ & *rl++;
1197 *dl++ = *ll++ & *rl++;
1198 *dl++ = *ll++ & *rl++;
1199 *dl++ = *ll++ & *rl++;
1200 }
1201 break;
1202 case OP_BIT_XOR:
1203 while (len--) {
1204 *dl++ = *ll++ ^ *rl++;
1205 *dl++ = *ll++ ^ *rl++;
1206 *dl++ = *ll++ ^ *rl++;
1207 *dl++ = *ll++ ^ *rl++;
1208 }
1209 break;
1210 case OP_BIT_OR:
1211 while (len--) {
1212 *dl++ = *ll++ | *rl++;
1213 *dl++ = *ll++ | *rl++;
1214 *dl++ = *ll++ | *rl++;
1215 *dl++ = *ll++ | *rl++;
1216 }
1217 }
1218
1219 dc = (char*)dl;
1220 lc = (char*)ll;
1221 rc = (char*)rl;
1222
1223 len = remainder;
1224 }
1225#endif
1226 {
1227 switch (optype) {
1228 case OP_BIT_AND:
1229 while (len--)
1230 *dc++ = *lc++ & *rc++;
1231 break;
1232 case OP_BIT_XOR:
1233 while (len--)
1234 *dc++ = *lc++ ^ *rc++;
1235 goto mop_up;
1236 case OP_BIT_OR:
1237 while (len--)
1238 *dc++ = *lc++ | *rc++;
1239 mop_up:
1240 len = lensave;
1241 if (rightlen > len)
1242 sv_catpvn(sv, rsave + len, rightlen - len);
1243 else if (leftlen > len)
1244 sv_catpvn(sv, lsave + len, leftlen - len);
1245 else
1246 *SvEND(sv) = '\0';
1247 break;
1248 }
1249 }
1250finish:
1251 SvTAINT(sv);
1252}
1253
1254OP *
1255Perl_do_kv(pTHX)
1256{
1257 dSP;
1258 HV *hv = (HV*)POPs;
1259 HV *keys;
1260 register HE *entry;
1261 SV *tmpstr;
1262 I32 gimme = GIMME_V;
1263 I32 dokeys = (PL_op->op_type == OP_KEYS);
1264 I32 dovalues = (PL_op->op_type == OP_VALUES);
1265 I32 realhv = (SvTYPE(hv) == SVt_PVHV);
1266
1267 if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV)
1268 dokeys = dovalues = TRUE;
1269
1270 if (!hv) {
1271 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1272 dTARGET; /* make sure to clear its target here */
1273 if (SvTYPE(TARG) == SVt_PVLV)
1274 LvTARG(TARG) = Nullsv;
1275 PUSHs(TARG);
1276 }
1277 RETURN;
1278 }
1279
1280 keys = realhv ? hv : avhv_keys((AV*)hv);
1281 (void)hv_iterinit(keys); /* always reset iterator regardless */
1282
1283 if (gimme == G_VOID)
1284 RETURN;
1285
1286 if (gimme == G_SCALAR) {
1287 IV i;
1288 dTARGET;
1289
1290 if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */
1291 if (SvTYPE(TARG) < SVt_PVLV) {
1292 sv_upgrade(TARG, SVt_PVLV);
1293 sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0);
1294 }
1295 LvTYPE(TARG) = 'k';
1296 if (LvTARG(TARG) != (SV*)keys) {
1297 if (LvTARG(TARG))
1298 SvREFCNT_dec(LvTARG(TARG));
1299 LvTARG(TARG) = SvREFCNT_inc(keys);
1300 }
1301 PUSHs(TARG);
1302 RETURN;
1303 }
1304
1305 if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied))
1306 i = HvKEYS(keys);
1307 else {
1308 i = 0;
1309 /*SUPPRESS 560*/
1310 while (hv_iternext(keys)) i++;
1311 }
1312 PUSHi( i );
1313 RETURN;
1314 }
1315
1316 EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues));
1317
1318 PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */
1319 while ((entry = hv_iternext(keys))) {
1320 SPAGAIN;
1321 if (dokeys)
1322 XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */
1323 if (dovalues) {
1324 PUTBACK;
1325 tmpstr = realhv ?
1326 hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry);
1327 DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu",
1328 (unsigned long)HeHASH(entry),
1329 HvMAX(keys)+1,
1330 (unsigned long)(HeHASH(entry) & HvMAX(keys))));
1331 SPAGAIN;
1332 XPUSHs(tmpstr);
1333 }
1334 PUTBACK;
1335 }
1336 return NORMAL;
1337}
1338