Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
bc89e66f |
3 | * Copyright (c) 1991-2001, 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" |
864dbfa3 |
15 | #define PERL_IN_DOOP_C |
79072805 |
16 | #include "perl.h" |
17 | |
64ca3a65 |
18 | #ifndef PERL_MICRO |
79072805 |
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 |
b250498f |
25 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 |
26 | { |
4757a243 |
27 | U8 *s; |
b250498f |
28 | U8 *d; |
4757a243 |
29 | U8 *send; |
b250498f |
30 | U8 *dstart; |
4757a243 |
31 | I32 matches = 0; |
9b877dbb |
32 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
33 | STRLEN len; |
4757a243 |
34 | short *tbl; |
35 | I32 ch; |
79072805 |
36 | |
4757a243 |
37 | tbl = (short*)cPVOP->op_pv; |
38 | if (!tbl) |
7d85a32c |
39 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
40 | |
4757a243 |
41 | s = (U8*)SvPV(sv, len); |
42 | send = s + len; |
43 | |
b250498f |
44 | /* First, take care of non-UTF8 input strings, because they're easy */ |
1aa99e6b |
45 | if (!SvUTF8(sv)) { |
01ec43d0 |
46 | while (s < send) { |
036b4402 |
47 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 |
48 | matches++; |
49 | *s++ = ch; |
50 | } |
51 | else |
52 | s++; |
53 | } |
54 | SvSETMAGIC(sv); |
b250498f |
55 | return matches; |
56 | } |
4757a243 |
57 | |
b250498f |
58 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
59 | if (grows) |
60 | New(0, d, len*2+1, U8); |
61 | else |
62 | d = s; |
b250498f |
63 | dstart = d; |
64 | while (s < send) { |
ba210ebe |
65 | STRLEN ulen; |
9b877dbb |
66 | UV c; |
b250498f |
67 | |
b250498f |
68 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 |
69 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb |
70 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
71 | matches++; |
9041c2e3 |
72 | d = uvchr_to_utf8(d, ch); |
b250498f |
73 | s += ulen; |
01ec43d0 |
74 | } |
75 | else { /* No match -> copy */ |
9b877dbb |
76 | Copy(s, d, ulen, U8); |
77 | d += ulen; |
78 | s += ulen; |
b250498f |
79 | } |
80 | } |
9b877dbb |
81 | if (grows) { |
82 | sv_setpvn(sv, (char*)dstart, d - dstart); |
83 | Safefree(dstart); |
84 | } |
85 | else { |
86 | *d = '\0'; |
87 | SvCUR_set(sv, d - dstart); |
88 | } |
b250498f |
89 | SvUTF8_on(sv); |
b250498f |
90 | SvSETMAGIC(sv); |
4757a243 |
91 | return matches; |
92 | } |
93 | |
942e002e |
94 | STATIC I32 |
036b4402 |
95 | S_do_trans_count(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 |
96 | { |
4757a243 |
97 | U8 *s; |
98 | U8 *send; |
99 | I32 matches = 0; |
100 | STRLEN len; |
101 | short *tbl; |
8973db79 |
102 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 |
103 | |
104 | tbl = (short*)cPVOP->op_pv; |
105 | if (!tbl) |
7d85a32c |
106 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
107 | |
108 | s = (U8*)SvPV(sv, len); |
109 | send = s + len; |
110 | |
1aa99e6b |
111 | if (!SvUTF8(sv)) |
112 | while (s < send) { |
113 | if (tbl[*s++] >= 0) |
036b4402 |
114 | matches++; |
1aa99e6b |
115 | } |
116 | else |
117 | while (s < send) { |
118 | UV c; |
119 | STRLEN ulen; |
9041c2e3 |
120 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 |
121 | if (c < 0x100) { |
122 | if (tbl[c] >= 0) |
123 | matches++; |
124 | } else if (complement) |
1aa99e6b |
125 | matches++; |
126 | s += ulen; |
127 | } |
4757a243 |
128 | |
129 | return matches; |
130 | } |
131 | |
942e002e |
132 | STATIC I32 |
b250498f |
133 | S_do_trans_complex(pTHX_ SV *sv)/* SPC - NOT OK */ |
4757a243 |
134 | { |
4757a243 |
135 | U8 *s; |
136 | U8 *send; |
137 | U8 *d; |
5e44153e |
138 | U8 *dstart; |
1aa99e6b |
139 | I32 isutf8; |
4757a243 |
140 | I32 matches = 0; |
9b877dbb |
141 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 |
142 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
143 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
144 | STRLEN len, rlen; |
4757a243 |
145 | short *tbl; |
146 | I32 ch; |
147 | |
148 | tbl = (short*)cPVOP->op_pv; |
149 | if (!tbl) |
7d85a32c |
150 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
151 | |
152 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
153 | isutf8 = SvUTF8(sv); |
4757a243 |
154 | send = s + len; |
155 | |
1aa99e6b |
156 | if (!isutf8) { |
157 | dstart = d = s; |
158 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
159 | U8* p = send; |
160 | while (s < send) { |
161 | if ((ch = tbl[*s]) >= 0) { |
036b4402 |
162 | *d = ch; |
163 | matches++; |
1aa99e6b |
164 | if (p != d - 1 || *p != *d) |
165 | p = d++; |
166 | } |
167 | else if (ch == -1) /* -1 is unmapped character */ |
168 | *d++ = *s; |
169 | else if (ch == -2) /* -2 is delete character */ |
170 | matches++; |
171 | s++; |
172 | } |
a0ed51b3 |
173 | } |
1aa99e6b |
174 | else { |
175 | while (s < send) { |
176 | if ((ch = tbl[*s]) >= 0) { |
177 | matches++; |
178 | *d++ = ch; |
179 | } |
180 | else if (ch == -1) /* -1 is unmapped character */ |
181 | *d++ = *s; |
182 | else if (ch == -2) /* -2 is delete character */ |
183 | matches++; |
184 | s++; |
185 | } |
186 | } |
187 | SvCUR_set(sv, d - dstart); |
4757a243 |
188 | } |
1aa99e6b |
189 | else { /* isutf8 */ |
9b877dbb |
190 | if (grows) |
191 | New(0, d, len*2+1, U8); |
192 | else |
193 | d = s; |
1aa99e6b |
