Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* doop.c |
79072805 | 2 | * |
1129b882 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, |
f4b6e4b3 | 4 | * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2009 by Larry Wall and others |
79072805 LW |
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 LW |
9 | */ |
10 | ||
11 | /* | |
4ac71550 TC |
12 | * 'So that was the job I felt I had to do when I started,' thought Sam. |
13 | * | |
14 | * [p.934 of _The Lord of the Rings_, VI/iii: "Mount Doom"] | |
79072805 LW |
15 | */ |
16 | ||
166f8a29 | 17 | /* This file contains some common functions needed to carry out certain |
347f3823 | 18 | * ops. For example, both pp_sprintf() and pp_prtf() call the function |
fc7d284e | 19 | * do_sprintf() found in this file. |
166f8a29 DM |
20 | */ |
21 | ||
79072805 | 22 | #include "EXTERN.h" |
864dbfa3 | 23 | #define PERL_IN_DOOP_C |
79072805 LW |
24 | #include "perl.h" |
25 | ||
64ca3a65 | 26 | #ifndef PERL_MICRO |
79072805 LW |
27 | #include <signal.h> |
28 | #endif | |
29 | ||
334c6444 DM |
30 | |
31 | /* Helper function for do_trans(). | |
32 | * Handles non-utf8 cases(*) not involving the /c, /d, /s flags, | |
33 | * and where search and replacement charlists aren't identical. | |
34 | * (*) i.e. where the search and replacement charlists are non-utf8. sv may | |
35 | * or may not be utf8. | |
36 | */ | |
37 | ||
f0fd0980 | 38 | STATIC Size_t |
c395bd6c | 39 | S_do_trans_simple(pTHX_ SV * const sv) |
79072805 | 40 | { |
f0fd0980 | 41 | Size_t matches = 0; |
463ee0b2 | 42 | STRLEN len; |
9138d6ca | 43 | U8 *s = (U8*)SvPV_nomg(sv,len); |
c395bd6c | 44 | U8 * const send = s+len; |
c1048fcf | 45 | const OPtrans_map * const tbl = (OPtrans_map*)cPVOP->op_pv; |
7918f24d NC |
46 | |
47 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE; | |
48 | ||
4757a243 | 49 | if (!tbl) |
7d85a32c | 50 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 51 | |
1e54db1a | 52 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 53 | if (!SvUTF8(sv)) { |
01ec43d0 | 54 | while (s < send) { |
e4ee4c1b | 55 | const short ch = tbl->map[*s]; |
f54cb97a | 56 | if (ch >= 0) { |
01ec43d0 | 57 | matches++; |
c4420975 | 58 | *s = (U8)ch; |
01ec43d0 | 59 | } |
c4420975 | 60 | s++; |
01ec43d0 GS |
61 | } |
62 | SvSETMAGIC(sv); | |
9b877dbb IH |
63 | } |
64 | else { | |
e4ee4c1b | 65 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); |
c395bd6c AL |
66 | U8 *d; |
67 | U8 *dstart; | |
68 | ||
69 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ | |
70 | if (grows) | |
71 | Newx(d, len*2+1, U8); | |
72 | else | |
73 | d = s; | |
74 | dstart = d; | |
75 | while (s < send) { | |
76 | STRLEN ulen; | |
e4ee4c1b | 77 | short ch; |
c395bd6c AL |
78 | |
79 | /* Need to check this, otherwise 128..255 won't match */ | |
80 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); | |
c1048fcf | 81 | if (c < 0x100 && (ch = tbl->map[c]) >= 0) { |
c395bd6c | 82 | matches++; |
e4ee4c1b | 83 | d = uvchr_to_utf8(d, (UV)ch); |
c395bd6c AL |
84 | s += ulen; |
85 | } | |
86 | else { /* No match -> copy */ | |
87 | Move(s, d, ulen, U8); | |
88 | d += ulen; | |
89 | s += ulen; | |
90 | } | |
91 | } | |
92 | if (grows) { | |
93 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
94 | Safefree(dstart); | |
95 | } | |
96 | else { | |
97 | *d = '\0'; | |
98 | SvCUR_set(sv, d - dstart); | |
99 | } | |
100 | SvUTF8_on(sv); | |
101 | SvSETMAGIC(sv); | |
9b877dbb | 102 | } |
4757a243 LW |
103 | return matches; |
104 | } | |
105 | ||
334c6444 DM |
106 | |
107 | /* Helper function for do_trans(). | |
108 | * Handles non-utf8 cases(*) where search and replacement charlists are | |
109 | * identical: so the string isn't modified, and only a count of modifiable | |
110 | * chars is needed. | |
111 | * Note that it doesn't handle /d or /s, since these modify the string | |
112 | * even if the replacement list is empty. | |
113 | * (*) i.e. where the search and replacement charlists are non-utf8. sv may | |
114 | * or may not be utf8. | |
115 | */ | |
116 | ||
f0fd0980 | 117 | STATIC Size_t |
c395bd6c | 118 | S_do_trans_count(pTHX_ SV * const sv) |
4757a243 | 119 | { |
4757a243 | 120 | STRLEN len; |
9138d6ca | 121 | const U8 *s = (const U8*)SvPV_nomg_const(sv, len); |
c395bd6c | 122 | const U8 * const send = s + len; |
f0fd0980 | 123 | Size_t matches = 0; |
c1048fcf | 124 | const OPtrans_map * const tbl = (OPtrans_map*)cPVOP->op_pv; |
7918f24d NC |
125 | |
126 | PERL_ARGS_ASSERT_DO_TRANS_COUNT; | |
127 | ||
4757a243 | 128 | if (!tbl) |
7d85a32c | 129 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 | 130 | |
c395bd6c | 131 | if (!SvUTF8(sv)) { |
1aa99e6b | 132 | while (s < send) { |
c1048fcf | 133 | if (tbl->map[*s++] >= 0) |
036b4402 | 134 | matches++; |
1aa99e6b | 135 | } |
c395bd6c | 136 | } |
fabdb6c0 | 137 | else { |
e4ee4c1b | 138 | const bool complement = cBOOL(PL_op->op_private & OPpTRANS_COMPLEMENT); |
1aa99e6b | 139 | while (s < send) { |
1aa99e6b | 140 | STRLEN ulen; |
9f7f3913 | 141 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 | 142 | if (c < 0x100) { |
c1048fcf | 143 | if (tbl->map[c] >= 0) |
8973db79 JH |
144 | matches++; |
145 | } else if (complement) | |
1aa99e6b IH |
146 | matches++; |
147 | s += ulen; | |
148 | } | |
fabdb6c0 | 149 | } |
4757a243 LW |
150 | |
151 | return matches; | |
152 | } | |
153 | ||
334c6444 DM |
154 | |
155 | /* Helper function for do_trans(). | |
156 | * Handles non-utf8 cases(*) involving the /c, /d, /s flags, | |
157 | * and where search and replacement charlists aren't identical. | |
158 | * (*) i.e. where the search and replacement charlists are non-utf8. sv may | |
159 | * or may not be utf8. | |
160 | */ | |
161 | ||
f0fd0980 | 162 | STATIC Size_t |
c395bd6c | 163 | S_do_trans_complex(pTHX_ SV * const sv) |
4757a243 | 164 | { |
c395bd6c | 165 | STRLEN len; |
9138d6ca | 166 | U8 *s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 167 | U8 * const send = s+len; |
f0fd0980 | 168 | Size_t matches = 0; |
0b9a13c3 | 169 | const OPtrans_map * const tbl = (OPtrans_map*)cPVOP->op_pv; |
7918f24d NC |
170 | |
