Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* pp.c |
79072805 | 2 | * |
1129b882 NC |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, |
4 | * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others | |
79072805 | 5 | * |
a0d0e21e LW |
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. | |
79072805 | 8 | * |
a0d0e21e LW |
9 | */ |
10 | ||
11 | /* | |
4ac71550 TC |
12 | * 'It's a big house this, and very peculiar. Always a bit more |
13 | * to discover, and no knowing what you'll find round a corner. | |
14 | * And Elves, sir!' --Samwise Gamgee | |
15 | * | |
16 | * [p.225 of _The Lord of the Rings_, II/i: "Many Meetings"] | |
a0d0e21e | 17 | */ |
79072805 | 18 | |
166f8a29 DM |
19 | /* This file contains general pp ("push/pop") functions that execute the |
20 | * opcodes that make up a perl program. A typical pp function expects to | |
21 | * find its arguments on the stack, and usually pushes its results onto | |
22 | * the stack, hence the 'pp' terminology. Each OP structure contains | |
23 | * a pointer to the relevant pp_foo() function. | |
24 | */ | |
25 | ||
79072805 | 26 | #include "EXTERN.h" |
864dbfa3 | 27 | #define PERL_IN_PP_C |
79072805 | 28 | #include "perl.h" |
77bc9082 | 29 | #include "keywords.h" |
79072805 | 30 | |
a4af207c JH |
31 | #include "reentr.h" |
32 | ||
dfe9444c AD |
33 | /* XXX I can't imagine anyone who doesn't have this actually _needs_ |
34 | it, since pid_t is an integral type. | |
35 | --AD 2/20/1998 | |
36 | */ | |
37 | #ifdef NEED_GETPID_PROTO | |
38 | extern Pid_t getpid (void); | |
8ac85365 NIS |
39 | #endif |
40 | ||
0630166f SP |
41 | /* |
42 | * Some BSDs and Cygwin default to POSIX math instead of IEEE. | |
43 | * This switches them over to IEEE. | |
44 | */ | |
45 | #if defined(LIBM_LIB_VERSION) | |
46 | _LIB_VERSION_TYPE _LIB_VERSION = _IEEE_; | |
47 | #endif | |
48 | ||
13017935 SM |
49 | /* variations on pp_null */ |
50 | ||
93a17b20 LW |
51 | PP(pp_stub) |
52 | { | |
97aff369 | 53 | dVAR; |
39644a26 | 54 | dSP; |
54310121 | 55 | if (GIMME_V == G_SCALAR) |
3280af22 | 56 | XPUSHs(&PL_sv_undef); |
93a17b20 LW |
57 | RETURN; |
58 | } | |
59 | ||
79072805 LW |
60 | /* Pushy stuff. */ |
61 | ||
93a17b20 LW |
62 | PP(pp_padav) |
63 | { | |
97aff369 | 64 | dVAR; dSP; dTARGET; |
13017935 | 65 | I32 gimme; |
e190e9b4 | 66 | assert(SvTYPE(TARG) == SVt_PVAV); |
533c011a | 67 | if (PL_op->op_private & OPpLVAL_INTRO) |
a5911867 RGS |
68 | if (!(PL_op->op_private & OPpPAD_STATE)) |
69 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); | |
85e6fe83 | 70 | EXTEND(SP, 1); |
533c011a | 71 | if (PL_op->op_flags & OPf_REF) { |
85e6fe83 | 72 | PUSHs(TARG); |
93a17b20 | 73 | RETURN; |
78f9721b SM |
74 | } else if (LVRET) { |
75 | if (GIMME == G_SCALAR) | |
76 | Perl_croak(aTHX_ "Can't return array to lvalue scalar context"); | |
77 | PUSHs(TARG); | |
78 | RETURN; | |
85e6fe83 | 79 | } |
13017935 SM |
80 | gimme = GIMME_V; |
81 | if (gimme == G_ARRAY) { | |
502c6561 | 82 | const I32 maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
85e6fe83 | 83 | EXTEND(SP, maxarg); |
93965878 NIS |
84 | if (SvMAGICAL(TARG)) { |
85 | U32 i; | |
eb160463 | 86 | for (i=0; i < (U32)maxarg; i++) { |
502c6561 | 87 | SV * const * const svp = av_fetch(MUTABLE_AV(TARG), i, FALSE); |
3280af22 | 88 | SP[i+1] = (svp) ? *svp : &PL_sv_undef; |
93965878 NIS |
89 | } |
90 | } | |
91 | else { | |
502c6561 | 92 | Copy(AvARRAY((const AV *)TARG), SP+1, maxarg, SV*); |
93965878 | 93 | } |
85e6fe83 LW |
94 | SP += maxarg; |
95 | } | |
13017935 | 96 | else if (gimme == G_SCALAR) { |
1b6737cc | 97 | SV* const sv = sv_newmortal(); |
502c6561 | 98 | const I32 maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
85e6fe83 LW |
99 | sv_setiv(sv, maxarg); |
100 | PUSHs(sv); | |
101 | } | |
102 | RETURN; | |
93a17b20 LW |
103 | } |
104 | ||
105 | PP(pp_padhv) | |
106 | { | |
97aff369 | 107 | dVAR; dSP; dTARGET; |
54310121 | 108 | I32 gimme; |
109 | ||
e190e9b4 | 110 | assert(SvTYPE(TARG) == SVt_PVHV); |
93a17b20 | 111 | XPUSHs(TARG); |
533c011a | 112 | if (PL_op->op_private & OPpLVAL_INTRO) |
a5911867 RGS |
113 | if (!(PL_op->op_private & OPpPAD_STATE)) |
114 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); | |
533c011a | 115 | if (PL_op->op_flags & OPf_REF) |
93a17b20 | 116 | RETURN; |
78f9721b SM |
117 | else if (LVRET) { |
118 | if (GIMME == G_SCALAR) | |
119 | Perl_croak(aTHX_ "Can't return hash to lvalue scalar context"); | |
120 | RETURN; | |
121 | } | |
54310121 | 122 | gimme = GIMME_V; |
123 | if (gimme == G_ARRAY) { | |
cea2e8a9 | 124 | RETURNOP(do_kv()); |
85e6fe83 | 125 | } |
54310121 | 126 | else if (gimme == G_SCALAR) { |
85fbaab2 | 127 | SV* const sv = Perl_hv_scalar(aTHX_ MUTABLE_HV(TARG)); |
85e6fe83 | 128 | SETs(sv); |
85e6fe83 | 129 | } |
54310121 | 130 | RETURN; |
93a17b20 LW |
131 | } |
132 | ||
79072805 LW |
133 | /* Translations. */ |
134 | ||
def89bff NC |
135 | const char S_no_symref_sv[] = |
136 | "Can't use string (\"%" SVf32 "\"%s) as %s ref while \"strict refs\" in use"; | |
137 | ||
79072805 LW |
138 | PP(pp_rv2gv) |
139 | { | |
97aff369 | 140 | dVAR; dSP; dTOPss; |
8ec5e241 | 141 | |
ed6116ce | 142 | if (SvROK(sv)) { |
a0d0e21e | 143 | wasref: |
f5284f61 IZ |
144 | tryAMAGICunDEREF(to_gv); |
145 | ||
ed6116ce | 146 | sv = SvRV(sv); |
b1dadf13 | 147 | if (SvTYPE(sv) == SVt_PVIO) { |
159b6efe | 148 | GV * const gv = MUTABLE_GV(sv_newmortal()); |
b1dadf13 | 149 | gv_init(gv, 0, "", 0, 0); |
a45c7426 | 150 | GvIOp(gv) = MUTABLE_IO(sv); |
b37c2d43 | 151 | SvREFCNT_inc_void_NN(sv); |
ad64d0ec | 152 | sv = MUTABLE_SV(gv); |
ef54e1a4 | 153 | } |
6e592b3a | 154 | else if (!isGV_with_GP(sv)) |
cea2e8a9 | 155 | DIE(aTHX_ "Not a GLOB reference"); |
79072805 LW |
156 | } |
157 | else { | |
6e592b3a | 158 | if (!isGV_with_GP(sv)) { |
a0d0e21e LW |
159 | if (SvGMAGICAL(sv)) { |
160 | mg_get(sv); | |
161 | if (SvROK(sv)) | |
162 | goto wasref; | |
163 | } | |
afd1915d | 164 | if (!SvOK(sv) && sv != &PL_sv_undef) { |
b13b2135 | 165 | /* If this is a 'my' scalar and flag is set then vivify |
853846ea | 166 | * NI-S 1999/05/07 |
b13b2135 | 167 | */ |
ac53db4c | 168 | if (SvREADONLY(sv)) |
f1f66076 | 169 | Perl_croak(aTHX_ "%s", PL_no_modify); |
1d8d4d2a | 170 | if (PL_op->op_private & OPpDEREF) { |
2c8ac474 GS |
171 | GV *gv; |
172 | if (cUNOP->op_targ) { | |
173 | STRLEN len; | |
0bd48802 AL |
174 | SV * const namesv = PAD_SV(cUNOP->op_targ); |
175 | const char * const name = SvPV(namesv, len); | |
159b6efe | 176 | gv = MUTABLE_GV(newSV(0)); |
2c8ac474 GS |
177 | gv_init(gv, CopSTASH(PL_curcop), name, len, 0); |
178 | } | |
179 | else { | |
0bd48802 | 180 | const char * const name = CopSTASHPV(PL_curcop); |
2c8ac474 | 181 | gv = newGVgen(name); |
1d8d4d2a | 182 | } |
43230e26 | 183 | prepare_SV_for_RV(sv); |
ad64d0ec | 184 | SvRV_set(sv, MUTABLE_SV(gv)); |
853846ea | 185 | SvROK_on(sv); |
1d8d4d2a | 186 | SvSETMAGIC(sv); |
853846ea | 187 | goto wasref; |
2c8ac474 | 188 | } |
533c011a NIS |
189 | if (PL_op->op_flags & OPf_REF || |
190 | PL_op->op_private & HINT_STRICT_REFS) | |
cea2e8a9 | 191 | DIE(aTHX_ PL_no_usym, "a symbol"); |
599cee73 | 192 | if (ckWARN(WARN_UNINITIALIZED)) |
29489e7c | 193 | report_uninit(sv); |
a0d0e21e LW |
194 | RETSETUNDEF; |
195 | } | |
35cd451c GS |
196 | if ((PL_op->op_flags & OPf_SPECIAL) && |
197 | !(PL_op->op_flags & OPf_MOD)) | |
198 | { | |
ad64d0ec | 199 | SV * const temp = MUTABLE_SV(gv_fetchsv(sv, 0, SVt_PVGV)); |
7a5fd60d NC |
200 | if (!temp |
201 | && (!is_gv_magical_sv(sv,0) | |
ad64d0ec NC |
202 | || !(sv = MUTABLE_SV(gv_fetchsv(sv, GV_ADD, |
203 | SVt_PVGV))))) { | |
35cd451c | 204 | RETSETUNDEF; |
c9d5ac95 | 205 | } |
7a5fd60d | 206 | sv = temp; |
35cd451c GS |
207 | } |
208 | else { | |
209 | if (PL_op->op_private & HINT_STRICT_REFS) | |
def89bff | 210 | DIE(aTHX_ S_no_symref_sv, sv, (SvCUR(sv)>32 ? "..." : ""), "a symbol"); |
e26df76a NC |
211 | if ((PL_op->op_private & (OPpLVAL_INTRO|OPpDONT_INIT_GV)) |
212 | == OPpDONT_INIT_GV) { | |
213 | /* We are the target of a coderef assignment. Return | |
214 | the scalar unchanged, and let pp_sasssign deal with | |
215 | things. */ | |
216 | RETURN; | |
217 | } | |
ad64d0ec | 218 | sv = MUTABLE_SV(gv_fetchsv(sv, GV_ADD, SVt_PVGV)); |
35cd451c | 219 | } |
93a17b20 | 220 | } |
79072805 | 221 | } |
533c011a | 222 | if (PL_op->op_private & OPpLVAL_INTRO) |
159b6efe | 223 | save_gp(MUTABLE_GV(sv), !(PL_op->op_flags & OPf_SPECIAL)); |
79072805 LW |
224 | SETs(sv); |
225 | RETURN; | |
226 | } | |
227 | ||
dc3c76f8 NC |
228 | /* Helper function for pp_rv2sv and pp_rv2av */ |
229 | GV * | |
fe9845cc RB |
230 | Perl_softref2xv(pTHX_ SV *const sv, const char *const what, |
231 | const svtype type, SV ***spp) | |
dc3c76f8 NC |
232 | { |
233 | dVAR; | |
234 | GV *gv; | |
235 | ||
7918f24d NC |
236 | PERL_ARGS_ASSERT_SOFTREF2XV; |
237 | ||
dc3c76f8 NC |
238 | if (PL_op->op_private & HINT_STRICT_REFS) { |
239 | if (SvOK(sv)) | |
def89bff | 240 | Perl_die(aTHX_ S_no_symref_sv, sv, (SvCUR(sv)>32 ? "..." : ""), what); |
dc3c76f8 NC |
241 | else |
242 | Perl_die(aTHX_ PL_no_usym, what); | |
243 | } | |
244 | if (!SvOK(sv)) { | |
245 | if (PL_op->op_flags & OPf_REF) | |
246 | Perl_die(aTHX_ PL_no_usym, what); | |
247 | if (ckWARN(WARN_UNINITIALIZED)) | |
248 | report_uninit(sv); | |
249 | if (type != SVt_PV && GIMME_V == G_ARRAY) { | |
250 | (*spp)--; | |
251 | return NULL; | |
252 | } | |
253 | **spp = &PL_sv_undef; | |
254 | return NULL; | |
255 | } | |
256 | if ((PL_op->op_flags & OPf_SPECIAL) && | |
257 | !(PL_op->op_flags & OPf_MOD)) | |
258 | { | |
81e3fc25 | 259 | gv = gv_fetchsv(sv, 0, type); |
dc3c76f8 NC |
260 | if (!gv |
261 | && (!is_gv_magical_sv(sv,0) | |
81e3fc25 | 262 | || !(gv = gv_fetchsv(sv, GV_ADD, type)))) |
dc3c76f8 NC |
263 | { |
264 | **spp = &PL_sv_undef; | |
265 | return NULL; | |
266 | } | |
267 | } | |
268 | else { | |
81e3fc25 | 269 | gv = gv_fetchsv(sv, GV_ADD, type); |
dc3c76f8 NC |
270 | } |
271 | return gv; | |
272 | } | |
273 | ||
79072805 LW |
274 | PP(pp_rv2sv) |
275 | { | |
97aff369 | 276 | dVAR; dSP; dTOPss; |
c445ea15 | 277 | GV *gv = NULL; |
79072805 | 278 | |
ed6116ce | 279 | if (SvROK(sv)) { |
a0d0e21e | 280 | wasref: |
f5284f61 IZ |
281 | tryAMAGICunDEREF(to_sv); |
282 | ||
ed6116ce | 283 | sv = SvRV(sv); |
79072805 LW |
284 | switch (SvTYPE(sv)) { |
285 | case SVt_PVAV: | |
286 | case SVt_PVHV: | |
287 | case SVt_PVCV: | |
cbae9b9f YST |
288 | case SVt_PVFM: |
289 | case SVt_PVIO: | |
cea2e8a9 | 290 | DIE(aTHX_ "Not a SCALAR reference"); |
42d0e0b7 | 291 | default: NOOP; |
79072805 LW |
292 | } |
293 | } | |
294 | else { | |
159b6efe | 295 | gv = MUTABLE_GV(sv); |
748a9306 | 296 | |
6e592b3a | 297 | if (!isGV_with_GP(gv)) { |
a0d0e21e LW |
298 | if (SvGMAGICAL(sv)) { |
299 | mg_get(sv); | |
300 | if (SvROK(sv)) | |
301 | goto wasref; | |
302 | } | |
dc3c76f8 NC |
303 | gv = Perl_softref2xv(aTHX_ sv, "a SCALAR", SVt_PV, &sp); |
304 | if (!gv) | |
305 | RETURN; | |
463ee0b2 | 306 | } |
29c711a3 | 307 | sv = GvSVn(gv); |
a0d0e21e | 308 | } |
533c011a | 309 | if (PL_op->op_flags & OPf_MOD) { |
82d03984 RGS |
310 | if (PL_op->op_private & OPpLVAL_INTRO) { |
311 | if (cUNOP->op_first->op_type == OP_NULL) | |
159b6efe | 312 | sv = save_scalar(MUTABLE_GV(TOPs)); |
82d03984 RGS |
313 | else if (gv) |
314 | sv = save_scalar(gv); | |
315 | else | |
f1f66076 | 316 | Perl_croak(aTHX_ "%s", PL_no_localize_ref); |
82d03984 | 317 | } |
533c011a NIS |
318 | else if (PL_op->op_private & OPpDEREF) |
319 | vivify_ref(sv, PL_op->op_private & OPpDEREF); | |
79072805 | 320 | } |
a0d0e21e | 321 | SETs(sv); |
79072805 LW |
322 | RETURN; |
323 | } | |
324 | ||
325 | PP(pp_av2arylen) | |
326 | { | |
97aff369 | 327 | dVAR; dSP; |
502c6561 | 328 | AV * const av = MUTABLE_AV(TOPs); |
02d85cc3 EB |
329 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
330 | if (lvalue) { | |
331 | SV ** const sv = Perl_av_arylen_p(aTHX_ MUTABLE_AV(av)); | |
332 | if (!*sv) { | |
333 | *sv = newSV_type(SVt_PVMG); | |
334 | sv_magic(*sv, MUTABLE_SV(av), PERL_MAGIC_arylen, NULL, 0); | |
335 | } | |
336 | SETs(*sv); | |
337 | } else { | |
338 | SETs(sv_2mortal(newSViv( | |
339 | AvFILL(MUTABLE_AV(av)) + CopARYBASE_get(PL_curcop) | |
340 | ))); | |
79072805 | 341 | } |
79072805 LW |
342 | RETURN; |
343 | } | |
344 | ||
a0d0e21e LW |
345 | PP(pp_pos) |
346 | { | |
97aff369 | 347 | dVAR; dSP; dTARGET; dPOPss; |
8ec5e241 | 348 | |
78f9721b | 349 | if (PL_op->op_flags & OPf_MOD || LVRET) { |
5f05dabc | 350 | if (SvTYPE(TARG) < SVt_PVLV) { |
351 | sv_upgrade(TARG, SVt_PVLV); | |
c445ea15 | 352 | sv_magic(TARG, NULL, PERL_MAGIC_pos, NULL, 0); |
5f05dabc | 353 | } |
354 | ||
355 | LvTYPE(TARG) = '.'; | |
6ff81951 | 356 | if (LvTARG(TARG) != sv) { |
cb39f75f | 357 | SvREFCNT_dec(LvTARG(TARG)); |
b37c2d43 | 358 | LvTARG(TARG) = SvREFCNT_inc_simple(sv); |
6ff81951 | 359 | } |
a0d0e21e LW |
360 | PUSHs(TARG); /* no SvSETMAGIC */ |
361 | RETURN; | |
362 | } | |
363 | else { | |
a0d0e21e | 364 | if (SvTYPE(sv) >= SVt_PVMG && SvMAGIC(sv)) { |
1b6737cc | 365 | const MAGIC * const mg = mg_find(sv, PERL_MAGIC_regex_global); |
565764a8 | 366 | if (mg && mg->mg_len >= 0) { |
a0ed51b3 | 367 | I32 i = mg->mg_len; |
7e2040f0 | 368 | if (DO_UTF8(sv)) |
