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 |
b250498f |
29 | S_do_trans_simple(pTHX_ SV *sv) |
79072805 |
30 | { |
97aff369 |
31 | dVAR; |
4757a243 |
32 | U8 *s; |
b250498f |
33 | U8 *d; |
f54cb97a |
34 | const U8 *send; |
b250498f |
35 | U8 *dstart; |
4757a243 |
36 | I32 matches = 0; |
f54cb97a |
37 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
463ee0b2 |
38 | STRLEN len; |
79072805 |
39 | |
c4420975 |
40 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
41 | if (!tbl) |
7d85a32c |
42 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 |
43 | |
4757a243 |
44 | s = (U8*)SvPV(sv, len); |
45 | send = s + len; |
46 | |
1e54db1a |
47 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b |
48 | if (!SvUTF8(sv)) { |
01ec43d0 |
49 | while (s < send) { |
f54cb97a |
50 | const I32 ch = tbl[*s]; |
51 | if (ch >= 0) { |
01ec43d0 |
52 | matches++; |
c4420975 |
53 | *s = (U8)ch; |
01ec43d0 |
54 | } |
c4420975 |
55 | s++; |
01ec43d0 |
56 | } |
57 | SvSETMAGIC(sv); |
b250498f |
58 | return matches; |
59 | } |
4757a243 |
60 | |
b250498f |
61 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ |
9b877dbb |
62 | if (grows) |
a02a5408 |
63 | Newx(d, len*2+1, U8); |
9b877dbb |
64 | else |
65 | d = s; |
b250498f |
66 | dstart = d; |
67 | while (s < send) { |
ba210ebe |
68 | STRLEN ulen; |
f54cb97a |
69 | I32 ch; |
b250498f |
70 | |
b250498f |
71 | /* Need to check this, otherwise 128..255 won't match */ |
9f7f3913 |
72 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
9b877dbb |
73 | if (c < 0x100 && (ch = tbl[c]) >= 0) { |
b250498f |
74 | matches++; |
9041c2e3 |
75 | d = uvchr_to_utf8(d, ch); |
b250498f |
76 | s += ulen; |
01ec43d0 |
77 | } |
78 | else { /* No match -> copy */ |
0376ff32 |
79 | Move(s, d, ulen, U8); |
9b877dbb |
80 | d += ulen; |
81 | s += ulen; |
b250498f |
82 | } |
83 | } |
9b877dbb |
84 | if (grows) { |
85 | sv_setpvn(sv, (char*)dstart, d - dstart); |
86 | Safefree(dstart); |
87 | } |
88 | else { |
89 | *d = '\0'; |
90 | SvCUR_set(sv, d - dstart); |
91 | } |
b250498f |
92 | SvUTF8_on(sv); |
b250498f |
93 | SvSETMAGIC(sv); |
4757a243 |
94 | return matches; |
95 | } |
96 | |
942e002e |
97 | STATIC I32 |
21304a3b |
98 | S_do_trans_count(pTHX_ SV *sv) |
4757a243 |
99 | { |
97aff369 |
100 | dVAR; |
83003860 |
101 | const U8 *s; |
102 | const U8 *send; |
4757a243 |
103 | I32 matches = 0; |
104 | STRLEN len; |
4757a243 |
105 | |
f54cb97a |
106 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
107 | if (!tbl) |
7d85a32c |
108 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 |
109 | |
83003860 |
110 | s = (const U8*)SvPV_const(sv, len); |
4757a243 |
111 | send = s + len; |
112 | |
1aa99e6b |
113 | if (!SvUTF8(sv)) |
114 | while (s < send) { |
115 | if (tbl[*s++] >= 0) |
036b4402 |
116 | matches++; |
1aa99e6b |
117 | } |
fabdb6c0 |
118 | else { |
119 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
1aa99e6b |
120 | while (s < send) { |
1aa99e6b |
121 | STRLEN ulen; |
9f7f3913 |
122 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 |
123 | if (c < 0x100) { |
124 | if (tbl[c] >= 0) |
125 | matches++; |
126 | } else if (complement) |
1aa99e6b |
127 | matches++; |
128 | s += ulen; |
129 | } |
fabdb6c0 |
130 | } |
4757a243 |
131 | |
132 | return matches; |
133 | } |
134 | |
942e002e |
135 | STATIC I32 |
21304a3b |
136 | S_do_trans_complex(pTHX_ SV *sv) |
4757a243 |
137 | { |
97aff369 |
138 | dVAR; |
4757a243 |
139 | U8 *s; |
140 | U8 *send; |
141 | U8 *d; |
5e44153e |
142 | U8 *dstart; |
1aa99e6b |
143 | I32 isutf8; |
4757a243 |
144 | I32 matches = 0; |
b7953727 |
145 | STRLEN len, rlen = 0; |
4757a243 |
146 | |
f54cb97a |
147 | const short * const tbl = (short*)cPVOP->op_pv; |
4757a243 |
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) { |
f54cb97a |
158 | const U8* p = send; |
1aa99e6b |
159 | while (s < send) { |
f54cb97a |
160 | const I32 ch = tbl[*s]; |
161 | if (ch >= 0) { |
eb160463 |
162 | *d = (U8)ch; |
036b4402 |
163 | matches++; |
1aa99e6b |
164 | if (p != d - 1 || *p != *d) |
165 | p = d++; |
166 | } |
167 | else if (ch == -1) /* -1 is unmapped character */ |
168 | *d++ = *s; |
169 | else if (ch == -2) /* -2 is delete character */ |
170 | matches++; |
171 | s++; |
172 | } |
a0ed51b3 |
173 | } |
1aa99e6b |
174 | else { |
175 | while (s < send) { |
f54cb97a |
176 | const I32 ch = tbl[*s]; |
177 | if (ch >= 0) { |
1aa99e6b |
178 | matches++; |
eb160463 |
179 | *d++ = (U8)ch; |
1aa99e6b |
180 | } |
181 | else if (ch == -1) /* -1 is unmapped character */ |
182 | *d++ = *s; |
183 | else if (ch == -2) /* -2 is delete character */ |
184 | matches++; |
185 | s++; |
186 | } |
187 | } |
76ef7183 |
188 | *d = '\0'; |
1aa99e6b |
189 | SvCUR_set(sv, d - dstart); |
4757a243 |
190 | } |
1aa99e6b |
191 | else { /* isutf8 */ |
fabdb6c0 |
192 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
193 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
194 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
195 | |
9b877dbb |
196 | if (grows) |
a02a5408 |
197 | Newx(d, len*2+1, U8); |
9b877dbb |
198 | else |
199 | d = s; |
1aa99e6b |
200 | dstart = d; |
8973db79 |
201 | if (complement && !del) |
202 | rlen = tbl[0x100]; |
1aa99e6b |
203 | |
4aecb5b5 |
204 | #ifdef MACOS_TRADITIONAL |
205 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ |
206 | #endif |
207 | |
1aa99e6b |
208 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b |
209 | UV pch = 0xfeedface; |
210 | while (s < send) { |
211 | STRLEN len; |
9f7f3913 |
212 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
213 | UTF8_ALLOW_DEFAULT); |
f54cb97a |
214 | I32 ch; |
1aa99e6b |
215 | |
8973db79 |
216 | if (comp > 0xff) { |
217 | if (!complement) { |
218 | Copy(s, d, len, U8); |
219 | d += len; |
220 | } |
221 | else { |
222 | matches++; |
223 | if (!del) { |
05d340b8 |
224 | ch = (rlen == 0) ? comp : |
225 | (comp - 0x100 < rlen) ? |
