Commit | Line | Data |
---|---|---|
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 | 24 | STATIC I32 |
cea2e8a9 | 25 | S_do_trans_CC_simple(pTHX_ SV *sv) |
79072805 | 26 | { |
11343788 | 27 | dTHR; |
4757a243 LW |
28 | U8 *s; |
29 | U8 *send; | |
30 | I32 matches = 0; | |
463ee0b2 | 31 | STRLEN len; |
4757a243 LW |
32 | short *tbl; |
33 | I32 ch; | |
79072805 | 34 | |
4757a243 LW |
35 | tbl = (short*)cPVOP->op_pv; |
36 | if (!tbl) | |
cea2e8a9 | 37 | Perl_croak(aTHX_ "panic: do_trans"); |
a0ed51b3 | 38 | |
4757a243 LW |
39 | s = (U8*)SvPV(sv, len); |
40 | send = s + len; | |
41 | ||
42 | while (s < send) { | |
43 | if ((ch = tbl[*s]) >= 0) { | |
44 | matches++; | |
45 | *s = ch; | |
79072805 | 46 | } |
4757a243 LW |
47 | s++; |
48 | } | |
49 | SvSETMAGIC(sv); | |
50 | ||
51 | return matches; | |
52 | } | |
53 | ||
942e002e | 54 | STATIC I32 |
cea2e8a9 | 55 | S_do_trans_CC_count(pTHX_ SV *sv) |
4757a243 LW |
56 | { |
57 | dTHR; | |
58 | U8 *s; | |
59 | U8 *send; | |
60 | I32 matches = 0; | |
61 | STRLEN len; | |
62 | short *tbl; | |
63 | ||
64 | tbl = (short*)cPVOP->op_pv; | |
65 | if (!tbl) | |
cea2e8a9 | 66 | Perl_croak(aTHX_ "panic: do_trans"); |
4757a243 LW |
67 | |
68 | s = (U8*)SvPV(sv, len); | |
69 | send = s + len; | |
70 | ||
71 | while (s < send) { | |
72 | if (tbl[*s] >= 0) | |
73 | matches++; | |
74 | s++; | |
75 | } | |
76 | ||
77 | return matches; | |
78 | } | |
79 | ||
942e002e | 80 | STATIC I32 |
cea2e8a9 | 81 | S_do_trans_CC_complex(pTHX_ SV *sv) |
4757a243 LW |
82 | { |
83 | dTHR; | |
84 | U8 *s; | |
85 | U8 *send; | |
86 | U8 *d; | |
87 | I32 matches = 0; | |
88 | STRLEN len; | |
89 | short *tbl; | |
90 | I32 ch; | |
91 | ||
92 | tbl = (short*)cPVOP->op_pv; | |
93 | if (!tbl) | |
cea2e8a9 | 94 | Perl_croak(aTHX_ "panic: do_trans"); |
4757a243 LW |
95 | |
96 | s = (U8*)SvPV(sv, len); | |
97 | send = s + len; | |
98 | ||
99 | d = s; | |
100 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
101 | U8* p = send; | |
102 | ||
103 | while (s < send) { | |
104 | if ((ch = tbl[*s]) >= 0) { | |
105 | *d = ch; | |
106 | matches++; | |
107 | if (p == d - 1 && *p == *d) | |
108 | matches--; | |
a0ed51b3 | 109 | else |
4757a243 | 110 | p = d++; |
a0ed51b3 | 111 | } |
4757a243 LW |
112 | else if (ch == -1) /* -1 is unmapped character */ |
113 | *d++ = *s; /* -2 is delete character */ | |
114 | s++; | |
a0ed51b3 | 115 | } |
4757a243 LW |
116 | } |
117 | else { | |
118 | while (s < send) { | |
119 | if ((ch = tbl[*s]) >= 0) { | |
120 | *d = ch; | |
121 | matches++; | |
122 | d++; | |
a0ed51b3 | 123 | } |
4757a243 LW |
124 | else if (ch == -1) /* -1 is unmapped character */ |
125 | *d++ = *s; /* -2 is delete character */ | |
126 | s++; | |
5d06d08e | 127 | } |
4757a243 LW |
128 | } |
129 | matches += send - d; /* account for disappeared chars */ | |
130 | *d = '\0'; | |
131 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
132 | SvSETMAGIC(sv); | |
133 | ||
134 | return matches; | |
135 | } | |
136 | ||
942e002e | 137 | STATIC I32 |
cea2e8a9 | 138 | S_do_trans_UU_simple(pTHX_ SV *sv) |
4757a243 LW |
139 | { |
140 | dTHR; | |
141 | U8 *s; | |
142 | U8 *send; | |
143 | U8 *d; | |
144 | I32 matches = 0; | |
145 | STRLEN len; | |
146 | ||
147 | SV* rv = (SV*)cSVOP->op_sv; | |
148 | HV* hv = (HV*)SvRV(rv); | |
149 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
150 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
151 | UV extra = none + 1; | |
152 | UV final; | |
153 | UV uv; | |
154 | ||
155 | s = (U8*)SvPV(sv, len); | |
156 | send = s + len; | |
157 | ||
158 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
159 | if (svp) | |
160 | final = SvUV(*svp); | |
161 | ||
162 | d = s; | |
163 | while (s < send) { | |
164 | if ((uv = swash_fetch(rv, s)) < none) { | |
165 | s += UTF8SKIP(s); | |
166 | matches++; | |
167 | d = uv_to_utf8(d, uv); | |
168 | } | |
169 | else if (uv == none) { | |
170 | int i; | |
171 | for (i = UTF8SKIP(s); i; i--) | |
172 | *d++ = *s++; | |
173 | } | |
174 | else if (uv == extra) { | |
175 | s += UTF8SKIP(s); | |
176 | matches++; | |
177 | d = uv_to_utf8(d, final); | |
178 | } | |
179 | else | |
180 | s += UTF8SKIP(s); | |
181 | } | |
182 | *d = '\0'; | |
183 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
184 | SvSETMAGIC(sv); | |
185 | ||
186 | return matches; | |
187 | } | |
188 | ||
942e002e | 189 | STATIC I32 |
cea2e8a9 | 190 | S_do_trans_UU_count(pTHX_ SV *sv) |
4757a243 LW |
191 | { |
192 | dTHR; | |
193 | U8 *s; | |
194 | U8 *send; | |
195 | I32 matches = 0; | |
196 | STRLEN len; | |
197 | ||
198 | SV* rv = (SV*)cSVOP->op_sv; | |
199 | HV* hv = (HV*)SvRV(rv); | |
200 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
201 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
202 | UV uv; | |
203 | ||
204 | s = (U8*)SvPV(sv, len); | |
205 | send = s + len; | |