a0ed51b3 | 369 | sv_pos_b2u(sv, &i); |
fc15ae8f | 370 | PUSHi(i + CopARYBASE_get(PL_curcop)); |
a0d0e21e LW |
371 | RETURN; |
372 | } | |
373 | } | |
374 | RETPUSHUNDEF; | |
375 | } | |
376 | } | |
377 | ||
79072805 LW |
378 | PP(pp_rv2cv) |
379 | { | |
97aff369 | 380 | dVAR; dSP; |
79072805 | 381 | GV *gv; |
1eced8f8 | 382 | HV *stash_unused; |
c445ea15 AL |
383 | const I32 flags = (PL_op->op_flags & OPf_SPECIAL) |
384 | ? 0 | |
385 | : ((PL_op->op_private & (OPpLVAL_INTRO|OPpMAY_RETURN_CONSTANT)) == OPpMAY_RETURN_CONSTANT) | |
386 | ? GV_ADD|GV_NOEXPAND | |
387 | : GV_ADD; | |
4633a7c4 LW |
388 | /* We usually try to add a non-existent subroutine in case of AUTOLOAD. */ |
389 | /* (But not in defined().) */ | |
e26df76a | 390 | |
1eced8f8 | 391 | CV *cv = sv_2cv(TOPs, &stash_unused, &gv, flags); |
07055b4c CS |
392 | if (cv) { |
393 | if (CvCLONE(cv)) | |
ad64d0ec | 394 | cv = MUTABLE_CV(sv_2mortal(MUTABLE_SV(cv_clone(cv)))); |
d32f2495 SC |
395 | if ((PL_op->op_private & OPpLVAL_INTRO)) { |
396 | if (gv && GvCV(gv) == cv && (gv = gv_autoload4(GvSTASH(gv), GvNAME(gv), GvNAMELEN(gv), FALSE))) | |
397 | cv = GvCV(gv); | |
398 | if (!CvLVALUE(cv)) | |
399 | DIE(aTHX_ "Can't modify non-lvalue subroutine call"); | |
400 | } | |
07055b4c | 401 | } |
e26df76a | 402 | else if ((flags == (GV_ADD|GV_NOEXPAND)) && gv && SvROK(gv)) { |
ea726b52 | 403 | cv = MUTABLE_CV(gv); |
e26df76a | 404 | } |
07055b4c | 405 | else |
ea726b52 | 406 | cv = MUTABLE_CV(&PL_sv_undef); |
ad64d0ec | 407 | SETs(MUTABLE_SV(cv)); |
79072805 LW |
408 | RETURN; |
409 | } | |
410 | ||
c07a80fd | 411 | PP(pp_prototype) |
412 | { | |
97aff369 | 413 | dVAR; dSP; |
c07a80fd | 414 | CV *cv; |
415 | HV *stash; | |
416 | GV *gv; | |
fabdb6c0 | 417 | SV *ret = &PL_sv_undef; |
c07a80fd | 418 | |
b6c543e3 | 419 | if (SvPOK(TOPs) && SvCUR(TOPs) >= 7) { |
e3f73d4e | 420 | const char * s = SvPVX_const(TOPs); |
b6c543e3 | 421 | if (strnEQ(s, "CORE::", 6)) { |
5458a98a | 422 | const int code = keyword(s + 6, SvCUR(TOPs) - 6, 1); |
b6c543e3 IZ |
423 | if (code < 0) { /* Overridable. */ |
424 | #define MAX_ARGS_OP ((sizeof(I32) - 1) * 2) | |
59b085e1 | 425 | int i = 0, n = 0, seen_question = 0, defgv = 0; |
b6c543e3 IZ |
426 | I32 oa; |
427 | char str[ MAX_ARGS_OP * 2 + 2 ]; /* One ';', one '\0' */ | |
428 | ||
bdf1bb36 | 429 | if (code == -KEY_chop || code == -KEY_chomp |
f23102e2 | 430 | || code == -KEY_exec || code == -KEY_system) |
77bc9082 | 431 | goto set; |
d116c547 | 432 | if (code == -KEY_mkdir) { |
84bafc02 | 433 | ret = newSVpvs_flags("_;$", SVs_TEMP); |
d116c547 RGS |
434 | goto set; |
435 | } | |
7c8178a1 RGS |
436 | if (code == -KEY_keys || code == -KEY_values || code == -KEY_each) { |
437 | ret = newSVpvs_flags("\\[@%]", SVs_TEMP); | |
438 | goto set; | |
439 | } | |
e3f73d4e RGS |
440 | if (code == -KEY_readpipe) { |
441 | s = "CORE::backtick"; | |
442 | } | |
b6c543e3 | 443 | while (i < MAXO) { /* The slow way. */ |
22c35a8c GS |
444 | if (strEQ(s + 6, PL_op_name[i]) |
445 | || strEQ(s + 6, PL_op_desc[i])) | |
446 | { | |
b6c543e3 | 447 | goto found; |
22c35a8c | 448 | } |
b6c543e3 IZ |
449 | i++; |
450 | } | |
451 | goto nonesuch; /* Should not happen... */ | |
452 | found: | |
59b085e1 | 453 | defgv = PL_opargs[i] & OA_DEFGV; |
22c35a8c | 454 | oa = PL_opargs[i] >> OASHIFT; |
b6c543e3 | 455 | while (oa) { |
59b085e1 | 456 | if (oa & OA_OPTIONAL && !seen_question && !defgv) { |
b6c543e3 IZ |
457 | seen_question = 1; |
458 | str[n++] = ';'; | |
ef54e1a4 | 459 | } |
b13b2135 | 460 | if ((oa & (OA_OPTIONAL - 1)) >= OA_AVREF |
6e97e420 SC |
461 | && (oa & (OA_OPTIONAL - 1)) <= OA_SCALARREF |
462 | /* But globs are already references (kinda) */ | |
463 | && (oa & (OA_OPTIONAL - 1)) != OA_FILEREF | |
464 | ) { | |
b6c543e3 IZ |
465 | str[n++] = '\\'; |
466 | } | |
b6c543e3 IZ |
467 | str[n++] = ("?$@@%&*$")[oa & (OA_OPTIONAL - 1)]; |
468 | oa = oa >> 4; | |
469 | } | |
59b085e1 RGS |
470 | if (defgv && str[n - 1] == '$') |
471 | str[n - 1] = '_'; | |
b6c543e3 | 472 | str[n++] = '\0'; |
59cd0e26 | 473 | ret = newSVpvn_flags(str, n - 1, SVs_TEMP); |
ef54e1a4 JH |
474 | } |
475 | else if (code) /* Non-Overridable */ | |
b6c543e3 IZ |
476 | goto set; |
477 | else { /* None such */ | |
478 | nonesuch: | |
d470f89e | 479 | DIE(aTHX_ "Can't find an opnumber for \"%s\"", s+6); |
b6c543e3 IZ |
480 | } |
481 | } | |
482 | } | |
f2c0649b | 483 | cv = sv_2cv(TOPs, &stash, &gv, 0); |
5f05dabc | 484 | if (cv && SvPOK(cv)) |
59cd0e26 | 485 | ret = newSVpvn_flags(SvPVX_const(cv), SvCUR(cv), SVs_TEMP); |
b6c543e3 | 486 | set: |
c07a80fd | 487 | SETs(ret); |
488 | RETURN; | |
489 | } | |
490 | ||
a0d0e21e LW |
491 | PP(pp_anoncode) |
492 | { | |
97aff369 | 493 | dVAR; dSP; |
ea726b52 | 494 | CV *cv = MUTABLE_CV(PAD_SV(PL_op->op_targ)); |
a5f75d66 | 495 | if (CvCLONE(cv)) |
ad64d0ec | 496 | cv = MUTABLE_CV(sv_2mortal(MUTABLE_SV(cv_clone(cv)))); |
5f05dabc | 497 | EXTEND(SP,1); |
ad64d0ec | 498 | PUSHs(MUTABLE_SV(cv)); |
a0d0e21e LW |
499 | RETURN; |
500 | } | |
501 | ||
502 | PP(pp_srefgen) | |
79072805 | 503 | { |
97aff369 | 504 | dVAR; dSP; |
71be2cbc | 505 | *SP = refto(*SP); |
79072805 | 506 | RETURN; |
8ec5e241 | 507 | } |
a0d0e21e LW |
508 | |
509 | PP(pp_refgen) | |
510 | { | |
97aff369 | 511 | dVAR; dSP; dMARK; |
a0d0e21e | 512 | if (GIMME != G_ARRAY) { |
5f0b1d4e GS |
513 | if (++MARK <= SP) |
514 | *MARK = *SP; | |
515 | else | |
3280af22 | 516 | *MARK = &PL_sv_undef; |
5f0b1d4e GS |
517 | *MARK = refto(*MARK); |
518 | SP = MARK; | |
519 | RETURN; | |
a0d0e21e | 520 | } |
bbce6d69 | 521 | EXTEND_MORTAL(SP - MARK); |
71be2cbc | 522 | while (++MARK <= SP) |
523 | *MARK = refto(*MARK); | |
a0d0e21e | 524 | RETURN; |
79072805 LW |
525 | } |
526 | ||
76e3520e | 527 | STATIC SV* |
cea2e8a9 | 528 | S_refto(pTHX_ SV *sv) |
71be2cbc | 529 | { |
97aff369 | 530 | dVAR; |
71be2cbc | 531 | SV* rv; |
532 | ||
7918f24d NC |
533 | PERL_ARGS_ASSERT_REFTO; |
534 | ||
71be2cbc | 535 | if (SvTYPE(sv) == SVt_PVLV && LvTYPE(sv) == 'y') { |
536 | if (LvTARGLEN(sv)) | |
68dc0745 | 537 | vivify_defelem(sv); |
538 | if (!(sv = LvTARG(sv))) | |
3280af22 | 539 | sv = &PL_sv_undef; |
0dd88869 | 540 | else |
b37c2d43 | 541 | SvREFCNT_inc_void_NN(sv); |
71be2cbc | 542 | } |
d8b46c1b | 543 | else if (SvTYPE(sv) == SVt_PVAV) { |
502c6561 NC |
544 | if (!AvREAL((const AV *)sv) && AvREIFY((const AV *)sv)) |
545 | av_reify(MUTABLE_AV(sv)); | |
d8b46c1b | 546 | SvTEMP_off(sv); |
b37c2d43 | 547 | SvREFCNT_inc_void_NN(sv); |
d8b46c1b | 548 | } |
f2933f5f DM |
549 | else if (SvPADTMP(sv) && !IS_PADGV(sv)) |
550 | sv = newSVsv(sv); | |
71be2cbc | 551 | else { |
552 | SvTEMP_off(sv); | |
b37c2d43 | 553 | SvREFCNT_inc_void_NN(sv); |
71be2cbc | 554 | } |
555 | rv = sv_newmortal(); | |
4df7f6af | 556 | sv_upgrade(rv, SVt_IV); |
b162af07 | 557 | SvRV_set(rv, sv); |
71be2cbc | 558 | SvROK_on(rv); |
559 | return rv; | |
560 | } | |
561 | ||
79072805 LW |
562 | PP(pp_ref) |
563 | { | |
97aff369 | 564 | dVAR; dSP; dTARGET; |
e1ec3a88 | 565 | const char *pv; |
1b6737cc | 566 | SV * const sv = POPs; |
f12c7020 | 567 | |
5b295bef RD |
568 | if (sv) |
569 | SvGETMAGIC(sv); | |
f12c7020 | 570 | |
a0d0e21e | 571 | if (!sv || !SvROK(sv)) |
4633a7c4 | 572 | RETPUSHNO; |
79072805 | 573 | |
1b6737cc | 574 | pv = sv_reftype(SvRV(sv),TRUE); |
463ee0b2 | 575 | PUSHp(pv, strlen(pv)); |
79072805 LW |
576 | RETURN; |
577 | } | |
578 | ||
579 | PP(pp_bless) | |
580 | { | |
97aff369 | 581 | dVAR; dSP; |
463ee0b2 | 582 | HV *stash; |
79072805 | 583 | |
463ee0b2 | 584 | if (MAXARG == 1) |
11faa288 | 585 | stash = CopSTASH(PL_curcop); |
7b8d334a | 586 | else { |
1b6737cc | 587 | SV * const ssv = POPs; |
7b8d334a | 588 | STRLEN len; |
e1ec3a88 | 589 | const char *ptr; |
81689caa | 590 | |
016a42f3 | 591 | if (ssv && !SvGMAGICAL(ssv) && !SvAMAGIC(ssv) && SvROK(ssv)) |
81689caa | 592 | Perl_croak(aTHX_ "Attempt to bless into a reference"); |
5c144d81 | 593 | ptr = SvPV_const(ssv,len); |
a2a5de95 NC |
594 | if (len == 0) |
595 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), | |
596 | "Explicit blessing to '' (assuming package main)"); | |
da51bb9b | 597 | stash = gv_stashpvn(ptr, len, GV_ADD); |
7b8d334a | 598 | } |
a0d0e21e | 599 | |
5d3fdfeb | 600 | (void)sv_bless(TOPs, stash); |
79072805 LW |
601 | RETURN; |
602 | } | |
603 | ||
fb73857a | 604 | PP(pp_gelem) |
605 | { | |
97aff369 | 606 | dVAR; dSP; |
b13b2135 | 607 | |
1b6737cc AL |
608 | SV *sv = POPs; |
609 | const char * const elem = SvPV_nolen_const(sv); | |
159b6efe | 610 | GV * const gv = MUTABLE_GV(POPs); |
c445ea15 | 611 | SV * tmpRef = NULL; |
1b6737cc | 612 | |
c445ea15 | 613 | sv = NULL; |
c4ba80c3 NC |
614 | if (elem) { |
615 | /* elem will always be NUL terminated. */ | |
1b6737cc | 616 | const char * const second_letter = elem + 1; |
c4ba80c3 NC |
617 | switch (*elem) { |
618 | case 'A': | |
1b6737cc | 619 | if (strEQ(second_letter, "RRAY")) |
ad64d0ec | 620 | tmpRef = MUTABLE_SV(GvAV(gv)); |
c4ba80c3 NC |
621 | break; |
622 | case 'C': | |
1b6737cc | 623 | if (strEQ(second_letter, "ODE")) |
ad64d0ec | 624 | tmpRef = MUTABLE_SV(GvCVu(gv)); |
c4ba80c3 NC |
625 | break; |
626 | case 'F': | |
1b6737cc | 627 | if (strEQ(second_letter, "ILEHANDLE")) { |
c4ba80c3 NC |
628 | /* finally deprecated in 5.8.0 */ |
629 | deprecate("*glob{FILEHANDLE}"); | |
ad64d0ec | 630 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
c4ba80c3 NC |
631 | } |
632 | else | |
1b6737cc | 633 | if (strEQ(second_letter, "ORMAT")) |
ad64d0ec | 634 | tmpRef = MUTABLE_SV(GvFORM(gv)); |
c4ba80c3 NC |
635 | break; |
636 | case 'G': | |
1b6737cc | 637 | if (strEQ(second_letter, "LOB")) |
ad64d0ec | 638 | tmpRef = MUTABLE_SV(gv); |
c4ba80c3 NC |
639 | break; |
640 | case 'H': | |
1b6737cc | 641 | if (strEQ(second_letter, "ASH")) |
ad64d0ec | 642 | tmpRef = MUTABLE_SV(GvHV(gv)); |
c4ba80c3 NC |
643 | break; |
644 | case 'I': | |
1b6737cc | 645 | if (*second_letter == 'O' && !elem[2]) |
ad64d0ec | 646 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
c4ba80c3 NC |
647 | break; |
648 | case 'N': | |
1b6737cc | 649 | if (strEQ(second_letter, "AME")) |
a663657d | 650 | sv = newSVhek(GvNAME_HEK(gv)); |
c4ba80c3 NC |
651 | break; |
652 | case 'P': | |
1b6737cc | 653 | if (strEQ(second_letter, "ACKAGE")) { |
7fa3a4ab NC |
654 | const HV * const stash = GvSTASH(gv); |
655 | const HEK * const hek = stash ? HvNAME_HEK(stash) : NULL; | |
396482e1 | 656 | sv = hek ? newSVhek(hek) : newSVpvs("__ANON__"); |
c4ba80c3 NC |
657 | } |
658 | break; | |
659 | case 'S': | |
1b6737cc | 660 | if (strEQ(second_letter, "CALAR")) |
f9d52e31 | 661 | tmpRef = GvSVn(gv); |
c4ba80c3 | 662 | break; |
39b99f21 | 663 | } |
fb73857a | 664 | } |
76e3520e GS |
665 | if (tmpRef) |
666 | sv = newRV(tmpRef); | |
fb73857a | 667 | if (sv) |
668 | sv_2mortal(sv); | |
669 | else | |
3280af22 | 670 | sv = &PL_sv_undef; |
fb73857a | 671 | XPUSHs(sv); |
672 | RETURN; | |
673 | } | |
674 | ||
a0d0e21e | 675 | /* Pattern matching */ |
79072805 | 676 | |
a0d0e21e | 677 | PP(pp_study) |
79072805 | 678 | { |
97aff369 | 679 | dVAR; dSP; dPOPss; |
a0d0e21e LW |
680 | register unsigned char *s; |
681 | register I32 pos; | |
682 | register I32 ch; | |
683 | register I32 *sfirst; | |
684 | register I32 *snext; | |
a0d0e21e LW |
685 | STRLEN len; |
686 | ||
3280af22 | 687 | if (sv == PL_lastscream) { |
1e422769 | 688 | if (SvSCREAM(sv)) |
689 | RETPUSHYES; | |
690 | } | |
a4f4e906 NC |
691 | s = (unsigned char*)(SvPV(sv, len)); |
692 | pos = len; | |
c9b9f909 | 693 | if (pos <= 0 || !SvPOK(sv) || SvUTF8(sv)) { |
a4f4e906 NC |
694 | /* No point in studying a zero length string, and not safe to study |
695 | anything that doesn't appear to be a simple scalar (and hence might | |
696 | change between now and when the regexp engine runs without our set | |
bd473224 | 697 | magic ever running) such as a reference to an object with overloaded |
a4f4e906 NC |
698 | stringification. */ |
699 | RETPUSHNO; | |
700 | } | |
701 | ||
702 | if (PL_lastscream) { | |
703 | SvSCREAM_off(PL_lastscream); | |
704 | SvREFCNT_dec(PL_lastscream); | |
c07a80fd | 705 | } |
b37c2d43 | 706 | PL_lastscream = SvREFCNT_inc_simple(sv); |
1e422769 | 707 | |
708 | s = (unsigned char*)(SvPV(sv, len)); | |
709 | pos = len; | |
710 | if (pos <= 0) | |
711 | RETPUSHNO; | |
3280af22 NIS |
712 | if (pos > PL_maxscream) { |
713 | if (PL_maxscream < 0) { | |
714 | PL_maxscream = pos + 80; | |
a02a5408 JC |
715 | Newx(PL_screamfirst, 256, I32); |
716 | Newx(PL_screamnext, PL_maxscream, I32); | |
79072805 LW |
717 | } |
718 | else { | |
3280af22 NIS |
719 | PL_maxscream = pos + pos / 4; |
720 | Renew(PL_screamnext, PL_maxscream, I32); | |
79072805 | 721 | } |
79072805 | 722 | } |
a0d0e21e | 723 | |
3280af22 NIS |
724 | sfirst = PL_screamfirst; |
725 | snext = PL_screamnext; | |
a0d0e21e LW |
726 | |
727 | if (!sfirst || !snext) | |
cea2e8a9 | 728 | DIE(aTHX_ "do_study: out of memory"); |
a0d0e21e LW |
729 | |
730 | for (ch = 256; ch; --ch) | |
731 | *sfirst++ = -1; | |
732 | sfirst -= 256; | |
733 | ||
734 | while (--pos >= 0) { | |
1b6737cc | 735 | register const I32 ch = s[pos]; |
a0d0e21e LW |
736 | if (sfirst[ch] >= 0) |
737 | snext[pos] = sfirst[ch] - pos; | |
738 | else | |
