Commit | Line | Data |
a0d0e21e |
1 | /* doop.c |
79072805 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
b94e2f88 |
4 | * 2000, 2001, 2002, 2004, 2005, 2006, 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 | |
166f8a29 |
15 | /* This file contains some common functions needed to carry out certain |
16 | * ops. For example both pp_schomp() and pp_chomp() - scalar and array |
17 | * chomp operations - call the function do_chomp() found in this file. |
18 | */ |
19 | |
79072805 |
20 | #include "EXTERN.h" |
864dbfa3 |
21 | #define PERL_IN_DOOP_C |
79072805 |
22 | #include "perl.h" |
23 | |
64ca3a65 |
24 | #ifndef PERL_MICRO |
79072805 |
25 | #include <signal.h> |
26 | #endif |
27 | |
942e002e |
28 | STATIC I32 |
c395bd6c |
29 | S_do_trans_simple(pTHX_ SV * const sv) |
79072805 |
30 | { |
97aff369 |
31 | dVAR; |
4757a243 |
32 | I32 matches = 0; |
463ee0b2 |
33 | STRLEN len; |
c395bd6c |
34 | U8 *s = (U8*)SvPV(sv,len); |
35 | U8 * const send = s+len; |
79072805 |
36 | |
c4420975 |
37 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
38 | if (!tbl) |
7d85a32c |
39 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
40 | |
1e54db1a |
41 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
42 | if (!SvUTF8(sv)) { |
01ec43d0 |
43 | while (s < send) { |
f54cb97a |
44 | const I32 ch = tbl[*s]; |
45 | if (ch >= 0) { |
01ec43d0 |
46 | matches++; |
c4420975 |
47 | *s = (U8)ch; |
01ec43d0 |
48 | } |
c4420975 |
49 | s++; |
01ec43d0 |
50 | } |
51 | SvSETMAGIC(sv); |
9b877dbb |
52 | } |
53 | else { |
c395bd6c |
54 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
55 | U8 *d; |
56 | U8 *dstart; |
57 | |
58 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
59 | if (grows) |
60 | Newx(d, len*2+1, U8); |
61 | else |
62 | d = s; |
63 | dstart = d; |
64 | while (s < send) { |
65 | STRLEN ulen; |
66 | I32 ch; |
67 | |
68 | /* Need to check this, otherwise 128..255 won't match */ |
69 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
70 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
71 | matches++; |
72 | d = uvchr_to_utf8(d, ch); |
73 | s += ulen; |
74 | } |
75 | else { /* No match -> copy */ |
76 | Move(s, d, ulen, U8); |
77 | d += ulen; |
78 | s += ulen; |
79 | } |
80 | } |
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 | } |
89 | SvUTF8_on(sv); |
90 | SvSETMAGIC(sv); |
9b877dbb |
91 | } |
4757a243 |
92 | return matches; |
93 | } |
94 | |
942e002e |
95 | STATIC I32 |
c395bd6c |
96 | S_do_trans_count(pTHX_ SV * const sv) |
4757a243 |
97 | { |
97aff369 |
98 | dVAR; |
4757a243 |
99 | STRLEN len; |
c395bd6c |
100 | const U8 *s = (const U8*)SvPV_const(sv, len); |
101 | const U8 * const send = s + len; |
102 | I32 matches = 0; |
4757a243 |
103 | |
f54cb97a |
104 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
105 | if (!tbl) |
7d85a32c |
106 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
107 | |
c395bd6c |
108 | if (!SvUTF8(sv)) { |
1aa99e6b |
109 | while (s < send) { |
110 | if (tbl[*s++] >= 0) |
036b4402 |
111 | matches++; |
1aa99e6b |
112 | } |
c395bd6c |
113 | } |
fabdb6c0 |
114 | else { |
115 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
1aa99e6b |
116 | while (s < send) { |
1aa99e6b |
117 | STRLEN ulen; |
9f7f3913 |
118 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 |
119 | if (c < 0x100) { |
120 | if (tbl[c] >= 0) |
121 | matches++; |
122 | } else if (complement) |
1aa99e6b |
123 | matches++; |
124 | s += ulen; |
125 | } |
fabdb6c0 |
126 | } |
4757a243 |
127 | |
128 | return matches; |
129 | } |
130 | |
942e002e |
131 | STATIC I32 |
c395bd6c |
132 | S_do_trans_complex(pTHX_ SV * const sv) |
4757a243 |
133 | { |
97aff369 |
134 | dVAR; |
c395bd6c |
135 | STRLEN len; |
136 | U8 *s = (U8*)SvPV(sv, len); |
137 | U8 * const send = s+len; |
4757a243 |
138 | I32 matches = 0; |
4757a243 |
139 | |
f54cb97a |
140 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
141 | if (!tbl) |
7d85a32c |
142 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 |
143 | |
c395bd6c |
144 | if (!SvUTF8(sv)) { |
145 | U8 *d = s; |
146 | U8 * const dstart = d; |
4757a243 |
147 | |
1aa99e6b |
148 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a |
149 | const U8* p = send; |
1aa99e6b |
150 | while (s < send) { |
f54cb97a |
151 | const I32 ch = tbl[*s]; |
152 | if (ch >= 0) { |
eb160463 |
153 | *d = (U8)ch; |
036b4402 |
154 | matches++; |
1aa99e6b |
155 | if (p != d - 1 || *p != *d) |
156 | p = d++; |
157 | } |
158 | else if (ch == -1) /* -1 is unmapped character */ |
159 | *d++ = *s; |
160 | else if (ch == -2) /* -2 is delete character */ |
161 | matches++; |
162 | s++; |
163 | } |
a0ed51b3 |
164 | } |
1aa99e6b |
165 | else { |
166 | while (s < send) { |
f54cb97a |
167 | const I32 ch = tbl[*s]; |
168 | if (ch >= 0) { |
1aa99e6b |
169 | matches++; |
eb160463 |
170 | *d++ = (U8)ch; |
1aa99e6b |
171 | } |
172 | else if (ch == -1) /* -1 is unmapped character */ |
173 | *d++ = *s; |
174 | else if (ch == -2) /* -2 is delete character */ |
175 | matches++; |
176 | s++; |
177 | } |
178 | } |
76ef7183 |
179 | *d = '\0'; |
1aa99e6b |
180 | SvCUR_set(sv, d - dstart); |
4757a243 |
181 | } |
c395bd6c |
182 | else { /* is utf8 */ |
fabdb6c0 |
183 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
184 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
185 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
c395bd6c |
186 | U8 *d; |
187 | U8 *dstart; |
188 | STRLEN rlen = 0; |
fabdb6c0 |
189 | |
9b877dbb |
190 | if (grows) |
a02a5408 |
191 | Newx(d, len*2+1, U8); |
9b877dbb |
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; |
9f7f3913 |
206 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
207 | UTF8_ALLOW_DEFAULT); |
f54cb97a |
208 | I32 ch; |
1aa99e6b |
209 | |
8973db79 |
210 | if (comp > 0xff) { |
211 | if (!complement) { |
640af7f1 |
212 | Move(s, d, len, U8); |
8973db79 |
213 | d += len; |
214 | } |
215 | else { |
216 | matches++; |
217 | if (!del) { |
bb7a0f54 |
218 | ch = (rlen == 0) ? (I32)comp : |
05d340b8 |
219 | (comp - 0x100 < rlen) ? |
8973db79 |
220 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 |