171 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX; | |
172 | ||
0b9a13c3 | 173 | if (!tbl) |
7d85a32c | 174 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 | 175 | |
c395bd6c AL |
176 | if (!SvUTF8(sv)) { |
177 | U8 *d = s; | |
178 | U8 * const dstart = d; | |
4757a243 | 179 | |
1aa99e6b | 180 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a | 181 | const U8* p = send; |
1aa99e6b | 182 | while (s < send) { |
e4ee4c1b | 183 | const short ch = tbl->map[*s]; |
f54cb97a | 184 | if (ch >= 0) { |
eb160463 | 185 | *d = (U8)ch; |
036b4402 | 186 | matches++; |
1aa99e6b IH |
187 | if (p != d - 1 || *p != *d) |
188 | p = d++; | |
189 | } | |
190 | else if (ch == -1) /* -1 is unmapped character */ | |
191 | *d++ = *s; | |
192 | else if (ch == -2) /* -2 is delete character */ | |
193 | matches++; | |
194 | s++; | |
195 | } | |
a0ed51b3 | 196 | } |
1aa99e6b IH |
197 | else { |
198 | while (s < send) { | |
e4ee4c1b | 199 | const short ch = tbl->map[*s]; |
f54cb97a | 200 | if (ch >= 0) { |
1aa99e6b | 201 | matches++; |
eb160463 | 202 | *d++ = (U8)ch; |
1aa99e6b IH |
203 | } |
204 | else if (ch == -1) /* -1 is unmapped character */ | |
205 | *d++ = *s; | |
206 | else if (ch == -2) /* -2 is delete character */ | |
207 | matches++; | |
208 | s++; | |
209 | } | |
210 | } | |
76ef7183 | 211 | *d = '\0'; |
1aa99e6b | 212 | SvCUR_set(sv, d - dstart); |
4757a243 | 213 | } |
c395bd6c | 214 | else { /* is utf8 */ |
e4ee4c1b DM |
215 | const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH); |
216 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); | |
c395bd6c AL |
217 | U8 *d; |
218 | U8 *dstart; | |
0b9a13c3 DM |
219 | Size_t size = tbl->size; |
220 | UV pch = 0xfeedface; | |
fabdb6c0 | 221 | |
9b877dbb | 222 | if (grows) |
a02a5408 | 223 | Newx(d, len*2+1, U8); |
9b877dbb IH |
224 | else |
225 | d = s; | |
1aa99e6b IH |
226 | dstart = d; |
227 | ||
4eaf16e8 DM |
228 | while (s < send) { |
229 | STRLEN len; | |
230 | const UV comp = utf8n_to_uvchr(s, send - s, &len, | |
231 | UTF8_ALLOW_DEFAULT); | |
232 | UV ch; | |
233 | short sch; | |
234 | ||
235 | sch = tbl->map[comp >= size ? size : comp]; | |
236 | ||
237 | if (sch >= 0) { | |
238 | ch = (UV)sch; | |
239 | replace: | |
240 | matches++; | |
241 | if (LIKELY(!squash || ch != pch)) { | |
242 | d = uvchr_to_utf8(d, ch); | |
243 | pch = ch; | |
0b9a13c3 | 244 | } |
4eaf16e8 DM |
245 | s += len; |
246 | continue; | |
247 | } | |
248 | else if (sch == -1) { /* -1 is unmapped character */ | |
249 | Move(s, d, len, U8); | |
250 | d += len; | |
251 | } | |
252 | else if (sch == -2) /* -2 is delete character */ | |
253 | matches++; | |
254 | else { | |
255 | assert(sch == -3); /* -3 is empty replacement */ | |
256 | ch = comp; | |
257 | goto replace; | |
258 | } | |
259 | ||
260 | s += len; | |
261 | pch = 0xfeedface; | |
262 | } | |
0b9a13c3 | 263 | |
9b877dbb IH |
264 | if (grows) { |
265 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
266 | Safefree(dstart); | |
267 | } | |
268 | else { | |
269 | *d = '\0'; | |
270 | SvCUR_set(sv, d - dstart); | |
271 | } | |
1aa99e6b | 272 | SvUTF8_on(sv); |
4757a243 | 273 | } |
5e44153e | 274 | SvSETMAGIC(sv); |
4757a243 LW |
275 | return matches; |
276 | } | |
277 | ||
334c6444 DM |
278 | |
279 | /* Helper function for do_trans(). | |
280 | * Handles utf8 cases(*) not involving the /c, /d, /s flags, | |
281 | * and where search and replacement charlists aren't identical. | |
282 | * (*) i.e. where the search or replacement charlists are utf8. sv may | |
283 | * or may not be utf8. | |
284 | */ | |
285 | ||
f0fd0980 | 286 | STATIC Size_t |
c395bd6c | 287 | S_do_trans_simple_utf8(pTHX_ SV * const sv) |
4757a243 | 288 | { |
4757a243 LW |
289 | U8 *s; |
290 | U8 *send; | |
291 | U8 *d; | |
036b4402 | 292 | U8 *start; |
1aa99e6b | 293 | U8 *dstart, *dend; |
f0fd0980 | 294 | Size_t matches = 0; |
e4ee4c1b | 295 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); |
4757a243 | 296 | STRLEN len; |
043e41b8 DM |
297 | SV* const rv = |
298 | #ifdef USE_ITHREADS | |
299 | PAD_SVl(cPADOP->op_padix); | |
300 | #else | |
ad64d0ec | 301 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 302 | #endif |
85fbaab2 | 303 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 304 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
305 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
306 | const UV extra = none + 1; | |
b7953727 | 307 | UV final = 0; |
1aa99e6b | 308 | U8 hibit = 0; |
4757a243 | 309 | |
7918f24d NC |
310 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8; |
311 | ||
9138d6ca | 312 | s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 313 | if (!SvUTF8(sv)) { |
40b85b80 KW |
314 | hibit = ! is_utf8_invariant_string(s, len); |
315 | if (hibit) { | |
316 | s = bytes_to_utf8(s, &len); | |
c4d5f83a | 317 | } |
1aa99e6b | 318 | } |
4757a243 | 319 | send = s + len; |
036b4402 | 320 | start = s; |
4757a243 | 321 | |
a4fc7abc | 322 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 LW |
323 | if (svp) |
324 | final = SvUV(*svp); | |
325 | ||
9b877dbb IH |
326 | if (grows) { |
327 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 328 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
329 | dend = d + len * 3; |
330 | dstart = d; | |
331 | } | |
332 | else { | |
333 | dstart = d = s; | |
334 | dend = d + len; | |
335 | } | |
1aa99e6b | 336 | |
4757a243 | 337 | while (s < send) { |
0bcc34c2 AL |
338 | const UV uv = swash_fetch(rv, s, TRUE); |
339 | if (uv < none) { | |
4757a243 LW |
340 | s += UTF8SKIP(s); |
341 | matches++; | |
c80e42f3 | 342 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
343 | } |
344 | else if (uv == none) { | |
4373e329 | 345 | const int i = UTF8SKIP(s); |
0376ff32 | 346 | Move(s, d, i, U8); |
9b877dbb IH |
347 | d += i; |
348 | s += i; | |
4757a243 LW |
349 | } |
350 | else if (uv == extra) { | |
4373e329 | 351 | s += UTF8SKIP(s); |
4757a243 | 352 | matches++; |
c80e42f3 | 353 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
354 | } |
355 | else | |
356 | s += UTF8SKIP(s); | |
1aa99e6b | 357 | |
9b877dbb | 358 | if (d > dend) { |
4373e329 AL |
359 | const STRLEN clen = d - dstart; |
