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; |
c395bd6c AL |
36 | U8 *s = (U8*)SvPV(sv,len); |
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; |
c395bd6c AL |
104 | const U8 *s = (const U8*)SvPV_const(sv, len); |
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 AL |
141 | STRLEN len; |
142 | U8 *s = (U8*)SvPV(sv, len); | |
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 | 205 | |
4aecb5b5 JH |
206 | #ifdef MACOS_TRADITIONAL |
207 | #define comp CoMP /* "comp" is a keyword in some compilers ... */ | |
208 | #endif | |
209 | ||
1aa99e6b | 210 | if (PL_op->op_private & OPpTRANS_SQUASH) { |
1aa99e6b IH |
211 | UV pch = 0xfeedface; |
212 | while (s < send) { | |
213 | STRLEN len; | |
9f7f3913 TS |
214 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
215 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 216 | I32 ch; |
1aa99e6b | 217 | |
8973db79 JH |
218 | if (comp > 0xff) { |
219 | if (!complement) { | |
640af7f1 | 220 | Move(s, d, len, U8); |
8973db79 JH |
221 | d += len; |
222 | } | |
223 | else { | |
224 | matches++; | |
225 | if (!del) { | |
bb7a0f54 | 226 | ch = (rlen == 0) ? (I32)comp : |
05d340b8 | 227 | (comp - 0x100 < rlen) ? |
8973db79 | 228 | tbl[comp+1] : tbl[0x100+rlen]; |
eb160463 | 229 | if ((UV)ch != pch) { |
9041c2e3 | 230 | d = uvchr_to_utf8(d, ch); |
eb160463 | 231 | pch = (UV)ch; |
8973db79 JH |
232 | } |
233 | s += len; | |
234 | continue; | |
235 | } | |
236 | } | |
9b877dbb | 237 | } |
1aa99e6b IH |
238 | else if ((ch = tbl[comp]) >= 0) { |
239 | matches++; | |
eb160463 | 240 | if ((UV)ch != pch) { |
9041c2e3 | 241 | d = uvchr_to_utf8(d, ch); |
eb160463 | 242 | pch = (UV)ch; |
1aa99e6b IH |
243 | } |
244 | s += len; | |
245 | continue; | |
246 | } | |
9b877dbb | 247 | else if (ch == -1) { /* -1 is unmapped character */ |
640af7f1 | 248 | Move(s, d, len, U8); |
9b877dbb IH |
249 | d += len; |
250 | } | |
1aa99e6b IH |
251 | else if (ch == -2) /* -2 is delete character */ |
252 | matches++; | |
253 | s += len; | |
254 | pch = 0xfeedface; | |
255 | } | |
5d06d08e | 256 | } |
1aa99e6b IH |
257 | else { |
258 | while (s < send) { | |
259 | STRLEN len; | |
9f7f3913 TS |
260 | const UV comp = utf8n_to_uvchr(s, send - s, &len, |
261 | UTF8_ALLOW_DEFAULT); | |
f54cb97a | 262 | I32 ch; |
8973db79 JH |
263 | if (comp > 0xff) { |
264 | if (!complement) { | |
0376ff32 | 265 | Move(s, d, len, U8); |
8973db79 JH |
266 | d += len; |
267 | } | |
268 | else { | |
269 | matches++; | |
270 | if (!del) { | |
9041c2e3 NIS |
271 | if (comp - 0x100 < rlen) |
272 | d = uvchr_to_utf8(d, tbl[comp+1]); | |
8973db79 | 273 | else |
9041c2e3 | 274 | d = uvchr_to_utf8(d, tbl[0x100+rlen]); |
8973db79 JH |
275 | } |
276 | } | |
9b877dbb | 277 | } |
1aa99e6b | 278 | else if ((ch = tbl[comp]) >= 0) { |
9041c2e3 | 279 | d = uvchr_to_utf8(d, ch); |
1aa99e6b IH |
280 | matches++; |
281 | } | |
282 | else if (ch == -1) { /* -1 is unmapped character */ | |
640af7f1 | 283 | Move(s, d, len, U8); |
9b877dbb | 284 | d += len; |
1aa99e6b IH |
285 | } |
286 | else if (ch == -2) /* -2 is delete character */ | |
287 | matches++; | |
288 | s += len; | |
289 | } | |
290 | } | |
9b877dbb IH |
291 | if (grows) { |
292 | sv_setpvn(sv, (char*)dstart, d - dstart); | |
293 | Safefree(dstart); | |
294 | } | |
295 | else { | |
296 | *d = '\0'; | |
297 | SvCUR_set(sv, d - dstart); | |
298 | } | |
1aa99e6b | 299 | SvUTF8_on(sv); |
4757a243 | 300 | } |
5e44153e | 301 | SvSETMAGIC(sv); |
4757a243 LW |
302 | return matches; |
303 | } | |
304 | ||
942e002e | 305 | STATIC I32 |
c395bd6c | 306 | S_do_trans_simple_utf8(pTHX_ SV * const sv) |
4757a243 | 307 | { |
97aff369 | 308 | dVAR; |
4757a243 LW |
309 | U8 *s; |
310 | U8 *send; | |
311 | U8 *d; | |
036b4402 | 312 | U8 *start; |
1aa99e6b | 313 | U8 *dstart, *dend; |
4757a243 | 314 | I32 matches = 0; |
f54cb97a | 315 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; |
4757a243 | 316 | STRLEN len; |
043e41b8 DM |
317 | SV* const rv = |
318 | #ifdef USE_ITHREADS | |
319 | PAD_SVl(cPADOP->op_padix); | |
320 | #else | |
ad64d0ec | 321 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 322 | #endif |
85fbaab2 | 323 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 324 | SV* const * svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
325 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
326 | const UV extra = none + 1; | |
b7953727 | 327 | UV final = 0; |
1aa99e6b | 328 | U8 hibit = 0; |
4757a243 | 329 | |
7918f24d NC |
330 | PERL_ARGS_ASSERT_DO_TRANS_SIMPLE_UTF8; |
331 | ||
4757a243 | 332 | s = (U8*)SvPV(sv, len); |
c395bd6c | 333 | if (!SvUTF8(sv)) { |
c4420975 AL |
334 | const U8 *t = s; |
335 | const U8 * const e = s + len; | |
c4d5f83a | 336 | while (t < e) { |
f54cb97a | 337 | const U8 ch = *t++; |
fabdb6c0 AL |
338 | hibit = !NATIVE_IS_INVARIANT(ch); |
339 | if (hibit) { | |
340 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 341 | break; |
fabdb6c0 | 342 | } |
c4d5f83a | 343 | } |
1aa99e6b | 344 | } |
4757a243 | 345 | send = s + len; |
036b4402 | 346 | start = s; |
4757a243 | 347 | |
a4fc7abc | 348 | svp = hv_fetchs(hv, "FINAL", FALSE); |
4757a243 LW |
349 | if (svp) |
350 | final = SvUV(*svp); | |
351 | ||
9b877dbb IH |
352 | if (grows) { |
353 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 354 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
355 | dend = d + len * 3; |
356 | dstart = d; | |
357 | } | |
358 | else { | |
359 | dstart = d = s; | |
360 | dend = d + len; | |
361 | } | |
1aa99e6b | 362 | |
4757a243 | 363 | while (s < send) { |
0bcc34c2 AL |
364 | const UV uv = swash_fetch(rv, s, TRUE); |
365 | if (uv < none) { | |
4757a243 LW |
366 | s += UTF8SKIP(s); |
367 | matches++; | |
7d85a32c | 368 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
369 | } |
370 | else if (uv == none) { | |
4373e329 | 371 | const int i = UTF8SKIP(s); |
0376ff32 | 372 | Move(s, d, i, U8); |
9b877dbb IH |
373 | d += i; |
374 | s += i; | |