221 | if ((UV)ch != pch) { |
9041c2e3 |
222 | d = uvchr_to_utf8(d, ch); |
eb160463 |
223 | pch = (UV)ch; |
8973db79 |
224 | } |
225 | s += len; |
226 | continue; |
227 | } |
228 | } |
9b877dbb |
229 | } |
1aa99e6b |
230 | else if ((ch = tbl[comp]) >= 0) { |
231 | matches++; |
eb160463 |
232 | if ((UV)ch != pch) { |
9041c2e3 |
233 | d = uvchr_to_utf8(d, ch); |
eb160463 |
234 | pch = (UV)ch; |
1aa99e6b |
235 | } |
236 | s += len; |
237 | continue; |
238 | } |
9b877dbb |
239 | else if (ch == -1) { /* -1 is unmapped character */ |
640af7f1 |
240 | Move(s, d, len, U8); |
9b877dbb |
241 | d += len; |
242 | } |
1aa99e6b |
243 | else if (ch == -2) /* -2 is delete character */ |
244 | matches++; |
245 | s += len; |
246 | pch = 0xfeedface; |
247 | } |
5d06d08e |
248 | } |
1aa99e6b |
249 | else { |
250 | while (s < send) { |
251 | STRLEN len; |
9f7f3913 |
252 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
253 | UTF8_ALLOW_DEFAULT); |
f54cb97a |
254 | I32 ch; |
8973db79 |
255 | if (comp > 0xff) { |
256 | if (!complement) { |
0376ff32 |
257 | Move(s, d, len, U8); |
8973db79 |
258 | d += len; |
259 | } |
260 | else { |
261 | matches++; |
262 | if (!del) { |
9041c2e3 |
263 | if (comp - 0x100 < rlen) |
264 | d = uvchr_to_utf8(d, tbl[comp+1]); |
8973db79 |
265 | else |
9041c2e3 |
266 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 |
267 | } |
268 | } |
9b877dbb |
269 | } |
1aa99e6b |
270 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 |
271 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
272 | matches++; |
273 | } |
274 | else if (ch == -1) { /* -1 is unmapped character */ |
640af7f1 |
275 | Move(s, d, len, U8); |
9b877dbb |
276 | d += len; |
1aa99e6b |
277 | } |
278 | else if (ch == -2) /* -2 is delete character */ |
279 | matches++; |
280 | s += len; |
281 | } |
282 | } |
9b877dbb |
283 | if (grows) { |
284 | sv_setpvn(sv, (char*)dstart, d - dstart); |
285 | Safefree(dstart); |
286 | } |
287 | else { |
288 | *d = '\0'; |
289 | SvCUR_set(sv, d - dstart); |
290 | } |
1aa99e6b |
291 | SvUTF8_on(sv); |
4757a243 |
292 | } |
5e44153e |
293 | SvSETMAGIC(sv); |
4757a243 |
294 | return matches; |
295 | } |
296 | |
942e002e |
297 | STATIC I32 |
c395bd6c |
298 | S_do_trans_simple_utf8(pTHX_ SV * const sv) |
4757a243 |
299 | { |
97aff369 |
300 | dVAR; |
4757a243 |
301 | U8 *s; |
302 | U8 *send; |
303 | U8 *d; |
036b4402 |
304 | U8 *start; |
1aa99e6b |
305 | U8 *dstart, *dend; |
4757a243 |
306 | I32 matches = 0; |
f54cb97a |
307 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
308 | STRLEN len; |
309 | |
53c1dcc0 |
310 | SV* const rv = (SV*)cSVOP->op_sv; |
311 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
312 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
313 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
314 | const UV extra = none + 1; |
b7953727 |
315 | UV final = 0; |
1aa99e6b |
316 | U8 hibit = 0; |
4757a243 |
317 | |
318 | s = (U8*)SvPV(sv, len); |
c395bd6c |
319 | if (!SvUTF8(sv)) { |
c4420975 |
320 | const U8 *t = s; |
321 | const U8 * const e = s + len; |
c4d5f83a |
322 | while (t < e) { |
f54cb97a |
323 | const U8 ch = *t++; |
fabdb6c0 |
324 | hibit = !NATIVE_IS_INVARIANT(ch); |
325 | if (hibit) { |
326 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
327 | break; |
fabdb6c0 |
328 | } |
c4d5f83a |
329 | } |
1aa99e6b |
330 | } |
4757a243 |
331 | send = s + len; |
036b4402 |
332 | start = s; |
4757a243 |
333 | |
a4fc7abc |
334 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 |
335 | if (svp) |
336 | final = SvUV(*svp); |
337 | |
9b877dbb |
338 | if (grows) { |
339 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
340 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
341 | dend = d + len * 3; |
342 | dstart = d; |
343 | } |
344 | else { |
345 | dstart = d = s; |
346 | dend = d + len; |
347 | } |
1aa99e6b |
348 | |
4757a243 |
349 | while (s < send) { |
0bcc34c2 |
350 | const UV uv = swash_fetch(rv, s, TRUE); |
351 | if (uv < none) { |
4757a243 |
352 | s += UTF8SKIP(s); |
353 | matches++; |
7d85a32c |
354 | d = uvuni_to_utf8(d, uv); |
4757a243 |
355 | } |
356 | else if (uv == none) { |
4373e329 |
357 | const int i = UTF8SKIP(s); |
0376ff32 |
358 | Move(s, d, i, U8); |
9b877dbb |
359 | d += i; |
360 | s += i; |
4757a243 |
361 | } |
362 | else if (uv == extra) { |
4373e329 |
363 | s += UTF8SKIP(s); |
4757a243 |
364 | matches++; |
7d85a32c |
365 | d = uvuni_to_utf8(d, final); |
4757a243 |
366 | } |
367 | else |
368 | s += UTF8SKIP(s); |
1aa99e6b |
369 | |
9b877dbb |
370 | if (d > dend) { |
4373e329 |
371 | const STRLEN clen = d - dstart; |
372 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
373 | if (!grows) |
7d85a32c |
374 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 |
375 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
376 | d = dstart + clen; |
377 | dend = dstart + nlen; |
378 | } |
4757a243 |
379 | } |
16ec844d |
380 | if (grows || hibit) { |
9b877dbb |
381 | sv_setpvn(sv, (char*)dstart, d - dstart); |
382 | Safefree(dstart); |
16ec844d |
383 | if (grows && hibit) |
384 | Safefree(start); |
9b877dbb |
385 | } |
386 | else { |
387 | *d = '\0'; |
388 | SvCUR_set(sv, d - dstart); |
389 | } |
4757a243 |
390 | SvSETMAGIC(sv); |
1aa99e6b |
391 | SvUTF8_on(sv); |
4757a243 |
392 | |
393 | return matches; |
394 | } |
395 | |
942e002e |
396 | STATIC I32 |
c395bd6c |
397 | S_do_trans_count_utf8(pTHX_ SV * const sv) |
4757a243 |
398 | { |
97aff369 |
399 | dVAR; |
93524f2b |
400 | const U8 *s; |
cbbf8932 |
401 | const U8 *start = NULL; |
402 | const U8 *send; |
4757a243 |
403 | I32 matches = 0; |
404 | STRLEN len; |
405 | |
53c1dcc0 |
406 | SV* const rv = (SV*)cSVOP->op_sv; |
407 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
408 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
409 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
410 | const UV extra = none + 1; |
1aa99e6b |
411 | U8 hibit = 0; |
4757a243 |
412 | |
93524f2b |
413 | s = (const U8*)SvPV_const(sv, len); |
1aa99e6b |
414 | if (!SvUTF8(sv)) { |
4373e329 |