360 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 361 | if (!grows) |
7d85a32c | 362 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 | 363 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
364 | d = dstart + clen; |
365 | dend = dstart + nlen; | |
366 | } | |
4757a243 | 367 | } |
16ec844d | 368 | if (grows || hibit) { |
9b877dbb IH |
369 | sv_setpvn(sv, (char*)dstart, d - dstart); |
370 | Safefree(dstart); | |
16ec844d IH |
371 | if (grows && hibit) |
372 | Safefree(start); | |
9b877dbb IH |
373 | } |
374 | else { | |
375 | *d = '\0'; | |
376 | SvCUR_set(sv, d - dstart); | |
377 | } | |
4757a243 | 378 | SvSETMAGIC(sv); |
1aa99e6b | 379 | SvUTF8_on(sv); |
4757a243 LW |
380 | |
381 | return matches; | |
382 | } | |
383 | ||
334c6444 DM |
384 | |
385 | /* Helper function for do_trans(). | |
386 | * Handles utf8 cases(*) where search and replacement charlists are | |
387 | * identical: so the string isn't modified, and only a count of modifiable | |
388 | * chars is needed. | |
389 | * Note that it doesn't handle /d or /s, since these modify the string | |
390 | * even if the replacement charlist is empty. | |
391 | * (*) i.e. where the search or replacement charlists are utf8. sv may | |
392 | * or may not be utf8. | |
393 | */ | |
394 | ||
f0fd0980 | 395 | STATIC Size_t |
c395bd6c | 396 | S_do_trans_count_utf8(pTHX_ SV * const sv) |
4757a243 | 397 | { |
93524f2b | 398 | const U8 *s; |
cbbf8932 AL |
399 | const U8 *start = NULL; |
400 | const U8 *send; | |
f0fd0980 | 401 | Size_t matches = 0; |
4757a243 | 402 | STRLEN len; |
043e41b8 DM |
403 | SV* const rv = |
404 | #ifdef USE_ITHREADS | |
405 | PAD_SVl(cPADOP->op_padix); | |
406 | #else | |
ad64d0ec | 407 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 408 | #endif |
85fbaab2 | 409 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 410 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
411 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
412 | const UV extra = none + 1; | |
1aa99e6b | 413 | U8 hibit = 0; |
4757a243 | 414 | |
7918f24d NC |
415 | PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8; |
416 | ||
9138d6ca | 417 | s = (const U8*)SvPV_nomg_const(sv, len); |
1aa99e6b | 418 | if (!SvUTF8(sv)) { |
40b85b80 KW |
419 | hibit = ! is_utf8_invariant_string(s, len); |
420 | if (hibit) { | |
421 | start = s = bytes_to_utf8(s, &len); | |
c4d5f83a | 422 | } |
1aa99e6b | 423 | } |
4757a243 LW |
424 | send = s + len; |
425 | ||
426 | while (s < send) { | |
c4420975 AL |
427 | const UV uv = swash_fetch(rv, s, TRUE); |
428 | if (uv < none || uv == extra) | |
4757a243 | 429 | matches++; |
834a4ddd | 430 | s += UTF8SKIP(s); |
4757a243 | 431 | } |
1aa99e6b IH |
432 | if (hibit) |
433 | Safefree(start); | |
4757a243 LW |
434 | |
435 | return matches; | |
436 | } | |
437 | ||
334c6444 DM |
438 | |
439 | /* Helper function for do_trans(). | |
440 | * Handles utf8 cases(*) involving the /c, /d, /s flags, | |
441 | * and where search and replacement charlists aren't identical. | |
442 | * (*) i.e. where the search or replacement charlists are utf8. sv may | |
443 | * or may not be utf8. | |
444 | */ | |
445 | ||
f0fd0980 | 446 | STATIC Size_t |
c395bd6c | 447 | S_do_trans_complex_utf8(pTHX_ SV * const sv) |
4757a243 | 448 | { |
1aa99e6b | 449 | U8 *start, *send; |
4757a243 | 450 | U8 *d; |
f0fd0980 | 451 | Size_t matches = 0; |
e4ee4c1b DM |
452 | const bool squash = cBOOL(PL_op->op_private & OPpTRANS_SQUASH); |
453 | const bool del = cBOOL(PL_op->op_private & OPpTRANS_DELETE); | |
454 | const bool grows = cBOOL(PL_op->op_private & OPpTRANS_GROWS); | |
043e41b8 DM |
455 | SV* const rv = |
456 | #ifdef USE_ITHREADS | |
457 | PAD_SVl(cPADOP->op_padix); | |
458 | #else | |
ad64d0ec | 459 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 460 | #endif |
85fbaab2 | 461 | HV * const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 462 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
463 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
464 | const UV extra = none + 1; | |
b7953727 | 465 | UV final = 0; |
45005bfb | 466 | bool havefinal = FALSE; |
4757a243 | 467 | STRLEN len; |
1aa99e6b | 468 | U8 *dstart, *dend; |
1aa99e6b | 469 | U8 hibit = 0; |
9138d6ca | 470 | U8 *s = (U8*)SvPV_nomg(sv, len); |
7918f24d NC |
471 | |
472 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8; | |
473 | ||
c395bd6c | 474 | if (!SvUTF8(sv)) { |
40b85b80 KW |
475 | hibit = ! is_utf8_invariant_string(s, len); |
476 | if (hibit) { | |
477 | s = bytes_to_utf8(s, &len); | |
c4d5f83a | 478 | } |
1aa99e6b | 479 | } |
4757a243 | 480 | send = s + len; |
1aa99e6b | 481 | start = s; |
4757a243 | 482 | |
a4fc7abc | 483 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb | 484 | if (svp) { |
4757a243 | 485 | final = SvUV(*svp); |
45005bfb JH |
486 | havefinal = TRUE; |
487 | } | |
4757a243 | 488 | |
9b877dbb IH |
489 | if (grows) { |
490 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 491 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
492 | dend = d + len * 3; |
493 | dstart = d; | |
494 | } | |
495 | else { | |
496 | dstart = d = s; | |
497 | dend = d + len; | |
498 | } | |
4757a243 LW |
499 | |
500 | if (squash) { | |
501 | UV puv = 0xfeedface; | |
502 | while (s < send) { | |
f54cb97a | 503 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 504 | |
9b877dbb | 505 | if (d > dend) { |
4373e329 AL |
506 | const STRLEN clen = d - dstart; |
507 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 508 | if (!grows) |
7d85a32c | 509 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 510 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
511 | d = dstart + clen; |
512 | dend = dstart + nlen; | |
4757a243 LW |
513 | } |
514 | if (uv < none) { | |
515 | matches++; | |
94472101 | 516 | s += UTF8SKIP(s); |
4757a243 | 517 | if (uv != puv) { |
c80e42f3 | 518 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
519 | puv = uv; |
520 | } | |
4757a243 LW |
521 | continue; |
522 | } | |
523 | else if (uv == none) { /* "none" is unmapped character */ | |
4373e329 | 524 | const int i = UTF8SKIP(s); |
