Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
9607fc9c |
3 | * Copyright (c) 1991-1997, Larry Wall |
79072805 |
4 | * |
5 | * You may distribute under the terms of either the GNU General Public |
6 | * License or the Artistic License, as specified in the README file. |
7 | * |
a0d0e21e |
8 | */ |
9 | |
10 | /* |
11 | * "'So that was the job I felt I had to do when I started,' thought Sam." |
79072805 |
12 | */ |
13 | |
14 | #include "EXTERN.h" |
15 | #include "perl.h" |
16 | |
17 | #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) |
18 | #include <signal.h> |
19 | #endif |
20 | |
4757a243 |
21 | static I32 |
22 | do_trans_CC_simple(SV *sv) |
79072805 |
23 | { |
11343788 |
24 | dTHR; |
4757a243 |
25 | U8 *s; |
26 | U8 *send; |
27 | I32 matches = 0; |
463ee0b2 |
28 | STRLEN len; |
4757a243 |
29 | short *tbl; |
30 | I32 ch; |
79072805 |
31 | |
4757a243 |
32 | tbl = (short*)cPVOP->op_pv; |
33 | if (!tbl) |
34 | croak("panic: do_trans"); |
a0ed51b3 |
35 | |
4757a243 |
36 | s = (U8*)SvPV(sv, len); |
37 | send = s + len; |
38 | |
39 | while (s < send) { |
40 | if ((ch = tbl[*s]) >= 0) { |
41 | matches++; |
42 | *s = ch; |
79072805 |
43 | } |
4757a243 |
44 | s++; |
45 | } |
46 | SvSETMAGIC(sv); |
47 | |
48 | return matches; |
49 | } |
50 | |
51 | static I32 |
52 | do_trans_CC_count(SV *sv) |
53 | { |
54 | dTHR; |
55 | U8 *s; |
56 | U8 *send; |
57 | I32 matches = 0; |
58 | STRLEN len; |
59 | short *tbl; |
60 | |
61 | tbl = (short*)cPVOP->op_pv; |
62 | if (!tbl) |
63 | croak("panic: do_trans"); |
64 | |
65 | s = (U8*)SvPV(sv, len); |
66 | send = s + len; |
67 | |
68 | while (s < send) { |
69 | if (tbl[*s] >= 0) |
70 | matches++; |
71 | s++; |
72 | } |
73 | |
74 | return matches; |
75 | } |
76 | |
77 | static I32 |
78 | do_trans_CC_complex(SV *sv) |
79 | { |
80 | dTHR; |
81 | U8 *s; |
82 | U8 *send; |
83 | U8 *d; |
84 | I32 matches = 0; |
85 | STRLEN len; |
86 | short *tbl; |
87 | I32 ch; |
88 | |
89 | tbl = (short*)cPVOP->op_pv; |
90 | if (!tbl) |
91 | croak("panic: do_trans"); |
92 | |
93 | s = (U8*)SvPV(sv, len); |
94 | send = s + len; |
95 | |
96 | d = s; |
97 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
98 | U8* p = send; |
99 | |
100 | while (s < send) { |
101 | if ((ch = tbl[*s]) >= 0) { |
102 | *d = ch; |
103 | matches++; |
104 | if (p == d - 1 && *p == *d) |
105 | matches--; |
a0ed51b3 |
106 | else |
4757a243 |
107 | p = d++; |
a0ed51b3 |
108 | } |
4757a243 |
109 | else if (ch == -1) /* -1 is unmapped character */ |
110 | *d++ = *s; /* -2 is delete character */ |
111 | s++; |
a0ed51b3 |
112 | } |
4757a243 |
113 | } |
114 | else { |
115 | while (s < send) { |
116 | if ((ch = tbl[*s]) >= 0) { |
117 | *d = ch; |
118 | matches++; |
119 | d++; |
a0ed51b3 |
120 | } |
4757a243 |
121 | else if (ch == -1) /* -1 is unmapped character */ |
122 | *d++ = *s; /* -2 is delete character */ |
123 | s++; |
5d06d08e |
124 | } |
4757a243 |
125 | } |
126 | matches += send - d; /* account for disappeared chars */ |
127 | *d = '\0'; |
128 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); |
129 | SvSETMAGIC(sv); |
130 | |
131 | return matches; |
132 | } |
133 | |
134 | static I32 |
135 | do_trans_UU_simple(SV *sv) |
136 | { |
137 | dTHR; |
138 | U8 *s; |
139 | U8 *send; |
140 | U8 *d; |
141 | I32 matches = 0; |
142 | STRLEN len; |
143 | |
144 | SV* rv = (SV*)cSVOP->op_sv; |
145 | HV* hv = (HV*)SvRV(rv); |
146 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
147 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
148 | UV extra = none + 1; |
149 | UV final; |
150 | UV uv; |
151 | |
152 | s = (U8*)SvPV(sv, len); |
