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 FC |
18 | * ops. For example, both pp_sprintf() and pp_prtf() call the function |
19 | * do_printf() 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 | ||
942e002e | 30 | STATIC I32 |
c395bd6c | 31 | S_do_trans_simple(pTHX_ SV * const sv) |
79072805 | 32 | { |
4757a243 | 33 | I32 matches = 0; |
463ee0b2 | 34 | STRLEN len; |
9138d6ca | 35 | U8 *s = (U8*)SvPV_nomg(sv,len); |
c395bd6c | 36 | U8 * const send = s+len; |
c4420975 | 37 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
38 | |
39 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE; | |
40 | ||
4757a243 | 41 | if (!tbl) |
7d85a32c | 42 | Perl_croak(aTHX_ "panic: do_trans_simple line %d",__LINE__); |
a0ed51b3 | 43 | |
1e54db1a | 44 | /* First, take care of non-UTF-8 input strings, because they're easy */ |
1aa99e6b | 45 | if (!SvUTF8(sv)) { |
01ec43d0 | 46 | while (s < send) { |
f54cb97a AL |
47 | const I32 ch = tbl[*s]; |
48 | if (ch >= 0) { | |
01ec43d0 | 49 | matches++; |
c4420975 | 50 | *s = (U8)ch; |
01ec43d0 | 51 | } |
c4420975 | 52 | s++; |
01ec43d0 GS |
53 | } |
54 | SvSETMAGIC(sv); | |
9b877dbb IH |
55 | } |
56 | else { | |
c395bd6c AL |
57 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
58 | U8 *d; | |
59 | U8 *dstart; | |
60 | ||
61 | /* Allow for expansion: $_="a".chr(400); tr/a/\xFE/, FE needs encoding */ | |
62 | if (grows) | |
63 | Newx(d, len*2+1, U8); | |
64 | else | |
65 | d = s; | |
66 | dstart = d; | |
67 | while (s < send) { | |
68 | STRLEN ulen; | |
69 | I32 ch; | |
70 | ||
71 | /* Need to check this, otherwise 128..255 won't match */ | |
72 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); | |
73 | if (c < 0x100 && (ch = tbl[c]) >= 0) { | |
74 | matches++; | |
75 | d = uvchr_to_utf8(d, ch); | |
76 | s += ulen; | |
77 | } | |
78 | else { /* No match -> copy */ | |
79 | Move(s, d, ulen, U8); | |
80 | d += ulen; | |
81 | s += ulen; | |
82 | } | |
83 | } | |
84 | if (grows) { | |
85 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
86 | Safefree(dstart); | |
87 | } | |
88 | else { | |
89 | *d = '\0'; | |
90 | SvCUR_set(sv, d - dstart); | |
91 | } | |
92 | SvUTF8_on(sv); | |
93 | SvSETMAGIC(sv); | |
9b877dbb | 94 | } |
4757a243 LW |
95 | return matches; |
96 | } | |
97 | ||
942e002e | 98 | STATIC I32 |
c395bd6c | 99 | S_do_trans_count(pTHX_ SV * const sv) |
4757a243 | 100 | { |
4757a243 | 101 | STRLEN len; |
9138d6ca | 102 | const U8 *s = (const U8*)SvPV_nomg_const(sv, len); |
c395bd6c AL |
103 | const U8 * const send = s + len; |
104 | I32 matches = 0; | |
f54cb97a | 105 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
106 | |
107 | PERL_ARGS_ASSERT_DO_TRANS_COUNT; | |
108 | ||
4757a243 | 109 | if (!tbl) |
7d85a32c | 110 | Perl_croak(aTHX_ "panic: do_trans_count line %d",__LINE__); |
4757a243 | 111 | |
c395bd6c | 112 | if (!SvUTF8(sv)) { |
1aa99e6b IH |
113 | while (s < send) { |
114 | if (tbl[*s++] >= 0) | |
036b4402 | 115 | matches++; |
1aa99e6b | 116 | } |
c395bd6c | 117 | } |
fabdb6c0 AL |
118 | else { |
119 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; | |
1aa99e6b | 120 | while (s < send) { |
1aa99e6b | 121 | STRLEN ulen; |
9f7f3913 | 122 | const UV c = utf8n_to_uvchr(s, send - s, &ulen, UTF8_ALLOW_DEFAULT); |
8973db79 JH |
123 | if (c < 0x100) { |
124 | if (tbl[c] >= 0) | |
125 | matches++; | |
126 | } else if (complement) | |
1aa99e6b IH |
127 | matches++; |
128 | s += ulen; | |
129 | } | |
fabdb6c0 | 130 | } |
4757a243 LW |
131 | |
132 | return matches; | |
133 | } | |
134 | ||
942e002e | 135 | STATIC I32 |
c395bd6c | 136 | S_do_trans_complex(pTHX_ SV * const sv) |
4757a243 | 137 | { |
c395bd6c | 138 | STRLEN len; |
9138d6ca | 139 | U8 *s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 140 | U8 * const send = s+len; |
4757a243 | 141 | I32 matches = 0; |
f54cb97a | 142 | const short * const tbl = (short*)cPVOP->op_pv; |
7918f24d NC |
143 | |
144 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX; | |
145 | ||
4757a243 | 146 | if (!tbl) |
7d85a32c | 147 | Perl_croak(aTHX_ "panic: do_trans_complex line %d",__LINE__); |
4757a243 | 148 | |
c395bd6c AL |
149 | if (!SvUTF8(sv)) { |
150 | U8 *d = s; | |
151 | U8 * const dstart = d; | |
4757a243 | 152 | |
1aa99e6b | 153 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
f54cb97a | 154 | const U8* p = send; |
1aa99e6b | 155 | while (s < send) { |
f54cb97a AL |
156 | const I32 ch = tbl[*s]; |
157 | if (ch >= 0) { | |
eb160463 | 158 | *d = (U8)ch; |
036b4402 | 159 | matches++; |
1aa99e6b IH |
160 | if (p != d - 1 || *p != *d) |
161 | p = d++; | |
162 | } | |
163 | else if (ch == -1) /* -1 is unmapped character */ | |
164 | *d++ = *s; | |
165 | else if (ch == -2) /* -2 is delete character */ | |
166 | matches++; | |
167 | s++; | |
168 | } | |
a0ed51b3 | 169 | } |
1aa99e6b IH |
170 | else { |
171 | while (s < send) { | |
f54cb97a AL |
172 | const I32 ch = tbl[*s]; |
173 | if (ch >= 0) { | |
1aa99e6b | 174 | matches++; |
eb160463 | 175 | *d++ = (U8)ch; |
1aa99e6b IH |
176 | } |
177 | else if (ch == -1) /* -1 is unmapped character */ | |
178 | *d++ = *s; | |
179 | else if (ch == -2) /* -2 is delete character */ | |
180 | matches++; | |
181 | s++; | |
182 | } | |
183 | } | |
76ef7183 | 184 | *d = '\0'; |
1aa99e6b | 185 | SvCUR_set(sv, d - dstart); |
4757a243 | 186 | } |
c395bd6c | 187 | else { /* is utf8 */ |
fabdb6c0 AL |
188 | const I32 complement = PL_op->op_private & OPpTRANS_COMPLEMENT; |
189 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; | |
190 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