8973db79 |
226 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 |
227 | if ((UV)ch != pch) { |
9041c2e3 |
228 | d = uvchr_to_utf8(d, ch); |
eb160463 |
229 | pch = (UV)ch; |
8973db79 |
230 | } |
231 | s += len; |
232 | continue; |
233 | } |
234 | } |
9b877dbb |
235 | } |
1aa99e6b |
236 | else if ((ch = tbl[comp]) >= 0) { |
237 | matches++; |
eb160463 |
238 | if ((UV)ch != pch) { |
9041c2e3 |
239 | d = uvchr_to_utf8(d, ch); |
eb160463 |
240 | pch = (UV)ch; |
1aa99e6b |
241 | } |
242 | s += len; |
243 | continue; |
244 | } |
9b877dbb |
245 | else if (ch == -1) { /* -1 is unmapped character */ |
246 | Copy(s, d, len, U8); |
247 | d += len; |
248 | } |
1aa99e6b |
249 | else if (ch == -2) /* -2 is delete character */ |
250 | matches++; |
251 | s += len; |
252 | pch = 0xfeedface; |
253 | } |
5d06d08e |
254 | } |
1aa99e6b |
255 | else { |
256 | while (s < send) { |
257 | STRLEN len; |
9f7f3913 |
258 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
259 | UTF8_ALLOW_DEFAULT); |
f54cb97a |
260 | I32 ch; |
8973db79 |
261 | if (comp > 0xff) { |
262 | if (!complement) { |
0376ff32 |
263 | Move(s, d, len, U8); |
8973db79 |
264 | d += len; |
265 | } |
266 | else { |
267 | matches++; |
268 | if (!del) { |
9041c2e3 |
269 | if (comp - 0x100 < rlen) |
270 | d = uvchr_to_utf8(d, tbl[comp+1]); |
8973db79 |
271 | else |
9041c2e3 |
272 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 |
273 | } |
274 | } |
9b877dbb |
275 | } |
1aa99e6b |
276 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 |
277 | d = uvchr_to_utf8(d, ch); |
1aa99e6b |
278 | matches++; |
279 | } |
280 | else if (ch == -1) { /* -1 is unmapped character */ |
9b877dbb |
281 | Copy(s, d, len, U8); |
282 | d += len; |
1aa99e6b |
283 | } |
284 | else if (ch == -2) /* -2 is delete character */ |
285 | matches++; |
286 | s += len; |
287 | } |
288 | } |
9b877dbb |
289 | if (grows) { |
290 | sv_setpvn(sv, (char*)dstart, d - dstart); |
291 | Safefree(dstart); |
292 | } |
293 | else { |
294 | *d = '\0'; |
295 | SvCUR_set(sv, d - dstart); |
296 | } |
1aa99e6b |
297 | SvUTF8_on(sv); |
4757a243 |
298 | } |
5e44153e |
299 | SvSETMAGIC(sv); |
4757a243 |
300 | return matches; |
301 | } |
302 | |
942e002e |
303 | STATIC I32 |
21304a3b |
304 | S_do_trans_simple_utf8(pTHX_ SV *sv) |
4757a243 |
305 | { |
97aff369 |
306 | dVAR; |
4757a243 |
307 | U8 *s; |
308 | U8 *send; |
309 | U8 *d; |
036b4402 |
310 | U8 *start; |
1aa99e6b |
311 | U8 *dstart, *dend; |
4757a243 |
312 | I32 matches = 0; |
f54cb97a |
313 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 |
314 | STRLEN len; |
315 | |
53c1dcc0 |
316 | SV* const rv = (SV*)cSVOP->op_sv; |
317 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
318 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
319 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
320 | const UV extra = none + 1; |
b7953727 |
321 | UV final = 0; |
1aa99e6b |
322 | I32 isutf8; |
323 | U8 hibit = 0; |
4757a243 |
324 | |
325 | s = (U8*)SvPV(sv, len); |
1aa99e6b |
326 | isutf8 = SvUTF8(sv); |
327 | if (!isutf8) { |
c4420975 |
328 | const U8 *t = s; |
329 | const U8 * const e = s + len; |
c4d5f83a |
330 | while (t < e) { |
f54cb97a |
331 | const U8 ch = *t++; |
fabdb6c0 |
332 | hibit = !NATIVE_IS_INVARIANT(ch); |
333 | if (hibit) { |
334 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
335 | break; |
fabdb6c0 |
336 | } |
c4d5f83a |
337 | } |
1aa99e6b |
338 | } |
4757a243 |
339 | send = s + len; |
036b4402 |
340 | start = s; |
4757a243 |
341 | |
a4fc7abc |
342 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 |
343 | if (svp) |
344 | final = SvUV(*svp); |
345 | |
9b877dbb |
346 | if (grows) { |
347 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
348 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
349 | dend = d + len * 3; |
350 | dstart = d; |
351 | } |
352 | else { |
353 | dstart = d = s; |
354 | dend = d + len; |
355 | } |
1aa99e6b |
356 | |
4757a243 |
357 | while (s < send) { |
0bcc34c2 |
358 | const UV uv = swash_fetch(rv, s, TRUE); |
359 | if (uv < none) { |
4757a243 |
360 | s += UTF8SKIP(s); |
361 | matches++; |
7d85a32c |
362 | d = uvuni_to_utf8(d, uv); |
4757a243 |
363 | } |
364 | else if (uv == none) { |
4373e329 |
365 | const int i = UTF8SKIP(s); |
0376ff32 |
366 | Move(s, d, i, U8); |
9b877dbb |
367 | d += i; |
368 | s += i; |
4757a243 |
369 | } |
370 | else if (uv == extra) { |
4373e329 |
371 | s += UTF8SKIP(s); |
4757a243 |
372 | matches++; |
7d85a32c |
373 | d = uvuni_to_utf8(d, final); |
4757a243 |
374 | } |
375 | else |
376 | s += UTF8SKIP(s); |
1aa99e6b |
377 | |
9b877dbb |
378 | if (d > dend) { |
4373e329 |
379 | const STRLEN clen = d - dstart; |
380 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
381 | if (!grows) |
7d85a32c |
382 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 |
383 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
384 | d = dstart + clen; |
385 | dend = dstart + nlen; |
386 | } |
4757a243 |
387 | } |
16ec844d |
388 | if (grows || hibit) { |
9b877dbb |
389 | sv_setpvn(sv, (char*)dstart, d - dstart); |
390 | Safefree(dstart); |
16ec844d |
391 | if (grows && hibit) |
392 | Safefree(start); |
9b877dbb |
393 | } |
394 | else { |
395 | *d = '\0'; |
396 | SvCUR_set(sv, d - dstart); |
397 | } |
4757a243 |
398 | SvSETMAGIC(sv); |
1aa99e6b |
399 | SvUTF8_on(sv); |
4757a243 |
400 | |
401 | return matches; |
402 | } |
403 | |
942e002e |
404 | STATIC I32 |
21304a3b |
405 | S_do_trans_count_utf8(pTHX_ SV *sv) |
4757a243 |
406 | { |
97aff369 |
407 | dVAR; |
93524f2b |
408 | const U8 *s; |
cbbf8932 |
409 | const U8 *start = NULL; |
410 | const U8 *send; |
4757a243 |
411 | I32 matches = 0; |
412 | STRLEN len; |
413 | |
53c1dcc0 |
414 | SV* const rv = (SV*)cSVOP->op_sv; |
415 | HV* const hv = (HV*)SvRV(rv); |
a4fc7abc |
416 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