153 | send = s + len; |
154 | |
155 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
156 | if (svp) |
157 | final = SvUV(*svp); |
158 | |
159 | d = s; |
160 | while (s < send) { |
161 | if ((uv = swash_fetch(rv, s)) < none) { |
162 | s += UTF8SKIP(s); |
163 | matches++; |
164 | d = uv_to_utf8(d, uv); |
165 | } |
166 | else if (uv == none) { |
167 | int i; |
168 | for (i = UTF8SKIP(s); i; i--) |
169 | *d++ = *s++; |
170 | } |
171 | else if (uv == extra) { |
172 | s += UTF8SKIP(s); |
173 | matches++; |
174 | d = uv_to_utf8(d, final); |
175 | } |
176 | else |
177 | s += UTF8SKIP(s); |
178 | } |
179 | *d = '\0'; |
180 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); |
181 | SvSETMAGIC(sv); |
182 | |
183 | return matches; |
184 | } |
185 | |
186 | static I32 |
187 | do_trans_UU_count(SV *sv) |
188 | { |
189 | dTHR; |
190 | U8 *s; |
191 | U8 *send; |
192 | I32 matches = 0; |
193 | STRLEN len; |
194 | |
195 | SV* rv = (SV*)cSVOP->op_sv; |
196 | HV* hv = (HV*)SvRV(rv); |
197 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
198 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
199 | UV uv; |
200 | |
201 | s = (U8*)SvPV(sv, len); |
202 | send = s + len; |
203 | |
204 | while (s < send) { |
205 | if ((uv = swash_fetch(rv, s)) < none) { |
206 | s += UTF8SKIP(s); |
207 | matches++; |
208 | } |
209 | } |
210 | |
211 | return matches; |
212 | } |
213 | |
214 | static I32 |
215 | do_trans_UC_simple(SV *sv) |
216 | { |
217 | dTHR; |
218 | U8 *s; |
219 | U8 *send; |
220 | U8 *d; |
221 | I32 matches = 0; |
222 | STRLEN len; |
223 | |
224 | SV* rv = (SV*)cSVOP->op_sv; |
225 | HV* hv = (HV*)SvRV(rv); |
226 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
227 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
228 | UV extra = none + 1; |
229 | UV final; |
230 | UV uv; |
231 | |
232 | s = (U8*)SvPV(sv, len); |
233 | send = s + len; |
234 | |
235 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
236 | if (svp) |
237 | final = SvUV(*svp); |
238 | |
239 | d = s; |
240 | while (s < send) { |
241 | if ((uv = swash_fetch(rv, s)) < none) { |
242 | s += UTF8SKIP(s); |
243 | matches++; |
244 | *d++ = (U8)uv; |
245 | } |
246 | else if (uv == none) { |
247 | I32 ulen; |
248 | uv = utf8_to_uv(s, &ulen); |
249 | s += ulen; |
250 | *d++ = (U8)uv; |
251 | } |
252 | else if (uv == extra) { |
253 | s += UTF8SKIP(s); |
254 | matches++; |
255 | *d++ = (U8)final; |
256 | } |
257 | else |
258 | s += UTF8SKIP(s); |
259 | } |
260 | *d = '\0'; |
261 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); |
262 | SvSETMAGIC(sv); |
263 | |
264 | return matches; |
265 | } |
266 | |
267 | static I32 |
268 | do_trans_CU_simple(SV *sv) |
269 | { |
270 | dTHR; |
271 | U8 *s; |
272 | U8 *send; |
273 | U8 *d; |
274 | U8 *dst; |
275 | I32 matches = 0; |
276 | STRLEN len; |
277 | |
278 | SV* rv = (SV*)cSVOP->op_sv; |
279 | HV* hv = (HV*)SvRV(rv); |
280 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
281 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
282 | UV extra = none + 1; |
283 | UV final; |
284 | UV uv; |
285 | U8 tmpbuf[10]; |
286 | I32 bits = 16; |
287 | |
288 | s = (U8*)SvPV(sv, len); |
289 | send = s + len; |
290 | |
291 | svp = hv_fetch(hv, "BITS", 4, FALSE); |
292 | if (svp) |
293 | bits = (I32)SvIV(*svp); |
294 | |
295 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
296 | if (svp) |
297 | final = SvUV(*svp); |
298 | |
299 | Newz(801, d, len * (bits >> 3) + 1, U8); |
300 | dst = d; |
301 | |
302 | while (s < send) { |
303 | uv = *s++; |
304 | if (uv < 0x80) |
305 | tmpbuf[0] = uv; |
306 | else { |
307 | tmpbuf[0] = (( uv >> 6) | 0xc0); |
