Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
4eb8286e | 3 | * Copyright (c) 1991-1999, 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 | ||
18 | #if !defined(NSIG) || defined(M_UNIX) || defined(M_XENIX) | |
19 | #include <signal.h> | |
20 | #endif | |
21 | ||
942e002e | 22 | STATIC I32 |
cea2e8a9 | 23 | S_do_trans_CC_simple(pTHX_ SV *sv) |
79072805 | 24 | { |
11343788 | 25 | dTHR; |
4757a243 LW |
26 | U8 *s; |
27 | U8 *send; | |
28 | I32 matches = 0; | |
463ee0b2 | 29 | STRLEN len; |
4757a243 LW |
30 | short *tbl; |
31 | I32 ch; | |
79072805 | 32 | |
4757a243 LW |
33 | tbl = (short*)cPVOP->op_pv; |
34 | if (!tbl) | |
cea2e8a9 | 35 | Perl_croak(aTHX_ "panic: do_trans"); |
a0ed51b3 | 36 | |
4757a243 LW |
37 | s = (U8*)SvPV(sv, len); |
38 | send = s + len; | |
39 | ||
40 | while (s < send) { | |
41 | if ((ch = tbl[*s]) >= 0) { | |
42 | matches++; | |
43 | *s = ch; | |
79072805 | 44 | } |
4757a243 LW |
45 | s++; |
46 | } | |
47 | SvSETMAGIC(sv); | |
48 | ||
49 | return matches; | |
50 | } | |
51 | ||
942e002e | 52 | STATIC I32 |
cea2e8a9 | 53 | S_do_trans_CC_count(pTHX_ SV *sv) |
4757a243 LW |
54 | { |
55 | dTHR; | |
56 | U8 *s; | |
57 | U8 *send; | |
58 | I32 matches = 0; | |
59 | STRLEN len; | |
60 | short *tbl; | |
61 | ||
62 | tbl = (short*)cPVOP->op_pv; | |
63 | if (!tbl) | |
cea2e8a9 | 64 | Perl_croak(aTHX_ "panic: do_trans"); |
4757a243 LW |
65 | |
66 | s = (U8*)SvPV(sv, len); | |
67 | send = s + len; | |
68 | ||
69 | while (s < send) { | |
70 | if (tbl[*s] >= 0) | |
71 | matches++; | |
72 | s++; | |
73 | } | |
74 | ||
75 | return matches; | |
76 | } | |
77 | ||
942e002e | 78 | STATIC I32 |
cea2e8a9 | 79 | S_do_trans_CC_complex(pTHX_ SV *sv) |
4757a243 LW |
80 | { |
81 | dTHR; | |
82 | U8 *s; | |
83 | U8 *send; | |
84 | U8 *d; | |
85 | I32 matches = 0; | |
86 | STRLEN len; | |
87 | short *tbl; | |
88 | I32 ch; | |
89 | ||
90 | tbl = (short*)cPVOP->op_pv; | |
91 | if (!tbl) | |
cea2e8a9 | 92 | Perl_croak(aTHX_ "panic: do_trans"); |
4757a243 LW |
93 | |
94 | s = (U8*)SvPV(sv, len); | |
95 | send = s + len; | |
96 | ||
97 | d = s; | |
98 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
99 | U8* p = send; | |
100 | ||
101 | while (s < send) { | |
102 | if ((ch = tbl[*s]) >= 0) { | |
103 | *d = ch; | |
104 | matches++; | |
105 | if (p == d - 1 && *p == *d) | |
106 | matches--; | |
a0ed51b3 | 107 | else |
4757a243 | 108 | p = d++; |
a0ed51b3 | 109 | } |
4757a243 LW |
110 | else if (ch == -1) /* -1 is unmapped character */ |
111 | *d++ = *s; /* -2 is delete character */ | |
112 | s++; | |
a0ed51b3 | 113 | } |
4757a243 LW |
114 | } |
115 | else { | |
116 | while (s < send) { | |
117 | if ((ch = tbl[*s]) >= 0) { | |
118 | *d = ch; | |
119 | matches++; | |
120 | d++; | |
a0ed51b3 | 121 | } |
4757a243 LW |
122 | else if (ch == -1) /* -1 is unmapped character */ |
123 | *d++ = *s; /* -2 is delete character */ | |
124 | s++; | |
5d06d08e | 125 | } |
4757a243 LW |
126 | } |
127 | matches += send - d; /* account for disappeared chars */ | |
128 | *d = '\0'; | |
129 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
130 | SvSETMAGIC(sv); | |
131 | ||
132 | return matches; | |
133 | } | |