417 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
418 | const UV extra = none + 1; |
1aa99e6b |
419 | U8 hibit = 0; |
4757a243 |
420 | |
93524f2b |
421 | s = (const U8*)SvPV_const(sv, len); |
1aa99e6b |
422 | if (!SvUTF8(sv)) { |
4373e329 |
423 | const U8 *t = s; |
c4420975 |
424 | const U8 * const e = s + len; |
c4d5f83a |
425 | while (t < e) { |
4373e329 |
426 | const U8 ch = *t++; |
fabdb6c0 |
427 | hibit = !NATIVE_IS_INVARIANT(ch); |
428 | if (hibit) { |
429 | start = s = bytes_to_utf8(s, &len); |
1aa99e6b |
430 | break; |
fabdb6c0 |
431 | } |
c4d5f83a |
432 | } |
1aa99e6b |
433 | } |
4757a243 |
434 | send = s + len; |
435 | |
436 | while (s < send) { |
c4420975 |
437 | const UV uv = swash_fetch(rv, s, TRUE); |
438 | if (uv < none || uv == extra) |
4757a243 |
439 | matches++; |
834a4ddd |
440 | s += UTF8SKIP(s); |
4757a243 |
441 | } |
1aa99e6b |
442 | if (hibit) |
443 | Safefree(start); |
4757a243 |
444 | |
445 | return matches; |
446 | } |
447 | |
942e002e |
448 | STATIC I32 |
21304a3b |
449 | S_do_trans_complex_utf8(pTHX_ SV *sv) |
4757a243 |
450 | { |
97aff369 |
451 | dVAR; |
1aa99e6b |
452 | U8 *start, *send; |
4757a243 |
453 | U8 *d; |
454 | I32 matches = 0; |
4373e329 |
455 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
456 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; |
457 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
53c1dcc0 |
458 | SV * const rv = (SV*)cSVOP->op_sv; |
459 | HV * const hv = (HV*)SvRV(rv); |
a4fc7abc |
460 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a |
461 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
462 | const UV extra = none + 1; |
b7953727 |
463 | UV final = 0; |
45005bfb |
464 | bool havefinal = FALSE; |
4757a243 |
465 | STRLEN len; |
1aa99e6b |
466 | U8 *dstart, *dend; |
1aa99e6b |
467 | U8 hibit = 0; |
4757a243 |
468 | |
f54cb97a |
469 | U8 *s = (U8*)SvPV(sv, len); |
470 | const I32 isutf8 = SvUTF8(sv); |
1aa99e6b |
471 | if (!isutf8) { |
53c1dcc0 |
472 | const U8 *t = s; |
473 | const U8 * const e = s + len; |
c4d5f83a |
474 | while (t < e) { |
f54cb97a |
475 | const U8 ch = *t++; |
fabdb6c0 |
476 | hibit = !NATIVE_IS_INVARIANT(ch); |
477 | if (hibit) { |
478 | s = bytes_to_utf8(s, &len); |
1aa99e6b |
479 | break; |
fabdb6c0 |
480 | } |
c4d5f83a |
481 | } |
1aa99e6b |
482 | } |
4757a243 |
483 | send = s + len; |
1aa99e6b |
484 | start = s; |
4757a243 |
485 | |
a4fc7abc |
486 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb |
487 | if (svp) { |
4757a243 |
488 | final = SvUV(*svp); |
45005bfb |
489 | havefinal = TRUE; |
490 | } |
4757a243 |
491 | |
9b877dbb |
492 | if (grows) { |
493 | /* d needs to be bigger than s, in case e.g. upgrading is required */ |
a02a5408 |
494 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb |
495 | dend = d + len * 3; |
496 | dstart = d; |
497 | } |
498 | else { |
499 | dstart = d = s; |
500 | dend = d + len; |
501 | } |
4757a243 |
502 | |
503 | if (squash) { |
504 | UV puv = 0xfeedface; |
505 | while (s < send) { |
f54cb97a |
506 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 |
507 | |
9b877dbb |
508 | if (d > dend) { |
4373e329 |
509 | const STRLEN clen = d - dstart; |
510 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
511 | if (!grows) |
7d85a32c |
512 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
513 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
514 | d = dstart + clen; |
515 | dend = dstart + nlen; |
4757a243 |
516 | } |
517 | if (uv < none) { |
518 | matches++; |
94472101 |
519 | s += UTF8SKIP(s); |
4757a243 |
520 | if (uv != puv) { |
7d85a32c |
521 | d = uvuni_to_utf8(d, uv); |
4757a243 |
522 | puv = uv; |
523 | } |
4757a243 |
524 | continue; |
525 | } |
526 | else if (uv == none) { /* "none" is unmapped character */ |
4373e329 |
527 | const int i = UTF8SKIP(s); |
0376ff32 |
528 | Move(s, d, i, U8); |
9b877dbb |
529 | d += i; |
530 | s += i; |
4757a243 |
531 | puv = 0xfeedface; |
532 | continue; |
a0ed51b3 |
533 | } |
4757a243 |
534 | else if (uv == extra && !del) { |
535 | matches++; |
45005bfb |
536 | if (havefinal) { |
537 | s += UTF8SKIP(s); |
538 | if (puv != final) { |
7d85a32c |
539 | d = uvuni_to_utf8(d, final); |
45005bfb |
540 | puv = final; |
541 | } |
542 | } |
543 | else { |
544 | STRLEN len; |
9f7f3913 |
545 | uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb |
546 | if (uv != puv) { |
0376ff32 |
547 | Move(s, d, len, U8); |
45005bfb |
548 | d += len; |
549 | puv = uv; |
550 | } |
551 | s += len; |
4757a243 |
552 | } |
4757a243 |
553 | continue; |
554 | } |
01ec43d0 |
555 | matches++; /* "none+1" is delete character */ |
556 | s += UTF8SKIP(s); |
a0ed51b3 |
557 | } |
79072805 |
558 | } |
559 | else { |
4757a243 |
560 | while (s < send) { |
f54cb97a |
561 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb |
562 | if (d > dend) { |
f54cb97a |
563 | const STRLEN clen = d - dstart; |
564 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; |
9b877dbb |
565 | if (!grows) |
7d85a32c |
566 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 |
567 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b |
568 | d = dstart + clen; |
569 | dend = dstart + nlen; |
a0ed51b3 |
570 | } |
4757a243 |
571 | if (uv < none) { |
572 | matches++; |
01ec43d0 |
573 | s += UTF8SKIP(s); |
7d85a32c |
574 | d = uvuni_to_utf8(d, uv); |
4757a243 |
575 | continue; |
a0ed51b3 |
576 | } |
4757a243 |
577 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a |
578 | const int i = UTF8SKIP(s); |
0376ff32 |
579 | Move(s, d, i, U8); |
9b877dbb |
580 | d += i; |
581 | s += i; |
4757a243 |
582 | continue; |
79072805 |
583 | } |
4757a243 |
584 | else if (uv == extra && !del) { |
585 | matches++; |
01ec43d0 |
586 | s += UTF8SKIP(s); |
7d85a32c |
587 | d = uvuni_to_utf8(d, final); |
4757a243 |
588 | continue; |
589 | } |
01ec43d0 |
590 | matches++; /* "none+1" is delete character */ |