308 | tmpbuf[1] = (( uv & 0x3f) | 0x80); |
a0ed51b3 |
309 | } |
4757a243 |
310 | |
311 | if ((uv = swash_fetch(rv, tmpbuf)) < none) { |
312 | matches++; |
313 | d = uv_to_utf8(d, uv); |
314 | } |
315 | else if (uv == none) |
316 | d = uv_to_utf8(d, s[-1]); |
317 | else if (uv == extra) { |
318 | matches++; |
319 | d = uv_to_utf8(d, final); |
320 | } |
321 | } |
322 | *d = '\0'; |
323 | sv_usepvn_mg(sv, (char*)dst, d - dst); |
324 | |
325 | return matches; |
326 | } |
327 | |
328 | /* utf-8 to latin-1 */ |
329 | |
330 | static I32 |
331 | do_trans_UC_trivial(SV *sv) |
332 | { |
333 | dTHR; |
334 | U8 *s; |
335 | U8 *send; |
336 | U8 *d; |
337 | STRLEN len; |
338 | |
339 | s = (U8*)SvPV(sv, len); |
340 | send = s + len; |
341 | |
342 | d = s; |
343 | while (s < send) { |
344 | if (*s < 0x80) |
345 | *d++ = *s++; |
a0ed51b3 |
346 | else { |
4757a243 |
347 | I32 ulen; |
348 | UV uv = utf8_to_uv(s, &ulen); |
349 | s += ulen; |
350 | *d++ = (U8)uv; |
351 | } |
352 | } |
353 | *d = '\0'; |
354 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); |
355 | SvSETMAGIC(sv); |
a0ed51b3 |
356 | |
4757a243 |
357 | return SvCUR(sv); |
358 | } |
a0ed51b3 |
359 | |
4757a243 |
360 | /* latin-1 to utf-8 */ |
a0ed51b3 |
361 | |
4757a243 |
362 | static I32 |
363 | do_trans_CU_trivial(SV *sv) |
364 | { |
365 | dTHR; |
366 | U8 *s; |
367 | U8 *send; |
368 | U8 *d; |
369 | U8 *dst; |
370 | I32 matches; |
371 | STRLEN len; |
a0ed51b3 |
372 | |
4757a243 |
373 | s = (U8*)SvPV(sv, len); |
374 | send = s + len; |
375 | |
376 | Newz(801, d, len * 2 + 1, U8); |
377 | dst = d; |
378 | |
379 | matches = send - s; |
380 | |
381 | while (s < send) { |
382 | if (*s < 0x80) |
383 | *d++ = *s++; |
384 | else { |
385 | UV uv = *s++; |
386 | *d++ = (( uv >> 6) | 0xc0); |
387 | *d++ = (( uv & 0x3f) | 0x80); |
388 | } |
389 | } |
390 | *d = '\0'; |
391 | sv_usepvn_mg(sv, (char*)dst, d - dst); |
392 | |
393 | return matches; |
394 | } |
395 | |
396 | static I32 |
397 | do_trans_UU_complex(SV *sv) |
398 | { |
399 | dTHR; |
400 | U8 *s; |
401 | U8 *send; |
402 | U8 *d; |
403 | I32 matches = 0; |
404 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
405 | I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF; |
406 | I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF; |
407 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
408 | SV* rv = (SV*)cSVOP->op_sv; |
409 | HV* hv = (HV*)SvRV(rv); |
410 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
411 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
412 | UV extra = none + 1; |
413 | UV final; |
414 | UV uv; |
415 | STRLEN len; |
416 | U8 *dst; |
417 | |
418 | s = (U8*)SvPV(sv, len); |
419 | send = s + len; |
420 | |
421 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
422 | if (svp) |
423 | final = SvUV(*svp); |
424 | |
425 | if (PL_op->op_private & OPpTRANS_GROWS) { |
426 | I32 bits = 16; |
427 | |
428 | svp = hv_fetch(hv, "BITS", 4, FALSE); |
429 | if (svp) |
430 | bits = (I32)SvIV(*svp); |
431 | |
432 | Newz(801, d, len * (bits >> 3) + 1, U8); |
433 | dst = d; |
434 | } |
435 | else { |
436 | d = s; |
437 | dst = 0; |
438 | } |
439 | |
440 | if (squash) { |
441 | UV puv = 0xfeedface; |
442 | while (s < send) { |
443 | if (from_utf) { |
444 | uv = swash_fetch(rv, s); |
a0ed51b3 |
445 | } |
446 | else { |
4757a243 |
447 | U8 tmpbuf[2]; |
448 | uv = *s++; |
449 | if (uv < 0x80) |
450 | tmpbuf[0] = uv; |
451 | else { |
452 | tmpbuf[0] = (( uv >> 6) | 0xc0); |
453 | tmpbuf[1] = (( uv & 0x3f) | 0x80); |
454 | } |
455 | uv = swash_fetch(rv, tmpbuf); |
456 | } |
457 | if (uv < none) { |