4757a243 LW |
375 | } |
376 | else if (uv == extra) { | |
4373e329 | 377 | s += UTF8SKIP(s); |
4757a243 | 378 | matches++; |
7d85a32c | 379 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
380 | } |
381 | else | |
382 | s += UTF8SKIP(s); | |
1aa99e6b | 383 | |
9b877dbb | 384 | if (d > dend) { |
4373e329 AL |
385 | const STRLEN clen = d - dstart; |
386 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 387 | if (!grows) |
7d85a32c | 388 | Perl_croak(aTHX_ "panic: do_trans_simple_utf8 line %d",__LINE__); |
89ebb4a3 | 389 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
390 | d = dstart + clen; |
391 | dend = dstart + nlen; | |
392 | } | |
4757a243 | 393 | } |
16ec844d | 394 | if (grows || hibit) { |
9b877dbb IH |
395 | sv_setpvn(sv, (char*)dstart, d - dstart); |
396 | Safefree(dstart); | |
16ec844d IH |
397 | if (grows && hibit) |
398 | Safefree(start); | |
9b877dbb IH |
399 | } |
400 | else { | |
401 | *d = '\0'; | |
402 | SvCUR_set(sv, d - dstart); | |
403 | } | |
4757a243 | 404 | SvSETMAGIC(sv); |
1aa99e6b | 405 | SvUTF8_on(sv); |
4757a243 LW |
406 | |
407 | return matches; | |
408 | } | |
409 | ||
942e002e | 410 | STATIC I32 |
c395bd6c | 411 | S_do_trans_count_utf8(pTHX_ SV * const sv) |
4757a243 | 412 | { |
97aff369 | 413 | dVAR; |
93524f2b | 414 | const U8 *s; |
cbbf8932 AL |
415 | const U8 *start = NULL; |
416 | const U8 *send; | |
4757a243 LW |
417 | I32 matches = 0; |
418 | STRLEN len; | |
043e41b8 DM |
419 | SV* const rv = |
420 | #ifdef USE_ITHREADS | |
421 | PAD_SVl(cPADOP->op_padix); | |
422 | #else | |
ad64d0ec | 423 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 424 | #endif |
85fbaab2 | 425 | HV* const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 426 | SV* const * const svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
427 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
428 | const UV extra = none + 1; | |
1aa99e6b | 429 | U8 hibit = 0; |
4757a243 | 430 | |
7918f24d NC |
431 | PERL_ARGS_ASSERT_DO_TRANS_COUNT_UTF8; |
432 | ||
93524f2b | 433 | s = (const U8*)SvPV_const(sv, len); |
1aa99e6b | 434 | if (!SvUTF8(sv)) { |
4373e329 | 435 | const U8 *t = s; |
c4420975 | 436 | const U8 * const e = s + len; |
c4d5f83a | 437 | while (t < e) { |
4373e329 | 438 | const U8 ch = *t++; |
fabdb6c0 AL |
439 | hibit = !NATIVE_IS_INVARIANT(ch); |
440 | if (hibit) { | |
441 | start = s = bytes_to_utf8(s, &len); | |
1aa99e6b | 442 | break; |
fabdb6c0 | 443 | } |
c4d5f83a | 444 | } |
1aa99e6b | 445 | } |
4757a243 LW |
446 | send = s + len; |
447 | ||
448 | while (s < send) { | |
c4420975 AL |
449 | const UV uv = swash_fetch(rv, s, TRUE); |
450 | if (uv < none || uv == extra) | |
4757a243 | 451 | matches++; |
834a4ddd | 452 | s += UTF8SKIP(s); |
4757a243 | 453 | } |
1aa99e6b IH |
454 | if (hibit) |
455 | Safefree(start); | |
4757a243 LW |
456 | |
457 | return matches; | |
458 | } | |
459 | ||
942e002e | 460 | STATIC I32 |
c395bd6c | 461 | S_do_trans_complex_utf8(pTHX_ SV * const sv) |
4757a243 | 462 | { |
97aff369 | 463 | dVAR; |
1aa99e6b | 464 | U8 *start, *send; |
4757a243 LW |
465 | U8 *d; |
466 | I32 matches = 0; | |
4373e329 AL |
467 | const I32 squash = PL_op->op_private & OPpTRANS_SQUASH; |
468 | const I32 del = PL_op->op_private & OPpTRANS_DELETE; | |
469 | const I32 grows = PL_op->op_private & OPpTRANS_GROWS; | |
043e41b8 DM |
470 | SV* const rv = |
471 | #ifdef USE_ITHREADS | |
472 | PAD_SVl(cPADOP->op_padix); | |
473 | #else | |
ad64d0ec | 474 | MUTABLE_SV(cSVOP->op_sv); |
043e41b8 | 475 | #endif |
85fbaab2 | 476 | HV * const hv = MUTABLE_HV(SvRV(rv)); |
a4fc7abc | 477 | SV * const *svp = hv_fetchs(hv, "NONE", FALSE); |
f54cb97a AL |
478 | const UV none = svp ? SvUV(*svp) : 0x7fffffff; |
479 | const UV extra = none + 1; | |
b7953727 | 480 | UV final = 0; |
45005bfb | 481 | bool havefinal = FALSE; |
4757a243 | 482 | STRLEN len; |
1aa99e6b | 483 | U8 *dstart, *dend; |
1aa99e6b | 484 | U8 hibit = 0; |
f54cb97a | 485 | U8 *s = (U8*)SvPV(sv, len); |
7918f24d NC |
486 | |
487 | PERL_ARGS_ASSERT_DO_TRANS_COMPLEX_UTF8; | |
488 | ||
c395bd6c | 489 | if (!SvUTF8(sv)) { |
53c1dcc0 AL |
490 | const U8 *t = s; |
491 | const U8 * const e = s + len; | |
c4d5f83a | 492 | while (t < e) { |
f54cb97a | 493 | const U8 ch = *t++; |
fabdb6c0 AL |
494 | hibit = !NATIVE_IS_INVARIANT(ch); |
495 | if (hibit) { | |
496 | s = bytes_to_utf8(s, &len); | |
1aa99e6b | 497 | break; |
fabdb6c0 | 498 | } |
c4d5f83a | 499 | } |
1aa99e6b | 500 | } |
4757a243 | 501 | send = s + len; |
1aa99e6b | 502 | start = s; |
4757a243 | 503 | |
a4fc7abc | 504 | svp = hv_fetchs(hv, "FINAL", FALSE); |
45005bfb | 505 | if (svp) { |
4757a243 | 506 | final = SvUV(*svp); |
45005bfb JH |
507 | havefinal = TRUE; |
508 | } | |
4757a243 | 509 | |
9b877dbb IH |
510 | if (grows) { |
511 | /* d needs to be bigger than s, in case e.g. upgrading is required */ | |
a02a5408 | 512 | Newx(d, len * 3 + UTF8_MAXBYTES, U8); |
9b877dbb IH |
513 | dend = d + len * 3; |
514 | dstart = d; | |
515 | } | |
516 | else { | |
517 | dstart = d = s; | |
518 | dend = d + len; | |
519 | } | |
4757a243 LW |
520 | |
521 | if (squash) { | |
522 | UV puv = 0xfeedface; | |
523 | while (s < send) { | |
f54cb97a | 524 | UV uv = swash_fetch(rv, s, TRUE); |
9041c2e3 | 525 | |
9b877dbb | 526 | if (d > dend) { |
4373e329 AL |
527 | const STRLEN clen = d - dstart; |
528 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 529 | if (!grows) |
7d85a32c | 530 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 531 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
532 | d = dstart + clen; |
533 | dend = dstart + nlen; | |
4757a243 LW |
534 | } |
535 | if (uv < none) { | |
536 | matches++; | |
94472101 | 537 | s += UTF8SKIP(s); |
4757a243 | 538 | if (uv != puv) { |
7d85a32c | 539 | d = uvuni_to_utf8(d, uv); |
4757a243 LW |
540 | puv = uv; |
541 | } | |
4757a243 LW |
542 | continue; |
543 | } | |
544 | else if (uv == none) { /* "none" is unmapped character */ | |