591 | s += UTF8SKIP(s); |
79072805 |
592 | } |
4757a243 |
593 | } |
16ec844d |
594 | if (grows || hibit) { |
9b877dbb |
595 | sv_setpvn(sv, (char*)dstart, d - dstart); |
596 | Safefree(dstart); |
16ec844d |
597 | if (grows && hibit) |
598 | Safefree(start); |
9b877dbb |
599 | } |
600 | else { |
601 | *d = '\0'; |
602 | SvCUR_set(sv, d - dstart); |
603 | } |
1aa99e6b |
604 | SvUTF8_on(sv); |
4757a243 |
605 | SvSETMAGIC(sv); |
606 | |
607 | return matches; |
608 | } |
609 | |
610 | I32 |
864dbfa3 |
611 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 |
612 | { |
97aff369 |
613 | dVAR; |
4757a243 |
614 | STRLEN len; |
4373e329 |
615 | const I32 hasutf = (PL_op->op_private & |
036b4402 |
616 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 |
617 | |
2233f375 |
618 | if (SvREADONLY(sv)) { |
765f542d |
619 | if (SvIsCOW(sv)) |
620 | sv_force_normal_flags(sv, 0); |
2233f375 |
621 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) |
622 | Perl_croak(aTHX_ PL_no_modify); |
623 | } |
10516c54 |
624 | (void)SvPV_const(sv, len); |
4757a243 |
625 | if (!len) |
626 | return 0; |
d59e14db |
627 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
628 | if (!SvPOKp(sv)) |
629 | (void)SvPV_force(sv, len); |
2de7b02f |
630 | (void)SvPOK_only_UTF8(sv); |
d59e14db |
631 | } |
4757a243 |
632 | |
cea2e8a9 |
633 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 |
634 | |
a77e643a |
635 | switch (PL_op->op_private & ~hasutf & ( |
636 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| |
637 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { |
4757a243 |
638 | case 0: |
01ec43d0 |
639 | if (hasutf) |
640 | return do_trans_simple_utf8(sv); |
641 | else |
642 | return do_trans_simple(sv); |
4757a243 |
643 | |
644 | case OPpTRANS_IDENTICAL: |
05d340b8 |
645 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 |
646 | if (hasutf) |
647 | return do_trans_count_utf8(sv); |
648 | else |
649 | return do_trans_count(sv); |
4757a243 |
650 | |
651 | default: |
01ec43d0 |
652 | if (hasutf) |
036b4402 |
653 | return do_trans_complex_utf8(sv); |
4757a243 |
654 | else |
036b4402 |
655 | return do_trans_complex(sv); |
79072805 |
656 | } |
79072805 |
657 | } |
658 | |
659 | void |
864dbfa3 |
660 | Perl_do_join(pTHX_ register SV *sv, SV *del, register SV **mark, register SV **sp) |
79072805 |
661 | { |
97aff369 |
662 | dVAR; |
53c1dcc0 |
663 | SV ** const oldmark = mark; |
79072805 |
664 | register I32 items = sp - mark; |
79072805 |
665 | register STRLEN len; |
463ee0b2 |
666 | STRLEN delimlen; |
79072805 |
667 | |
5c144d81 |
668 | (void) SvPV_const(del, delimlen); /* stringify and get the delimlen */ |
5ba99574 |
669 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 |
670 | |
79072805 |
671 | mark++; |
672 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); |
862a34c6 |
673 | SvUPGRADE(sv, SVt_PV); |
79072805 |
674 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
675 | while (items-- > 0) { |
1426bbf4 |
676 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a |
677 | STRLEN tmplen; |
5c144d81 |
678 | SvPV_const(*mark, tmplen); |
463ee0b2 |
679 | len += tmplen; |
79072805 |
680 | } |
681 | mark++; |
682 | } |
683 | SvGROW(sv, len + 1); /* so try to pre-extend */ |
684 | |
685 | mark = oldmark; |
db7c17d7 |
686 | items = sp - mark; |
79072805 |
687 | ++mark; |
688 | } |
689 | |
fc0199cd |
690 | sv_setpvn(sv, "", 0); |
e4803c42 |
691 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 |
692 | SvUTF8_off(sv); |
e4803c42 |
693 | |
8d6d96c1 |
694 | if (PL_tainting && SvMAGICAL(sv)) |
695 | SvTAINTED_off(sv); |
696 | |
463ee0b2 |
697 | if (items-- > 0) { |
92d29cee |
698 | if (*mark) |
699 | sv_catsv(sv, *mark); |
463ee0b2 |
700 | mark++; |
701 | } |
8d6d96c1 |
702 | |
c512ce4f |
703 | if (delimlen) { |
79072805 |
704 | for (; items > 0; items--,mark++) { |
c512ce4f |
705 | sv_catsv(sv,del); |
79072805 |
706 | sv_catsv(sv,*mark); |
707 | } |
708 | } |
709 | else { |
710 | for (; items > 0; items--,mark++) |
711 | sv_catsv(sv,*mark); |
712 | } |
713 | SvSETMAGIC(sv); |
714 | } |
715 | |
716 | void |
864dbfa3 |
717 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 |
718 | { |
97aff369 |
719 | dVAR; |
46fc3d4c |
720 | STRLEN patlen; |
53c1dcc0 |
721 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c |
722 | bool do_taint = FALSE; |
723 | |
5b781b5b |
724 | SvUTF8_off(sv); |
2cf2cfc6 |
725 | if (DO_UTF8(*sarg)) |
726 | SvUTF8_on(sv); |
4608196e |
727 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint); |
79072805 |
728 | SvSETMAGIC(sv); |
46fc3d4c |
729 | if (do_taint) |
730 | SvTAINTED_on(sv); |
79072805 |
731 | } |
732 | |
33b45480 |
733 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 |
734 | UV |
735 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) |
736 | { |
97aff369 |
737 | dVAR; |
81e118e0 |
738 | STRLEN srclen, len; |
10516c54 |
739 | const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); |
81e118e0 |
740 | UV retnum = 0; |
741 | |
a50d7633 |
742 | if (offset < 0) |
81e118e0 |
743 | return retnum; |
8e84507e |
744 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
745 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 |
746 | |
dcad2880 |
747 | if (SvUTF8(sv)) |
33b45480 |
748 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 |
749 | |
81e118e0 |
750 | offset *= size; /* turn into bit offset */ |
751 | len = (offset + size + 7) / 8; /* required number of bytes */ |
752 | if (len > srclen) { |
753 | if (size <= 8) |
754 | retnum = 0; |
755 | else { |
756 | offset >>= 3; /* turn into byte offset */ |
757 | if (size == 16) { |
eb160463 |
758 | if ((STRLEN)offset >= srclen) |
81e118e0 |
759 | retnum = 0; |
760 | else |
628e1a40 |
761 | retnum = (UV) s[offset] << 8; |
81e118e0 |
762 | } |
763 | else if (size == 32) { |
eb160463 |
764 | if ((STRLEN)offset >= srclen) |
81e118e0 |