206 | ||
207 | while (s < send) { | |
834a4ddd | 208 | if ((uv = swash_fetch(rv, s)) < none) |
4757a243 | 209 | matches++; |
834a4ddd | 210 | s += UTF8SKIP(s); |
4757a243 LW |
211 | } |
212 | ||
213 | return matches; | |
214 | } | |
215 | ||
942e002e | 216 | STATIC I32 |
cea2e8a9 | 217 | S_do_trans_UC_simple(pTHX_ SV *sv) |
4757a243 LW |
218 | { |
219 | dTHR; | |
220 | U8 *s; | |
221 | U8 *send; | |
222 | U8 *d; | |
223 | I32 matches = 0; | |
224 | STRLEN len; | |
225 | ||
226 | SV* rv = (SV*)cSVOP->op_sv; | |
227 | HV* hv = (HV*)SvRV(rv); | |
228 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
229 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
230 | UV extra = none + 1; | |
231 | UV final; | |
232 | UV uv; | |
233 | ||
234 | s = (U8*)SvPV(sv, len); | |
235 | send = s + len; | |
236 | ||
237 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
238 | if (svp) | |
239 | final = SvUV(*svp); | |
240 | ||
241 | d = s; | |
242 | while (s < send) { | |
243 | if ((uv = swash_fetch(rv, s)) < none) { | |
244 | s += UTF8SKIP(s); | |
245 | matches++; | |
246 | *d++ = (U8)uv; | |
247 | } | |
248 | else if (uv == none) { | |
249 | I32 ulen; | |
250 | uv = utf8_to_uv(s, &ulen); | |
251 | s += ulen; | |
252 | *d++ = (U8)uv; | |
253 | } | |
254 | else if (uv == extra) { | |
255 | s += UTF8SKIP(s); | |
256 | matches++; | |
257 | *d++ = (U8)final; | |
258 | } | |
259 | else | |
260 | s += UTF8SKIP(s); | |
261 | } | |
262 | *d = '\0'; | |
263 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
264 | SvSETMAGIC(sv); | |
265 | ||
266 | return matches; | |
267 | } | |
268 | ||
942e002e | 269 | STATIC I32 |
cea2e8a9 | 270 | S_do_trans_CU_simple(pTHX_ SV *sv) |
4757a243 LW |
271 | { |
272 | dTHR; | |
273 | U8 *s; | |
274 | U8 *send; | |
275 | U8 *d; | |
276 | U8 *dst; | |
277 | I32 matches = 0; | |
278 | STRLEN len; | |
279 | ||
280 | SV* rv = (SV*)cSVOP->op_sv; | |
281 | HV* hv = (HV*)SvRV(rv); | |
282 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
283 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
284 | UV extra = none + 1; | |
285 | UV final; | |
286 | UV uv; | |
806e7201 | 287 | U8 tmpbuf[UTF8_MAXLEN]; |
4757a243 LW |
288 | I32 bits = 16; |
289 | ||
290 | s = (U8*)SvPV(sv, len); | |
291 | send = s + len; | |
292 | ||
293 | svp = hv_fetch(hv, "BITS", 4, FALSE); | |
294 | if (svp) | |
295 | bits = (I32)SvIV(*svp); | |
296 | ||
297 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
298 | if (svp) | |
299 | final = SvUV(*svp); | |
300 | ||
301 | Newz(801, d, len * (bits >> 3) + 1, U8); | |
302 | dst = d; | |
303 | ||
304 | while (s < send) { | |
305 | uv = *s++; | |
306 | if (uv < 0x80) | |
307 | tmpbuf[0] = uv; | |
308 | else { | |
309 | tmpbuf[0] = (( uv >> 6) | 0xc0); | |
310 | tmpbuf[1] = (( uv & 0x3f) | 0x80); | |
a0ed51b3 | 311 | } |
4757a243 LW |
312 | |
313 | if ((uv = swash_fetch(rv, tmpbuf)) < none) { | |
314 | matches++; | |
315 | d = uv_to_utf8(d, uv); | |
316 | } | |
317 | else if (uv == none) | |
318 | d = uv_to_utf8(d, s[-1]); | |
319 | else if (uv == extra) { | |
320 | matches++; | |
321 | d = uv_to_utf8(d, final); | |
322 | } | |
323 | } | |
324 | *d = '\0'; | |
325 | sv_usepvn_mg(sv, (char*)dst, d - dst); | |
326 | ||
327 | return matches; | |
328 | } | |
329 | ||
330 | /* utf-8 to latin-1 */ | |
331 | ||
942e002e | 332 | STATIC I32 |
cea2e8a9 | 333 | S_do_trans_UC_trivial(pTHX_ SV *sv) |
4757a243 LW |
334 | { |
335 | dTHR; | |
336 | U8 *s; | |
337 | U8 *send; | |
338 | U8 *d; | |
339 | STRLEN len; | |
340 | ||
341 | s = (U8*)SvPV(sv, len); | |
342 | send = s + len; | |
343 | ||
344 | d = s; | |
345 | while (s < send) { | |
346 | if (*s < 0x80) | |
347 | *d++ = *s++; | |
a0ed51b3 | 348 | else { |
4757a243 LW |
349 | I32 ulen; |
350 | UV uv = utf8_to_uv(s, &ulen); | |
351 | s += ulen; | |
352 | *d++ = (U8)uv; | |
353 | } | |
354 | } | |
355 | *d = '\0'; | |
356 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
357 | SvSETMAGIC(sv); | |
a0ed51b3 | 358 | |
4757a243 LW |
359 | return SvCUR(sv); |
360 | } | |
a0ed51b3 | 361 | |
4757a243 | 362 | /* latin-1 to utf-8 */ |
a0ed51b3 | 363 | |
942e002e | 364 | STATIC I32 |
cea2e8a9 | 365 | S_do_trans_CU_trivial(pTHX_ SV *sv) |
4757a243 LW |
366 | { |
367 | dTHR; | |
368 | U8 *s; | |
369 | U8 *send; | |
370 | U8 *d; | |
371 | U8 *dst; | |
372 | I32 matches; | |
373 | STRLEN len; | |
a0ed51b3 | 374 | |
4757a243 LW |
375 | s = (U8*)SvPV(sv, len); |
376 | send = s + len; | |
377 | ||
378 | Newz(801, d, len * 2 + 1, U8); | |
379 | dst = d; | |
380 | ||
381 | matches = send - s; | |
382 | ||
383 | while (s < send) { | |
384 | if (*s < 0x80) | |
385 | *d++ = *s++; | |
386 | else { | |
387 | UV uv = *s++; | |
388 | *d++ = (( uv >> 6) | 0xc0); | |
389 | *d++ = (( uv & 0x3f) | 0x80); | |
390 | } | |
391 | } | |
392 | *d = '\0'; | |