458 | matches++; |
459 | if (uv != puv) { |
460 | if (uv >= 0x80 && to_utf) |
461 | d = uv_to_utf8(d, uv); |
462 | else |
463 | *d++ = (U8)uv; |
464 | puv = uv; |
465 | } |
466 | if (from_utf) |
467 | s += UTF8SKIP(s); |
468 | continue; |
469 | } |
470 | else if (uv == none) { /* "none" is unmapped character */ |
471 | if (from_utf) { |
472 | if (*s < 0x80) |
473 | *d++ = *s++; |
474 | else if (to_utf) { |
a0ed51b3 |
475 | int i; |
4757a243 |
476 | for (i = UTF8SKIP(s); i; --i) |
477 | *d++ = *s++; |
a0ed51b3 |
478 | } |
4757a243 |
479 | else { |
480 | I32 ulen; |
481 | *d++ = (U8)utf8_to_uv(s, &ulen); |
482 | s += ulen; |
a0ed51b3 |
483 | } |
a0ed51b3 |
484 | } |
4757a243 |
485 | else { /* must be to_utf only */ |
486 | d = uv_to_utf8(d, s[-1]); |
487 | } |
488 | puv = 0xfeedface; |
489 | continue; |
a0ed51b3 |
490 | } |
4757a243 |
491 | else if (uv == extra && !del) { |
492 | matches++; |
493 | if (uv != puv) { |
494 | if (final >= 0x80 && to_utf) |
495 | d = uv_to_utf8(d, final); |
496 | else |
497 | *d++ = (U8)final; |
498 | puv = final; |
499 | } |
500 | if (from_utf) |
501 | s += UTF8SKIP(s); |
502 | continue; |
503 | } |
504 | matches++; /* "none+1" is delete character */ |
505 | if (from_utf) |
506 | s += UTF8SKIP(s); |
a0ed51b3 |
507 | } |
79072805 |
508 | } |
509 | else { |
4757a243 |
510 | while (s < send) { |
511 | if (from_utf) { |
512 | uv = swash_fetch(rv, s); |
513 | } |
514 | else { |
515 | U8 tmpbuf[2]; |
516 | uv = *s++; |
517 | if (uv < 0x80) |
518 | tmpbuf[0] = uv; |
519 | else { |
520 | tmpbuf[0] = (( uv >> 6) | 0xc0); |
521 | tmpbuf[1] = (( uv & 0x3f) | 0x80); |
a0ed51b3 |
522 | } |
4757a243 |
523 | uv = swash_fetch(rv, tmpbuf); |
a0ed51b3 |
524 | } |
4757a243 |
525 | if (uv < none) { |
526 | matches++; |
527 | if (uv >= 0x80 && to_utf) |
528 | d = uv_to_utf8(d, uv); |
529 | else |
530 | *d++ = (U8)uv; |
531 | if (from_utf) |
532 | s += UTF8SKIP(s); |
533 | continue; |
a0ed51b3 |
534 | } |
4757a243 |
535 | else if (uv == none) { /* "none" is unmapped character */ |
536 | if (from_utf) { |
537 | if (*s < 0x80) |
538 | *d++ = *s++; |
539 | else if (to_utf) { |
540 | int i; |
541 | for (i = UTF8SKIP(s); i; --i) |
542 | *d++ = *s++; |
543 | } |
544 | else { |
545 | I32 ulen; |
546 | *d++ = (U8)utf8_to_uv(s, &ulen); |
547 | s += ulen; |
a0ed51b3 |
548 | } |
79072805 |
549 | } |
4757a243 |
550 | else { /* must be to_utf only */ |
551 | d = uv_to_utf8(d, s[-1]); |
552 | } |
553 | continue; |
79072805 |
554 | } |
4757a243 |
555 | else if (uv == extra && !del) { |
556 | matches++; |
557 | if (final >= 0x80 && to_utf) |
558 | d = uv_to_utf8(d, final); |
559 | else |
560 | *d++ = (U8)final; |
561 | if (from_utf) |
562 | s += UTF8SKIP(s); |
563 | continue; |
564 | } |
565 | matches++; /* "none+1" is delete character */ |
566 | if (from_utf) |
567 | s += UTF8SKIP(s); |
79072805 |
568 | } |
4757a243 |
569 | } |
570 | if (dst) |
571 | sv_usepvn(sv, (char*)dst, d - dst); |
572 | else { |
573 | *d = '\0'; |
574 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); |
575 | } |
576 | SvSETMAGIC(sv); |
577 | |
578 | return matches; |
579 | } |
580 | |
581 | I32 |
582 | do_trans(SV *sv) |
583 | { |
584 | STRLEN len; |
585 | |
586 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
587 | croak(no_modify); |
588 | |
589 | (void)SvPV(sv, len); |
590 | if (!len) |
591 | return 0; |
592 | if (!SvPOKp(sv)) |
593 | (void)SvPV_force(sv, len); |
594 | (void)SvPOK_only(sv); |
595 | |
596 | DEBUG_t( deb("2.TBL\n")); |
597 | |
598 | switch (PL_op->op_private & 63) { |
599 | case 0: |
600 | return do_trans_CC_simple(sv); |