739 | snext[pos] = -pos; | |
740 | sfirst[ch] = pos; | |
79072805 LW |
741 | } |
742 | ||
c07a80fd | 743 | SvSCREAM_on(sv); |
14befaf4 | 744 | /* piggyback on m//g magic */ |
c445ea15 | 745 | sv_magic(sv, NULL, PERL_MAGIC_regex_global, NULL, 0); |
1e422769 | 746 | RETPUSHYES; |
79072805 LW |
747 | } |
748 | ||
a0d0e21e | 749 | PP(pp_trans) |
79072805 | 750 | { |
97aff369 | 751 | dVAR; dSP; dTARG; |
a0d0e21e LW |
752 | SV *sv; |
753 | ||
533c011a | 754 | if (PL_op->op_flags & OPf_STACKED) |
a0d0e21e | 755 | sv = POPs; |
59f00321 RGS |
756 | else if (PL_op->op_private & OPpTARGET_MY) |
757 | sv = GETTARGET; | |
79072805 | 758 | else { |
54b9620d | 759 | sv = DEFSV; |
a0d0e21e | 760 | EXTEND(SP,1); |
79072805 | 761 | } |
adbc6bb1 | 762 | TARG = sv_newmortal(); |
4757a243 | 763 | PUSHi(do_trans(sv)); |
a0d0e21e | 764 | RETURN; |
79072805 LW |
765 | } |
766 | ||
a0d0e21e | 767 | /* Lvalue operators. */ |
79072805 | 768 | |
a0d0e21e LW |
769 | PP(pp_schop) |
770 | { | |
97aff369 | 771 | dVAR; dSP; dTARGET; |
a0d0e21e LW |
772 | do_chop(TARG, TOPs); |
773 | SETTARG; | |
774 | RETURN; | |
79072805 LW |
775 | } |
776 | ||
a0d0e21e | 777 | PP(pp_chop) |
79072805 | 778 | { |
97aff369 | 779 | dVAR; dSP; dMARK; dTARGET; dORIGMARK; |
2ec6af5f RG |
780 | while (MARK < SP) |
781 | do_chop(TARG, *++MARK); | |
782 | SP = ORIGMARK; | |
b59aed67 | 783 | XPUSHTARG; |
a0d0e21e | 784 | RETURN; |
79072805 LW |
785 | } |
786 | ||
a0d0e21e | 787 | PP(pp_schomp) |
79072805 | 788 | { |
97aff369 | 789 | dVAR; dSP; dTARGET; |
a0d0e21e LW |
790 | SETi(do_chomp(TOPs)); |
791 | RETURN; | |
79072805 LW |
792 | } |
793 | ||
a0d0e21e | 794 | PP(pp_chomp) |
79072805 | 795 | { |
97aff369 | 796 | dVAR; dSP; dMARK; dTARGET; |
a0d0e21e | 797 | register I32 count = 0; |
8ec5e241 | 798 | |
a0d0e21e LW |
799 | while (SP > MARK) |
800 | count += do_chomp(POPs); | |
b59aed67 | 801 | XPUSHi(count); |
a0d0e21e | 802 | RETURN; |
79072805 LW |
803 | } |
804 | ||
a0d0e21e LW |
805 | PP(pp_undef) |
806 | { | |
97aff369 | 807 | dVAR; dSP; |
a0d0e21e LW |
808 | SV *sv; |
809 | ||
533c011a | 810 | if (!PL_op->op_private) { |
774d564b | 811 | EXTEND(SP, 1); |
a0d0e21e | 812 | RETPUSHUNDEF; |
774d564b | 813 | } |
79072805 | 814 | |
a0d0e21e LW |
815 | sv = POPs; |
816 | if (!sv) | |
817 | RETPUSHUNDEF; | |
85e6fe83 | 818 | |
765f542d | 819 | SV_CHECK_THINKFIRST_COW_DROP(sv); |
85e6fe83 | 820 | |
a0d0e21e LW |
821 | switch (SvTYPE(sv)) { |
822 | case SVt_NULL: | |
823 | break; | |
824 | case SVt_PVAV: | |
502c6561 | 825 | av_undef(MUTABLE_AV(sv)); |
a0d0e21e LW |
826 | break; |
827 | case SVt_PVHV: | |
85fbaab2 | 828 | hv_undef(MUTABLE_HV(sv)); |
a0d0e21e LW |
829 | break; |
830 | case SVt_PVCV: | |
a2a5de95 NC |
831 | if (cv_const_sv((const CV *)sv)) |
832 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "Constant subroutine %s undefined", | |
833 | CvANON((const CV *)sv) ? "(anonymous)" | |
834 | : GvENAME(CvGV((const CV *)sv))); | |
5f66b61c | 835 | /* FALLTHROUGH */ |
9607fc9c | 836 | case SVt_PVFM: |
6fc92669 GS |
837 | { |
838 | /* let user-undef'd sub keep its identity */ | |
ea726b52 NC |
839 | GV* const gv = CvGV((const CV *)sv); |
840 | cv_undef(MUTABLE_CV(sv)); | |
841 | CvGV((const CV *)sv) = gv; | |
6fc92669 | 842 | } |
a0d0e21e | 843 | break; |
8e07c86e | 844 | case SVt_PVGV: |
6e592b3a | 845 | if (SvFAKE(sv)) { |
3280af22 | 846 | SvSetMagicSV(sv, &PL_sv_undef); |
6e592b3a BM |
847 | break; |
848 | } | |
849 | else if (isGV_with_GP(sv)) { | |
20408e3c | 850 | GP *gp; |
dd69841b BB |
851 | HV *stash; |
852 | ||
853 | /* undef *Foo:: */ | |
159b6efe | 854 | if((stash = GvHV((const GV *)sv)) && HvNAME_get(stash)) |
dd69841b BB |
855 | mro_isa_changed_in(stash); |
856 | /* undef *Pkg::meth_name ... */ | |
159b6efe NC |
857 | else if(GvCVu((const GV *)sv) && (stash = GvSTASH((const GV *)sv)) |
858 | && HvNAME_get(stash)) | |
dd69841b BB |
859 | mro_method_changed_in(stash); |
860 | ||
159b6efe | 861 | gp_free(MUTABLE_GV(sv)); |
a02a5408 | 862 | Newxz(gp, 1, GP); |
20408e3c | 863 | GvGP(sv) = gp_ref(gp); |
561b68a9 | 864 | GvSV(sv) = newSV(0); |
57843af0 | 865 | GvLINE(sv) = CopLINE(PL_curcop); |
159b6efe | 866 | GvEGV(sv) = MUTABLE_GV(sv); |
20408e3c | 867 | GvMULTI_on(sv); |
6e592b3a | 868 | break; |
20408e3c | 869 | } |
6e592b3a | 870 | /* FALL THROUGH */ |
a0d0e21e | 871 | default: |
b15aece3 | 872 | if (SvTYPE(sv) >= SVt_PV && SvPVX_const(sv) && SvLEN(sv)) { |
8bd4d4c5 | 873 | SvPV_free(sv); |
c445ea15 | 874 | SvPV_set(sv, NULL); |
4633a7c4 | 875 | SvLEN_set(sv, 0); |
a0d0e21e | 876 | } |
0c34ef67 | 877 | SvOK_off(sv); |
4633a7c4 | 878 | SvSETMAGIC(sv); |
79072805 | 879 | } |
a0d0e21e LW |
880 | |
881 | RETPUSHUNDEF; | |
79072805 LW |
882 | } |
883 | ||
a0d0e21e | 884 | PP(pp_predec) |
79072805 | 885 | { |
97aff369 | 886 | dVAR; dSP; |
6e592b3a | 887 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
f1f66076 | 888 | DIE(aTHX_ "%s", PL_no_modify); |
3510b4a1 NC |
889 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
890 | && SvIVX(TOPs) != IV_MIN) | |
55497cff | 891 | { |
45977657 | 892 | SvIV_set(TOPs, SvIVX(TOPs) - 1); |
55497cff | 893 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
748a9306 LW |
894 | } |
895 | else | |
896 | sv_dec(TOPs); | |
a0d0e21e LW |
897 | SvSETMAGIC(TOPs); |
898 | return NORMAL; | |
899 | } | |
79072805 | 900 | |
a0d0e21e LW |
901 | PP(pp_postinc) |
902 | { | |
97aff369 | 903 | dVAR; dSP; dTARGET; |
6e592b3a | 904 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
f1f66076 | 905 | DIE(aTHX_ "%s", PL_no_modify); |
a0d0e21e | 906 | sv_setsv(TARG, TOPs); |
3510b4a1 NC |
907 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
908 | && SvIVX(TOPs) != IV_MAX) | |
55497cff | 909 | { |
45977657 | 910 | SvIV_set(TOPs, SvIVX(TOPs) + 1); |
55497cff | 911 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
748a9306 LW |
912 | } |
913 | else | |
914 | sv_inc(TOPs); | |
a0d0e21e | 915 | SvSETMAGIC(TOPs); |
1e54a23f | 916 | /* special case for undef: see thread at 2003-03/msg00536.html in archive */ |
a0d0e21e LW |
917 | if (!SvOK(TARG)) |
918 | sv_setiv(TARG, 0); | |
919 | SETs(TARG); | |
920 | return NORMAL; | |
921 | } | |
79072805 | 922 | |
a0d0e21e LW |
923 | PP(pp_postdec) |
924 | { | |
97aff369 | 925 | dVAR; dSP; dTARGET; |
6e592b3a | 926 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
f1f66076 | 927 | DIE(aTHX_ "%s", PL_no_modify); |
a0d0e21e | 928 | sv_setsv(TARG, TOPs); |
3510b4a1 NC |
929 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
930 | && SvIVX(TOPs) != IV_MIN) | |
55497cff | 931 | { |
45977657 | 932 | SvIV_set(TOPs, SvIVX(TOPs) - 1); |
55497cff | 933 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
748a9306 LW |
934 | } |
935 | else | |
936 | sv_dec(TOPs); | |
a0d0e21e LW |
937 | SvSETMAGIC(TOPs); |
938 | SETs(TARG); | |
939 | return NORMAL; | |
940 | } | |
79072805 | 941 | |
a0d0e21e LW |
942 | /* Ordinary operators. */ |
943 | ||
944 | PP(pp_pow) | |
945 | { | |
800401ee | 946 | dVAR; dSP; dATARGET; SV *svl, *svr; |
58d76dfd | 947 | #ifdef PERL_PRESERVE_IVUV |
52a96ae6 HS |
948 | bool is_int = 0; |
949 | #endif | |
950 | tryAMAGICbin(pow,opASSIGN); | |
800401ee JH |
951 | svl = sv_2num(TOPm1s); |
952 | svr = sv_2num(TOPs); | |
52a96ae6 HS |
953 | #ifdef PERL_PRESERVE_IVUV |
954 | /* For integer to integer power, we do the calculation by hand wherever | |
955 | we're sure it is safe; otherwise we call pow() and try to convert to | |
956 | integer afterwards. */ | |
58d76dfd | 957 | { |
800401ee JH |
958 | SvIV_please(svr); |
959 | if (SvIOK(svr)) { | |
960 | SvIV_please(svl); | |
961 | if (SvIOK(svl)) { | |
900658e3 PF |
962 | UV power; |
963 | bool baseuok; | |
964 | UV baseuv; | |
965 | ||
800401ee JH |
966 | if (SvUOK(svr)) { |
967 | power = SvUVX(svr); | |
900658e3 | 968 | } else { |
800401ee | 969 | const IV iv = SvIVX(svr); |
900658e3 PF |
970 | if (iv >= 0) { |
971 | power = iv; | |
972 | } else { | |
973 | goto float_it; /* Can't do negative powers this way. */ | |
974 | } | |
975 | } | |
976 | ||
800401ee | 977 | baseuok = SvUOK(svl); |
900658e3 | 978 | if (baseuok) { |
800401ee | 979 | baseuv = SvUVX(svl); |
900658e3 | 980 | } else { |
800401ee | 981 | const IV iv = SvIVX(svl); |
900658e3 PF |
982 | if (iv >= 0) { |
983 | baseuv = iv; | |
984 | baseuok = TRUE; /* effectively it's a UV now */ | |
985 | } else { | |
986 | baseuv = -iv; /* abs, baseuok == false records sign */ | |
987 | } | |
988 | } | |
52a96ae6 HS |
989 | /* now we have integer ** positive integer. */ |
990 | is_int = 1; | |
991 | ||
992 | /* foo & (foo - 1) is zero only for a power of 2. */ | |
58d76dfd | 993 | if (!(baseuv & (baseuv - 1))) { |
52a96ae6 | 994 | /* We are raising power-of-2 to a positive integer. |
58d76dfd JH |
995 | The logic here will work for any base (even non-integer |
996 | bases) but it can be less accurate than | |
997 | pow (base,power) or exp (power * log (base)) when the | |
998 | intermediate values start to spill out of the mantissa. | |
999 | With powers of 2 we know this can't happen. | |
1000 | And powers of 2 are the favourite thing for perl | |
1001 | programmers to notice ** not doing what they mean. */ | |
1002 | NV result = 1.0; | |
1003 | NV base = baseuok ? baseuv : -(NV)baseuv; | |
900658e3 PF |
1004 | |
1005 | if (power & 1) { | |
1006 | result *= base; | |
1007 | } | |
1008 | while (power >>= 1) { | |
1009 | base *= base; | |
1010 | if (power & 1) { | |
1011 | result *= base; | |
1012 | } | |
1013 | } | |
58d76dfd JH |
1014 | SP--; |
1015 | SETn( result ); | |
800401ee | 1016 | SvIV_please(svr); |
58d76dfd | 1017 | RETURN; |
52a96ae6 HS |
1018 | } else { |
1019 | register unsigned int highbit = 8 * sizeof(UV); | |
900658e3 PF |
1020 | register unsigned int diff = 8 * sizeof(UV); |
1021 | while (diff >>= 1) { | |
1022 | highbit -= diff; | |
1023 | if (baseuv >> highbit) { | |
1024 | highbit += diff; | |
1025 | } | |
52a96ae6 HS |
1026 | } |
1027 | /* we now have baseuv < 2 ** highbit */ | |
1028 | if (power * highbit <= 8 * sizeof(UV)) { | |
1029 | /* result will definitely fit in UV, so use UV math | |
1030 | on same algorithm as above */ | |
1031 | register UV result = 1; | |
1032 | register UV base = baseuv; | |
900658e3 PF |
1033 | const bool odd_power = (bool)(power & 1); |
1034 | if (odd_power) { | |
1035 | result *= base; | |
1036 | } | |
1037 | while (power >>= 1) { | |
1038 | base *= base; | |
1039 | if (power & 1) { | |
52a96ae6 | 1040 | result *= base; |
52a96ae6 HS |
1041 | } |
1042 | } | |
1043 | SP--; | |
0615a994 | 1044 | if (baseuok || !odd_power) |
52a96ae6 HS |
1045 | /* answer is positive */ |
1046 | SETu( result ); | |
1047 | else if (result <= (UV)IV_MAX) | |
1048 | /* answer negative, fits in IV */ | |
1049 | SETi( -(IV)result ); | |
1050 | else if (result == (UV)IV_MIN) | |
1051 | /* 2's complement assumption: special case IV_MIN */ | |
1052 | SETi( IV_MIN ); | |
1053 | else | |
1054 | /* answer negative, doesn't fit */ | |
1055 | SETn( -(NV)result ); | |
1056 | RETURN; | |
1057 | } | |
1058 | } | |
1059 | } | |
1060 | } | |
58d76dfd | 1061 | } |
52a96ae6 | 1062 | float_it: |
58d76dfd | 1063 | #endif |
a0d0e21e | 1064 | { |
4efa5a16 RD |
1065 | NV right = SvNV(svr); |
1066 | NV left = SvNV(svl); | |
1067 | (void)POPs; | |
3aaeb624 JA |
1068 | |
1069 | #if defined(USE_LONG_DOUBLE) && defined(HAS_AIX_POWL_NEG_BASE_BUG) | |
1070 | /* | |
1071 | We are building perl with long double support and are on an AIX OS | |
1072 | afflicted with a powl() function that wrongly returns NaNQ for any | |
1073 | negative base. This was reported to IBM as PMR #23047-379 on | |
1074 | 03/06/2006. The problem exists in at least the following versions | |
1075 | of AIX and the libm fileset, and no doubt others as well: | |
1076 | ||
1077 | AIX 4.3.3-ML10 bos.adt.libm 4.3.3.50 | |
1078 | AIX 5.1.0-ML04 bos.adt.libm 5.1.0.29 | |
1079 | AIX 5.2.0 bos.adt.libm 5.2.0.85 | |
1080 | ||
1081 | So, until IBM fixes powl(), we provide the following workaround to | |
1082 | handle the problem ourselves. Our logic is as follows: for | |
1083 | negative bases (left), we use fmod(right, 2) to check if the | |
1084 | exponent is an odd or even integer: | |
1085 | ||
1086 | - if odd, powl(left, right) == -powl(-left, right) | |
1087 | - if even, powl(left, right) == powl(-left, right) | |
1088 | ||
1089 | If the exponent is not an integer, the result is rightly NaNQ, so | |
1090 | we just return that (as NV_NAN). | |
1091 | */ | |
1092 | ||
1093 | if (left < 0.0) { | |
1094 | NV mod2 = Perl_fmod( right, 2.0 ); | |
1095 | if (mod2 == 1.0 || mod2 == -1.0) { /* odd integer */ | |
1096 | SETn( -Perl_pow( -left, right) ); | |
1097 | } else if (mod2 == 0.0) { /* even integer */ | |
1098 | SETn( Perl_pow( -left, right) ); | |
1099 | } else { /* fractional power */ | |
1100 | SETn( NV_NAN ); | |
1101 | } | |
1102 | } else { | |
1103 | SETn( Perl_pow( left, right) ); | |
1104 | } | |
1105 | #else | |
52a96ae6 | 1106 | SETn( Perl_pow( left, right) ); |
3aaeb624 JA |
1107 | #endif /* HAS_AIX_POWL_NEG_BASE_BUG */ |
1108 | ||
52a96ae6 HS |
1109 | #ifdef PERL_PRESERVE_IVUV |
1110 | if (is_int) | |
800401ee | 1111 | SvIV_please(svr); |
52a96ae6 HS |
1112 | #endif |
1113 | RETURN; | |
93a17b20 | 1114 | } |
