Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
4 | * 2000, 2001, 2002, by Larry Wall and others |
79072805 |
5 | * |
6 | * You may distribute under the terms of either the GNU General Public |
7 | * License or the Artistic License, as specified in the README file. |
8 | * |
a0d0e21e |
9 | */ |
10 | |
11 | /* |
12 | * "'So that was the job I felt I had to do when I started,' thought Sam." |
79072805 |
13 | */ |
14 | |
15 | #include "EXTERN.h" |
864dbfa3 |
16 | #define PERL_IN_DOOP_C |
79072805 |
17 | #include "perl.h" |
18 | |
64ca3a65 |
19 | #ifndef PERL_MICRO |
79072805 |
20 | #include <signal.h> |
21 | #endif |
22 | |
942e002e |
23 | STATIC I32 |
b250498f |
24 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 |
25 | { |
4757a243 |
26 | U8 *s; |
b250498f |
27 | U8 *d; |
4757a243 |
28 | U8 *send; |
b250498f |
29 | U8 *dstart; |
4757a243 |
30 | I32 matches = 0; |
9b877dbb |
31 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
32 | STRLEN len; |
4757a243 |
33 | short *tbl; |
34 | I32 ch; |
79072805 |
35 | |
4757a243 |
36 | tbl = (short*)cPVOP->op_pv; |
37 | if (!tbl) |
7d85a32c |
38 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
39 | |
4757a243 |
40 | s = (U8*)SvPV(sv, len); |
41 | send = s + len; |
42 | |
1e54db1a |
43 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
44 | if (!SvUTF8(sv)) { |
01ec43d0 |
45 | while (s < send) { |
036b4402 |
46 | if ((ch = tbl[*s]) >= 0) { |
01ec43d0 |
47 | matches++; |
eb160463 |
48 | *s++ = (U8)ch; |
01ec43d0 |
49 | } |
50 | else |
51 | s++; |
52 | } |
53 | SvSETMAGIC(sv); |
b250498f |
54 | return matches; |
55 | } |
4757a243 |
56 | |
b250498f |
57 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
58 | if (grows) |
59 | New(0, d, len*2+1, U8); |
60 | else |
61 | d = s; |
b250498f |
62 | dstart = d; |
63 | while (s < send) { |
ba210ebe |
64 | STRLEN ulen; |
9b877dbb |
65 | UV c; |
b250498f |
66 | |
b250498f |
67 | /* Need to check this, otherwise 128..255 won't match */ |
9041c2e3 |
68 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
9b877dbb |
69 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
70 | matches++; |
9041c2e3 |
71 | d = uvchr_to_utf8(d, ch); |
b250498f |
72 | s += ulen; |
01ec43d0 |
73 | } |
74 | else { /* No match -> copy */ |
0376ff32 |
75 | Move(s, d, ulen, U8); |
9b877dbb |
76 | d += ulen; |
77 | s += ulen; |
b250498f |
78 | } |
79 | } |
9b877dbb |
80 | if (grows) { |
81 | sv_setpvn(sv, (char*)dstart, d - dstart); |
82 | Safefree(dstart); |
83 | } |
84 | else { |
85 | *d = '\0'; |
86 | SvCUR_set(sv, d - dstart); |
87 | } |
b250498f |
88 | SvUTF8_on(sv); |
b250498f |
89 | SvSETMAGIC(sv); |
4757a243 |
90 | return matches; |
91 | } |
92 | |
942e002e |
93 | STATIC I32 |
21304a3b |
94 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 |
95 | { |
4757a243 |
96 | U8 *s; |
97 | U8 *send; |
98 | I32 matches = 0; |
99 | STRLEN len; |
100 | short *tbl; |
8973db79 |
101 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
4757a243 |
102 | |
103 | tbl = (short*)cPVOP->op_pv; |
104 | if (!tbl) |
7d85a32c |
105 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
106 | |
107 | s = (U8*)SvPV(sv, len); |
108 | send = s + len; |
109 | |
1aa99e6b |
110 | if (!SvUTF8(sv)) |
111 | while (s < send) { |
112 | if (tbl[*s++] >= 0) |
036b4402 |
113 | matches++; |
1aa99e6b |
114 | } |
115 | else |
116 | while (s < send) { |
117 | UV c; |
118 | STRLEN ulen; |
9041c2e3 |
119 | c = utf8n_to_uvchr(s, send - s, &ulen, 0); |
8973db79 |
120 | if (c < 0x100) { |
121 | if (tbl[c] >= 0) |
122 | matches++; |
123 | } else if (complement) |
1aa99e6b |
124 | matches++; |
125 | s += ulen; |
126 | } |
4757a243 |
127 | |
128 | return matches; |
129 | } |
130 | |
942e002e |
131 | STATIC I32 |
21304a3b |
132 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 |
133 | { |
4757a243 |
134 | U8 *s; |
135 | U8 *send; |
136 | U8 *d; |
5e44153e |
137 | U8 *dstart; |
1aa99e6b |
138 | I32 isutf8; |
4757a243 |
139 | I32 matches = 0; |
9b877dbb |
140 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
8973db79 |
141 | I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
142 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
b7953727 |
143 | STRLEN len, rlen = 0; |
4757a243 |
144 | short *tbl; |
145 | I32 ch; |
146 | |
147 | tbl = (short*)cPVOP->op_pv; |
148 | if (!tbl) |
7d85a32c |
149 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
150 | |
151 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
152 | isutf8 = SvUTF8(sv); |
4757a243 |
153 | send = s + len; |
154 | |
1aa99e6b |
155 | if (!isutf8) { |
156 | dstart = d = s; |
157 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
158 | U8* p = send; |
159 | while (s < send) { |
160 | if ((ch = tbl[*s]) >= 0) { |
eb160463 |
161 | *d = (U8)ch; |
036b4402 |
162 | matches++; |
1aa99e6b |