0376ff32 | 525 | Move(s, d, i, U8); |
9b877dbb IH |
526 | d += i; |
527 | s += i; | |
4757a243 LW |
528 | puv = 0xfeedface; |
529 | continue; | |
a0ed51b3 | 530 | } |
4757a243 LW |
531 | else if (uv == extra && !del) { |
532 | matches++; | |
45005bfb JH |
533 | if (havefinal) { |
534 | s += UTF8SKIP(s); | |
535 | if (puv != final) { | |
c80e42f3 | 536 | d = uvchr_to_utf8(d, final); |
45005bfb JH |
537 | puv = final; |
538 | } | |
539 | } | |
540 | else { | |
541 | STRLEN len; | |
c80e42f3 | 542 | uv = utf8n_to_uvchr(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb | 543 | if (uv != puv) { |
0376ff32 | 544 | Move(s, d, len, U8); |
45005bfb JH |
545 | d += len; |
546 | puv = uv; | |
547 | } | |
548 | s += len; | |
4757a243 | 549 | } |
4757a243 LW |
550 | continue; |
551 | } | |
01ec43d0 GS |
552 | matches++; /* "none+1" is delete character */ |
553 | s += UTF8SKIP(s); | |
a0ed51b3 | 554 | } |
79072805 LW |
555 | } |
556 | else { | |
4757a243 | 557 | while (s < send) { |
f54cb97a | 558 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb | 559 | if (d > dend) { |
f54cb97a AL |
560 | const STRLEN clen = d - dstart; |
561 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 562 | if (!grows) |
7d85a32c | 563 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 564 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
565 | d = dstart + clen; |
566 | dend = dstart + nlen; | |
a0ed51b3 | 567 | } |
4757a243 LW |
568 | if (uv < none) { |
569 | matches++; | |
01ec43d0 | 570 | s += UTF8SKIP(s); |
c80e42f3 | 571 | d = uvchr_to_utf8(d, uv); |
4757a243 | 572 | continue; |
a0ed51b3 | 573 | } |
4757a243 | 574 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a | 575 | const int i = UTF8SKIP(s); |
0376ff32 | 576 | Move(s, d, i, U8); |
9b877dbb IH |
577 | d += i; |
578 | s += i; | |
4757a243 | 579 | continue; |
79072805 | 580 | } |
4757a243 LW |
581 | else if (uv == extra && !del) { |
582 | matches++; | |
01ec43d0 | 583 | s += UTF8SKIP(s); |
c80e42f3 | 584 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
585 | continue; |
586 | } | |
01ec43d0 GS |
587 | matches++; /* "none+1" is delete character */ |
588 | s += UTF8SKIP(s); | |
79072805 | 589 | } |
4757a243 | 590 | } |
16ec844d | 591 | if (grows || hibit) { |
9b877dbb IH |
592 | sv_setpvn(sv, (char*)dstart, d - dstart); |
593 | Safefree(dstart); | |
16ec844d IH |
594 | if (grows && hibit) |
595 | Safefree(start); | |
9b877dbb IH |
596 | } |
597 | else { | |
598 | *d = '\0'; | |
599 | SvCUR_set(sv, d - dstart); | |
600 | } | |
1aa99e6b | 601 | SvUTF8_on(sv); |
4757a243 LW |
602 | SvSETMAGIC(sv); |
603 | ||
604 | return matches; | |
605 | } | |
606 | ||
334c6444 DM |
607 | |
608 | /* Execute a tr//. sv is the value to be translated, while PL_op | |
609 | * should be an OP_TRANS or OP_TRANSR op, whose op_pv field contains a | |
610 | * translation table or whose op_sv field contains a swash. | |
611 | * Returns a count of number of characters translated | |
612 | */ | |
613 | ||
f0fd0980 | 614 | Size_t |
864dbfa3 | 615 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
616 | { |
617 | STRLEN len; | |
e4ee4c1b DM |
618 | const U8 flags = PL_op->op_private; |
619 | const U8 hasutf = flags & (OPpTRANS_FROM_UTF | OPpTRANS_TO_UTF); | |
4757a243 | 620 | |
7918f24d NC |
621 | PERL_ARGS_ASSERT_DO_TRANS; |
622 | ||
a53bfdae HS |
623 | if (SvREADONLY(sv) && !(flags & OPpTRANS_IDENTICAL)) { |
624 | Perl_croak_no_modify(); | |
2233f375 | 625 | } |
10516c54 | 626 | (void)SvPV_const(sv, len); |
4757a243 LW |
627 | if (!len) |
628 | return 0; | |
a53bfdae | 629 | if (!(flags & OPpTRANS_IDENTICAL)) { |
4499db73 | 630 | if (!SvPOKp(sv) || SvTHINKFIRST(sv)) |
b4cc4d79 | 631 | (void)SvPV_force_nomg(sv, len); |
2de7b02f | 632 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 633 | } |
4757a243 | 634 | |
a53bfdae HS |
635 | /* If we use only OPpTRANS_IDENTICAL to bypass the READONLY check, |
636 | * we must also rely on it to choose the readonly strategy. | |
637 | */ | |
638 | if (flags & OPpTRANS_IDENTICAL) { | |
639 | return hasutf ? do_trans_count_utf8(sv) : do_trans_count(sv); | |
640 | } else if (flags & (OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
641 | return hasutf ? do_trans_complex_utf8(sv) : do_trans_complex(sv); | |
642 | } else { | |
643 | return hasutf ? do_trans_simple_utf8(sv) : do_trans_simple(sv); | |
79072805 | 644 | } |
79072805 LW |
645 | } |
646 | ||
647 | void | |
5aaab254 | 648 | Perl_do_join(pTHX_ SV *sv, SV *delim, SV **mark, SV **sp) |
79072805 | 649 | { |
53c1dcc0 | 650 | SV ** const oldmark = mark; |
eb578fdb KW |
651 | I32 items = sp - mark; |
652 | STRLEN len; | |
463ee0b2 | 653 | STRLEN delimlen; |
810a07df | 654 | const char * const delims = SvPV_const(delim, delimlen); |
79072805 | 655 | |
7918f24d NC |
656 | PERL_ARGS_ASSERT_DO_JOIN; |
657 | ||
79072805 LW |
658 | mark++; |
659 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
862a34c6 | 660 | SvUPGRADE(sv, SVt_PV); |
79072805 LW |
661 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
662 | while (items-- > 0) { | |
1426bbf4 | 663 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a | 664 | STRLEN tmplen; |
5c144d81 | 665 | SvPV_const(*mark, tmplen); |
463ee0b2 | 666 | len += tmplen; |
79072805 LW |
667 | } |
668 | mark++; | |
669 | } | |
670 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
671 | ||
672 | mark = oldmark; | |
db7c17d7 | 673 | items = sp - mark; |
79072805 LW |
674 | ++mark; |
675 | } | |
676 | ||
f607343f | 677 | SvPVCLEAR(sv); |
e4803c42 | 678 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 679 | SvUTF8_off(sv); |
e4803c42 | 680 | |
284167a5 | 681 | if (TAINTING_get && SvMAGICAL(sv)) |
8d6d96c1 HS |
682 | SvTAINTED_off(sv); |
683 | ||
463ee0b2 | 684 | if (items-- > 0) { |
92d29cee JH |
685 | if (*mark) |
686 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
687 | mark++; |
688 | } | |
8d6d96c1 | 689 | |
c512ce4f | 690 | if (delimlen) { |
810a07df | 691 | const U32 delimflag = DO_UTF8(delim) ? SV_CATUTF8 : SV_CATBYTES; |
79072805 | 692 | for (; items > 0; items--,mark++) { |
f0ee3863 FC |
693 | STRLEN len; |
694 | const char *s; | |