4373e329 | 545 | const int i = UTF8SKIP(s); |
0376ff32 | 546 | Move(s, d, i, U8); |
9b877dbb IH |
547 | d += i; |
548 | s += i; | |
4757a243 LW |
549 | puv = 0xfeedface; |
550 | continue; | |
a0ed51b3 | 551 | } |
4757a243 LW |
552 | else if (uv == extra && !del) { |
553 | matches++; | |
45005bfb JH |
554 | if (havefinal) { |
555 | s += UTF8SKIP(s); | |
556 | if (puv != final) { | |
7d85a32c | 557 | d = uvuni_to_utf8(d, final); |
45005bfb JH |
558 | puv = final; |
559 | } | |
560 | } | |
561 | else { | |
562 | STRLEN len; | |
9f7f3913 | 563 | uv = utf8n_to_uvuni(s, send - s, &len, UTF8_ALLOW_DEFAULT); |
45005bfb | 564 | if (uv != puv) { |
0376ff32 | 565 | Move(s, d, len, U8); |
45005bfb JH |
566 | d += len; |
567 | puv = uv; | |
568 | } | |
569 | s += len; | |
4757a243 | 570 | } |
4757a243 LW |
571 | continue; |
572 | } | |
01ec43d0 GS |
573 | matches++; /* "none+1" is delete character */ |
574 | s += UTF8SKIP(s); | |
a0ed51b3 | 575 | } |
79072805 LW |
576 | } |
577 | else { | |
4757a243 | 578 | while (s < send) { |
f54cb97a | 579 | const UV uv = swash_fetch(rv, s, TRUE); |
9b877dbb | 580 | if (d > dend) { |
f54cb97a AL |
581 | const STRLEN clen = d - dstart; |
582 | const STRLEN nlen = dend - dstart + len + UTF8_MAXBYTES; | |
9b877dbb | 583 | if (!grows) |
7d85a32c | 584 | Perl_croak(aTHX_ "panic: do_trans_complex_utf8 line %d",__LINE__); |
89ebb4a3 | 585 | Renew(dstart, nlen + UTF8_MAXBYTES, U8); |
1aa99e6b IH |
586 | d = dstart + clen; |
587 | dend = dstart + nlen; | |
a0ed51b3 | 588 | } |
4757a243 LW |
589 | if (uv < none) { |
590 | matches++; | |
01ec43d0 | 591 | s += UTF8SKIP(s); |
7d85a32c | 592 | d = uvuni_to_utf8(d, uv); |
4757a243 | 593 | continue; |
a0ed51b3 | 594 | } |
4757a243 | 595 | else if (uv == none) { /* "none" is unmapped character */ |
f54cb97a | 596 | const int i = UTF8SKIP(s); |
0376ff32 | 597 | Move(s, d, i, U8); |
9b877dbb IH |
598 | d += i; |
599 | s += i; | |
4757a243 | 600 | continue; |
79072805 | 601 | } |
4757a243 LW |
602 | else if (uv == extra && !del) { |
603 | matches++; | |
01ec43d0 | 604 | s += UTF8SKIP(s); |
7d85a32c | 605 | d = uvuni_to_utf8(d, final); |
4757a243 LW |
606 | continue; |
607 | } | |
01ec43d0 GS |
608 | matches++; /* "none+1" is delete character */ |
609 | s += UTF8SKIP(s); | |
79072805 | 610 | } |
4757a243 | 611 | } |
16ec844d | 612 | if (grows || hibit) { |
9b877dbb IH |
613 | sv_setpvn(sv, (char*)dstart, d - dstart); |
614 | Safefree(dstart); | |
16ec844d IH |
615 | if (grows && hibit) |
616 | Safefree(start); | |
9b877dbb IH |
617 | } |
618 | else { | |
619 | *d = '\0'; | |
620 | SvCUR_set(sv, d - dstart); | |
621 | } | |
1aa99e6b | 622 | SvUTF8_on(sv); |
4757a243 LW |
623 | SvSETMAGIC(sv); |
624 | ||
625 | return matches; | |
626 | } | |
627 | ||
628 | I32 | |
864dbfa3 | 629 | Perl_do_trans(pTHX_ SV *sv) |
4757a243 | 630 | { |
97aff369 | 631 | dVAR; |
4757a243 | 632 | STRLEN len; |
4373e329 | 633 | const I32 hasutf = (PL_op->op_private & |
036b4402 | 634 | (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)); |
4757a243 | 635 | |
7918f24d NC |
636 | PERL_ARGS_ASSERT_DO_TRANS; |
637 | ||
3788ef8f | 638 | if (SvREADONLY(sv) && !(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
765f542d NC |
639 | if (SvIsCOW(sv)) |
640 | sv_force_normal_flags(sv, 0); | |
3788ef8f | 641 | if (SvREADONLY(sv)) |
f1f66076 | 642 | Perl_croak(aTHX_ "%s", PL_no_modify); |
2233f375 | 643 | } |
10516c54 | 644 | (void)SvPV_const(sv, len); |
4757a243 LW |
645 | if (!len) |
646 | return 0; | |
d59e14db RGS |
647 | if (!(PL_op->op_private & OPpTRANS_IDENTICAL)) { |
648 | if (!SvPOKp(sv)) | |
649 | (void)SvPV_force(sv, len); | |
2de7b02f | 650 | (void)SvPOK_only_UTF8(sv); |
d59e14db | 651 | } |
4757a243 | 652 | |
cea2e8a9 | 653 | DEBUG_t( Perl_deb(aTHX_ "2.TBL\n")); |
4757a243 | 654 | |
a77e643a RGS |
655 | switch (PL_op->op_private & ~hasutf & ( |
656 | OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF|OPpTRANS_IDENTICAL| | |
657 | OPpTRANS_SQUASH|OPpTRANS_DELETE|OPpTRANS_COMPLEMENT)) { | |
4757a243 | 658 | case 0: |
01ec43d0 GS |
659 | if (hasutf) |
660 | return do_trans_simple_utf8(sv); | |
661 | else | |
662 | return do_trans_simple(sv); | |
4757a243 LW |
663 | |
664 | case OPpTRANS_IDENTICAL: | |
05d340b8 | 665 | case OPpTRANS_IDENTICAL|OPpTRANS_COMPLEMENT: |
01ec43d0 GS |
666 | if (hasutf) |
667 | return do_trans_count_utf8(sv); | |
668 | else | |
669 | return do_trans_count(sv); | |
4757a243 LW |
670 | |
671 | default: | |
01ec43d0 | 672 | if (hasutf) |
036b4402 | 673 | return do_trans_complex_utf8(sv); |
4757a243 | 674 | else |
036b4402 | 675 | return do_trans_complex(sv); |
79072805 | 676 | } |
79072805 LW |
677 | } |
678 | ||
679 | void | |
b9ac451d | 680 | Perl_do_join(pTHX_ register SV *sv, SV *delim, register SV **mark, register SV **sp) |
79072805 | 681 | { |
97aff369 | 682 | dVAR; |
53c1dcc0 | 683 | SV ** const oldmark = mark; |
79072805 | 684 | register I32 items = sp - mark; |
79072805 | 685 | register STRLEN len; |
463ee0b2 | 686 | STRLEN delimlen; |
79072805 | 687 | |
7918f24d NC |
688 | PERL_ARGS_ASSERT_DO_JOIN; |
689 | ||
b9ac451d | 690 | (void) SvPV_const(delim, delimlen); /* stringify and get the delimlen */ |
5ba99574 | 691 | /* SvCUR assumes it's SvPOK() and woe betide you if it's not. */ |
516a5887 | 692 | |
79072805 LW |
693 | mark++; |
694 | len = (items > 0 ? (delimlen * (items - 1) ) : 0); | |
862a34c6 | 695 | SvUPGRADE(sv, SVt_PV); |
79072805 LW |
696 | if (SvLEN(sv) < len + items) { /* current length is way too short */ |
697 | while (items-- > 0) { | |
1426bbf4 | 698 | if (*mark && !SvGAMAGIC(*mark) && SvOK(*mark)) { |
f54cb97a | 699 | STRLEN tmplen; |
5c144d81 | 700 | SvPV_const(*mark, tmplen); |
463ee0b2 | 701 | len += tmplen; |
79072805 LW |
702 | } |
703 | mark++; | |
704 | } | |
705 | SvGROW(sv, len + 1); /* so try to pre-extend */ | |
706 | ||
707 | mark = oldmark; | |
db7c17d7 | 708 | items = sp - mark; |
79072805 LW |
709 | ++mark; |