a0d0e21e LW |
1115 | } |
1116 | ||
1117 | PP(pp_multiply) | |
1118 | { | |
800401ee JH |
1119 | dVAR; dSP; dATARGET; SV *svl, *svr; |
1120 | tryAMAGICbin(mult,opASSIGN); | |
1121 | svl = sv_2num(TOPm1s); | |
1122 | svr = sv_2num(TOPs); | |
28e5dec8 | 1123 | #ifdef PERL_PRESERVE_IVUV |
800401ee JH |
1124 | SvIV_please(svr); |
1125 | if (SvIOK(svr)) { | |
28e5dec8 JH |
1126 | /* Unless the left argument is integer in range we are going to have to |
1127 | use NV maths. Hence only attempt to coerce the right argument if | |
1128 | we know the left is integer. */ | |
1129 | /* Left operand is defined, so is it IV? */ | |
800401ee JH |
1130 | SvIV_please(svl); |
1131 | if (SvIOK(svl)) { | |
1132 | bool auvok = SvUOK(svl); | |
1133 | bool buvok = SvUOK(svr); | |
28e5dec8 JH |
1134 | const UV topmask = (~ (UV)0) << (4 * sizeof (UV)); |
1135 | const UV botmask = ~((~ (UV)0) << (4 * sizeof (UV))); | |
1136 | UV alow; | |
1137 | UV ahigh; | |
1138 | UV blow; | |
1139 | UV bhigh; | |
1140 | ||
1141 | if (auvok) { | |
800401ee | 1142 | alow = SvUVX(svl); |
28e5dec8 | 1143 | } else { |
800401ee | 1144 | const IV aiv = SvIVX(svl); |
28e5dec8 JH |
1145 | if (aiv >= 0) { |
1146 | alow = aiv; | |
1147 | auvok = TRUE; /* effectively it's a UV now */ | |
1148 | } else { | |
1149 | alow = -aiv; /* abs, auvok == false records sign */ | |
1150 | } | |
1151 | } | |
1152 | if (buvok) { | |
800401ee | 1153 | blow = SvUVX(svr); |
28e5dec8 | 1154 | } else { |
800401ee | 1155 | const IV biv = SvIVX(svr); |
28e5dec8 JH |
1156 | if (biv >= 0) { |
1157 | blow = biv; | |
1158 | buvok = TRUE; /* effectively it's a UV now */ | |
1159 | } else { | |
1160 | blow = -biv; /* abs, buvok == false records sign */ | |
1161 | } | |
1162 | } | |
1163 | ||
1164 | /* If this does sign extension on unsigned it's time for plan B */ | |
1165 | ahigh = alow >> (4 * sizeof (UV)); | |
1166 | alow &= botmask; | |
1167 | bhigh = blow >> (4 * sizeof (UV)); | |
1168 | blow &= botmask; | |
1169 | if (ahigh && bhigh) { | |
6f207bd3 | 1170 | NOOP; |
28e5dec8 JH |
1171 | /* eg 32 bit is at least 0x10000 * 0x10000 == 0x100000000 |
1172 | which is overflow. Drop to NVs below. */ | |
1173 | } else if (!ahigh && !bhigh) { | |
1174 | /* eg 32 bit is at most 0xFFFF * 0xFFFF == 0xFFFE0001 | |
1175 | so the unsigned multiply cannot overflow. */ | |
c445ea15 | 1176 | const UV product = alow * blow; |
28e5dec8 JH |
1177 | if (auvok == buvok) { |
1178 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
1179 | SP--; | |
1180 | SETu( product ); | |
1181 | RETURN; | |
1182 | } else if (product <= (UV)IV_MIN) { | |
1183 | /* 2s complement assumption that (UV)-IV_MIN is correct. */ | |
1184 | /* -ve result, which could overflow an IV */ | |
1185 | SP--; | |
25716404 | 1186 | SETi( -(IV)product ); |
28e5dec8 JH |
1187 | RETURN; |
1188 | } /* else drop to NVs below. */ | |
1189 | } else { | |
1190 | /* One operand is large, 1 small */ | |
1191 | UV product_middle; | |
1192 | if (bhigh) { | |
1193 | /* swap the operands */ | |
1194 | ahigh = bhigh; | |
1195 | bhigh = blow; /* bhigh now the temp var for the swap */ | |
1196 | blow = alow; | |
1197 | alow = bhigh; | |
1198 | } | |
1199 | /* now, ((ahigh * blow) << half_UV_len) + (alow * blow) | |
1200 | multiplies can't overflow. shift can, add can, -ve can. */ | |
1201 | product_middle = ahigh * blow; | |
1202 | if (!(product_middle & topmask)) { | |
1203 | /* OK, (ahigh * blow) won't lose bits when we shift it. */ | |
1204 | UV product_low; | |
1205 | product_middle <<= (4 * sizeof (UV)); | |
1206 | product_low = alow * blow; | |
1207 | ||
1208 | /* as for pp_add, UV + something mustn't get smaller. | |
1209 | IIRC ANSI mandates this wrapping *behaviour* for | |
1210 | unsigned whatever the actual representation*/ | |
1211 | product_low += product_middle; | |
1212 | if (product_low >= product_middle) { | |
1213 | /* didn't overflow */ | |
1214 | if (auvok == buvok) { | |
1215 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
1216 | SP--; | |
1217 | SETu( product_low ); | |
1218 | RETURN; | |
1219 | } else if (product_low <= (UV)IV_MIN) { | |
1220 | /* 2s complement assumption again */ | |
1221 | /* -ve result, which could overflow an IV */ | |
1222 | SP--; | |
25716404 | 1223 | SETi( -(IV)product_low ); |
28e5dec8 JH |
1224 | RETURN; |
1225 | } /* else drop to NVs below. */ | |
1226 | } | |
1227 | } /* product_middle too large */ | |
1228 | } /* ahigh && bhigh */ | |
800401ee JH |
1229 | } /* SvIOK(svl) */ |
1230 | } /* SvIOK(svr) */ | |
28e5dec8 | 1231 | #endif |
a0d0e21e | 1232 | { |
4efa5a16 RD |
1233 | NV right = SvNV(svr); |
1234 | NV left = SvNV(svl); | |
1235 | (void)POPs; | |
a0d0e21e LW |
1236 | SETn( left * right ); |
1237 | RETURN; | |
79072805 | 1238 | } |
a0d0e21e LW |
1239 | } |
1240 | ||
1241 | PP(pp_divide) | |
1242 | { | |
800401ee JH |
1243 | dVAR; dSP; dATARGET; SV *svl, *svr; |
1244 | tryAMAGICbin(div,opASSIGN); | |
1245 | svl = sv_2num(TOPm1s); | |
1246 | svr = sv_2num(TOPs); | |
5479d192 | 1247 | /* Only try to do UV divide first |
68795e93 | 1248 | if ((SLOPPYDIVIDE is true) or |
5479d192 NC |
1249 | (PERL_PRESERVE_IVUV is true and one or both SV is a UV too large |
1250 | to preserve)) | |
1251 | The assumption is that it is better to use floating point divide | |
1252 | whenever possible, only doing integer divide first if we can't be sure. | |
1253 | If NV_PRESERVES_UV is true then we know at compile time that no UV | |
1254 | can be too large to preserve, so don't need to compile the code to | |
1255 | test the size of UVs. */ | |
1256 | ||
a0d0e21e | 1257 | #ifdef SLOPPYDIVIDE |
5479d192 NC |
1258 | # define PERL_TRY_UV_DIVIDE |
1259 | /* ensure that 20./5. == 4. */ | |
a0d0e21e | 1260 | #else |
5479d192 NC |
1261 | # ifdef PERL_PRESERVE_IVUV |
1262 | # ifndef NV_PRESERVES_UV | |
1263 | # define PERL_TRY_UV_DIVIDE | |
1264 | # endif | |
1265 | # endif | |
a0d0e21e | 1266 | #endif |
5479d192 NC |
1267 | |
1268 | #ifdef PERL_TRY_UV_DIVIDE | |
800401ee JH |
1269 | SvIV_please(svr); |
1270 | if (SvIOK(svr)) { | |
1271 | SvIV_please(svl); | |
1272 | if (SvIOK(svl)) { | |
1273 | bool left_non_neg = SvUOK(svl); | |
1274 | bool right_non_neg = SvUOK(svr); | |
5479d192 NC |
1275 | UV left; |
1276 | UV right; | |
1277 | ||
1278 | if (right_non_neg) { | |
800401ee | 1279 | right = SvUVX(svr); |
5479d192 NC |
1280 | } |
1281 | else { | |
800401ee | 1282 | const IV biv = SvIVX(svr); |
5479d192 NC |
1283 | if (biv >= 0) { |
1284 | right = biv; | |
1285 | right_non_neg = TRUE; /* effectively it's a UV now */ | |
1286 | } | |
1287 | else { | |
1288 | right = -biv; | |
1289 | } | |
1290 | } | |
1291 | /* historically undef()/0 gives a "Use of uninitialized value" | |
1292 | warning before dieing, hence this test goes here. | |
1293 | If it were immediately before the second SvIV_please, then | |
1294 | DIE() would be invoked before left was even inspected, so | |
1295 | no inpsection would give no warning. */ | |
1296 | if (right == 0) | |
1297 | DIE(aTHX_ "Illegal division by zero"); | |
1298 | ||
1299 | if (left_non_neg) { | |
800401ee | 1300 | left = SvUVX(svl); |
5479d192 NC |
1301 | } |
1302 | else { | |
800401ee | 1303 | const IV aiv = SvIVX(svl); |
5479d192 NC |
1304 | if (aiv >= 0) { |
1305 | left = aiv; | |
1306 | left_non_neg = TRUE; /* effectively it's a UV now */ | |
1307 | } | |
1308 | else { | |
1309 | left = -aiv; | |
1310 | } | |
1311 | } | |
1312 | ||
1313 | if (left >= right | |
1314 | #ifdef SLOPPYDIVIDE | |
1315 | /* For sloppy divide we always attempt integer division. */ | |
1316 | #else | |
1317 | /* Otherwise we only attempt it if either or both operands | |
1318 | would not be preserved by an NV. If both fit in NVs | |
0c2ee62a NC |
1319 | we fall through to the NV divide code below. However, |
1320 | as left >= right to ensure integer result here, we know that | |
1321 | we can skip the test on the right operand - right big | |
1322 | enough not to be preserved can't get here unless left is | |
1323 | also too big. */ | |
1324 | ||
1325 | && (left > ((UV)1 << NV_PRESERVES_UV_BITS)) | |
5479d192 NC |
1326 | #endif |
1327 | ) { | |
1328 | /* Integer division can't overflow, but it can be imprecise. */ | |
1b6737cc | 1329 | const UV result = left / right; |
5479d192 NC |
1330 | if (result * right == left) { |
1331 | SP--; /* result is valid */ | |
1332 | if (left_non_neg == right_non_neg) { | |
1333 | /* signs identical, result is positive. */ | |
1334 | SETu( result ); | |
1335 | RETURN; | |
1336 | } | |
1337 | /* 2s complement assumption */ | |
1338 | if (result <= (UV)IV_MIN) | |
91f3b821 | 1339 | SETi( -(IV)result ); |
5479d192 NC |
1340 | else { |
1341 | /* It's exact but too negative for IV. */ | |
1342 | SETn( -(NV)result ); | |
1343 | } | |
1344 | RETURN; | |
1345 | } /* tried integer divide but it was not an integer result */ | |
32fdb065 | 1346 | } /* else (PERL_ABS(result) < 1.0) or (both UVs in range for NV) */ |
5479d192 NC |
1347 | } /* left wasn't SvIOK */ |
1348 | } /* right wasn't SvIOK */ | |
1349 | #endif /* PERL_TRY_UV_DIVIDE */ | |
1350 | { | |
4efa5a16 RD |
1351 | NV right = SvNV(svr); |
1352 | NV left = SvNV(svl); | |
1353 | (void)POPs;(void)POPs; | |
ebc6a117 PD |
1354 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
1355 | if (! Perl_isnan(right) && right == 0.0) | |
1356 | #else | |
5479d192 | 1357 | if (right == 0.0) |
ebc6a117 | 1358 | #endif |
5479d192 NC |
1359 | DIE(aTHX_ "Illegal division by zero"); |
1360 | PUSHn( left / right ); | |
1361 | RETURN; | |
79072805 | 1362 | } |
a0d0e21e LW |
1363 | } |
1364 | ||
1365 | PP(pp_modulo) | |
1366 | { | |
97aff369 | 1367 | dVAR; dSP; dATARGET; tryAMAGICbin(modulo,opASSIGN); |
a0d0e21e | 1368 | { |
9c5ffd7c JH |
1369 | UV left = 0; |
1370 | UV right = 0; | |
dc656993 JH |
1371 | bool left_neg = FALSE; |
1372 | bool right_neg = FALSE; | |
e2c88acc NC |
1373 | bool use_double = FALSE; |
1374 | bool dright_valid = FALSE; | |
9c5ffd7c JH |
1375 | NV dright = 0.0; |
1376 | NV dleft = 0.0; | |
800401ee JH |
1377 | SV * svl; |
1378 | SV * const svr = sv_2num(TOPs); | |
1379 | SvIV_please(svr); | |
1380 | if (SvIOK(svr)) { | |
1381 | right_neg = !SvUOK(svr); | |
e2c88acc | 1382 | if (!right_neg) { |
800401ee | 1383 | right = SvUVX(svr); |
e2c88acc | 1384 | } else { |
800401ee | 1385 | const IV biv = SvIVX(svr); |
e2c88acc NC |
1386 | if (biv >= 0) { |
1387 | right = biv; | |
1388 | right_neg = FALSE; /* effectively it's a UV now */ | |
1389 | } else { | |
1390 | right = -biv; | |
1391 | } | |
1392 | } | |
1393 | } | |
1394 | else { | |
4efa5a16 | 1395 | dright = SvNV(svr); |
787eafbd IZ |
1396 | right_neg = dright < 0; |
1397 | if (right_neg) | |
1398 | dright = -dright; | |
e2c88acc NC |
1399 | if (dright < UV_MAX_P1) { |
1400 | right = U_V(dright); | |
1401 | dright_valid = TRUE; /* In case we need to use double below. */ | |
1402 | } else { | |
1403 | use_double = TRUE; | |
1404 | } | |
787eafbd | 1405 | } |
4efa5a16 | 1406 | sp--; |
a0d0e21e | 1407 | |
e2c88acc NC |
1408 | /* At this point use_double is only true if right is out of range for |
1409 | a UV. In range NV has been rounded down to nearest UV and | |
1410 | use_double false. */ | |
800401ee JH |
1411 | svl = sv_2num(TOPs); |
1412 | SvIV_please(svl); | |
1413 | if (!use_double && SvIOK(svl)) { | |
1414 | if (SvIOK(svl)) { | |
1415 | left_neg = !SvUOK(svl); | |
e2c88acc | 1416 | if (!left_neg) { |
800401ee | 1417 | left = SvUVX(svl); |
e2c88acc | 1418 | } else { |
800401ee | 1419 | const IV aiv = SvIVX(svl); |
e2c88acc NC |
1420 | if (aiv >= 0) { |
1421 | left = aiv; | |
1422 | left_neg = FALSE; /* effectively it's a UV now */ | |
1423 | } else { | |
1424 | left = -aiv; | |
1425 | } | |
1426 | } | |
1427 | } | |
1428 | } | |
787eafbd | 1429 | else { |
4efa5a16 | 1430 | dleft = SvNV(svl); |
787eafbd IZ |
1431 | left_neg = dleft < 0; |
1432 | if (left_neg) | |
1433 | dleft = -dleft; | |
68dc0745 | 1434 | |
e2c88acc NC |
1435 | /* This should be exactly the 5.6 behaviour - if left and right are |
1436 | both in range for UV then use U_V() rather than floor. */ | |
1437 | if (!use_double) { | |
1438 | if (dleft < UV_MAX_P1) { | |
1439 | /* right was in range, so is dleft, so use UVs not double. | |
1440 | */ | |
1441 | left = U_V(dleft); | |
1442 | } | |
1443 | /* left is out of range for UV, right was in range, so promote | |
1444 | right (back) to double. */ | |
1445 | else { | |
1446 | /* The +0.5 is used in 5.6 even though it is not strictly | |
1447 | consistent with the implicit +0 floor in the U_V() | |
1448 | inside the #if 1. */ | |
1449 | dleft = Perl_floor(dleft + 0.5); | |
1450 | use_double = TRUE; | |
1451 | if (dright_valid) | |
1452 | dright = Perl_floor(dright + 0.5); | |
1453 | else | |
1454 | dright = right; | |
1455 | } | |
1456 | } | |
1457 | } | |
4efa5a16 | 1458 | sp--; |
787eafbd | 1459 | if (use_double) { |
65202027 | 1460 | NV dans; |
787eafbd | 1461 | |
787eafbd | 1462 | if (!dright) |
cea2e8a9 | 1463 | DIE(aTHX_ "Illegal modulus zero"); |
787eafbd | 1464 | |
65202027 | 1465 | dans = Perl_fmod(dleft, dright); |
787eafbd IZ |
1466 | if ((left_neg != right_neg) && dans) |
1467 | dans = dright - dans; | |
1468 | if (right_neg) | |
1469 | dans = -dans; | |