c395bd6c AL |
191 | U8 *d; |
192 | U8 *dstart; | |
193 | STRLEN rlen = 0; | |
fabdb6c0 | 194 | |
9b877dbb | 195 | if (grows) |
a02a5408 | 196 | Newx(d, len*2+1, U8); |
9b877dbb IH |
197 | else |
198 | d = s; | |
1aa99e6b | 199 | dstart = d; |
8973db79 JH |
200 | if (complement && !del) |
201 | rlen = tbl[0x100]; | |
1aa99e6b IH |
202 | |
203 | if (PL_op->op_private & OPpTRANS_SQUASH) { | |
1aa99e6b IH |
204 | UV pch = 0xfeedface; |
205 | while (s < send) { | |
206 | STRLEN len; | |
9f7f3913 TS |
207 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
208 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 209 | I32 ch; |
1aa99e6b | 210 | |
8973db79 JH |
211 | if (comp > 0xff) { |
212 | if (!complement) { | |
640af7f1 | 213 | Move(s, d, len, U8); |
8973db79 JH |
214 | d += len; |
215 | } | |
216 | else { | |
217 | matches++; | |
218 | if (!del) { | |
bb7a0f54 | 219 | ch = (rlen == 0) ? (I32)comp : |
05d340b8 | 220 | (comp - 0x100 < rlen) ? |
8973db79 | 221 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 222 | if ((UV)ch != pch) { |
9041c2e3 | 223 | d = uvchr_to_utf8(d, ch); |
eb160463 | 224 | pch = (UV)ch; |
8973db79 JH |
225 | } |
226 | s += len; | |
227 | continue; | |
228 | } | |
229 | } | |
9b877dbb | 230 | } |
1aa99e6b IH |
231 | else if ((ch = tbl[comp]) >= 0) { |
232 | matches++; | |
eb160463 | 233 | if ((UV)ch != pch) { |
9041c2e3 | 234 | d = uvchr_to_utf8(d, ch); |
eb160463 | 235 | pch = (UV)ch; |
1aa99e6b IH |
236 | } |
237 | s += len; | |
238 | continue; | |
239 | } | |
9b877dbb | 240 | else if (ch == -1) { /* -1 is unmapped character */ |
640af7f1 | 241 | Move(s, d, len, U8); |
9b877dbb IH |
242 | d += len; |
243 | } | |
1aa99e6b IH |
244 | else if (ch == -2) /* -2 is delete character */ |
245 | matches++; | |
246 | s += len; | |
247 | pch = 0xfeedface; | |
248 | } | |
5d06d08e | 249 | } |
1aa99e6b IH |
250 | else { |
251 | while (s < send) { | |
252 | STRLEN len; | |
9f7f3913 TS |
253 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
254 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 255 | I32 ch; |
8973db79 JH |
256 | if (comp > 0xff) { |
257 | if (!complement) { | |
0376ff32 | 258 | Move(s, d, len, U8); |
8973db79 JH |
259 | d += len; |
260 | } | |
261 | else { | |
262 | matches++; | |
263 | if (!del) { | |
9041c2e3 NIS |
264 | if (comp - 0x100 < rlen) |
265 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 266 | else |
9041c2e3 | 267 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
268 | } |
269 | } | |
9b877dbb | 270 | } |
1aa99e6b | 271 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 272 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
273 | matches++; |
274 | } | |
275 | else if (ch == -1) { /* -1 is unmapped character */ | |
640af7f1 | 276 | Move(s, d, len, U8); |
9b877dbb | 277 | d += len; |
1aa99e6b IH |
278 | } |
279 | else if (ch == -2) /* -2 is delete character */ | |
280 | matches++; | |
281 | s += len; | |
282 | } | |
283 | } | |
9b877dbb IH |
284 | if (grows) { |
285 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
286 | Safefree(dstart); | |
287 | } | |
288 | else { | |
289 | *d = '\0'; | |
290 | SvCUR_set(sv, d - dstart); | |
291 | } | |
1aa99e6b | 292 | SvUTF8_on(sv); |
4757a243 | 293 | } |
5e44153e | 294 | SvSETMAGIC(sv); |
4757a243 LW |
295 | return matches; |
296 | } | |
297 | ||
942e002e | 298 | STATIC I32 |
c395bd6c | 299 | S_do_trans_simple_utf8(pTHX_ SV * const sv) |
4757a243 | 300 | { |
4757a243 LW |
301 | U8 *s; |
302 | U8 *send; | |
303 | U8 *d; | |
036b4402 | 304 | U8 *start; |
1aa99e6b | 305 | U8 *dstart, *dend; |
4757a243 | 306 | I32 matches = 0; |
f54cb97a | 307 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 | 308 | STRLEN len; |
043e41b8 DM |
309 | SV* const rv = |
310 | #ifdef USE_ITHREADS | |
311 | PAD_SVl(cPADOP->op_padix); | |
312 | #else | |
ad64d0ec | 313 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 314 | #endif |
85fbaab2 | 315 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 316 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
317 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
318 | const UV extra = none + 1; | |
b7953727 | 319 | UV final = 0; |
1aa99e6b | 320 | U8 hibit = 0; |
4757a243 | 321 | |
7918f24d NC |
322 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8; |
323 | ||
9138d6ca | 324 | s = (U8*)SvPV_nomg(sv, len); |
c395bd6c | 325 | if (!SvUTF8(sv)) { |
c4420975 AL |
326 | const U8 *t = s; |
327 | const U8 * const e = s + len; | |
c4d5f83a | 328 | while (t < e) { |
f54cb97a | 329 | const U8 ch = *t++; |
6f2d5cbc | 330 | hibit = !NATIVE_BYTE_IS_INVARIANT(ch); |
fabdb6c0 AL |
331 | if (hibit) { |
332 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 333 | break; |
fabdb6c0 | 334 | } |
c4d5f83a | 335 | } |
1aa99e6b | 336 | } |
4757a243 | 337 | send = s + len; |
036b4402 | 338 | start = s; |
4757a243 | 339 | |
a4fc7abc | 340 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 LW |
341 | if (svp) |
342 | final = SvUV(*svp); | |
343 | ||
9b877dbb IH |
344 | if (grows) { |
345 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 346 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
347 | dend = d + len * 3; |
348 | dstart = d; | |
349 | } | |
350 | else { | |
351 | dstart = d = s; | |
352 | dend = d + len; | |
353 | } | |
1aa99e6b | 354 | |
4757a243 | 355 | while (s < send) { |
0bcc34c2 AL |
356 | const UV uv = swash_fetch(rv, s, TRUE); |
357 | if (uv < none) { | |
4757a243 LW |
358 | s += UTF8SKIP(s); |
359 | matches++; | |
c80e42f3 | 360 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