393 | sv_usepvn_mg(sv, (char*)dst, d - dst); | |
394 | ||
395 | return matches; | |
396 | } | |
397 | ||
942e002e | 398 | STATIC I32 |
cea2e8a9 | 399 | S_do_trans_UU_complex(pTHX_ SV *sv) |
4757a243 LW |
400 | { |
401 | dTHR; | |
402 | U8 *s; | |
403 | U8 *send; | |
404 | U8 *d; | |
405 | I32 matches = 0; | |
406 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; | |
407 | I32 from_utf = PL_op->op_private & OPpTRANS_FROM_UTF; | |
408 | I32 to_utf = PL_op->op_private & OPpTRANS_TO_UTF; | |
409 | I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
410 | SV* rv = (SV*)cSVOP->op_sv; | |
411 | HV* hv = (HV*)SvRV(rv); | |
412 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
413 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
414 | UV extra = none + 1; | |
415 | UV final; | |
416 | UV uv; | |
417 | STRLEN len; | |
418 | U8 *dst; | |
419 | ||
420 | s = (U8*)SvPV(sv, len); | |
421 | send = s + len; | |
422 | ||
423 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
424 | if (svp) | |
425 | final = SvUV(*svp); | |
426 | ||
427 | if (PL_op->op_private & OPpTRANS_GROWS) { | |
428 | I32 bits = 16; | |
429 | ||
430 | svp = hv_fetch(hv, "BITS", 4, FALSE); | |
431 | if (svp) | |
432 | bits = (I32)SvIV(*svp); | |
433 | ||
434 | Newz(801, d, len * (bits >> 3) + 1, U8); | |
435 | dst = d; | |
436 | } | |
437 | else { | |
438 | d = s; | |
439 | dst = 0; | |
440 | } | |
441 | ||
442 | if (squash) { | |
443 | UV puv = 0xfeedface; | |
444 | while (s < send) { | |
445 | if (from_utf) { | |
446 | uv = swash_fetch(rv, s); | |
a0ed51b3 LW |
447 | } |
448 | else { | |
4757a243 LW |
449 | U8 tmpbuf[2]; |
450 | uv = *s++; | |
451 | if (uv < 0x80) | |
452 | tmpbuf[0] = uv; | |
453 | else { | |
454 | tmpbuf[0] = (( uv >> 6) | 0xc0); | |
455 | tmpbuf[1] = (( uv & 0x3f) | 0x80); | |
456 | } | |
457 | uv = swash_fetch(rv, tmpbuf); | |
458 | } | |
459 | if (uv < none) { | |
460 | matches++; | |
461 | if (uv != puv) { | |
462 | if (uv >= 0x80 && to_utf) | |
463 | d = uv_to_utf8(d, uv); | |
464 | else | |
465 | *d++ = (U8)uv; | |
466 | puv = uv; | |
467 | } | |
468 | if (from_utf) | |
469 | s += UTF8SKIP(s); | |
470 | continue; | |
471 | } | |
472 | else if (uv == none) { /* "none" is unmapped character */ | |
473 | if (from_utf) { | |
474 | if (*s < 0x80) | |
475 | *d++ = *s++; | |
476 | else if (to_utf) { | |
a0ed51b3 | 477 | int i; |
4757a243 LW |
478 | for (i = UTF8SKIP(s); i; --i) |
479 | *d++ = *s++; | |
a0ed51b3 | 480 | } |
4757a243 LW |
481 | else { |
482 | I32 ulen; | |
483 | *d++ = (U8)utf8_to_uv(s, &ulen); | |
484 | s += ulen; | |
a0ed51b3 | 485 | } |
a0ed51b3 | 486 | } |
4757a243 LW |
487 | else { /* must be to_utf only */ |
488 | d = uv_to_utf8(d, s[-1]); | |
489 | } | |
490 | puv = 0xfeedface; | |
491 | continue; | |
a0ed51b3 | 492 | } |
4757a243 LW |
493 | else if (uv == extra && !del) { |
494 | matches++; | |
495 | if (uv != puv) { | |
496 | if (final >= 0x80 && to_utf) | |
497 | d = uv_to_utf8(d, final); | |
498 | else | |
499 | *d++ = (U8)final; | |
500 | puv = final; | |
501 | } | |
502 | if (from_utf) | |
503 | s += UTF8SKIP(s); | |
504 | continue; | |
505 | } | |
506 | matches++; /* "none+1" is delete character */ | |
507 | if (from_utf) | |
508 | s += UTF8SKIP(s); | |
a0ed51b3 | 509 | } |
79072805 LW |
510 | } |
511 | else { | |
4757a243 LW |
512 | while (s < send) { |
513 | if (from_utf) { | |
514 | uv = swash_fetch(rv, s); | |
515 | } | |
516 | else { | |
517 | U8 tmpbuf[2]; | |
518 | uv = *s++; | |
519 | if (uv < 0x80) | |
520 | tmpbuf[0] = uv; | |
521 | else { | |
522 | tmpbuf[0] = (( uv >> 6) | 0xc0); | |
523 | tmpbuf[1] = (( uv & 0x3f) | 0x80); | |
a0ed51b3 | 524 | } |
4757a243 | 525 | uv = swash_fetch(rv, tmpbuf); |
a0ed51b3 | 526 | } |
4757a243 LW |
527 | if (uv < none) { |
528 | matches++; | |
529 | if (uv >= 0x80 && to_utf) | |
530 | d = uv_to_utf8(d, uv); | |
531 | else | |
532 | *d++ = (U8)uv; | |
533 | if (from_utf) | |
534 | s += UTF8SKIP(s); | |
535 | continue; | |
a0ed51b3 | 536 | } |
4757a243 LW |
537 | else if (uv == none) { /* "none" is unmapped character */ |
538 | if (from_utf) { | |
539 | if (*s < 0x80) | |
540 | *d++ = *s++; | |
541 | else if (to_utf) { | |
542 | int i; | |
543 | for (i = UTF8SKIP(s); i; --i) | |
544 | *d++ = *s++; | |
545 | } | |
546 | else { | |
547 | I32 ulen; | |
548 | *d++ = (U8)utf8_to_uv(s, &ulen); | |
549 | s += ulen; | |
a0ed51b3 | 550 | } |
79072805 | 551 | } |
4757a243 LW |
552 | else { /* must be to_utf only */ |
553 | d = uv_to_utf8(d, s[-1]); | |
554 | } | |
555 | continue; | |
79072805 | 556 | } |
4757a243 LW |
557 | else if (uv == extra && !del) { |
558 | matches++; | |
559 | if (final >= 0x80 && to_utf) | |
560 | d = uv_to_utf8(d, final); | |
561 | else | |
562 | *d++ = (U8)final; | |
563 | if (from_utf) | |
564 | s += UTF8SKIP(s); | |
565 | continue; | |
566 | } | |