710 | } | |
711 | ||
76f68e9b | 712 | sv_setpvs(sv, ""); |
e4803c42 | 713 | /* sv_setpv retains old UTF8ness [perl #24846] */ |
fb622db0 | 714 | SvUTF8_off(sv); |
e4803c42 | 715 | |
8d6d96c1 HS |
716 | if (PL_tainting && SvMAGICAL(sv)) |
717 | SvTAINTED_off(sv); | |
718 | ||
463ee0b2 | 719 | if (items-- > 0) { |
92d29cee JH |
720 | if (*mark) |
721 | sv_catsv(sv, *mark); | |
463ee0b2 LW |
722 | mark++; |
723 | } | |
8d6d96c1 | 724 | |
c512ce4f | 725 | if (delimlen) { |
79072805 | 726 | for (; items > 0; items--,mark++) { |
b9ac451d | 727 | sv_catsv(sv,delim); |
79072805 LW |
728 | sv_catsv(sv,*mark); |
729 | } | |
730 | } | |
731 | else { | |
732 | for (; items > 0; items--,mark++) | |
733 | sv_catsv(sv,*mark); | |
734 | } | |
735 | SvSETMAGIC(sv); | |
736 | } | |
737 | ||
738 | void | |
864dbfa3 | 739 | Perl_do_sprintf(pTHX_ SV *sv, I32 len, SV **sarg) |
79072805 | 740 | { |
97aff369 | 741 | dVAR; |
46fc3d4c | 742 | STRLEN patlen; |
53c1dcc0 | 743 | const char * const pat = SvPV_const(*sarg, patlen); |
46fc3d4c | 744 | bool do_taint = FALSE; |
745 | ||
7918f24d NC |
746 | PERL_ARGS_ASSERT_DO_SPRINTF; |
747 | ||
5b781b5b | 748 | SvUTF8_off(sv); |
2cf2cfc6 A |
749 | if (DO_UTF8(*sarg)) |
750 | SvUTF8_on(sv); | |
4608196e | 751 | sv_vsetpvfn(sv, pat, patlen, NULL, sarg + 1, len - 1, &do_taint); |
79072805 | 752 | SvSETMAGIC(sv); |
46fc3d4c | 753 | if (do_taint) |
754 | SvTAINTED_on(sv); | |
79072805 LW |
755 | } |
756 | ||
33b45480 | 757 | /* currently converts input to bytes if possible, but doesn't sweat failure */ |
81e118e0 JH |
758 | UV |
759 | Perl_do_vecget(pTHX_ SV *sv, I32 offset, I32 size) | |
760 | { | |
97aff369 | 761 | dVAR; |
bbb8a7e0 | 762 | STRLEN srclen, len, uoffset, bitoffs = 0; |
10516c54 | 763 | const unsigned char *s = (const unsigned char *) SvPV_const(sv, srclen); |
81e118e0 JH |
764 | UV retnum = 0; |
765 | ||
7918f24d NC |
766 | PERL_ARGS_ASSERT_DO_VECGET; |
767 | ||
a50d7633 | 768 | if (offset < 0) |
61f9802b | 769 | return 0; |
8e84507e | 770 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 771 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
246fae53 | 772 | |
dcad2880 | 773 | if (SvUTF8(sv)) |
33b45480 | 774 | (void) Perl_sv_utf8_downgrade(aTHX_ sv, TRUE); |
246fae53 | 775 | |
bbb8a7e0 MHM |
776 | if (size < 8) { |
777 | bitoffs = ((offset%8)*size)%8; | |
778 | uoffset = offset/(8/size); | |
779 | } | |
780 | else if (size > 8) | |
781 | uoffset = offset*(size/8); | |
782 | else | |
783 | uoffset = offset; | |
784 | ||
785 | len = uoffset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
81e118e0 JH |
786 | if (len > srclen) { |
787 | if (size <= 8) | |
788 | retnum = 0; | |
789 | else { | |
81e118e0 | 790 | if (size == 16) { |
bb7a0f54 | 791 | if (uoffset >= srclen) |
81e118e0 JH |
792 | retnum = 0; |
793 | else | |
bb7a0f54 | 794 | retnum = (UV) s[uoffset] << 8; |
81e118e0 JH |
795 | } |
796 | else if (size == 32) { | |
bb7a0f54 | 797 | if (uoffset >= srclen) |
81e118e0 | 798 | retnum = 0; |
bb7a0f54 | 799 | else if (uoffset + 1 >= srclen) |
81e118e0 | 800 | retnum = |
bb7a0f54 MHM |
801 | ((UV) s[uoffset ] << 24); |
802 | else if (uoffset + 2 >= srclen) | |
81e118e0 | 803 | retnum = |
bb7a0f54 MHM |
804 | ((UV) s[uoffset ] << 24) + |
805 | ((UV) s[uoffset + 1] << 16); | |
81e118e0 JH |
806 | else |
807 | retnum = | |
bb7a0f54 MHM |
808 | ((UV) s[uoffset ] << 24) + |
809 | ((UV) s[uoffset + 1] << 16) + | |
810 | ( s[uoffset + 2] << 8); | |
81e118e0 | 811 | } |
d7d93a81 | 812 | #ifdef UV_IS_QUAD |
c5a0f51a | 813 | else if (size == 64) { |
c5a0f51a | 814 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 815 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 816 | "Bit vector size > 32 non-portable"); |
bb7a0f54 | 817 | if (uoffset >= srclen) |
c5a0f51a | 818 | retnum = 0; |
bb7a0f54 | 819 | else if (uoffset + 1 >= srclen) |
c5a0f51a | 820 | retnum = |
bb7a0f54 MHM |
821 | (UV) s[uoffset ] << 56; |
822 | else if (uoffset + 2 >= srclen) | |
c5a0f51a | 823 | retnum = |
bb7a0f54 MHM |
824 | ((UV) s[uoffset ] << 56) + |
825 | ((UV) s[uoffset + 1] << 48); | |
826 | else if (uoffset + 3 >= srclen) | |
c5a0f51a | 827 | retnum = |
bb7a0f54 MHM |
828 | ((UV) s[uoffset ] << 56) + |
829 | ((UV) s[uoffset + 1] << 48) + | |
830 | ((UV) s[uoffset + 2] << 40); | |
831 | else if (uoffset + 4 >= srclen) | |
c5a0f51a | 832 | retnum = |
bb7a0f54 MHM |
833 | ((UV) s[uoffset ] << 56) + |
834 | ((UV) s[uoffset + 1] << 48) + | |
835 | ((UV) s[uoffset + 2] << 40) + | |
836 | ((UV) s[uoffset + 3] << 32); | |
837 | else if (uoffset + 5 >= srclen) | |
c5a0f51a | 838 | retnum = |
bb7a0f54 MHM |
839 | ((UV) s[uoffset ] << 56) + |
840 | ((UV) s[uoffset + 1] << 48) + | |
841 | ((UV) s[uoffset + 2] << 40) + | |
842 | ((UV) s[uoffset + 3] << 32) + | |
843 | ( s[uoffset + 4] << 24); | |
844 | else if (uoffset + 6 >= 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 | ((UV) s[uoffset + 4] << 24) + | |
851 | ((UV) s[uoffset + 5] << 16); | |
c5a0f51a | 852 | else |
8e84507e | 853 | retnum = |
bb7a0f54 MHM |
854 | ((UV) s[uoffset ] << 56) + |
855 | ((UV) s[uoffset + 1] << 48) + | |
856 | ((UV) s[uoffset + 2] << 40) + | |
857 | ((UV) s[uoffset + 3] << 32) + | |
858 | ((UV) s[uoffset + 4] << 24) + | |
859 | ((UV) s[uoffset + 5] << 16) + | |
860 | ( s[uoffset + 6] << 8); | |
c5a0f51a JH |
861 | } |
862 | #endif | |
81e118e0 JH |
863 | } |
864 | } | |
865 | else if (size < 8) | |
bbb8a7e0 | 866 | retnum = (s[uoffset] >> bitoffs) & ((1 << size) - 1); |
81e118e0 | 867 | else { |
81e118e0 | 868 | if (size == 8) |
bb7a0f54 | 869 | retnum = s[uoffset]; |
81e118e0 JH |
870 | else if (size == 16) |
871 | retnum = | |
bb7a0f54 MHM |
872 | ((UV) s[uoffset] << 8) + |
873 | s[uoffset + 1]; | |
81e118e0 JH |
874 | else if (size == 32) |
875 | retnum = | |
bb7a0f54 MHM |
876 | ((UV) s[uoffset ] << 24) + |
877 | ((UV) s[uoffset + 1] << 16) + | |
878 | ( s[uoffset + 2] << 8) + | |
879 | s[uoffset + 3]; | |
