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 DM |
17 | /* This file contains some common functions needed to carry out certain |
18 | * ops. For example both pp_schomp() and pp_chomp() - scalar and array | |
19 | * chomp operations - call the function do_chomp() found in this file. | |
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 | ||
942e002e | 30 | STATIC I32 |
c395bd6c | 31 | S_do_trans_simple(pTHX_ SV * const sv) |
79072805 | 32 | { |
97aff369 | 33 | dVAR; |
4757a243 | 34 | I32 matches = 0; |
463ee0b2 | 35 | STRLEN len; |
9138d6ca | 36 | U8 *s = (U8*)SvPV_nomg(sv,len); |
c395bd6c | 37 | U8 * const send = s+len; |
c4420975 | 38 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
39 | |
40 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE; | |
41 | ||
4757a243 | 42 | if (!tbl) |
7d85a32c | 43 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 44 | |
1e54db1a | 45 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 46 | if (!SvUTF8(sv)) { |
01ec43d0 | 47 | while (s < send) { |
f54cb97a AL |
48 | const I32 ch = tbl[*s]; |
49 | if (ch >= 0) { | |
01ec43d0 | 50 | matches++; |
c4420975 | 51 | *s = (U8)ch; |
01ec43d0 | 52 | } |
c4420975 | 53 | s++; |
01ec43d0 GS |
54 | } |
55 | SvSETMAGIC(sv); | |
9b877dbb IH |
56 | } |
57 | else { | |
c395bd6c AL |
58 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
59 | U8 *d; | |
60 | U8 *dstart; | |
61 | ||
62 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ | |
63 | if (grows) | |
64 | Newx(d, len*2+1, U8); | |
65 | else | |
66 | d = s; | |
67 | dstart = d; | |
68 | while (s < send) { | |
69 | STRLEN ulen; | |
70 | I32 ch; | |
71 | ||
72 | /* Need to check this, otherwise 128..255 won't match */ | |
73 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); | |
74 | if (c < 0x100 && (ch = tbl[c]) >= 0) { | |
75 | matches++; | |
76 | d = uvchr_to_utf8(d, ch); | |
77 | s += ulen; | |
78 | } | |
79 | else { /* No match -> copy */ | |
80 | Move(s, d, ulen, U8); | |
81 | d += ulen; | |
82 | s += ulen; | |
83 | } | |
84 | } | |
85 | if (grows) { | |
86 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
87 | Safefree(dstart); | |
88 | } | |
89 | else { | |
90 | *d = '\0'; | |
91 | SvCUR_set(sv, d - dstart); | |
92 | } | |
93 | SvUTF8_on(sv); | |
94 | SvSETMAGIC(sv); | |
9b877dbb | 95 | } |
4757a243 LW |
96 | return matches; |
97 | } | |
98 | ||
942e002e | 99 | STATIC I32 |
c395bd6c | 100 | S_do_trans_count(pTHX_ SV * const sv) |
4757a243 | 101 | { |
97aff369 | 102 | dVAR; |
4757a243 | 103 | STRLEN len; |
9138d6ca | 104 | const U8 *s = (const U8*)SvPV_nomg_const(sv, len); |
c395bd6c AL |
105 | const U8 * const send = s + len; |
106 | I32 matches = 0; | |
f54cb97a | 107 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
108 | |
109 | PERL_ARGS_ASSERT_DO_TRANS_COUNT; | |
110 | ||
4757a243 | 111 | if (!tbl) |
7d85a32c | 112 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 | 113 | |
c395bd6c | 114 | if (!SvUTF8(sv)) { |
1aa99e6b IH |
115 | while (s < send) { |
116 | if (tbl[*s++] >= 0) | |
036b4402 | 117 | matches++; |
1aa99e6b | 118 | } |
c395bd6c | 119 | } |
fabdb6c0 AL |
120 | else { |
121 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; | |
1aa99e6b | 122 | while (s < send) { |
1aa99e6b | 123 | STRLEN ulen; |
9f7f3913 | 124 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 JH |
125 | if (c < 0x100) { |
126 | if (tbl[c] >= 0) | |
127 | matches++; | |
128 | } else if (complement) | |
1aa99e6b IH |
129 | matches++; |
130 | s += ulen; | |
131 | } | |
fabdb6c0 | 132 | } |
4757a243 LW |
133 | |
134 | return matches; | |
135 | } | |
136 | ||
942e002e | 137 | STATIC I32 |
c395bd6c | 138 | S_do_trans_complex(pTHX_ SV * const sv) |
4757a243 | 139 | { |
97aff369 | 140 | dVAR; |
c395bd6c | 141 | STRLEN len; |
9138d6ca | 142 | U8 *s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 143 | U8 * const send = s+len; |
4757a243 | 144 | I32 matches = 0; |
f54cb97a | 145 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
146 | |
147 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX; | |
148 | ||
4757a243 | 149 | if (!tbl) |
7d85a32c | 150 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 | 151 | |
c395bd6c AL |
152 | if (!SvUTF8(sv)) { |
153 | U8 *d = s; | |
154 | U8 * const dstart = d; | |
4757a243 | 155 | |
1aa99e6b | 156 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a | 157 | const U8* p = send; |
1aa99e6b | 158 | while (s < send) { |
f54cb97a AL |
159 | const I32 ch = tbl[*s]; |
160 | if (ch >= 0) { | |
eb160463 | 161 | *d = (U8)ch; |
036b4402 | 162 | matches++; |
1aa99e6b IH |
163 | if (p != d - 1 || *p != *d) |
164 | p = d++; | |
165 | } | |
166 | else if (ch == -1) /* -1 is unmapped character */ | |
167 | *d++ = *s; | |
168 | else if (ch == -2) /* -2 is delete character */ | |
169 | matches++; | |
170 | s++; | |
171 | } | |
a0ed51b3 | 172 | } |
1aa99e6b IH |
173 | else { |
174 | while (s < send) { | |
f54cb97a AL |
175 | const I32 ch = tbl[*s]; |
176 | if (ch >= 0) { | |
1aa99e6b | 177 | matches++; |
eb160463 | 178 | *d++ = (U8)ch; |
1aa99e6b IH |
179 | } |
180 | else if (ch == -1) /* -1 is unmapped character */ | |
181 | *d++ = *s; | |
182 | else if (ch == -2) /* -2 is delete character */ | |
183 | matches++; | |
184 | s++; | |
185 | } | |
186 | } | |
76ef7183 | 187 | *d = '\0'; |
1aa99e6b | 188 | SvCUR_set(sv, d - dstart); |
4757a243 | 189 | } |
c395bd6c | 190 | else { /* is utf8 */ |
fabdb6c0 AL |
191 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
192 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; | |
193 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