163 | if (p != d - 1 || *p != *d) |
164 | p = d++; |
165 | } |
166 | else if (ch == -1) /* -1 is unmapped character */ |
167 | *d++ = *s; |
168 | else if (ch == -2) /* -2 is delete character */ |
169 | matches++; |
170 | s++; |
171 | } |
a0ed51b3 |
172 | } |
1aa99e6b |
173 | else { |
174 | while (s < send) { |
175 | if ((ch = tbl[*s]) >= 0) { |
176 | matches++; |
eb160463 |
177 | *d++ = (U8)ch; |
1aa99e6b |
178 | } |
179 | else if (ch == -1) /* -1 is unmapped character */ |
180 | *d++ = *s; |
181 | else if (ch == -2) /* -2 is delete character */ |
182 | matches++; |
183 | s++; |
184 | } |
185 | } |
76ef7183 |
186 | *d = '\0'; |
1aa99e6b |
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]; |
eb160463 |
219 | if ((UV)ch != pch) { |
9041c2e3 |
220 | d = uvchr_to_utf8(d, ch); |
eb160463 |
221 | pch = (UV)ch; |
8973db79 |
222 | } |
223 | s += len; |
224 | continue; |
225 | } |
226 | } |
9b877dbb |
227 | } |
1aa99e6b |
228 | else if ((ch = tbl[comp]) >= 0) { |
229 | matches++; |
eb160463 |
230 | if ((UV)ch != pch) { |
9041c2e3 |
231 | d = uvchr_to_utf8(d, ch); |
eb160463 |
232 | pch = (UV)ch; |
1aa99e6b |
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) { |
0376ff32 |
253 | Move(s, d, len, U8); |
8973db79 |
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 |
21304a3b |
294 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
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; |
b7953727 |
310 | UV final = 0; |
4757a243 |
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) { |
3568d838 |
346 | if ((uv = swash_fetch(rv, s, TRUE)) < none) { |
4757a243 |
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); |
0376ff32 |
353 | Move(s, d, i, U8); |
9b877dbb |
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); |
4757a243 |
388 | |
389 | return matches; |
390 | } |
391 | |
942e002e |
392 | STATIC I32 |
21304a3b |
393 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 |
394 | { |
4757a243 |
395 | U8 *s; |
b7953727 |
396 | U8 *start = 0, *send; |
4757a243 |
397 | I32 matches = 0; |
398 | STRLEN len; |
399 | |
400 | SV* rv = (SV*)cSVOP->op_sv; |
401 | HV* hv = (HV*)SvRV(rv); |
402 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
403 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
45005bfb |
404 | UV extra = none + 1; |
4757a243 |
405 | UV uv; |
1aa99e6b |
406 | U8 hibit = 0; |
4757a243 |
407 | |
408 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
409 | if (!SvUTF8(sv)) { |
410 | U8 *t = s, *e = s + len; |
c4d5f83a |
411 | while (t < e) { |
412 | U8 ch = *t++; |
413 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
414 | break; |
c4d5f83a |
415 | } |
1aa99e6b |
416 | if (hibit) |
417 | start = s = bytes_to_utf8(s, &len); |
418 | } |
4757a243 |
419 | send = s + len; |
420 | |
421 | while (s < send) { |
3568d838 |
422 | if ((uv = swash_fetch(rv, s, TRUE)) < none || uv == extra) |
4757a243 |
423 | matches++; |
834a4ddd |
424 | s += UTF8SKIP(s); |
4757a243 |
425 | } |
1aa99e6b |
426 | if (hibit) |
427 | Safefree(start); |
4757a243 |
428 | |
429 | return matches; |
430 | } |
431 | |
942e002e |
432 | STATIC I32 |
21304a3b |
433 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 |
434 | { |
4757a243 |
435 | U8 *s; |
1aa99e6b |
436 | U8 *start, *send; |
4757a243 |
437 | U8 *d; |
438 | I32 matches = 0; |
439 | I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
4757a243 |
440 | I32 del = PL_op->op_private & OPpTRANS_DELETE; |
9b877dbb |
441 | I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
442 | SV* rv = (SV*)cSVOP->op_sv; |
443 | HV* hv = (HV*)SvRV(rv); |
444 | SV** svp = hv_fetch(hv, "NONE", 4, FALSE); |
445 | UV none = svp ? SvUV(*svp) : 0x7fffffff; |
446 | UV extra = none + 1; |
b7953727 |
447 | UV final = 0; |
45005bfb |
448 | bool havefinal = FALSE; |
4757a243 |
449 | UV uv; |
450 | STRLEN len; |
1aa99e6b |
451 | U8 *dstart, *dend; |
452 | I32 isutf8; |
453 | U8 hibit = 0; |
4757a243 |
454 | |
455 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
456 | isutf8 = SvUTF8(sv); |
457 | if (!isutf8) { |
458 | U8 *t = s, *e = s + len; |
c4d5f83a |
459 | while (t < e) { |
460 | U8 ch = *t++; |
461 | if ((hibit = !NATIVE_IS_INVARIANT(ch))) |
1aa99e6b |
462 | break; |
c4d5f83a |
463 | } |
1aa99e6b |
464 | if (hibit) |
465 | s = bytes_to_utf8(s, &len); |
466 | } |
4757a243 |
467 | send = s + len; |
1aa99e6b |
468 | start = s; |
4757a243 |
469 | |
470 | svp = hv_fetch(hv, "FINAL", 5, FALSE); |
45005bfb |
471 | if (svp) { |
4757a243 |
472 | final = SvUV(*svp); |
45005bfb |
473 | havefinal = TRUE; |
474 | } |
4757a243 |
475 | |
9b877dbb |
476 | if (grows) { |
477 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
478 | New(0, d, len*3+UTF8_MAXLEN, U8); |
479 | dend = d + len * 3; |
480 | dstart = d; |
481 | } |
482 | else { |
483 | dstart = d = s; |