1470 | sv_setnv(TARG, dans); | |
1471 | } | |
1472 | else { | |
1473 | UV ans; | |
1474 | ||
787eafbd | 1475 | if (!right) |
cea2e8a9 | 1476 | DIE(aTHX_ "Illegal modulus zero"); |
787eafbd IZ |
1477 | |
1478 | ans = left % right; | |
1479 | if ((left_neg != right_neg) && ans) | |
1480 | ans = right - ans; | |
1481 | if (right_neg) { | |
1482 | /* XXX may warn: unary minus operator applied to unsigned type */ | |
1483 | /* could change -foo to be (~foo)+1 instead */ | |
1484 | if (ans <= ~((UV)IV_MAX)+1) | |
1485 | sv_setiv(TARG, ~ans+1); | |
1486 | else | |
65202027 | 1487 | sv_setnv(TARG, -(NV)ans); |
787eafbd IZ |
1488 | } |
1489 | else | |
1490 | sv_setuv(TARG, ans); | |
1491 | } | |
1492 | PUSHTARG; | |
1493 | RETURN; | |
79072805 | 1494 | } |
a0d0e21e | 1495 | } |
79072805 | 1496 | |
a0d0e21e LW |
1497 | PP(pp_repeat) |
1498 | { | |
97aff369 | 1499 | dVAR; dSP; dATARGET; tryAMAGICbin(repeat,opASSIGN); |
748a9306 | 1500 | { |
2b573ace JH |
1501 | register IV count; |
1502 | dPOPss; | |
5b295bef | 1503 | SvGETMAGIC(sv); |
2b573ace JH |
1504 | if (SvIOKp(sv)) { |
1505 | if (SvUOK(sv)) { | |
1b6737cc | 1506 | const UV uv = SvUV(sv); |
2b573ace JH |
1507 | if (uv > IV_MAX) |
1508 | count = IV_MAX; /* The best we can do? */ | |
1509 | else | |
1510 | count = uv; | |
1511 | } else { | |
0bd48802 | 1512 | const IV iv = SvIV(sv); |
2b573ace JH |
1513 | if (iv < 0) |
1514 | count = 0; | |
1515 | else | |
1516 | count = iv; | |
1517 | } | |
1518 | } | |
1519 | else if (SvNOKp(sv)) { | |
1b6737cc | 1520 | const NV nv = SvNV(sv); |
2b573ace JH |
1521 | if (nv < 0.0) |
1522 | count = 0; | |
1523 | else | |
1524 | count = (IV)nv; | |
1525 | } | |
1526 | else | |
4ea561bc | 1527 | count = SvIV(sv); |
533c011a | 1528 | if (GIMME == G_ARRAY && PL_op->op_private & OPpREPEAT_DOLIST) { |
a0d0e21e | 1529 | dMARK; |
0bd48802 AL |
1530 | static const char oom_list_extend[] = "Out of memory during list extend"; |
1531 | const I32 items = SP - MARK; | |
1532 | const I32 max = items * count; | |
79072805 | 1533 | |
2b573ace JH |
1534 | MEM_WRAP_CHECK_1(max, SV*, oom_list_extend); |
1535 | /* Did the max computation overflow? */ | |
27d5b266 | 1536 | if (items > 0 && max > 0 && (max < items || max < count)) |
2b573ace | 1537 | Perl_croak(aTHX_ oom_list_extend); |
a0d0e21e LW |
1538 | MEXTEND(MARK, max); |
1539 | if (count > 1) { | |
1540 | while (SP > MARK) { | |
976c8a39 JH |
1541 | #if 0 |
1542 | /* This code was intended to fix 20010809.028: | |
1543 | ||
1544 | $x = 'abcd'; | |
1545 | for (($x =~ /./g) x 2) { | |
1546 | print chop; # "abcdabcd" expected as output. | |
1547 | } | |
1548 | ||
1549 | * but that change (#11635) broke this code: | |
1550 | ||
1551 | $x = [("foo")x2]; # only one "foo" ended up in the anonlist. | |
1552 | ||
1553 | * I can't think of a better fix that doesn't introduce | |
1554 | * an efficiency hit by copying the SVs. The stack isn't | |
1555 | * refcounted, and mortalisation obviously doesn't | |
1556 | * Do The Right Thing when the stack has more than | |
1557 | * one pointer to the same mortal value. | |
1558 | * .robin. | |
1559 | */ | |
e30acc16 RH |
1560 | if (*SP) { |
1561 | *SP = sv_2mortal(newSVsv(*SP)); | |
1562 | SvREADONLY_on(*SP); | |
1563 | } | |
976c8a39 JH |
1564 | #else |
1565 | if (*SP) | |
1566 | SvTEMP_off((*SP)); | |
1567 | #endif | |
a0d0e21e | 1568 | SP--; |
79072805 | 1569 | } |
a0d0e21e LW |
1570 | MARK++; |
1571 | repeatcpy((char*)(MARK + items), (char*)MARK, | |
ad64d0ec | 1572 | items * sizeof(const SV *), count - 1); |
a0d0e21e | 1573 | SP += max; |
79072805 | 1574 | } |
a0d0e21e LW |
1575 | else if (count <= 0) |
1576 | SP -= items; | |
79072805 | 1577 | } |
a0d0e21e | 1578 | else { /* Note: mark already snarfed by pp_list */ |
0bd48802 | 1579 | SV * const tmpstr = POPs; |
a0d0e21e | 1580 | STRLEN len; |
9b877dbb | 1581 | bool isutf; |
2b573ace JH |
1582 | static const char oom_string_extend[] = |
1583 | "Out of memory during string extend"; | |
a0d0e21e | 1584 | |
a0d0e21e LW |
1585 | SvSetSV(TARG, tmpstr); |
1586 | SvPV_force(TARG, len); | |
9b877dbb | 1587 | isutf = DO_UTF8(TARG); |
8ebc5c01 | 1588 | if (count != 1) { |
1589 | if (count < 1) | |
1590 | SvCUR_set(TARG, 0); | |
1591 | else { | |
c445ea15 | 1592 | const STRLEN max = (UV)count * len; |
19a94d75 | 1593 | if (len > MEM_SIZE_MAX / count) |
2b573ace JH |
1594 | Perl_croak(aTHX_ oom_string_extend); |
1595 | MEM_WRAP_CHECK_1(max, char, oom_string_extend); | |
8569b9dc | 1596 | SvGROW(TARG, max + 1); |
a0d0e21e | 1597 | repeatcpy(SvPVX(TARG) + len, SvPVX(TARG), len, count - 1); |
b162af07 | 1598 | SvCUR_set(TARG, SvCUR(TARG) * count); |
7a4c00b4 | 1599 | } |
a0d0e21e | 1600 | *SvEND(TARG) = '\0'; |
a0d0e21e | 1601 | } |
dfcb284a GS |
1602 | if (isutf) |
1603 | (void)SvPOK_only_UTF8(TARG); | |
1604 | else | |
1605 | (void)SvPOK_only(TARG); | |
b80b6069 RH |
1606 | |
1607 | if (PL_op->op_private & OPpREPEAT_DOLIST) { | |
1608 | /* The parser saw this as a list repeat, and there | |
1609 | are probably several items on the stack. But we're | |
1610 | in scalar context, and there's no pp_list to save us | |
1611 | now. So drop the rest of the items -- robin@kitsite.com | |
1612 | */ | |
1613 | dMARK; | |
1614 | SP = MARK; | |
1615 | } | |
a0d0e21e | 1616 | PUSHTARG; |
79072805 | 1617 | } |
a0d0e21e | 1618 | RETURN; |
748a9306 | 1619 | } |
a0d0e21e | 1620 | } |
79072805 | 1621 | |
a0d0e21e LW |
1622 | PP(pp_subtract) |
1623 | { | |
800401ee JH |
1624 | dVAR; dSP; dATARGET; bool useleft; SV *svl, *svr; |
1625 | tryAMAGICbin(subtr,opASSIGN); | |
1626 | svl = sv_2num(TOPm1s); | |
1627 | svr = sv_2num(TOPs); | |
1628 | useleft = USE_LEFT(svl); | |
28e5dec8 | 1629 | #ifdef PERL_PRESERVE_IVUV |
7dca457a NC |
1630 | /* See comments in pp_add (in pp_hot.c) about Overflow, and how |
1631 | "bad things" happen if you rely on signed integers wrapping. */ | |
800401ee JH |
1632 | SvIV_please(svr); |
1633 | if (SvIOK(svr)) { | |
28e5dec8 JH |
1634 | /* Unless the left argument is integer in range we are going to have to |
1635 | use NV maths. Hence only attempt to coerce the right argument if | |
1636 | we know the left is integer. */ | |
9c5ffd7c JH |
1637 | register UV auv = 0; |
1638 | bool auvok = FALSE; | |
7dca457a NC |
1639 | bool a_valid = 0; |
1640 | ||
28e5dec8 | 1641 | if (!useleft) { |
7dca457a NC |
1642 | auv = 0; |
1643 | a_valid = auvok = 1; | |
1644 | /* left operand is undef, treat as zero. */ | |
28e5dec8 JH |
1645 | } else { |
1646 | /* Left operand is defined, so is it IV? */ | |
800401ee JH |
1647 | SvIV_please(svl); |
1648 | if (SvIOK(svl)) { | |
1649 | if ((auvok = SvUOK(svl))) | |
1650 | auv = SvUVX(svl); | |
7dca457a | 1651 | else { |
800401ee | 1652 | register const IV aiv = SvIVX(svl); |
7dca457a NC |
1653 | if (aiv >= 0) { |
1654 | auv = aiv; | |
1655 | auvok = 1; /* Now acting as a sign flag. */ | |
1656 | } else { /* 2s complement assumption for IV_MIN */ | |
1657 | auv = (UV)-aiv; | |
28e5dec8 | 1658 | } |
7dca457a NC |
1659 | } |
1660 | a_valid = 1; | |
1661 | } | |
1662 | } | |
1663 | if (a_valid) { | |
1664 | bool result_good = 0; | |
1665 | UV result; | |
1666 | register UV buv; | |
800401ee | 1667 | bool buvok = SvUOK(svr); |
9041c2e3 | 1668 | |
7dca457a | 1669 | if (buvok) |
800401ee | 1670 | buv = SvUVX(svr); |
7dca457a | 1671 | else { |
800401ee | 1672 | register const IV biv = SvIVX(svr); |
7dca457a NC |
1673 | if (biv >= 0) { |
1674 | buv = biv; | |
1675 | buvok = 1; | |
1676 | } else | |
1677 | buv = (UV)-biv; | |
1678 | } | |
1679 | /* ?uvok if value is >= 0. basically, flagged as UV if it's +ve, | |
602f51c4 | 1680 | else "IV" now, independent of how it came in. |
7dca457a NC |
1681 | if a, b represents positive, A, B negative, a maps to -A etc |
1682 | a - b => (a - b) | |
1683 | A - b => -(a + b) | |
1684 | a - B => (a + b) | |
1685 | A - B => -(a - b) | |
1686 | all UV maths. negate result if A negative. | |
1687 | subtract if signs same, add if signs differ. */ | |
1688 | ||
1689 | if (auvok ^ buvok) { | |
1690 | /* Signs differ. */ | |
1691 | result = auv + buv; | |
1692 | if (result >= auv) | |
1693 | result_good = 1; | |
1694 | } else { | |
1695 | /* Signs same */ | |
1696 | if (auv >= buv) { | |
1697 | result = auv - buv; | |
1698 | /* Must get smaller */ | |
1699 | if (result <= auv) | |
1700 | result_good = 1; | |
1701 | } else { | |
1702 | result = buv - auv; | |
1703 | if (result <= buv) { | |
1704 | /* result really should be -(auv-buv). as its negation | |
1705 | of true value, need to swap our result flag */ | |
1706 | auvok = !auvok; | |
1707 | result_good = 1; | |
28e5dec8 | 1708 | } |
28e5dec8 JH |
1709 | } |
1710 | } | |
7dca457a NC |
1711 | if (result_good) { |
1712 | SP--; | |
1713 | if (auvok) | |
1714 | SETu( result ); | |
1715 | else { | |
1716 | /* Negate result */ | |
1717 | if (result <= (UV)IV_MIN) | |
1718 | SETi( -(IV)result ); | |
1719 | else { | |
1720 | /* result valid, but out of range for IV. */ | |
1721 | SETn( -(NV)result ); | |
1722 | } | |
1723 | } | |
1724 | RETURN; | |
1725 | } /* Overflow, drop through to NVs. */ | |
28e5dec8 JH |
1726 | } |
1727 | } | |
1728 | #endif | |
a0d0e21e | 1729 | { |
4efa5a16 RD |
1730 | NV value = SvNV(svr); |
1731 | (void)POPs; | |
1732 | ||
28e5dec8 JH |
1733 | if (!useleft) { |
1734 | /* left operand is undef, treat as zero - value */ | |
1735 | SETn(-value); | |
1736 | RETURN; | |
1737 | } | |
4efa5a16 | 1738 | SETn( SvNV(svl) - value ); |
28e5dec8 | 1739 | RETURN; |
79072805 | 1740 | } |
a0d0e21e | 1741 | } |
79072805 | 1742 | |
a0d0e21e LW |
1743 | PP(pp_left_shift) |
1744 | { | |
97aff369 | 1745 | dVAR; dSP; dATARGET; tryAMAGICbin(lshift,opASSIGN); |
a0d0e21e | 1746 | { |
1b6737cc | 1747 | const IV shift = POPi; |
d0ba1bd2 | 1748 | if (PL_op->op_private & HINT_INTEGER) { |
c445ea15 | 1749 | const IV i = TOPi; |
972b05a9 | 1750 | SETi(i << shift); |
d0ba1bd2 JH |
1751 | } |
1752 | else { | |
c445ea15 | 1753 | const UV u = TOPu; |
972b05a9 | 1754 | SETu(u << shift); |
d0ba1bd2 | 1755 | } |
55497cff | 1756 | RETURN; |
79072805 | 1757 | } |
a0d0e21e | 1758 | } |
79072805 | 1759 | |
a0d0e21e LW |
1760 | PP(pp_right_shift) |
1761 | { | |
97aff369 | 1762 | dVAR; dSP; dATARGET; tryAMAGICbin(rshift,opASSIGN); |
a0d0e21e | 1763 | { |
1b6737cc | 1764 | const IV shift = POPi; |
d0ba1bd2 | 1765 | if (PL_op->op_private & HINT_INTEGER) { |
0bd48802 | 1766 | const IV i = TOPi; |
972b05a9 | 1767 | SETi(i >> shift); |
d0ba1bd2 JH |
1768 | } |
1769 | else { | |
0bd48802 | 1770 | const UV u = TOPu; |
972b05a9 | 1771 | SETu(u >> shift); |
d0ba1bd2 | 1772 | } |
a0d0e21e | 1773 | RETURN; |
93a17b20 | 1774 | } |
79072805 LW |
1775 | } |
1776 | ||
a0d0e21e | 1777 | PP(pp_lt) |
79072805 | 1778 | { |
97aff369 | 1779 | dVAR; dSP; tryAMAGICbinSET(lt,0); |
28e5dec8 JH |
1780 | #ifdef PERL_PRESERVE_IVUV |
1781 | SvIV_please(TOPs); | |
1782 | if (SvIOK(TOPs)) { | |
1783 | SvIV_please(TOPm1s); | |
1784 | if (SvIOK(TOPm1s)) { | |
1785 | bool auvok = SvUOK(TOPm1s); | |
1786 | bool buvok = SvUOK(TOPs); | |
a227d84d | 1787 | |
28e5dec8 | 1788 | if (!auvok && !buvok) { /* ## IV < IV ## */ |
1b6737cc AL |
1789 | const IV aiv = SvIVX(TOPm1s); |
1790 | const IV biv = SvIVX(TOPs); | |
28e5dec8 JH |
1791 | |
1792 | SP--; | |
1793 | SETs(boolSV(aiv < biv)); | |
1794 | RETURN; | |
1795 | } | |
1796 | if (auvok && buvok) { /* ## UV < UV ## */ | |
1b6737cc AL |
1797 | const UV auv = SvUVX(TOPm1s); |
1798 | const UV buv = SvUVX(TOPs); | |
28e5dec8 JH |
1799 | |
1800 | SP--; | |
1801 | SETs(boolSV(auv < buv)); | |
1802 | RETURN; | |
1803 | } | |
1804 | if (auvok) { /* ## UV < IV ## */ | |
1805 | UV auv; | |
1b6737cc | 1806 | const IV biv = SvIVX(TOPs); |
28e5dec8 JH |
1807 | SP--; |
1808 | if (biv < 0) { | |
1809 | /* As (a) is a UV, it's >=0, so it cannot be < */ | |
1810 | SETs(&PL_sv_no); | |
1811 | RETURN; | |
1812 | } | |
1813 | auv = SvUVX(TOPs); | |
28e5dec8 JH |
1814 | SETs(boolSV(auv < (UV)biv)); |
1815 | RETURN; | |
1816 | } | |
1817 | { /* ## IV < UV ## */ | |
1b6737cc | 1818 | const IV aiv = SvIVX(TOPm1s); |
28e5dec8 JH |
1819 | UV buv; |
1820 | ||
28e5dec8 JH |
1821 | if (aiv < 0) { |
1822 | /* As (b) is a UV, it's >=0, so it must be < */ | |
1823 | SP--; | |
1824 | SETs(&PL_sv_yes); | |
1825 | RETURN; | |
1826 | } | |
1827 | buv = SvUVX(TOPs); | |
1828 | SP--; | |
28e5dec8 JH |
1829 | SETs(boolSV((UV)aiv < buv)); |
1830 | RETURN; | |
1831 | } | |
1832 | } | |
1833 | } | |
1834 | #endif | |
30de85b6 | 1835 | #ifndef NV_PRESERVES_UV |
50fb3111 NC |
1836 | #ifdef PERL_PRESERVE_IVUV |
1837 | else | |
1838 | #endif | |
0bdaccee NC |
1839 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
1840 | SP--; | |
1841 | SETs(boolSV(SvRV(TOPs) < SvRV(TOPp1s))); | |
1842 | RETURN; | |
1843 | } | |
30de85b6 | 1844 | #endif |
a0d0e21e | 1845 | { |
cab190d4 JD |
1846 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
1847 | dPOPTOPnnrl; | |
1848 | if (Perl_isnan(left) || Perl_isnan(right)) | |
1849 | RETSETNO; | |
1850 | SETs(boolSV(left < right)); | |
1851 | #else | |
a0d0e21e | 1852 | dPOPnv; |
54310121 | 1853 | SETs(boolSV(TOPn < value)); |
cab190d4 | 1854 | #endif |
a0d0e21e | 1855 | RETURN; |
79072805 | 1856 | } |
a0d0e21e | 1857 | } |
79072805 | 1858 | |
a0d0e21e LW |