567 | matches++; /* "none+1" is delete character */ | |
568 | if (from_utf) | |
569 | s += UTF8SKIP(s); | |
79072805 | 570 | } |
4757a243 LW |
571 | } |
572 | if (dst) | |
573 | sv_usepvn(sv, (char*)dst, d - dst); | |
574 | else { | |
575 | *d = '\0'; | |
576 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
577 | } | |
578 | SvSETMAGIC(sv); | |
579 | ||
580 | return matches; | |
581 | } | |
582 | ||
583 | I32 | |
864dbfa3 | 584 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 | 585 | { |
46124e9e | 586 | dTHR; |
4757a243 LW |
587 | STRLEN len; |
588 | ||
589 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) | |
cea2e8a9 | 590 | Perl_croak(aTHX_ PL_no_modify); |
4757a243 LW |
591 | |
592 | (void)SvPV(sv, len); | |
593 | if (!len) | |
594 | return 0; | |
595 | if (!SvPOKp(sv)) | |
596 | (void)SvPV_force(sv, len); | |
2de7b02f GS |
597 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) |
598 | (void)SvPOK_only_UTF8(sv); | |
4757a243 | 599 | |
cea2e8a9 | 600 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 LW |
601 | |
602 | switch (PL_op->op_private & 63) { | |
603 | case 0: | |
604 | return do_trans_CC_simple(sv); | |
605 | ||
606 | case OPpTRANS_FROM_UTF: | |
607 | return do_trans_UC_simple(sv); | |
608 | ||
609 | case OPpTRANS_TO_UTF: | |
610 | return do_trans_CU_simple(sv); | |
611 | ||
612 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF: | |
613 | return do_trans_UU_simple(sv); | |
614 | ||
615 | case OPpTRANS_IDENTICAL: | |
616 | return do_trans_CC_count(sv); | |
617 | ||
618 | case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL: | |
619 | return do_trans_UC_trivial(sv); | |
620 | ||
621 | case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: | |
622 | return do_trans_CU_trivial(sv); | |
623 | ||
624 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: | |
625 | return do_trans_UU_count(sv); | |
626 | ||
627 | default: | |
628 | if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)) | |
629 | return do_trans_UU_complex(sv); /* could be UC or CU too */ | |
630 | else | |
631 | return do_trans_CC_complex(sv); | |
79072805 | 632 | } |
79072805 LW |
633 | } |
634 | ||
635 | void | |
864dbfa3 | 636 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
637 | { |
638 | SV **oldmark = mark; | |
639 | register I32 items = sp - mark; | |
79072805 | 640 | register STRLEN len; |
463ee0b2 LW |
641 | STRLEN delimlen; |
642 | register char *delim = SvPV(del, delimlen); | |
643 | STRLEN tmplen; | |
79072805 LW |
644 | |
645 | mark++; | |
646 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 647 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
648 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
649 | while (items-- > 0) { | |
48c036b1 | 650 | if (*mark && !SvGMAGICAL(*mark) && SvOK(*mark)) { |
463ee0b2 LW |
651 | SvPV(*mark, tmplen); |
652 | len += tmplen; | |
79072805 LW |
653 | } |
654 | mark++; | |
655 | } | |
656 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
657 | ||
658 | mark = oldmark; | |
db7c17d7 | 659 | items = sp - mark; |
79072805 LW |
660 | ++mark; |
661 | } | |
662 | ||
463ee0b2 | 663 | if (items-- > 0) { |
8990e307 LW |
664 | char *s; |
665 | ||
92d29cee JH |
666 | sv_setpv(sv, ""); |
667 | if (*mark) | |
668 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
669 | mark++; |
670 | } | |
79072805 LW |
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 | |
864dbfa3 | 688 | Perl_do_sprintf(pTHX_ 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 LW |
698 | } |
699 | ||
4ebbc975 | 700 | /* XXX SvUTF8 support missing! */ |
81e118e0 JH |
701 | UV |
702 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
703 | { | |
704 | STRLEN srclen, len; | |
705 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); | |
706 | UV retnum = 0; | |
707 | ||
a50d7633 | 708 | if (offset < 0) |
81e118e0 | 709 | return retnum; |
a50d7633 JH |
710 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
711 | Perl_croak(aTHX_ "Illegal number of bits in vec"); | |
81e118e0 JH |
712 | offset *= size; /* turn into bit offset */ |
713 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
714 | if (len > srclen) { | |
715 | if (size <= 8) | |
716 | retnum = 0; | |
717 | else { | |
718 | offset >>= 3; /* turn into byte offset */ | |
719 | if (size == 16) { | |
720 | if (offset >= srclen) | |
721 | retnum = 0; | |
722 | else | |
628e1a40 | 723 | retnum = (UV) s[offset] << 8; |
81e118e0 JH |
724 | } |
725 | else if (size == 32) { | |
726 | if (offset >= srclen) | |
727 | retnum = 0; | |
728 | else if (offset + 1 >= srclen) | |
729 | retnum = | |
730 | ((UV) s[offset ] << 24); | |
731 | else if (offset + 2 >= srclen) | |
732 | retnum = | |
733 | ((UV) s[offset ] << 24) + | |
734 | ((UV) s[offset + 1] << 16); | |
735 | else | |
736 | retnum = | |
737 | ((UV) s[offset ] << 24) + | |