361 | } |
362 | else if (uv == none) { | |
4373e329 | 363 | const int i = UTF8SKIP(s); |
0376ff32 | 364 | Move(s, d, i, U8); |
9b877dbb IH |
365 | d += i; |
366 | s += i; | |
4757a243 LW |
367 | } |
368 | else if (uv == extra) { | |
4373e329 | 369 | s += UTF8SKIP(s); |
4757a243 | 370 | matches++; |
c80e42f3 | 371 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
372 | } |
373 | else | |
374 | s += UTF8SKIP(s); | |
1aa99e6b | 375 | |
9b877dbb | 376 | if (d > dend) { |
4373e329 AL |
377 | const STRLEN clen = d - dstart; |
378 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 379 | if (!grows) |
7d85a32c | 380 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 | 381 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
382 | d = dstart + clen; |
383 | dend = dstart + nlen; | |
384 | } | |
4757a243 | 385 | } |
16ec844d | 386 | if (grows || hibit) { |
9b877dbb IH |
387 | sv_setpvn(sv, (char*)dstart, d - dstart); |
388 | Safefree(dstart); | |
16ec844d IH |
389 | if (grows && hibit) |
390 | Safefree(start); | |
9b877dbb IH |
391 | } |
392 | else { | |
393 | *d = '\0'; | |
394 | SvCUR_set(sv, d - dstart); | |
395 | } | |
4757a243 | 396 | SvSETMAGIC(sv); |
1aa99e6b | 397 | SvUTF8_on(sv); |
4757a243 LW |
398 | |
399 | return matches; | |
400 | } | |
401 | ||
942e002e | 402 | STATIC I32 |
c395bd6c | 403 | S_do_trans_count_utf8(pTHX_ SV * const sv) |
4757a243 | 404 | { |
93524f2b | 405 | const U8 *s; |
cbbf8932 AL |
406 | const U8 *start = NULL; |
407 | const U8 *send; | |
4757a243 LW |
408 | I32 matches = 0; |
409 | STRLEN len; | |
043e41b8 DM |
410 | SV* const rv = |
411 | #ifdef USE_ITHREADS | |
412 | PAD_SVl(cPADOP->op_padix); | |
413 | #else | |
ad64d0ec | 414 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 415 | #endif |
85fbaab2 | 416 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 417 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
418 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
419 | const UV extra = none + 1; | |
1aa99e6b | 420 | U8 hibit = 0; |
4757a243 | 421 | |
7918f24d NC |
422 | PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8; |
423 | ||
9138d6ca | 424 | s = (const U8*)SvPV_nomg_const(sv, len); |
1aa99e6b | 425 | if (!SvUTF8(sv)) { |
4373e329 | 426 | const U8 *t = s; |
c4420975 | 427 | const U8 * const e = s + len; |
c4d5f83a | 428 | while (t < e) { |
4373e329 | 429 | const U8 ch = *t++; |
6f2d5cbc | 430 | hibit = !NATIVE_BYTE_IS_INVARIANT(ch); |
fabdb6c0 AL |
431 | if (hibit) { |
432 | start = s = bytes_to_utf8(s, &len); | |
1aa99e6b | 433 | break; |
fabdb6c0 | 434 | } |
c4d5f83a | 435 | } |
1aa99e6b | 436 | } |
4757a243 LW |
437 | send = s + len; |
438 | ||
439 | while (s < send) { | |
c4420975 AL |
440 | const UV uv = swash_fetch(rv, s, TRUE); |
441 | if (uv < none || uv == extra) | |
4757a243 | 442 | matches++; |
834a4ddd | 443 | s += UTF8SKIP(s); |
4757a243 | 444 | } |
1aa99e6b IH |
445 | if (hibit) |
446 | Safefree(start); | |
4757a243 LW |
447 | |
448 | return matches; | |
449 | } | |
450 | ||
942e002e | 451 | STATIC I32 |
c395bd6c | 452 | S_do_trans_complex_utf8(pTHX_ SV * const sv) |
4757a243 | 453 | { |
1aa99e6b | 454 | U8 *start, *send; |
4757a243 LW |
455 | U8 *d; |
456 | I32 matches = 0; | |
4373e329 AL |
457 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
458 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
459 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; | |
043e41b8 DM |
460 | SV* const rv = |
461 | #ifdef USE_ITHREADS | |
462 | PAD_SVl(cPADOP->op_padix); | |
463 | #else | |
ad64d0ec | 464 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 465 | #endif |
85fbaab2 | 466 | HV * const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 467 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
468 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
469 | const UV extra = none + 1; | |
b7953727 | 470 | UV final = 0; |
45005bfb | 471 | bool havefinal = FALSE; |
4757a243 | 472 | STRLEN len; |
1aa99e6b | 473 | U8 *dstart, *dend; |
1aa99e6b | 474 | U8 hibit = 0; |
9138d6ca | 475 | U8 *s = (U8*)SvPV_nomg(sv, len); |
7918f24d NC |
476 | |
477 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8; | |
478 | ||
c395bd6c | 479 | if (!SvUTF8(sv)) { |
53c1dcc0 AL |
480 | const U8 *t = s; |
481 | const U8 * const e = s + len; | |
c4d5f83a | 482 | while (t < e) { |
f54cb97a | 483 | const U8 ch = *t++; |
6f2d5cbc | 484 | hibit = !NATIVE_BYTE_IS_INVARIANT(ch); |
fabdb6c0 AL |
485 | if (hibit) { |
486 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 487 | break; |
fabdb6c0 | 488 | } |
c4d5f83a | 489 | } |
1aa99e6b | 490 | } |
4757a243 | 491 | send = s + len; |
1aa99e6b | 492 | start = s; |
4757a243 | 493 | |
a4fc7abc | 494 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb | 495 | if (svp) { |
4757a243 | 496 | final = SvUV(*svp); |
45005bfb JH |
497 | havefinal = TRUE; |
498 | } | |
4757a243 | 499 | |
9b877dbb IH |
500 | if (grows) { |
501 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 502 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
503 | dend = d + len * 3; |
504 | dstart = d; | |
505 | } | |
506 | else { | |
507 | dstart = d = s; | |
508 | dend = d + len; | |
509 | } | |
4757a243 LW |
510 | |
511 | if (squash) { | |
512 | UV puv = 0xfeedface; | |
513 | while (s < send) { | |
f54cb97a | 514 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 515 | |
9b877dbb | 516 | if (d > dend) { |
4373e329 AL |
517 | const STRLEN clen = d - dstart; |
518 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 519 | if (!grows) |