415 | const U8 *t = s; |
c4420975 |
416 | const U8 * const e = s + len; |
c4d5f83a |
417 | while (t < e) { |
4373e329 |
418 | const U8 ch = *t++; |
fabdb6c0 |
419 | hibit = !NATIVE_IS_INVARIANT(ch); |
420 | if (hibit) { |
421 | start = s = bytes_to_utf8(s, &len); |
1aa99e6b |
422 | break; |
fabdb6c0 |
423 | } |
c4d5f83a |
424 | } |
1aa99e6b |
425 | } |
4757a243 |
426 | send = s + len; |
427 | |
428 | while (s < send) { |
c4420975 |
429 | const UV uv = swash_fetch(rv, s, TRUE); |
430 | if (uv < none || uv == extra) |
4757a243 |
431 | matches++; |
834a4ddd |
432 | s += UTF8SKIP(s); |
4757a243 |
433 | } |
1aa99e6b |
434 | if (hibit) |
435 | Safefree(start); |
4757a243 |
436 | |
437 | return matches; |
438 | } |
439 | |
942e002e |
440 | STATIC I32 |
c395bd6c |
441 | S_do_trans_complex_utf8(pTHX_ SV * const sv) |
4757a243 |
442 | { |
97aff369 |
443 | dVAR; |
1aa99e6b |
444 | U8 *start, *send; |
4757a243 |
445 | U8 *d; |
446 | I32 matches = 0; |
4373e329 |
447 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
448 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
449 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
53c1dcc0 |
450 | SV * const rv = (SV*)cSVOP->op_sv; |
451 | HV * const hv = (HV*)SvRV(rv); |
a4fc7abc |
452 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
453 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
454 | const UV extra = none + 1; |
b7953727 |
455 | UV final = 0; |
45005bfb |
456 | bool havefinal = FALSE; |
4757a243 |
457 | STRLEN len; |
1aa99e6b |
458 | U8 *dstart, *dend; |
1aa99e6b |
459 | U8 hibit = 0; |
4757a243 |
460 | |
f54cb97a |
461 | U8 *s = (U8*)SvPV(sv, len); |
c395bd6c |
462 | if (!SvUTF8(sv)) { |
53c1dcc0 |
463 | const U8 *t = s; |
464 | const U8 * const e = s + len; |
c4d5f83a |
465 | while (t < e) { |
f54cb97a |
466 | const U8 ch = *t++; |
fabdb6c0 |
467 | hibit = !NATIVE_IS_INVARIANT(ch); |
468 | if (hibit) { |
469 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
470 | break; |
fabdb6c0 |
471 | } |
c4d5f83a |
472 | } |
1aa99e6b |
473 | } |
4757a243 |
474 | send = s + len; |
1aa99e6b |
475 | start = s; |
4757a243 |
476 | |
a4fc7abc |
477 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb |
478 | if (svp) { |
4757a243 |
479 | final = SvUV(*svp); |
45005bfb |
480 | havefinal = TRUE; |
481 | } |
4757a243 |
482 | |
9b877dbb |
483 | if (grows) { |
484 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
485 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
486 | dend = d + len * 3; |
487 | dstart = d; |
488 | } |
489 | else { |
490 | dstart = d = s; |
491 | dend = d + len; |
492 | } |
4757a243 |
493 | |
494 | if (squash) { |
495 | UV puv = 0xfeedface; |
496 | while (s < send) { |
f54cb97a |
497 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
498 | |
9b877dbb |
499 | if (d > dend) { |
4373e329 |
500 | const STRLEN clen = d - dstart; |
501 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
502 | if (!grows) |
7d85a32c |
503 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
504 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
505 | d = dstart + clen; |
506 | dend = dstart + nlen; |
4757a243 |
507 | } |
508 | if (uv < none) { |
509 | matches++; |
94472101 |
510 | s += UTF8SKIP(s); |
4757a243 |
511 | if (uv != puv) { |
7d85a32c |
512 | d = uvuni_to_utf8(d, uv); |
4757a243 |
513 | puv = uv; |
514 | } |
4757a243 |
515 | continue; |
516 | } |
517 | else if (uv == none) { /* "none" is unmapped character */ |
4373e329 |
518 | const int i = UTF8SKIP(s); |
0376ff32 |
519 | Move(s, d, i, U8); |
9b877dbb |
520 | d += i; |
521 | s += i; |
4757a243 |
522 | puv = 0xfeedface; |
523 | continue; |
a0ed51b3 |
524 | } |
4757a243 |
525 | else if (uv == extra && !del) { |
526 | matches++; |
45005bfb |
527 | if (havefinal) { |
528 | s += UTF8SKIP(s); |
529 | if (puv != final) { |
7d85a32c |
530 | d = uvuni_to_utf8(d, final); |
45005bfb |
531 | puv = final; |
532 | } |
533 | } |
534 | else { |
535 | STRLEN len; |
9f7f3913 |
536 | uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb |
537 | if (uv != puv) { |
0376ff32 |
538 | Move(s, d, len, U8); |
45005bfb |
539 | d += len; |
540 | puv = uv; |
541 | } |
542 | s += len; |
4757a243 |
543 | } |
4757a243 |
544 | continue; |
545 | } |
01ec43d0 |
546 | matches++; /* "none+1" is delete character */ |
547 | s += UTF8SKIP(s); |
a0ed51b3 |
548 | } |
79072805 |
549 | } |
550 | else { |
4757a243 |
551 | while (s < send) { |
f54cb97a |
552 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
553 | if (d > dend) { |
f54cb97a |
554 | const STRLEN clen = d - dstart; |
555 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
556 | if (!grows) |
7d85a32c |
557 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
558 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
559 | d = dstart + clen; |
560 | dend = dstart + nlen; |
a0ed51b3 |
561 | } |
4757a243 |
562 | if (uv < none) { |
563 | matches++; |
01ec43d0 |
564 | s += UTF8SKIP(s); |
7d85a32c |
565 | d = uvuni_to_utf8(d, uv); |
4757a243 |
566 | continue; |
a0ed51b3 |
567 | } |
4757a243 |
568 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a |
569 | const int i = UTF8SKIP(s); |
0376ff32 |
570 | Move(s, d, i, U8); |
9b877dbb |
571 | d += i; |
572 | s += i; |
4757a243 |
573 | continue; |
79072805 |
574 | } |
4757a243 |
575 | else if (uv == extra && !del) { |
576 | matches++; |
01ec43d0 |
577 | s += UTF8SKIP(s); |
7d85a32c |
578 | d = uvuni_to_utf8(d, final); |
4757a243 |
579 | continue; |
580 | } |
01ec43d0 |
581 | matches++; /* "none+1" is delete character */ |
582 | s += UTF8SKIP(s); |
79072805 |
583 | } |
4757a243 |
584 | } |
16ec844d |
585 | if (grows || hibit) { |
9b877dbb |
586 | sv_setpvn(sv, (char*)dstart, d - dstart); |
587 | Safefree(dstart); |
16ec844d |
588 | if (grows && hibit) |
589 | Safefree(start); |
9b877dbb |
590 | } |
591 | else { |
592 | *d = '\0'; |
593 | SvCUR_set(sv, d - dstart); |
594 | } |
1aa99e6b |
595 | SvUTF8_on(sv); |
4757a243 |
596 | SvSETMAGIC(sv); |
597 | |
598 | return matches; |
599 | } |
600 | |
601 | I32 |
864dbfa3 |
602 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