738 | ((UV) s[offset + 1] << 16) + | |
739 | ( s[offset + 2] << 8); | |
740 | } | |
d7d93a81 | 741 | #ifdef UV_IS_QUAD |
c5a0f51a JH |
742 | else if (size == 64) { |
743 | dTHR; | |
744 | if (ckWARN(WARN_PORTABLE)) | |
745 | Perl_warner(aTHX_ WARN_PORTABLE, | |
746 | "Bit vector size > 32 non-portable"); | |
747 | if (offset >= srclen) | |
748 | retnum = 0; | |
749 | else if (offset + 1 >= srclen) | |
750 | retnum = | |
751 | (UV) s[offset ] << 56; | |
752 | else if (offset + 2 >= srclen) | |
753 | retnum = | |
754 | ((UV) s[offset ] << 56) + | |
755 | ((UV) s[offset + 1] << 48); | |
756 | else if (offset + 3 >= srclen) | |
757 | retnum = | |
758 | ((UV) s[offset ] << 56) + | |
759 | ((UV) s[offset + 1] << 48) + | |
760 | ((UV) s[offset + 2] << 40); | |
761 | else if (offset + 4 >= srclen) | |
762 | retnum = | |
763 | ((UV) s[offset ] << 56) + | |
764 | ((UV) s[offset + 1] << 48) + | |
765 | ((UV) s[offset + 2] << 40) + | |
766 | ((UV) s[offset + 3] << 32); | |
767 | else if (offset + 5 >= 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 | ( s[offset + 4] << 24); | |
774 | else if (offset + 6 >= srclen) | |
775 | retnum = | |
776 | ((UV) s[offset ] << 56) + | |
777 | ((UV) s[offset + 1] << 48) + | |
778 | ((UV) s[offset + 2] << 40) + | |
779 | ((UV) s[offset + 3] << 32) + | |
780 | ((UV) s[offset + 4] << 24) + | |
781 | ((UV) s[offset + 5] << 16); | |
782 | else | |
783 | retnum = | |
784 | ((UV) s[offset ] << 56) + | |
785 | ((UV) s[offset + 1] << 48) + | |
786 | ((UV) s[offset + 2] << 40) + | |
787 | ((UV) s[offset + 3] << 32) + | |
788 | ((UV) s[offset + 4] << 24) + | |
789 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 790 | ( s[offset + 6] << 8); |
c5a0f51a JH |
791 | } |
792 | #endif | |
81e118e0 JH |
793 | } |
794 | } | |
795 | else if (size < 8) | |
796 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); | |
797 | else { | |
798 | offset >>= 3; /* turn into byte offset */ | |
799 | if (size == 8) | |
800 | retnum = s[offset]; | |
801 | else if (size == 16) | |
802 | retnum = | |
628e1a40 | 803 | ((UV) s[offset] << 8) + |
81e118e0 JH |
804 | s[offset + 1]; |
805 | else if (size == 32) | |
806 | retnum = | |
807 | ((UV) s[offset ] << 24) + | |
808 | ((UV) s[offset + 1] << 16) + | |
809 | ( s[offset + 2] << 8) + | |
810 | s[offset + 3]; | |
d7d93a81 | 811 | #ifdef UV_IS_QUAD |
c5a0f51a JH |
812 | else if (size == 64) { |
813 | dTHR; | |
814 | if (ckWARN(WARN_PORTABLE)) | |
815 | Perl_warner(aTHX_ WARN_PORTABLE, | |
816 | "Bit vector size > 32 non-portable"); | |
817 | retnum = | |
818 | ((UV) s[offset ] << 56) + | |
819 | ((UV) s[offset + 1] << 48) + | |
820 | ((UV) s[offset + 2] << 40) + | |
821 | ((UV) s[offset + 3] << 32) + | |
822 | ((UV) s[offset + 4] << 24) + | |
823 | ((UV) s[offset + 5] << 16) + | |
628e1a40 | 824 | ( s[offset + 6] << 8) + |
c5a0f51a JH |
825 | s[offset + 7]; |
826 | } | |
827 | #endif | |
81e118e0 JH |
828 | } |
829 | ||
830 | return retnum; | |
831 | } | |
832 | ||
4ebbc975 | 833 | /* XXX SvUTF8 support missing! */ |
79072805 | 834 | void |
864dbfa3 | 835 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
836 | { |
837 | SV *targ = LvTARG(sv); | |
838 | register I32 offset; | |
839 | register I32 size; | |
8990e307 | 840 | register unsigned char *s; |
81e118e0 | 841 | register UV lval; |
79072805 | 842 | I32 mask; |
a0d0e21e LW |
843 | STRLEN targlen; |
844 | STRLEN len; | |
79072805 | 845 | |
8990e307 LW |
846 | if (!targ) |
847 | return; | |
a0d0e21e | 848 | s = (unsigned char*)SvPV_force(targ, targlen); |
4ebbc975 | 849 | (void)SvPOK_only(targ); |
81e118e0 | 850 | lval = SvUV(sv); |
79072805 LW |
851 | offset = LvTARGOFF(sv); |
852 | size = LvTARGLEN(sv); | |
a50d7633 JH |
853 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
854 | Perl_croak(aTHX_ "Illegal number of bits in vec"); | |
a0d0e21e | 855 | |
81e118e0 JH |
856 | offset *= size; /* turn into bit offset */ |
857 | len = (offset + size + 7) / 8; /* required number of bytes */ | |
a0d0e21e LW |
858 | if (len > targlen) { |
859 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 860 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
861 | SvCUR_set(targ, len); |
862 | } | |
863 | ||
79072805 LW |
864 | if (size < 8) { |
865 | mask = (1 << size) - 1; | |
866 | size = offset & 7; | |
867 | lval &= mask; | |
81e118e0 | 868 | offset >>= 3; /* turn into byte offset */ |
79072805 LW |
869 | s[offset] &= ~(mask << size); |
870 | s[offset] |= lval << size; | |
871 | } | |
872 | else { | |
81e118e0 | 873 | offset >>= 3; /* turn into byte offset */ |
79072805 | 874 | if (size == 8) |
c5a0f51a | 875 | s[offset ] = lval & 0xff; |
79072805 | 876 | else if (size == 16) { |
c5a0f51a JH |
877 | s[offset ] = (lval >> 8) & 0xff; |
878 | s[offset+1] = lval & 0xff; | |
79072805 LW |
879 | } |