134 | ||
942e002e | 135 | STATIC I32 |
cea2e8a9 | 136 | S_do_trans_UU_simple(pTHX_ SV *sv) |
4757a243 LW |
137 | { |
138 | dTHR; | |
139 | U8 *s; | |
140 | U8 *send; | |
141 | U8 *d; | |
142 | I32 matches = 0; | |
143 | STRLEN len; | |
144 | ||
145 | SV* rv = (SV*)cSVOP->op_sv; | |
146 | HV* hv = (HV*)SvRV(rv); | |
147 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
148 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
149 | UV extra = none + 1; | |
150 | UV final; | |
151 | UV uv; | |
152 | ||
153 | s = (U8*)SvPV(sv, len); | |
154 | send = s + len; | |
155 | ||
156 | svp = hv_fetch(hv, "FINAL", 5, FALSE); | |
157 | if (svp) | |
158 | final = SvUV(*svp); | |
159 | ||
160 | d = s; | |
161 | while (s < send) { | |
162 | if ((uv = swash_fetch(rv, s)) < none) { | |
163 | s += UTF8SKIP(s); | |
164 | matches++; | |
165 | d = uv_to_utf8(d, uv); | |
166 | } | |
167 | else if (uv == none) { | |
168 | int i; | |
169 | for (i = UTF8SKIP(s); i; i--) | |
170 | *d++ = *s++; | |
171 | } | |
172 | else if (uv == extra) { | |
173 | s += UTF8SKIP(s); | |
174 | matches++; | |
175 | d = uv_to_utf8(d, final); | |
176 | } | |
177 | else | |
178 | s += UTF8SKIP(s); | |
179 | } | |
180 | *d = '\0'; | |
181 | SvCUR_set(sv, d - (U8*)SvPVX(sv)); | |
182 | SvSETMAGIC(sv); | |
183 | ||
184 | return matches; | |
185 | } | |
186 | ||
942e002e | 187 | STATIC I32 |
cea2e8a9 | 188 | S_do_trans_UU_count(pTHX_ SV *sv) |
4757a243 LW |
189 | { |
190 | dTHR; | |
191 | U8 *s; | |
192 | U8 *send; | |
193 | I32 matches = 0; | |
194 | STRLEN len; | |
195 | ||
196 | SV* rv = (SV*)cSVOP->op_sv; | |
197 | HV* hv = (HV*)SvRV(rv); | |
198 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); | |
199 | UV none = svp ? SvUV(*svp) : 0x7fffffff; | |
200 | UV uv; | |
201 | ||
202 | s = (U8*)SvPV(sv, len); | |
203 | send = s + len; | |
204 | ||
205 | while (s < send) { | |
834a4ddd | 206 | if ((uv = swash_fetch(rv, s)) < none) |
4757a243 | 207 | matches++; |
834a4ddd | 208 | s += UTF8SKIP(s); |
4757a243 LW |
209 | } |
210 | ||
211 | return matches; | |
212 | } | |
213 | ||
942e002e | 214 | STATIC I32 |
cea2e8a9 | 215 | S_do_trans_UC_simple(pTHX_ SV *sv) |
4757a243 LW |
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 | ||
942e002e | 267 | STATIC I32 |
cea2e8a9 | 268 | S_do_trans_CU_simple(pTHX_ SV *sv) |
4757a243 LW |
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 LW |
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 | ||
942e002e | 330 | STATIC I32 |
cea2e8a9 | 331 | S_do_trans_UC_trivial(pTHX_ SV *sv) |
4757a243 LW |
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 LW |
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 LW |
357 | return SvCUR(sv); |
358 | } | |
a0ed51b3 | 359 | |
4757a243 | 360 | /* latin-1 to utf-8 */ |
a0ed51b3 | 361 | |
942e002e | 362 | STATIC I32 |
cea2e8a9 | 363 | S_do_trans_CU_trivial(pTHX_ SV *sv) |
4757a243 LW |
364 | { |
365 | dTHR; | |
366 | U8 *s; | |
367 | U8 *send; | |
368 | U8 *d; | |
369 | U8 *dst; | |
370 | I32 matches; | |
371 | STRLEN len; | |
a0ed51b3 | 372 | |
4757a243 LW |
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 | ||
942e002e | 396 | STATIC I32 |