194 | dstart = d; |
8973db79 |
195 | if (complement && !del) |
196 | rlen = tbl[0x100]; |
1aa99e6b |
197 | |
4aecb5b5 |
198 | #ifdef MACOS_TRADITIONAL |
199 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ |
200 | #endif |
201 | |
1aa99e6b |
202 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b |
203 | UV pch = 0xfeedface; |
204 | while (s < send) { |
205 | STRLEN len; |
9041c2e3 |
206 | UV comp = utf8_to_uvchr(s, &len); |
1aa99e6b |
207 | |
8973db79 |
208 | if (comp > 0xff) { |
209 | if (!complement) { |
210 | Copy(s, d, len, U8); |
211 | d += len; |
212 | } |
213 | else { |
214 | matches++; |
215 | if (!del) { |
05d340b8 |
216 | ch = (rlen == 0) ? comp : |
217 | (comp - 0x100 < rlen) ? |
8973db79 |
218 | tbl[comp+1] : tbl[0x100+rlen]; |
219 | if (ch != pch) { |
9041c2e3 |
220 | d = uvchr_to_utf8(d, ch); |
8973db79 |
221 | pch = ch; |
222 | } |
223 | s += len; |
224 | continue; |
225 | } |
226 | } |
9b877dbb |
227 | } |
1aa99e6b |
228 | else if ((ch = tbl[comp]) >= 0) { |
229 | matches++; |
230 | if (ch != pch) { |
9041c2e3 |
231 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
232 | pch = ch; |
233 | } |
234 | s += len; |
235 | continue; |
236 | } |
9b877dbb |
237 | else if (ch == -1) { /* -1 is unmapped character */ |
238 | Copy(s, d, len, U8); |
239 | d += len; |
240 | } |
1aa99e6b |
241 | else if (ch == -2) /* -2 is delete character */ |
242 | matches++; |
243 | s += len; |
244 | pch = 0xfeedface; |
245 | } |
5d06d08e |
246 | } |
1aa99e6b |
247 | else { |
248 | while (s < send) { |
249 | STRLEN len; |
9041c2e3 |
250 | UV comp = utf8_to_uvchr(s, &len); |
8973db79 |
251 | if (comp > 0xff) { |
252 | if (!complement) { |
253 | Copy(s, d, len, U8); |
254 | d += len; |
255 | } |
256 | else { |
257 | matches++; |
258 | if (!del) { |
9041c2e3 |
259 | if (comp - 0x100 < rlen) |
260 | d = uvchr_to_utf8(d, tbl[comp+1]); |
8973db79 |
261 | else |
9041c2e3 |
262 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 |
263 | } |
264 | } |
9b877dbb |
265 | } |
1aa99e6b |
266 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 |
267 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
268 | matches++; |
269 | } |
270 | else if (ch == -1) { /* -1 is unmapped character */ |
9b877dbb |
271 | Copy(s, d, len, U8); |
272 | d += len; |
1aa99e6b |
273 | } |
274 | else if (ch == -2) /* -2 is delete character */ |
275 | matches++; |
276 | s += len; |
277 | } |
278 | } |
9b877dbb |
279 | if (grows) { |
280 | sv_setpvn(sv, (char*)dstart, d - dstart); |
281 | Safefree(dstart); |
282 | } |
283 | else { |
284 | *d = '\0'; |
285 | SvCUR_set(sv, d - dstart); |
286 | } |
1aa99e6b |
287 | SvUTF8_on(sv); |
4757a243 |
288 | } |
5e44153e |
289 | SvSETMAGIC(sv); |
4757a243 |
290 | return matches; |
291 | } |
292 | |
942e002e |
293 | STATIC I32 |
036b4402 |
294 | S_do_trans_simple_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 |
295 | { |
4757a243 |
296 | U8 *s; |
297 | U8 *send; |
298 | U8 *d; |
036b4402 |
299 | U8 *start; |
1aa99e6b |
300 | U8 *dstart, *dend; |
4757a243 |
301 | I32 matches = 0; |
9b877dbb |
302 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
303 | STRLEN len; |
304 | |
305 | SV* rv = (SV*)cSVOP->op_sv; |
306 | HV* hv = (HV*)SvRV(rv); |
307 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
308 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
309 | UV extra = none + 1; |
310 | UV final; |
311 | UV uv; |
1aa99e6b |
312 | I32 isutf8; |
313 | U8 hibit = 0; |
4757a243 |
314 | |
315 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
316 | isutf8 = SvUTF8(sv); |
317 | if (!isutf8) { |
318 | U8 *t = s, *e = s + len; |
c4d5f83a |
319 | while (t < e) { |
320 | U8 ch = *t++; |
321 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
322 | break; |
c4d5f83a |
323 | } |
1aa99e6b |
324 | if (hibit) |
325 | s = bytes_to_utf8(s, &len); |
326 | } |
4757a243 |
327 | send = s + len; |
036b4402 |
328 | start = s; |
4757a243 |
329 | |
330 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
331 | if (svp) |
332 | final = SvUV(*svp); |
333 | |
9b877dbb |
334 | if (grows) { |
335 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
336 | New(0, d, len*3+UTF8_MAXLEN, U8); |
337 | dend = d + len * 3; |
338 | dstart = d; |
339 | } |
340 | else { |
341 | dstart = d = s; |
342 | dend = d + len; |
343 | } |
1aa99e6b |
344 | |
4757a243 |
345 | while (s < send) { |
346 | if ((uv = swash_fetch(rv, s)) < none) { |
347 | s += UTF8SKIP(s); |
348 | matches++; |
7d85a32c |
349 | d = uvuni_to_utf8(d, uv); |
4757a243 |
350 | } |
351 | else if (uv == none) { |
1aa99e6b |
352 | int i = UTF8SKIP(s); |
9b877dbb |
353 | Copy(s, d, i, U8); |
354 | d += i; |
355 | s += i; |
4757a243 |
356 | } |
357 | else if (uv == extra) { |
1aa99e6b |
358 | int i = UTF8SKIP(s); |
036b4402 |
359 | s += i; |
4757a243 |
360 | matches++; |
7d85a32c |
361 | d = uvuni_to_utf8(d, final); |
4757a243 |
362 | } |
363 | else |
364 | s += UTF8SKIP(s); |
1aa99e6b |
365 | |
9b877dbb |
366 | if (d > dend) { |
1aa99e6b |
367 | STRLEN clen = d - dstart; |
368 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
9b877dbb |
369 | if (!grows) |
7d85a32c |
370 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
1aa99e6b |