484 | dend = d + len; |
485 | } |
4757a243 |
486 | |
487 | if (squash) { |
488 | UV puv = 0xfeedface; |
489 | while (s < send) { |
3568d838 |
490 | uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
491 | |
9b877dbb |
492 | if (d > dend) { |
493 | STRLEN clen = d - dstart; |
494 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
495 | if (!grows) |
7d85a32c |
496 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
497 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
498 | d = dstart + clen; |
499 | dend = dstart + nlen; |
4757a243 |
500 | } |
501 | if (uv < none) { |
502 | matches++; |
94472101 |
503 | s += UTF8SKIP(s); |
4757a243 |
504 | if (uv != puv) { |
7d85a32c |
505 | d = uvuni_to_utf8(d, uv); |
4757a243 |
506 | puv = uv; |
507 | } |
4757a243 |
508 | continue; |
509 | } |
510 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
511 | int i = UTF8SKIP(s); |
0376ff32 |
512 | Move(s, d, i, U8); |
9b877dbb |
513 | d += i; |
514 | s += i; |
4757a243 |
515 | puv = 0xfeedface; |
516 | continue; |
a0ed51b3 |
517 | } |
4757a243 |
518 | else if (uv == extra && !del) { |
519 | matches++; |
45005bfb |
520 | if (havefinal) { |
521 | s += UTF8SKIP(s); |
522 | if (puv != final) { |
7d85a32c |
523 | d = uvuni_to_utf8(d, final); |
45005bfb |
524 | puv = final; |
525 | } |
526 | } |
527 | else { |
528 | STRLEN len; |
7d85a32c |
529 | uv = utf8_to_uvuni(s, &len); |
45005bfb |
530 | if (uv != puv) { |
0376ff32 |
531 | Move(s, d, len, U8); |
45005bfb |
532 | d += len; |
533 | puv = uv; |
534 | } |
535 | s += len; |
4757a243 |
536 | } |
4757a243 |
537 | continue; |
538 | } |
01ec43d0 |
539 | matches++; /* "none+1" is delete character */ |
540 | s += UTF8SKIP(s); |
a0ed51b3 |
541 | } |
79072805 |
542 | } |
543 | else { |
4757a243 |
544 | while (s < send) { |
3568d838 |
545 | uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
546 | if (d > dend) { |
547 | STRLEN clen = d - dstart; |
548 | STRLEN nlen = dend - dstart + len + UTF8_MAXLEN; |
549 | if (!grows) |
7d85a32c |
550 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
1aa99e6b |
551 | Renew(dstart, nlen+UTF8_MAXLEN, U8); |
552 | d = dstart + clen; |
553 | dend = dstart + nlen; |
a0ed51b3 |
554 | } |
4757a243 |
555 | if (uv < none) { |
556 | matches++; |
01ec43d0 |
557 | s += UTF8SKIP(s); |
7d85a32c |
558 | d = uvuni_to_utf8(d, uv); |
4757a243 |
559 | continue; |
a0ed51b3 |
560 | } |
4757a243 |
561 | else if (uv == none) { /* "none" is unmapped character */ |
1aa99e6b |
562 | int i = UTF8SKIP(s); |
0376ff32 |
563 | Move(s, d, i, U8); |
9b877dbb |
564 | d += i; |
565 | s += i; |
4757a243 |
566 | continue; |
79072805 |
567 | } |
4757a243 |
568 | else if (uv == extra && !del) { |
569 | matches++; |
01ec43d0 |
570 | s += UTF8SKIP(s); |
7d85a32c |
571 | d = uvuni_to_utf8(d, final); |
4757a243 |
572 | continue; |
573 | } |
01ec43d0 |
574 | matches++; /* "none+1" is delete character */ |
575 | s += UTF8SKIP(s); |
79072805 |
576 | } |
4757a243 |
577 | } |
16ec844d |
578 | if (grows || hibit) { |
9b877dbb |
579 | sv_setpvn(sv, (char*)dstart, d - dstart); |
580 | Safefree(dstart); |
16ec844d |
581 | if (grows && hibit) |
582 | Safefree(start); |
9b877dbb |
583 | } |
584 | else { |
585 | *d = '\0'; |
586 | SvCUR_set(sv, d - dstart); |
587 | } |
1aa99e6b |
588 | SvUTF8_on(sv); |
4757a243 |
589 | SvSETMAGIC(sv); |
590 | |
591 | return matches; |
592 | } |
593 | |
594 | I32 |
864dbfa3 |
595 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
596 | { |
597 | STRLEN len; |
8e84507e |
598 | I32 hasutf = (PL_op->op_private & |
036b4402 |
599 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
600 | |
2233f375 |
601 | if (SvREADONLY(sv)) { |
765f542d |
602 | if (SvIsCOW(sv)) |
603 | sv_force_normal_flags(sv, 0); |
2233f375 |
604 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
605 | Perl_croak(aTHX_ PL_no_modify); |
606 | } |
4757a243 |
607 | (void)SvPV(sv, len); |
608 | if (!len) |
609 | return 0; |
d59e14db |
610 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
611 | if (!SvPOKp(sv)) |
612 | (void)SvPV_force(sv, len); |
2de7b02f |
613 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
614 | } |
4757a243 |
615 | |
cea2e8a9 |
616 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
617 | |
036b4402 |
618 | switch (PL_op->op_private & ~hasutf & 63) { |
4757a243 |
619 | case 0: |
01ec43d0 |
620 | if (hasutf) |
621 | return do_trans_simple_utf8(sv); |
622 | else |
623 | return do_trans_simple(sv); |
4757a243 |
624 | |
625 | case OPpTRANS_IDENTICAL: |
05d340b8 |
626 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
627 | if (hasutf) |
628 | return do_trans_count_utf8(sv); |
629 | else |
630 | return do_trans_count(sv); |
4757a243 |
631 | |
632 | default: |
01ec43d0 |
633 | if (hasutf) |
036b4402 |
634 | return do_trans_complex_utf8(sv); |