7d85a32c | 520 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 521 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
522 | d = dstart + clen; |
523 | dend = dstart + nlen; | |
4757a243 LW |
524 | } |
525 | if (uv < none) { | |
526 | matches++; | |
94472101 | 527 | s += UTF8SKIP(s); |
4757a243 | 528 | if (uv != puv) { |
c80e42f3 | 529 | d = uvchr_to_utf8(d, uv); |
4757a243 LW |
530 | puv = uv; |
531 | } | |
4757a243 LW |
532 | continue; |
533 | } | |
534 | else if (uv == none) { /* "none" is unmapped character */ | |
4373e329 | 535 | const int i = UTF8SKIP(s); |
0376ff32 | 536 | Move(s, d, i, U8); |
9b877dbb IH |
537 | d += i; |
538 | s += i; | |
4757a243 LW |
539 | puv = 0xfeedface; |
540 | continue; | |
a0ed51b3 | 541 | } |
4757a243 LW |
542 | else if (uv == extra && !del) { |
543 | matches++; | |
45005bfb JH |
544 | if (havefinal) { |
545 | s += UTF8SKIP(s); | |
546 | if (puv != final) { | |
c80e42f3 | 547 | d = uvchr_to_utf8(d, final); |
45005bfb JH |
548 | puv = final; |
549 | } | |
550 | } | |
551 | else { | |
552 | STRLEN len; | |
c80e42f3 | 553 | uv = utf8n_to_uvchr(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb | 554 | if (uv != puv) { |
0376ff32 | 555 | Move(s, d, len, U8); |
45005bfb JH |
556 | d += len; |
557 | puv = uv; | |
558 | } | |
559 | s += len; | |
4757a243 | 560 | } |
4757a243 LW |
561 | continue; |
562 | } | |
01ec43d0 GS |
563 | matches++; /* "none+1" is delete character */ |
564 | s += UTF8SKIP(s); | |
a0ed51b3 | 565 | } |
79072805 LW |
566 | } |
567 | else { | |
4757a243 | 568 | while (s < send) { |
f54cb97a | 569 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb | 570 | if (d > dend) { |
f54cb97a AL |
571 | const STRLEN clen = d - dstart; |
572 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 573 | if (!grows) |
7d85a32c | 574 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 575 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
576 | d = dstart + clen; |
577 | dend = dstart + nlen; | |
a0ed51b3 | 578 | } |
4757a243 LW |
579 | if (uv < none) { |
580 | matches++; | |
01ec43d0 | 581 | s += UTF8SKIP(s); |
c80e42f3 | 582 | d = uvchr_to_utf8(d, uv); |
4757a243 | 583 | continue; |
a0ed51b3 | 584 | } |
4757a243 | 585 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a | 586 | const int i = UTF8SKIP(s); |
0376ff32 | 587 | Move(s, d, i, U8); |
9b877dbb IH |
588 | d += i; |
589 | s += i; | |
4757a243 | 590 | continue; |
79072805 | 591 | } |
4757a243 LW |
592 | else if (uv == extra && !del) { |
593 | matches++; | |
01ec43d0 | 594 | s += UTF8SKIP(s); |
c80e42f3 | 595 | d = uvchr_to_utf8(d, final); |
4757a243 LW |
596 | continue; |
597 | } | |
01ec43d0 GS |
598 | matches++; /* "none+1" is delete character */ |
599 | s += UTF8SKIP(s); | |
79072805 | 600 | } |
4757a243 | 601 | } |
16ec844d | 602 | if (grows || hibit) { |
9b877dbb IH |
603 | sv_setpvn(sv, (char*)dstart, d - dstart); |
604 | Safefree(dstart); | |
16ec844d IH |
605 | if (grows && hibit) |
606 | Safefree(start); | |
9b877dbb IH |
607 | } |
608 | else { | |
609 | *d = '\0'; | |
610 | SvCUR_set(sv, d - dstart); | |
611 | } | |
1aa99e6b | 612 | SvUTF8_on(sv); |
4757a243 LW |
613 | SvSETMAGIC(sv); |
614 | ||
615 | return matches; | |
616 | } | |
617 | ||
618 | I32 | |
864dbfa3 | 619 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 LW |
620 | { |
621 | STRLEN len; | |
a53bfdae HS |
622 | const I32 flags = PL_op->op_private; |
623 | const I32 hasutf = flags & (OPpTRANS_FROM_UTF | OPpTRANS_TO_UTF); | |
4757a243 | 624 | |
7918f24d NC |
625 | PERL_ARGS_ASSERT_DO_TRANS; |
626 | ||
a53bfdae HS |
627 | if (SvREADONLY(sv) && !(flags & OPpTRANS_IDENTICAL)) { |
628 | Perl_croak_no_modify(); | |
2233f375 | 629 | } |
10516c54 | 630 | (void)SvPV_const(sv, len); |
4757a243 LW |
631 | if (!len) |
632 | return 0; | |
a53bfdae | 633 | if (!(flags & OPpTRANS_IDENTICAL)) { |
4499db73 | 634 | if (!SvPOKp(sv) || SvTHINKFIRST(sv)) |
b4cc4d79 | 635 | (void)SvPV_force_nomg(sv, len); |
2de7b02f | 636 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 637 | } |
4757a243 | 638 | |
cea2e8a9 | 639 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 640 | |
a53bfdae HS |
641 | /* If we use only OPpTRANS_IDENTICAL to bypass the READONLY check, |
642 | * we must also rely on it to choose the readonly strategy. | |
643 | */ | |
644 | if (flags & OPpTRANS_IDENTICAL) { | |
645 | return hasutf ? do_trans_count_utf8(sv) : do_trans_count(sv); | |
646 | } else if (flags & (OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
647 | return hasutf ? do_trans_complex_utf8(sv) : do_trans_complex(sv); | |
648 | } else { | |
649 | return hasutf ? do_trans_simple_utf8(sv) : do_trans_simple(sv); | |
79072805 | 650 | } |
79072805 LW |
651 | } |
652 | ||
653 | void | |
5aaab254 | 654 | Perl_do_join(pTHX_ SV *sv, SV *delim, SV **mark, SV **sp) |
79072805 | 655 | { |
53c1dcc0 | 656 | SV ** const oldmark = mark; |
eb578fdb KW |
657 | I32 items = sp - mark; |
658 | STRLEN len; | |
463ee0b2 | 659 | STRLEN delimlen; |
810a07df | 660 | const char * const delims = SvPV_const(delim, delimlen); |
79072805 | 661 | |
7918f24d NC |
662 | PERL_ARGS_ASSERT_DO_JOIN; |
663 | ||
79072805 LW |
664 | mark++; |
665 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
862a34c6 | 666 | SvUPGRADE(sv, SVt_PV); |
79072805 LW |
667 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
668 | while (items-- > 0) { | |
1426bbf4 | 669 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a | 670 | STRLEN tmplen; |
5c144d81 | 671 | SvPV_const(*mark, tmplen); |