371 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
372 | d = dstart + clen; |
373 | dend = dstart + nlen; |
374 | } |
4757a243 |
375 | } |
16ec844d |
376 | if (grows || hibit) { |
9b877dbb |
377 | sv_setpvn(sv, (char*)dstart, d - dstart); |
378 | Safefree(dstart); |
16ec844d |
379 | if (grows && hibit) |
380 | Safefree(start); |
9b877dbb |
381 | } |
382 | else { |
383 | *d = '\0'; |
384 | SvCUR_set(sv, d - dstart); |
385 | } |
4757a243 |
386 | SvSETMAGIC(sv); |
1aa99e6b |
387 | SvUTF8_on(sv); |
9041c2e3 |
388 | /* Downgrading just 'cos it will is suspect - NI-S */ |
1aa99e6b |
389 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
390 | sv_utf8_downgrade(sv, TRUE); |
4757a243 |
391 | |
392 | return matches; |
393 | } |
394 | |
942e002e |
395 | STATIC I32 |
036b4402 |
396 | S_do_trans_count_utf8(pTHX_ SV *sv)/* SPC - OK */ |
4757a243 |
397 | { |
4757a243 |
398 | U8 *s; |
1aa99e6b |
399 | U8 *start, *send; |
4757a243 |
400 | I32 matches = 0; |
401 | STRLEN len; |
402 | |
403 | SV* rv = (SV*)cSVOP->op_sv; |
404 | HV* hv = (HV*)SvRV(rv); |
405 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
406 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
45005bfb |
407 | UV extra = none + 1; |
4757a243 |
408 | UV uv; |
1aa99e6b |
409 | U8 hibit = 0; |
4757a243 |
410 | |
411 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
412 | if (!SvUTF8(sv)) { |
413 | U8 *t = s, *e = s + len; |
c4d5f83a |
414 | while (t < e) { |
415 | U8 ch = *t++; |
416 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
417 | break; |
c4d5f83a |
418 | } |
1aa99e6b |
419 | if (hibit) |
420 | start = s = bytes_to_utf8(s, &len); |
421 | } |
4757a243 |
422 | send = s + len; |
423 | |
424 | while (s < send) { |
45005bfb |
425 | if ((uv = swash_fetch(rv, s)) < none || uv == extra) |
4757a243 |
426 | matches++; |
834a4ddd |
427 | s += UTF8SKIP(s); |
4757a243 |
428 | } |
1aa99e6b |
429 | if (hibit) |
430 | Safefree(start); |
4757a243 |
431 | |
432 | return matches; |
433 | } |
434 | |
942e002e |
435 | STATIC I32 |
036b4402 |
436 | S_do_trans_complex_utf8(pTHX_ SV *sv) /* SPC - NOT OK */ |
4757a243 |
437 | { |
4757a243 |
438 | U8 *s; |
1aa99e6b |
439 | U8 *start, *send; |
4757a243 |
440 | U8 *d; |
441 | I32 matches = 0; |
442 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
4757a243 |
443 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb |
444 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
445 | SV* rv = (SV*)cSVOP->op_sv; |
446 | HV* hv = (HV*)SvRV(rv); |
447 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
448 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
449 | UV extra = none + 1; |
450 | UV final; |
45005bfb |
451 | bool havefinal = FALSE; |
4757a243 |
452 | UV uv; |
453 | STRLEN len; |
1aa99e6b |
454 | U8 *dstart, *dend; |
455 | I32 isutf8; |
456 | U8 hibit = 0; |
4757a243 |
457 | |
458 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
459 | isutf8 = SvUTF8(sv); |
460 | if (!isutf8) { |
461 | U8 *t = s, *e = s + len; |
c4d5f83a |
462 | while (t < e) { |
463 | U8 ch = *t++; |
464 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
465 | break; |
c4d5f83a |
466 | } |
1aa99e6b |
467 | if (hibit) |
468 | s = bytes_to_utf8(s, &len); |
469 | } |
4757a243 |
470 | send = s + len; |
1aa99e6b |
471 | start = s; |
4757a243 |
472 | |
473 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
45005bfb |
474 | if (svp) { |
4757a243 |
475 | final = SvUV(*svp); |
45005bfb |
476 | havefinal = TRUE; |
477 | } |
4757a243 |
478 | |
9b877dbb |
479 | if (grows) { |
480 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
481 | New(0, d, len*3+UTF8_MAXLEN, U8); |
482 | dend = d + len * 3; |
483 | dstart = d; |
484 | } |
485 | else { |
486 | dstart = d = s; |
487 | dend = d + len; |
488 | } |
4757a243 |
489 | |
490 | if (squash) { |
491 | UV puv = 0xfeedface; |
492 | while (s < send) { |
1aa99e6b |
493 | uv = swash_fetch(rv, s); |
9041c2e3 |
494 | |
9b877dbb |
495 | if (d > dend) { |
496 | STRLEN clen = d - dstart; |
497 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
498 | if (!grows) |
7d85a32c |
499 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
500 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
501 | d = dstart + clen; |
502 | dend = dstart + nlen; |
4757a243 |
503 | } |
504 | if (uv < none) { |
505 | matches++; |
94472101 |
506 | s += UTF8SKIP(s); |
4757a243 |
507 | if (uv != puv) { |
7d85a32c |
508 | d = uvuni_to_utf8(d, uv); |
4757a243 |
509 | puv = uv; |
510 | } |
4757a243 |
511 | continue; |
512 | } |
513 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
514 | int i = UTF8SKIP(s); |
9b877dbb |
515 | Copy(s, d, i, U8); |
516 | d += i; |
517 | s += i; |
4757a243 |
518 | puv = 0xfeedface; |
519 | continue; |
a0ed51b3 |
520 | } |
4757a243 |
521 | else if (uv == extra && !del) { |
522 | matches++; |
45005bfb |
523 | if (havefinal) { |
524 | s += UTF8SKIP(s); |
525 | if (puv != final) { |
7d85a32c |
526 | d = uvuni_to_utf8(d, final); |
45005bfb |
527 | puv = final; |
528 | } |
529 | } |
530 | else { |
531 | STRLEN len; |
7d85a32c |
532 | uv = utf8_to_uvuni(s, &len); |
45005bfb |
533 | if (uv != puv) { |
534 | Copy(s, d, len, U8); |
535 | d += len; |
536 | puv = uv; |
537 | } |
538 | s += len; |