c395bd6c AL |
194 | U8 *d; |
195 | U8 *dstart; | |
196 | STRLEN rlen = 0; | |
fabdb6c0 | 197 | |
9b877dbb | 198 | if (grows) |
a02a5408 | 199 | Newx(d, len*2+1, U8); |
9b877dbb IH |
200 | else |
201 | d = s; | |
1aa99e6b | 202 | dstart = d; |
8973db79 JH |
203 | if (complement && !del) |
204 | rlen = tbl[0x100]; | |
1aa99e6b IH |
205 | |
206 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
1aa99e6b IH |
207 | UV pch = 0xfeedface; |
208 | while (s < send) { | |
209 | STRLEN len; | |
9f7f3913 TS |
210 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
211 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 212 | I32 ch; |
1aa99e6b | 213 | |
8973db79 JH |
214 | if (comp > 0xff) { |
215 | if (!complement) { | |
640af7f1 | 216 | Move(s, d, len, U8); |
8973db79 JH |
217 | d += len; |
218 | } | |
219 | else { | |
220 | matches++; | |
221 | if (!del) { | |
bb7a0f54 | 222 | ch = (rlen == 0) ? (I32)comp : |
05d340b8 | 223 | (comp - 0x100 < rlen) ? |
8973db79 | 224 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 225 | if ((UV)ch != pch) { |
9041c2e3 | 226 | d = uvchr_to_utf8(d, ch); |
eb160463 | 227 | pch = (UV)ch; |
8973db79 JH |
228 | } |
229 | s += len; | |
230 | continue; | |
231 | } | |
232 | } | |
9b877dbb | 233 | } |
1aa99e6b IH |
234 | else if ((ch = tbl[comp]) >= 0) { |
235 | matches++; | |
eb160463 | 236 | if ((UV)ch != pch) { |
9041c2e3 | 237 | d = uvchr_to_utf8(d, ch); |
eb160463 | 238 | pch = (UV)ch; |
1aa99e6b IH |
239 | } |
240 | s += len; | |
241 | continue; | |
242 | } | |
9b877dbb | 243 | else if (ch == -1) { /* -1 is unmapped character */ |
640af7f1 | 244 | Move(s, d, len, U8); |
9b877dbb IH |
245 | d += len; |
246 | } | |
1aa99e6b IH |
247 | else if (ch == -2) /* -2 is delete character */ |
248 | matches++; | |
249 | s += len; | |
250 | pch = 0xfeedface; | |
251 | } | |
5d06d08e | 252 | } |
1aa99e6b IH |
253 | else { |
254 | while (s < send) { | |
255 | STRLEN len; | |
9f7f3913 TS |
256 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
257 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 258 | I32 ch; |
8973db79 JH |
259 | if (comp > 0xff) { |
260 | if (!complement) { | |
0376ff32 | 261 | Move(s, d, len, U8); |
8973db79 JH |
262 | d += len; |
263 | } | |
264 | else { | |
265 | matches++; | |
266 | if (!del) { | |
9041c2e3 NIS |
267 | if (comp - 0x100 < rlen) |
268 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 269 | else |
9041c2e3 | 270 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
271 | } |
272 | } | |
9b877dbb | 273 | } |
1aa99e6b | 274 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 275 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
276 | matches++; |
277 | } | |
278 | else if (ch == -1) { /* -1 is unmapped character */ | |
640af7f1 | 279 | Move(s, d, len, U8); |
9b877dbb | 280 | d += len; |
1aa99e6b IH |
281 | } |
282 | else if (ch == -2) /* -2 is delete character */ | |
283 | matches++; | |
284 | s += len; | |
285 | } | |
286 | } | |
9b877dbb IH |
287 | if (grows) { |
288 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
289 | Safefree(dstart); | |
290 | } | |
291 | else { | |
292 | *d = '\0'; | |
293 | SvCUR_set(sv, d - dstart); | |
294 | } | |
1aa99e6b | 295 | SvUTF8_on(sv); |
4757a243 | 296 | } |
5e44153e | 297 | SvSETMAGIC(sv); |
4757a243 LW |
298 | return matches; |
299 | } | |
300 | ||
942e002e | 301 | STATIC I32 |
c395bd6c | 302 | S_do_trans_simple_utf8(pTHX_ SV * const sv) |
4757a243 | 303 | { |
97aff369 | 304 | dVAR; |
4757a243 LW |
305 | U8 *s; |
306 | U8 *send; | |
307 | U8 *d; | |
036b4402 | 308 | U8 *start; |
1aa99e6b | 309 | U8 *dstart, *dend; |
4757a243 | 310 | I32 matches = 0; |
f54cb97a | 311 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 | 312 | STRLEN len; |
043e41b8 DM |
313 | SV* const rv = |
314 | #ifdef USE_ITHREADS | |
315 | PAD_SVl(cPADOP->op_padix); | |
316 | #else | |
ad64d0ec | 317 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 318 | #endif |
85fbaab2 | 319 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 320 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
321 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
322 | const UV extra = none + 1; | |
b7953727 | 323 | UV final = 0; |
1aa99e6b | 324 | U8 hibit = 0; |
4757a243 | 325 | |
7918f24d NC |
326 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8; |
327 | ||
9138d6ca | 328 | s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 329 | if (!SvUTF8(sv)) { |
c4420975 AL |
330 | const U8 *t = s; |
331 | const U8 * const e = s + len; | |
c4d5f83a | 332 | while (t < e) { |
f54cb97a | 333 | const U8 ch = *t++; |
fabdb6c0 AL |
334 | hibit = !NATIVE_IS_INVARIANT(ch); |
335 | if (hibit) { | |
336 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 337 | break; |
fabdb6c0 | 338 | } |
c4d5f83a | 339 | } |
1aa99e6b | 340 | } |
4757a243 | 341 | send = s + len; |
036b4402 | 342 | start = s; |
4757a243 | 343 | |
a4fc7abc | 344 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 LW |
345 | if (svp) |
346 | final = SvUV(*svp); | |
347 | ||
9b877dbb IH |
348 | if (grows) { |
349 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 350 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
351 | dend = d + len * 3; |
352 | dstart = d; | |
353 | } | |
354 | else { | |
355 | dstart = d = s; | |
356 | dend = d + len; | |
357 | } | |
1aa99e6b | 358 | |
4757a243 | 359 | while (s < send) { |
0bcc34c2 AL |
360 | const UV uv = swash_fetch(rv, s, TRUE); |
361 | if (uv < none) { | |
4757a243 LW |
362 | s += UTF8SKIP(s); |
363 | matches++; | |
7d85a32c | 364 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
365 | } |