603 | { |
97aff369 |
604 | dVAR; |
4757a243 |
605 | STRLEN len; |
4373e329 |
606 | const I32 hasutf = (PL_op->op_private & |
036b4402 |
607 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
608 | |
2233f375 |
609 | if (SvREADONLY(sv)) { |
765f542d |
610 | if (SvIsCOW(sv)) |
611 | sv_force_normal_flags(sv, 0); |
2233f375 |
612 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
613 | Perl_croak(aTHX_ PL_no_modify); |
614 | } |
10516c54 |
615 | (void)SvPV_const(sv, len); |
4757a243 |
616 | if (!len) |
617 | return 0; |
d59e14db |
618 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
619 | if (!SvPOKp(sv)) |
620 | (void)SvPV_force(sv, len); |
2de7b02f |
621 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
622 | } |
4757a243 |
623 | |
cea2e8a9 |
624 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
625 | |
a77e643a |
626 | switch (PL_op->op_private & ~hasutf & ( |
627 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| |
628 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { |
4757a243 |
629 | case 0: |
01ec43d0 |
630 | if (hasutf) |
631 | return do_trans_simple_utf8(sv); |
632 | else |
633 | return do_trans_simple(sv); |
4757a243 |
634 | |
635 | case OPpTRANS_IDENTICAL: |
05d340b8 |
636 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
637 | if (hasutf) |
638 | return do_trans_count_utf8(sv); |
639 | else |
640 | return do_trans_count(sv); |
4757a243 |
641 | |
642 | default: |
01ec43d0 |
643 | if (hasutf) |
036b4402 |
644 | return do_trans_complex_utf8(sv); |
4757a243 |
645 | else |
036b4402 |
646 | return do_trans_complex(sv); |
79072805 |
647 | } |
79072805 |
648 | } |
649 | |
650 | void |
b9ac451d |
651 | Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp) |
79072805 |
652 | { |
97aff369 |
653 | dVAR; |
53c1dcc0 |
654 | SV ** const oldmark = mark; |
79072805 |
655 | register I32 items = sp - mark; |
79072805 |
656 | register STRLEN len; |
463ee0b2 |
657 | STRLEN delimlen; |
79072805 |
658 | |
b9ac451d |
659 | (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */ |
5ba99574 |
660 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
661 | |
79072805 |
662 | mark++; |
663 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
862a34c6 |
664 | SvUPGRADE(sv, SVt_PV); |
79072805 |
665 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
666 | while (items-- > 0) { |
1426bbf4 |
667 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a |
668 | STRLEN tmplen; |
5c144d81 |
669 | SvPV_const(*mark, tmplen); |
463ee0b2 |
670 | len += tmplen; |
79072805 |
671 | } |
672 | mark++; |
673 | } |
674 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
675 | |
676 | mark = oldmark; |
db7c17d7 |
677 | items = sp - mark; |
79072805 |
678 | ++mark; |
679 | } |
680 | |
fc0199cd |
681 | sv_setpvn(sv, "", 0); |
e4803c42 |
682 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 |
683 | SvUTF8_off(sv); |
e4803c42 |
684 | |
8d6d96c1 |
685 | if (PL_tainting && SvMAGICAL(sv)) |
686 | SvTAINTED_off(sv); |
687 | |
463ee0b2 |
688 | if (items-- > 0) { |
92d29cee |
689 | if (*mark) |
690 | sv_catsv(sv, *mark); |
463ee0b2 |
691 | mark++; |
692 | } |
8d6d96c1 |
693 | |
c512ce4f |
694 | if (delimlen) { |
79072805 |
695 | for (; items > 0; items--,mark++) { |
b9ac451d |
696 | sv_catsv(sv,delim); |
79072805 |
697 | sv_catsv(sv,*mark); |
698 | } |
699 | } |
700 | else { |
701 | for (; items > 0; items--,mark++) |
702 | sv_catsv(sv,*mark); |
703 | } |
704 | SvSETMAGIC(sv); |
705 | } |
706 | |
707 | void |
864dbfa3 |
708 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
709 | { |
97aff369 |
710 | dVAR; |
46fc3d4c |
711 | STRLEN patlen; |
53c1dcc0 |
712 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c |
713 | bool do_taint = FALSE; |
714 | |
5b781b5b |
715 | SvUTF8_off(sv); |
2cf2cfc6 |
716 | if (DO_UTF8(*sarg)) |
717 | SvUTF8_on(sv); |
4608196e |
718 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint); |
79072805 |
719 | SvSETMAGIC(sv); |
46fc3d4c |
720 | if (do_taint) |
721 | SvTAINTED_on(sv); |
79072805 |
722 | } |
723 | |
33b45480 |
724 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
725 | UV |
726 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
727 | { |
97aff369 |
728 | dVAR; |
bb7a0f54 |
729 | STRLEN srclen, len, uoffset; |
10516c54 |
730 | const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); |
81e118e0 |
731 | UV retnum = 0; |
732 | |
a50d7633 |
733 | if (offset < 0) |
81e118e0 |
734 | return retnum; |
8e84507e |
735 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
736 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
737 | |
dcad2880 |
738 | if (SvUTF8(sv)) |
33b45480 |
739 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
740 | |
bb7a0f54 |
741 | uoffset = offset*size; /* turn into bit offset */ |
742 | len = (uoffset + size + 7) / 8; /* required number of bytes */ |
81e118e0 |
743 | if (len > srclen) { |
744 | if (size <= 8) |
745 | retnum = 0; |
746 | else { |
bb7a0f54 |
747 | uoffset >>= 3; /* turn into byte offset */ |
81e118e0 |
748 | if (size == 16) { |
bb7a0f54 |
749 | if (uoffset >= srclen) |
81e118e0 |
750 | retnum = 0; |
751 | else |
bb7a0f54 |
752 | retnum = (UV) s[uoffset] << 8; |
81e118e0 |
753 | } |
754 | else if (size == 32) { |
bb7a0f54 |
755 | if (uoffset >= srclen) |
81e118e0 |
756 | retnum = 0; |
bb7a0f54 |
757 | else if (uoffset + 1 >= srclen) |
81e118e0 |
758 | retnum = |
bb7a0f54 |
759 | ((UV) s[uoffset ] << 24); |
760 | else if (uoffset + 2 >= srclen) |
81e118e0 |
761 | retnum = |
bb7a0f54 |
762 | ((UV) s[uoffset ] << 24) + |
763 | ((UV) s[uoffset + 1] << 16); |
81e118e0 |
764 | else |
765 | retnum = |
bb7a0f54 |
766 | ((UV) s[uoffset ] << 24) + |
767 | ((UV) s[uoffset + 1] << 16) + |
768 | ( s[uoffset + 2] << 8); |
81e118e0 |
769 | } |
d7d93a81 |
770 | #ifdef UV_IS_QUAD |
c5a0f51a |
771 | else if (size == 64) { |
c5a0f51a |
772 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
773 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
774 | "Bit vector size > 32 non-portable"); |
bb7a0f54 |
775 | if (uoffset >= srclen) |