4757a243 |
539 | } |
4757a243 |
540 | continue; |
541 | } |
01ec43d0 |
542 | matches++; /* "none+1" is delete character */ |
543 | s += UTF8SKIP(s); |
a0ed51b3 |
544 | } |
79072805 |
545 | } |
546 | else { |
4757a243 |
547 | while (s < send) { |
1aa99e6b |
548 | uv = swash_fetch(rv, s); |
9b877dbb |
549 | if (d > dend) { |
550 | STRLEN clen = d - dstart; |
551 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
552 | if (!grows) |
7d85a32c |
553 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
554 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
555 | d = dstart + clen; |
556 | dend = dstart + nlen; |
a0ed51b3 |
557 | } |
4757a243 |
558 | if (uv < none) { |
559 | matches++; |
01ec43d0 |
560 | s += UTF8SKIP(s); |
7d85a32c |
561 | d = uvuni_to_utf8(d, uv); |
4757a243 |
562 | continue; |
a0ed51b3 |
563 | } |
4757a243 |
564 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
565 | int i = UTF8SKIP(s); |
9b877dbb |
566 | Copy(s, d, i, U8); |
567 | d += i; |
568 | s += i; |
4757a243 |
569 | continue; |
79072805 |
570 | } |
4757a243 |
571 | else if (uv == extra && !del) { |
572 | matches++; |
01ec43d0 |
573 | s += UTF8SKIP(s); |
7d85a32c |
574 | d = uvuni_to_utf8(d, final); |
4757a243 |
575 | continue; |
576 | } |
01ec43d0 |
577 | matches++; /* "none+1" is delete character */ |
578 | s += UTF8SKIP(s); |
79072805 |
579 | } |
4757a243 |
580 | } |
16ec844d |
581 | if (grows || hibit) { |
9b877dbb |
582 | sv_setpvn(sv, (char*)dstart, d - dstart); |
583 | Safefree(dstart); |
16ec844d |
584 | if (grows && hibit) |
585 | Safefree(start); |
9b877dbb |
586 | } |
587 | else { |
588 | *d = '\0'; |
589 | SvCUR_set(sv, d - dstart); |
590 | } |
1aa99e6b |
591 | SvUTF8_on(sv); |
1aa99e6b |
592 | if (!isutf8 && !(PL_hints & HINT_UTF8)) |
593 | sv_utf8_downgrade(sv, TRUE); |
4757a243 |
594 | SvSETMAGIC(sv); |
595 | |
596 | return matches; |
597 | } |
598 | |
599 | I32 |
864dbfa3 |
600 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
601 | { |
602 | STRLEN len; |
8e84507e |
603 | I32 hasutf = (PL_op->op_private & |
036b4402 |
604 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
605 | |
606 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
cea2e8a9 |
607 | Perl_croak(aTHX_ PL_no_modify); |
4757a243 |
608 | |
609 | (void)SvPV(sv, len); |
610 | if (!len) |
611 | return 0; |
612 | if (!SvPOKp(sv)) |
613 | (void)SvPV_force(sv, len); |
2de7b02f |
614 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) |
615 | (void)SvPOK_only_UTF8(sv); |
4757a243 |
616 | |
cea2e8a9 |
617 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
618 | |
036b4402 |
619 | switch (PL_op->op_private & ~hasutf & 63) { |
4757a243 |
620 | case 0: |
01ec43d0 |
621 | if (hasutf) |
622 | return do_trans_simple_utf8(sv); |
623 | else |
624 | return do_trans_simple(sv); |
4757a243 |
625 | |
626 | case OPpTRANS_IDENTICAL: |
05d340b8 |
627 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
628 | if (hasutf) |
629 | return do_trans_count_utf8(sv); |
630 | else |
631 | return do_trans_count(sv); |
4757a243 |
632 | |
633 | default: |
01ec43d0 |
634 | if (hasutf) |
036b4402 |
635 | return do_trans_complex_utf8(sv); |
4757a243 |
636 | else |
036b4402 |
637 | return do_trans_complex(sv); |
79072805 |
638 | } |
79072805 |
639 | } |
640 | |
641 | void |
864dbfa3 |
642 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
643 | { |
644 | SV **oldmark = mark; |
645 | register I32 items = sp - mark; |
79072805 |
646 | register STRLEN len; |
463ee0b2 |
647 | STRLEN delimlen; |
648 | register char *delim = SvPV(del, delimlen); |
649 | STRLEN tmplen; |
79072805 |
650 | |
651 | mark++; |
652 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
07f14f54 |
653 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 |
654 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
655 | while (items-- > 0) { |
1426bbf4 |
656 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 |
657 | SvPV(*mark, tmplen); |
658 | len += tmplen; |
79072805 |
659 | } |
660 | mark++; |
661 | } |
662 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
663 | |
664 | mark = oldmark; |
db7c17d7 |
665 | items = sp - mark; |
79072805 |
666 | ++mark; |
667 | } |
668 | |
463ee0b2 |
669 | if (items-- > 0) { |
92d29cee |
670 | sv_setpv(sv, ""); |
671 | if (*mark) |
672 | sv_catsv(sv, *mark); |
463ee0b2 |
673 | mark++; |
674 | } |
79072805 |
675 | else |
676 | sv_setpv(sv,""); |
c512ce4f |
677 | if (delimlen) { |
79072805 |
678 | for (; items > 0; items--,mark++) { |
c512ce4f |
679 | sv_catsv(sv,del); |
79072805 |
680 | sv_catsv(sv,*mark); |
681 | } |
682 | } |
683 | else { |
684 | for (; items > 0; items--,mark++) |
685 | sv_catsv(sv,*mark); |
686 | } |
687 | SvSETMAGIC(sv); |
688 | } |
689 | |
690 | void |
864dbfa3 |
691 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
692 | { |
46fc3d4c |
693 | STRLEN patlen; |
694 | char *pat = SvPV(*sarg, patlen); |
695 | bool do_taint = FALSE; |
696 | |
697 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
698 | SvSETMAGIC(sv); |
46fc3d4c |
699 | if (do_taint) |
700 | SvTAINTED_on(sv); |
79072805 |
701 | } |
702 | |
33b45480 |
703 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
704 | UV |