463ee0b2 | 672 | len += tmplen; |
79072805 LW |
673 | } |
674 | mark++; | |
675 | } | |
676 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
677 | ||
678 | mark = oldmark; | |
db7c17d7 | 679 | items = sp - mark; |
79072805 LW |
680 | ++mark; |
681 | } | |
682 | ||
76f68e9b | 683 | sv_setpvs(sv, ""); |
e4803c42 | 684 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 685 | SvUTF8_off(sv); |
e4803c42 | 686 | |
284167a5 | 687 | if (TAINTING_get && SvMAGICAL(sv)) |
8d6d96c1 HS |
688 | SvTAINTED_off(sv); |
689 | ||
463ee0b2 | 690 | if (items-- > 0) { |
92d29cee JH |
691 | if (*mark) |
692 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
693 | mark++; |
694 | } | |
8d6d96c1 | 695 | |
c512ce4f | 696 | if (delimlen) { |
810a07df | 697 | const U32 delimflag = DO_UTF8(delim) ? SV_CATUTF8 : SV_CATBYTES; |
79072805 | 698 | for (; items > 0; items--,mark++) { |
f0ee3863 FC |
699 | STRLEN len; |
700 | const char *s; | |
810a07df | 701 | sv_catpvn_flags(sv,delims,delimlen,delimflag); |
f0ee3863 FC |
702 | s = SvPV_const(*mark,len); |
703 | sv_catpvn_flags(sv,s,len, | |
704 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
79072805 LW |
705 | } |
706 | } | |
707 | else { | |
708 | for (; items > 0; items--,mark++) | |
f0ee3863 FC |
709 | { |
710 | STRLEN len; | |
711 | const char *s = SvPV_const(*mark,len); | |
712 | sv_catpvn_flags(sv,s,len, | |
713 | DO_UTF8(*mark) ? SV_CATUTF8 : SV_CATBYTES); | |
714 | } | |
79072805 LW |
715 | } |
716 | SvSETMAGIC(sv); | |
717 | } | |
718 | ||
719 | void | |
864dbfa3 | 720 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 721 | { |
46fc3d4c | 722 | STRLEN patlen; |
53c1dcc0 | 723 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c | 724 | bool do_taint = FALSE; |
725 | ||
7918f24d NC |
726 | PERL_ARGS_ASSERT_DO_SPRINTF; |
727 | ||
e06d98fb DM |
728 | if (SvTAINTED(*sarg)) |
729 | TAINT_PROPER( | |
730 | (PL_op && PL_op->op_type < OP_max) | |
731 | ? (PL_op->op_type == OP_PRTF) | |
732 | ? "printf" | |
733 | : PL_op_name[PL_op->op_type] | |
734 | : "(unknown)" | |
735 | ); | |
5b781b5b | 736 | SvUTF8_off(sv); |
2cf2cfc6 A |
737 | if (DO_UTF8(*sarg)) |
738 | SvUTF8_on(sv); | |
4608196e | 739 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint); |
79072805 | 740 | SvSETMAGIC(sv); |
46fc3d4c | 741 | if (do_taint) |
742 | SvTAINTED_on(sv); | |
79072805 LW |
743 | } |
744 | ||
33b45480 | 745 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 | 746 | UV |
fcb8da00 | 747 | Perl_do_vecget(pTHX_ SV *sv, SSize_t offset, int size) |
81e118e0 | 748 | { |
bbb8a7e0 | 749 | STRLEN srclen, len, uoffset, bitoffs = 0; |
fc9668ae DM |
750 | const I32 svpv_flags = ((PL_op->op_flags & OPf_MOD || LVRET) |
751 | ? SV_UNDEF_RETURNS_NULL : 0); | |
752 | unsigned char *s = (unsigned char *) | |
753 | SvPV_flags(sv, srclen, (svpv_flags|SV_GMAGIC)); | |
81e118e0 JH |
754 | UV retnum = 0; |
755 | ||
032061d2 | 756 | if (!s) { |
fc9668ae | 757 | s = (unsigned char *)""; |
032061d2 BF |
758 | } |
759 | ||
7918f24d NC |
760 | PERL_ARGS_ASSERT_DO_VECGET; |
761 | ||
a50d7633 | 762 | if (offset < 0) |
61f9802b | 763 | return 0; |
8e84507e | 764 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 765 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 766 | |
fc9668ae | 767 | if (SvUTF8(sv)) { |
33b45480 | 768 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
fc9668ae DM |
769 | /* PVX may have changed */ |
770 | s = (unsigned char *) SvPV_flags(sv, srclen, svpv_flags); | |
771 | } | |
246fae53 | 772 | |
bbb8a7e0 MHM |
773 | if (size < 8) { |
774 | bitoffs = ((offset%8)*size)%8; | |
775 | uoffset = offset/(8/size); | |
776 | } | |
777 | else if (size > 8) | |
778 | uoffset = offset*(size/8); | |
779 | else | |
780 | uoffset = offset; | |
781 | ||
782 | len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
81e118e0 JH |
783 | if (len > srclen) { |
784 | if (size <= 8) | |
785 | retnum = 0; | |
786 | else { | |
81e118e0 | 787 | if (size == 16) { |
bb7a0f54 | 788 | if (uoffset >= srclen) |
81e118e0 JH |
789 | retnum = 0; |
790 | else | |
bb7a0f54 | 791 | retnum = (UV) s[uoffset] << 8; |
81e118e0 JH |
792 | } |
793 | else if (size == 32) { | |
bb7a0f54 | 794 | if (uoffset >= srclen) |
81e118e0 | 795 | retnum = 0; |
bb7a0f54 | 796 | else if (uoffset + 1 >= srclen) |
81e118e0 | 797 | retnum = |
bb7a0f54 MHM |
798 | ((UV) s[uoffset ] << 24); |
799 | else if (uoffset + 2 >= srclen) | |
81e118e0 | 800 | retnum = |
bb7a0f54 MHM |
801 | ((UV) s[uoffset ] << 24) + |
802 | ((UV) s[uoffset + 1] << 16); | |
81e118e0 JH |
803 | else |
804 | retnum = | |
bb7a0f54 MHM |
805 | ((UV) s[uoffset ] << 24) + |
806 | ((UV) s[uoffset + 1] << 16) + | |
807 | ( s[uoffset + 2] << 8); | |
81e118e0 | 808 | } |
d7d93a81 | 809 | #ifdef UV_IS_QUAD |
c5a0f51a | 810 | else if (size == 64) { |
a2a5de95 NC |
811 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
812 | "Bit vector size > 32 non-portable"); | |
bb7a0f54 | 813 | if (uoffset >= srclen) |
c5a0f51a | 814 | retnum = 0; |
bb7a0f54 | 815 | else if (uoffset + 1 >= srclen) |
c5a0f51a | 816 | retnum = |
bb7a0f54 MHM |
817 | (UV) s[uoffset ] << 56; |
818 | else if (uoffset + 2 >= srclen) | |
c5a0f51a | 819 | retnum = |
bb7a0f54 MHM |
820 | ((UV) s[uoffset ] << 56) + |
821 | ((UV) s[uoffset + 1] << 48); | |
822 | else if (uoffset + 3 >= srclen) | |
c5a0f51a | 823 | retnum = |
bb7a0f54 MHM |
824 | ((UV) s[uoffset ] << 56) + |
825 | ((UV) s[uoffset + 1] << 48) + | |
826 | ((UV) s[uoffset + 2] << 40); | |
827 | else if (uoffset + 4 >= srclen) | |
c5a0f51a | 828 | retnum = |