366 | else if (uv == none) { | |
4373e329 | 367 | const int i = UTF8SKIP(s); |
0376ff32 | 368 | Move(s, d, i, U8); |
9b877dbb IH |
369 | d += i; |
370 | s += i; | |
4757a243 LW |
371 | } |
372 | else if (uv == extra) { | |
4373e329 | 373 | s += UTF8SKIP(s); |
4757a243 | 374 | matches++; |
7d85a32c | 375 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
376 | } |
377 | else | |
378 | s += UTF8SKIP(s); | |
1aa99e6b | 379 | |
9b877dbb | 380 | if (d > dend) { |
4373e329 AL |
381 | const STRLEN clen = d - dstart; |
382 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 383 | if (!grows) |
7d85a32c | 384 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 | 385 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
386 | d = dstart + clen; |
387 | dend = dstart + nlen; | |
388 | } | |
4757a243 | 389 | } |
16ec844d | 390 | if (grows || hibit) { |
9b877dbb IH |
391 | sv_setpvn(sv, (char*)dstart, d - dstart); |
392 | Safefree(dstart); | |
16ec844d IH |
393 | if (grows && hibit) |
394 | Safefree(start); | |
9b877dbb IH |
395 | } |
396 | else { | |
397 | *d = '\0'; | |
398 | SvCUR_set(sv, d - dstart); | |
399 | } | |
4757a243 | 400 | SvSETMAGIC(sv); |
1aa99e6b | 401 | SvUTF8_on(sv); |
4757a243 LW |
402 | |
403 | return matches; | |
404 | } | |
405 | ||
942e002e | 406 | STATIC I32 |
c395bd6c | 407 | S_do_trans_count_utf8(pTHX_ SV * const sv) |
4757a243 | 408 | { |
97aff369 | 409 | dVAR; |
93524f2b | 410 | const U8 *s; |
cbbf8932 AL |
411 | const U8 *start = NULL; |
412 | const U8 *send; | |
4757a243 LW |
413 | I32 matches = 0; |
414 | STRLEN len; | |
043e41b8 DM |
415 | SV* const rv = |
416 | #ifdef USE_ITHREADS | |
417 | PAD_SVl(cPADOP->op_padix); | |
418 | #else | |
ad64d0ec | 419 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 420 | #endif |
85fbaab2 | 421 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 422 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
423 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
424 | const UV extra = none + 1; | |
1aa99e6b | 425 | U8 hibit = 0; |
4757a243 | 426 | |
7918f24d NC |
427 | PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8; |
428 | ||
9138d6ca | 429 | s = (const U8*)SvPV_nomg_const(sv, len); |
1aa99e6b | 430 | if (!SvUTF8(sv)) { |
4373e329 | 431 | const U8 *t = s; |
c4420975 | 432 | const U8 * const e = s + len; |
c4d5f83a | 433 | while (t < e) { |
4373e329 | 434 | const U8 ch = *t++; |
fabdb6c0 AL |
435 | hibit = !NATIVE_IS_INVARIANT(ch); |
436 | if (hibit) { | |
437 | start = s = bytes_to_utf8(s, &len); | |
1aa99e6b | 438 | break; |
fabdb6c0 | 439 | } |
c4d5f83a | 440 | } |
1aa99e6b | 441 | } |
4757a243 LW |
442 | send = s + len; |
443 | ||
444 | while (s < send) { | |
c4420975 AL |
445 | const UV uv = swash_fetch(rv, s, TRUE); |
446 | if (uv < none || uv == extra) | |
4757a243 | 447 | matches++; |
834a4ddd | 448 | s += UTF8SKIP(s); |
4757a243 | 449 | } |
1aa99e6b IH |
450 | if (hibit) |
451 | Safefree(start); | |
4757a243 LW |
452 | |
453 | return matches; | |
454 | } | |
455 | ||
942e002e | 456 | STATIC I32 |
c395bd6c | 457 | S_do_trans_complex_utf8(pTHX_ SV * const sv) |
4757a243 | 458 | { |
97aff369 | 459 | dVAR; |
1aa99e6b | 460 | U8 *start, *send; |
4757a243 LW |
461 | U8 *d; |
462 | I32 matches = 0; | |
4373e329 AL |
463 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
464 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
465 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; | |
043e41b8 DM |
466 | SV* const rv = |
467 | #ifdef USE_ITHREADS | |
468 | PAD_SVl(cPADOP->op_padix); | |
469 | #else | |
ad64d0ec | 470 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 471 | #endif |
85fbaab2 | 472 | HV * const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 473 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
474 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
475 | const UV extra = none + 1; | |
b7953727 | 476 | UV final = 0; |
45005bfb | 477 | bool havefinal = FALSE; |
4757a243 | 478 | STRLEN len; |
1aa99e6b | 479 | U8 *dstart, *dend; |
1aa99e6b | 480 | U8 hibit = 0; |
9138d6ca | 481 | U8 *s = (U8*)SvPV_nomg(sv, len); |
7918f24d NC |
482 | |
483 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8; | |
484 | ||
c395bd6c | 485 | if (!SvUTF8(sv)) { |
53c1dcc0 AL |
486 | const U8 *t = s; |
487 | const U8 * const e = s + len; | |
c4d5f83a | 488 | while (t < e) { |
f54cb97a | 489 | const U8 ch = *t++; |
fabdb6c0 AL |
490 | hibit = !NATIVE_IS_INVARIANT(ch); |
491 | if (hibit) { | |
492 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 493 | break; |
fabdb6c0 | 494 | } |
c4d5f83a | 495 | } |
1aa99e6b | 496 | } |
4757a243 | 497 | send = s + len; |
1aa99e6b | 498 | start = s; |
4757a243 | 499 | |
a4fc7abc | 500 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb | 501 | if (svp) { |
4757a243 | 502 | final = SvUV(*svp); |
45005bfb JH |
503 | havefinal = TRUE; |
504 | } | |
4757a243 | 505 | |
9b877dbb IH |
506 | if (grows) { |
507 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 508 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
509 | dend = d + len * 3; |
510 | dstart = d; | |
511 | } | |
512 | else { | |
513 | dstart = d = s; | |
514 | dend = d + len; | |
515 | } | |
4757a243 LW |
516 | |
517 | if (squash) { | |
518 | UV puv = 0xfeedface; | |
519 | while (s < send) { | |
f54cb97a | 520 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 521 | |
9b877dbb | 522 | if (d > dend) { |
4373e329 AL |
523 | const STRLEN clen = d - dstart; |
524 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 525 | if (!grows) |