d7d93a81 | 880 | #ifdef UV_IS_QUAD |
c5a0f51a | 881 | else if (size == 64) { |
c5a0f51a | 882 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 883 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a JH |
884 | "Bit vector size > 32 non-portable"); |
885 | retnum = | |
bb7a0f54 MHM |
886 | ((UV) s[uoffset ] << 56) + |
887 | ((UV) s[uoffset + 1] << 48) + | |
888 | ((UV) s[uoffset + 2] << 40) + | |
889 | ((UV) s[uoffset + 3] << 32) + | |
890 | ((UV) s[uoffset + 4] << 24) + | |
891 | ((UV) s[uoffset + 5] << 16) + | |
892 | ( s[uoffset + 6] << 8) + | |
893 | s[uoffset + 7]; | |
c5a0f51a JH |
894 | } |
895 | #endif | |
81e118e0 JH |
896 | } |
897 | ||
898 | return retnum; | |
899 | } | |
900 | ||
33b45480 SB |
901 | /* currently converts input to bytes if possible but doesn't sweat failures, |
902 | * although it does ensure that the string it clobbers is not marked as | |
903 | * utf8-valid any more | |
904 | */ | |
79072805 | 905 | void |
864dbfa3 | 906 | Perl_do_vecset(pTHX_ SV *sv) |
79072805 | 907 | { |
97aff369 | 908 | dVAR; |
bbb8a7e0 | 909 | register I32 offset, bitoffs = 0; |
79072805 | 910 | register I32 size; |
8990e307 | 911 | register unsigned char *s; |
81e118e0 | 912 | register UV lval; |
79072805 | 913 | I32 mask; |
a0d0e21e LW |
914 | STRLEN targlen; |
915 | STRLEN len; | |
c4420975 | 916 | SV * const targ = LvTARG(sv); |
79072805 | 917 | |
7918f24d NC |
918 | PERL_ARGS_ASSERT_DO_VECSET; |
919 | ||
8990e307 LW |
920 | if (!targ) |
921 | return; | |
a0d0e21e | 922 | s = (unsigned char*)SvPV_force(targ, targlen); |
246fae53 | 923 | if (SvUTF8(targ)) { |
33b45480 SB |
924 | /* This is handled by the SvPOK_only below... |
925 | if (!Perl_sv_utf8_downgrade(aTHX_ targ, TRUE)) | |
926 | SvUTF8_off(targ); | |
927 | */ | |
928 | (void) Perl_sv_utf8_downgrade(aTHX_ targ, TRUE); | |
246fae53 MG |
929 | } |
930 | ||
4ebbc975 | 931 | (void)SvPOK_only(targ); |
81e118e0 | 932 | lval = SvUV(sv); |
79072805 | 933 | offset = LvTARGOFF(sv); |
fe58ced6 | 934 | if (offset < 0) |
ed9aa3b7 | 935 | Perl_croak(aTHX_ "Negative offset to vec in lvalue context"); |
79072805 | 936 | size = LvTARGLEN(sv); |
8e84507e | 937 | if (size < 1 || (size & (size-1))) /* size < 1 or not a power of two */ |
a50d7633 | 938 | Perl_croak(aTHX_ "Illegal number of bits in vec"); |
8e84507e | 939 | |
bbb8a7e0 MHM |
940 | if (size < 8) { |
941 | bitoffs = ((offset%8)*size)%8; | |
942 | offset /= 8/size; | |
943 | } | |
944 | else if (size > 8) | |
945 | offset *= size/8; | |
946 | ||
947 | len = offset + (bitoffs + size + 7)/8; /* required number of bytes */ | |
a0d0e21e LW |
948 | if (len > targlen) { |
949 | s = (unsigned char*)SvGROW(targ, len + 1); | |
12ae5dfc | 950 | (void)memzero((char *)(s + targlen), len - targlen + 1); |
a0d0e21e LW |
951 | SvCUR_set(targ, len); |
952 | } | |
8e84507e | 953 | |
79072805 LW |
954 | if (size < 8) { |
955 | mask = (1 << size) - 1; | |
79072805 | 956 | lval &= mask; |
bbb8a7e0 MHM |
957 | s[offset] &= ~(mask << bitoffs); |
958 | s[offset] |= lval << bitoffs; | |
79072805 LW |
959 | } |
960 | else { | |
961 | if (size == 8) | |
eb160463 | 962 | s[offset ] = (U8)( lval & 0xff); |
79072805 | 963 | else if (size == 16) { |
eb160463 GS |
964 | s[offset ] = (U8)((lval >> 8) & 0xff); |
965 | s[offset+1] = (U8)( lval & 0xff); | |
79072805 LW |
966 | } |
967 | else if (size == 32) { | |
eb160463 GS |
968 | s[offset ] = (U8)((lval >> 24) & 0xff); |
969 | s[offset+1] = (U8)((lval >> 16) & 0xff); | |
970 | s[offset+2] = (U8)((lval >> 8) & 0xff); | |
971 | s[offset+3] = (U8)( lval & 0xff); | |
c5a0f51a | 972 | } |
d7d93a81 | 973 | #ifdef UV_IS_QUAD |
c5a0f51a | 974 | else if (size == 64) { |
c5a0f51a | 975 | if (ckWARN(WARN_PORTABLE)) |
9014280d | 976 | Perl_warner(aTHX_ packWARN(WARN_PORTABLE), |
c5a0f51a | 977 | "Bit vector size > 32 non-portable"); |
eb160463 GS |
978 | s[offset ] = (U8)((lval >> 56) & 0xff); |
979 | s[offset+1] = (U8)((lval >> 48) & 0xff); | |
980 | s[offset+2] = (U8)((lval >> 40) & 0xff); | |
981 | s[offset+3] = (U8)((lval >> 32) & 0xff); | |
982 | s[offset+4] = (U8)((lval >> 24) & 0xff); | |
983 | s[offset+5] = (U8)((lval >> 16) & 0xff); | |
984 | s[offset+6] = (U8)((lval >> 8) & 0xff); | |
985 | s[offset+7] = (U8)( lval & 0xff); | |
79072805 | 986 | } |
dc1e3f56 | 987 | #endif |
79072805 | 988 | } |
7bb043c3 | 989 | SvSETMAGIC(targ); |
79072805 LW |
990 | } |
991 | ||
992 | void | |
864dbfa3 | 993 | Perl_do_chop(pTHX_ register SV *astr, register SV *sv) |
79072805 | 994 | { |
97aff369 | 995 | dVAR; |
463ee0b2 | 996 | STRLEN len; |
a0d0e21e | 997 | char *s; |
8e84507e | 998 | |
7918f24d NC |
999 | PERL_ARGS_ASSERT_DO_CHOP; |
1000 | ||
79072805 | 1001 | if (SvTYPE(sv) == SVt_PVAV) { |
a0d0e21e | 1002 | register I32 i; |
502c6561 | 1003 | AV *const av = MUTABLE_AV(sv); |
f54cb97a AL |
1004 | const I32 max = AvFILL(av); |
1005 | ||
1006 | for (i = 0; i <= max; i++) { | |
ad64d0ec | 1007 | sv = MUTABLE_SV(av_fetch(av, i, FALSE)); |
3280af22 | 1008 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1009 | do_chop(astr, sv); |
1010 | } | |
1011 | return; | |
79072805 | 1012 | } |
aa854799 | 1013 | else if (SvTYPE(sv) == SVt_PVHV) { |
85fbaab2 | 1014 | HV* const hv = MUTABLE_HV(sv); |
a0d0e21e LW |
1015 | HE* entry; |
1016 | (void)hv_iterinit(hv); | |
155aba94 | 1017 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1018 | do_chop(astr,hv_iterval(hv,entry)); |
1019 | return; | |
79072805 | 1020 | } |
991e6d41 NC |
1021 | else if (SvREADONLY(sv)) { |
1022 | if (SvFAKE(sv)) { | |
1023 | /* SV is copy-on-write */ | |
1024 | sv_force_normal_flags(sv, 0); | |
1025 | } | |
1026 | if (SvREADONLY(sv)) | |
f1f66076 | 1027 | Perl_croak(aTHX_ "%s", PL_no_modify); |
991e6d41 | 1028 | } |
565a3db3 RGS |
1029 | |
1030 | if (PL_encoding && !SvUTF8(sv)) { | |
1031 | /* like in do_chomp(), utf8-ize the sv as a side-effect | |
1032 | * if we're using encoding. */ | |
1033 | sv_recode_to_utf8(sv, PL_encoding); | |