601 | |
602 | case OPpTRANS_FROM_UTF: |
603 | return do_trans_UC_simple(sv); |
604 | |
605 | case OPpTRANS_TO_UTF: |
606 | return do_trans_CU_simple(sv); |
607 | |
608 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF: |
609 | return do_trans_UU_simple(sv); |
610 | |
611 | case OPpTRANS_IDENTICAL: |
612 | return do_trans_CC_count(sv); |
613 | |
614 | case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL: |
615 | return do_trans_UC_trivial(sv); |
616 | |
617 | case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: |
618 | return do_trans_CU_trivial(sv); |
619 | |
620 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: |
621 | return do_trans_UU_count(sv); |
622 | |
623 | default: |
624 | if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)) |
625 | return do_trans_UU_complex(sv); /* could be UC or CU too */ |
626 | else |
627 | return do_trans_CC_complex(sv); |
79072805 |
628 | } |
79072805 |
629 | } |
630 | |
631 | void |
8ac85365 |
632 | do_join(register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
633 | { |
634 | SV **oldmark = mark; |
635 | register I32 items = sp - mark; |
79072805 |
636 | register STRLEN len; |
463ee0b2 |
637 | STRLEN delimlen; |
638 | register char *delim = SvPV(del, delimlen); |
639 | STRLEN tmplen; |
79072805 |
640 | |
641 | mark++; |
642 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
643 | if (SvTYPE(sv) < SVt_PV) |
644 | sv_upgrade(sv, SVt_PV); |
645 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
646 | while (items-- > 0) { |
48c036b1 |
647 | if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) { |
463ee0b2 |
648 | SvPV(*mark, tmplen); |
649 | len += tmplen; |
79072805 |
650 | } |
651 | mark++; |
652 | } |
653 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
654 | |
655 | mark = oldmark; |
656 | items = sp - mark;; |
657 | ++mark; |
658 | } |
659 | |
463ee0b2 |
660 | if (items-- > 0) { |
8990e307 |
661 | char *s; |
662 | |
663 | if (*mark) { |
664 | s = SvPV(*mark, tmplen); |
665 | sv_setpvn(sv, s, tmplen); |
666 | } |
667 | else |
668 | sv_setpv(sv, ""); |
463ee0b2 |
669 | mark++; |
670 | } |
79072805 |
671 | else |
672 | sv_setpv(sv,""); |
673 | len = delimlen; |
674 | if (len) { |
675 | for (; items > 0; items--,mark++) { |
676 | sv_catpvn(sv,delim,len); |
677 | sv_catsv(sv,*mark); |
678 | } |
679 | } |
680 | else { |
681 | for (; items > 0; items--,mark++) |
682 | sv_catsv(sv,*mark); |
683 | } |
684 | SvSETMAGIC(sv); |
685 | } |
686 | |
687 | void |
8ac85365 |
688 | do_sprintf(SV *sv, I32 len, SV **sarg) |
79072805 |
689 | { |
46fc3d4c |
690 | STRLEN patlen; |
691 | char *pat = SvPV(*sarg, patlen); |
692 | bool do_taint = FALSE; |
693 | |
694 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
695 | SvSETMAGIC(sv); |
46fc3d4c |
696 | if (do_taint) |
697 | SvTAINTED_on(sv); |
79072805 |
698 | } |
699 | |
700 | void |
8ac85365 |
701 | do_vecset(SV *sv) |
79072805 |
702 | { |
703 | SV *targ = LvTARG(sv); |
704 | register I32 offset; |
705 | register I32 size; |
8990e307 |
706 | register unsigned char *s; |
707 | register unsigned long lval; |
79072805 |
708 | I32 mask; |
a0d0e21e |
709 | STRLEN targlen; |
710 | STRLEN len; |
79072805 |
711 | |
8990e307 |
712 | if (!targ) |
713 | return; |
a0d0e21e |
714 | s = (unsigned char*)SvPV_force(targ, targlen); |
8990e307 |
715 | lval = U_L(SvNV(sv)); |
79072805 |
716 | offset = LvTARGOFF(sv); |
717 | size = LvTARGLEN(sv); |
a0d0e21e |
718 | |
719 | len = (offset + size + 7) / 8; |
720 | if (len > targlen) { |
721 | s = (unsigned char*)SvGROW(targ, len + 1); |
722 | (void)memzero(s + targlen, len - targlen + 1); |