7d85a32c | 526 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 527 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
528 | d = dstart + clen; |
529 | dend = dstart + nlen; | |
4757a243 LW |
530 | } |
531 | if (uv < none) { | |
532 | matches++; | |
94472101 | 533 | s += UTF8SKIP(s); |
4757a243 | 534 | if (uv != puv) { |
7d85a32c | 535 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
536 | puv = uv; |
537 | } | |
4757a243 LW |
538 | continue; |
539 | } | |
540 | else if (uv == none) { /* "none" is unmapped character */ | |
4373e329 | 541 | const int i = UTF8SKIP(s); |
0376ff32 | 542 | Move(s, d, i, U8); |
9b877dbb IH |
543 | d += i; |
544 | s += i; | |
4757a243 LW |
545 | puv = 0xfeedface; |
546 | continue; | |
a0ed51b3 | 547 | } |
4757a243 LW |
548 | else if (uv == extra && !del) { |
549 | matches++; | |
45005bfb JH |
550 | if (havefinal) { |
551 | s += UTF8SKIP(s); | |
552 | if (puv != final) { | |
7d85a32c | 553 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
554 | puv = final; |
555 | } | |
556 | } | |
557 | else { | |
558 | STRLEN len; | |
9f7f3913 | 559 | uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb | 560 | if (uv != puv) { |
0376ff32 | 561 | Move(s, d, len, U8); |
45005bfb JH |
562 | d += len; |
563 | puv = uv; | |
564 | } | |
565 | s += len; | |
4757a243 | 566 | } |
4757a243 LW |
567 | continue; |
568 | } | |
01ec43d0 GS |
569 | matches++; /* "none+1" is delete character */ |
570 | s += UTF8SKIP(s); | |
a0ed51b3 | 571 | } |
79072805 LW |
572 | } |
573 | else { | |
4757a243 | 574 | while (s < send) { |
f54cb97a | 575 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb | 576 | if (d > dend) { |
f54cb97a AL |
577 | const STRLEN clen = d - dstart; |
578 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 579 | if (!grows) |
7d85a32c | 580 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 581 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
582 | d = dstart + clen; |
583 | dend = dstart + nlen; | |
a0ed51b3 | 584 | } |
4757a243 LW |
585 | if (uv < none) { |
586 | matches++; | |
01ec43d0 | 587 | s += UTF8SKIP(s); |
7d85a32c | 588 | d = uvuni_to_utf8(d, uv); |
4757a243 | 589 | continue; |
a0ed51b3 | 590 | } |
4757a243 | 591 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a | 592 | const int i = UTF8SKIP(s); |
0376ff32 | 593 | Move(s, d, i, U8); |
9b877dbb IH |
594 | d += i; |
595 | s += i; | |
4757a243 | 596 | continue; |
79072805 | 597 | } |
4757a243 LW |
598 | else if (uv == extra && !del) { |
599 | matches++; | |
01ec43d0 | 600 | s += UTF8SKIP(s); |
7d85a32c | 601 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
602 | continue; |
603 | } | |
01ec43d0 GS |
604 | matches++; /* "none+1" is delete character */ |
605 | s += UTF8SKIP(s); | |
79072805 | 606 | } |
4757a243 | 607 | } |
16ec844d | 608 | if (grows || hibit) { |
9b877dbb IH |
609 | sv_setpvn(sv, (char*)dstart, d - dstart); |
610 | Safefree(dstart); | |
16ec844d IH |
611 | if (grows && hibit) |
612 | Safefree(start); | |
9b877dbb IH |
613 | } |
614 | else { | |
615 | *d = '\0'; | |
616 | SvCUR_set(sv, d - dstart); | |
617 | } | |
1aa99e6b | 618 | SvUTF8_on(sv); |
4757a243 LW |
619 | SvSETMAGIC(sv); |
620 | ||
621 | return matches; | |
622 | } | |
623 | ||
624 | I32 | |
864dbfa3 | 625 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 | 626 | { |
97aff369 | 627 | dVAR; |
4757a243 | 628 | STRLEN len; |
4373e329 | 629 | const I32 hasutf = (PL_op->op_private & |
036b4402 | 630 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 631 | |
7918f24d NC |
632 | PERL_ARGS_ASSERT_DO_TRANS; |
633 | ||
3788ef8f | 634 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
deaf58b8 | 635 | if (!SvIsCOW(sv)) |
6ad8f254 | 636 | Perl_croak_no_modify(aTHX); |
2233f375 | 637 | } |
10516c54 | 638 | (void)SvPV_const(sv, len); |
4757a243 LW |
639 | if (!len) |
640 | return 0; | |
d59e14db | 641 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
4499db73 | 642 | if (!SvPOKp(sv) || SvTHINKFIRST(sv)) |
b4cc4d79 | 643 | (void)SvPV_force_nomg(sv, len); |
2de7b02f | 644 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 645 | } |
4757a243 | 646 | |
cea2e8a9 | 647 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 648 | |
a77e643a RGS |
649 | switch (PL_op->op_private & ~hasutf & ( |
650 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| | |
651 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
4757a243 | 652 | case 0: |
01ec43d0 GS |
653 | if (hasutf) |
654 | return do_trans_simple_utf8(sv); | |
655 | else | |
656 | return do_trans_simple(sv); | |
4757a243 LW |
657 | |
658 | case OPpTRANS_IDENTICAL: | |
05d340b8 | 659 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 GS |
660 | if (hasutf) |
661 | return do_trans_count_utf8(sv); | |
662 | else | |
663 | return do_trans_count(sv); | |
4757a243 LW |
664 | |
665 | default: | |
01ec43d0 | 666 | if (hasutf) |
036b4402 | 667 | return do_trans_complex_utf8(sv); |
4757a243 | 668 | else |
036b4402 | 669 | return do_trans_complex(sv); |
79072805 | 670 | } |
79072805 LW |
671 | } |
672 | ||
673 | void | |
b9ac451d | 674 | Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp) |
79072805 | 675 | { |
97aff369 | 676 | dVAR; |
53c1dcc0 | 677 | SV ** const oldmark = mark; |
eb578fdb KW |
678 | I32 items = sp - mark; |
679 | STRLEN len; | |
463ee0b2 | 680 | STRLEN delimlen; |
79072805 | 681 | |
7918f24d NC |
682 | PERL_ARGS_ASSERT_DO_JOIN; |
683 | ||
b9ac451d | 684 | (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */ |
5ba99574 | 685 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 | 686 | |
79072805 LW |
687 | mark++; |
688 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