880 | else if (size == 32) { | |
c5a0f51a JH |
881 | s[offset ] = (lval >> 24) & 0xff; |
882 | s[offset+1] = (lval >> 16) & 0xff; | |
883 | s[offset+2] = (lval >> 8) & 0xff; | |
884 | s[offset+3] = lval & 0xff; | |
885 | } | |
d7d93a81 | 886 | #ifdef UV_IS_QUAD |
c5a0f51a JH |
887 | else if (size == 64) { |
888 | dTHR; | |
889 | if (ckWARN(WARN_PORTABLE)) | |
890 | Perl_warner(aTHX_ WARN_PORTABLE, | |
891 | "Bit vector size > 32 non-portable"); | |
892 | s[offset ] = (lval >> 56) & 0xff; | |
893 | s[offset+1] = (lval >> 48) & 0xff; | |
894 | s[offset+2] = (lval >> 40) & 0xff; | |
895 | s[offset+3] = (lval >> 32) & 0xff; | |
896 | s[offset+4] = (lval >> 24) & 0xff; | |
897 | s[offset+5] = (lval >> 16) & 0xff; | |
898 | s[offset+6] = (lval >> 8) & 0xff; | |
899 | s[offset+7] = lval & 0xff; | |
79072805 | 900 | } |
dc1e3f56 | 901 | #endif |
79072805 | 902 | } |
7bb043c3 | 903 | SvSETMAGIC(targ); |
79072805 LW |
904 | } |
905 | ||
906 | void | |
864dbfa3 | 907 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 908 | { |
463ee0b2 | 909 | STRLEN len; |
a0d0e21e | 910 | char *s; |
c485e607 | 911 | dTHR; |
a0d0e21e | 912 | |
79072805 | 913 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
914 | register I32 i; |
915 | I32 max; | |
916 | AV* av = (AV*)sv; | |
917 | max = AvFILL(av); | |
918 | for (i = 0; i <= max; i++) { | |
919 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 920 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
921 | do_chop(astr, sv); |
922 | } | |
923 | return; | |
79072805 | 924 | } |
aa854799 | 925 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
926 | HV* hv = (HV*)sv; |
927 | HE* entry; | |
928 | (void)hv_iterinit(hv); | |
929 | /*SUPPRESS 560*/ | |
155aba94 | 930 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
931 | do_chop(astr,hv_iterval(hv,entry)); |
932 | return; | |
79072805 | 933 | } |
aa854799 | 934 | else if (SvREADONLY(sv)) |
cea2e8a9 | 935 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e | 936 | s = SvPV(sv, len); |
748a9306 | 937 | if (len && !SvPOK(sv)) |
a0d0e21e | 938 | s = SvPV_force(sv, len); |
7e2040f0 | 939 | if (DO_UTF8(sv)) { |
a0ed51b3 LW |
940 | if (s && len) { |
941 | char *send = s + len; | |
942 | char *start = s; | |
943 | s = send - 1; | |
944 | while ((*s & 0xc0) == 0x80) | |
945 | --s; | |
0453d815 PM |
946 | if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8)) |
947 | Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character"); | |
a0ed51b3 LW |
948 | sv_setpvn(astr, s, send - s); |
949 | *s = '\0'; | |
950 | SvCUR_set(sv, s - start); | |
951 | SvNIOK_off(sv); | |
7e2040f0 | 952 | SvUTF8_on(astr); |
a0ed51b3 LW |
953 | } |
954 | else | |
955 | sv_setpvn(astr, "", 0); | |
956 | } | |
7e2040f0 | 957 | else if (s && len) { |
a0d0e21e LW |
958 | s += --len; |
959 | sv_setpvn(astr, s, 1); | |
960 | *s = '\0'; | |
961 | SvCUR_set(sv, len); | |
2c19a612 | 962 | SvUTF8_off(sv); |
a0d0e21e | 963 | SvNIOK_off(sv); |
79072805 LW |
964 | } |
965 | else | |
a0d0e21e LW |
966 | sv_setpvn(astr, "", 0); |
967 | SvSETMAGIC(sv); | |
7e2040f0 | 968 | } |
a0d0e21e LW |
969 | |
970 | I32 | |
864dbfa3 | 971 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 972 | { |
aeea060c | 973 | dTHR; |
c07a80fd | 974 | register I32 count; |
a0d0e21e LW |
975 | STRLEN len; |
976 | char *s; | |
c07a80fd | 977 | |
3280af22 | 978 | if (RsSNARF(PL_rs)) |
c07a80fd | 979 | return 0; |
4c5a6083 GS |
980 | if (RsRECORD(PL_rs)) |
981 | return 0; | |
c07a80fd | 982 | count = 0; |
a0d0e21e LW |
983 | if (SvTYPE(sv) == SVt_PVAV) { |
984 | register I32 i; | |
985 | I32 max; | |
986 | AV* av = (AV*)sv; | |
987 | max = AvFILL(av); | |
988 | for (i = 0; i <= max; i++) { | |
989 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 990 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
991 | count += do_chomp(sv); |
992 | } | |
993 | return count; | |
994 | } | |
aa854799 | 995 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
996 | HV* hv = (HV*)sv; |
997 | HE* entry; | |
998 | (void)hv_iterinit(hv); | |
999 | /*SUPPRESS 560*/ | |
155aba94 | 1000 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1001 | count += do_chomp(hv_iterval(hv,entry)); |
1002 | return count; | |
1003 | } | |
aa854799 | 1004 | else if (SvREADONLY(sv)) |
cea2e8a9 | 1005 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e LW |
1006 | s = SvPV(sv, len); |
1007 | if (len && !SvPOKp(sv)) | |
1008 | s = SvPV_force(sv, len); | |
1009 | if (s && len) { | |
1010 | s += --len; | |
3280af22 | 1011 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1012 | if (*s != '\n') |
1013 | goto nope; | |
1014 | ++count; | |
1015 | while (len && s[-1] == '\n') { | |
1016 | --len; | |
1017 | --s; | |
1018 | ++count; | |
1019 | } | |
1020 | } | |
a0d0e21e | 1021 | else { |
c07a80fd | 1022 | STRLEN rslen; |