765 | retnum = 0; |
eb160463 |
766 | else if ((STRLEN)(offset + 1) >= srclen) |
81e118e0 |
767 | retnum = |
768 | ((UV) s[offset ] << 24); |
eb160463 |
769 | else if ((STRLEN)(offset + 2) >= srclen) |
81e118e0 |
770 | retnum = |
771 | ((UV) s[offset ] << 24) + |
772 | ((UV) s[offset + 1] << 16); |
773 | else |
774 | retnum = |
775 | ((UV) s[offset ] << 24) + |
776 | ((UV) s[offset + 1] << 16) + |
777 | ( s[offset + 2] << 8); |
778 | } |
d7d93a81 |
779 | #ifdef UV_IS_QUAD |
c5a0f51a |
780 | else if (size == 64) { |
c5a0f51a |
781 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
782 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
783 | "Bit vector size > 32 non-portable"); |
784 | if (offset >= srclen) |
785 | retnum = 0; |
786 | else if (offset + 1 >= srclen) |
787 | retnum = |
788 | (UV) s[offset ] << 56; |
789 | else if (offset + 2 >= srclen) |
790 | retnum = |
791 | ((UV) s[offset ] << 56) + |
792 | ((UV) s[offset + 1] << 48); |
793 | else if (offset + 3 >= srclen) |
794 | retnum = |
795 | ((UV) s[offset ] << 56) + |
796 | ((UV) s[offset + 1] << 48) + |
797 | ((UV) s[offset + 2] << 40); |
798 | else if (offset + 4 >= srclen) |
799 | retnum = |
800 | ((UV) s[offset ] << 56) + |
801 | ((UV) s[offset + 1] << 48) + |
802 | ((UV) s[offset + 2] << 40) + |
803 | ((UV) s[offset + 3] << 32); |
804 | else if (offset + 5 >= srclen) |
805 | retnum = |
806 | ((UV) s[offset ] << 56) + |
807 | ((UV) s[offset + 1] << 48) + |
808 | ((UV) s[offset + 2] << 40) + |
809 | ((UV) s[offset + 3] << 32) + |
810 | ( s[offset + 4] << 24); |
811 | else if (offset + 6 >= srclen) |
812 | retnum = |
813 | ((UV) s[offset ] << 56) + |
814 | ((UV) s[offset + 1] << 48) + |
815 | ((UV) s[offset + 2] << 40) + |
816 | ((UV) s[offset + 3] << 32) + |
817 | ((UV) s[offset + 4] << 24) + |
818 | ((UV) s[offset + 5] << 16); |
819 | else |
8e84507e |
820 | retnum = |
c5a0f51a |
821 | ((UV) s[offset ] << 56) + |
822 | ((UV) s[offset + 1] << 48) + |
823 | ((UV) s[offset + 2] << 40) + |
824 | ((UV) s[offset + 3] << 32) + |
825 | ((UV) s[offset + 4] << 24) + |
826 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
827 | ( s[offset + 6] << 8); |
c5a0f51a |
828 | } |
829 | #endif |
81e118e0 |
830 | } |
831 | } |
832 | else if (size < 8) |
833 | retnum = (s[offset >> 3] >> (offset & 7)) & ((1 << size) - 1); |
834 | else { |
835 | offset >>= 3; /* turn into byte offset */ |
836 | if (size == 8) |
837 | retnum = s[offset]; |
838 | else if (size == 16) |
839 | retnum = |
628e1a40 |
840 | ((UV) s[offset] << 8) + |
81e118e0 |
841 | s[offset + 1]; |
842 | else if (size == 32) |
843 | retnum = |
844 | ((UV) s[offset ] << 24) + |
845 | ((UV) s[offset + 1] << 16) + |
846 | ( s[offset + 2] << 8) + |
847 | s[offset + 3]; |
d7d93a81 |
848 | #ifdef UV_IS_QUAD |
c5a0f51a |
849 | else if (size == 64) { |
c5a0f51a |
850 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
851 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
852 | "Bit vector size > 32 non-portable"); |
853 | retnum = |
854 | ((UV) s[offset ] << 56) + |
855 | ((UV) s[offset + 1] << 48) + |
856 | ((UV) s[offset + 2] << 40) + |
857 | ((UV) s[offset + 3] << 32) + |
858 | ((UV) s[offset + 4] << 24) + |
859 | ((UV) s[offset + 5] << 16) + |
628e1a40 |
860 | ( s[offset + 6] << 8) + |
c5a0f51a |
861 | s[offset + 7]; |
862 | } |
863 | #endif |
81e118e0 |
864 | } |
865 | |
866 | return retnum; |
867 | } |
868 | |
33b45480 |
869 | /* currently converts input to bytes if possible but doesn't sweat failures, |
870 | * although it does ensure that the string it clobbers is not marked as |
871 | * utf8-valid any more |
872 | */ |
79072805 |
873 | void |
864dbfa3 |
874 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 |
875 | { |
97aff369 |
876 | dVAR; |
79072805 |
877 | register I32 offset; |
878 | register I32 size; |
8990e307 |
879 | register unsigned char *s; |
81e118e0 |
880 | register UV lval; |
79072805 |
881 | I32 mask; |
a0d0e21e |
882 | STRLEN targlen; |
883 | STRLEN len; |
c4420975 |
884 | SV * const targ = LvTARG(sv); |
79072805 |
885 | |
8990e307 |
886 | if (!targ) |
887 | return; |
a0d0e21e |
888 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 |
889 | if (SvUTF8(targ)) { |
33b45480 |
890 | /* This is handled by the SvPOK_only below... |
891 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) |
892 | SvUTF8_off(targ); |
893 | */ |
894 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); |
246fae53 |
895 | } |
896 | |
4ebbc975 |
897 | (void)SvPOK_only(targ); |
81e118e0 |
898 | lval = SvUV(sv); |
79072805 |
899 | offset = LvTARGOFF(sv); |
fe58ced6 |
900 | if (offset < 0) |
ed9aa3b7 |
901 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 |
902 | size = LvTARGLEN(sv); |
8e84507e |
903 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 |
904 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e |
905 | |
81e118e0 |
906 | offset *= size; /* turn into bit offset */ |
907 | len = (offset + size + 7) / 8; /* required number of bytes */ |
a0d0e21e |
908 | if (len > targlen) { |
909 | s = (unsigned char*)SvGROW(targ, len + 1); |
12ae5dfc |
910 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e |
911 | SvCUR_set(targ, len); |
912 | } |
8e84507e |
913 | |
79072805 |
914 | if (size < 8) { |
915 | mask = (1 << size) - 1; |
916 | size = offset & 7; |
917 | lval &= mask; |
81e118e0 |
918 | offset >>= 3; /* turn into byte offset */ |
79072805 |
919 | s[offset] &= ~(mask << size); |
920 | s[offset] |= lval << size; |
921 | } |
922 | else { |
81e118e0 |
923 | offset >>= 3; /* turn into byte offset */ |
79072805 |
924 | if (size == 8) |
eb160463 |
925 | s[offset ] = (U8)( lval & 0xff); |
79072805 |
926 | else if (size == 16) { |
eb160463 |
927 | s[offset ] = (U8)((lval >> 8) & 0xff); |
928 | s[offset+1] = (U8)( lval & 0xff); |
79072805 |
929 | } |
930 | else if (size == 32) { |
eb160463 |