862a34c6 | 689 | SvUPGRADE(sv, SVt_PV); |
79072805 LW |
690 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
691 | while (items-- > 0) { | |
1426bbf4 | 692 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a | 693 | STRLEN tmplen; |
5c144d81 | 694 | SvPV_const(*mark, tmplen); |
463ee0b2 | 695 | len += tmplen; |
79072805 LW |
696 | } |
697 | mark++; | |
698 | } | |
699 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
700 | ||
701 | mark = oldmark; | |
db7c17d7 | 702 | items = sp - mark; |
79072805 LW |
703 | ++mark; |
704 | } | |
705 | ||
76f68e9b | 706 | sv_setpvs(sv, ""); |
e4803c42 | 707 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 708 | SvUTF8_off(sv); |
e4803c42 | 709 | |
8d6d96c1 HS |
710 | if (PL_tainting && SvMAGICAL(sv)) |
711 | SvTAINTED_off(sv); | |
712 | ||
463ee0b2 | 713 | if (items-- > 0) { |
92d29cee JH |
714 | if (*mark) |
715 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
716 | mark++; |
717 | } | |
8d6d96c1 | 718 | |
c512ce4f | 719 | if (delimlen) { |
79072805 | 720 | for (; items > 0; items--,mark++) { |
760209f8 | 721 | sv_catsv_nomg(sv,delim); |
79072805 LW |
722 | sv_catsv(sv,*mark); |
723 | } | |
724 | } | |
725 | else { | |
726 | for (; items > 0; items--,mark++) | |
727 | sv_catsv(sv,*mark); | |
728 | } | |
729 | SvSETMAGIC(sv); | |
730 | } | |
731 | ||
732 | void | |
864dbfa3 | 733 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 734 | { |
97aff369 | 735 | dVAR; |
46fc3d4c | 736 | STRLEN patlen; |
53c1dcc0 | 737 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c | 738 | bool do_taint = FALSE; |
739 | ||
7918f24d NC |
740 | PERL_ARGS_ASSERT_DO_SPRINTF; |
741 | ||
e06d98fb DM |
742 | if (SvTAINTED(*sarg)) |
743 | TAINT_PROPER( | |
744 | (PL_op && PL_op->op_type < OP_max) | |
745 | ? (PL_op->op_type == OP_PRTF) | |
746 | ? "printf" | |
747 | : PL_op_name[PL_op->op_type] | |
748 | : "(unknown)" | |
749 | ); | |
5b781b5b | 750 | SvUTF8_off(sv); |
2cf2cfc6 A |
751 | if (DO_UTF8(*sarg)) |
752 | SvUTF8_on(sv); | |
4608196e | 753 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint); |
79072805 | 754 | SvSETMAGIC(sv); |
46fc3d4c | 755 | if (do_taint) |
756 | SvTAINTED_on(sv); | |
79072805 LW |
757 | } |
758 | ||
33b45480 | 759 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 | 760 | UV |
fcb8da00 | 761 | Perl_do_vecget(pTHX_ SV *sv, SSize_t offset, int size) |
81e118e0 | 762 | { |
97aff369 | 763 | dVAR; |
bbb8a7e0 | 764 | STRLEN srclen, len, uoffset, bitoffs = 0; |
032061d2 BF |
765 | const unsigned char *s = (const unsigned char *) SvPV_flags_const(sv, srclen, |
766 | SV_GMAGIC | ((PL_op->op_flags & OPf_MOD || LVRET) | |
767 | ? SV_UNDEF_RETURNS_NULL : 0)); | |
81e118e0 JH |
768 | UV retnum = 0; |
769 | ||
032061d2 BF |
770 | if (!s) { |
771 | s = (const unsigned char *)""; | |
772 | } | |
773 | ||
7918f24d NC |
774 | PERL_ARGS_ASSERT_DO_VECGET; |
775 | ||
a50d7633 | 776 | if (offset < 0) |
61f9802b | 777 | return 0; |
8e84507e | 778 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 779 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 780 | |
dcad2880 | 781 | if (SvUTF8(sv)) |
33b45480 | 782 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 783 | |
bbb8a7e0 MHM |
784 | if (size < 8) { |
785 | bitoffs = ((offset%8)*size)%8; | |
786 | uoffset = offset/(8/size); | |
787 | } | |
788 | else if (size > 8) | |
789 | uoffset = offset*(size/8); | |
790 | else | |
791 | uoffset = offset; | |
792 | ||
793 | len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
81e118e0 JH |
794 | if (len > srclen) { |
795 | if (size <= 8) | |
796 | retnum = 0; | |
797 | else { | |
81e118e0 | 798 | if (size == 16) { |
bb7a0f54 | 799 | if (uoffset >= srclen) |
81e118e0 JH |
800 | retnum = 0; |
801 | else | |
bb7a0f54 | 802 | retnum = (UV) s[uoffset] << 8; |
81e118e0 JH |
803 | } |
804 | else if (size == 32) { | |
bb7a0f54 | 805 | if (uoffset >= srclen) |
81e118e0 | 806 | retnum = 0; |
bb7a0f54 | 807 | else if (uoffset + 1 >= srclen) |
81e118e0 | 808 | retnum = |
bb7a0f54 MHM |
809 | ((UV) s[uoffset ] << 24); |
810 | else if (uoffset + 2 >= srclen) | |
81e118e0 | 811 | retnum = |
bb7a0f54 MHM |
812 | ((UV) s[uoffset ] << 24) + |
813 | ((UV) s[uoffset + 1] << 16); | |
81e118e0 JH |
814 | else |
815 | retnum = | |
bb7a0f54 MHM |
816 | ((UV) s[uoffset ] << 24) + |
817 | ((UV) s[uoffset + 1] << 16) + | |
818 | ( s[uoffset + 2] << 8); | |
81e118e0 | 819 | } |
d7d93a81 | 820 | #ifdef UV_IS_QUAD |
c5a0f51a | 821 | else if (size == 64) { |
a2a5de95 NC |
822 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
823 | "Bit vector size > 32 non-portable"); | |
bb7a0f54 | 824 | if (uoffset >= srclen) |
c5a0f51a | 825 | retnum = 0; |
bb7a0f54 | 826 | else if (uoffset + 1 >= srclen) |
c5a0f51a | 827 | retnum = |
bb7a0f54 MHM |
828 | (UV) s[uoffset ] << 56; |
829 | else if (uoffset + 2 >= srclen) | |
c5a0f51a | 830 | retnum = |
bb7a0f54 MHM |
831 | ((UV) s[uoffset ] << 56) + |
832 | ((UV) s[uoffset + 1] << 48); | |
833 | else if (uoffset + 3 >= srclen) | |
c5a0f51a | 834 | retnum = |
bb7a0f54 MHM |
835 | ((UV) s[uoffset ] << 56) + |
836 | ((UV) s[uoffset + 1] << 48) + | |
837 | ((UV) s[uoffset + 2] << 40); | |
838 | else if (uoffset + 4 >= srclen) | |
c5a0f51a | 839 | retnum = |
bb7a0f54 MHM |
840 | ((UV) s[uoffset ] << 56) + |
841 | ((UV) s[uoffset + 1] << 48) + | |
842 | ((UV) s[uoffset + 2] << 40) + | |
843 | ((UV) s[uoffset + 3] << 32); | |
844 | else if (uoffset + 5 >= srclen) | |
c5a0f51a | 845 | retnum = |
bb7a0f54 MHM |
846 | ((UV) s[uoffset ] << 56) + |
847 | ((UV) s[uoffset + 1] << 48) + | |
848 | ((UV) s[uoffset + 2] << 40) + | |