810a07df | 695 | sv_catpvn_flags(sv,delims,delimlen,delimflag); |
f0ee3863 FC |
696 | s = SvPV_const(*mark,len); |
697 | sv_catpvn_flags(sv,s,len, | |
698 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
79072805 LW |
699 | } |
700 | } | |
701 | else { | |
702 | for (; items > 0; items--,mark++) | |
f0ee3863 FC |
703 | { |
704 | STRLEN len; | |
705 | const char *s = SvPV_const(*mark,len); | |
706 | sv_catpvn_flags(sv,s,len, | |
707 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
708 | } | |
79072805 LW |
709 | } |
710 | SvSETMAGIC(sv); | |
711 | } | |
712 | ||
713 | void | |
03a22d83 | 714 | Perl_do_sprintf(pTHX_ SV *sv, SSize_t len, SV **sarg) |
79072805 | 715 | { |
46fc3d4c | 716 | STRLEN patlen; |
53c1dcc0 | 717 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c | 718 | bool do_taint = FALSE; |
719 | ||
7918f24d | 720 | PERL_ARGS_ASSERT_DO_SPRINTF; |
03a22d83 | 721 | assert(len >= 1); |
7918f24d | 722 | |
e06d98fb DM |
723 | if (SvTAINTED(*sarg)) |
724 | TAINT_PROPER( | |
725 | (PL_op && PL_op->op_type < OP_max) | |
726 | ? (PL_op->op_type == OP_PRTF) | |
727 | ? "printf" | |
728 | : PL_op_name[PL_op->op_type] | |
729 | : "(unknown)" | |
730 | ); | |
5b781b5b | 731 | SvUTF8_off(sv); |
2cf2cfc6 A |
732 | if (DO_UTF8(*sarg)) |
733 | SvUTF8_on(sv); | |
03a22d83 | 734 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, (Size_t)(len - 1), &do_taint); |
79072805 | 735 | SvSETMAGIC(sv); |
46fc3d4c | 736 | if (do_taint) |
737 | SvTAINTED_on(sv); | |
79072805 LW |
738 | } |
739 | ||
33b45480 | 740 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 | 741 | UV |
d69c4304 | 742 | Perl_do_vecget(pTHX_ SV *sv, STRLEN offset, int size) |
81e118e0 | 743 | { |
67dd6f35 | 744 | STRLEN srclen, len, avail, uoffset, bitoffs = 0; |
fc9668ae DM |
745 | const I32 svpv_flags = ((PL_op->op_flags & OPf_MOD || LVRET) |
746 | ? SV_UNDEF_RETURNS_NULL : 0); | |
747 | unsigned char *s = (unsigned char *) | |
748 | SvPV_flags(sv, srclen, (svpv_flags|SV_GMAGIC)); | |
81e118e0 JH |
749 | UV retnum = 0; |
750 | ||
032061d2 | 751 | if (!s) { |
fc9668ae | 752 | s = (unsigned char *)""; |
032061d2 BF |
753 | } |
754 | ||
7918f24d NC |
755 | PERL_ARGS_ASSERT_DO_VECGET; |
756 | ||
8e84507e | 757 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 758 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 759 | |
fc9668ae | 760 | if (SvUTF8(sv)) { |
315f3fc1 KW |
761 | if (Perl_sv_utf8_downgrade(aTHX_ sv, TRUE)) { |
762 | /* PVX may have changed */ | |
763 | s = (unsigned char *) SvPV_flags(sv, srclen, svpv_flags); | |
764 | } | |
765 | else { | |
766 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), | |
767 | "Use of strings with code points over 0xFF as" | |
768 | " arguments to vec is deprecated. This will" | |
769 | " be a fatal error in Perl 5.32"); | |
770 | } | |
fc9668ae | 771 | } |
246fae53 | 772 | |
bbb8a7e0 MHM |
773 | if (size < 8) { |
774 | bitoffs = ((offset%8)*size)%8; | |
775 | uoffset = offset/(8/size); | |
776 | } | |
67dd6f35 DM |
777 | else if (size > 8) { |
778 | int n = size/8; | |
779 | if (offset > Size_t_MAX / n - 1) /* would overflow */ | |
780 | return 0; | |
781 | uoffset = offset*n; | |
782 | } | |
bbb8a7e0 MHM |
783 | else |
784 | uoffset = offset; | |
785 | ||
67dd6f35 DM |
786 | if (uoffset >= srclen) |
787 | return 0; | |
788 | ||
789 | len = (bitoffs + size + 7)/8; /* required number of bytes */ | |
790 | avail = srclen - uoffset; /* available number of bytes */ | |
791 | ||
792 | /* Does the byte range overlap the end of the string? If so, | |
793 | * handle specially. */ | |
794 | if (avail < len) { | |
81e118e0 JH |
795 | if (size <= 8) |
796 | retnum = 0; | |
797 | else { | |
81e118e0 | 798 | if (size == 16) { |
67dd6f35 DM |
799 | assert(avail == 1); |
800 | retnum = (UV) s[uoffset] << 8; | |
81e118e0 JH |
801 | } |
802 | else if (size == 32) { | |
67dd6f35 DM |
803 | assert(avail >= 1 && avail <= 3); |
804 | if (avail == 1) | |
81e118e0 | 805 | retnum = |
bb7a0f54 | 806 | ((UV) s[uoffset ] << 24); |
67dd6f35 | 807 | else if (avail == 2) |
81e118e0 | 808 | retnum = |
bb7a0f54 MHM |
809 | ((UV) s[uoffset ] << 24) + |
810 | ((UV) s[uoffset + 1] << 16); | |
81e118e0 JH |
811 | else |
812 | retnum = | |
bb7a0f54 MHM |
813 | ((UV) s[uoffset ] << 24) + |
814 | ((UV) s[uoffset + 1] << 16) + | |
815 | ( s[uoffset + 2] << 8); | |
81e118e0 | 816 | } |
d7d93a81 | 817 | #ifdef UV_IS_QUAD |
c5a0f51a | 818 | else if (size == 64) { |
a2a5de95 NC |
819 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
820 | "Bit vector size > 32 non-portable"); | |
67dd6f35 DM |
821 | assert(avail >= 1 && avail <= 7); |
822 | if (avail == 1) | |
c5a0f51a | 823 | retnum = |
bb7a0f54 | 824 | (UV) s[uoffset ] << 56; |
67dd6f35 | 825 | else if (avail == 2) |
c5a0f51a | 826 | retnum = |
bb7a0f54 MHM |
827 | ((UV) s[uoffset ] << 56) + |
828 | ((UV) s[uoffset + 1] << 48); | |
67dd6f35 | 829 | else if (avail == 3) |
c5a0f51a | 830 | retnum = |
bb7a0f54 MHM |
831 | ((UV) s[uoffset ] << 56) + |
832 | ((UV) s[uoffset + 1] << 48) + | |
833 | ((UV) s[uoffset + 2] << 40); | |
67dd6f35 | 834 | else if (avail == 4) |
c5a0f51a | 835 | retnum = |
bb7a0f54 MHM |
836 | ((UV) s[uoffset ] << 56) + |
837 | ((UV) s[uoffset + 1] << 48) + | |
838 | ((UV) s[uoffset + 2] << 40) + | |
839 | ((UV) s[uoffset + 3] << 32); | |
67dd6f35 | 840 | else if (avail == 5) |
c5a0f51a | 841 | retnum = |
bb7a0f54 MHM |
842 | ((UV) s[uoffset ] << 56) + |
843 | ((UV) s[uoffset + 1] << 48) + | |
844 | ((UV) s[uoffset + 2] << 40) + | |
845 | ((UV) s[uoffset + 3] << 32) + | |
e7aca353 | 846 | ((UV) s[uoffset + 4] << 24); |
67dd6f35 | 847 | else if (avail == 6) |
c5a0f51a | 848 | retnum = |
bb7a0f54 MHM |
849 | ((UV) s[uoffset ] << 56) + |
850 | ((UV) s[uoffset + 1] << 48) + | |
851 | ((UV) s[uoffset + 2] << 40) + | |
852 | ((UV) s[uoffset + 3] << 32) + | |
853 | ((UV) s[uoffset + 4] << 24) + | |
854 | ((UV) s[uoffset + 5] << 16); | |
c5a0f51a | 855 | else |
8e84507e | 856 | retnum = |
bb7a0f54 MHM |
857 | ((UV) s[uoffset ] << 56) + |
858 | ((UV) s[uoffset + 1] << 48) + | |
859 | ((UV) s[uoffset + 2] << 40) + | |