705 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
706 | { |
707 | STRLEN srclen, len; |
708 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); |
709 | UV retnum = 0; |
710 | |
a50d7633 |
711 | if (offset < 0) |
81e118e0 |
712 | return retnum; |
8e84507e |
713 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
714 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
715 | |
dcad2880 |
716 | if (SvUTF8(sv)) |
33b45480 |
717 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
718 | |
81e118e0 |
719 | offset *= size; /* turn into bit offset */ |
720 | len = (offset + size + 7) / 8; /* required number of bytes */ |
721 | if (len > srclen) { |
722 | if (size <= 8) |
723 | retnum = 0; |
724 | else { |
725 | offset >>= 3; /* turn into byte offset */ |
726 | if (size == 16) { |
727 | if (offset >= srclen) |
728 | retnum = 0; |
729 | else |
628e1a40 |
730 | retnum = (UV) s[offset] << 8; |
81e118e0 |
731 | } |
732 | else if (size == 32) { |
733 | if (offset >= srclen) |
734 | retnum = 0; |
735 | else if (offset + 1 >= srclen) |
736 | retnum = |
737 | ((UV) s[offset ] << 24); |
738 | else if (offset + 2 >= srclen) |
739 | retnum = |
740 | ((UV) s[offset ] << 24) + |
741 | ((UV) s[offset + 1] << 16); |
742 | else |
743 | retnum = |
744 | ((UV) s[offset ] << 24) + |
745 | ((UV) s[offset + 1] << 16) + |
746 | ( s[offset + 2] << 8); |
747 | } |
d7d93a81 |
748 | #ifdef UV_IS_QUAD |
c5a0f51a |
749 | else if (size == 64) { |
c5a0f51a |
750 | if (ckWARN(WARN_PORTABLE)) |
751 | Perl_warner(aTHX_ WARN_PORTABLE, |
752 | "Bit vector size > 32 non-portable"); |
753 | if (offset >= srclen) |
754 | retnum = 0; |
755 | else if (offset + 1 >= srclen) |
756 | retnum = |
757 | (UV) s[offset ] << 56; |
758 | else if (offset + 2 >= srclen) |
759 | retnum = |
760 | ((UV) s[offset ] << 56) + |
761 | ((UV) s[offset + 1] << 48); |
762 | else if (offset + 3 >= srclen) |
763 | retnum = |
764 | ((UV) s[offset ] << 56) + |
765 | ((UV) s[offset + 1] << 48) + |
766 | ((UV) s[offset + 2] << 40); |
767 | else if (offset + 4 >= srclen) |
768 | retnum = |
769 | ((UV) s[offset ] << 56) + |
770 | ((UV) s[offset + 1] << 48) + |
771 | ((UV) s[offset + 2] << 40) + |
772 | ((UV) s[offset + 3] << 32); |
773 | else if (offset + 5 >= srclen) |
774 | retnum = |
775 | ((UV) s[offset ] << 56) + |
776 | ((UV) s[offset + 1] << 48) + |
777 | ((UV) s[offset + 2] << 40) + |
778 | ((UV) s[offset + 3] << 32) + |
779 | ( s[offset + 4] << 24); |
780 | else if (offset + 6 >= srclen) |
781 | retnum = |
782 | ((UV) s[offset ] << 56) + |
783 | ((UV) s[offset + 1] << 48) + |
784 | ((UV) s[offset + 2] << 40) + |
785 | ((UV) s[offset + 3] << 32) + |
786 | ((UV) s[offset + 4] << 24) + |
787 | ((UV) s[offset + 5] << 16); |
788 | else |
8e84507e |
789 | retnum = |
c5a0f51a |
790 | ((UV) s[offset ] << 56) + |
791 | ((UV) s[offset + 1] << 48) + |
792 | ((UV) s[offset + 2] << 40) + |
793 | ((UV) s[offset + 3] << 32) + |
794 | ((UV) s[offset + 4] << 24) + |
795 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
796 | ( s[offset + 6] << 8); |
c5a0f51a |
797 | } |
798 | #endif |
81e118e0 |
799 | } |
800 | } |
801 | else if (size < 8) |
802 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
803 | else { |
804 | offset >>= 3; /* turn into byte offset */ |
805 | if (size == 8) |
806 | retnum = s[offset]; |
807 | else if (size == 16) |
808 | retnum = |
628e1a40 |
809 | ((UV) s[offset] << 8) + |
81e118e0 |
810 | s[offset + 1]; |
811 | else if (size == 32) |
812 | retnum = |
813 | ((UV) s[offset ] << 24) + |
814 | ((UV) s[offset + 1] << 16) + |
815 | ( s[offset + 2] << 8) + |
816 | s[offset + 3]; |
d7d93a81 |
817 | #ifdef UV_IS_QUAD |
c5a0f51a |
818 | else if (size == 64) { |
c5a0f51a |
819 | if (ckWARN(WARN_PORTABLE)) |
820 | Perl_warner(aTHX_ WARN_PORTABLE, |
821 | "Bit vector size > 32 non-portable"); |
822 | retnum = |
823 | ((UV) s[offset ] << 56) + |
824 | ((UV) s[offset + 1] << 48) + |
825 | ((UV) s[offset + 2] << 40) + |
826 | ((UV) s[offset + 3] << 32) + |
827 | ((UV) s[offset + 4] << 24) + |
828 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
829 | ( s[offset + 6] << 8) + |
c5a0f51a |
830 | s[offset + 7]; |
831 | } |
832 | #endif |
81e118e0 |
833 | } |
834 | |
835 | return retnum; |
836 | } |
837 | |
33b45480 |
838 | /* currently converts input to bytes if possible but doesn't sweat failures, |
839 | * although it does ensure that the string it clobbers is not marked as |
840 | * utf8-valid any more |
841 | */ |
79072805 |
842 | void |
864dbfa3 |
843 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
844 | { |
845 | SV *targ = LvTARG(sv); |
846 | register I32 offset; |
847 | register I32 size; |
8990e307 |
848 | register unsigned char *s; |
81e118e0 |
849 | register UV lval; |
79072805 |
850 | I32 mask; |
a0d0e21e |
851 | STRLEN targlen; |
852 | STRLEN len; |
79072805 |
853 | |
8990e307 |
854 | if (!targ) |
855 | return; |
a0d0e21e |
856 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
857 | if (SvUTF8(targ)) { |
33b45480 |
858 | /* This is handled by the SvPOK_only below... |
859 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
860 | SvUTF8_off(targ); |
861 | */ |
862 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
863 | } |
864 | |
4ebbc975 |