849 | ((UV) s[uoffset + 3] << 32) + | |
850 | ( s[uoffset + 4] << 24); | |
851 | else if (uoffset + 6 >= srclen) | |
c5a0f51a | 852 | retnum = |
bb7a0f54 MHM |
853 | ((UV) s[uoffset ] << 56) + |
854 | ((UV) s[uoffset + 1] << 48) + | |
855 | ((UV) s[uoffset + 2] << 40) + | |
856 | ((UV) s[uoffset + 3] << 32) + | |
857 | ((UV) s[uoffset + 4] << 24) + | |
858 | ((UV) s[uoffset + 5] << 16); | |
c5a0f51a | 859 | else |
8e84507e | 860 | retnum = |
bb7a0f54 MHM |
861 | ((UV) s[uoffset ] << 56) + |
862 | ((UV) s[uoffset + 1] << 48) + | |
863 | ((UV) s[uoffset + 2] << 40) + | |
864 | ((UV) s[uoffset + 3] << 32) + | |
865 | ((UV) s[uoffset + 4] << 24) + | |
866 | ((UV) s[uoffset + 5] << 16) + | |
867 | ( s[uoffset + 6] << 8); | |
c5a0f51a JH |
868 | } |
869 | #endif | |
81e118e0 JH |
870 | } |
871 | } | |
872 | else if (size < 8) | |
bbb8a7e0 | 873 | retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1); |
81e118e0 | 874 | else { |
81e118e0 | 875 | if (size == 8) |
bb7a0f54 | 876 | retnum = s[uoffset]; |
81e118e0 JH |
877 | else if (size == 16) |
878 | retnum = | |
bb7a0f54 MHM |
879 | ((UV) s[uoffset] << 8) + |
880 | s[uoffset + 1]; | |
81e118e0 JH |
881 | else if (size == 32) |
882 | retnum = | |
bb7a0f54 MHM |
883 | ((UV) s[uoffset ] << 24) + |
884 | ((UV) s[uoffset + 1] << 16) + | |
885 | ( s[uoffset + 2] << 8) + | |
886 | s[uoffset + 3]; | |
d7d93a81 | 887 | #ifdef UV_IS_QUAD |
c5a0f51a | 888 | else if (size == 64) { |
a2a5de95 NC |
889 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
890 | "Bit vector size > 32 non-portable"); | |
c5a0f51a | 891 | retnum = |
bb7a0f54 MHM |
892 | ((UV) s[uoffset ] << 56) + |
893 | ((UV) s[uoffset + 1] << 48) + | |
894 | ((UV) s[uoffset + 2] << 40) + | |
895 | ((UV) s[uoffset + 3] << 32) + | |
896 | ((UV) s[uoffset + 4] << 24) + | |
897 | ((UV) s[uoffset + 5] << 16) + | |
898 | ( s[uoffset + 6] << 8) + | |
899 | s[uoffset + 7]; | |
c5a0f51a JH |
900 | } |
901 | #endif | |
81e118e0 JH |
902 | } |
903 | ||
904 | return retnum; | |
905 | } | |
906 | ||
33b45480 SB |
907 | /* currently converts input to bytes if possible but doesn't sweat failures, |
908 | * although it does ensure that the string it clobbers is not marked as | |
909 | * utf8-valid any more | |
910 | */ | |
79072805 | 911 | void |
864dbfa3 | 912 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 | 913 | { |
97aff369 | 914 | dVAR; |
eb578fdb KW |
915 | SSize_t offset, bitoffs = 0; |
916 | int size; | |
917 | unsigned char *s; | |
918 | UV lval; | |
79072805 | 919 | I32 mask; |
a0d0e21e LW |
920 | STRLEN targlen; |
921 | STRLEN len; | |
c4420975 | 922 | SV * const targ = LvTARG(sv); |
79072805 | 923 | |
7918f24d NC |
924 | PERL_ARGS_ASSERT_DO_VECSET; |
925 | ||
8990e307 LW |
926 | if (!targ) |
927 | return; | |
032061d2 BF |
928 | s = (unsigned char*)SvPV_force_flags(targ, targlen, |
929 | SV_GMAGIC | SV_UNDEF_RETURNS_NULL); | |
246fae53 | 930 | if (SvUTF8(targ)) { |
33b45480 SB |
931 | /* This is handled by the SvPOK_only below... |
932 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
933 | SvUTF8_off(targ); | |
934 | */ | |
935 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
936 | } |
937 | ||
4ebbc975 | 938 | (void)SvPOK_only(targ); |
81e118e0 | 939 | lval = SvUV(sv); |
79072805 | 940 | offset = LvTARGOFF(sv); |
fe58ced6 | 941 | if (offset < 0) |
ed9aa3b7 | 942 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 943 | size = LvTARGLEN(sv); |
8e84507e | 944 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 945 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 946 | |
bbb8a7e0 MHM |
947 | if (size < 8) { |
948 | bitoffs = ((offset%8)*size)%8; | |
949 | offset /= 8/size; | |
950 | } | |
951 | else if (size > 8) | |
952 | offset *= size/8; | |
953 | ||
954 | len = offset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
a0d0e21e LW |
955 | if (len > targlen) { |
956 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 957 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
958 | SvCUR_set(targ, len); |
959 | } | |
8e84507e | 960 | |
79072805 LW |
961 | if (size < 8) { |
962 | mask = (1 << size) - 1; | |
79072805 | 963 | lval &= mask; |
bbb8a7e0 MHM |
964 | s[offset] &= ~(mask << bitoffs); |
965 | s[offset] |= lval << bitoffs; | |
79072805 LW |
966 | } |
967 | else { | |
968 | if (size == 8) | |
eb160463 | 969 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 970 | else if (size == 16) { |
eb160463 GS |
971 | s[offset ] = (U8)((lval >> 8) & 0xff); |
972 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
973 | } |
974 | else if (size == 32) { | |
eb160463 GS |
975 | s[offset ] = (U8)((lval >> 24) & 0xff); |
976 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
977 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
978 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 979 | } |
d7d93a81 | 980 | #ifdef UV_IS_QUAD |
c5a0f51a | 981 | else if (size == 64) { |
a2a5de95 NC |
982 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
983 | "Bit vector size > 32 non-portable"); | |
eb160463 GS |
984 | s[offset ] = (U8)((lval >> 56) & 0xff); |
985 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
986 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
987 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
988 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
989 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
990 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
991 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 992 | } |
dc1e3f56 | 993 | #endif |
79072805 | 994 | } |
7bb043c3 | 995 | SvSETMAGIC(targ); |