1859 | PP(pp_gt) |
1860 | { | |
97aff369 | 1861 | dVAR; dSP; tryAMAGICbinSET(gt,0); |
28e5dec8 JH |
1862 | #ifdef PERL_PRESERVE_IVUV |
1863 | SvIV_please(TOPs); | |
1864 | if (SvIOK(TOPs)) { | |
1865 | SvIV_please(TOPm1s); | |
1866 | if (SvIOK(TOPm1s)) { | |
1867 | bool auvok = SvUOK(TOPm1s); | |
1868 | bool buvok = SvUOK(TOPs); | |
a227d84d | 1869 | |
28e5dec8 | 1870 | if (!auvok && !buvok) { /* ## IV > IV ## */ |
1b6737cc AL |
1871 | const IV aiv = SvIVX(TOPm1s); |
1872 | const IV biv = SvIVX(TOPs); | |
1873 | ||
28e5dec8 JH |
1874 | SP--; |
1875 | SETs(boolSV(aiv > biv)); | |
1876 | RETURN; | |
1877 | } | |
1878 | if (auvok && buvok) { /* ## UV > UV ## */ | |
1b6737cc AL |
1879 | const UV auv = SvUVX(TOPm1s); |
1880 | const UV buv = SvUVX(TOPs); | |
28e5dec8 JH |
1881 | |
1882 | SP--; | |
1883 | SETs(boolSV(auv > buv)); | |
1884 | RETURN; | |
1885 | } | |
1886 | if (auvok) { /* ## UV > IV ## */ | |
1887 | UV auv; | |
1b6737cc AL |
1888 | const IV biv = SvIVX(TOPs); |
1889 | ||
28e5dec8 JH |
1890 | SP--; |
1891 | if (biv < 0) { | |
1892 | /* As (a) is a UV, it's >=0, so it must be > */ | |
1893 | SETs(&PL_sv_yes); | |
1894 | RETURN; | |
1895 | } | |
1896 | auv = SvUVX(TOPs); | |
28e5dec8 JH |
1897 | SETs(boolSV(auv > (UV)biv)); |
1898 | RETURN; | |
1899 | } | |
1900 | { /* ## IV > UV ## */ | |
1b6737cc | 1901 | const IV aiv = SvIVX(TOPm1s); |
28e5dec8 JH |
1902 | UV buv; |
1903 | ||
28e5dec8 JH |
1904 | if (aiv < 0) { |
1905 | /* As (b) is a UV, it's >=0, so it cannot be > */ | |
1906 | SP--; | |
1907 | SETs(&PL_sv_no); | |
1908 | RETURN; | |
1909 | } | |
1910 | buv = SvUVX(TOPs); | |
1911 | SP--; | |
28e5dec8 JH |
1912 | SETs(boolSV((UV)aiv > buv)); |
1913 | RETURN; | |
1914 | } | |
1915 | } | |
1916 | } | |
1917 | #endif | |
30de85b6 | 1918 | #ifndef NV_PRESERVES_UV |
50fb3111 NC |
1919 | #ifdef PERL_PRESERVE_IVUV |
1920 | else | |
1921 | #endif | |
0bdaccee | 1922 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
30de85b6 NC |
1923 | SP--; |
1924 | SETs(boolSV(SvRV(TOPs) > SvRV(TOPp1s))); | |
1925 | RETURN; | |
1926 | } | |
1927 | #endif | |
a0d0e21e | 1928 | { |
cab190d4 JD |
1929 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
1930 | dPOPTOPnnrl; | |
1931 | if (Perl_isnan(left) || Perl_isnan(right)) | |
1932 | RETSETNO; | |
1933 | SETs(boolSV(left > right)); | |
1934 | #else | |
a0d0e21e | 1935 | dPOPnv; |
54310121 | 1936 | SETs(boolSV(TOPn > value)); |
cab190d4 | 1937 | #endif |
a0d0e21e | 1938 | RETURN; |
79072805 | 1939 | } |
a0d0e21e LW |
1940 | } |
1941 | ||
1942 | PP(pp_le) | |
1943 | { | |
97aff369 | 1944 | dVAR; dSP; tryAMAGICbinSET(le,0); |
28e5dec8 JH |
1945 | #ifdef PERL_PRESERVE_IVUV |
1946 | SvIV_please(TOPs); | |
1947 | if (SvIOK(TOPs)) { | |
1948 | SvIV_please(TOPm1s); | |
1949 | if (SvIOK(TOPm1s)) { | |
1950 | bool auvok = SvUOK(TOPm1s); | |
1951 | bool buvok = SvUOK(TOPs); | |
a227d84d | 1952 | |
28e5dec8 | 1953 | if (!auvok && !buvok) { /* ## IV <= IV ## */ |
1b6737cc AL |
1954 | const IV aiv = SvIVX(TOPm1s); |
1955 | const IV biv = SvIVX(TOPs); | |
28e5dec8 JH |
1956 | |
1957 | SP--; | |
1958 | SETs(boolSV(aiv <= biv)); | |
1959 | RETURN; | |
1960 | } | |
1961 | if (auvok && buvok) { /* ## UV <= UV ## */ | |
1962 | UV auv = SvUVX(TOPm1s); | |
1963 | UV buv = SvUVX(TOPs); | |
1964 | ||
1965 | SP--; | |
1966 | SETs(boolSV(auv <= buv)); | |
1967 | RETURN; | |
1968 | } | |
1969 | if (auvok) { /* ## UV <= IV ## */ | |
1970 | UV auv; | |
1b6737cc AL |
1971 | const IV biv = SvIVX(TOPs); |
1972 | ||
28e5dec8 JH |
1973 | SP--; |
1974 | if (biv < 0) { | |
1975 | /* As (a) is a UV, it's >=0, so a cannot be <= */ | |
1976 | SETs(&PL_sv_no); | |
1977 | RETURN; | |
1978 | } | |
1979 | auv = SvUVX(TOPs); | |
28e5dec8 JH |
1980 | SETs(boolSV(auv <= (UV)biv)); |
1981 | RETURN; | |
1982 | } | |
1983 | { /* ## IV <= UV ## */ | |
1b6737cc | 1984 | const IV aiv = SvIVX(TOPm1s); |
28e5dec8 | 1985 | UV buv; |
1b6737cc | 1986 | |
28e5dec8 JH |
1987 | if (aiv < 0) { |
1988 | /* As (b) is a UV, it's >=0, so a must be <= */ | |
1989 | SP--; | |
1990 | SETs(&PL_sv_yes); | |
1991 | RETURN; | |
1992 | } | |
1993 | buv = SvUVX(TOPs); | |
1994 | SP--; | |
28e5dec8 JH |
1995 | SETs(boolSV((UV)aiv <= buv)); |
1996 | RETURN; | |
1997 | } | |
1998 | } | |
1999 | } | |
2000 | #endif | |
30de85b6 | 2001 | #ifndef NV_PRESERVES_UV |
50fb3111 NC |
2002 | #ifdef PERL_PRESERVE_IVUV |
2003 | else | |
2004 | #endif | |
0bdaccee | 2005 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
30de85b6 NC |
2006 | SP--; |
2007 | SETs(boolSV(SvRV(TOPs) <= SvRV(TOPp1s))); | |
2008 | RETURN; | |
2009 | } | |
2010 | #endif | |
a0d0e21e | 2011 | { |
cab190d4 JD |
2012 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
2013 | dPOPTOPnnrl; | |
2014 | if (Perl_isnan(left) || Perl_isnan(right)) | |
2015 | RETSETNO; | |
2016 | SETs(boolSV(left <= right)); | |
2017 | #else | |
a0d0e21e | 2018 | dPOPnv; |
54310121 | 2019 | SETs(boolSV(TOPn <= value)); |
cab190d4 | 2020 | #endif |
a0d0e21e | 2021 | RETURN; |
79072805 | 2022 | } |
a0d0e21e LW |
2023 | } |
2024 | ||
2025 | PP(pp_ge) | |
2026 | { | |
97aff369 | 2027 | dVAR; dSP; tryAMAGICbinSET(ge,0); |
28e5dec8 JH |
2028 | #ifdef PERL_PRESERVE_IVUV |
2029 | SvIV_please(TOPs); | |
2030 | if (SvIOK(TOPs)) { | |
2031 | SvIV_please(TOPm1s); | |
2032 | if (SvIOK(TOPm1s)) { | |
2033 | bool auvok = SvUOK(TOPm1s); | |
2034 | bool buvok = SvUOK(TOPs); | |
a227d84d | 2035 | |
28e5dec8 | 2036 | if (!auvok && !buvok) { /* ## IV >= IV ## */ |
1b6737cc AL |
2037 | const IV aiv = SvIVX(TOPm1s); |
2038 | const IV biv = SvIVX(TOPs); | |
2039 | ||
28e5dec8 JH |
2040 | SP--; |
2041 | SETs(boolSV(aiv >= biv)); | |
2042 | RETURN; | |
2043 | } | |
2044 | if (auvok && buvok) { /* ## UV >= UV ## */ | |
1b6737cc AL |
2045 | const UV auv = SvUVX(TOPm1s); |
2046 | const UV buv = SvUVX(TOPs); | |
2047 | ||
28e5dec8 JH |
2048 | SP--; |
2049 | SETs(boolSV(auv >= buv)); | |
2050 | RETURN; | |
2051 | } | |
2052 | if (auvok) { /* ## UV >= IV ## */ | |
2053 | UV auv; | |
1b6737cc AL |
2054 | const IV biv = SvIVX(TOPs); |
2055 | ||
28e5dec8 JH |
2056 | SP--; |
2057 | if (biv < 0) { | |
2058 | /* As (a) is a UV, it's >=0, so it must be >= */ | |
2059 | SETs(&PL_sv_yes); | |
2060 | RETURN; | |
2061 | } | |
2062 | auv = SvUVX(TOPs); | |
28e5dec8 JH |
2063 | SETs(boolSV(auv >= (UV)biv)); |
2064 | RETURN; | |
2065 | } | |
2066 | { /* ## IV >= UV ## */ | |
1b6737cc | 2067 | const IV aiv = SvIVX(TOPm1s); |
28e5dec8 | 2068 | UV buv; |
1b6737cc | 2069 | |
28e5dec8 JH |
2070 | if (aiv < 0) { |
2071 | /* As (b) is a UV, it's >=0, so a cannot be >= */ | |
2072 | SP--; | |
2073 | SETs(&PL_sv_no); | |
2074 | RETURN; | |
2075 | } | |
2076 | buv = SvUVX(TOPs); | |
2077 | SP--; | |
28e5dec8 JH |
2078 | SETs(boolSV((UV)aiv >= buv)); |
2079 | RETURN; | |
2080 | } | |
2081 | } | |
2082 | } | |
2083 | #endif | |
30de85b6 | 2084 | #ifndef NV_PRESERVES_UV |
50fb3111 NC |
2085 | #ifdef PERL_PRESERVE_IVUV |
2086 | else | |
2087 | #endif | |
0bdaccee | 2088 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
30de85b6 NC |
2089 | SP--; |
2090 | SETs(boolSV(SvRV(TOPs) >= SvRV(TOPp1s))); | |
2091 | RETURN; | |
2092 | } | |
2093 | #endif | |
a0d0e21e | 2094 | { |
cab190d4 JD |
2095 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
2096 | dPOPTOPnnrl; | |
2097 | if (Perl_isnan(left) || Perl_isnan(right)) | |
2098 | RETSETNO; | |
2099 | SETs(boolSV(left >= right)); | |
2100 | #else | |
a0d0e21e | 2101 | dPOPnv; |
54310121 | 2102 | SETs(boolSV(TOPn >= value)); |
cab190d4 | 2103 | #endif |
a0d0e21e | 2104 | RETURN; |
79072805 | 2105 | } |
a0d0e21e | 2106 | } |
79072805 | 2107 | |
a0d0e21e LW |
2108 | PP(pp_ne) |
2109 | { | |
97aff369 | 2110 | dVAR; dSP; tryAMAGICbinSET(ne,0); |
3bb2c415 | 2111 | #ifndef NV_PRESERVES_UV |
0bdaccee | 2112 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
e61d22ef NC |
2113 | SP--; |
2114 | SETs(boolSV(SvRV(TOPs) != SvRV(TOPp1s))); | |
3bb2c415 JH |
2115 | RETURN; |
2116 | } | |
2117 | #endif | |
28e5dec8 JH |
2118 | #ifdef PERL_PRESERVE_IVUV |
2119 | SvIV_please(TOPs); | |
2120 | if (SvIOK(TOPs)) { | |
2121 | SvIV_please(TOPm1s); | |
2122 | if (SvIOK(TOPm1s)) { | |
0bd48802 AL |
2123 | const bool auvok = SvUOK(TOPm1s); |
2124 | const bool buvok = SvUOK(TOPs); | |
a227d84d | 2125 | |
30de85b6 NC |
2126 | if (auvok == buvok) { /* ## IV == IV or UV == UV ## */ |
2127 | /* Casting IV to UV before comparison isn't going to matter | |
2128 | on 2s complement. On 1s complement or sign&magnitude | |
2129 | (if we have any of them) it could make negative zero | |
2130 | differ from normal zero. As I understand it. (Need to | |
2131 | check - is negative zero implementation defined behaviour | |
2132 | anyway?). NWC */ | |
1b6737cc AL |
2133 | const UV buv = SvUVX(POPs); |
2134 | const UV auv = SvUVX(TOPs); | |
2135 | ||
28e5dec8 JH |
2136 | SETs(boolSV(auv != buv)); |
2137 | RETURN; | |
2138 | } | |
2139 | { /* ## Mixed IV,UV ## */ | |
2140 | IV iv; | |
2141 | UV uv; | |
2142 | ||
2143 | /* != is commutative so swap if needed (save code) */ | |
2144 | if (auvok) { | |
2145 | /* swap. top of stack (b) is the iv */ | |
2146 | iv = SvIVX(TOPs); | |
2147 | SP--; | |
2148 | if (iv < 0) { | |
2149 | /* As (a) is a UV, it's >0, so it cannot be == */ | |
2150 | SETs(&PL_sv_yes); | |
2151 | RETURN; | |
2152 | } | |
2153 | uv = SvUVX(TOPs); | |
2154 | } else { | |
2155 | iv = SvIVX(TOPm1s); | |
2156 | SP--; | |
2157 | if (iv < 0) { | |
2158 | /* As (b) is a UV, it's >0, so it cannot be == */ | |
2159 | SETs(&PL_sv_yes); | |
2160 | RETURN; | |
2161 | } | |
2162 | uv = SvUVX(*(SP+1)); /* Do I want TOPp1s() ? */ | |
2163 | } | |
28e5dec8 JH |
2164 | SETs(boolSV((UV)iv != uv)); |
2165 | RETURN; | |
2166 | } | |
2167 | } | |
2168 | } | |
2169 | #endif | |
a0d0e21e | 2170 | { |
cab190d4 JD |
2171 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
2172 | dPOPTOPnnrl; | |
2173 | if (Perl_isnan(left) || Perl_isnan(right)) | |
2174 | RETSETYES; | |
2175 | SETs(boolSV(left != right)); | |
2176 | #else | |
a0d0e21e | 2177 | dPOPnv; |
54310121 | 2178 | SETs(boolSV(TOPn != value)); |
cab190d4 | 2179 | #endif |
a0d0e21e LW |
2180 | RETURN; |
2181 | } | |
79072805 LW |
2182 | } |
2183 | ||
a0d0e21e | 2184 | PP(pp_ncmp) |
79072805 | 2185 | { |
97aff369 | 2186 | dVAR; dSP; dTARGET; tryAMAGICbin(ncmp,0); |
d8c7644e | 2187 | #ifndef NV_PRESERVES_UV |
0bdaccee | 2188 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
0bd48802 AL |
2189 | const UV right = PTR2UV(SvRV(POPs)); |
2190 | const UV left = PTR2UV(SvRV(TOPs)); | |
e61d22ef | 2191 | SETi((left > right) - (left < right)); |
d8c7644e JH |
2192 | RETURN; |
2193 | } | |
2194 | #endif | |
28e5dec8 JH |
2195 | #ifdef PERL_PRESERVE_IVUV |
2196 | /* Fortunately it seems NaN isn't IOK */ | |
2197 | SvIV_please(TOPs); | |
2198 | if (SvIOK(TOPs)) { | |
2199 | SvIV_please(TOPm1s); | |
2200 | if (SvIOK(TOPm1s)) { | |
1b6737cc AL |
2201 | const bool leftuvok = SvUOK(TOPm1s); |
2202 | const bool rightuvok = SvUOK(TOPs); | |
28e5dec8 JH |
2203 | I32 value; |
2204 | if (!leftuvok && !rightuvok) { /* ## IV <=> IV ## */ | |
1b6737cc AL |
2205 | const IV leftiv = SvIVX(TOPm1s); |
2206 | const IV rightiv = SvIVX(TOPs); | |
28e5dec8 JH |
2207 | |
2208 | if (leftiv > rightiv) | |
2209 | value = 1; | |
2210 | else if (leftiv < rightiv) | |
2211 | value = -1; | |
2212 | else | |
2213 | value = 0; | |
2214 | } else if (leftuvok && rightuvok) { /* ## UV <=> UV ## */ | |
1b6737cc AL |
2215 | const UV leftuv = SvUVX(TOPm1s); |
2216 | const UV rightuv = SvUVX(TOPs); | |
28e5dec8 JH |
2217 | |
2218 | if (leftuv > rightuv) | |
2219 | value = 1; | |
2220 | else if (leftuv < rightuv) | |
2221 | value = -1; | |
2222 | else | |
2223 | value = 0; | |
2224 | } else if (leftuvok) { /* ## UV <=> IV ## */ | |
1b6737cc | 2225 | const IV rightiv = SvIVX(TOPs); |
28e5dec8 JH |
2226 | if (rightiv < 0) { |
2227 | /* As (a) is a UV, it's >=0, so it cannot be < */ | |
2228 | value = 1; | |
2229 | } else { | |
1b6737cc | 2230 | const UV leftuv = SvUVX(TOPm1s); |
83bac5dd | 2231 | if (leftuv > (UV)rightiv) { |
28e5dec8 JH |
2232 | value = 1; |
2233 | } else if (leftuv < (UV)rightiv) { | |
2234 | value = -1; | |
2235 | } else { | |
2236 | value = 0; | |
2237 | } | |
2238 | } | |
2239 | } else { /* ## IV <=> UV ## */ | |
1b6737cc | 2240 | const IV leftiv = SvIVX(TOPm1s); |
28e5dec8 JH |
2241 | if (leftiv < 0) { |
2242 | /* As (b) is a UV, it's >=0, so it must be < */ | |
2243 | value = -1; | |
2244 | } else { | |
1b6737cc | 2245 | const UV rightuv = SvUVX(TOPs); |
83bac5dd | 2246 | if ((UV)leftiv > rightuv) { |
28e5dec8 | 2247 | value = 1; |
83bac5dd | 2248 | } else if ((UV)leftiv < rightuv) { |
28e5dec8 JH |
2249 | value = -1; |
2250 | } else { | |
2251 | value = 0; | |
2252 | } | |
2253 | } | |
2254 | } | |
2255 | SP--; | |
2256 | SETi(value); | |
2257 | RETURN; | |
2258 | } | |
2259 | } | |
2260 | #endif | |
a0d0e21e LW |
2261 | { |
2262 | dPOPTOPnnrl; | |
2263 | I32 value; | |
79072805 | 2264 | |
a3540c92 | 2265 | #ifdef Perl_isnan |
1ad04cfd JH |
2266 | if (Perl_isnan(left) || Perl_isnan(right)) { |
2267 | SETs(&PL_sv_undef); | |
2268 | RETURN; | |
2269 | } | |
2270 | value = (left > right) - (left < right); | |
2271 | #else | |