c5a0f51a |
776 | retnum = 0; |
bb7a0f54 |
777 | else if (uoffset + 1 >= srclen) |
c5a0f51a |
778 | retnum = |
bb7a0f54 |
779 | (UV) s[uoffset ] << 56; |
780 | else if (uoffset + 2 >= srclen) |
c5a0f51a |
781 | retnum = |
bb7a0f54 |
782 | ((UV) s[uoffset ] << 56) + |
783 | ((UV) s[uoffset + 1] << 48); |
784 | else if (uoffset + 3 >= srclen) |
c5a0f51a |
785 | retnum = |
bb7a0f54 |
786 | ((UV) s[uoffset ] << 56) + |
787 | ((UV) s[uoffset + 1] << 48) + |
788 | ((UV) s[uoffset + 2] << 40); |
789 | else if (uoffset + 4 >= srclen) |
c5a0f51a |
790 | retnum = |
bb7a0f54 |
791 | ((UV) s[uoffset ] << 56) + |
792 | ((UV) s[uoffset + 1] << 48) + |
793 | ((UV) s[uoffset + 2] << 40) + |
794 | ((UV) s[uoffset + 3] << 32); |
795 | else if (uoffset + 5 >= srclen) |
c5a0f51a |
796 | retnum = |
bb7a0f54 |
797 | ((UV) s[uoffset ] << 56) + |
798 | ((UV) s[uoffset + 1] << 48) + |
799 | ((UV) s[uoffset + 2] << 40) + |
800 | ((UV) s[uoffset + 3] << 32) + |
801 | ( s[uoffset + 4] << 24); |
802 | else if (uoffset + 6 >= srclen) |
c5a0f51a |
803 | retnum = |
bb7a0f54 |
804 | ((UV) s[uoffset ] << 56) + |
805 | ((UV) s[uoffset + 1] << 48) + |
806 | ((UV) s[uoffset + 2] << 40) + |
807 | ((UV) s[uoffset + 3] << 32) + |
808 | ((UV) s[uoffset + 4] << 24) + |
809 | ((UV) s[uoffset + 5] << 16); |
c5a0f51a |
810 | else |
8e84507e |
811 | retnum = |
bb7a0f54 |
812 | ((UV) s[uoffset ] << 56) + |
813 | ((UV) s[uoffset + 1] << 48) + |
814 | ((UV) s[uoffset + 2] << 40) + |
815 | ((UV) s[uoffset + 3] << 32) + |
816 | ((UV) s[uoffset + 4] << 24) + |
817 | ((UV) s[uoffset + 5] << 16) + |
818 | ( s[uoffset + 6] << 8); |
c5a0f51a |
819 | } |
820 | #endif |
81e118e0 |
821 | } |
822 | } |
823 | else if (size < 8) |
bb7a0f54 |
824 | retnum = (s[uoffset >> 3] >> (uoffset & 7)) & ((1 << size) - 1); |
81e118e0 |
825 | else { |
bb7a0f54 |
826 | uoffset >>= 3; /* turn into byte offset */ |
81e118e0 |
827 | if (size == 8) |
bb7a0f54 |
828 | retnum = s[uoffset]; |
81e118e0 |
829 | else if (size == 16) |
830 | retnum = |
bb7a0f54 |
831 | ((UV) s[uoffset] << 8) + |
832 | s[uoffset + 1]; |
81e118e0 |
833 | else if (size == 32) |
834 | retnum = |
bb7a0f54 |
835 | ((UV) s[uoffset ] << 24) + |
836 | ((UV) s[uoffset + 1] << 16) + |
837 | ( s[uoffset + 2] << 8) + |
838 | s[uoffset + 3]; |
d7d93a81 |
839 | #ifdef UV_IS_QUAD |
c5a0f51a |
840 | else if (size == 64) { |
c5a0f51a |
841 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
842 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
843 | "Bit vector size > 32 non-portable"); |
844 | retnum = |
bb7a0f54 |
845 | ((UV) s[uoffset ] << 56) + |
846 | ((UV) s[uoffset + 1] << 48) + |
847 | ((UV) s[uoffset + 2] << 40) + |
848 | ((UV) s[uoffset + 3] << 32) + |
849 | ((UV) s[uoffset + 4] << 24) + |
850 | ((UV) s[uoffset + 5] << 16) + |
851 | ( s[uoffset + 6] << 8) + |
852 | s[uoffset + 7]; |
c5a0f51a |
853 | } |
854 | #endif |
81e118e0 |
855 | } |
856 | |
857 | return retnum; |
858 | } |
859 | |
33b45480 |
860 | /* currently converts input to bytes if possible but doesn't sweat failures, |
861 | * although it does ensure that the string it clobbers is not marked as |
862 | * utf8-valid any more |
863 | */ |
79072805 |
864 | void |
864dbfa3 |
865 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
866 | { |
97aff369 |
867 | dVAR; |
79072805 |
868 | register I32 offset; |
869 | register I32 size; |
8990e307 |
870 | register unsigned char *s; |
81e118e0 |
871 | register UV lval; |
79072805 |
872 | I32 mask; |
a0d0e21e |
873 | STRLEN targlen; |
874 | STRLEN len; |
c4420975 |
875 | SV * const targ = LvTARG(sv); |
79072805 |
876 | |
8990e307 |
877 | if (!targ) |
878 | return; |
a0d0e21e |
879 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
880 | if (SvUTF8(targ)) { |
33b45480 |
881 | /* This is handled by the SvPOK_only below... |
882 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
883 | SvUTF8_off(targ); |
884 | */ |
885 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
886 | } |
887 | |
4ebbc975 |
888 | (void)SvPOK_only(targ); |
81e118e0 |
889 | lval = SvUV(sv); |
79072805 |
890 | offset = LvTARGOFF(sv); |
fe58ced6 |
891 | if (offset < 0) |
ed9aa3b7 |
892 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
893 | size = LvTARGLEN(sv); |
8e84507e |
894 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
895 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
896 | |
81e118e0 |
897 | offset *= size; /* turn into bit offset */ |
898 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
899 | if (len > targlen) { |
900 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
901 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
902 | SvCUR_set(targ, len); |
903 | } |
8e84507e |
904 | |
79072805 |
905 | if (size < 8) { |
906 | mask = (1 << size) - 1; |
907 | size = offset & 7; |
908 | lval &= mask; |
81e118e0 |
909 | offset >>= 3; /* turn into byte offset */ |
79072805 |
910 | s[offset] &= ~(mask << size); |
911 | s[offset] |= lval << size; |
912 | } |
913 | else { |
81e118e0 |
914 | offset >>= 3; /* turn into byte offset */ |
79072805 |
915 | if (size == 8) |
eb160463 |
916 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
917 | else if (size == 16) { |
eb160463 |
918 | s[offset ] = (U8)((lval >> 8) & 0xff); |
919 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
920 | } |
921 | else if (size == 32) { |
eb160463 |
922 | s[offset ] = (U8)((lval >> 24) & 0xff); |
923 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
924 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
925 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
926 | } |
d7d93a81 |
927 | #ifdef UV_IS_QUAD |
c5a0f51a |
928 | else if (size == 64) { |
c5a0f51a |
929 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
930 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
931 | "Bit vector size > 32 non-portable"); |
eb160463 |
932 | s[offset ] = (U8)((lval >> 56) & 0xff); |
933 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
934 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
935 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