cea2e8a9 | 397 | S_do_trans_UU_complex(pTHX_ SV *sv) |
4757a243 LW |
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 LW |
445 | } |
446 | else { | |
4757a243 LW |
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 LW |
476 | for (i = UTF8SKIP(s); i; --i) |
477 | *d++ = *s++; | |
a0ed51b3 | 478 | } |
4757a243 LW |
479 | else { |
480 | I32 ulen; | |
481 | *d++ = (U8)utf8_to_uv(s, &ulen); | |
482 | s += ulen; | |
a0ed51b3 | 483 | } |
a0ed51b3 | 484 | } |
4757a243 LW |
485 | else { /* must be to_utf only */ |
486 | d = uv_to_utf8(d, s[-1]); | |
487 | } | |
488 | puv = 0xfeedface; | |
489 | continue; | |
a0ed51b3 | 490 | } |
4757a243 LW |
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 LW |
508 | } |
509 | else { | |
4757a243 LW |
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 LW |
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 LW |
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 LW |
550 | else { /* must be to_utf only */ |
551 | d = uv_to_utf8(d, s[-1]); | |
552 | } | |
553 | continue; | |
79072805 | 554 | } |
4757a243 LW |
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 LW |
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 | |
864dbfa3 | 582 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 | 583 | { |
46124e9e | 584 | dTHR; |
4757a243 LW |
585 | STRLEN len; |
586 | ||
587 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) | |
cea2e8a9 | 588 | Perl_croak(aTHX_ PL_no_modify); |
4757a243 LW |
589 | |
590 | (void)SvPV(sv, len); | |
591 | if (!len) | |
592 | return 0; | |
593 | if (!SvPOKp(sv)) | |
594 | (void)SvPV_force(sv, len); | |
595 | (void)SvPOK_only(sv); | |
596 | ||
cea2e8a9 | 597 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 LW |
598 | |
599 | switch (PL_op->op_private & 63) { | |
600 | case 0: | |
601 | return do_trans_CC_simple(sv); | |
602 | ||
603 | case OPpTRANS_FROM_UTF: | |
604 | return do_trans_UC_simple(sv); | |
605 | ||
606 | case OPpTRANS_TO_UTF: | |
607 | return do_trans_CU_simple(sv); | |
608 | ||
609 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF: | |
610 | return do_trans_UU_simple(sv); | |
611 | ||
612 | case OPpTRANS_IDENTICAL: | |
613 | return do_trans_CC_count(sv); | |
614 | ||
615 | case OPpTRANS_FROM_UTF|OPpTRANS_IDENTICAL: | |
616 | return do_trans_UC_trivial(sv); | |
617 | ||
618 | case OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: | |
619 | return do_trans_CU_trivial(sv); | |
620 | ||
621 | case OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL: | |
622 | return do_trans_UU_count(sv); | |
623 | ||
624 | default: | |
625 | if (PL_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)) | |
626 | return do_trans_UU_complex(sv); /* could be UC or CU too */ | |
627 | else | |
628 | return do_trans_CC_complex(sv); | |
79072805 | 629 | } |
79072805 LW |
630 | } |
631 | ||
632 | void | |
864dbfa3 | 633 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 LW |
634 | { |
635 | SV **oldmark = mark; | |
636 | register I32 items = sp - mark; | |
79072805 | 637 | register STRLEN len; |
463ee0b2 LW |
638 | STRLEN delimlen; |
639 | register char *delim = SvPV(del, delimlen); | |
640 | STRLEN tmplen; | |