79072805 LW |
996 | } |
997 | ||
998 | void | |
864dbfa3 | 999 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 | 1000 | { |
97aff369 | 1001 | dVAR; |
79072805 | 1002 | #ifdef LIBERAL |
eb578fdb KW |
1003 | long *dl; |
1004 | long *ll; | |
1005 | long *rl; | |
79072805 | 1006 | #endif |
eb578fdb | 1007 | char *dc; |
463ee0b2 LW |
1008 | STRLEN leftlen; |
1009 | STRLEN rightlen; | |
eb578fdb KW |
1010 | const char *lc; |
1011 | const char *rc; | |
1012 | STRLEN len; | |
bb7a0f54 | 1013 | STRLEN lensave; |
e62f0680 NC |
1014 | const char *lsave; |
1015 | const char *rsave; | |
12abf4f0 NC |
1016 | bool left_utf; |
1017 | bool right_utf; | |
bb7a0f54 | 1018 | STRLEN needlen = 0; |
0c57e439 | 1019 | |
7918f24d | 1020 | PERL_ARGS_ASSERT_DO_VOP; |
79072805 | 1021 | |
6b349a5c | 1022 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv))) |
76f68e9b | 1023 | sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */ |
8c8eee82 BM |
1024 | if (sv == left) { |
1025 | lsave = lc = SvPV_force_nomg(left, leftlen); | |
1026 | } | |
1027 | else { | |
1028 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1029 | SvPV_force_nomg_nolen(sv); | |
1030 | } | |
e62f0680 | 1031 | rsave = rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 NC |
1032 | |
1033 | /* This need to come after SvPV to ensure that string overloading has | |
1034 | fired off. */ | |
1035 | ||
1036 | left_utf = DO_UTF8(left); | |
1037 | right_utf = DO_UTF8(right); | |
1038 | ||
1039 | if (left_utf && !right_utf) { | |
1040 | /* Avoid triggering overloading again by using temporaries. | |
1041 | Maybe there should be a variant of sv_utf8_upgrade that takes pvn | |
1042 | */ | |
59cd0e26 | 1043 | right = newSVpvn_flags(rsave, rightlen, SVs_TEMP); |
12abf4f0 NC |
1044 | sv_utf8_upgrade(right); |
1045 | rsave = rc = SvPV_nomg_const(right, rightlen); | |
1046 | right_utf = TRUE; | |
1047 | } | |
1048 | else if (!left_utf && right_utf) { | |
59cd0e26 | 1049 | left = newSVpvn_flags(lsave, leftlen, SVs_TEMP); |
12abf4f0 NC |
1050 | sv_utf8_upgrade(left); |
1051 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1052 | left_utf = TRUE; | |
1053 | } | |
1054 | ||
93a17b20 | 1055 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1056 | lensave = len; |
9fdd7463 JH |
1057 | SvCUR_set(sv, len); |
1058 | (void)SvPOK_only(sv); | |
c9b3c8d0 JH |
1059 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1060 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
a02a5408 | 1061 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 JH |
1062 | } |
1063 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2596d9fe | 1064 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 MHM |
1065 | if (SvLEN(sv) < len + 1) { |
1066 | dc = SvGROW(sv, len + 1); | |
ff68c719 | 1067 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1068 | } | |
c9b3c8d0 JH |
1069 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1070 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1071 | } |
1072 | else { | |
bb7a0f54 MHM |
1073 | needlen = optype == OP_BIT_AND |
1074 | ? len : (leftlen > rightlen ? leftlen : rightlen); | |
a02a5408 | 1075 | Newxz(dc, needlen + 1, char); |
aa0a69cb | 1076 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 | 1077 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 | 1078 | } |
0c57e439 GS |
1079 | if (left_utf || right_utf) { |
1080 | UV duc, luc, ruc; | |
794a0d33 JH |
1081 | char *dcorig = dc; |
1082 | char *dcsave = NULL; | |
0c57e439 GS |
1083 | STRLEN lulen = leftlen; |
1084 | STRLEN rulen = rightlen; | |
ba210ebe | 1085 | STRLEN ulen; |
0c57e439 GS |
1086 | |
1087 | switch (optype) { | |
1088 | case OP_BIT_AND: | |
1089 | while (lulen && rulen) { | |
9041c2e3 | 1090 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1091 | lc += ulen; |
1092 | lulen -= ulen; | |
9041c2e3 | 1093 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1094 | rc += ulen; |
1095 | rulen -= ulen; | |
1096 | duc = luc & ruc; | |
9041c2e3 | 1097 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1098 | } |
3ce4c532 NC |
1099 | if (sv == left || sv == right) |
1100 | (void)sv_usepvn(sv, dcorig, needlen); | |
794a0d33 | 1101 | SvCUR_set(sv, dc - dcorig); |
0c57e439 GS |
1102 | break; |
1103 | case OP_BIT_XOR: | |
1104 | while (lulen && rulen) { | |
9041c2e3 | 1105 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1106 | lc += ulen; |
1107 | lulen -= ulen; | |
9041c2e3 | 1108 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1109 | rc += ulen; |
1110 | rulen -= ulen; | |
1111 | duc = luc ^ ruc; | |
9041c2e3 | 1112 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1113 | } |
1114 | goto mop_up_utf; | |
1115 | case OP_BIT_OR: | |
1116 | while (lulen && rulen) { | |
9041c2e3 | 1117 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1118 | lc += ulen; |
1119 | lulen -= ulen; | |
9041c2e3 | 1120 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1121 | rc += ulen; |
1122 | rulen -= ulen; | |
1123 | duc = luc | ruc; | |
9041c2e3 | 1124 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1125 | } |
1126 | mop_up_utf: | |
794a0d33 JH |
1127 | if (rulen) |
1128 | dcsave = savepvn(rc, rulen); | |
1129 | else if (lulen) | |
1130 | dcsave = savepvn(lc, lulen); | |
3ce4c532 | 1131 | if (sv == left || sv == right) |
4bac9ae4 | 1132 | (void)sv_usepvn(sv, dcorig, needlen); /* uses Renew(); defaults to nomg */ |
794a0d33 | 1133 | SvCUR_set(sv, dc - dcorig); |
0c57e439 | 1134 | if (rulen) |
4bac9ae4 | 1135 | sv_catpvn_nomg(sv, dcsave, rulen); |
0c57e439 | 1136 | else if (lulen) |
4bac9ae4 | 1137 | sv_catpvn_nomg(sv, dcsave, lulen); |