936 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
937 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
938 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
939 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
940 | } |
dc1e3f56 |
941 | #endif |
79072805 |
942 | } |
7bb043c3 |
943 | SvSETMAGIC(targ); |
79072805 |
944 | } |
945 | |
946 | void |
864dbfa3 |
947 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
948 | { |
97aff369 |
949 | dVAR; |
463ee0b2 |
950 | STRLEN len; |
a0d0e21e |
951 | char *s; |
8e84507e |
952 | |
79072805 |
953 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
954 | register I32 i; |
c4420975 |
955 | AV* const av = (AV*)sv; |
f54cb97a |
956 | const I32 max = AvFILL(av); |
957 | |
958 | for (i = 0; i <= max; i++) { |
a0d0e21e |
959 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
960 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
961 | do_chop(astr, sv); |
962 | } |
963 | return; |
79072805 |
964 | } |
aa854799 |
965 | else if (SvTYPE(sv) == SVt_PVHV) { |
c4420975 |
966 | HV* const hv = (HV*)sv; |
a0d0e21e |
967 | HE* entry; |
968 | (void)hv_iterinit(hv); |
155aba94 |
969 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
970 | do_chop(astr,hv_iterval(hv,entry)); |
971 | return; |
79072805 |
972 | } |
991e6d41 |
973 | else if (SvREADONLY(sv)) { |
974 | if (SvFAKE(sv)) { |
975 | /* SV is copy-on-write */ |
976 | sv_force_normal_flags(sv, 0); |
977 | } |
978 | if (SvREADONLY(sv)) |
979 | Perl_croak(aTHX_ PL_no_modify); |
980 | } |
565a3db3 |
981 | |
982 | if (PL_encoding && !SvUTF8(sv)) { |
983 | /* like in do_chomp(), utf8-ize the sv as a side-effect |
984 | * if we're using encoding. */ |
985 | sv_recode_to_utf8(sv, PL_encoding); |
986 | } |
987 | |
a0d0e21e |
988 | s = SvPV(sv, len); |
748a9306 |
989 | if (len && !SvPOK(sv)) |
a0d0e21e |
990 | s = SvPV_force(sv, len); |
7e2040f0 |
991 | if (DO_UTF8(sv)) { |
a0ed51b3 |
992 | if (s && len) { |
c4420975 |
993 | char * const send = s + len; |
0bcc34c2 |
994 | char * const start = s; |
a0ed51b3 |
995 | s = send - 1; |
a0dbb045 |
996 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
997 | s--; |
ace7757b |
998 | if (is_utf8_string((U8*)s, send - s)) { |
a0dbb045 |
999 | sv_setpvn(astr, s, send - s); |
1000 | *s = '\0'; |
1001 | SvCUR_set(sv, s - start); |
1002 | SvNIOK_off(sv); |
1003 | SvUTF8_on(astr); |
1004 | } |
a0ed51b3 |
1005 | } |
1006 | else |
1007 | sv_setpvn(astr, "", 0); |
1008 | } |
7e2040f0 |
1009 | else if (s && len) { |
a0d0e21e |
1010 | s += --len; |
1011 | sv_setpvn(astr, s, 1); |
1012 | *s = '\0'; |
1013 | SvCUR_set(sv, len); |
2c19a612 |
1014 | SvUTF8_off(sv); |
a0d0e21e |
1015 | SvNIOK_off(sv); |
79072805 |
1016 | } |
1017 | else |
a0d0e21e |
1018 | sv_setpvn(astr, "", 0); |
1019 | SvSETMAGIC(sv); |
7e2040f0 |
1020 | } |
a0d0e21e |
1021 | |
1022 | I32 |
864dbfa3 |
1023 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1024 | { |
97aff369 |
1025 | dVAR; |
c07a80fd |
1026 | register I32 count; |
a0d0e21e |
1027 | STRLEN len; |
1028 | char *s; |
c4c87a06 |
1029 | char *temp_buffer = NULL; |
a0714e2c |
1030 | SV* svrecode = NULL; |
c07a80fd |
1031 | |
3280af22 |
1032 | if (RsSNARF(PL_rs)) |
c07a80fd |
1033 | return 0; |
4c5a6083 |
1034 | if (RsRECORD(PL_rs)) |
1035 | return 0; |
c07a80fd |
1036 | count = 0; |
a0d0e21e |
1037 | if (SvTYPE(sv) == SVt_PVAV) { |
1038 | register I32 i; |
890ce7af |
1039 | AV* const av = (AV*)sv; |
f54cb97a |
1040 | const I32 max = AvFILL(av); |
1041 | |
1042 | for (i = 0; i <= max; i++) { |
a0d0e21e |
1043 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1044 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1045 | count += do_chomp(sv); |
1046 | } |
1047 | return count; |
1048 | } |
aa854799 |
1049 | else if (SvTYPE(sv) == SVt_PVHV) { |
890ce7af |
1050 | HV* const hv = (HV*)sv; |
a0d0e21e |
1051 | HE* entry; |
1052 | (void)hv_iterinit(hv); |
155aba94 |
1053 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1054 | count += do_chomp(hv_iterval(hv,entry)); |
1055 | return count; |
1056 | } |
991e6d41 |
1057 | else if (SvREADONLY(sv)) { |
1058 | if (SvFAKE(sv)) { |
1059 | /* SV is copy-on-write */ |
1060 | sv_force_normal_flags(sv, 0); |
1061 | } |
1062 | if (SvREADONLY(sv)) |
1063 | Perl_croak(aTHX_ PL_no_modify); |
1064 | } |
a6aa349d |
1065 | |
1066 | if (PL_encoding) { |
1067 | if (!SvUTF8(sv)) { |
1068 | /* XXX, here sv is utf8-ized as a side-effect! |
1069 | If encoding.pm is used properly, almost string-generating |
1070 | operations, including literal strings, chr(), input data, etc. |
1071 | should have been utf8-ized already, right? |
1072 | */ |
1073 | sv_recode_to_utf8(sv, PL_encoding); |
1074 | } |
1075 | } |
1076 | |
a0d0e21e |
1077 | s = SvPV(sv, len); |
a0d0e21e |
1078 | if (s && len) { |
1079 | s += --len; |
3280af22 |
1080 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1081 | if (*s != '\n') |
1082 | goto nope; |
1083 | ++count; |
1084 | while (len && s[-1] == '\n') { |
1085 | --len; |
1086 | --s; |
1087 | ++count; |
1088 | } |
1089 | } |
a0d0e21e |
1090 | else { |
a6aa349d |
1091 | STRLEN rslen, rs_charlen; |
93524f2b |
1092 | const char *rsptr = SvPV_const(PL_rs, rslen); |
a6aa349d |
1093 | |
1094 | rs_charlen = SvUTF8(PL_rs) |
1095 | ? sv_len_utf8(PL_rs) |
1096 | : rslen; |
1097 | |
c4c87a06 |
1098 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1099 | /* Assumption is that rs is shorter than the scalar. */ |
1100 | if (SvUTF8(PL_rs)) { |
1101 | /* RS is utf8, scalar is 8 bit. */ |
1102 | bool is_utf8 = TRUE; |
1103 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, |
1104 | &rslen, &is_utf8); |
1105 | if (is_utf8) { |
1106 | /* Cannot downgrade, therefore cannot possibly match |
1107 | */ |
1108 | assert (temp_buffer == rsptr); |
1109 | temp_buffer = NULL; |
1110 | goto nope; |
1111 | } |
1112 | rsptr = temp_buffer; |
a6aa349d |
1113 | } |
1114 | else if (PL_encoding) { |
1115 | /* RS is 8 bit, encoding.pm is used. |
1116 | * Do not recode PL_rs as a side-effect. */ |
1117 | svrecode = newSVpvn(rsptr, rslen); |