723 | SvCUR_set(targ, len); |
724 | } |
725 | |
79072805 |
726 | if (size < 8) { |
727 | mask = (1 << size) - 1; |
728 | size = offset & 7; |
729 | lval &= mask; |
730 | offset >>= 3; |
731 | s[offset] &= ~(mask << size); |
732 | s[offset] |= lval << size; |
733 | } |
734 | else { |
a0d0e21e |
735 | offset >>= 3; |
79072805 |
736 | if (size == 8) |
737 | s[offset] = lval & 255; |
738 | else if (size == 16) { |
739 | s[offset] = (lval >> 8) & 255; |
740 | s[offset+1] = lval & 255; |
741 | } |
742 | else if (size == 32) { |
743 | s[offset] = (lval >> 24) & 255; |
744 | s[offset+1] = (lval >> 16) & 255; |
745 | s[offset+2] = (lval >> 8) & 255; |
746 | s[offset+3] = lval & 255; |
747 | } |
748 | } |
749 | } |
750 | |
751 | void |
8ac85365 |
752 | do_chop(register SV *astr, register SV *sv) |
79072805 |
753 | { |
463ee0b2 |
754 | STRLEN len; |
a0d0e21e |
755 | char *s; |
c485e607 |
756 | dTHR; |
a0d0e21e |
757 | |
79072805 |
758 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
759 | register I32 i; |
760 | I32 max; |
761 | AV* av = (AV*)sv; |
762 | max = AvFILL(av); |
763 | for (i = 0; i <= max; i++) { |
764 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
765 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
766 | do_chop(astr, sv); |
767 | } |
768 | return; |
79072805 |
769 | } |
770 | if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
771 | HV* hv = (HV*)sv; |
772 | HE* entry; |
773 | (void)hv_iterinit(hv); |
774 | /*SUPPRESS 560*/ |
775 | while (entry = hv_iternext(hv)) |
776 | do_chop(astr,hv_iterval(hv,entry)); |
777 | return; |
79072805 |
778 | } |
a0d0e21e |
779 | s = SvPV(sv, len); |
748a9306 |
780 | if (len && !SvPOK(sv)) |
a0d0e21e |
781 | s = SvPV_force(sv, len); |
a0ed51b3 |
782 | if (IN_UTF8) { |
783 | if (s && len) { |
784 | char *send = s + len; |
785 | char *start = s; |
786 | s = send - 1; |
787 | while ((*s & 0xc0) == 0x80) |
788 | --s; |
789 | if (UTF8SKIP(s) != send - s) |
790 | warn("Malformed UTF-8 character"); |
791 | sv_setpvn(astr, s, send - s); |
792 | *s = '\0'; |
793 | SvCUR_set(sv, s - start); |
794 | SvNIOK_off(sv); |
795 | } |
796 | else |
797 | sv_setpvn(astr, "", 0); |
798 | } |
799 | else |
a0d0e21e |
800 | if (s && len) { |
801 | s += --len; |
802 | sv_setpvn(astr, s, 1); |
803 | *s = '\0'; |
804 | SvCUR_set(sv, len); |
805 | SvNIOK_off(sv); |
79072805 |
806 | } |
807 | else |
a0d0e21e |
808 | sv_setpvn(astr, "", 0); |
809 | SvSETMAGIC(sv); |
810 | } |
811 | |
812 | I32 |
8ac85365 |
813 | do_chomp(register SV *sv) |
a0d0e21e |
814 | { |
aeea060c |
815 | dTHR; |
c07a80fd |
816 | register I32 count; |
a0d0e21e |
817 | STRLEN len; |
818 | char *s; |
c07a80fd |
819 | |
3280af22 |
820 | if (RsSNARF(PL_rs)) |
c07a80fd |
821 | return 0; |
822 | count = 0; |
a0d0e21e |
823 | if (SvTYPE(sv) == SVt_PVAV) { |
824 | register I32 i; |
825 | I32 max; |
826 | AV* av = (AV*)sv; |
827 | max = AvFILL(av); |
828 | for (i = 0; i <= max; i++) { |
829 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
830 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
831 | count += do_chomp(sv); |
832 | } |
833 | return count; |
834 | } |
835 | if (SvTYPE(sv) == SVt_PVHV) { |
836 | HV* hv = (HV*)sv; |
837 | HE* entry; |
838 | (void)hv_iterinit(hv); |
839 | /*SUPPRESS 560*/ |
840 | while (entry = hv_iternext(hv)) |
841 | count += do_chomp(hv_iterval(hv,entry)); |
842 | return count; |
843 | } |
844 | s = SvPV(sv, len); |
845 | if (len && !SvPOKp(sv)) |
846 | s = SvPV_force(sv, len); |
847 | if (s && len) { |
848 | s += --len; |
3280af22 |
849 | if (RsPARA(PL_rs)) { |
a0d0e21e |
850 | if (*s != '\n') |