4757a243 |
635 | else |
036b4402 |
636 | return do_trans_complex(sv); |
79072805 |
637 | } |
79072805 |
638 | } |
639 | |
640 | void |
864dbfa3 |
641 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
642 | { |
643 | SV **oldmark = mark; |
644 | register I32 items = sp - mark; |
79072805 |
645 | register STRLEN len; |
463ee0b2 |
646 | STRLEN delimlen; |
463ee0b2 |
647 | STRLEN tmplen; |
79072805 |
648 | |
5ba99574 |
649 | (void) SvPV(del, delimlen); /* stringify and get the delimlen */ |
650 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
651 | |
79072805 |
652 | mark++; |
653 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
07f14f54 |
654 | (void)SvUPGRADE(sv, SVt_PV); |
79072805 |
655 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
656 | while (items-- > 0) { |
1426bbf4 |
657 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
463ee0b2 |
658 | SvPV(*mark, tmplen); |
659 | len += tmplen; |
79072805 |
660 | } |
661 | mark++; |
662 | } |
663 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
664 | |
665 | mark = oldmark; |
db7c17d7 |
666 | items = sp - mark; |
79072805 |
667 | ++mark; |
668 | } |
669 | |
8d6d96c1 |
670 | sv_setpv(sv, ""); |
671 | if (PL_tainting && SvMAGICAL(sv)) |
672 | SvTAINTED_off(sv); |
673 | |
463ee0b2 |
674 | if (items-- > 0) { |
92d29cee |
675 | if (*mark) |
676 | sv_catsv(sv, *mark); |
463ee0b2 |
677 | mark++; |
678 | } |
8d6d96c1 |
679 | |
c512ce4f |
680 | if (delimlen) { |
79072805 |
681 | for (; items > 0; items--,mark++) { |
c512ce4f |
682 | sv_catsv(sv,del); |
79072805 |
683 | sv_catsv(sv,*mark); |
684 | } |
685 | } |
686 | else { |
687 | for (; items > 0; items--,mark++) |
688 | sv_catsv(sv,*mark); |
689 | } |
690 | SvSETMAGIC(sv); |
691 | } |
692 | |
693 | void |
864dbfa3 |
694 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
695 | { |
46fc3d4c |
696 | STRLEN patlen; |
697 | char *pat = SvPV(*sarg, patlen); |
698 | bool do_taint = FALSE; |
699 | |
5b781b5b |
700 | SvUTF8_off(sv); |
2cf2cfc6 |
701 | if (DO_UTF8(*sarg)) |
702 | SvUTF8_on(sv); |
46fc3d4c |
703 | sv_vsetpvfn(sv, pat, patlen, Null(va_list*), sarg + 1, len - 1, &do_taint); |
79072805 |
704 | SvSETMAGIC(sv); |
46fc3d4c |
705 | if (do_taint) |
706 | SvTAINTED_on(sv); |
79072805 |
707 | } |
708 | |
33b45480 |
709 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
710 | UV |
711 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
712 | { |
713 | STRLEN srclen, len; |
714 | unsigned char *s = (unsigned char *) SvPV(sv, srclen); |
715 | UV retnum = 0; |
716 | |
a50d7633 |
717 | if (offset < 0) |
81e118e0 |
718 | return retnum; |
8e84507e |
719 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
720 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
721 | |
dcad2880 |
722 | if (SvUTF8(sv)) |
33b45480 |
723 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
724 | |
81e118e0 |
725 | offset *= size; /* turn into bit offset */ |
726 | len = (offset + size + 7) / 8; /* required number of bytes */ |
727 | if (len > srclen) { |
728 | if (size <= 8) |
729 | retnum = 0; |
730 | else { |
731 | offset >>= 3; /* turn into byte offset */ |
732 | if (size == 16) { |
eb160463 |
733 | if ((STRLEN)offset >= srclen) |
81e118e0 |
734 | retnum = 0; |
735 | else |
628e1a40 |
736 | retnum = (UV) s[offset] << 8; |
81e118e0 |
737 | } |
738 | else if (size == 32) { |
eb160463 |
739 | if ((STRLEN)offset >= srclen) |
81e118e0 |
740 | retnum = 0; |
eb160463 |
741 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 |
742 | retnum = |
743 | ((UV) s[offset ] << 24); |
eb160463 |
744 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 |
745 | retnum = |
746 | ((UV) s[offset ] << 24) + |
747 | ((UV) s[offset + 1] << 16); |
748 | else |
749 | retnum = |
750 | ((UV) s[offset ] << 24) + |
751 | ((UV) s[offset + 1] << 16) + |
752 | ( s[offset + 2] << 8); |
753 | } |
d7d93a81 |
754 | #ifdef UV_IS_QUAD |
c5a0f51a |
755 | else if (size == 64) { |
c5a0f51a |
756 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
757 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
758 | "Bit vector size > 32 non-portable"); |
759 | if (offset >= srclen) |
760 | retnum = 0; |
761 | else if (offset + 1 >= srclen) |
762 | retnum = |
763 | (UV) s[offset ] << 56; |
764 | else if (offset + 2 >= srclen) |
765 | retnum = |
766 | ((UV) s[offset ] << 56) + |
767 | ((UV) s[offset + 1] << 48); |
768 | else if (offset + 3 >= srclen) |
769 | retnum = |
770 | ((UV) s[offset ] << 56) + |
771 | ((UV) s[offset + 1] << 48) + |
772 | ((UV) s[offset + 2] << 40); |
773 | else if (offset + 4 >= 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 | else if (offset + 5 >= srclen) |
780 | retnum = |
781 | ((UV) s[offset ] << 56) + |
782 | ((UV) s[offset + 1] << 48) + |
783 | ((UV) s[offset + 2] << 40) + |