1034 | } | |
1035 | ||
a0d0e21e | 1036 | s = SvPV(sv, len); |
748a9306 | 1037 | if (len && !SvPOK(sv)) |
1e968d83 | 1038 | s = SvPV_force_nomg(sv, len); |
7e2040f0 | 1039 | if (DO_UTF8(sv)) { |
a0ed51b3 | 1040 | if (s && len) { |
c4420975 | 1041 | char * const send = s + len; |
0bcc34c2 | 1042 | char * const start = s; |
a0ed51b3 | 1043 | s = send - 1; |
a0dbb045 JH |
1044 | while (s > start && UTF8_IS_CONTINUATION(*s)) |
1045 | s--; | |
ace7757b | 1046 | if (is_utf8_string((U8*)s, send - s)) { |
a0dbb045 JH |
1047 | sv_setpvn(astr, s, send - s); |
1048 | *s = '\0'; | |
1049 | SvCUR_set(sv, s - start); | |
1050 | SvNIOK_off(sv); | |
1051 | SvUTF8_on(astr); | |
1052 | } | |
a0ed51b3 LW |
1053 | } |
1054 | else | |
76f68e9b | 1055 | sv_setpvs(astr, ""); |
a0ed51b3 | 1056 | } |
7e2040f0 | 1057 | else if (s && len) { |
a0d0e21e LW |
1058 | s += --len; |
1059 | sv_setpvn(astr, s, 1); | |
1060 | *s = '\0'; | |
1061 | SvCUR_set(sv, len); | |
2c19a612 | 1062 | SvUTF8_off(sv); |
a0d0e21e | 1063 | SvNIOK_off(sv); |
79072805 LW |
1064 | } |
1065 | else | |
76f68e9b | 1066 | sv_setpvs(astr, ""); |
a0d0e21e | 1067 | SvSETMAGIC(sv); |
7e2040f0 | 1068 | } |
a0d0e21e LW |
1069 | |
1070 | I32 | |
864dbfa3 | 1071 | Perl_do_chomp(pTHX_ register SV *sv) |
a0d0e21e | 1072 | { |
97aff369 | 1073 | dVAR; |
c07a80fd | 1074 | register I32 count; |
a0d0e21e LW |
1075 | STRLEN len; |
1076 | char *s; | |
c4c87a06 | 1077 | char *temp_buffer = NULL; |
a0714e2c | 1078 | SV* svrecode = NULL; |
c07a80fd | 1079 | |
7918f24d NC |
1080 | PERL_ARGS_ASSERT_DO_CHOMP; |
1081 | ||
3280af22 | 1082 | if (RsSNARF(PL_rs)) |
c07a80fd | 1083 | return 0; |
4c5a6083 GS |
1084 | if (RsRECORD(PL_rs)) |
1085 | return 0; | |
c07a80fd | 1086 | count = 0; |
a0d0e21e LW |
1087 | if (SvTYPE(sv) == SVt_PVAV) { |
1088 | register I32 i; | |
502c6561 | 1089 | AV *const av = MUTABLE_AV(sv); |
f54cb97a AL |
1090 | const I32 max = AvFILL(av); |
1091 | ||
1092 | for (i = 0; i <= max; i++) { | |
ad64d0ec | 1093 | sv = MUTABLE_SV(av_fetch(av, i, FALSE)); |
3280af22 | 1094 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) |
a0d0e21e LW |
1095 | count += do_chomp(sv); |
1096 | } | |
1097 | return count; | |
1098 | } | |
aa854799 | 1099 | else if (SvTYPE(sv) == SVt_PVHV) { |
85fbaab2 | 1100 | HV* const hv = MUTABLE_HV(sv); |
a0d0e21e LW |
1101 | HE* entry; |
1102 | (void)hv_iterinit(hv); | |
155aba94 | 1103 | while ((entry = hv_iternext(hv))) |
a0d0e21e LW |
1104 | count += do_chomp(hv_iterval(hv,entry)); |
1105 | return count; | |
1106 | } | |
991e6d41 NC |
1107 | else if (SvREADONLY(sv)) { |
1108 | if (SvFAKE(sv)) { | |
1109 | /* SV is copy-on-write */ | |
1110 | sv_force_normal_flags(sv, 0); | |
1111 | } | |
1112 | if (SvREADONLY(sv)) | |
f1f66076 | 1113 | Perl_croak(aTHX_ "%s", PL_no_modify); |
991e6d41 | 1114 | } |
a6aa349d TS |
1115 | |
1116 | if (PL_encoding) { | |
1117 | if (!SvUTF8(sv)) { | |
1118 | /* XXX, here sv is utf8-ized as a side-effect! | |
1119 | If encoding.pm is used properly, almost string-generating | |
1120 | operations, including literal strings, chr(), input data, etc. | |
1121 | should have been utf8-ized already, right? | |
1122 | */ | |
1123 | sv_recode_to_utf8(sv, PL_encoding); | |
1124 | } | |
1125 | } | |
1126 | ||
a0d0e21e | 1127 | s = SvPV(sv, len); |
a0d0e21e LW |
1128 | if (s && len) { |
1129 | s += --len; | |
3280af22 | 1130 | if (RsPARA(PL_rs)) { |
a0d0e21e LW |
1131 | if (*s != '\n') |
1132 | goto nope; | |
1133 | ++count; | |
1134 | while (len && s[-1] == '\n') { | |
1135 | --len; | |
1136 | --s; | |
1137 | ++count; | |
1138 | } | |
1139 | } | |
a0d0e21e | 1140 | else { |
a6aa349d | 1141 | STRLEN rslen, rs_charlen; |
93524f2b | 1142 | const char *rsptr = SvPV_const(PL_rs, rslen); |
a6aa349d TS |
1143 | |
1144 | rs_charlen = SvUTF8(PL_rs) | |
1145 | ? sv_len_utf8(PL_rs) | |
1146 | : rslen; | |
1147 | ||
c4c87a06 NC |
1148 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { |
1149 | /* Assumption is that rs is shorter than the scalar. */ | |
1150 | if (SvUTF8(PL_rs)) { | |
1151 | /* RS is utf8, scalar is 8 bit. */ | |
1152 | bool is_utf8 = TRUE; | |
1153 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, | |
1154 | &rslen, &is_utf8); | |
1155 | if (is_utf8) { | |
1156 | /* Cannot downgrade, therefore cannot possibly match | |
1157 | */ | |
1158 | assert (temp_buffer == rsptr); | |
1159 | temp_buffer = NULL; | |
1160 | goto nope; | |
1161 | } | |
1162 | rsptr = temp_buffer; | |
a6aa349d TS |
1163 | } |
1164 | else if (PL_encoding) { | |
1165 | /* RS is 8 bit, encoding.pm is used. | |
1166 | * Do not recode PL_rs as a side-effect. */ | |
1167 | svrecode = newSVpvn(rsptr, rslen); | |
1168 | sv_recode_to_utf8(svrecode, PL_encoding); | |
93524f2b | 1169 | rsptr = SvPV_const(svrecode, rslen); |
a6aa349d TS |
1170 | rs_charlen = sv_len_utf8(svrecode); |
1171 | } | |
1172 | else { | |
c4c87a06 NC |
1173 | /* RS is 8 bit, scalar is utf8. */ |
1174 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); | |
1175 | rsptr = temp_buffer; | |
1176 | } | |
1177 | } | |
c07a80fd | 1178 | if (rslen == 1) { |
1179 | if (*s != *rsptr) | |
1180 | goto nope; | |
1181 | ++count; | |
1182 | } | |
1183 | else { | |
8c2cee6f | 1184 | if (len < rslen - 1) |
c07a80fd | 1185 | goto nope; |
1186 | len -= rslen - 1; | |
1187 | s -= rslen - 1; | |
36477c24 | 1188 | if (memNE(s, rsptr, rslen)) |
c07a80fd | 1189 | goto nope; |
a6aa349d | 1190 | count += rs_charlen; |
c07a80fd | 1191 | } |
a0d0e21e | 1192 | } |
8b6b16e7 | 1193 | s = SvPV_force_nolen(sv); |
a0d0e21e | 1194 | SvCUR_set(sv, len); |
9b33ce3b | 1195 | *SvEND(sv) = '\0'; |
a0d0e21e | 1196 | SvNIOK_off(sv); |
9b33ce3b | 1197 | SvSETMAGIC(sv); |
a0d0e21e LW |
1198 | } |
1199 | nope: | |
a6aa349d TS |
1200 | |
1201 | if (svrecode) | |
1202 | SvREFCNT_dec(svrecode); | |
1203 | ||
c4c87a06 | 1204 | Safefree(temp_buffer); |
a0d0e21e | 1205 | return count; |
8e84507e | 1206 | } |
79072805 LW |
1207 | |