79072805 LW |
641 | |
642 | mark++; | |
643 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
07f14f54 | 644 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 LW |
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 LW |
648 | SvPV(*mark, tmplen); |
649 | len += tmplen; | |
79072805 LW |
650 | } |
651 | mark++; | |
652 | } | |
653 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
654 | ||
655 | mark = oldmark; | |
db7c17d7 | 656 | items = sp - mark; |
79072805 LW |
657 | ++mark; |
658 | } | |
659 | ||
463ee0b2 | 660 | if (items-- > 0) { |
8990e307 LW |
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 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 | ||
700 | void | |
864dbfa3 | 701 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 LW |
702 | { |
703 | SV *targ = LvTARG(sv); | |
704 | register I32 offset; | |
705 | register I32 size; | |
8990e307 LW |
706 | register unsigned char *s; |
707 | register unsigned long lval; | |
79072805 | 708 | I32 mask; |
a0d0e21e LW |
709 | STRLEN targlen; |
710 | STRLEN len; | |
79072805 | 711 | |
8990e307 LW |
712 | if (!targ) |
713 | return; | |
a0d0e21e | 714 | s = (unsigned char*)SvPV_force(targ, targlen); |
8990e307 | 715 | lval = U_L(SvNV(sv)); |
79072805 LW |
716 | offset = LvTARGOFF(sv); |
717 | size = LvTARGLEN(sv); | |
a0d0e21e LW |
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 LW |
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 LW |
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 | } | |
7bb043c3 | 749 | SvSETMAGIC(targ); |
79072805 LW |
750 | } |
751 | ||
752 | void | |
864dbfa3 | 753 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 754 | { |
463ee0b2 | 755 | STRLEN len; |
a0d0e21e | 756 | char *s; |
c485e607 | 757 | dTHR; |
a0d0e21e | 758 | |
79072805 | 759 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e LW |
760 | register I32 i; |
761 | I32 max; | |
762 | AV* av = (AV*)sv; | |
763 | max = AvFILL(av); | |
764 | for (i = 0; i <= max; i++) { | |
765 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 766 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
767 | do_chop(astr, sv); |
768 | } | |
769 | return; | |
79072805 | 770 | } |
aa854799 | 771 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
772 | HV* hv = (HV*)sv; |
773 | HE* entry; | |
774 | (void)hv_iterinit(hv); | |
775 | /*SUPPRESS 560*/ | |
776 | while (entry = hv_iternext(hv)) | |
777 | do_chop(astr,hv_iterval(hv,entry)); | |
778 | return; | |
79072805 | 779 | } |
aa854799 | 780 | else if (SvREADONLY(sv)) |
cea2e8a9 | 781 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e | 782 | s = SvPV(sv, len); |
748a9306 | 783 | if (len && !SvPOK(sv)) |
a0d0e21e | 784 | s = SvPV_force(sv, len); |
a0ed51b3 LW |
785 | if (IN_UTF8) { |
786 | if (s && len) { | |
787 | char *send = s + len; | |
788 | char *start = s; | |
789 | s = send - 1; | |
790 | while ((*s & 0xc0) == 0x80) | |
791 | --s; | |
0453d815 PM |
792 | if (UTF8SKIP(s) != send - s && ckWARN_d(WARN_UTF8)) |
793 | Perl_warner(aTHX_ WARN_UTF8, "Malformed UTF-8 character"); | |
a0ed51b3 LW |
794 | sv_setpvn(astr, s, send - s); |
795 | *s = '\0'; | |
796 | SvCUR_set(sv, s - start); | |