784 | ((UV) s[offset + 3] << 32) + |
785 | ( s[offset + 4] << 24); |
786 | else if (offset + 6 >= srclen) |
787 | retnum = |
788 | ((UV) s[offset ] << 56) + |
789 | ((UV) s[offset + 1] << 48) + |
790 | ((UV) s[offset + 2] << 40) + |
791 | ((UV) s[offset + 3] << 32) + |
792 | ((UV) s[offset + 4] << 24) + |
793 | ((UV) s[offset + 5] << 16); |
794 | else |
8e84507e |
795 | retnum = |
c5a0f51a |
796 | ((UV) s[offset ] << 56) + |
797 | ((UV) s[offset + 1] << 48) + |
798 | ((UV) s[offset + 2] << 40) + |
799 | ((UV) s[offset + 3] << 32) + |
800 | ((UV) s[offset + 4] << 24) + |
801 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
802 | ( s[offset + 6] << 8); |
c5a0f51a |
803 | } |
804 | #endif |
81e118e0 |
805 | } |
806 | } |
807 | else if (size < 8) |
808 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
809 | else { |
810 | offset >>= 3; /* turn into byte offset */ |
811 | if (size == 8) |
812 | retnum = s[offset]; |
813 | else if (size == 16) |
814 | retnum = |
628e1a40 |
815 | ((UV) s[offset] << 8) + |
81e118e0 |
816 | s[offset + 1]; |
817 | else if (size == 32) |
818 | retnum = |
819 | ((UV) s[offset ] << 24) + |
820 | ((UV) s[offset + 1] << 16) + |
821 | ( s[offset + 2] << 8) + |
822 | s[offset + 3]; |
d7d93a81 |
823 | #ifdef UV_IS_QUAD |
c5a0f51a |
824 | else if (size == 64) { |
c5a0f51a |
825 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
826 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
827 | "Bit vector size > 32 non-portable"); |
828 | retnum = |
829 | ((UV) s[offset ] << 56) + |
830 | ((UV) s[offset + 1] << 48) + |
831 | ((UV) s[offset + 2] << 40) + |
832 | ((UV) s[offset + 3] << 32) + |
833 | ((UV) s[offset + 4] << 24) + |
834 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
835 | ( s[offset + 6] << 8) + |
c5a0f51a |
836 | s[offset + 7]; |
837 | } |
838 | #endif |
81e118e0 |
839 | } |
840 | |
841 | return retnum; |
842 | } |
843 | |
33b45480 |
844 | /* currently converts input to bytes if possible but doesn't sweat failures, |
845 | * although it does ensure that the string it clobbers is not marked as |
846 | * utf8-valid any more |
847 | */ |
79072805 |
848 | void |
864dbfa3 |
849 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
850 | { |
851 | SV *targ = LvTARG(sv); |
852 | register I32 offset; |
853 | register I32 size; |
8990e307 |
854 | register unsigned char *s; |
81e118e0 |
855 | register UV lval; |
79072805 |
856 | I32 mask; |
a0d0e21e |
857 | STRLEN targlen; |
858 | STRLEN len; |
79072805 |
859 | |
8990e307 |
860 | if (!targ) |
861 | return; |
a0d0e21e |
862 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
863 | if (SvUTF8(targ)) { |
33b45480 |
864 | /* This is handled by the SvPOK_only below... |
865 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
866 | SvUTF8_off(targ); |
867 | */ |
868 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
869 | } |
870 | |
4ebbc975 |
871 | (void)SvPOK_only(targ); |
81e118e0 |
872 | lval = SvUV(sv); |
79072805 |
873 | offset = LvTARGOFF(sv); |
fe58ced6 |
874 | if (offset < 0) |
ed9aa3b7 |
875 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
876 | size = LvTARGLEN(sv); |
8e84507e |
877 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
878 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
879 | |
81e118e0 |
880 | offset *= size; /* turn into bit offset */ |
881 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
882 | if (len > targlen) { |
883 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
884 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
885 | SvCUR_set(targ, len); |
886 | } |
8e84507e |
887 | |
79072805 |
888 | if (size < 8) { |
889 | mask = (1 << size) - 1; |
890 | size = offset & 7; |
891 | lval &= mask; |
81e118e0 |
892 | offset >>= 3; /* turn into byte offset */ |
79072805 |
893 | s[offset] &= ~(mask << size); |
894 | s[offset] |= lval << size; |
895 | } |
896 | else { |
81e118e0 |
897 | offset >>= 3; /* turn into byte offset */ |
79072805 |
898 | if (size == 8) |
eb160463 |
899 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
900 | else if (size == 16) { |
eb160463 |
901 | s[offset ] = (U8)((lval >> 8) & 0xff); |
902 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
903 | } |
904 | else if (size == 32) { |
eb160463 |
905 | s[offset ] = (U8)((lval >> 24) & 0xff); |
906 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
907 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
908 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
909 | } |
d7d93a81 |
910 | #ifdef UV_IS_QUAD |
c5a0f51a |
911 | else if (size == 64) { |
c5a0f51a |
912 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
913 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
914 | "Bit vector size > 32 non-portable"); |
eb160463 |
915 | s[offset ] = (U8)((lval >> 56) & 0xff); |