ff0cee69 | 2272 | if (left == right) |
a0d0e21e | 2273 | value = 0; |
a0d0e21e LW |
2274 | else if (left < right) |
2275 | value = -1; | |
44a8e56a | 2276 | else if (left > right) |
2277 | value = 1; | |
2278 | else { | |
3280af22 | 2279 | SETs(&PL_sv_undef); |
44a8e56a | 2280 | RETURN; |
2281 | } | |
1ad04cfd | 2282 | #endif |
a0d0e21e LW |
2283 | SETi(value); |
2284 | RETURN; | |
79072805 | 2285 | } |
a0d0e21e | 2286 | } |
79072805 | 2287 | |
afd9910b | 2288 | PP(pp_sle) |
a0d0e21e | 2289 | { |
97aff369 | 2290 | dVAR; dSP; |
79072805 | 2291 | |
afd9910b NC |
2292 | int amg_type = sle_amg; |
2293 | int multiplier = 1; | |
2294 | int rhs = 1; | |
79072805 | 2295 | |
afd9910b NC |
2296 | switch (PL_op->op_type) { |
2297 | case OP_SLT: | |
2298 | amg_type = slt_amg; | |
2299 | /* cmp < 0 */ | |
2300 | rhs = 0; | |
2301 | break; | |
2302 | case OP_SGT: | |
2303 | amg_type = sgt_amg; | |
2304 | /* cmp > 0 */ | |
2305 | multiplier = -1; | |
2306 | rhs = 0; | |
2307 | break; | |
2308 | case OP_SGE: | |
2309 | amg_type = sge_amg; | |
2310 | /* cmp >= 0 */ | |
2311 | multiplier = -1; | |
2312 | break; | |
79072805 | 2313 | } |
79072805 | 2314 | |
afd9910b | 2315 | tryAMAGICbinSET_var(amg_type,0); |
a0d0e21e LW |
2316 | { |
2317 | dPOPTOPssrl; | |
1b6737cc | 2318 | const int cmp = (IN_LOCALE_RUNTIME |
bbce6d69 | 2319 | ? sv_cmp_locale(left, right) |
2320 | : sv_cmp(left, right)); | |
afd9910b | 2321 | SETs(boolSV(cmp * multiplier < rhs)); |
a0d0e21e LW |
2322 | RETURN; |
2323 | } | |
2324 | } | |
79072805 | 2325 | |
36477c24 | 2326 | PP(pp_seq) |
2327 | { | |
97aff369 | 2328 | dVAR; dSP; tryAMAGICbinSET(seq,0); |
36477c24 | 2329 | { |
2330 | dPOPTOPssrl; | |
54310121 | 2331 | SETs(boolSV(sv_eq(left, right))); |
a0d0e21e LW |
2332 | RETURN; |
2333 | } | |
2334 | } | |
79072805 | 2335 | |
a0d0e21e | 2336 | PP(pp_sne) |
79072805 | 2337 | { |
97aff369 | 2338 | dVAR; dSP; tryAMAGICbinSET(sne,0); |
a0d0e21e LW |
2339 | { |
2340 | dPOPTOPssrl; | |
54310121 | 2341 | SETs(boolSV(!sv_eq(left, right))); |
a0d0e21e | 2342 | RETURN; |
463ee0b2 | 2343 | } |
79072805 LW |
2344 | } |
2345 | ||
a0d0e21e | 2346 | PP(pp_scmp) |
79072805 | 2347 | { |
97aff369 | 2348 | dVAR; dSP; dTARGET; tryAMAGICbin(scmp,0); |
a0d0e21e LW |
2349 | { |
2350 | dPOPTOPssrl; | |
1b6737cc | 2351 | const int cmp = (IN_LOCALE_RUNTIME |
bbce6d69 | 2352 | ? sv_cmp_locale(left, right) |
2353 | : sv_cmp(left, right)); | |
2354 | SETi( cmp ); | |
a0d0e21e LW |
2355 | RETURN; |
2356 | } | |
2357 | } | |
79072805 | 2358 | |
55497cff | 2359 | PP(pp_bit_and) |
2360 | { | |
97aff369 | 2361 | dVAR; dSP; dATARGET; tryAMAGICbin(band,opASSIGN); |
a0d0e21e LW |
2362 | { |
2363 | dPOPTOPssrl; | |
5b295bef RD |
2364 | SvGETMAGIC(left); |
2365 | SvGETMAGIC(right); | |
4633a7c4 | 2366 | if (SvNIOKp(left) || SvNIOKp(right)) { |
d0ba1bd2 | 2367 | if (PL_op->op_private & HINT_INTEGER) { |
1b6737cc | 2368 | const IV i = SvIV_nomg(left) & SvIV_nomg(right); |
972b05a9 | 2369 | SETi(i); |
d0ba1bd2 JH |
2370 | } |
2371 | else { | |
1b6737cc | 2372 | const UV u = SvUV_nomg(left) & SvUV_nomg(right); |
972b05a9 | 2373 | SETu(u); |
d0ba1bd2 | 2374 | } |
a0d0e21e LW |
2375 | } |
2376 | else { | |
533c011a | 2377 | do_vop(PL_op->op_type, TARG, left, right); |
a0d0e21e LW |
2378 | SETTARG; |
2379 | } | |
2380 | RETURN; | |
2381 | } | |
2382 | } | |
79072805 | 2383 | |
a0d0e21e LW |
2384 | PP(pp_bit_or) |
2385 | { | |
3658c1f1 NC |
2386 | dVAR; dSP; dATARGET; |
2387 | const int op_type = PL_op->op_type; | |
2388 | ||
2389 | tryAMAGICbin_var((op_type == OP_BIT_OR ? bor_amg : bxor_amg), opASSIGN); | |
a0d0e21e LW |
2390 | { |
2391 | dPOPTOPssrl; | |
5b295bef RD |
2392 | SvGETMAGIC(left); |
2393 | SvGETMAGIC(right); | |
4633a7c4 | 2394 | if (SvNIOKp(left) || SvNIOKp(right)) { |
d0ba1bd2 | 2395 | if (PL_op->op_private & HINT_INTEGER) { |
3658c1f1 NC |
2396 | const IV l = (USE_LEFT(left) ? SvIV_nomg(left) : 0); |
2397 | const IV r = SvIV_nomg(right); | |
2398 | const IV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
2399 | SETi(result); | |
d0ba1bd2 JH |
2400 | } |
2401 | else { | |
3658c1f1 NC |
2402 | const UV l = (USE_LEFT(left) ? SvUV_nomg(left) : 0); |
2403 | const UV r = SvUV_nomg(right); | |
2404 | const UV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
2405 | SETu(result); | |
d0ba1bd2 | 2406 | } |
a0d0e21e LW |
2407 | } |
2408 | else { | |
3658c1f1 | 2409 | do_vop(op_type, TARG, left, right); |
a0d0e21e LW |
2410 | SETTARG; |
2411 | } | |
2412 | RETURN; | |
79072805 | 2413 | } |
a0d0e21e | 2414 | } |
79072805 | 2415 | |
a0d0e21e LW |
2416 | PP(pp_negate) |
2417 | { | |
97aff369 | 2418 | dVAR; dSP; dTARGET; tryAMAGICun(neg); |
a0d0e21e | 2419 | { |
800401ee | 2420 | SV * const sv = sv_2num(TOPs); |
1b6737cc | 2421 | const int flags = SvFLAGS(sv); |
5b295bef | 2422 | SvGETMAGIC(sv); |
28e5dec8 JH |
2423 | if ((flags & SVf_IOK) || ((flags & (SVp_IOK | SVp_NOK)) == SVp_IOK)) { |
2424 | /* It's publicly an integer, or privately an integer-not-float */ | |
2425 | oops_its_an_int: | |
9b0e499b GS |
2426 | if (SvIsUV(sv)) { |
2427 | if (SvIVX(sv) == IV_MIN) { | |
28e5dec8 | 2428 | /* 2s complement assumption. */ |
9b0e499b GS |
2429 | SETi(SvIVX(sv)); /* special case: -((UV)IV_MAX+1) == IV_MIN */ |
2430 | RETURN; | |
2431 | } | |
2432 | else if (SvUVX(sv) <= IV_MAX) { | |
beccb14c | 2433 | SETi(-SvIVX(sv)); |
9b0e499b GS |
2434 | RETURN; |
2435 | } | |
2436 | } | |
2437 | else if (SvIVX(sv) != IV_MIN) { | |
2438 | SETi(-SvIVX(sv)); | |
2439 | RETURN; | |
2440 | } | |
28e5dec8 JH |
2441 | #ifdef PERL_PRESERVE_IVUV |
2442 | else { | |
2443 | SETu((UV)IV_MIN); | |
2444 | RETURN; | |
2445 | } | |
2446 | #endif | |
9b0e499b GS |
2447 | } |
2448 | if (SvNIOKp(sv)) | |
a0d0e21e | 2449 | SETn(-SvNV(sv)); |
4633a7c4 | 2450 | else if (SvPOKp(sv)) { |
a0d0e21e | 2451 | STRLEN len; |
c445ea15 | 2452 | const char * const s = SvPV_const(sv, len); |
bbce6d69 | 2453 | if (isIDFIRST(*s)) { |
76f68e9b | 2454 | sv_setpvs(TARG, "-"); |
a0d0e21e | 2455 | sv_catsv(TARG, sv); |
79072805 | 2456 | } |
a0d0e21e LW |
2457 | else if (*s == '+' || *s == '-') { |
2458 | sv_setsv(TARG, sv); | |
2459 | *SvPV_force(TARG, len) = *s == '-' ? '+' : '-'; | |
79072805 | 2460 | } |
8eb28a70 JH |
2461 | else if (DO_UTF8(sv)) { |
2462 | SvIV_please(sv); | |
2463 | if (SvIOK(sv)) | |
2464 | goto oops_its_an_int; | |
2465 | if (SvNOK(sv)) | |
2466 | sv_setnv(TARG, -SvNV(sv)); | |
2467 | else { | |
76f68e9b | 2468 | sv_setpvs(TARG, "-"); |
8eb28a70 JH |
2469 | sv_catsv(TARG, sv); |
2470 | } | |
834a4ddd | 2471 | } |
28e5dec8 | 2472 | else { |
8eb28a70 JH |
2473 | SvIV_please(sv); |
2474 | if (SvIOK(sv)) | |
2475 | goto oops_its_an_int; | |
2476 | sv_setnv(TARG, -SvNV(sv)); | |
28e5dec8 | 2477 | } |
a0d0e21e | 2478 | SETTARG; |
79072805 | 2479 | } |
4633a7c4 LW |
2480 | else |
2481 | SETn(-SvNV(sv)); | |
79072805 | 2482 | } |
a0d0e21e | 2483 | RETURN; |
79072805 LW |
2484 | } |
2485 | ||
a0d0e21e | 2486 | PP(pp_not) |
79072805 | 2487 | { |
97aff369 | 2488 | dVAR; dSP; tryAMAGICunSET(not); |
3280af22 | 2489 | *PL_stack_sp = boolSV(!SvTRUE(*PL_stack_sp)); |
a0d0e21e | 2490 | return NORMAL; |
79072805 LW |
2491 | } |
2492 | ||
a0d0e21e | 2493 | PP(pp_complement) |
79072805 | 2494 | { |
97aff369 | 2495 | dVAR; dSP; dTARGET; tryAMAGICun(compl); |
a0d0e21e LW |
2496 | { |
2497 | dTOPss; | |
5b295bef | 2498 | SvGETMAGIC(sv); |
4633a7c4 | 2499 | if (SvNIOKp(sv)) { |
d0ba1bd2 | 2500 | if (PL_op->op_private & HINT_INTEGER) { |
1b6737cc | 2501 | const IV i = ~SvIV_nomg(sv); |
972b05a9 | 2502 | SETi(i); |
d0ba1bd2 JH |
2503 | } |
2504 | else { | |
1b6737cc | 2505 | const UV u = ~SvUV_nomg(sv); |
972b05a9 | 2506 | SETu(u); |
d0ba1bd2 | 2507 | } |
a0d0e21e LW |
2508 | } |
2509 | else { | |
51723571 | 2510 | register U8 *tmps; |
55497cff | 2511 | register I32 anum; |
a0d0e21e LW |
2512 | STRLEN len; |
2513 | ||
10516c54 | 2514 | (void)SvPV_nomg_const(sv,len); /* force check for uninit var */ |
891f9566 | 2515 | sv_setsv_nomg(TARG, sv); |
51723571 | 2516 | tmps = (U8*)SvPV_force(TARG, len); |
a0d0e21e | 2517 | anum = len; |
1d68d6cd | 2518 | if (SvUTF8(TARG)) { |
a1ca4561 | 2519 | /* Calculate exact length, let's not estimate. */ |
1d68d6cd | 2520 | STRLEN targlen = 0; |
ba210ebe | 2521 | STRLEN l; |
a1ca4561 YST |
2522 | UV nchar = 0; |
2523 | UV nwide = 0; | |
01f6e806 | 2524 | U8 * const send = tmps + len; |
74d49cd0 TS |
2525 | U8 * const origtmps = tmps; |
2526 | const UV utf8flags = UTF8_ALLOW_ANYUV; | |
1d68d6cd | 2527 | |
1d68d6cd | 2528 | while (tmps < send) { |
74d49cd0 TS |
2529 | const UV c = utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
2530 | tmps += l; | |
5bbb0b5a | 2531 | targlen += UNISKIP(~c); |
a1ca4561 YST |
2532 | nchar++; |
2533 | if (c > 0xff) | |
2534 | nwide++; | |
1d68d6cd SC |
2535 | } |
2536 | ||
2537 | /* Now rewind strings and write them. */ | |
74d49cd0 | 2538 | tmps = origtmps; |
a1ca4561 YST |
2539 | |
2540 | if (nwide) { | |
01f6e806 AL |
2541 | U8 *result; |
2542 | U8 *p; | |
2543 | ||
74d49cd0 | 2544 | Newx(result, targlen + 1, U8); |
01f6e806 | 2545 | p = result; |
a1ca4561 | 2546 | while (tmps < send) { |
74d49cd0 TS |
2547 | const UV c = utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
2548 | tmps += l; | |
01f6e806 | 2549 | p = uvchr_to_utf8_flags(p, ~c, UNICODE_ALLOW_ANY); |
a1ca4561 | 2550 | } |
01f6e806 | 2551 | *p = '\0'; |
c1c21316 NC |
2552 | sv_usepvn_flags(TARG, (char*)result, targlen, |
2553 | SV_HAS_TRAILING_NUL); | |
a1ca4561 YST |
2554 | SvUTF8_on(TARG); |
2555 | } | |
2556 | else { | |
01f6e806 AL |
2557 | U8 *result; |
2558 | U8 *p; | |
2559 | ||
74d49cd0 | 2560 | Newx(result, nchar + 1, U8); |
01f6e806 | 2561 | p = result; |
a1ca4561 | 2562 | while (tmps < send) { |
74d49cd0 TS |
2563 | const U8 c = (U8)utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
2564 | tmps += l; | |
01f6e806 | 2565 | *p++ = ~c; |
a1ca4561 | 2566 | } |
01f6e806 | 2567 | *p = '\0'; |
c1c21316 | 2568 | sv_usepvn_flags(TARG, (char*)result, nchar, SV_HAS_TRAILING_NUL); |
d0a21e00 | 2569 | SvUTF8_off(TARG); |
1d68d6cd | 2570 | } |
ec93b65f | 2571 | SETTARG; |
1d68d6cd SC |
2572 | RETURN; |
2573 | } | |
a0d0e21e | 2574 | #ifdef LIBERAL |
51723571 JH |
2575 | { |
2576 | register long *tmpl; | |
2577 | for ( ; anum && (unsigned long)tmps % sizeof(long); anum--, tmps++) | |
2578 | *tmps = ~*tmps; | |
2579 | tmpl = (long*)tmps; | |
bb7a0f54 | 2580 | for ( ; anum >= (I32)sizeof(long); anum -= (I32)sizeof(long), tmpl++) |
51723571 JH |
2581 | *tmpl = ~*tmpl; |
2582 | tmps = (U8*)tmpl; | |
2583 | } | |
a0d0e21e LW |
2584 | #endif |
2585 | for ( ; anum > 0; anum--, tmps++) | |
2586 | *tmps = ~*tmps; | |
ec93b65f | 2587 | SETTARG; |
a0d0e21e LW |
2588 | } |
2589 | RETURN; | |
2590 | } | |
79072805 LW |
2591 | } |
2592 | ||
a0d0e21e LW |
2593 | /* integer versions of some of the above */ |
2594 | ||
a0d0e21e | 2595 | PP(pp_i_multiply) |
79072805 | 2596 | { |
97aff369 | 2597 | dVAR; dSP; dATARGET; tryAMAGICbin(mult,opASSIGN); |
a0d0e21e LW |
2598 | { |
2599 | dPOPTOPiirl; | |
2600 | SETi( left * right ); | |
2601 | RETURN; | |
2602 | } | |
79072805 LW |
2603 | } |
2604 | ||
a0d0e21e | 2605 | PP(pp_i_divide) |
79072805 | 2606 | { |
ece1bcef | 2607 | IV num; |
97aff369 | 2608 | dVAR; dSP; dATARGET; tryAMAGICbin(div,opASSIGN); |
a0d0e21e LW |
2609 | { |
2610 | dPOPiv; | |
2611 | if (value == 0) | |
ece1bcef SP |
2612 | DIE(aTHX_ "Illegal division by zero"); |
2613 | num = POPi; | |
a0cec769 YST |
2614 | |
2615 | /* avoid FPE_INTOVF on some platforms when num is IV_MIN */ | |
2616 | if (value == -1) | |
2617 | value = - num; | |
2618 | else | |
2619 | value = num / value; | |
a0d0e21e LW |
2620 | PUSHi( value ); |
2621 | RETURN; | |
2622 | } | |
79072805 LW |
2623 | } |
2624 | ||
befad5d1 | 2625 | #if defined(__GLIBC__) && IVSIZE == 8 |
224ec323 JH |
2626 | STATIC |
2627 | PP(pp_i_modulo_0) | |
befad5d1 NC |
2628 | #else |
2629 | PP(pp_i_modulo) | |
2630 | #endif | |
224ec323 JH |
2631 | { |
2632 | /* This is the vanilla old i_modulo. */ | |
27da23d5 | 2633 | dVAR; dSP; dATARGET; tryAMAGICbin(modulo,opASSIGN); |
224ec323 JH |
2634 | { |
2635 | dPOPTOPiirl; | |
2636 | if (!right) | |
2637 | DIE(aTHX_ "Illegal modulus zero"); | |
a0cec769 YST |
2638 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
2639 | if (right == -1) | |
2640 | SETi( 0 ); | |
2641 | else | |
2642 | SETi( left % right ); | |
224ec323 JH |
2643 | RETURN; |
2644 | } | |
2645 | } | |
2646 | ||
11010fa3 | 2647 | #if defined(__GLIBC__) && IVSIZE == 8 |
224ec323 JH |
2648 | STATIC |
2649 | PP(pp_i_modulo_1) | |
befad5d1 | 2650 | |
224ec323 | 2651 | { |
224ec323 | 2652 | /* This is the i_modulo with the workaround for the _moddi3 bug |
fce2b89e | 2653 | * in (at least) glibc 2.2.5 (the PERL_ABS() the workaround). |
224ec323 | 2654 | * See below for pp_i_modulo. */ |
5186cc12 | 2655 | dVAR; dSP; dATARGET; tryAMAGICbin(modulo,opASSIGN); |
224ec323 JH |
2656 | { |
2657 | dPOPTOPiirl; | |
2658 | if (!right) | |
2659 | DIE(aTHX_ "Illegal modulus zero"); | |
a0cec769 YST |
2660 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
2661 | if (right == -1) | |
2662 | SETi( 0 ); | |
2663 | else | |
2664 | SETi( left % PERL_ABS(right) ); | |
224ec323 JH |
2665 | RETURN; |