1118 | sv_recode_to_utf8(svrecode, PL_encoding); |
93524f2b |
1119 | rsptr = SvPV_const(svrecode, rslen); |
a6aa349d |
1120 | rs_charlen = sv_len_utf8(svrecode); |
1121 | } |
1122 | else { |
c4c87a06 |
1123 | /* RS is 8 bit, scalar is utf8. */ |
1124 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); |
1125 | rsptr = temp_buffer; |
1126 | } |
1127 | } |
c07a80fd |
1128 | if (rslen == 1) { |
1129 | if (*s != *rsptr) |
1130 | goto nope; |
1131 | ++count; |
1132 | } |
1133 | else { |
8c2cee6f |
1134 | if (len < rslen - 1) |
c07a80fd |
1135 | goto nope; |
1136 | len -= rslen - 1; |
1137 | s -= rslen - 1; |
36477c24 |
1138 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1139 | goto nope; |
a6aa349d |
1140 | count += rs_charlen; |
c07a80fd |
1141 | } |
a0d0e21e |
1142 | } |
8b6b16e7 |
1143 | s = SvPV_force_nolen(sv); |
a0d0e21e |
1144 | SvCUR_set(sv, len); |
9b33ce3b |
1145 | *SvEND(sv) = '\0'; |
a0d0e21e |
1146 | SvNIOK_off(sv); |
9b33ce3b |
1147 | SvSETMAGIC(sv); |
a0d0e21e |
1148 | } |
1149 | nope: |
a6aa349d |
1150 | |
1151 | if (svrecode) |
1152 | SvREFCNT_dec(svrecode); |
1153 | |
c4c87a06 |
1154 | Safefree(temp_buffer); |
a0d0e21e |
1155 | return count; |
8e84507e |
1156 | } |
79072805 |
1157 | |
1158 | void |
864dbfa3 |
1159 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1160 | { |
97aff369 |
1161 | dVAR; |
79072805 |
1162 | #ifdef LIBERAL |
1163 | register long *dl; |
1164 | register long *ll; |
1165 | register long *rl; |
1166 | #endif |
1167 | register char *dc; |
463ee0b2 |
1168 | STRLEN leftlen; |
1169 | STRLEN rightlen; |
e62f0680 |
1170 | register const char *lc; |
1171 | register const char *rc; |
bb7a0f54 |
1172 | register STRLEN len; |
1173 | STRLEN lensave; |
e62f0680 |
1174 | const char *lsave; |
1175 | const char *rsave; |
12abf4f0 |
1176 | bool left_utf; |
1177 | bool right_utf; |
bb7a0f54 |
1178 | STRLEN needlen = 0; |
0c57e439 |
1179 | |
79072805 |
1180 | |
1fbd88dc |
1181 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1182 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
e62f0680 |
1183 | lsave = lc = SvPV_nomg_const(left, leftlen); |
1184 | rsave = rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 |
1185 | |
1186 | /* This need to come after SvPV to ensure that string overloading has |
1187 | fired off. */ |
1188 | |
1189 | left_utf = DO_UTF8(left); |
1190 | right_utf = DO_UTF8(right); |
1191 | |
1192 | if (left_utf && !right_utf) { |
1193 | /* Avoid triggering overloading again by using temporaries. |
1194 | Maybe there should be a variant of sv_utf8_upgrade that takes pvn |
1195 | */ |
1196 | right = sv_2mortal(newSVpvn(rsave, rightlen)); |
1197 | sv_utf8_upgrade(right); |
1198 | rsave = rc = SvPV_nomg_const(right, rightlen); |
1199 | right_utf = TRUE; |
1200 | } |
1201 | else if (!left_utf && right_utf) { |
1202 | left = sv_2mortal(newSVpvn(lsave, leftlen)); |
1203 | sv_utf8_upgrade(left); |
1204 | lsave = lc = SvPV_nomg_const(left, leftlen); |
1205 | left_utf = TRUE; |
1206 | } |
1207 | |
93a17b20 |
1208 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1209 | lensave = len; |
9fdd7463 |
1210 | SvCUR_set(sv, len); |
1211 | (void)SvPOK_only(sv); |
c9b3c8d0 |
1212 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1213 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
a02a5408 |
1214 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 |
1215 | } |
1216 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe |
1217 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 |
1218 | if (SvLEN(sv) < len + 1) { |
1219 | dc = SvGROW(sv, len + 1); |
ff68c719 |
1220 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1221 | } |
c9b3c8d0 |
1222 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1223 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1224 | } |
1225 | else { |
bb7a0f54 |
1226 | needlen = optype == OP_BIT_AND |
1227 | ? len : (leftlen > rightlen ? leftlen : rightlen); |
a02a5408 |
1228 | Newxz(dc, needlen + 1, char); |
aa0a69cb |
1229 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 |
1230 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1231 | } |
0c57e439 |
1232 | if (left_utf || right_utf) { |
1233 | UV duc, luc, ruc; |
794a0d33 |
1234 | char *dcorig = dc; |
1235 | char *dcsave = NULL; |
0c57e439 |
1236 | STRLEN lulen = leftlen; |
1237 | STRLEN rulen = rightlen; |
ba210ebe |
1238 | STRLEN ulen; |
0c57e439 |
1239 | |
1240 | switch (optype) { |
1241 | case OP_BIT_AND: |
1242 | while (lulen && rulen) { |
9041c2e3 |
1243 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1244 | lc += ulen; |
1245 | lulen -= ulen; |
9041c2e3 |
1246 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1247 | rc += ulen; |
1248 | rulen -= ulen; |
1249 | duc = luc & ruc; |
9041c2e3 |
1250 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1251 | } |
3ce4c532 |
1252 | if (sv == left || sv == right) |
1253 | (void)sv_usepvn(sv, dcorig, needlen); |
794a0d33 |
1254 | SvCUR_set(sv, dc - dcorig); |
0c57e439 |
1255 | break; |
1256 | case OP_BIT_XOR: |
1257 | while (lulen && rulen) { |
9041c2e3 |
1258 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1259 | lc += ulen; |
1260 | lulen -= ulen; |
9041c2e3 |
1261 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1262 | rc += ulen; |
1263 | rulen -= ulen; |
1264 | duc = luc ^ ruc; |
9041c2e3 |
1265 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1266 | } |
1267 | goto mop_up_utf; |
1268 | case OP_BIT_OR: |
1269 | while (lulen && rulen) { |
9041c2e3 |
1270 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1271 | lc += ulen; |
1272 | lulen -= ulen; |
9041c2e3 |
1273 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1274 | rc += ulen; |
1275 | rulen -= ulen; |
1276 | duc = luc | ruc; |
9041c2e3 |
1277 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1278 | } |
1279 | mop_up_utf: |
794a0d33 |
1280 | if (rulen) |
1281 | dcsave = savepvn(rc, rulen); |
1282 | else if (lulen) |
1283 | dcsave = savepvn(lc, lulen); |
3ce4c532 |
1284 | if (sv == left || sv == right) |
794a0d33 |
1285 | (void)sv_usepvn(sv, dcorig, needlen); /* Uses Renew(). */ |