851 | goto nope; |
852 | ++count; |
853 | while (len && s[-1] == '\n') { |
854 | --len; |
855 | --s; |
856 | ++count; |
857 | } |
858 | } |
a0d0e21e |
859 | else { |
c07a80fd |
860 | STRLEN rslen; |
3280af22 |
861 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd |
862 | if (rslen == 1) { |
863 | if (*s != *rsptr) |
864 | goto nope; |
865 | ++count; |
866 | } |
867 | else { |
8c2cee6f |
868 | if (len < rslen - 1) |
c07a80fd |
869 | goto nope; |
870 | len -= rslen - 1; |
871 | s -= rslen - 1; |
36477c24 |
872 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
873 | goto nope; |
874 | count += rslen; |
875 | } |
a0d0e21e |
876 | } |
a0d0e21e |
877 | *s = '\0'; |
878 | SvCUR_set(sv, len); |
879 | SvNIOK_off(sv); |
880 | } |
881 | nope: |
882 | SvSETMAGIC(sv); |
883 | return count; |
884 | } |
79072805 |
885 | |
886 | void |
8ac85365 |
887 | do_vop(I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
888 | { |
aeea060c |
889 | dTHR; /* just for taint */ |
79072805 |
890 | #ifdef LIBERAL |
891 | register long *dl; |
892 | register long *ll; |
893 | register long *rl; |
894 | #endif |
895 | register char *dc; |
463ee0b2 |
896 | STRLEN leftlen; |
897 | STRLEN rightlen; |
7a4c00b4 |
898 | register char *lc; |
899 | register char *rc; |
79072805 |
900 | register I32 len; |
a0d0e21e |
901 | I32 lensave; |
7a4c00b4 |
902 | char *lsave; |
903 | char *rsave; |
79072805 |
904 | |
1fbd88dc |
905 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
906 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
7a4c00b4 |
907 | lsave = lc = SvPV(left, leftlen); |
908 | rsave = rc = SvPV(right, rightlen); |
93a17b20 |
909 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
910 | lensave = len; |
7a4c00b4 |
911 | if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
3280af22 |
912 | dc = SvPV_force(sv, PL_na); |
ff68c719 |
913 | if (SvCUR(sv) < len) { |
914 | dc = SvGROW(sv, len + 1); |
915 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
916 | } |
917 | } |
918 | else { |
919 | I32 needlen = ((optype == OP_BIT_AND) |
920 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
921 | Newz(801, dc, needlen + 1, char); |
922 | (void)sv_usepvn(sv, dc, needlen); |
923 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
924 | } |
a0d0e21e |
925 | SvCUR_set(sv, len); |
926 | (void)SvPOK_only(sv); |
79072805 |
927 | #ifdef LIBERAL |
928 | if (len >= sizeof(long)*4 && |
929 | !((long)dc % sizeof(long)) && |
930 | !((long)lc % sizeof(long)) && |
931 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
932 | { |
933 | I32 remainder = len % (sizeof(long)*4); |
934 | len /= (sizeof(long)*4); |
935 | |
936 | dl = (long*)dc; |
937 | ll = (long*)lc; |
938 | rl = (long*)rc; |
939 | |
940 | switch (optype) { |
941 | case OP_BIT_AND: |
942 | while (len--) { |
943 | *dl++ = *ll++ & *rl++; |
944 | *dl++ = *ll++ & *rl++; |
945 | *dl++ = *ll++ & *rl++; |
946 | *dl++ = *ll++ & *rl++; |
947 | } |
948 | break; |
a0d0e21e |
949 | case OP_BIT_XOR: |
79072805 |
950 | while (len--) { |
951 | *dl++ = *ll++ ^ *rl++; |
952 | *dl++ = *ll++ ^ *rl++; |
953 | *dl++ = *ll++ ^ *rl++; |
954 | *dl++ = *ll++ ^ *rl++; |
955 | } |
956 | break; |
957 | case OP_BIT_OR: |
958 | while (len--) { |
959 | *dl++ = *ll++ | *rl++; |
960 | *dl++ = *ll++ | *rl++; |
961 | *dl++ = *ll++ | *rl++; |
962 | *dl++ = *ll++ | *rl++; |
963 | } |
964 | } |
965 | |
966 | dc = (char*)dl; |
967 | lc = (char*)ll; |
968 | rc = (char*)rl; |
969 | |
970 | len = remainder; |
971 | } |
972 | #endif |
a0d0e21e |
973 | { |
a0d0e21e |
974 | switch (optype) { |
975 | case OP_BIT_AND: |
976 | while (len--) |