3280af22 | 1023 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 1024 | if (rslen == 1) { |
1025 | if (*s != *rsptr) | |
1026 | goto nope; | |
1027 | ++count; | |
1028 | } | |
1029 | else { | |
8c2cee6f | 1030 | if (len < rslen - 1) |
c07a80fd | 1031 | goto nope; |
1032 | len -= rslen - 1; | |
1033 | s -= rslen - 1; | |
36477c24 | 1034 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1035 | goto nope; |
1036 | count += rslen; | |
1037 | } | |
a0d0e21e | 1038 | } |
a0d0e21e LW |
1039 | *s = '\0'; |
1040 | SvCUR_set(sv, len); | |
1041 | SvNIOK_off(sv); | |
1042 | } | |
1043 | nope: | |
1044 | SvSETMAGIC(sv); | |
1045 | return count; | |
1046 | } | |
79072805 LW |
1047 | |
1048 | void | |
864dbfa3 | 1049 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 | 1050 | { |
aeea060c | 1051 | dTHR; /* just for taint */ |
79072805 LW |
1052 | #ifdef LIBERAL |
1053 | register long *dl; | |
1054 | register long *ll; | |
1055 | register long *rl; | |
1056 | #endif | |
1057 | register char *dc; | |
463ee0b2 LW |
1058 | STRLEN leftlen; |
1059 | STRLEN rightlen; | |
7a4c00b4 | 1060 | register char *lc; |
1061 | register char *rc; | |
79072805 | 1062 | register I32 len; |
a0d0e21e | 1063 | I32 lensave; |
7a4c00b4 | 1064 | char *lsave; |
1065 | char *rsave; | |
0c57e439 GS |
1066 | bool left_utf = DO_UTF8(left); |
1067 | bool right_utf = DO_UTF8(right); | |
1068 | ||
1069 | if (left_utf && !right_utf) | |
1070 | sv_utf8_upgrade(right); | |
1071 | if (!left_utf && right_utf) | |
1072 | sv_utf8_upgrade(left); | |
79072805 | 1073 | |
1fbd88dc CS |
1074 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1075 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 1076 | lsave = lc = SvPV(left, leftlen); |
1077 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 1078 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1079 | lensave = len; |
7a4c00b4 | 1080 | if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2d8e6c8d GS |
1081 | STRLEN n_a; |
1082 | dc = SvPV_force(sv, n_a); | |
ff68c719 | 1083 | if (SvCUR(sv) < len) { |
1084 | dc = SvGROW(sv, len + 1); | |
1085 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); | |
1086 | } | |
1087 | } | |
1088 | else { | |
1089 | I32 needlen = ((optype == OP_BIT_AND) | |
1090 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
1091 | Newz(801, dc, needlen + 1, char); | |
1092 | (void)sv_usepvn(sv, dc, needlen); | |
1093 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 1094 | } |
a0d0e21e LW |
1095 | SvCUR_set(sv, len); |
1096 | (void)SvPOK_only(sv); | |
0c57e439 GS |
1097 | if (left_utf || right_utf) { |
1098 | UV duc, luc, ruc; | |
1099 | STRLEN lulen = leftlen; | |
1100 | STRLEN rulen = rightlen; | |
1101 | STRLEN dulen = 0; | |
1102 | I32 ulen; | |
1103 | ||
6b7c0e6e GS |
1104 | if (optype != OP_BIT_AND) |
1105 | dc = SvGROW(sv, leftlen+rightlen+1); | |
1106 | ||
0c57e439 GS |
1107 | switch (optype) { |
1108 | case OP_BIT_AND: | |
1109 | while (lulen && rulen) { | |
1110 | luc = utf8_to_uv((U8*)lc, &ulen); | |
1111 | lc += ulen; | |
1112 | lulen -= ulen; | |
1113 | ruc = utf8_to_uv((U8*)rc, &ulen); | |
1114 | rc += ulen; | |
1115 | rulen -= ulen; | |
1116 | duc = luc & ruc; | |
1117 | dc = (char*)uv_to_utf8((U8*)dc, duc); | |
1118 | } | |
1119 | dulen = dc - SvPVX(sv); | |
1120 | SvCUR_set(sv, dulen); | |
1121 | break; | |
1122 | case OP_BIT_XOR: | |
1123 | while (lulen && rulen) { | |
1124 | luc = utf8_to_uv((U8*)lc, &ulen); | |
1125 | lc += ulen; | |
1126 | lulen -= ulen; | |
1127 | ruc = utf8_to_uv((U8*)rc, &ulen); | |
1128 | rc += ulen; | |
1129 | rulen -= ulen; | |
1130 | duc = luc ^ ruc; | |
1131 | dc = (char*)uv_to_utf8((U8*)dc, duc); | |
1132 | } | |
1133 | goto mop_up_utf; | |
1134 | case OP_BIT_OR: | |
1135 | while (lulen && rulen) { | |
1136 | luc = utf8_to_uv((U8*)lc, &ulen); | |
1137 | lc += ulen; | |
1138 | lulen -= ulen; | |
1139 | ruc = utf8_to_uv((U8*)rc, &ulen); | |
1140 | rc += ulen; | |
1141 | rulen -= ulen; | |
1142 | duc = luc | ruc; | |
1143 | dc = (char*)uv_to_utf8((U8*)dc, duc); | |
1144 | } | |
1145 | mop_up_utf: | |
1146 | dulen = dc - SvPVX(sv); | |
1147 | SvCUR_set(sv, dulen); | |
1148 | if (rulen) | |
1149 | sv_catpvn(sv, rc, rulen); | |
1150 | else if (lulen) | |
1151 | sv_catpvn(sv, lc, lulen); | |
1152 | else | |
1153 | *SvEND(sv) = '\0'; | |
1154 | break; | |
1155 | } | |
1156 | SvUTF8_on(sv); | |
1157 | goto finish; | |
1158 | } | |
1159 | else | |
79072805 LW |
1160 | #ifdef LIBERAL |
1161 | if (len >= sizeof(long)*4 && | |
1162 | !((long)dc % sizeof(long)) && | |
1163 | !((long)lc % sizeof(long)) && | |
1164 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
1165 | { | |
1166 | I32 remainder = len % (sizeof(long)*4); | |
1167 | len /= (sizeof(long)*4); | |
1168 | ||
1169 | dl = (long*)dc; | |
1170 | ll = (long*)lc; | |
1171 | rl = (long*)rc; | |