916 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
917 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
918 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
919 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
920 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
921 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
922 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
923 | } |
dc1e3f56 |
924 | #endif |
79072805 |
925 | } |
7bb043c3 |
926 | SvSETMAGIC(targ); |
79072805 |
927 | } |
928 | |
929 | void |
864dbfa3 |
930 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
931 | { |
463ee0b2 |
932 | STRLEN len; |
a0d0e21e |
933 | char *s; |
8e84507e |
934 | |
79072805 |
935 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
936 | register I32 i; |
937 | I32 max; |
938 | AV* av = (AV*)sv; |
939 | max = AvFILL(av); |
940 | for (i = 0; i <= max; i++) { |
941 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
942 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
943 | do_chop(astr, sv); |
944 | } |
945 | return; |
79072805 |
946 | } |
aa854799 |
947 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
948 | HV* hv = (HV*)sv; |
949 | HE* entry; |
950 | (void)hv_iterinit(hv); |
951 | /*SUPPRESS 560*/ |
155aba94 |
952 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
953 | do_chop(astr,hv_iterval(hv,entry)); |
954 | return; |
79072805 |
955 | } |
991e6d41 |
956 | else if (SvREADONLY(sv)) { |
957 | if (SvFAKE(sv)) { |
958 | /* SV is copy-on-write */ |
959 | sv_force_normal_flags(sv, 0); |
960 | } |
961 | if (SvREADONLY(sv)) |
962 | Perl_croak(aTHX_ PL_no_modify); |
963 | } |
a0d0e21e |
964 | s = SvPV(sv, len); |
748a9306 |
965 | if (len && !SvPOK(sv)) |
a0d0e21e |
966 | s = SvPV_force(sv, len); |
7e2040f0 |
967 | if (DO_UTF8(sv)) { |
a0ed51b3 |
968 | if (s && len) { |
969 | char *send = s + len; |
970 | char *start = s; |
971 | s = send - 1; |
a0dbb045 |
972 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
973 | s--; |
9041c2e3 |
974 | if (utf8_to_uvchr((U8*)s, 0)) { |
a0dbb045 |
975 | sv_setpvn(astr, s, send - s); |
976 | *s = '\0'; |
977 | SvCUR_set(sv, s - start); |
978 | SvNIOK_off(sv); |
979 | SvUTF8_on(astr); |
980 | } |
a0ed51b3 |
981 | } |
982 | else |
983 | sv_setpvn(astr, "", 0); |
984 | } |
7e2040f0 |
985 | else if (s && len) { |
a0d0e21e |
986 | s += --len; |
987 | sv_setpvn(astr, s, 1); |
988 | *s = '\0'; |
989 | SvCUR_set(sv, len); |
2c19a612 |
990 | SvUTF8_off(sv); |
a0d0e21e |
991 | SvNIOK_off(sv); |
79072805 |
992 | } |
993 | else |
a0d0e21e |
994 | sv_setpvn(astr, "", 0); |
995 | SvSETMAGIC(sv); |
7e2040f0 |
996 | } |
a0d0e21e |
997 | |
998 | I32 |
864dbfa3 |
999 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1000 | { |
c07a80fd |
1001 | register I32 count; |
a0d0e21e |
1002 | STRLEN len; |
9b33ce3b |
1003 | STRLEN n_a; |
a0d0e21e |
1004 | char *s; |
c07a80fd |
1005 | |
3280af22 |
1006 | if (RsSNARF(PL_rs)) |
c07a80fd |
1007 | return 0; |
4c5a6083 |
1008 | if (RsRECORD(PL_rs)) |
1009 | return 0; |
c07a80fd |
1010 | count = 0; |
a0d0e21e |
1011 | if (SvTYPE(sv) == SVt_PVAV) { |
1012 | register I32 i; |
1013 | I32 max; |
1014 | AV* av = (AV*)sv; |
1015 | max = AvFILL(av); |
1016 | for (i = 0; i <= max; i++) { |
1017 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1018 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1019 | count += do_chomp(sv); |
1020 | } |
1021 | return count; |
1022 | } |
aa854799 |
1023 | else if (SvTYPE(sv) == SVt_PVHV) { |
a0d0e21e |
1024 | HV* hv = (HV*)sv; |
1025 | HE* entry; |
1026 | (void)hv_iterinit(hv); |
1027 | /*SUPPRESS 560*/ |
155aba94 |
1028 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1029 | count += do_chomp(hv_iterval(hv,entry)); |
1030 | return count; |
1031 | } |
991e6d41 |
1032 | else if (SvREADONLY(sv)) { |
1033 | if (SvFAKE(sv)) { |
1034 | /* SV is copy-on-write */ |
1035 | sv_force_normal_flags(sv, 0); |
1036 | } |
1037 | if (SvREADONLY(sv)) |
1038 | Perl_croak(aTHX_ PL_no_modify); |
1039 | } |
a0d0e21e |
1040 | s = SvPV(sv, len); |
a0d0e21e |
1041 | if (s && len) { |
1042 | s += --len; |
3280af22 |
1043 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1044 | if (*s != '\n') |
1045 | goto nope; |
1046 | ++count; |
1047 | while (len && s[-1] == '\n') { |
1048 | --len; |
1049 | --s; |
1050 | ++count; |
1051 | } |
1052 | } |
a0d0e21e |
1053 | else { |
c07a80fd |
1054 | STRLEN rslen; |
3280af22 |
1055 | char *rsptr = SvPV(PL_rs, rslen); |
c07a80fd |
1056 | if (rslen == 1) { |
1057 | if (*s != *rsptr) |
1058 | goto nope; |
1059 | ++count; |
1060 | } |
1061 | else { |
8c2cee6f |
1062 | if (len < rslen - 1) |
c07a80fd |
1063 | goto nope; |
1064 | len -= rslen - 1; |
1065 | s -= rslen - 1; |
36477c24 |
1066 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1067 | goto nope; |