865 | (void)SvPOK_only(targ); |
81e118e0 |
866 | lval = SvUV(sv); |
79072805 |
867 | offset = LvTARGOFF(sv); |
fe58ced6 |
868 | if (offset < 0) |
869 | Perl_croak(aTHX_ "Assigning to negative offset in vec"); |
79072805 |
870 | size = LvTARGLEN(sv); |
8e84507e |
871 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
872 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
873 | |
81e118e0 |
874 | offset *= size; /* turn into bit offset */ |
875 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
876 | if (len > targlen) { |
877 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
878 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
879 | SvCUR_set(targ, len); |
880 | } |
8e84507e |
881 | |
79072805 |
882 | if (size < 8) { |
883 | mask = (1 << size) - 1; |
884 | size = offset & 7; |
885 | lval &= mask; |
81e118e0 |
886 | offset >>= 3; /* turn into byte offset */ |
79072805 |
887 | s[offset] &= ~(mask << size); |
888 | s[offset] |= lval << size; |
889 | } |
890 | else { |
81e118e0 |
891 | offset >>= 3; /* turn into byte offset */ |
79072805 |
892 | if (size == 8) |
c5a0f51a |
893 | s[offset ] = lval & 0xff; |
79072805 |
894 | else if (size == 16) { |
c5a0f51a |
895 | s[offset ] = (lval >> 8) & 0xff; |
896 | s[offset+1] = lval & 0xff; |
79072805 |
897 | } |
898 | else if (size == 32) { |
c5a0f51a |
899 | s[offset ] = (lval >> 24) & 0xff; |
900 | s[offset+1] = (lval >> 16) & 0xff; |
901 | s[offset+2] = (lval >> 8) & 0xff; |
902 | s[offset+3] = lval & 0xff; |
903 | } |
d7d93a81 |
904 | #ifdef UV_IS_QUAD |
c5a0f51a |
905 | else if (size == 64) { |
c5a0f51a |
906 | if (ckWARN(WARN_PORTABLE)) |
907 | Perl_warner(aTHX_ WARN_PORTABLE, |
908 | "Bit vector size > 32 non-portable"); |
909 | s[offset ] = (lval >> 56) & 0xff; |
910 | s[offset+1] = (lval >> 48) & 0xff; |
911 | s[offset+2] = (lval >> 40) & 0xff; |
912 | s[offset+3] = (lval >> 32) & 0xff; |
913 | s[offset+4] = (lval >> 24) & 0xff; |
914 | s[offset+5] = (lval >> 16) & 0xff; |
915 | s[offset+6] = (lval >> 8) & 0xff; |
916 | s[offset+7] = lval & 0xff; |
79072805 |
917 | } |
dc1e3f56 |
918 | #endif |
79072805 |
919 | } |
7bb043c3 |
920 | SvSETMAGIC(targ); |
79072805 |
921 | } |
922 | |
923 | void |
864dbfa3 |
924 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
925 | { |
463ee0b2 |
926 | STRLEN len; |
a0d0e21e |
927 | char *s; |
8e84507e |
928 | |
79072805 |
929 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
930 | register I32 i; |
931 | I32 max; |
932 | AV* av = (AV*)sv; |
933 | max = AvFILL(av); |
934 | for (i = 0; i <= max; i++) { |
935 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
936 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
937 | do_chop(astr, sv); |
938 | } |
939 | return; |
79072805 |
940 | } |
aa854799 |
941 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
942 | HV* hv = (HV*)sv; |
943 | HE* entry; |
944 | (void)hv_iterinit(hv); |
945 | /*SUPPRESS 560*/ |
155aba94 |
946 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
947 | do_chop(astr,hv_iterval(hv,entry)); |
948 | return; |
79072805 |
949 | } |
aa854799 |
950 | else if (SvREADONLY(sv)) |
cea2e8a9 |
951 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e |
952 | s = SvPV(sv, len); |
748a9306 |
953 | if (len && !SvPOK(sv)) |
a0d0e21e |
954 | s = SvPV_force(sv, len); |
7e2040f0 |
955 | if (DO_UTF8(sv)) { |
a0ed51b3 |
956 | if (s && len) { |
957 | char *send = s + len; |
958 | char *start = s; |
959 | s = send - 1; |
a0dbb045 |
960 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
961 | s--; |
9041c2e3 |
962 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 |
963 | sv_setpvn(astr, s, send - s); |
964 | *s = '\0'; |
965 | SvCUR_set(sv, s - start); |
966 | SvNIOK_off(sv); |
967 | SvUTF8_on(astr); |
968 | } |
a0ed51b3 |
969 | } |
970 | else |
971 | sv_setpvn(astr, "", 0); |
972 | } |
7e2040f0 |
973 | else if (s && len) { |
a0d0e21e |
974 | s += --len; |
975 | sv_setpvn(astr, s, 1); |
976 | *s = '\0'; |
977 | SvCUR_set(sv, len); |
2c19a612 |
978 | SvUTF8_off(sv); |
a0d0e21e |
979 | SvNIOK_off(sv); |
79072805 |
980 | } |
981 | else |
a0d0e21e |
982 | sv_setpvn(astr, "", 0); |
983 | SvSETMAGIC(sv); |
7e2040f0 |
984 | } |
a0d0e21e |
985 | |
986 | I32 |
864dbfa3 |
987 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
988 | { |
c07a80fd |
989 | register I32 count; |
a0d0e21e |
990 | STRLEN len; |
991 | char *s; |
c07a80fd |
992 | |
3280af22 |
993 | if (RsSNARF(PL_rs)) |
c07a80fd |
994 | return 0; |
4c5a6083 |
995 | if (RsRECORD(PL_rs)) |
996 | return 0; |
c07a80fd |
997 | count = 0; |
a0d0e21e |
998 | if (SvTYPE(sv) == SVt_PVAV) { |
999 | register I32 i; |
1000 | I32 max; |
1001 | AV* av = (AV*)sv; |
1002 | max = AvFILL(av); |
1003 | for (i = 0; i <= max; i++) { |
1004 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1005 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1006 | count += do_chomp(sv); |
1007 | } |
1008 | return count; |
1009 | } |
aa854799 |
1010 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
1011 | HV* hv = (HV*)sv; |
1012 | HE* entry; |
1013 | (void)hv_iterinit(hv); |
1014 | /*SUPPRESS 560*/ |
155aba94 |
1015 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1016 | count += do_chomp(hv_iterval(hv,entry)); |