0c57e439 GS |
1138 | else |
1139 | *SvEND(sv) = '\0'; | |
794a0d33 | 1140 | Safefree(dcsave); |
0c57e439 | 1141 | break; |
41547412 NC |
1142 | default: |
1143 | if (sv == left || sv == right) | |
1144 | Safefree(dcorig); | |
c395bd6c AL |
1145 | Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)", |
1146 | (unsigned)optype, PL_op_name[optype]); | |
0c57e439 GS |
1147 | } |
1148 | SvUTF8_on(sv); | |
1149 | goto finish; | |
1150 | } | |
1151 | else | |
79072805 LW |
1152 | #ifdef LIBERAL |
1153 | if (len >= sizeof(long)*4 && | |
bb7a0f54 MHM |
1154 | !((unsigned long)dc % sizeof(long)) && |
1155 | !((unsigned long)lc % sizeof(long)) && | |
1156 | !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ | |
79072805 | 1157 | { |
bb7a0f54 | 1158 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 LW |
1159 | len /= (sizeof(long)*4); |
1160 | ||
1161 | dl = (long*)dc; | |
1162 | ll = (long*)lc; | |
1163 | rl = (long*)rc; | |
1164 | ||
1165 | switch (optype) { | |
1166 | case OP_BIT_AND: | |
1167 | while (len--) { | |
1168 | *dl++ = *ll++ & *rl++; | |
1169 | *dl++ = *ll++ & *rl++; | |
1170 | *dl++ = *ll++ & *rl++; | |
1171 | *dl++ = *ll++ & *rl++; | |
1172 | } | |
1173 | break; | |
a0d0e21e | 1174 | case OP_BIT_XOR: |
79072805 LW |
1175 | while (len--) { |
1176 | *dl++ = *ll++ ^ *rl++; | |
1177 | *dl++ = *ll++ ^ *rl++; | |
1178 | *dl++ = *ll++ ^ *rl++; | |
1179 | *dl++ = *ll++ ^ *rl++; | |
1180 | } | |
1181 | break; | |
1182 | case OP_BIT_OR: | |
1183 | while (len--) { | |
1184 | *dl++ = *ll++ | *rl++; | |
1185 | *dl++ = *ll++ | *rl++; | |
1186 | *dl++ = *ll++ | *rl++; | |
1187 | *dl++ = *ll++ | *rl++; | |
1188 | } | |
1189 | } | |
1190 | ||
1191 | dc = (char*)dl; | |
1192 | lc = (char*)ll; | |
1193 | rc = (char*)rl; | |
1194 | ||
1195 | len = remainder; | |
1196 | } | |
1197 | #endif | |
a0d0e21e | 1198 | { |
a0d0e21e LW |
1199 | switch (optype) { |
1200 | case OP_BIT_AND: | |
1201 | while (len--) | |
1202 | *dc++ = *lc++ & *rc++; | |
1a787b95 | 1203 | *dc = '\0'; |
a0d0e21e LW |
1204 | break; |
1205 | case OP_BIT_XOR: | |
1206 | while (len--) | |
1207 | *dc++ = *lc++ ^ *rc++; | |
1208 | goto mop_up; | |
1209 | case OP_BIT_OR: | |
1210 | while (len--) | |
1211 | *dc++ = *lc++ | *rc++; | |
1212 | mop_up: | |
1213 | len = lensave; | |
bb7a0f54 | 1214 | if (rightlen > len) |
4bac9ae4 | 1215 | sv_catpvn_nomg(sv, rsave + len, rightlen - len); |
eb160463 | 1216 | else if (leftlen > (STRLEN)len) |
4bac9ae4 | 1217 | sv_catpvn_nomg(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1218 | else |
1219 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1220 | break; |
1221 | } | |
79072805 | 1222 | } |
0c57e439 | 1223 | finish: |
fb73857a | 1224 | SvTAINT(sv); |
79072805 | 1225 | } |
463ee0b2 LW |
1226 | |
1227 | OP * | |
cea2e8a9 | 1228 | Perl_do_kv(pTHX) |
463ee0b2 | 1229 | { |
97aff369 | 1230 | dVAR; |
39644a26 | 1231 | dSP; |
73ff03e8 | 1232 | HV * const keys = MUTABLE_HV(POPs); |
eb578fdb | 1233 | HE *entry; |
f54cb97a AL |
1234 | const I32 gimme = GIMME_V; |
1235 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); | |
cba5a3b0 DG |
1236 | /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */ |
1237 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS || PL_op->op_type == OP_RKEYS); | |
1238 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES || PL_op->op_type == OP_RVALUES); | |
463ee0b2 | 1239 | |
800e9ae0 | 1240 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1241 | |
54310121 | 1242 | if (gimme == G_VOID) |
aa689395 | 1243 | RETURN; |
1244 | ||
54310121 | 1245 | if (gimme == G_SCALAR) { |
78f9721b | 1246 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
2154eca7 EB |
1247 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
1248 | sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0); | |
1249 | LvTYPE(ret) = 'k'; | |
1250 | LvTARG(ret) = SvREFCNT_inc_simple(keys); | |
1251 | PUSHs(ret); | |
81714fb9 | 1252 | } |
463ee0b2 | 1253 | else { |
2154eca7 EB |
1254 | IV i; |
1255 | dTARGET; | |
1256 | ||
1257 | if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) { | |
1b95d04f | 1258 | i = HvUSEDKEYS(keys); |
2154eca7 EB |
1259 | } |
1260 | else { | |
1261 | i = 0; | |
1262 | while (hv_iternext(keys)) i++; | |
1263 | } | |
1264 | PUSHi( i ); | |
463ee0b2 | 1265 | } |
463ee0b2 LW |
1266 | RETURN; |
1267 | } | |
1268 | ||
1b95d04f | 1269 | EXTEND(SP, HvUSEDKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1270 | |
463ee0b2 | 1271 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1272 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1273 | SPAGAIN; |
574c8022 | 1274 | if (dokeys) { |
890ce7af | 1275 | SV* const sv = hv_iterkeysv(entry); |
574c8022 JH |
1276 | XPUSHs(sv); /* won't clobber stack_sp */ |
1277 | } | |
463ee0b2 | 1278 | if (dovalues) { |
f54cb97a | 1279 | SV *tmpstr; |
463ee0b2 | 1280 | PUTBACK; |
73ff03e8 | 1281 | tmpstr = hv_iterval(keys,entry); |
cea2e8a9 | 1282 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1283 | (unsigned long)HeHASH(entry), |
cdd3ba14 | 1284 | (int)HvMAX(keys)+1, |
800e9ae0 | 1285 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 | 1286 | SPAGAIN; |
46fc3d4c | 1287 | XPUSHs(tmpstr); |
463ee0b2 LW |
1288 | } |
1289 | PUTBACK; | |
1290 | } | |
1291 | return NORMAL; | |
1292 | } | |
4e35701f | 1293 | |
af3babe4 NC |
1294 | /* |
1295 | * Local variables: | |
1296 | * c-indentation-style: bsd | |
1297 | * c-basic-offset: 4 | |
14d04a33 | 1298 | * indent-tabs-mode: nil |
af3babe4 NC |
1299 | * End: |
1300 | * | |
14d04a33 | 1301 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 1302 | */ |