bb7a0f54 MHM |
829 | ((UV) s[uoffset ] << 56) + |
830 | ((UV) s[uoffset + 1] << 48) + | |
831 | ((UV) s[uoffset + 2] << 40) + | |
832 | ((UV) s[uoffset + 3] << 32); | |
833 | else if (uoffset + 5 >= srclen) | |
c5a0f51a | 834 | retnum = |
bb7a0f54 MHM |
835 | ((UV) s[uoffset ] << 56) + |
836 | ((UV) s[uoffset + 1] << 48) + | |
837 | ((UV) s[uoffset + 2] << 40) + | |
838 | ((UV) s[uoffset + 3] << 32) + | |
e7aca353 | 839 | ((UV) s[uoffset + 4] << 24); |
bb7a0f54 | 840 | else if (uoffset + 6 >= srclen) |
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) + | |
846 | ((UV) s[uoffset + 4] << 24) + | |
847 | ((UV) s[uoffset + 5] << 16); | |
c5a0f51a | 848 | else |
8e84507e | 849 | retnum = |
bb7a0f54 MHM |
850 | ((UV) s[uoffset ] << 56) + |
851 | ((UV) s[uoffset + 1] << 48) + | |
852 | ((UV) s[uoffset + 2] << 40) + | |
853 | ((UV) s[uoffset + 3] << 32) + | |
854 | ((UV) s[uoffset + 4] << 24) + | |
855 | ((UV) s[uoffset + 5] << 16) + | |
e7aca353 | 856 | ((UV) s[uoffset + 6] << 8); |
c5a0f51a JH |
857 | } |
858 | #endif | |
81e118e0 JH |
859 | } |
860 | } | |
861 | else if (size < 8) | |
bbb8a7e0 | 862 | retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1); |
81e118e0 | 863 | else { |
81e118e0 | 864 | if (size == 8) |
bb7a0f54 | 865 | retnum = s[uoffset]; |
81e118e0 JH |
866 | else if (size == 16) |
867 | retnum = | |
bb7a0f54 MHM |
868 | ((UV) s[uoffset] << 8) + |
869 | s[uoffset + 1]; | |
81e118e0 JH |
870 | else if (size == 32) |
871 | retnum = | |
bb7a0f54 MHM |
872 | ((UV) s[uoffset ] << 24) + |
873 | ((UV) s[uoffset + 1] << 16) + | |
874 | ( s[uoffset + 2] << 8) + | |
875 | s[uoffset + 3]; | |
d7d93a81 | 876 | #ifdef UV_IS_QUAD |
c5a0f51a | 877 | else if (size == 64) { |
a2a5de95 NC |
878 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
879 | "Bit vector size > 32 non-portable"); | |
c5a0f51a | 880 | retnum = |
bb7a0f54 MHM |
881 | ((UV) s[uoffset ] << 56) + |
882 | ((UV) s[uoffset + 1] << 48) + | |
883 | ((UV) s[uoffset + 2] << 40) + | |
884 | ((UV) s[uoffset + 3] << 32) + | |
885 | ((UV) s[uoffset + 4] << 24) + | |
886 | ((UV) s[uoffset + 5] << 16) + | |
887 | ( s[uoffset + 6] << 8) + | |
888 | s[uoffset + 7]; | |
c5a0f51a JH |
889 | } |
890 | #endif | |
81e118e0 JH |
891 | } |
892 | ||
893 | return retnum; | |
894 | } | |
895 | ||
33b45480 SB |
896 | /* currently converts input to bytes if possible but doesn't sweat failures, |
897 | * although it does ensure that the string it clobbers is not marked as | |
898 | * utf8-valid any more | |
899 | */ | |
79072805 | 900 | void |
864dbfa3 | 901 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 | 902 | { |
eb578fdb KW |
903 | SSize_t offset, bitoffs = 0; |
904 | int size; | |
905 | unsigned char *s; | |
906 | UV lval; | |
79072805 | 907 | I32 mask; |
a0d0e21e LW |
908 | STRLEN targlen; |
909 | STRLEN len; | |
c4420975 | 910 | SV * const targ = LvTARG(sv); |
79072805 | 911 | |
7918f24d NC |
912 | PERL_ARGS_ASSERT_DO_VECSET; |
913 | ||
8990e307 LW |
914 | if (!targ) |
915 | return; | |
032061d2 BF |
916 | s = (unsigned char*)SvPV_force_flags(targ, targlen, |
917 | SV_GMAGIC | SV_UNDEF_RETURNS_NULL); | |
246fae53 | 918 | if (SvUTF8(targ)) { |
33b45480 SB |
919 | /* This is handled by the SvPOK_only below... |
920 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
921 | SvUTF8_off(targ); | |
922 | */ | |
923 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
924 | } |
925 | ||
4ebbc975 | 926 | (void)SvPOK_only(targ); |
81e118e0 | 927 | lval = SvUV(sv); |
79072805 | 928 | offset = LvTARGOFF(sv); |
fe58ced6 | 929 | if (offset < 0) |
ed9aa3b7 | 930 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 931 | size = LvTARGLEN(sv); |
8e84507e | 932 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 933 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 934 | |
bbb8a7e0 MHM |
935 | if (size < 8) { |
936 | bitoffs = ((offset%8)*size)%8; | |
937 | offset /= 8/size; | |
938 | } | |
939 | else if (size > 8) | |
940 | offset *= size/8; | |
941 | ||
942 | len = offset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
a0d0e21e LW |
943 | if (len > targlen) { |
944 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 945 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
946 | SvCUR_set(targ, len); |
947 | } | |
8e84507e | 948 | |
79072805 LW |
949 | if (size < 8) { |
950 | mask = (1 << size) - 1; | |
79072805 | 951 | lval &= mask; |
bbb8a7e0 MHM |
952 | s[offset] &= ~(mask << bitoffs); |
953 | s[offset] |= lval << bitoffs; | |
79072805 LW |
954 | } |
955 | else { | |
956 | if (size == 8) | |
eb160463 | 957 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 958 | else if (size == 16) { |
eb160463 GS |
959 | s[offset ] = (U8)((lval >> 8) & 0xff); |
960 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
961 | } |
962 | else if (size == 32) { | |
eb160463 GS |
963 | s[offset ] = (U8)((lval >> 24) & 0xff); |
964 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
965 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
966 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 967 | } |
d7d93a81 | 968 | #ifdef UV_IS_QUAD |
c5a0f51a | 969 | else if (size == 64) { |
a2a5de95 NC |
970 | Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE), |
971 | "Bit vector size > 32 non-portable"); | |
eb160463 GS |
972 | s[offset ] = (U8)((lval >> 56) & 0xff); |
973 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
974 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
975 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
976 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