1017 | return count; |
1018 | } |
aa854799 |
1019 | else if (SvREADONLY(sv)) |
cea2e8a9 |
1020 | Perl_croak(aTHX_ PL_no_modify); |
a0d0e21e |
1021 | s = SvPV(sv, len); |
1022 | if (len && !SvPOKp(sv)) |
1023 | s = SvPV_force(sv, len); |
1024 | if (s && len) { |
1025 | s += --len; |
3280af22 |
1026 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1027 | if (*s != '\n') |
1028 | goto nope; |
1029 | ++count; |
1030 | while (len && s[-1] == '\n') { |
1031 | --len; |
1032 | --s; |
1033 | ++count; |
1034 | } |
1035 | } |
a0d0e21e |
1036 | else { |
c07a80fd |
1037 | STRLEN rslen; |
3280af22 |
1038 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd |
1039 | if (rslen == 1) { |
1040 | if (*s != *rsptr) |
1041 | goto nope; |
1042 | ++count; |
1043 | } |
1044 | else { |
8c2cee6f |
1045 | if (len < rslen - 1) |
c07a80fd |
1046 | goto nope; |
1047 | len -= rslen - 1; |
1048 | s -= rslen - 1; |
36477c24 |
1049 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1050 | goto nope; |
1051 | count += rslen; |
1052 | } |
a0d0e21e |
1053 | } |
a0d0e21e |
1054 | *s = '\0'; |
1055 | SvCUR_set(sv, len); |
1056 | SvNIOK_off(sv); |
1057 | } |
1058 | nope: |
1059 | SvSETMAGIC(sv); |
1060 | return count; |
8e84507e |
1061 | } |
79072805 |
1062 | |
1063 | void |
864dbfa3 |
1064 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1065 | { |
1066 | #ifdef LIBERAL |
1067 | register long *dl; |
1068 | register long *ll; |
1069 | register long *rl; |
1070 | #endif |
1071 | register char *dc; |
463ee0b2 |
1072 | STRLEN leftlen; |
1073 | STRLEN rightlen; |
7a4c00b4 |
1074 | register char *lc; |
1075 | register char *rc; |
79072805 |
1076 | register I32 len; |
a0d0e21e |
1077 | I32 lensave; |
7a4c00b4 |
1078 | char *lsave; |
1079 | char *rsave; |
0c57e439 |
1080 | bool left_utf = DO_UTF8(left); |
1081 | bool right_utf = DO_UTF8(right); |
c9b3c8d0 |
1082 | I32 needlen; |
0c57e439 |
1083 | |
1084 | if (left_utf && !right_utf) |
1085 | sv_utf8_upgrade(right); |
a1ca4561 |
1086 | else if (!left_utf && right_utf) |
0c57e439 |
1087 | sv_utf8_upgrade(left); |
79072805 |
1088 | |
1fbd88dc |
1089 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1090 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
7a4c00b4 |
1091 | lsave = lc = SvPV(left, leftlen); |
1092 | rsave = rc = SvPV(right, rightlen); |
93a17b20 |
1093 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1094 | lensave = len; |
c9b3c8d0 |
1095 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1096 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
1097 | Newz(801, dc, needlen + 1, char); |
1098 | } |
1099 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2d8e6c8d |
1100 | STRLEN n_a; |
1101 | dc = SvPV_force(sv, n_a); |
ff68c719 |
1102 | if (SvCUR(sv) < len) { |
1103 | dc = SvGROW(sv, len + 1); |
1104 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1105 | } |
c9b3c8d0 |
1106 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1107 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1108 | } |
1109 | else { |
c9b3c8d0 |
1110 | needlen = ((optype == OP_BIT_AND) |
1111 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
ff68c719 |
1112 | Newz(801, dc, needlen + 1, char); |
1113 | (void)sv_usepvn(sv, dc, needlen); |
1114 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1115 | } |
a0d0e21e |
1116 | SvCUR_set(sv, len); |
1117 | (void)SvPOK_only(sv); |
0c57e439 |
1118 | if (left_utf || right_utf) { |
1119 | UV duc, luc, ruc; |
c9b3c8d0 |
1120 | char *dcsave = dc; |
0c57e439 |
1121 | STRLEN lulen = leftlen; |
1122 | STRLEN rulen = rightlen; |
ba210ebe |
1123 | STRLEN ulen; |
0c57e439 |
1124 | |
1125 | switch (optype) { |
1126 | case OP_BIT_AND: |
1127 | while (lulen && rulen) { |
9041c2e3 |
1128 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1129 | lc += ulen; |
1130 | lulen -= ulen; |
9041c2e3 |
1131 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1132 | rc += ulen; |
1133 | rulen -= ulen; |
1134 | duc = luc & ruc; |
9041c2e3 |
1135 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1136 | } |
c9b3c8d0 |
1137 | if (sv == left || sv == right) |
1138 | (void)sv_usepvn(sv, dcsave, needlen); |
1139 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1140 | break; |
1141 | case OP_BIT_XOR: |
1142 | while (lulen && rulen) { |
9041c2e3 |
1143 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1144 | lc += ulen; |
1145 | lulen -= ulen; |
9041c2e3 |
1146 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1147 | rc += ulen; |
1148 | rulen -= ulen; |
1149 | duc = luc ^ ruc; |
9041c2e3 |
1150 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1151 | } |
1152 | goto mop_up_utf; |
1153 | case OP_BIT_OR: |
1154 | while (lulen && rulen) { |
9041c2e3 |
1155 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1156 | lc += ulen; |
1157 | lulen -= ulen; |
9041c2e3 |
1158 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1159 | rc += ulen; |
1160 | rulen -= ulen; |
1161 | duc = luc | ruc; |
9041c2e3 |
1162 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1163 | } |
1164 | mop_up_utf: |
c9b3c8d0 |
1165 | if (sv == left || sv == right) |
1166 | (void)sv_usepvn(sv, dcsave, needlen); |
1167 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1168 | if (rulen) |