1068 | count += rslen; |
1069 | } |
a0d0e21e |
1070 | } |
9b33ce3b |
1071 | s = SvPV_force(sv, n_a); |
a0d0e21e |
1072 | SvCUR_set(sv, len); |
9b33ce3b |
1073 | *SvEND(sv) = '\0'; |
a0d0e21e |
1074 | SvNIOK_off(sv); |
9b33ce3b |
1075 | SvSETMAGIC(sv); |
a0d0e21e |
1076 | } |
1077 | nope: |
a0d0e21e |
1078 | return count; |
8e84507e |
1079 | } |
79072805 |
1080 | |
1081 | void |
864dbfa3 |
1082 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1083 | { |
1084 | #ifdef LIBERAL |
1085 | register long *dl; |
1086 | register long *ll; |
1087 | register long *rl; |
1088 | #endif |
1089 | register char *dc; |
463ee0b2 |
1090 | STRLEN leftlen; |
1091 | STRLEN rightlen; |
7a4c00b4 |
1092 | register char *lc; |
1093 | register char *rc; |
79072805 |
1094 | register I32 len; |
a0d0e21e |
1095 | I32 lensave; |
7a4c00b4 |
1096 | char *lsave; |
1097 | char *rsave; |
0c57e439 |
1098 | bool left_utf = DO_UTF8(left); |
1099 | bool right_utf = DO_UTF8(right); |
b7953727 |
1100 | I32 needlen = 0; |
0c57e439 |
1101 | |
1102 | if (left_utf && !right_utf) |
1103 | sv_utf8_upgrade(right); |
a1ca4561 |
1104 | else if (!left_utf && right_utf) |
0c57e439 |
1105 | sv_utf8_upgrade(left); |
79072805 |
1106 | |
1fbd88dc |
1107 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1108 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
7a4c00b4 |
1109 | lsave = lc = SvPV(left, leftlen); |
1110 | rsave = rc = SvPV(right, rightlen); |
93a17b20 |
1111 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1112 | lensave = len; |
c9b3c8d0 |
1113 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1114 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
1115 | Newz(801, dc, needlen + 1, char); |
1116 | } |
1117 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2d8e6c8d |
1118 | STRLEN n_a; |
1119 | dc = SvPV_force(sv, n_a); |
eb160463 |
1120 | if (SvCUR(sv) < (STRLEN)len) { |
1121 | dc = SvGROW(sv, (STRLEN)(len + 1)); |
ff68c719 |
1122 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1123 | } |
c9b3c8d0 |
1124 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1125 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1126 | } |
1127 | else { |
c9b3c8d0 |
1128 | needlen = ((optype == OP_BIT_AND) |
1129 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
ff68c719 |
1130 | Newz(801, dc, needlen + 1, char); |
1131 | (void)sv_usepvn(sv, dc, needlen); |
1132 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1133 | } |
a0d0e21e |
1134 | SvCUR_set(sv, len); |
1135 | (void)SvPOK_only(sv); |
0c57e439 |
1136 | if (left_utf || right_utf) { |
1137 | UV duc, luc, ruc; |
c9b3c8d0 |
1138 | char *dcsave = dc; |
0c57e439 |
1139 | STRLEN lulen = leftlen; |
1140 | STRLEN rulen = rightlen; |
ba210ebe |
1141 | STRLEN ulen; |
0c57e439 |
1142 | |
1143 | switch (optype) { |
1144 | case OP_BIT_AND: |
1145 | while (lulen && rulen) { |
9041c2e3 |
1146 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1147 | lc += ulen; |
1148 | lulen -= ulen; |
9041c2e3 |
1149 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1150 | rc += ulen; |
1151 | rulen -= ulen; |
1152 | duc = luc & ruc; |
9041c2e3 |
1153 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1154 | } |
c9b3c8d0 |
1155 | if (sv == left || sv == right) |
1156 | (void)sv_usepvn(sv, dcsave, needlen); |
1157 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1158 | break; |
1159 | case OP_BIT_XOR: |
1160 | while (lulen && rulen) { |
9041c2e3 |
1161 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1162 | lc += ulen; |
1163 | lulen -= ulen; |
9041c2e3 |
1164 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1165 | rc += ulen; |
1166 | rulen -= ulen; |
1167 | duc = luc ^ ruc; |
9041c2e3 |
1168 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1169 | } |
1170 | goto mop_up_utf; |
1171 | case OP_BIT_OR: |
1172 | while (lulen && rulen) { |
9041c2e3 |
1173 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1174 | lc += ulen; |
1175 | lulen -= ulen; |
9041c2e3 |
1176 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1177 | rc += ulen; |
1178 | rulen -= ulen; |
1179 | duc = luc | ruc; |
9041c2e3 |
1180 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1181 | } |
1182 | mop_up_utf: |
c9b3c8d0 |
1183 | if (sv == left || sv == right) |
1184 | (void)sv_usepvn(sv, dcsave, needlen); |
1185 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1186 | if (rulen) |
1187 | sv_catpvn(sv, rc, rulen); |
1188 | else if (lulen) |
1189 | sv_catpvn(sv, lc, lulen); |
1190 | else |
1191 | *SvEND(sv) = '\0'; |
1192 | break; |
1193 | } |
1194 | SvUTF8_on(sv); |
1195 | goto finish; |
1196 | } |
1197 | else |
79072805 |
1198 | #ifdef LIBERAL |
1199 | if (len >= sizeof(long)*4 && |
1200 | !((long)dc % sizeof(long)) && |
1201 | !((long)lc % sizeof(long)) && |