797 | SvNIOK_off(sv); | |
798 | } | |
799 | else | |
800 | sv_setpvn(astr, "", 0); | |
801 | } | |
802 | else | |
a0d0e21e LW |
803 | if (s && len) { |
804 | s += --len; | |
805 | sv_setpvn(astr, s, 1); | |
806 | *s = '\0'; | |
807 | SvCUR_set(sv, len); | |
808 | SvNIOK_off(sv); | |
79072805 LW |
809 | } |
810 | else | |
a0d0e21e LW |
811 | sv_setpvn(astr, "", 0); |
812 | SvSETMAGIC(sv); | |
813 | } | |
814 | ||
815 | I32 | |
864dbfa3 | 816 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 817 | { |
aeea060c | 818 | dTHR; |
c07a80fd | 819 | register I32 count; |
a0d0e21e LW |
820 | STRLEN len; |
821 | char *s; | |
c07a80fd | 822 | |
3280af22 | 823 | if (RsSNARF(PL_rs)) |
c07a80fd | 824 | return 0; |
4c5a6083 GS |
825 | if (RsRECORD(PL_rs)) |
826 | return 0; | |
c07a80fd | 827 | count = 0; |
a0d0e21e LW |
828 | if (SvTYPE(sv) == SVt_PVAV) { |
829 | register I32 i; | |
830 | I32 max; | |
831 | AV* av = (AV*)sv; | |
832 | max = AvFILL(av); | |
833 | for (i = 0; i <= max; i++) { | |
834 | sv = (SV*)av_fetch(av, i, FALSE); | |
3280af22 | 835 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
836 | count += do_chomp(sv); |
837 | } | |
838 | return count; | |
839 | } | |
aa854799 | 840 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e LW |
841 | HV* hv = (HV*)sv; |
842 | HE* entry; | |
843 | (void)hv_iterinit(hv); | |
844 | /*SUPPRESS 560*/ | |
845 | while (entry = hv_iternext(hv)) | |
846 | count += do_chomp(hv_iterval(hv,entry)); | |
847 | return count; | |
848 | } | |
aa854799 | 849 | else if (SvREADONLY(sv)) |
cea2e8a9 | 850 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e LW |
851 | s = SvPV(sv, len); |
852 | if (len && !SvPOKp(sv)) | |
853 | s = SvPV_force(sv, len); | |
854 | if (s && len) { | |
855 | s += --len; | |
3280af22 | 856 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
857 | if (*s != '\n') |
858 | goto nope; | |
859 | ++count; | |
860 | while (len && s[-1] == '\n') { | |
861 | --len; | |
862 | --s; | |
863 | ++count; | |
864 | } | |
865 | } | |
a0d0e21e | 866 | else { |
c07a80fd | 867 | STRLEN rslen; |
3280af22 | 868 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd | 869 | if (rslen == 1) { |
870 | if (*s != *rsptr) | |
871 | goto nope; | |
872 | ++count; | |
873 | } | |
874 | else { | |
8c2cee6f | 875 | if (len < rslen - 1) |
c07a80fd | 876 | goto nope; |
877 | len -= rslen - 1; | |
878 | s -= rslen - 1; | |
36477c24 | 879 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 880 | goto nope; |
881 | count += rslen; | |
882 | } | |
a0d0e21e | 883 | } |
a0d0e21e LW |
884 | *s = '\0'; |
885 | SvCUR_set(sv, len); | |
886 | SvNIOK_off(sv); | |
887 | } | |
888 | nope: | |
889 | SvSETMAGIC(sv); | |
890 | return count; | |
891 | } | |
79072805 LW |
892 | |
893 | void | |
864dbfa3 | 894 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 | 895 | { |
aeea060c | 896 | dTHR; /* just for taint */ |
79072805 LW |
897 | #ifdef LIBERAL |
898 | register long *dl; | |
899 | register long *ll; | |
900 | register long *rl; | |
901 | #endif | |
902 | register char *dc; | |
463ee0b2 LW |
903 | STRLEN leftlen; |
904 | STRLEN rightlen; | |
7a4c00b4 | 905 | register char *lc; |