1169 | sv_catpvn(sv, rc, rulen); |
1170 | else if (lulen) |
1171 | sv_catpvn(sv, lc, lulen); |
1172 | else |
1173 | *SvEND(sv) = '\0'; |
1174 | break; |
1175 | } |
1176 | SvUTF8_on(sv); |
1177 | goto finish; |
1178 | } |
1179 | else |
79072805 |
1180 | #ifdef LIBERAL |
1181 | if (len >= sizeof(long)*4 && |
1182 | !((long)dc % sizeof(long)) && |
1183 | !((long)lc % sizeof(long)) && |
1184 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1185 | { |
1186 | I32 remainder = len % (sizeof(long)*4); |
1187 | len /= (sizeof(long)*4); |
1188 | |
1189 | dl = (long*)dc; |
1190 | ll = (long*)lc; |
1191 | rl = (long*)rc; |
1192 | |
1193 | switch (optype) { |
1194 | case OP_BIT_AND: |
1195 | while (len--) { |
1196 | *dl++ = *ll++ & *rl++; |
1197 | *dl++ = *ll++ & *rl++; |
1198 | *dl++ = *ll++ & *rl++; |
1199 | *dl++ = *ll++ & *rl++; |
1200 | } |
1201 | break; |
a0d0e21e |
1202 | case OP_BIT_XOR: |
79072805 |
1203 | while (len--) { |
1204 | *dl++ = *ll++ ^ *rl++; |
1205 | *dl++ = *ll++ ^ *rl++; |
1206 | *dl++ = *ll++ ^ *rl++; |
1207 | *dl++ = *ll++ ^ *rl++; |
1208 | } |
1209 | break; |
1210 | case OP_BIT_OR: |
1211 | while (len--) { |
1212 | *dl++ = *ll++ | *rl++; |
1213 | *dl++ = *ll++ | *rl++; |
1214 | *dl++ = *ll++ | *rl++; |
1215 | *dl++ = *ll++ | *rl++; |
1216 | } |
1217 | } |
1218 | |
1219 | dc = (char*)dl; |
1220 | lc = (char*)ll; |
1221 | rc = (char*)rl; |
1222 | |
1223 | len = remainder; |
1224 | } |
1225 | #endif |
a0d0e21e |
1226 | { |
a0d0e21e |
1227 | switch (optype) { |
1228 | case OP_BIT_AND: |
1229 | while (len--) |
1230 | *dc++ = *lc++ & *rc++; |
1231 | break; |
1232 | case OP_BIT_XOR: |
1233 | while (len--) |
1234 | *dc++ = *lc++ ^ *rc++; |
1235 | goto mop_up; |
1236 | case OP_BIT_OR: |
1237 | while (len--) |
1238 | *dc++ = *lc++ | *rc++; |
1239 | mop_up: |
1240 | len = lensave; |
1241 | if (rightlen > len) |
1242 | sv_catpvn(sv, rsave + len, rightlen - len); |
1243 | else if (leftlen > len) |
1244 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1245 | else |
1246 | *SvEND(sv) = '\0'; |
a0d0e21e |
1247 | break; |
1248 | } |
79072805 |
1249 | } |
0c57e439 |
1250 | finish: |
fb73857a |
1251 | SvTAINT(sv); |
79072805 |
1252 | } |
463ee0b2 |
1253 | |
1254 | OP * |
cea2e8a9 |
1255 | Perl_do_kv(pTHX) |
463ee0b2 |
1256 | { |
39644a26 |
1257 | dSP; |
463ee0b2 |
1258 | HV *hv = (HV*)POPs; |
800e9ae0 |
1259 | HV *keys; |
463ee0b2 |
1260 | register HE *entry; |
463ee0b2 |
1261 | SV *tmpstr; |
54310121 |
1262 | I32 gimme = GIMME_V; |
533c011a |
1263 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1264 | I32 dovalues = (PL_op->op_type == OP_VALUES); |
c750a3ec |
1265 | I32 realhv = (SvTYPE(hv) == SVt_PVHV); |
8e84507e |
1266 | |
1267 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e |
1268 | dokeys = dovalues = TRUE; |
463ee0b2 |
1269 | |
85581909 |
1270 | if (!hv) { |
78f9721b |
1271 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1272 | dTARGET; /* make sure to clear its target here */ |
1273 | if (SvTYPE(TARG) == SVt_PVLV) |
1274 | LvTARG(TARG) = Nullsv; |
1275 | PUSHs(TARG); |
1276 | } |
463ee0b2 |
1277 | RETURN; |
85581909 |
1278 | } |
748a9306 |
1279 | |
800e9ae0 |
1280 | keys = realhv ? hv : avhv_keys((AV*)hv); |
1281 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1282 | |
54310121 |
1283 | if (gimme == G_VOID) |
aa689395 |
1284 | RETURN; |
1285 | |
54310121 |
1286 | if (gimme == G_SCALAR) { |
6ee623d5 |
1287 | IV i; |
463ee0b2 |
1288 | dTARGET; |
1289 | |
78f9721b |
1290 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1291 | if (SvTYPE(TARG) < SVt_PVLV) { |
1292 | sv_upgrade(TARG, SVt_PVLV); |
1293 | sv_magic(TARG, Nullsv, 'k', Nullch, 0); |
1294 | } |
1295 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1296 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1297 | if (LvTARG(TARG)) |
1298 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1299 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1300 | } |
85581909 |
1301 | PUSHs(TARG); |
1302 | RETURN; |
1303 | } |
1304 | |
33c27489 |
1305 | if (! SvTIED_mg((SV*)keys, 'P')) |
800e9ae0 |
1306 | i = HvKEYS(keys); |
463ee0b2 |
1307 | else { |
1308 | i = 0; |
463ee0b2 |
1309 | /*SUPPRESS 560*/ |
800e9ae0 |
1310 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1311 | } |
1312 | PUSHi( i ); |
1313 | RETURN; |
1314 | } |
1315 | |
8ed4b672 |
1316 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1317 | |
463ee0b2 |
1318 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1319 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1320 | SPAGAIN; |
6bfc225d |
1321 | if (dokeys) |
8c2cee6f |
1322 | XPUSHs(hv_iterkeysv(entry)); /* won't clobber stack_sp */ |
463ee0b2 |
1323 | if (dovalues) { |
463ee0b2 |
1324 | PUTBACK; |
b6429b1b |
1325 | tmpstr = realhv ? |
1326 | hv_iterval(hv,entry) : avhv_iterval((AV*)hv,entry); |
cea2e8a9 |
1327 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1328 | (unsigned long)HeHASH(entry), |
800e9ae0 |
1329 | HvMAX(keys)+1, |
1330 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1331 | SPAGAIN; |
46fc3d4c |
1332 | XPUSHs(tmpstr); |
463ee0b2 |
1333 | } |
1334 | PUTBACK; |
1335 | } |
1336 | return NORMAL; |
1337 | } |
4e35701f |
1338 | |