1208 | void | |
864dbfa3 | 1209 | Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right) |
79072805 | 1210 | { |
97aff369 | 1211 | dVAR; |
79072805 LW |
1212 | #ifdef LIBERAL |
1213 | register long *dl; | |
1214 | register long *ll; | |
1215 | register long *rl; | |
1216 | #endif | |
1217 | register char *dc; | |
463ee0b2 LW |
1218 | STRLEN leftlen; |
1219 | STRLEN rightlen; | |
e62f0680 NC |
1220 | register const char *lc; |
1221 | register const char *rc; | |
bb7a0f54 MHM |
1222 | register STRLEN len; |
1223 | STRLEN lensave; | |
e62f0680 NC |
1224 | const char *lsave; |
1225 | const char *rsave; | |
12abf4f0 NC |
1226 | bool left_utf; |
1227 | bool right_utf; | |
bb7a0f54 | 1228 | STRLEN needlen = 0; |
0c57e439 | 1229 | |
7918f24d | 1230 | PERL_ARGS_ASSERT_DO_VOP; |
79072805 | 1231 | |
1fbd88dc | 1232 | if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv))) |
76f68e9b | 1233 | sv_setpvs(sv, ""); /* avoid undef warning on |= and ^= */ |
8c8eee82 BM |
1234 | if (sv == left) { |
1235 | lsave = lc = SvPV_force_nomg(left, leftlen); | |
1236 | } | |
1237 | else { | |
1238 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1239 | SvPV_force_nomg_nolen(sv); | |
1240 | } | |
e62f0680 | 1241 | rsave = rc = SvPV_nomg_const(right, rightlen); |
12abf4f0 NC |
1242 | |
1243 | /* This need to come after SvPV to ensure that string overloading has | |
1244 | fired off. */ | |
1245 | ||
1246 | left_utf = DO_UTF8(left); | |
1247 | right_utf = DO_UTF8(right); | |
1248 | ||
1249 | if (left_utf && !right_utf) { | |
1250 | /* Avoid triggering overloading again by using temporaries. | |
1251 | Maybe there should be a variant of sv_utf8_upgrade that takes pvn | |
1252 | */ | |
59cd0e26 | 1253 | right = newSVpvn_flags(rsave, rightlen, SVs_TEMP); |
12abf4f0 NC |
1254 | sv_utf8_upgrade(right); |
1255 | rsave = rc = SvPV_nomg_const(right, rightlen); | |
1256 | right_utf = TRUE; | |
1257 | } | |
1258 | else if (!left_utf && right_utf) { | |
59cd0e26 | 1259 | left = newSVpvn_flags(lsave, leftlen, SVs_TEMP); |
12abf4f0 NC |
1260 | sv_utf8_upgrade(left); |
1261 | lsave = lc = SvPV_nomg_const(left, leftlen); | |
1262 | left_utf = TRUE; | |
1263 | } | |
1264 | ||
93a17b20 | 1265 | len = leftlen < rightlen ? leftlen : rightlen; |
a0d0e21e | 1266 | lensave = len; |
9fdd7463 JH |
1267 | SvCUR_set(sv, len); |
1268 | (void)SvPOK_only(sv); | |
c9b3c8d0 JH |
1269 | if ((left_utf || right_utf) && (sv == left || sv == right)) { |
1270 | needlen = optype == OP_BIT_AND ? len : leftlen + rightlen; | |
a02a5408 | 1271 | Newxz(dc, needlen + 1, char); |
c9b3c8d0 JH |
1272 | } |
1273 | else if (SvOK(sv) || SvTYPE(sv) > SVt_PVMG) { | |
2596d9fe | 1274 | dc = SvPV_force_nomg_nolen(sv); |
bb7a0f54 MHM |
1275 | if (SvLEN(sv) < len + 1) { |
1276 | dc = SvGROW(sv, len + 1); | |
ff68c719 | 1277 | (void)memzero(dc + SvCUR(sv), len - SvCUR(sv) + 1); |
1278 | } | |
c9b3c8d0 JH |
1279 | if (optype != OP_BIT_AND && (left_utf || right_utf)) |
1280 | dc = SvGROW(sv, leftlen + rightlen + 1); | |
ff68c719 | 1281 | } |
1282 | else { | |
bb7a0f54 MHM |
1283 | needlen = optype == OP_BIT_AND |
1284 | ? len : (leftlen > rightlen ? leftlen : rightlen); | |
a02a5408 | 1285 | Newxz(dc, needlen + 1, char); |
aa0a69cb | 1286 | sv_usepvn_flags(sv, dc, needlen, SV_HAS_TRAILING_NUL); |
ff68c719 | 1287 | dc = SvPVX(sv); /* sv_usepvn() calls Renew() */ |
79072805 | 1288 | } |
0c57e439 GS |
1289 | if (left_utf || right_utf) { |
1290 | UV duc, luc, ruc; | |
794a0d33 JH |
1291 | char *dcorig = dc; |
1292 | char *dcsave = NULL; | |
0c57e439 GS |
1293 | STRLEN lulen = leftlen; |
1294 | STRLEN rulen = rightlen; | |
ba210ebe | 1295 | STRLEN ulen; |
0c57e439 GS |
1296 | |
1297 | switch (optype) { | |
1298 | case OP_BIT_AND: | |
1299 | while (lulen && rulen) { | |
9041c2e3 | 1300 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1301 | lc += ulen; |
1302 | lulen -= ulen; | |
9041c2e3 | 1303 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1304 | rc += ulen; |
1305 | rulen -= ulen; | |
1306 | duc = luc & ruc; | |
9041c2e3 | 1307 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 | 1308 | } |
3ce4c532 NC |
1309 | if (sv == left || sv == right) |
1310 | (void)sv_usepvn(sv, dcorig, needlen); | |
794a0d33 | 1311 | SvCUR_set(sv, dc - dcorig); |
0c57e439 GS |
1312 | break; |
1313 | case OP_BIT_XOR: | |
1314 | while (lulen && rulen) { | |
9041c2e3 | 1315 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1316 | lc += ulen; |
1317 | lulen -= ulen; | |
9041c2e3 | 1318 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1319 | rc += ulen; |
1320 | rulen -= ulen; | |
1321 | duc = luc ^ ruc; | |
9041c2e3 | 1322 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1323 | } |
1324 | goto mop_up_utf; | |
1325 | case OP_BIT_OR: | |
1326 | while (lulen && rulen) { | |
9041c2e3 | 1327 | luc = utf8n_to_uvchr((U8*)lc, lulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1328 | lc += ulen; |
1329 | lulen -= ulen; | |
9041c2e3 | 1330 | ruc = utf8n_to_uvchr((U8*)rc, rulen, &ulen, UTF8_ALLOW_ANYUV); |
0c57e439 GS |
1331 | rc += ulen; |
1332 | rulen -= ulen; | |
1333 | duc = luc | ruc; | |
9041c2e3 | 1334 | dc = (char*)uvchr_to_utf8((U8*)dc, duc); |
0c57e439 GS |
1335 | } |
1336 | mop_up_utf: | |
794a0d33 JH |
1337 | if (rulen) |
1338 | dcsave = savepvn(rc, rulen); | |
1339 | else if (lulen) | |
1340 | dcsave = savepvn(lc, lulen); | |
3ce4c532 | 1341 | if (sv == left || sv == right) |
794a0d33 JH |
1342 | (void)sv_usepvn(sv, dcorig, needlen); /* Uses Renew(). */ |
1343 | SvCUR_set(sv, dc - dcorig); | |
0c57e439 | 1344 | if (rulen) |
794a0d33 | 1345 | sv_catpvn(sv, dcsave, rulen); |
0c57e439 | 1346 | else if (lulen) |
794a0d33 | 1347 | sv_catpvn(sv, dcsave, lulen); |
0c57e439 GS |
1348 | else |
1349 | *SvEND(sv) = '\0'; | |
794a0d33 | 1350 | Safefree(dcsave); |
0c57e439 | 1351 | break; |
41547412 NC |
1352 | default: |
1353 | if (sv == left || sv == right) | |
1354 | Safefree(dcorig); | |
c395bd6c AL |