1286 | SvCUR_set(sv, dc - dcorig); |
0c57e439 |
1287 | if (rulen) |
794a0d33 |
1288 | sv_catpvn(sv, dcsave, rulen); |
0c57e439 |
1289 | else if (lulen) |
794a0d33 |
1290 | sv_catpvn(sv, dcsave, lulen); |
0c57e439 |
1291 | else |
1292 | *SvEND(sv) = '\0'; |
794a0d33 |
1293 | Safefree(dcsave); |
0c57e439 |
1294 | break; |
41547412 |
1295 | default: |
1296 | if (sv == left || sv == right) |
1297 | Safefree(dcorig); |
c395bd6c |
1298 | Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)", |
1299 | (unsigned)optype, PL_op_name[optype]); |
0c57e439 |
1300 | } |
1301 | SvUTF8_on(sv); |
1302 | goto finish; |
1303 | } |
1304 | else |
79072805 |
1305 | #ifdef LIBERAL |
1306 | if (len >= sizeof(long)*4 && |
bb7a0f54 |
1307 | !((unsigned long)dc % sizeof(long)) && |
1308 | !((unsigned long)lc % sizeof(long)) && |
1309 | !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ |
79072805 |
1310 | { |
bb7a0f54 |
1311 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 |
1312 | len /= (sizeof(long)*4); |
1313 | |
1314 | dl = (long*)dc; |
1315 | ll = (long*)lc; |
1316 | rl = (long*)rc; |
1317 | |
1318 | switch (optype) { |
1319 | case OP_BIT_AND: |
1320 | while (len--) { |
1321 | *dl++ = *ll++ & *rl++; |
1322 | *dl++ = *ll++ & *rl++; |
1323 | *dl++ = *ll++ & *rl++; |
1324 | *dl++ = *ll++ & *rl++; |
1325 | } |
1326 | break; |
a0d0e21e |
1327 | case OP_BIT_XOR: |
79072805 |
1328 | while (len--) { |
1329 | *dl++ = *ll++ ^ *rl++; |
1330 | *dl++ = *ll++ ^ *rl++; |
1331 | *dl++ = *ll++ ^ *rl++; |
1332 | *dl++ = *ll++ ^ *rl++; |
1333 | } |
1334 | break; |
1335 | case OP_BIT_OR: |
1336 | while (len--) { |
1337 | *dl++ = *ll++ | *rl++; |
1338 | *dl++ = *ll++ | *rl++; |
1339 | *dl++ = *ll++ | *rl++; |
1340 | *dl++ = *ll++ | *rl++; |
1341 | } |
1342 | } |
1343 | |
1344 | dc = (char*)dl; |
1345 | lc = (char*)ll; |
1346 | rc = (char*)rl; |
1347 | |
1348 | len = remainder; |
1349 | } |
1350 | #endif |
a0d0e21e |
1351 | { |
a0d0e21e |
1352 | switch (optype) { |
1353 | case OP_BIT_AND: |
1354 | while (len--) |
1355 | *dc++ = *lc++ & *rc++; |
1a787b95 |
1356 | *dc = '\0'; |
a0d0e21e |
1357 | break; |
1358 | case OP_BIT_XOR: |
1359 | while (len--) |
1360 | *dc++ = *lc++ ^ *rc++; |
1361 | goto mop_up; |
1362 | case OP_BIT_OR: |
1363 | while (len--) |
1364 | *dc++ = *lc++ | *rc++; |
1365 | mop_up: |
1366 | len = lensave; |
bb7a0f54 |
1367 | if (rightlen > len) |
a0d0e21e |
1368 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1369 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1370 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1371 | else |
1372 | *SvEND(sv) = '\0'; |
a0d0e21e |
1373 | break; |
1374 | } |
79072805 |
1375 | } |
0c57e439 |
1376 | finish: |
fb73857a |
1377 | SvTAINT(sv); |
79072805 |
1378 | } |
463ee0b2 |
1379 | |
1380 | OP * |
cea2e8a9 |
1381 | Perl_do_kv(pTHX) |
463ee0b2 |
1382 | { |
97aff369 |
1383 | dVAR; |
39644a26 |
1384 | dSP; |
c4420975 |
1385 | HV * const hv = (HV*)POPs; |
800e9ae0 |
1386 | HV *keys; |
463ee0b2 |
1387 | register HE *entry; |
f54cb97a |
1388 | const I32 gimme = GIMME_V; |
1389 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); |
1390 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); |
1391 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); |
463ee0b2 |
1392 | |
85581909 |
1393 | if (!hv) { |
78f9721b |
1394 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1395 | dTARGET; /* make sure to clear its target here */ |
1396 | if (SvTYPE(TARG) == SVt_PVLV) |
a0714e2c |
1397 | LvTARG(TARG) = NULL; |
85581909 |
1398 | PUSHs(TARG); |
1399 | } |
463ee0b2 |
1400 | RETURN; |
85581909 |
1401 | } |
748a9306 |
1402 | |
6d822dc4 |
1403 | keys = hv; |
800e9ae0 |
1404 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1405 | |
54310121 |
1406 | if (gimme == G_VOID) |
aa689395 |
1407 | RETURN; |
1408 | |
54310121 |
1409 | if (gimme == G_SCALAR) { |
6ee623d5 |
1410 | IV i; |
463ee0b2 |
1411 | dTARGET; |
1412 | |
78f9721b |
1413 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1414 | if (SvTYPE(TARG) < SVt_PVLV) { |
1415 | sv_upgrade(TARG, SVt_PVLV); |
a0714e2c |
1416 | sv_magic(TARG, NULL, PERL_MAGIC_nkeys, NULL, 0); |
85581909 |
1417 | } |
1418 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1419 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1420 | if (LvTARG(TARG)) |
1421 | SvREFCNT_dec(LvTARG(TARG)); |
b37c2d43 |
1422 | LvTARG(TARG) = SvREFCNT_inc_simple(keys); |
6ff81951 |
1423 | } |
85581909 |
1424 | PUSHs(TARG); |
1425 | RETURN; |
1426 | } |
1427 | |
81714fb9 |
1428 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied) |
1429 | && ! SvTIED_mg((SV*)keys, PERL_MAGIC_regdata_names)) |
1430 | { |
a98face1 |
1431 | i = HvKEYS(keys); |
81714fb9 |
1432 | } |
463ee0b2 |
1433 | else { |
1434 | i = 0; |
800e9ae0 |
1435 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1436 | } |
1437 | PUSHi( i ); |
1438 | RETURN; |
1439 | } |
1440 | |
8ed4b672 |
1441 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1442 | |
463ee0b2 |
1443 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1444 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1445 | SPAGAIN; |
574c8022 |
1446 | if (dokeys) { |
890ce7af |
1447 | SV* const sv = hv_iterkeysv(entry); |
574c8022 |
1448 | XPUSHs(sv); /* won't clobber stack_sp */ |
1449 | } |
463ee0b2 |
1450 | if (dovalues) { |
f54cb97a |
1451 | SV *tmpstr; |
463ee0b2 |
1452 | PUTBACK; |
6d822dc4 |
1453 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1454 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1455 | (unsigned long)HeHASH(entry), |
cdd3ba14 |
1456 | (int)HvMAX(keys)+1, |
800e9ae0 |
1457 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1458 | SPAGAIN; |
46fc3d4c |
1459 | XPUSHs(tmpstr); |
463ee0b2 |
1460 | } |
1461 | PUTBACK; |
1462 | } |
1463 | return NORMAL; |
1464 | } |
4e35701f |
1465 | |
af3babe4 |
1466 | /* |
1467 | * Local variables: |
1468 | * c-indentation-style: bsd |
1469 | * c-basic-offset: 4 |
1470 | * indent-tabs-mode: t |
1471 | * End: |
1472 | * |
37442d52 |
1473 | * ex: set ts=8 sts=4 sw=4 noet: |
1474 | */ |