1202 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1203 | { |
1204 | I32 remainder = len % (sizeof(long)*4); |
1205 | len /= (sizeof(long)*4); |
1206 | |
1207 | dl = (long*)dc; |
1208 | ll = (long*)lc; |
1209 | rl = (long*)rc; |
1210 | |
1211 | switch (optype) { |
1212 | case OP_BIT_AND: |
1213 | while (len--) { |
1214 | *dl++ = *ll++ & *rl++; |
1215 | *dl++ = *ll++ & *rl++; |
1216 | *dl++ = *ll++ & *rl++; |
1217 | *dl++ = *ll++ & *rl++; |
1218 | } |
1219 | break; |
a0d0e21e |
1220 | case OP_BIT_XOR: |
79072805 |
1221 | while (len--) { |
1222 | *dl++ = *ll++ ^ *rl++; |
1223 | *dl++ = *ll++ ^ *rl++; |
1224 | *dl++ = *ll++ ^ *rl++; |
1225 | *dl++ = *ll++ ^ *rl++; |
1226 | } |
1227 | break; |
1228 | case OP_BIT_OR: |
1229 | while (len--) { |
1230 | *dl++ = *ll++ | *rl++; |
1231 | *dl++ = *ll++ | *rl++; |
1232 | *dl++ = *ll++ | *rl++; |
1233 | *dl++ = *ll++ | *rl++; |
1234 | } |
1235 | } |
1236 | |
1237 | dc = (char*)dl; |
1238 | lc = (char*)ll; |
1239 | rc = (char*)rl; |
1240 | |
1241 | len = remainder; |
1242 | } |
1243 | #endif |
a0d0e21e |
1244 | { |
a0d0e21e |
1245 | switch (optype) { |
1246 | case OP_BIT_AND: |
1247 | while (len--) |
1248 | *dc++ = *lc++ & *rc++; |
1249 | break; |
1250 | case OP_BIT_XOR: |
1251 | while (len--) |
1252 | *dc++ = *lc++ ^ *rc++; |
1253 | goto mop_up; |
1254 | case OP_BIT_OR: |
1255 | while (len--) |
1256 | *dc++ = *lc++ | *rc++; |
1257 | mop_up: |
1258 | len = lensave; |
eb160463 |
1259 | if (rightlen > (STRLEN)len) |
a0d0e21e |
1260 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1261 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1262 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1263 | else |
1264 | *SvEND(sv) = '\0'; |
a0d0e21e |
1265 | break; |
1266 | } |
79072805 |
1267 | } |
0c57e439 |
1268 | finish: |
fb73857a |
1269 | SvTAINT(sv); |
79072805 |
1270 | } |
463ee0b2 |
1271 | |
1272 | OP * |
cea2e8a9 |
1273 | Perl_do_kv(pTHX) |
463ee0b2 |
1274 | { |
39644a26 |
1275 | dSP; |
463ee0b2 |
1276 | HV *hv = (HV*)POPs; |
800e9ae0 |
1277 | HV *keys; |
463ee0b2 |
1278 | register HE *entry; |
463ee0b2 |
1279 | SV *tmpstr; |
54310121 |
1280 | I32 gimme = GIMME_V; |
533c011a |
1281 | I32 dokeys = (PL_op->op_type == OP_KEYS); |
1282 | I32 dovalues = (PL_op->op_type == OP_VALUES); |
8e84507e |
1283 | |
1284 | if (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV) |
a0d0e21e |
1285 | dokeys = dovalues = TRUE; |
463ee0b2 |
1286 | |
85581909 |
1287 | if (!hv) { |
78f9721b |
1288 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1289 | dTARGET; /* make sure to clear its target here */ |
1290 | if (SvTYPE(TARG) == SVt_PVLV) |
1291 | LvTARG(TARG) = Nullsv; |
1292 | PUSHs(TARG); |
1293 | } |
463ee0b2 |
1294 | RETURN; |
85581909 |
1295 | } |
748a9306 |
1296 | |
6d822dc4 |
1297 | keys = hv; |
800e9ae0 |
1298 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1299 | |
54310121 |
1300 | if (gimme == G_VOID) |
aa689395 |
1301 | RETURN; |
1302 | |
54310121 |
1303 | if (gimme == G_SCALAR) { |
6ee623d5 |
1304 | IV i; |
463ee0b2 |
1305 | dTARGET; |
1306 | |
78f9721b |
1307 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1308 | if (SvTYPE(TARG) < SVt_PVLV) { |
1309 | sv_upgrade(TARG, SVt_PVLV); |
14befaf4 |
1310 | sv_magic(TARG, Nullsv, PERL_MAGIC_nkeys, Nullch, 0); |
85581909 |
1311 | } |
1312 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1313 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1314 | if (LvTARG(TARG)) |
1315 | SvREFCNT_dec(LvTARG(TARG)); |
800e9ae0 |
1316 | LvTARG(TARG) = SvREFCNT_inc(keys); |
6ff81951 |
1317 | } |
85581909 |
1318 | PUSHs(TARG); |
1319 | RETURN; |
1320 | } |
1321 | |
14befaf4 |
1322 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 |
1323 | i = HvKEYS(keys); |
463ee0b2 |
1324 | else { |
1325 | i = 0; |
463ee0b2 |
1326 | /*SUPPRESS 560*/ |
800e9ae0 |
1327 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1328 | } |
1329 | PUSHi( i ); |
1330 | RETURN; |
1331 | } |
1332 | |
8ed4b672 |
1333 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1334 | |
463ee0b2 |
1335 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1336 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1337 | SPAGAIN; |
574c8022 |
1338 | if (dokeys) { |
1339 | SV* sv = hv_iterkeysv(entry); |
574c8022 |
1340 | XPUSHs(sv); /* won't clobber stack_sp */ |
1341 | } |
463ee0b2 |
1342 | if (dovalues) { |
463ee0b2 |
1343 | PUTBACK; |
6d822dc4 |
1344 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1345 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1346 | (unsigned long)HeHASH(entry), |
800e9ae0 |
1347 | HvMAX(keys)+1, |
1348 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1349 | SPAGAIN; |
46fc3d4c |
1350 | XPUSHs(tmpstr); |
463ee0b2 |
1351 | } |
1352 | PUTBACK; |
1353 | } |
1354 | return NORMAL; |
1355 | } |
4e35701f |
1356 | |