931 | s[offset ] = (U8)((lval >> 24) & 0xff); |
932 | s[offset+1] = (U8)((lval >> 16) & 0xff); |
933 | s[offset+2] = (U8)((lval >> 8) & 0xff); |
934 | s[offset+3] = (U8)( lval & 0xff); |
c5a0f51a |
935 | } |
d7d93a81 |
936 | #ifdef UV_IS_QUAD |
c5a0f51a |
937 | else if (size == 64) { |
c5a0f51a |
938 | if (ckWARN(WARN_PORTABLE)) |
9014280d |
939 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a |
940 | "Bit vector size > 32 non-portable"); |
eb160463 |
941 | s[offset ] = (U8)((lval >> 56) & 0xff); |
942 | s[offset+1] = (U8)((lval >> 48) & 0xff); |
943 | s[offset+2] = (U8)((lval >> 40) & 0xff); |
944 | s[offset+3] = (U8)((lval >> 32) & 0xff); |
945 | s[offset+4] = (U8)((lval >> 24) & 0xff); |
946 | s[offset+5] = (U8)((lval >> 16) & 0xff); |
947 | s[offset+6] = (U8)((lval >> 8) & 0xff); |
948 | s[offset+7] = (U8)( lval & 0xff); |
79072805 |
949 | } |
dc1e3f56 |
950 | #endif |
79072805 |
951 | } |
7bb043c3 |
952 | SvSETMAGIC(targ); |
79072805 |
953 | } |
954 | |
955 | void |
864dbfa3 |
956 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 |
957 | { |
97aff369 |
958 | dVAR; |
463ee0b2 |
959 | STRLEN len; |
a0d0e21e |
960 | char *s; |
8e84507e |
961 | |
79072805 |
962 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e |
963 | register I32 i; |
c4420975 |
964 | AV* const av = (AV*)sv; |
f54cb97a |
965 | const I32 max = AvFILL(av); |
966 | |
967 | for (i = 0; i <= max; i++) { |
a0d0e21e |
968 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
969 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
970 | do_chop(astr, sv); |
971 | } |
972 | return; |
79072805 |
973 | } |
aa854799 |
974 | else if (SvTYPE(sv) == SVt_PVHV) { |
c4420975 |
975 | HV* const hv = (HV*)sv; |
a0d0e21e |
976 | HE* entry; |
977 | (void)hv_iterinit(hv); |
155aba94 |
978 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
979 | do_chop(astr,hv_iterval(hv,entry)); |
980 | return; |
79072805 |
981 | } |
991e6d41 |
982 | else if (SvREADONLY(sv)) { |
983 | if (SvFAKE(sv)) { |
984 | /* SV is copy-on-write */ |
985 | sv_force_normal_flags(sv, 0); |
986 | } |
987 | if (SvREADONLY(sv)) |
988 | Perl_croak(aTHX_ PL_no_modify); |
989 | } |
565a3db3 |
990 | |
991 | if (PL_encoding && !SvUTF8(sv)) { |
992 | /* like in do_chomp(), utf8-ize the sv as a side-effect |
993 | * if we're using encoding. */ |
994 | sv_recode_to_utf8(sv, PL_encoding); |
995 | } |
996 | |
a0d0e21e |
997 | s = SvPV(sv, len); |
748a9306 |
998 | if (len && !SvPOK(sv)) |
a0d0e21e |
999 | s = SvPV_force(sv, len); |
7e2040f0 |
1000 | if (DO_UTF8(sv)) { |
a0ed51b3 |
1001 | if (s && len) { |
c4420975 |
1002 | char * const send = s + len; |
0bcc34c2 |
1003 | char * const start = s; |
a0ed51b3 |
1004 | s = send - 1; |
a0dbb045 |
1005 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
1006 | s--; |
ace7757b |
1007 | if (is_utf8_string((U8*)s, send - s)) { |
a0dbb045 |
1008 | sv_setpvn(astr, s, send - s); |
1009 | *s = '\0'; |
1010 | SvCUR_set(sv, s - start); |
1011 | SvNIOK_off(sv); |
1012 | SvUTF8_on(astr); |
1013 | } |
a0ed51b3 |
1014 | } |
1015 | else |
1016 | sv_setpvn(astr, "", 0); |
1017 | } |
7e2040f0 |
1018 | else if (s && len) { |
a0d0e21e |
1019 | s += --len; |
1020 | sv_setpvn(astr, s, 1); |
1021 | *s = '\0'; |
1022 | SvCUR_set(sv, len); |
2c19a612 |
1023 | SvUTF8_off(sv); |
a0d0e21e |
1024 | SvNIOK_off(sv); |
79072805 |
1025 | } |
1026 | else |
a0d0e21e |
1027 | sv_setpvn(astr, "", 0); |
1028 | SvSETMAGIC(sv); |
7e2040f0 |
1029 | } |
a0d0e21e |
1030 | |
1031 | I32 |
864dbfa3 |
1032 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e |
1033 | { |
97aff369 |
1034 | dVAR; |
c07a80fd |
1035 | register I32 count; |
a0d0e21e |
1036 | STRLEN len; |
1037 | char *s; |
c4c87a06 |
1038 | char *temp_buffer = NULL; |
a0714e2c |
1039 | SV* svrecode = NULL; |
c07a80fd |
1040 | |
3280af22 |
1041 | if (RsSNARF(PL_rs)) |
c07a80fd |
1042 | return 0; |
4c5a6083 |
1043 | if (RsRECORD(PL_rs)) |
1044 | return 0; |
c07a80fd |
1045 | count = 0; |
a0d0e21e |
1046 | if (SvTYPE(sv) == SVt_PVAV) { |
1047 | register I32 i; |
890ce7af |
1048 | AV* const av = (AV*)sv; |
f54cb97a |
1049 | const I32 max = AvFILL(av); |
1050 | |
1051 | for (i = 0; i <= max; i++) { |
a0d0e21e |
1052 | sv = (SV*)av_fetch(av, i, FALSE); |
3280af22 |
1053 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e |
1054 | count += do_chomp(sv); |
1055 | } |
1056 | return count; |
1057 | } |
aa854799 |
1058 | else if (SvTYPE(sv) == SVt_PVHV) { |
890ce7af |
1059 | HV* const hv = (HV*)sv; |
a0d0e21e |
1060 | HE* entry; |
1061 | (void)hv_iterinit(hv); |
155aba94 |
1062 | while ((entry = hv_iternext(hv))) |
a0d0e21e |
1063 | count += do_chomp(hv_iterval(hv,entry)); |
1064 | return count; |
1065 | } |
991e6d41 |
1066 | else if (SvREADONLY(sv)) { |
1067 | if (SvFAKE(sv)) { |
1068 | /* SV is copy-on-write */ |
1069 | sv_force_normal_flags(sv, 0); |
1070 | } |
1071 | if (SvREADONLY(sv)) |
1072 | Perl_croak(aTHX_ PL_no_modify); |
1073 | } |
a6aa349d |
1074 | |
1075 | if (PL_encoding) { |
1076 | if (!SvUTF8(sv)) { |
1077 | /* XXX, here sv is utf8-ized as a side-effect! |
1078 | If encoding.pm is used properly, almost string-generating |
1079 | operations, including literal strings, chr(), input data, etc. |
1080 | should have been utf8-ized already, right? |
1081 | */ |
1082 | sv_recode_to_utf8(sv, PL_encoding); |
1083 | } |
1084 | } |
1085 | |
a0d0e21e |
1086 | s = SvPV(sv, len); |
a0d0e21e |
1087 | if (s && len) { |
1088 | s += --len; |
3280af22 |
1089 | if (RsPARA(PL_rs)) { |
a0d0e21e |
1090 | if (*s != '\n') |
1091 | goto nope; |
1092 | ++count; |
1093 | while (len && s[-1] == '\n') { |
1094 | --len; |
1095 | --s; |
1096 | ++count; |
1097 | } |
1098 | } |
a0d0e21e |
1099 | else { |
a6aa349d |
1100 | STRLEN rslen, rs_charlen; |
93524f2b |
1101 | const char *rsptr = SvPV_const(PL_rs, rslen); |
a6aa349d |
1102 | |
1103 | rs_charlen = SvUTF8(PL_rs) |