860 | ((UV) s[uoffset + 3] << 32) + | |
861 | ((UV) s[uoffset + 4] << 24) + | |
862 | ((UV) s[uoffset + 5] << 16) + | |
e7aca353 | 863 | ((UV) s[uoffset + 6] << 8); |
c5a0f51a JH |
864 | } |
865 | #endif | |
81e118e0 JH |
866 | } |
867 | } | |
868 | else if (size < 8) | |
bbb8a7e0 | 869 | retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1); |
81e118e0 | 870 | else { |
81e118e0 | 871 | if (size == 8) |
bb7a0f54 | 872 | retnum = s[uoffset]; |
81e118e0 JH |
873 | else if (size == 16) |
874 | retnum = | |
bb7a0f54 MHM |
875 | ((UV) s[uoffset] << 8) + |
876 | s[uoffset + 1]; | |
81e118e0 JH |
877 | else if (size == 32) |
878 | retnum = | |
bb7a0f54 MHM |
879 | ((UV) s[uoffset ] << 24) + |
880 | ((UV) s[uoffset + 1] << 16) + | |
881 | ( s[uoffset + 2] << 8) + | |
882 | s[uoffset + 3]; | |
d7d93a81 | 883 | #ifdef UV_IS_QUAD |
c5a0f51a | 884 | else if (size == 64) { |
a2a5de95 NC |
885 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
886 | "Bit vector size > 32 non-portable"); | |
c5a0f51a | 887 | retnum = |
bb7a0f54 MHM |
888 | ((UV) s[uoffset ] << 56) + |
889 | ((UV) s[uoffset + 1] << 48) + | |
890 | ((UV) s[uoffset + 2] << 40) + | |
891 | ((UV) s[uoffset + 3] << 32) + | |
892 | ((UV) s[uoffset + 4] << 24) + | |
893 | ((UV) s[uoffset + 5] << 16) + | |
894 | ( s[uoffset + 6] << 8) + | |
895 | s[uoffset + 7]; | |
c5a0f51a JH |
896 | } |
897 | #endif | |
81e118e0 JH |
898 | } |
899 | ||
900 | return retnum; | |
901 | } | |
902 | ||
33b45480 SB |
903 | /* currently converts input to bytes if possible but doesn't sweat failures, |
904 | * although it does ensure that the string it clobbers is not marked as | |
905 | * utf8-valid any more | |
906 | */ | |
79072805 | 907 | void |
864dbfa3 | 908 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 | 909 | { |
67dd6f35 | 910 | STRLEN offset, bitoffs = 0; |
eb578fdb KW |
911 | int size; |
912 | unsigned char *s; | |
913 | UV lval; | |
79072805 | 914 | I32 mask; |
a0d0e21e LW |
915 | STRLEN targlen; |
916 | STRLEN len; | |
c4420975 | 917 | SV * const targ = LvTARG(sv); |
1b92e694 | 918 | char errflags = LvFLAGS(sv); |
79072805 | 919 | |
7918f24d NC |
920 | PERL_ARGS_ASSERT_DO_VECSET; |
921 | ||
1b92e694 DM |
922 | /* some out-of-range errors have been deferred if/until the LV is |
923 | * actually written to: f(vec($s,-1,8)) is not always fatal */ | |
924 | if (errflags) { | |
b063b0a8 DIM |
925 | assert(!(errflags & ~(LVf_NEG_OFF|LVf_OUT_OF_RANGE))); |
926 | if (errflags & LVf_NEG_OFF) | |
1b92e694 DM |
927 | Perl_croak_nocontext("Negative offset to vec in lvalue context"); |
928 | Perl_croak_nocontext("Out of memory!"); | |
929 | } | |
930 | ||
8990e307 LW |
931 | if (!targ) |
932 | return; | |
032061d2 BF |
933 | s = (unsigned char*)SvPV_force_flags(targ, targlen, |
934 | SV_GMAGIC | SV_UNDEF_RETURNS_NULL); | |
246fae53 | 935 | if (SvUTF8(targ)) { |
33b45480 SB |
936 | /* This is handled by the SvPOK_only below... |
937 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
938 | SvUTF8_off(targ); | |
939 | */ | |
940 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
941 | } |
942 | ||
4ebbc975 | 943 | (void)SvPOK_only(targ); |
81e118e0 | 944 | lval = SvUV(sv); |
79072805 LW |
945 | offset = LvTARGOFF(sv); |
946 | size = LvTARGLEN(sv); | |
67dd6f35 | 947 | |
8e84507e | 948 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 949 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 950 | |
bbb8a7e0 MHM |
951 | if (size < 8) { |
952 | bitoffs = ((offset%8)*size)%8; | |
953 | offset /= 8/size; | |
954 | } | |
67dd6f35 DM |
955 | else if (size > 8) { |
956 | int n = size/8; | |
957 | if (offset > Size_t_MAX / n - 1) /* would overflow */ | |
958 | Perl_croak_nocontext("Out of memory!"); | |
959 | offset *= n; | |
960 | } | |
bbb8a7e0 | 961 | |
67dd6f35 DM |
962 | len = (bitoffs + size + 7)/8; /* required number of bytes */ |
963 | if (targlen < offset || targlen - offset < len) { | |
964 | STRLEN newlen = offset > Size_t_MAX - len - 1 ? /* avoid overflow */ | |
965 | Size_t_MAX : offset + len + 1; | |
966 | s = (unsigned char*)SvGROW(targ, newlen); | |
967 | (void)memzero((char *)(s + targlen), newlen - targlen); | |
968 | SvCUR_set(targ, newlen - 1); | |
a0d0e21e | 969 | } |
8e84507e | 970 | |
79072805 LW |
971 | if (size < 8) { |
972 | mask = (1 << size) - 1; | |
79072805 | 973 | lval &= mask; |
bbb8a7e0 MHM |
974 | s[offset] &= ~(mask << bitoffs); |
975 | s[offset] |= lval << bitoffs; | |
79072805 LW |
976 | } |
977 | else { | |
978 | if (size == 8) | |
eb160463 | 979 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 980 | else if (size == 16) { |
eb160463 GS |
981 | s[offset ] = (U8)((lval >> 8) & 0xff); |
982 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
983 | } |
984 | else if (size == 32) { | |
eb160463 GS |
985 | s[offset ] = (U8)((lval >> 24) & 0xff); |
986 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
987 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
988 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 989 | } |
d7d93a81 | 990 | #ifdef UV_IS_QUAD |
c5a0f51a | 991 | else if (size == 64) { |
a2a5de95 NC |
992 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
993 | "Bit vector size > 32 non-portable"); | |
eb160463 GS |
994 | s[offset ] = (U8)((lval >> 56) & 0xff); |
995 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
996 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
997 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
998 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
999 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
1000 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
1001 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 1002 | } |
dc1e3f56 | 1003 | #endif |
79072805 | 1004 | } |
7bb043c3 | 1005 | SvSETMAGIC(targ); |
79072805 LW |
1006 | } |
1007 | ||
1008 | void | |
864dbfa3 | 1009 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
1010 | { |
1011 | #ifdef LIBERAL | |
eb578fdb KW |
1012 | long *dl; |
1013 | long *ll; | |
1014 | long *rl; | |
79072805 | 1015 | #endif |
eb578fdb | 1016 | char *dc; |
463ee0b2 LW |
1017 | STRLEN leftlen; |
1018 | STRLEN rightlen; | |