1172 | ||
1173 | switch (optype) { | |
1174 | case OP_BIT_AND: | |
1175 | while (len--) { | |
1176 | *dl++ = *ll++ & *rl++; | |
1177 | *dl++ = *ll++ & *rl++; | |
1178 | *dl++ = *ll++ & *rl++; | |
1179 | *dl++ = *ll++ & *rl++; | |
1180 | } | |
1181 | break; | |
a0d0e21e | 1182 | case OP_BIT_XOR: |
79072805 LW |
1183 | while (len--) { |
1184 | *dl++ = *ll++ ^ *rl++; | |
1185 | *dl++ = *ll++ ^ *rl++; | |
1186 | *dl++ = *ll++ ^ *rl++; | |
1187 | *dl++ = *ll++ ^ *rl++; | |
1188 | } | |
1189 | break; | |
1190 | case OP_BIT_OR: | |
1191 | while (len--) { | |
1192 | *dl++ = *ll++ | *rl++; | |
1193 | *dl++ = *ll++ | *rl++; | |
1194 | *dl++ = *ll++ | *rl++; | |
1195 | *dl++ = *ll++ | *rl++; | |
1196 | } | |
1197 | } | |
1198 | ||
1199 | dc = (char*)dl; | |
1200 | lc = (char*)ll; | |
1201 | rc = (char*)rl; | |
1202 | ||
1203 | len = remainder; | |
1204 | } | |
1205 | #endif | |
a0d0e21e | 1206 | { |
a0d0e21e LW |
1207 | switch (optype) { |
1208 | case OP_BIT_AND: | |
1209 | while (len--) | |
1210 | *dc++ = *lc++ & *rc++; | |
1211 | break; | |
1212 | case OP_BIT_XOR: | |
1213 | while (len--) | |
1214 | *dc++ = *lc++ ^ *rc++; | |
1215 | goto mop_up; | |
1216 | case OP_BIT_OR: | |
1217 | while (len--) | |
1218 | *dc++ = *lc++ | *rc++; | |
1219 | mop_up: | |
1220 | len = lensave; | |
1221 | if (rightlen > len) | |
1222 | sv_catpvn(sv, rsave + len, rightlen - len); | |
1223 | else if (leftlen > len) | |
1224 | sv_catpvn(sv, lsave + len, leftlen - len); | |
4633a7c4 LW |
1225 | else |
1226 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1227 | break; |
1228 | } | |
79072805 | 1229 | } |
0c57e439 | 1230 | finish: |
fb73857a | 1231 | SvTAINT(sv); |
79072805 | 1232 | } |
463ee0b2 LW |
1233 | |
1234 | OP * | |
cea2e8a9 | 1235 | Perl_do_kv(pTHX) |
463ee0b2 | 1236 | { |
4e35701f | 1237 | djSP; |
463ee0b2 | 1238 | HV *hv = (HV*)POPs; |
800e9ae0 | 1239 | HV *keys; |
463ee0b2 | 1240 | register HE *entry; |
463ee0b2 | 1241 | SV *tmpstr; |
54310121 | 1242 | I32 gimme = GIMME_V; |
533c011a NIS |
1243 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1244 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
c750a3ec MB |
1245 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
1246 | ||
533c011a | 1247 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e | 1248 | dokeys = dovalues = TRUE; |
463ee0b2 | 1249 | |
85581909 | 1250 | if (!hv) { |
533c011a | 1251 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 SB |
1252 | dTARGET; /* make sure to clear its target here */ |
1253 | if (SvTYPE(TARG) == SVt_PVLV) | |
1254 | LvTARG(TARG) = Nullsv; | |
1255 | PUSHs(TARG); | |
1256 | } | |
463ee0b2 | 1257 | RETURN; |
85581909 | 1258 | } |
748a9306 | 1259 | |
800e9ae0 JP |
1260 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1261 | (void)hv_iterinit(keys); /* always reset iterator regardless */ | |
748a9306 | 1262 | |
54310121 | 1263 | if (gimme == G_VOID) |
aa689395 | 1264 | RETURN; |
1265 | ||
54310121 | 1266 | if (gimme == G_SCALAR) { |
6ee623d5 | 1267 | IV i; |
463ee0b2 LW |
1268 | dTARGET; |
1269 | ||
533c011a | 1270 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 SB |
1271 | if (SvTYPE(TARG) < SVt_PVLV) { |
1272 | sv_upgrade(TARG, SVt_PVLV); | |
1273 | sv_magic(TARG, Nullsv, 'k', Nullch, 0); | |
1274 | } | |
1275 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1276 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1277 | if (LvTARG(TARG)) |
1278 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1279 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1280 | } |
85581909 SB |
1281 | PUSHs(TARG); |
1282 | RETURN; | |
1283 | } | |
1284 | ||
33c27489 | 1285 | if (! SvTIED_mg((SV*)keys, 'P')) |
800e9ae0 | 1286 | i = HvKEYS(keys); |
463ee0b2 LW |
1287 | else { |
1288 | i = 0; | |
463ee0b2 | 1289 | /*SUPPRESS 560*/ |
800e9ae0 | 1290 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1291 | } |
1292 | PUSHi( i ); | |
1293 | RETURN; | |
1294 | } | |
1295 | ||
8ed4b672 | 1296 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1297 | |
463ee0b2 | 1298 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1299 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1300 | SPAGAIN; |
8c2cee6f | 1301 | if (dokeys) |
1302 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ | |
463ee0b2 | 1303 | if (dovalues) { |
463ee0b2 | 1304 | PUTBACK; |
b6429b1b GS |
1305 | tmpstr = realhv ? |
1306 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); | |
cea2e8a9 | 1307 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1308 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1309 | HvMAX(keys)+1, |
1310 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1311 | SPAGAIN; |
46fc3d4c | 1312 | XPUSHs(tmpstr); |
463ee0b2 LW |
1313 | } |
1314 | PUTBACK; | |
1315 | } | |
1316 | return NORMAL; | |
1317 | } | |
4e35701f | 1318 |