977 | *dc++ = *lc++ & *rc++; |
978 | break; |
979 | case OP_BIT_XOR: |
980 | while (len--) |
981 | *dc++ = *lc++ ^ *rc++; |
982 | goto mop_up; |
983 | case OP_BIT_OR: |
984 | while (len--) |
985 | *dc++ = *lc++ | *rc++; |
986 | mop_up: |
987 | len = lensave; |
988 | if (rightlen > len) |
989 | sv_catpvn(sv, rsave + len, rightlen - len); |
990 | else if (leftlen > len) |
991 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
992 | else |
993 | *SvEND(sv) = '\0'; |
a0d0e21e |
994 | break; |
995 | } |
79072805 |
996 | } |
fb73857a |
997 | SvTAINT(sv); |
79072805 |
998 | } |
463ee0b2 |
999 | |
1000 | OP * |
8ac85365 |
1001 | do_kv(ARGSproto) |
463ee0b2 |
1002 | { |
4e35701f |
1003 | djSP; |
463ee0b2 |
1004 | HV *hv = (HV*)POPs; |
800e9ae0 |
1005 | HV *keys; |
463ee0b2 |
1006 | register HE *entry; |
463ee0b2 |
1007 | SV *tmpstr; |
54310121 |
1008 | I32 gimme = GIMME_V; |
533c011a |
1009 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1010 | I32 dovalues = (PL_op->op_type == OP_VALUES); |
c750a3ec |
1011 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
1012 | |
533c011a |
1013 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e |
1014 | dokeys = dovalues = TRUE; |
463ee0b2 |
1015 | |
85581909 |
1016 | if (!hv) { |
533c011a |
1017 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 |
1018 | dTARGET; /* make sure to clear its target here */ |
1019 | if (SvTYPE(TARG) == SVt_PVLV) |
1020 | LvTARG(TARG) = Nullsv; |
1021 | PUSHs(TARG); |
1022 | } |
463ee0b2 |
1023 | RETURN; |
85581909 |
1024 | } |
748a9306 |
1025 | |
800e9ae0 |
1026 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1027 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1028 | |
54310121 |
1029 | if (gimme == G_VOID) |
aa689395 |
1030 | RETURN; |
1031 | |
54310121 |
1032 | if (gimme == G_SCALAR) { |
6ee623d5 |
1033 | IV i; |
463ee0b2 |
1034 | dTARGET; |
1035 | |
533c011a |
1036 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 |
1037 | if (SvTYPE(TARG) < SVt_PVLV) { |
1038 | sv_upgrade(TARG, SVt_PVLV); |
1039 | sv_magic(TARG, Nullsv, 'k', Nullch, 0); |
1040 | } |
1041 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1042 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1043 | if (LvTARG(TARG)) |
1044 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1045 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1046 | } |
85581909 |
1047 | PUSHs(TARG); |
1048 | RETURN; |
1049 | } |
1050 | |
800e9ae0 |
1051 | if (!SvRMAGICAL(keys) || !mg_find((SV*)keys,'P')) |
1052 | i = HvKEYS(keys); |
463ee0b2 |
1053 | else { |
1054 | i = 0; |
463ee0b2 |
1055 | /*SUPPRESS 560*/ |
800e9ae0 |
1056 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1057 | } |
1058 | PUSHi( i ); |
1059 | RETURN; |
1060 | } |
1061 | |
8ed4b672 |
1062 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1063 | |
463ee0b2 |
1064 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
800e9ae0 |
1065 | while (entry = hv_iternext(keys)) { |
463ee0b2 |
1066 | SPAGAIN; |
8c2cee6f |
1067 | if (dokeys) |
1068 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ |
463ee0b2 |
1069 | if (dovalues) { |
463ee0b2 |
1070 | PUTBACK; |
b6429b1b |
1071 | tmpstr = realhv ? |
1072 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); |
46fc3d4c |
1073 | DEBUG_H(sv_setpvf(tmpstr, "%lu%%%d=%lu", |
1074 | (unsigned long)HeHASH(entry), |
800e9ae0 |
1075 | HvMAX(keys)+1, |
1076 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1077 | SPAGAIN; |
46fc3d4c |
1078 | XPUSHs(tmpstr); |
463ee0b2 |
1079 | } |
1080 | PUTBACK; |
1081 | } |
1082 | return NORMAL; |
1083 | } |
4e35701f |
1084 | |