1355 | Perl_croak(aTHX_ "panic: do_vop called for op %u (%s)", |
1356 | (unsigned)optype, PL_op_name[optype]); | |
0c57e439 GS |
1357 | } |
1358 | SvUTF8_on(sv); | |
1359 | goto finish; | |
1360 | } | |
1361 | else | |
79072805 LW |
1362 | #ifdef LIBERAL |
1363 | if (len >= sizeof(long)*4 && | |
bb7a0f54 MHM |
1364 | !((unsigned long)dc % sizeof(long)) && |
1365 | !((unsigned long)lc % sizeof(long)) && | |
1366 | !((unsigned long)rc % sizeof(long))) /* It's almost always aligned... */ | |
79072805 | 1367 | { |
bb7a0f54 | 1368 | const STRLEN remainder = len % (sizeof(long)*4); |
79072805 LW |
1369 | len /= (sizeof(long)*4); |
1370 | ||
1371 | dl = (long*)dc; | |
1372 | ll = (long*)lc; | |
1373 | rl = (long*)rc; | |
1374 | ||
1375 | switch (optype) { | |
1376 | case OP_BIT_AND: | |
1377 | while (len--) { | |
1378 | *dl++ = *ll++ & *rl++; | |
1379 | *dl++ = *ll++ & *rl++; | |
1380 | *dl++ = *ll++ & *rl++; | |
1381 | *dl++ = *ll++ & *rl++; | |
1382 | } | |
1383 | break; | |
a0d0e21e | 1384 | case OP_BIT_XOR: |
79072805 LW |
1385 | while (len--) { |
1386 | *dl++ = *ll++ ^ *rl++; | |
1387 | *dl++ = *ll++ ^ *rl++; | |
1388 | *dl++ = *ll++ ^ *rl++; | |
1389 | *dl++ = *ll++ ^ *rl++; | |
1390 | } | |
1391 | break; | |
1392 | case OP_BIT_OR: | |
1393 | while (len--) { | |
1394 | *dl++ = *ll++ | *rl++; | |
1395 | *dl++ = *ll++ | *rl++; | |
1396 | *dl++ = *ll++ | *rl++; | |
1397 | *dl++ = *ll++ | *rl++; | |
1398 | } | |
1399 | } | |
1400 | ||
1401 | dc = (char*)dl; | |
1402 | lc = (char*)ll; | |
1403 | rc = (char*)rl; | |
1404 | ||
1405 | len = remainder; | |
1406 | } | |
1407 | #endif | |
a0d0e21e | 1408 | { |
a0d0e21e LW |
1409 | switch (optype) { |
1410 | case OP_BIT_AND: | |
1411 | while (len--) | |
1412 | *dc++ = *lc++ & *rc++; | |
1a787b95 | 1413 | *dc = '\0'; |
a0d0e21e LW |
1414 | break; |
1415 | case OP_BIT_XOR: | |
1416 | while (len--) | |
1417 | *dc++ = *lc++ ^ *rc++; | |
1418 | goto mop_up; | |
1419 | case OP_BIT_OR: | |
1420 | while (len--) | |
1421 | *dc++ = *lc++ | *rc++; | |
1422 | mop_up: | |
1423 | len = lensave; | |
bb7a0f54 | 1424 | if (rightlen > len) |
a0d0e21e | 1425 | sv_catpvn(sv, rsave + len, rightlen - len); |
eb160463 | 1426 | else if (leftlen > (STRLEN)len) |
a0d0e21e | 1427 | sv_catpvn(sv, lsave + len, leftlen - len); |
4633a7c4 LW |
1428 | else |
1429 | *SvEND(sv) = '\0'; | |
a0d0e21e LW |
1430 | break; |
1431 | } | |
79072805 | 1432 | } |
0c57e439 | 1433 | finish: |
fb73857a | 1434 | SvTAINT(sv); |
79072805 | 1435 | } |
463ee0b2 LW |
1436 | |
1437 | OP * | |
cea2e8a9 | 1438 | Perl_do_kv(pTHX) |
463ee0b2 | 1439 | { |
97aff369 | 1440 | dVAR; |
39644a26 | 1441 | dSP; |
85fbaab2 | 1442 | HV * const hv = MUTABLE_HV(POPs); |
800e9ae0 | 1443 | HV *keys; |
463ee0b2 | 1444 | register HE *entry; |
f54cb97a AL |
1445 | const I32 gimme = GIMME_V; |
1446 | const I32 dokv = (PL_op->op_type == OP_RV2HV || PL_op->op_type == OP_PADHV); | |
1447 | const I32 dokeys = dokv || (PL_op->op_type == OP_KEYS); | |
1448 | const I32 dovalues = dokv || (PL_op->op_type == OP_VALUES); | |
463ee0b2 | 1449 | |
85581909 | 1450 | if (!hv) { |
78f9721b | 1451 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1452 | dTARGET; /* make sure to clear its target here */ |
1453 | if (SvTYPE(TARG) == SVt_PVLV) | |
a0714e2c | 1454 | LvTARG(TARG) = NULL; |
85581909 SB |
1455 | PUSHs(TARG); |
1456 | } | |
463ee0b2 | 1457 | RETURN; |
85581909 | 1458 | } |
748a9306 | 1459 | |
6d822dc4 | 1460 | keys = hv; |
800e9ae0 | 1461 | (void)hv_iterinit(keys); /* always reset iterator regardless */ |
748a9306 | 1462 | |
54310121 | 1463 | if (gimme == G_VOID) |
aa689395 | 1464 | RETURN; |
1465 | ||
54310121 | 1466 | if (gimme == G_SCALAR) { |
6ee623d5 | 1467 | IV i; |
463ee0b2 LW |
1468 | dTARGET; |
1469 | ||
78f9721b | 1470 | if (PL_op->op_flags & OPf_MOD || LVRET) { /* lvalue */ |
85581909 SB |
1471 | if (SvTYPE(TARG) < SVt_PVLV) { |
1472 | sv_upgrade(TARG, SVt_PVLV); | |
a0714e2c | 1473 | sv_magic(TARG, NULL, PERL_MAGIC_nkeys, NULL, 0); |
85581909 SB |
1474 | } |
1475 | LvTYPE(TARG) = 'k'; | |
ad64d0ec | 1476 | if (LvTARG(TARG) != (const SV *)keys) { |
6ff81951 GS |
1477 | if (LvTARG(TARG)) |
1478 | SvREFCNT_dec(LvTARG(TARG)); | |
b37c2d43 | 1479 | LvTARG(TARG) = SvREFCNT_inc_simple(keys); |
6ff81951 | 1480 | } |
85581909 SB |
1481 | PUSHs(TARG); |
1482 | RETURN; | |
1483 | } | |
1484 | ||
ad64d0ec | 1485 | if (! SvTIED_mg((const SV *)keys, PERL_MAGIC_tied) ) |
81714fb9 | 1486 | { |
a98face1 | 1487 | i = HvKEYS(keys); |
81714fb9 | 1488 | } |
463ee0b2 LW |
1489 | else { |
1490 | i = 0; | |
800e9ae0 | 1491 | while (hv_iternext(keys)) i++; |
463ee0b2 LW |
1492 | } |
1493 | PUSHi( i ); | |
1494 | RETURN; | |
1495 | } | |
1496 | ||
8ed4b672 | 1497 | EXTEND(SP, HvKEYS(keys) * (dokeys + dovalues)); |
463ee0b2 | 1498 | |
463ee0b2 | 1499 | PUTBACK; /* hv_iternext and hv_iterval might clobber stack_sp */ |
155aba94 | 1500 | while ((entry = hv_iternext(keys))) { |
463ee0b2 | 1501 | SPAGAIN; |
574c8022 | 1502 | if (dokeys) { |
890ce7af | 1503 | SV* const sv = hv_iterkeysv(entry); |
574c8022 JH |
1504 | XPUSHs(sv); /* won't clobber stack_sp */ |
1505 | } | |
463ee0b2 | 1506 | if (dovalues) { |
f54cb97a | 1507 | SV *tmpstr; |
463ee0b2 | 1508 | PUTBACK; |
6d822dc4 | 1509 | tmpstr = hv_iterval(hv,entry); |
cea2e8a9 | 1510 | DEBUG_H(Perl_sv_setpvf(aTHX_ tmpstr, "%lu%%%d=%lu", |
46fc3d4c | 1511 | (unsigned long)HeHASH(entry), |
cdd3ba14 | 1512 | (int)HvMAX(keys)+1, |
800e9ae0 | 1513 | (unsigned long)(HeHASH(entry) & HvMAX(keys)))); |
463ee0b2 | 1514 | SPAGAIN; |
46fc3d4c | 1515 | XPUSHs(tmpstr); |
463ee0b2 LW |
1516 | } |
1517 | PUTBACK; | |
1518 | } | |
1519 | return NORMAL; | |
1520 | } | |
4e35701f | 1521 | |
af3babe4 NC |
1522 | /* |
1523 | * Local variables: | |
1524 | * c-indentation-style: bsd | |
1525 | * c-basic-offset: 4 | |
1526 | * indent-tabs-mode: t | |
1527 | * End: | |
1528 | * | |
37442d52 RGS |
1529 | * ex: set ts=8 sts=4 sw=4 noet: |
1530 | */ |