977 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
978 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
979 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 980 | } |
dc1e3f56 | 981 | #endif |
79072805 | 982 | } |
7bb043c3 | 983 | SvSETMAGIC(targ); |
79072805 LW |
984 | } |
985 | ||
986 | void | |
864dbfa3 | 987 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 LW |
988 | { |
989 | #ifdef LIBERAL | |
eb578fdb KW |
990 | long *dl; |
991 | long *ll; | |
992 | long *rl; | |
79072805 | 993 | #endif |
eb578fdb | 994 | char *dc; |
463ee0b2 LW |
995 | STRLEN leftlen; |
996 | STRLEN rightlen; | |
eb578fdb KW |
997 | const char *lc; |
998 | const char *rc; | |
999 | STRLEN len; | |
bb7a0f54 | 1000 | STRLEN lensave; |
e62f0680 NC |
1001 | const char *lsave; |
1002 | const char *rsave; | |
12abf4f0 NC |
1003 | bool left_utf; |
1004 | bool right_utf; | |
bb7a0f54 | 1005 | STRLEN needlen = 0; |
0c57e439 | 1006 | |
7918f24d | 1007 | PERL_ARGS_ASSERT_DO_VOP; |
79072805 | 1008 | |
6b349a5c | 1009 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv))) |
76f68e9b | 1010 | sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */ |
8c8eee82 BM |
1011 | if (sv == left) { |
1012 | lsave = lc = SvPV_force_nomg(left, leftlen); | |
1013 | } | |
1014 | else { | |
1015 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1016 | SvPV_force_nomg_nolen(sv); | |
1017 | } | |
e62f0680 | 1018 | rsave = rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 NC |
1019 | |
1020 | /* This need to come after SvPV to ensure that string overloading has | |
1021 | fired off. */ | |
1022 | ||
1023 | left_utf = DO_UTF8(left); | |
1024 | right_utf = DO_UTF8(right); | |
1025 | ||
1026 | if (left_utf && !right_utf) { | |
1027 | /* Avoid triggering overloading again by using temporaries. | |
1028 | Maybe there should be a variant of sv_utf8_upgrade that takes pvn | |
1029 | */ | |
59cd0e26 | 1030 | right = newSVpvn_flags(rsave, rightlen, SVs_TEMP); |
12abf4f0 NC |
1031 | sv_utf8_upgrade(right); |
1032 | rsave = rc = SvPV_nomg_const(right, rightlen); | |
1033 | right_utf = TRUE; | |
1034 | } | |
1035 | else if (!left_utf && right_utf) { | |
59cd0e26 | 1036 | left = newSVpvn_flags(lsave, leftlen, SVs_TEMP); |
12abf4f0 NC |
1037 | sv_utf8_upgrade(left); |
1038 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1039 | left_utf = TRUE; | |
1040 | } | |
1041 | ||
93a17b20 | 1042 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1043 | lensave = len; |
9fdd7463 JH |
1044 | SvCUR_set(sv, len); |
1045 | (void)SvPOK_only(sv); | |
c9b3c8d0 JH |
1046 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1047 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
a02a5408 | 1048 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 JH |
1049 | } |
1050 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2596d9fe | 1051 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 MHM |
1052 | if (SvLEN(sv) < len + 1) { |
1053 | dc = SvGROW(sv, len + 1); | |
ff68c719 | 1054 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1055 | } | |
c9b3c8d0 JH |
1056 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1057 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1058 | } |
1059 | else { | |
bb7a0f54 MHM |
1060 | needlen = optype == OP_BIT_AND |
1061 | ? len : (leftlen > rightlen ? leftlen : rightlen); | |
a02a5408 | 1062 | Newxz(dc, needlen + 1, char); |
aa0a69cb | 1063 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 | 1064 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 | 1065 | } |
0c57e439 GS |
1066 | if (left_utf || right_utf) { |
1067 | UV duc, luc, ruc; | |
794a0d33 JH |
1068 | char *dcorig = dc; |
1069 | char *dcsave = NULL; | |
0c57e439 GS |
1070 | STRLEN lulen = leftlen; |
1071 | STRLEN rulen = rightlen; | |
ba210ebe | 1072 | STRLEN ulen; |
0c57e439 GS |
1073 | |
1074 | switch (optype) { | |
1075 | case OP_BIT_AND: | |
1076 | while (lulen && rulen) { | |
9041c2e3 | 1077 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1078 | lc += ulen; |
1079 | lulen -= ulen; | |
9041c2e3 | 1080 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1081 | rc += ulen; |
1082 | rulen -= ulen; | |
1083 | duc = luc & ruc; | |
9041c2e3 | 1084 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1085 | } |
3ce4c532 NC |
1086 | if (sv == left || sv == right) |
1087 | (void)sv_usepvn(sv, dcorig, needlen); | |
794a0d33 | 1088 | SvCUR_set(sv, dc - dcorig); |
0c57e439 GS |
1089 | break; |
1090 | case OP_BIT_XOR: | |
1091 | while (lulen && rulen) { | |
9041c2e3 | 1092 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1093 | lc += ulen; |
1094 | lulen -= ulen; | |
9041c2e3 | 1095 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1096 | rc += ulen; |
1097 | rulen -= ulen; | |
1098 | duc = luc ^ ruc; | |
9041c2e3 | 1099 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1100 | } |
1101 | goto mop_up_utf; | |
1102 | case OP_BIT_OR: | |
1103 | while (lulen && rulen) { | |
9041c2e3 | 1104 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1105 | lc += ulen; |
1106 | lulen -= ulen; | |
9041c2e3 | 1107 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1108 | rc += ulen; |
1109 | rulen -= ulen; | |
1110 | duc = luc | ruc; | |
9041c2e3 | 1111 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1112 | } |
1113 | mop_up_utf: | |
794a0d33 JH |
1114 | if (rulen) |
1115 | dcsave = savepvn(rc, rulen); | |
1116 | else if (lulen) | |
1117 | dcsave = savepvn(lc, lulen); | |
3ce4c532 | 1118 | if (sv == left || sv == right) |