906 | register char *rc; | |
79072805 | 907 | register I32 len; |
a0d0e21e | 908 | I32 lensave; |
7a4c00b4 | 909 | char *lsave; |
910 | char *rsave; | |
79072805 | 911 | |
1fbd88dc CS |
912 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
913 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ | |
7a4c00b4 | 914 | lsave = lc = SvPV(left, leftlen); |
915 | rsave = rc = SvPV(right, rightlen); | |
93a17b20 | 916 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 917 | lensave = len; |
7a4c00b4 | 918 | if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2d8e6c8d GS |
919 | STRLEN n_a; |
920 | dc = SvPV_force(sv, n_a); | |
ff68c719 | 921 | if (SvCUR(sv) < len) { |
922 | dc = SvGROW(sv, len + 1); | |
923 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); | |
924 | } | |
925 | } | |
926 | else { | |
927 | I32 needlen = ((optype == OP_BIT_AND) | |
928 | ? len : (leftlen > rightlen ? leftlen : rightlen)); | |
929 | Newz(801, dc, needlen + 1, char); | |
930 | (void)sv_usepvn(sv, dc, needlen); | |
931 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ | |
79072805 | 932 | } |
a0d0e21e LW |
933 | SvCUR_set(sv, len); |
934 | (void)SvPOK_only(sv); | |
79072805 LW |
935 | #ifdef LIBERAL |
936 | if (len >= sizeof(long)*4 && | |
937 | !((long)dc % sizeof(long)) && | |
938 | !((long)lc % sizeof(long)) && | |
939 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ | |
940 | { | |
941 | I32 remainder = len % (sizeof(long)*4); | |
942 | len /= (sizeof(long)*4); | |
943 | ||
944 | dl = (long*)dc; | |
945 | ll = (long*)lc; | |
946 | rl = (long*)rc; | |
947 | ||
948 | switch (optype) { | |
949 | case OP_BIT_AND: | |
950 | while (len--) { | |
951 | *dl++ = *ll++ & *rl++; | |
952 | *dl++ = *ll++ & *rl++; | |
953 | *dl++ = *ll++ & *rl++; | |
954 | *dl++ = *ll++ & *rl++; | |
955 | } | |
956 | break; | |
a0d0e21e | 957 | case OP_BIT_XOR: |
79072805 LW |
958 | while (len--) { |
959 | *dl++ = *ll++ ^ *rl++; | |
960 | *dl++ = *ll++ ^ *rl++; | |
961 | *dl++ = *ll++ ^ *rl++; | |
962 | *dl++ = *ll++ ^ *rl++; | |
963 | } | |
964 | break; | |
965 | case OP_BIT_OR: | |
966 | while (len--) { | |
967 | *dl++ = *ll++ | *rl++; | |
968 | *dl++ = *ll++ | *rl++; | |
969 | *dl++ = *ll++ | *rl++; | |
970 | *dl++ = *ll++ | *rl++; | |
971 | } | |
972 | } | |
973 | ||
974 | dc = (char*)dl; | |
975 | lc = (char*)ll; | |
976 | rc = (char*)rl; | |
977 | ||
978 | len = remainder; | |
979 | } | |
980 | #endif | |
a0d0e21e | 981 | { |
a0d0e21e LW |
982 | switch (optype) { |
983 | case OP_BIT_AND: | |
984 | while (len--) | |
985 | *dc++ = *lc++ & *rc++; | |
986 | break; | |
987 | case OP_BIT_XOR: | |
988 | while (len--) | |
989 | *dc++ = *lc++ ^ *rc++; | |
990 | goto mop_up; | |
991 | case OP_BIT_OR: | |
992 | while (len--) | |
993 | *dc++ = *lc++ | *rc++; | |
994 | mop_up: | |
995 | len = lensave; | |
996 | if (rightlen > len) | |
997 | sv_catpvn(sv, rsave + len, rightlen - len); | |
998 | else if (leftlen > len) | |
999 | sv_catpvn(sv, lsave + len, leftlen - len); | |
4633a7c4 LW |
1000 | else |
1001 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1002 | break; |
1003 | } | |
79072805 | 1004 | } |
fb73857a | 1005 | SvTAINT(sv); |