1104 | ? sv_len_utf8(PL_rs) |
1105 | : rslen; |
1106 | |
c4c87a06 |
1107 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1108 | /* Assumption is that rs is shorter than the scalar. */ |
1109 | if (SvUTF8(PL_rs)) { |
1110 | /* RS is utf8, scalar is 8 bit. */ |
1111 | bool is_utf8 = TRUE; |
1112 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, |
1113 | &rslen, &is_utf8); |
1114 | if (is_utf8) { |
1115 | /* Cannot downgrade, therefore cannot possibly match |
1116 | */ |
1117 | assert (temp_buffer == rsptr); |
1118 | temp_buffer = NULL; |
1119 | goto nope; |
1120 | } |
1121 | rsptr = temp_buffer; |
a6aa349d |
1122 | } |
1123 | else if (PL_encoding) { |
1124 | /* RS is 8 bit, encoding.pm is used. |
1125 | * Do not recode PL_rs as a side-effect. */ |
1126 | svrecode = newSVpvn(rsptr, rslen); |
1127 | sv_recode_to_utf8(svrecode, PL_encoding); |
93524f2b |
1128 | rsptr = SvPV_const(svrecode, rslen); |
a6aa349d |
1129 | rs_charlen = sv_len_utf8(svrecode); |
1130 | } |
1131 | else { |
c4c87a06 |
1132 | /* RS is 8 bit, scalar is utf8. */ |
1133 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); |
1134 | rsptr = temp_buffer; |
1135 | } |
1136 | } |
c07a80fd |
1137 | if (rslen == 1) { |
1138 | if (*s != *rsptr) |
1139 | goto nope; |
1140 | ++count; |
1141 | } |
1142 | else { |
8c2cee6f |
1143 | if (len < rslen - 1) |
c07a80fd |
1144 | goto nope; |
1145 | len -= rslen - 1; |
1146 | s -= rslen - 1; |
36477c24 |
1147 | if (memNE(s, rsptr, rslen)) |
c07a80fd |
1148 | goto nope; |
a6aa349d |
1149 | count += rs_charlen; |
c07a80fd |
1150 | } |
a0d0e21e |
1151 | } |
8b6b16e7 |
1152 | s = SvPV_force_nolen(sv); |
a0d0e21e |
1153 | SvCUR_set(sv, len); |
9b33ce3b |
1154 | *SvEND(sv) = '\0'; |
a0d0e21e |
1155 | SvNIOK_off(sv); |
9b33ce3b |
1156 | SvSETMAGIC(sv); |
a0d0e21e |
1157 | } |
1158 | nope: |
a6aa349d |
1159 | |
1160 | if (svrecode) |
1161 | SvREFCNT_dec(svrecode); |
1162 | |
c4c87a06 |
1163 | Safefree(temp_buffer); |
a0d0e21e |
1164 | return count; |
8e84507e |
1165 | } |
79072805 |
1166 | |
1167 | void |
864dbfa3 |
1168 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 |
1169 | { |
97aff369 |
1170 | dVAR; |
79072805 |
1171 | #ifdef LIBERAL |
1172 | register long *dl; |
1173 | register long *ll; |
1174 | register long *rl; |
1175 | #endif |
1176 | register char *dc; |
463ee0b2 |
1177 | STRLEN leftlen; |
1178 | STRLEN rightlen; |
e62f0680 |
1179 | register const char *lc; |
1180 | register const char *rc; |
79072805 |
1181 | register I32 len; |
a0d0e21e |
1182 | I32 lensave; |
e62f0680 |
1183 | const char *lsave; |
1184 | const char *rsave; |
f54cb97a |
1185 | const bool left_utf = DO_UTF8(left); |
1186 | const bool right_utf = DO_UTF8(right); |
b7953727 |
1187 | I32 needlen = 0; |
0c57e439 |
1188 | |
1189 | if (left_utf && !right_utf) |
1190 | sv_utf8_upgrade(right); |
a1ca4561 |
1191 | else if (!left_utf && right_utf) |
0c57e439 |
1192 | sv_utf8_upgrade(left); |
79072805 |
1193 | |
1fbd88dc |
1194 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
1195 | sv_setpvn(sv, "", 0); /* avoid undef warning on |= and ^= */ |
e62f0680 |
1196 | lsave = lc = SvPV_nomg_const(left, leftlen); |
1197 | rsave = rc = SvPV_nomg_const(right, rightlen); |
93a17b20 |
1198 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e |
1199 | lensave = len; |
c9b3c8d0 |
1200 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1201 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; |
a02a5408 |
1202 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 |
1203 | } |
1204 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe |
1205 | dc = SvPV_force_nomg_nolen(sv); |
1a787b95 |
1206 | if (SvLEN(sv) < (STRLEN)(len + 1)) { |
eb160463 |
1207 | dc = SvGROW(sv, (STRLEN)(len + 1)); |
ff68c719 |
1208 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1209 | } |
c9b3c8d0 |
1210 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1211 | dc = SvGROW(sv, leftlen + rightlen + 1); |
ff68c719 |
1212 | } |
1213 | else { |
c9b3c8d0 |
1214 | needlen = ((optype == OP_BIT_AND) |
1215 | ? len : (leftlen > rightlen ? leftlen : rightlen)); |
a02a5408 |
1216 | Newxz(dc, needlen + 1, char); |
ff68c719 |
1217 | (void)sv_usepvn(sv, dc, needlen); |
1218 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 |
1219 | } |
a0d0e21e |
1220 | SvCUR_set(sv, len); |
1221 | (void)SvPOK_only(sv); |
0c57e439 |
1222 | if (left_utf || right_utf) { |
1223 | UV duc, luc, ruc; |
c4420975 |
1224 | char * const dcsave = dc; |
0c57e439 |
1225 | STRLEN lulen = leftlen; |
1226 | STRLEN rulen = rightlen; |
ba210ebe |
1227 | STRLEN ulen; |
0c57e439 |
1228 | |
1229 | switch (optype) { |
1230 | case OP_BIT_AND: |
1231 | while (lulen && rulen) { |
9041c2e3 |
1232 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1233 | lc += ulen; |
1234 | lulen -= ulen; |
9041c2e3 |
1235 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1236 | rc += ulen; |
1237 | rulen -= ulen; |
1238 | duc = luc & ruc; |
9041c2e3 |
1239 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1240 | } |
c9b3c8d0 |
1241 | if (sv == left || sv == right) |
1242 | (void)sv_usepvn(sv, dcsave, needlen); |
1243 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1244 | break; |
1245 | case OP_BIT_XOR: |
1246 | while (lulen && rulen) { |
9041c2e3 |
1247 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1248 | lc += ulen; |
1249 | lulen -= ulen; |
9041c2e3 |
1250 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1251 | rc += ulen; |
1252 | rulen -= ulen; |
1253 | duc = luc ^ ruc; |
9041c2e3 |
1254 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1255 | } |
1256 | goto mop_up_utf; |
1257 | case OP_BIT_OR: |
1258 | while (lulen && rulen) { |
9041c2e3 |
1259 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1260 | lc += ulen; |
1261 | lulen -= ulen; |
9041c2e3 |
1262 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 |