eb578fdb KW |
1019 | const char *lc; |
1020 | const char *rc; | |
b404a7f5 | 1021 | STRLEN len = 0; |
bb7a0f54 | 1022 | STRLEN lensave; |
e62f0680 NC |
1023 | const char *lsave; |
1024 | const char *rsave; | |
bb7a0f54 | 1025 | STRLEN needlen = 0; |
08b6664b | 1026 | bool result_needs_to_be_utf8 = FALSE; |
b50535da KW |
1027 | bool left_utf8 = FALSE; |
1028 | bool right_utf8 = FALSE; | |
1029 | U8 * left_non_downgraded = NULL; | |
1030 | U8 * right_non_downgraded = NULL; | |
1031 | Size_t left_non_downgraded_len = 0; | |
1032 | Size_t right_non_downgraded_len = 0; | |
1033 | char * non_downgraded = NULL; | |
1034 | Size_t non_downgraded_len = 0; | |
0c57e439 | 1035 | |
7918f24d | 1036 | PERL_ARGS_ASSERT_DO_VOP; |
79072805 | 1037 | |
6b349a5c | 1038 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv))) |
f607343f | 1039 | SvPVCLEAR(sv); /* avoid undef warning on |= and ^= */ |
8c8eee82 | 1040 | if (sv == left) { |
08b6664b | 1041 | lc = SvPV_force_nomg(left, leftlen); |
8c8eee82 BM |
1042 | } |
1043 | else { | |
08b6664b | 1044 | lc = SvPV_nomg_const(left, leftlen); |
8c8eee82 BM |
1045 | SvPV_force_nomg_nolen(sv); |
1046 | } | |
08b6664b | 1047 | rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 | 1048 | |
0e6e7171 | 1049 | /* This needs to come after SvPV to ensure that string overloading has |
12abf4f0 NC |
1050 | fired off. */ |
1051 | ||
08b6664b KW |
1052 | /* Create downgraded temporaries of any UTF-8 encoded operands */ |
1053 | if (DO_UTF8(left)) { | |
b50535da | 1054 | const U8 * save_lc = (U8 *) lc; |
08b6664b | 1055 | |
b50535da | 1056 | left_utf8 = TRUE; |
08b6664b KW |
1057 | result_needs_to_be_utf8 = TRUE; |
1058 | ||
b50535da KW |
1059 | left_non_downgraded_len = leftlen; |
1060 | lc = (char *) bytes_from_utf8_loc((const U8 *) lc, &leftlen, | |
1061 | &left_utf8, | |
1062 | (const U8 **) &left_non_downgraded); | |
1063 | /* Calculate the number of trailing unconvertible bytes. This quantity | |
1064 | * is the original length minus the length of the converted portion. */ | |
1065 | left_non_downgraded_len -= left_non_downgraded - save_lc; | |
1066 | SAVEFREEPV(lc); | |
12abf4f0 | 1067 | } |
08b6664b | 1068 | if (DO_UTF8(right)) { |
b50535da | 1069 | const U8 * save_rc = (U8 *) rc; |
08b6664b | 1070 | |
b50535da | 1071 | right_utf8 = TRUE; |
08b6664b KW |
1072 | result_needs_to_be_utf8 = TRUE; |
1073 | ||
b50535da KW |
1074 | right_non_downgraded_len = rightlen; |
1075 | rc = (char *) bytes_from_utf8_loc((const U8 *) rc, &rightlen, | |
1076 | &right_utf8, | |
1077 | (const U8 **) &right_non_downgraded); | |
1078 | right_non_downgraded_len -= right_non_downgraded - save_rc; | |
1079 | SAVEFREEPV(rc); | |
1080 | } | |
1081 | ||
1082 | /* We set 'len' to the length that the operation actually operates on. The | |
1083 | * dangling part of the longer operand doesn't actually participate in the | |
1084 | * operation. What happens is that we pretend that the shorter operand has | |
1085 | * been extended to the right by enough imaginary zeros to match the length | |
1086 | * of the longer one. But we know in advance the result of the operation | |
1087 | * on zeros without having to do it. In the case of '&', the result is | |
1088 | * zero, and the dangling portion is simply discarded. For '|' and '^', the | |
1089 | * result is the same as the other operand, so the dangling part is just | |
1090 | * appended to the final result, unchanged. We currently accept above-FF | |
1091 | * code points in the dangling portion, as that's how it has long worked, | |
1092 | * and code depends on it staying that way. But it is now fatal for | |
1093 | * above-FF to appear in the portion that does get operated on. Hence, any | |
1094 | * above-FF must come only in the longer operand, and only in its dangling | |
1095 | * portion. That means that at least one of the operands has to be | |
1096 | * entirely non-UTF-8, and the length of that operand has to be before the | |
1097 | * first above-FF in the other */ | |
ba52ce15 | 1098 | if (left_utf8 || right_utf8) { |
3f04a466 KW |
1099 | if (left_utf8) { |
1100 | if (right_utf8 || rightlen > leftlen) { | |
814eedc8 | 1101 | Perl_croak(aTHX_ FATAL_ABOVE_FF_MSG, PL_op_desc[optype]); |
3f04a466 KW |
1102 | } |
1103 | len = rightlen; | |
08b6664b | 1104 | } |
3f04a466 KW |
1105 | else if (right_utf8) { |
1106 | if (leftlen > rightlen) { | |
814eedc8 | 1107 | Perl_croak(aTHX_ FATAL_ABOVE_FF_MSG, PL_op_desc[optype]); |
3f04a466 KW |
1108 | } |
1109 | len = leftlen; | |
b50535da | 1110 | } |
ba52ce15 KW |
1111 | |
1112 | Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), | |
814eedc8 | 1113 | DEPRECATED_ABOVE_FF_MSG, PL_op_desc[optype]); |
ba52ce15 | 1114 | } |
b50535da | 1115 | else { /* Neither is UTF-8 */ |
78ba9007 | 1116 | len = MIN(leftlen, rightlen); |
12abf4f0 NC |
1117 | } |
1118 | ||
b50535da | 1119 | lensave = len; |
08b6664b KW |
1120 | lsave = lc; |
1121 | rsave = rc; | |
b50535da | 1122 | |
9fdd7463 JH |
1123 | SvCUR_set(sv, len); |
1124 | (void)SvPOK_only(sv); | |
08b6664b | 1125 | if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { |
2596d9fe | 1126 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 MHM |
1127 | if (SvLEN(sv) < len + 1) { |
1128 | dc = SvGROW(sv, len + 1); | |
ff68c719 | 1129 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1130 | } | |
1131 | } | |
1132 | else { | |
bb7a0f54 MHM |
1133 | needlen = optype == OP_BIT_AND |
1134 | ? len : (leftlen > rightlen ? leftlen : rightlen); | |
a02a5408 | 1135 | Newxz(dc, needlen + 1, char); |
aa0a69cb | 1136 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 | 1137 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 | 1138 | } |
0c57e439 | 1139 | |
79072805 LW |
1140 | #ifdef LIBERAL |
1141 | if (len >= sizeof(long)*4 && | |
bb7a0f54 MHM |
1142 | !((unsigned long)dc % sizeof(long)) && |
1143 | !((unsigned long)lc % sizeof(long)) && | |
1144 | !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ | |
79072805 | 1145 | { |
bb7a0f54 | 1146 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 LW |
1147 | len /= (sizeof(long)*4); |
1148 | ||