79072805 | 1006 | } |
463ee0b2 LW |
1007 | |
1008 | OP * | |
cea2e8a9 | 1009 | Perl_do_kv(pTHX) |
463ee0b2 | 1010 | { |
4e35701f | 1011 | djSP; |
463ee0b2 | 1012 | HV *hv = (HV*)POPs; |
800e9ae0 | 1013 | HV *keys; |
463ee0b2 | 1014 | register HE *entry; |
463ee0b2 | 1015 | SV *tmpstr; |
54310121 | 1016 | I32 gimme = GIMME_V; |
533c011a NIS |
1017 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1018 | I32 dovalues = (PL_op->op_type == OP_VALUES); | |
c750a3ec MB |
1019 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
1020 | ||
533c011a | 1021 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e | 1022 | dokeys = dovalues = TRUE; |
463ee0b2 | 1023 | |
85581909 | 1024 | if (!hv) { |
533c011a | 1025 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 SB |
1026 | dTARGET; /* make sure to clear its target here */ |
1027 | if (SvTYPE(TARG) == SVt_PVLV) | |
1028 | LvTARG(TARG) = Nullsv; | |
1029 | PUSHs(TARG); | |
1030 | } | |
463ee0b2 | 1031 | RETURN; |
85581909 | 1032 | } |
748a9306 | 1033 | |
800e9ae0 JP |
1034 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1035 | (void)hv_iterinit(keys); /* always reset iterator regardless */ | |
748a9306 | 1036 | |
54310121 | 1037 | if (gimme == G_VOID) |
aa689395 | 1038 | RETURN; |
1039 | ||
54310121 | 1040 | if (gimme == G_SCALAR) { |
6ee623d5 | 1041 | IV i; |
463ee0b2 LW |
1042 | dTARGET; |
1043 | ||
533c011a | 1044 | if (PL_op->op_flags & OPf_MOD) { /* lvalue */ |
85581909 SB |
1045 | if (SvTYPE(TARG) < SVt_PVLV) { |
1046 | sv_upgrade(TARG, SVt_PVLV); | |
1047 | sv_magic(TARG, Nullsv, 'k', Nullch, 0); | |
1048 | } | |
1049 | LvTYPE(TARG) = 'k'; | |
800e9ae0 | 1050 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 GS |
1051 | if (LvTARG(TARG)) |
1052 | SvREFCNT_dec(LvTARG(TARG)); | |
800e9ae0 | 1053 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 | 1054 | } |
85581909 SB |
1055 | PUSHs(TARG); |
1056 | RETURN; | |
1057 | } | |
1058 | ||
33c27489 | 1059 | if (! SvTIED_mg((SV*)keys, 'P')) |
800e9ae0 | 1060 | i = HvKEYS(keys); |
463ee0b2 LW |
1061 | else { |
1062 | i = 0; | |
463ee0b2 | 1063 | /*SUPPRESS 560*/ |
800e9ae0 | 1064 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1065 | } |
1066 | PUSHi( i ); | |
1067 | RETURN; | |
1068 | } | |
1069 | ||
8ed4b672 | 1070 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1071 | |
463ee0b2 | 1072 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
800e9ae0 | 1073 | while (entry = hv_iternext(keys)) { |
463ee0b2 | 1074 | SPAGAIN; |
8c2cee6f | 1075 | if (dokeys) |
1076 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ | |
463ee0b2 | 1077 | if (dovalues) { |
463ee0b2 | 1078 | PUTBACK; |
b6429b1b GS |
1079 | tmpstr = realhv ? |
1080 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); | |
cea2e8a9 | 1081 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1082 | (unsigned long)HeHASH(entry), |
800e9ae0 JP |
1083 | HvMAX(keys)+1, |
1084 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); | |
463ee0b2 | 1085 | SPAGAIN; |
46fc3d4c | 1086 | XPUSHs(tmpstr); |
463ee0b2 LW |
1087 | } |
1088 | PUTBACK; | |
1089 | } | |
1090 | return NORMAL; | |
1091 | } | |
4e35701f | 1092 |