1263 | rc += ulen; |
1264 | rulen -= ulen; |
1265 | duc = luc | ruc; |
9041c2e3 |
1266 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 |
1267 | } |
1268 | mop_up_utf: |
c9b3c8d0 |
1269 | if (sv == left || sv == right) |
1270 | (void)sv_usepvn(sv, dcsave, needlen); |
1271 | SvCUR_set(sv, dc - dcsave); |
0c57e439 |
1272 | if (rulen) |
1273 | sv_catpvn(sv, rc, rulen); |
1274 | else if (lulen) |
1275 | sv_catpvn(sv, lc, lulen); |
1276 | else |
1277 | *SvEND(sv) = '\0'; |
1278 | break; |
1279 | } |
1280 | SvUTF8_on(sv); |
1281 | goto finish; |
1282 | } |
1283 | else |
79072805 |
1284 | #ifdef LIBERAL |
1285 | if (len >= sizeof(long)*4 && |
1286 | !((long)dc % sizeof(long)) && |
1287 | !((long)lc % sizeof(long)) && |
1288 | !((long)rc % sizeof(long))) /* It's almost always aligned... */ |
1289 | { |
f54cb97a |
1290 | const I32 remainder = len % (sizeof(long)*4); |
79072805 |
1291 | len /= (sizeof(long)*4); |
1292 | |
1293 | dl = (long*)dc; |
1294 | ll = (long*)lc; |
1295 | rl = (long*)rc; |
1296 | |
1297 | switch (optype) { |
1298 | case OP_BIT_AND: |
1299 | while (len--) { |
1300 | *dl++ = *ll++ & *rl++; |
1301 | *dl++ = *ll++ & *rl++; |
1302 | *dl++ = *ll++ & *rl++; |
1303 | *dl++ = *ll++ & *rl++; |
1304 | } |
1305 | break; |
a0d0e21e |
1306 | case OP_BIT_XOR: |
79072805 |
1307 | while (len--) { |
1308 | *dl++ = *ll++ ^ *rl++; |
1309 | *dl++ = *ll++ ^ *rl++; |
1310 | *dl++ = *ll++ ^ *rl++; |
1311 | *dl++ = *ll++ ^ *rl++; |
1312 | } |
1313 | break; |
1314 | case OP_BIT_OR: |
1315 | while (len--) { |
1316 | *dl++ = *ll++ | *rl++; |
1317 | *dl++ = *ll++ | *rl++; |
1318 | *dl++ = *ll++ | *rl++; |
1319 | *dl++ = *ll++ | *rl++; |
1320 | } |
1321 | } |
1322 | |
1323 | dc = (char*)dl; |
1324 | lc = (char*)ll; |
1325 | rc = (char*)rl; |
1326 | |
1327 | len = remainder; |
1328 | } |
1329 | #endif |
a0d0e21e |
1330 | { |
a0d0e21e |
1331 | switch (optype) { |
1332 | case OP_BIT_AND: |
1333 | while (len--) |
1334 | *dc++ = *lc++ & *rc++; |
1a787b95 |
1335 | *dc = '\0'; |
a0d0e21e |
1336 | break; |
1337 | case OP_BIT_XOR: |
1338 | while (len--) |
1339 | *dc++ = *lc++ ^ *rc++; |
1340 | goto mop_up; |
1341 | case OP_BIT_OR: |
1342 | while (len--) |
1343 | *dc++ = *lc++ | *rc++; |
1344 | mop_up: |
1345 | len = lensave; |
eb160463 |
1346 | if (rightlen > (STRLEN)len) |
a0d0e21e |
1347 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 |
1348 | else if (leftlen > (STRLEN)len) |
a0d0e21e |
1349 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 |
1350 | else |
1351 | *SvEND(sv) = '\0'; |
a0d0e21e |
1352 | break; |
1353 | } |
79072805 |
1354 | } |
0c57e439 |
1355 | finish: |
fb73857a |
1356 | SvTAINT(sv); |
79072805 |
1357 | } |
463ee0b2 |
1358 | |
1359 | OP * |
cea2e8a9 |
1360 | Perl_do_kv(pTHX) |
463ee0b2 |
1361 | { |
97aff369 |
1362 | dVAR; |
39644a26 |
1363 | dSP; |
c4420975 |
1364 | HV * const hv = (HV*)POPs; |
800e9ae0 |
1365 | HV *keys; |
463ee0b2 |
1366 | register HE *entry; |
f54cb97a |
1367 | const I32 gimme = GIMME_V; |
1368 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); |
1369 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); |
1370 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); |
463ee0b2 |
1371 | |
85581909 |
1372 | if (!hv) { |
78f9721b |
1373 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1374 | dTARGET; /* make sure to clear its target here */ |
1375 | if (SvTYPE(TARG) == SVt_PVLV) |
a0714e2c |
1376 | LvTARG(TARG) = NULL; |
85581909 |
1377 | PUSHs(TARG); |
1378 | } |
463ee0b2 |
1379 | RETURN; |
85581909 |
1380 | } |
748a9306 |
1381 | |
6d822dc4 |
1382 | keys = hv; |
800e9ae0 |
1383 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 |
1384 | |
54310121 |
1385 | if (gimme == G_VOID) |
aa689395 |
1386 | RETURN; |
1387 | |
54310121 |
1388 | if (gimme == G_SCALAR) { |
6ee623d5 |
1389 | IV i; |
463ee0b2 |
1390 | dTARGET; |
1391 | |
78f9721b |
1392 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 |
1393 | if (SvTYPE(TARG) < SVt_PVLV) { |
1394 | sv_upgrade(TARG, SVt_PVLV); |
a0714e2c |
1395 | sv_magic(TARG, NULL, PERL_MAGIC_nkeys, NULL, 0); |
85581909 |
1396 | } |
1397 | LvTYPE(TARG) = 'k'; |
800e9ae0 |
1398 | if (LvTARG(TARG) != (SV*)keys) { |
6ff81951 |
1399 | if (LvTARG(TARG)) |
1400 | SvREFCNT_dec(LvTARG(TARG)); |
b37c2d43 |
1401 | LvTARG(TARG) = SvREFCNT_inc_simple(keys); |
6ff81951 |
1402 | } |
85581909 |
1403 | PUSHs(TARG); |
1404 | RETURN; |
1405 | } |
1406 | |
14befaf4 |
1407 | if (! SvTIED_mg((SV*)keys, PERL_MAGIC_tied)) |
a98face1 |
1408 | i = HvKEYS(keys); |
463ee0b2 |
1409 | else { |
1410 | i = 0; |
800e9ae0 |
1411 | while (hv_iternext(keys)) i++; |
463ee0b2 |
1412 | } |
1413 | PUSHi( i ); |
1414 | RETURN; |
1415 | } |
1416 | |
8ed4b672 |
1417 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 |
1418 | |
463ee0b2 |
1419 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 |
1420 | while ((entry = hv_iternext(keys))) { |
463ee0b2 |
1421 | SPAGAIN; |
574c8022 |
1422 | if (dokeys) { |
890ce7af |
1423 | SV* const sv = hv_iterkeysv(entry); |
574c8022 |
1424 | XPUSHs(sv); /* won't clobber stack_sp */ |
1425 | } |
463ee0b2 |
1426 | if (dovalues) { |
f54cb97a |
1427 | SV *tmpstr; |
463ee0b2 |
1428 | PUTBACK; |
6d822dc4 |
1429 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 |
1430 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c |
1431 | (unsigned long)HeHASH(entry), |
cdd3ba14 |
1432 | (int)HvMAX(keys)+1, |
800e9ae0 |
1433 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 |
1434 | SPAGAIN; |
46fc3d4c |
1435 | XPUSHs(tmpstr); |
463ee0b2 |
1436 | } |
1437 | PUTBACK; |
1438 | } |
1439 | return NORMAL; |
1440 | } |
4e35701f |
1441 | |
af3babe4 |
1442 | /* |
1443 | * Local variables: |
1444 | * c-indentation-style: bsd |
1445 | * c-basic-offset: 4 |
1446 | * indent-tabs-mode: t |
1447 | * End: |
1448 | * |
37442d52 |
1449 | * ex: set ts=8 sts=4 sw=4 noet: |
1450 | */ |