1149 | dl = (long*)dc; | |
1150 | ll = (long*)lc; | |
1151 | rl = (long*)rc; | |
1152 | ||
1153 | switch (optype) { | |
1154 | case OP_BIT_AND: | |
1155 | while (len--) { | |
1156 | *dl++ = *ll++ & *rl++; | |
1157 | *dl++ = *ll++ & *rl++; | |
1158 | *dl++ = *ll++ & *rl++; | |
1159 | *dl++ = *ll++ & *rl++; | |
1160 | } | |
1161 | break; | |
a0d0e21e | 1162 | case OP_BIT_XOR: |
79072805 LW |
1163 | while (len--) { |
1164 | *dl++ = *ll++ ^ *rl++; | |
1165 | *dl++ = *ll++ ^ *rl++; | |
1166 | *dl++ = *ll++ ^ *rl++; | |
1167 | *dl++ = *ll++ ^ *rl++; | |
1168 | } | |
1169 | break; | |
1170 | case OP_BIT_OR: | |
1171 | while (len--) { | |
1172 | *dl++ = *ll++ | *rl++; | |
1173 | *dl++ = *ll++ | *rl++; | |
1174 | *dl++ = *ll++ | *rl++; | |
1175 | *dl++ = *ll++ | *rl++; | |
1176 | } | |
1177 | } | |
1178 | ||
1179 | dc = (char*)dl; | |
1180 | lc = (char*)ll; | |
1181 | rc = (char*)rl; | |
1182 | ||
1183 | len = remainder; | |
1184 | } | |
1185 | #endif | |
27a9d47d KW |
1186 | switch (optype) { |
1187 | case OP_BIT_AND: | |
1188 | while (len--) | |
1189 | *dc++ = *lc++ & *rc++; | |
1190 | *dc = '\0'; | |
1191 | break; | |
1192 | case OP_BIT_XOR: | |
1193 | while (len--) | |
1194 | *dc++ = *lc++ ^ *rc++; | |
1195 | goto mop_up; | |
1196 | case OP_BIT_OR: | |
1197 | while (len--) | |
1198 | *dc++ = *lc++ | *rc++; | |
1199 | mop_up: | |
1200 | len = lensave; | |
1201 | if (rightlen > len) { | |
1202 | if (dc == rc) | |
1203 | SvCUR(sv) = rightlen; | |
1204 | else | |
1205 | sv_catpvn_nomg(sv, rsave + len, rightlen - len); | |
1206 | } | |
1207 | else if (leftlen > len) { | |
1208 | if (dc == lc) | |
1209 | SvCUR(sv) = leftlen; | |
1210 | else | |
1211 | sv_catpvn_nomg(sv, lsave + len, leftlen - len); | |
1212 | } | |
1213 | *SvEND(sv) = '\0'; | |
392582f8 | 1214 | |
b50535da KW |
1215 | /* If there is trailing stuff that couldn't be converted from UTF-8, it |
1216 | * is appended as-is for the ^ and | operators. This preserves | |
1217 | * backwards compatibility */ | |
1218 | if (right_non_downgraded) { | |
1219 | non_downgraded = (char *) right_non_downgraded; | |
1220 | non_downgraded_len = right_non_downgraded_len; | |
1221 | } | |
1222 | else if (left_non_downgraded) { | |
1223 | non_downgraded = (char *) left_non_downgraded; | |
1224 | non_downgraded_len = left_non_downgraded_len; | |
1225 | } | |
1226 | ||
27a9d47d KW |
1227 | break; |
1228 | } | |
08b6664b KW |
1229 | |
1230 | if (result_needs_to_be_utf8) { | |
b50535da KW |
1231 | sv_utf8_upgrade_nomg(sv); |
1232 | ||
1233 | /* Append any trailing UTF-8 as-is. */ | |
1234 | if (non_downgraded) { | |
1235 | sv_catpvn_nomg(sv, non_downgraded, non_downgraded_len); | |
1236 | } | |
79072805 | 1237 | } |
08b6664b | 1238 | |
fb73857a | 1239 | SvTAINT(sv); |
79072805 | 1240 | } |
463ee0b2 | 1241 | |
b1c05ba5 | 1242 | |
a2232057 DM |
1243 | /* Perl_do_kv() may be: |
1244 | * * called directly as the pp function for pp_keys() and pp_values(); | |
a2232057 DM |
1245 | * * It may also be called directly when the op is OP_AVHVSWITCH, to |
1246 | * implement CORE::keys(), CORE::values(). | |
1247 | * | |
1248 | * In all cases it expects an HV on the stack and returns a list of keys, | |
1249 | * values, or key-value pairs, depending on PL_op. | |
1250 | */ | |
b1c05ba5 | 1251 | |
463ee0b2 | 1252 | OP * |
cea2e8a9 | 1253 | Perl_do_kv(pTHX) |
463ee0b2 | 1254 | { |
39644a26 | 1255 | dSP; |
73ff03e8 | 1256 | HV * const keys = MUTABLE_HV(POPs); |
1c23e2bd | 1257 | const U8 gimme = GIMME_V; |
a2232057 | 1258 | |
af3b1cba | 1259 | const I32 dokeys = (PL_op->op_type == OP_KEYS) |
a2232057 | 1260 | || ( PL_op->op_type == OP_AVHVSWITCH |
94184451 DM |
1261 | && (PL_op->op_private & OPpAVHVSWITCH_MASK) |
1262 | + OP_EACH == OP_KEYS); | |
a2232057 | 1263 | |
af3b1cba | 1264 | const I32 dovalues = (PL_op->op_type == OP_VALUES) |
a2232057 | 1265 | || ( PL_op->op_type == OP_AVHVSWITCH |
94184451 DM |
1266 | && (PL_op->op_private & OPpAVHVSWITCH_MASK) |
1267 | + OP_EACH == OP_VALUES); | |
a2232057 | 1268 | |
af3b1cba | 1269 | assert( PL_op->op_type == OP_KEYS |
a2232057 DM |
1270 | || PL_op->op_type == OP_VALUES |
1271 | || PL_op->op_type == OP_AVHVSWITCH); | |
463ee0b2 | 1272 | |
4fa080db DM |
1273 | assert(!( PL_op->op_type == OP_VALUES |
1274 | && (PL_op->op_private & OPpMAYBE_LVSUB))); | |
1275 | ||
800e9ae0 | 1276 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1277 | |
54310121 | 1278 | if (gimme == G_VOID) |
aa689395 | 1279 | RETURN; |
1280 | ||
54310121 | 1281 | if (gimme == G_SCALAR) { |
78f9721b | 1282 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
2154eca7 EB |
1283 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
1284 | sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0); | |
1285 | LvTYPE(ret) = 'k'; | |
1286 | LvTARG(ret) = SvREFCNT_inc_simple(keys); | |
1287 | PUSHs(ret); | |
81714fb9 | 1288 | } |
463ee0b2 | 1289 | else { |
2154eca7 EB |
1290 | IV i; |
1291 | dTARGET; | |
1292 | ||
af3b1cba DM |
1293 | /* note that in 'scalar(keys %h)' the OP_KEYS is usually |
1294 | * optimised away and the action is performed directly by the | |
1295 | * padhv or rv2hv op. We now only get here via OP_AVHVSWITCH | |
1296 | * and \&CORE::keys | |
1297 | */ | |
2154eca7 | 1298 | if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) { |
1b95d04f | 1299 | i = HvUSEDKEYS(keys); |
2154eca7 EB |
1300 | } |
1301 | else { | |
1302 | i = 0; | |
1303 | while (hv_iternext(keys)) i++; | |
1304 | } | |
1305 | PUSHi( i ); | |
463ee0b2 | 1306 | } |
463ee0b2 LW |
1307 | RETURN; |
1308 | } | |
1309 | ||
a061ab0b FC |
1310 | if (UNLIKELY(PL_op->op_private & OPpMAYBE_LVSUB)) { |
1311 | const I32 flags = is_lvalue_sub(); | |
1312 | if (flags && !(flags & OPpENTERSUB_INARGS)) | |
1313 | /* diag_listed_as: Can't modify %s in %s */ | |
1314 | Perl_croak(aTHX_ "Can't modify keys in list assignment"); | |
1315 | } | |
1316 | ||
8dc9003f DM |
1317 | PUTBACK; |
1318 | hv_pushkv(keys, (dokeys | (dovalues << 1))); | |
1319 | return NORMAL; | |
463ee0b2 | 1320 | } |
4e35701f | 1321 | |
af3babe4 | 1322 | /* |
14d04a33 | 1323 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 1324 | */ |