4bac9ae4 | 1119 | (void)sv_usepvn(sv, dcorig, needlen); /* uses Renew(); defaults to nomg */ |
794a0d33 | 1120 | SvCUR_set(sv, dc - dcorig); |
0c57e439 | 1121 | if (rulen) |
4bac9ae4 | 1122 | sv_catpvn_nomg(sv, dcsave, rulen); |
0c57e439 | 1123 | else if (lulen) |
4bac9ae4 | 1124 | sv_catpvn_nomg(sv, dcsave, lulen); |
0c57e439 GS |
1125 | else |
1126 | *SvEND(sv) = '\0'; | |
794a0d33 | 1127 | Safefree(dcsave); |
0c57e439 | 1128 | break; |
41547412 NC |
1129 | default: |
1130 | if (sv == left || sv == right) | |
1131 | Safefree(dcorig); | |
c395bd6c AL |
1132 | Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)", |
1133 | (unsigned)optype, PL_op_name[optype]); | |
0c57e439 GS |
1134 | } |
1135 | SvUTF8_on(sv); | |
1136 | goto finish; | |
1137 | } | |
1138 | else | |
79072805 LW |
1139 | #ifdef LIBERAL |
1140 | if (len >= sizeof(long)*4 && | |
bb7a0f54 MHM |
1141 | !((unsigned long)dc % sizeof(long)) && |
1142 | !((unsigned long)lc % sizeof(long)) && | |
1143 | !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ | |
79072805 | 1144 | { |
bb7a0f54 | 1145 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 LW |
1146 | len /= (sizeof(long)*4); |
1147 | ||
1148 | dl = (long*)dc; | |
1149 | ll = (long*)lc; | |
1150 | rl = (long*)rc; | |
1151 | ||
1152 | switch (optype) { | |
1153 | case OP_BIT_AND: | |
1154 | while (len--) { | |
1155 | *dl++ = *ll++ & *rl++; | |
1156 | *dl++ = *ll++ & *rl++; | |
1157 | *dl++ = *ll++ & *rl++; | |
1158 | *dl++ = *ll++ & *rl++; | |
1159 | } | |
1160 | break; | |
a0d0e21e | 1161 | case OP_BIT_XOR: |
79072805 LW |
1162 | while (len--) { |
1163 | *dl++ = *ll++ ^ *rl++; | |
1164 | *dl++ = *ll++ ^ *rl++; | |
1165 | *dl++ = *ll++ ^ *rl++; | |
1166 | *dl++ = *ll++ ^ *rl++; | |
1167 | } | |
1168 | break; | |
1169 | case OP_BIT_OR: | |
1170 | while (len--) { | |
1171 | *dl++ = *ll++ | *rl++; | |
1172 | *dl++ = *ll++ | *rl++; | |
1173 | *dl++ = *ll++ | *rl++; | |
1174 | *dl++ = *ll++ | *rl++; | |
1175 | } | |
1176 | } | |
1177 | ||
1178 | dc = (char*)dl; | |
1179 | lc = (char*)ll; | |
1180 | rc = (char*)rl; | |
1181 | ||
1182 | len = remainder; | |
1183 | } | |
1184 | #endif | |
a0d0e21e | 1185 | { |
a0d0e21e LW |
1186 | switch (optype) { |
1187 | case OP_BIT_AND: | |
1188 | while (len--) | |
1189 | *dc++ = *lc++ & *rc++; | |
1a787b95 | 1190 | *dc = '\0'; |
a0d0e21e LW |
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; | |
bb7a0f54 | 1201 | if (rightlen > len) |
4bac9ae4 | 1202 | sv_catpvn_nomg(sv, rsave + len, rightlen - len); |
eb160463 | 1203 | else if (leftlen > (STRLEN)len) |
4bac9ae4 | 1204 | sv_catpvn_nomg(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1205 | else |
1206 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1207 | break; |
1208 | } | |
79072805 | 1209 | } |
7b52d656 | 1210 | finish: |
fb73857a | 1211 | SvTAINT(sv); |
79072805 | 1212 | } |
463ee0b2 | 1213 | |
b1c05ba5 DM |
1214 | |
1215 | /* used for: pp_keys(), pp_values() */ | |
1216 | ||
463ee0b2 | 1217 | OP * |
cea2e8a9 | 1218 | Perl_do_kv(pTHX) |
463ee0b2 | 1219 | { |
39644a26 | 1220 | dSP; |
73ff03e8 | 1221 | HV * const keys = MUTABLE_HV(POPs); |
eb578fdb | 1222 | HE *entry; |
052a7c76 | 1223 | SSize_t extend_size; |
f54cb97a AL |
1224 | const I32 gimme = GIMME_V; |
1225 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); | |
cba5a3b0 | 1226 | /* op_type is OP_RKEYS/OP_RVALUES if pp_rkeys delegated to here */ |
26230909 AC |
1227 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); |
1228 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); | |
463ee0b2 | 1229 | |
800e9ae0 | 1230 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1231 | |
54310121 | 1232 | if (gimme == G_VOID) |
aa689395 | 1233 | RETURN; |
1234 | ||
54310121 | 1235 | if (gimme == G_SCALAR) { |
78f9721b | 1236 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
2154eca7 EB |
1237 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
1238 | sv_magic(ret, NULL, PERL_MAGIC_nkeys, NULL, 0); | |
1239 | LvTYPE(ret) = 'k'; | |
1240 | LvTARG(ret) = SvREFCNT_inc_simple(keys); | |
1241 | PUSHs(ret); | |
81714fb9 | 1242 | } |
463ee0b2 | 1243 | else { |
2154eca7 EB |
1244 | IV i; |
1245 | dTARGET; | |
1246 | ||
1247 | if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) { | |
1b95d04f | 1248 | i = HvUSEDKEYS(keys); |
2154eca7 EB |
1249 | } |
1250 | else { | |
1251 | i = 0; | |
1252 | while (hv_iternext(keys)) i++; | |
1253 | } | |
1254 | PUSHi( i ); | |
463ee0b2 | 1255 | } |
463ee0b2 LW |
1256 | RETURN; |
1257 | } | |
1258 | ||
052a7c76 DM |
1259 | /* 2*HvUSEDKEYS() should never be big enough to truncate or wrap */ |
1260 | assert(HvUSEDKEYS(keys) <= (SSize_t_MAX >> 1)); | |
1261 | extend_size = (SSize_t)HvUSEDKEYS(keys) * (dokeys + dovalues); | |
1262 | EXTEND(SP, extend_size); | |
463ee0b2 | 1263 | |
155aba94 | 1264 | while ((entry = hv_iternext(keys))) { |
574c8022 | 1265 | if (dokeys) { |
890ce7af | 1266 | SV* const sv = hv_iterkeysv(entry); |
2b32fed8 | 1267 | XPUSHs(sv); |
574c8022 | 1268 | } |
463ee0b2 | 1269 | if (dovalues) { |
86f45028 | 1270 | SV *tmpstr = hv_iterval(keys,entry); |
cea2e8a9 | 1271 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1272 | (unsigned long)HeHASH(entry), |
cdd3ba14 | 1273 | (int)HvMAX(keys)+1, |
800e9ae0 | 1274 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
46fc3d4c | 1275 | XPUSHs(tmpstr); |
463ee0b2 | 1276 | } |
463ee0b2 | 1277 | } |
2b32fed8 | 1278 | RETURN; |
463ee0b2 | 1279 | } |
4e35701f | 1280 | |
af3babe4 | 1281 | /* |
14d04a33 | 1282 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 1283 | */ |