2666 | } | |
224ec323 JH |
2667 | } |
2668 | ||
a0d0e21e | 2669 | PP(pp_i_modulo) |
79072805 | 2670 | { |
27da23d5 | 2671 | dVAR; dSP; dATARGET; tryAMAGICbin(modulo,opASSIGN); |
224ec323 JH |
2672 | { |
2673 | dPOPTOPiirl; | |
2674 | if (!right) | |
2675 | DIE(aTHX_ "Illegal modulus zero"); | |
2676 | /* The assumption is to use hereafter the old vanilla version... */ | |
2677 | PL_op->op_ppaddr = | |
2678 | PL_ppaddr[OP_I_MODULO] = | |
1c127fab | 2679 | Perl_pp_i_modulo_0; |
224ec323 JH |
2680 | /* .. but if we have glibc, we might have a buggy _moddi3 |
2681 | * (at least glicb 2.2.5 is known to have this bug), in other | |
2682 | * words our integer modulus with negative quad as the second | |
2683 | * argument might be broken. Test for this and re-patch the | |
2684 | * opcode dispatch table if that is the case, remembering to | |
2685 | * also apply the workaround so that this first round works | |
2686 | * right, too. See [perl #9402] for more information. */ | |
224ec323 JH |
2687 | { |
2688 | IV l = 3; | |
2689 | IV r = -10; | |
2690 | /* Cannot do this check with inlined IV constants since | |
2691 | * that seems to work correctly even with the buggy glibc. */ | |
2692 | if (l % r == -3) { | |
2693 | /* Yikes, we have the bug. | |
2694 | * Patch in the workaround version. */ | |
2695 | PL_op->op_ppaddr = | |
2696 | PL_ppaddr[OP_I_MODULO] = | |
2697 | &Perl_pp_i_modulo_1; | |
2698 | /* Make certain we work right this time, too. */ | |
32fdb065 | 2699 | right = PERL_ABS(right); |
224ec323 JH |
2700 | } |
2701 | } | |
a0cec769 YST |
2702 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
2703 | if (right == -1) | |
2704 | SETi( 0 ); | |
2705 | else | |
2706 | SETi( left % right ); | |
224ec323 JH |
2707 | RETURN; |
2708 | } | |
79072805 | 2709 | } |
befad5d1 | 2710 | #endif |
79072805 | 2711 | |
a0d0e21e | 2712 | PP(pp_i_add) |
79072805 | 2713 | { |
97aff369 | 2714 | dVAR; dSP; dATARGET; tryAMAGICbin(add,opASSIGN); |
a0d0e21e | 2715 | { |
5e66d4f1 | 2716 | dPOPTOPiirl_ul; |
a0d0e21e LW |
2717 | SETi( left + right ); |
2718 | RETURN; | |
79072805 | 2719 | } |
79072805 LW |
2720 | } |
2721 | ||
a0d0e21e | 2722 | PP(pp_i_subtract) |
79072805 | 2723 | { |
97aff369 | 2724 | dVAR; dSP; dATARGET; tryAMAGICbin(subtr,opASSIGN); |
a0d0e21e | 2725 | { |
5e66d4f1 | 2726 | dPOPTOPiirl_ul; |
a0d0e21e LW |
2727 | SETi( left - right ); |
2728 | RETURN; | |
79072805 | 2729 | } |
79072805 LW |
2730 | } |
2731 | ||
a0d0e21e | 2732 | PP(pp_i_lt) |
79072805 | 2733 | { |
97aff369 | 2734 | dVAR; dSP; tryAMAGICbinSET(lt,0); |
a0d0e21e LW |
2735 | { |
2736 | dPOPTOPiirl; | |
54310121 | 2737 | SETs(boolSV(left < right)); |
a0d0e21e LW |
2738 | RETURN; |
2739 | } | |
79072805 LW |
2740 | } |
2741 | ||
a0d0e21e | 2742 | PP(pp_i_gt) |
79072805 | 2743 | { |
97aff369 | 2744 | dVAR; dSP; tryAMAGICbinSET(gt,0); |
a0d0e21e LW |
2745 | { |
2746 | dPOPTOPiirl; | |
54310121 | 2747 | SETs(boolSV(left > right)); |
a0d0e21e LW |
2748 | RETURN; |
2749 | } | |
79072805 LW |
2750 | } |
2751 | ||
a0d0e21e | 2752 | PP(pp_i_le) |
79072805 | 2753 | { |
97aff369 | 2754 | dVAR; dSP; tryAMAGICbinSET(le,0); |
a0d0e21e LW |
2755 | { |
2756 | dPOPTOPiirl; | |
54310121 | 2757 | SETs(boolSV(left <= right)); |
a0d0e21e | 2758 | RETURN; |
85e6fe83 | 2759 | } |
79072805 LW |
2760 | } |
2761 | ||
a0d0e21e | 2762 | PP(pp_i_ge) |
79072805 | 2763 | { |
97aff369 | 2764 | dVAR; dSP; tryAMAGICbinSET(ge,0); |
a0d0e21e LW |
2765 | { |
2766 | dPOPTOPiirl; | |
54310121 | 2767 | SETs(boolSV(left >= right)); |
a0d0e21e LW |
2768 | RETURN; |
2769 | } | |
79072805 LW |
2770 | } |
2771 | ||
a0d0e21e | 2772 | PP(pp_i_eq) |
79072805 | 2773 | { |
97aff369 | 2774 | dVAR; dSP; tryAMAGICbinSET(eq,0); |
a0d0e21e LW |
2775 | { |
2776 | dPOPTOPiirl; | |
54310121 | 2777 | SETs(boolSV(left == right)); |
a0d0e21e LW |
2778 | RETURN; |
2779 | } | |
79072805 LW |
2780 | } |
2781 | ||
a0d0e21e | 2782 | PP(pp_i_ne) |
79072805 | 2783 | { |
97aff369 | 2784 | dVAR; dSP; tryAMAGICbinSET(ne,0); |
a0d0e21e LW |
2785 | { |
2786 | dPOPTOPiirl; | |
54310121 | 2787 | SETs(boolSV(left != right)); |
a0d0e21e LW |
2788 | RETURN; |
2789 | } | |
79072805 LW |
2790 | } |
2791 | ||
a0d0e21e | 2792 | PP(pp_i_ncmp) |
79072805 | 2793 | { |
97aff369 | 2794 | dVAR; dSP; dTARGET; tryAMAGICbin(ncmp,0); |
a0d0e21e LW |
2795 | { |
2796 | dPOPTOPiirl; | |
2797 | I32 value; | |
79072805 | 2798 | |
a0d0e21e | 2799 | if (left > right) |
79072805 | 2800 | value = 1; |
a0d0e21e | 2801 | else if (left < right) |
79072805 | 2802 | value = -1; |
a0d0e21e | 2803 | else |
79072805 | 2804 | value = 0; |
a0d0e21e LW |
2805 | SETi(value); |
2806 | RETURN; | |
79072805 | 2807 | } |
85e6fe83 LW |
2808 | } |
2809 | ||
2810 | PP(pp_i_negate) | |
2811 | { | |
97aff369 | 2812 | dVAR; dSP; dTARGET; tryAMAGICun(neg); |
85e6fe83 LW |
2813 | SETi(-TOPi); |
2814 | RETURN; | |
2815 | } | |
2816 | ||
79072805 LW |
2817 | /* High falutin' math. */ |
2818 | ||
2819 | PP(pp_atan2) | |
2820 | { | |
97aff369 | 2821 | dVAR; dSP; dTARGET; tryAMAGICbin(atan2,0); |
a0d0e21e LW |
2822 | { |
2823 | dPOPTOPnnrl; | |
a1021d57 | 2824 | SETn(Perl_atan2(left, right)); |
a0d0e21e LW |
2825 | RETURN; |
2826 | } | |
79072805 LW |
2827 | } |
2828 | ||
2829 | PP(pp_sin) | |
2830 | { | |
71302fe3 NC |
2831 | dVAR; dSP; dTARGET; |
2832 | int amg_type = sin_amg; | |
2833 | const char *neg_report = NULL; | |
bc81784a | 2834 | NV (*func)(NV) = Perl_sin; |
71302fe3 NC |
2835 | const int op_type = PL_op->op_type; |
2836 | ||
2837 | switch (op_type) { | |
2838 | case OP_COS: | |
2839 | amg_type = cos_amg; | |
bc81784a | 2840 | func = Perl_cos; |
71302fe3 NC |
2841 | break; |
2842 | case OP_EXP: | |
2843 | amg_type = exp_amg; | |
bc81784a | 2844 | func = Perl_exp; |
71302fe3 NC |
2845 | break; |
2846 | case OP_LOG: | |
2847 | amg_type = log_amg; | |
bc81784a | 2848 | func = Perl_log; |
71302fe3 NC |
2849 | neg_report = "log"; |
2850 | break; | |
2851 | case OP_SQRT: | |
2852 | amg_type = sqrt_amg; | |
bc81784a | 2853 | func = Perl_sqrt; |
71302fe3 NC |
2854 | neg_report = "sqrt"; |
2855 | break; | |
a0d0e21e | 2856 | } |
79072805 | 2857 | |
71302fe3 | 2858 | tryAMAGICun_var(amg_type); |
a0d0e21e | 2859 | { |
1b6737cc | 2860 | const NV value = POPn; |
71302fe3 NC |
2861 | if (neg_report) { |
2862 | if (op_type == OP_LOG ? (value <= 0.0) : (value < 0.0)) { | |
2863 | SET_NUMERIC_STANDARD(); | |
2864 | DIE(aTHX_ "Can't take %s of %"NVgf, neg_report, value); | |
2865 | } | |
2866 | } | |
2867 | XPUSHn(func(value)); | |
a0d0e21e LW |
2868 | RETURN; |
2869 | } | |
79072805 LW |
2870 | } |
2871 | ||
56cb0a1c AD |
2872 | /* Support Configure command-line overrides for rand() functions. |
2873 | After 5.005, perhaps we should replace this by Configure support | |
2874 | for drand48(), random(), or rand(). For 5.005, though, maintain | |
2875 | compatibility by calling rand() but allow the user to override it. | |
2876 | See INSTALL for details. --Andy Dougherty 15 July 1998 | |
2877 | */ | |
85ab1d1d JH |
2878 | /* Now it's after 5.005, and Configure supports drand48() and random(), |
2879 | in addition to rand(). So the overrides should not be needed any more. | |
2880 | --Jarkko Hietaniemi 27 September 1998 | |
2881 | */ | |
2882 | ||
2883 | #ifndef HAS_DRAND48_PROTO | |
20ce7b12 | 2884 | extern double drand48 (void); |
56cb0a1c AD |
2885 | #endif |
2886 | ||
79072805 LW |
2887 | PP(pp_rand) |
2888 | { | |
97aff369 | 2889 | dVAR; dSP; dTARGET; |
65202027 | 2890 | NV value; |
79072805 LW |
2891 | if (MAXARG < 1) |
2892 | value = 1.0; | |
2893 | else | |
2894 | value = POPn; | |
2895 | if (value == 0.0) | |
2896 | value = 1.0; | |
80252599 | 2897 | if (!PL_srand_called) { |
85ab1d1d | 2898 | (void)seedDrand01((Rand_seed_t)seed()); |
80252599 | 2899 | PL_srand_called = TRUE; |
93dc8474 | 2900 | } |
85ab1d1d | 2901 | value *= Drand01(); |
79072805 LW |
2902 | XPUSHn(value); |
2903 | RETURN; | |
2904 | } | |
2905 | ||
2906 | PP(pp_srand) | |
2907 | { | |
97aff369 | 2908 | dVAR; dSP; |
0bd48802 | 2909 | const UV anum = (MAXARG < 1) ? seed() : POPu; |
85ab1d1d | 2910 | (void)seedDrand01((Rand_seed_t)anum); |
80252599 | 2911 | PL_srand_called = TRUE; |
79072805 LW |
2912 | EXTEND(SP, 1); |
2913 | RETPUSHYES; | |
2914 | } | |
2915 | ||
79072805 LW |
2916 | PP(pp_int) |
2917 | { | |
97aff369 | 2918 | dVAR; dSP; dTARGET; tryAMAGICun(int); |
774d564b | 2919 | { |
800401ee JH |
2920 | SV * const sv = sv_2num(TOPs); |
2921 | const IV iv = SvIV(sv); | |
28e5dec8 JH |
2922 | /* XXX it's arguable that compiler casting to IV might be subtly |
2923 | different from modf (for numbers inside (IV_MIN,UV_MAX)) in which | |
2924 | else preferring IV has introduced a subtle behaviour change bug. OTOH | |
2925 | relying on floating point to be accurate is a bug. */ | |
2926 | ||
c781a409 | 2927 | if (!SvOK(sv)) { |
922c4365 | 2928 | SETu(0); |
c781a409 RD |
2929 | } |
2930 | else if (SvIOK(sv)) { | |
2931 | if (SvIsUV(sv)) | |
2932 | SETu(SvUV(sv)); | |
2933 | else | |
28e5dec8 | 2934 | SETi(iv); |
c781a409 | 2935 | } |
c781a409 RD |
2936 | else { |
2937 | const NV value = SvNV(sv); | |
1048ea30 | 2938 | if (value >= 0.0) { |
28e5dec8 JH |
2939 | if (value < (NV)UV_MAX + 0.5) { |
2940 | SETu(U_V(value)); | |
2941 | } else { | |
059a1014 | 2942 | SETn(Perl_floor(value)); |
28e5dec8 | 2943 | } |
1048ea30 | 2944 | } |
28e5dec8 JH |
2945 | else { |
2946 | if (value > (NV)IV_MIN - 0.5) { | |
2947 | SETi(I_V(value)); | |
2948 | } else { | |
1bbae031 | 2949 | SETn(Perl_ceil(value)); |
28e5dec8 JH |
2950 | } |
2951 | } | |
774d564b | 2952 | } |
79072805 | 2953 | } |
79072805 LW |
2954 | RETURN; |
2955 | } | |
2956 | ||
463ee0b2 LW |
2957 | PP(pp_abs) |
2958 | { | |
97aff369 | 2959 | dVAR; dSP; dTARGET; tryAMAGICun(abs); |
a0d0e21e | 2960 | { |
800401ee | 2961 | SV * const sv = sv_2num(TOPs); |
28e5dec8 | 2962 | /* This will cache the NV value if string isn't actually integer */ |
800401ee | 2963 | const IV iv = SvIV(sv); |
a227d84d | 2964 | |
800401ee | 2965 | if (!SvOK(sv)) { |
922c4365 | 2966 | SETu(0); |
800401ee JH |
2967 | } |
2968 | else if (SvIOK(sv)) { | |
28e5dec8 | 2969 | /* IVX is precise */ |
800401ee JH |
2970 | if (SvIsUV(sv)) { |
2971 | SETu(SvUV(sv)); /* force it to be numeric only */ | |
28e5dec8 JH |
2972 | } else { |
2973 | if (iv >= 0) { | |
2974 | SETi(iv); | |
2975 | } else { | |
2976 | if (iv != IV_MIN) { | |
2977 | SETi(-iv); | |
2978 | } else { | |
2979 | /* 2s complement assumption. Also, not really needed as | |
2980 | IV_MIN and -IV_MIN should both be %100...00 and NV-able */ | |
2981 | SETu(IV_MIN); | |
2982 | } | |
a227d84d | 2983 | } |
28e5dec8 JH |
2984 | } |
2985 | } else{ | |
800401ee | 2986 | const NV value = SvNV(sv); |
774d564b | 2987 | if (value < 0.0) |
1b6737cc | 2988 | SETn(-value); |
a4474c9e DD |
2989 | else |
2990 | SETn(value); | |
774d564b | 2991 | } |
a0d0e21e | 2992 | } |
774d564b | 2993 | RETURN; |
463ee0b2 LW |
2994 | } |
2995 | ||
79072805 LW |
2996 | PP(pp_oct) |
2997 | { | |
97aff369 | 2998 | dVAR; dSP; dTARGET; |
5c144d81 | 2999 | const char *tmps; |
53305cf1 | 3000 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES; |
6f894ead | 3001 | STRLEN len; |
53305cf1 NC |
3002 | NV result_nv; |
3003 | UV result_uv; | |
1b6737cc | 3004 | SV* const sv = POPs; |
79072805 | 3005 | |
349d4f2f | 3006 | tmps = (SvPV_const(sv, len)); |
2bc69dc4 NIS |
3007 | if (DO_UTF8(sv)) { |
3008 | /* If Unicode, try to downgrade | |
3009 | * If not possible, croak. */ | |
1b6737cc | 3010 | SV* const tsv = sv_2mortal(newSVsv(sv)); |
2bc69dc4 NIS |
3011 | |
3012 | SvUTF8_on(tsv); | |
3013 | sv_utf8_downgrade(tsv, FALSE); | |
349d4f2f | 3014 | tmps = SvPV_const(tsv, len); |
2bc69dc4 | 3015 | } |
daa2adfd NC |
3016 | if (PL_op->op_type == OP_HEX) |
3017 | goto hex; | |
3018 | ||
6f894ead | 3019 | while (*tmps && len && isSPACE(*tmps)) |
53305cf1 | 3020 | tmps++, len--; |
9e24b6e2 | 3021 | if (*tmps == '0') |
53305cf1 | 3022 | tmps++, len--; |
daa2adfd NC |
3023 | if (*tmps == 'x') { |
3024 | hex: | |
53305cf1 | 3025 | result_uv = grok_hex (tmps, &len, &flags, &result_nv); |
daa2adfd | 3026 | } |
9e24b6e2 | 3027 | else if (*tmps == 'b') |
53305cf1 | 3028 | result_uv = grok_bin (tmps, &len, &flags, &result_nv); |
464e2e8a | 3029 | else |
53305cf1 NC |
3030 | result_uv = grok_oct (tmps, &len, &flags, &result_nv); |
3031 | ||
3032 | if (flags & PERL_SCAN_GREATER_THAN_UV_MAX) { | |
3033 | XPUSHn(result_nv); | |
3034 | } | |
3035 | else { | |
3036 | XPUSHu(result_uv); | |
3037 | } | |
79072805 LW |
3038 | RETURN; |
3039 | } | |
3040 | ||
3041 | /* String stuff. */ | |
3042 | ||
3043 | PP(pp_length) | |
3044 | { | |
97aff369 | 3045 | dVAR; dSP; dTARGET; |
0bd48802 | 3046 | SV * const sv = TOPs; |
a0ed51b3 | 3047 | |
656266fc | 3048 | if (SvGAMAGIC(sv)) { |
9f621bb0 NC |
3049 | /* For an overloaded or magic scalar, we can't know in advance if |
3050 | it's going to be UTF-8 or not. Also, we can't call sv_len_utf8 as | |
3051 | it likes to cache the length. Maybe that should be a documented | |
3052 | feature of it. | |
92331800 NC |
3053 | */ |
3054 | STRLEN len; | |
9f621bb0 NC |
3055 | const char *const p |
3056 | = sv_2pv_flags(sv, &len, | |
3057 | SV_UNDEF_RETURNS_NULL|SV_CONST_RETURN|SV_GMAGIC); | |
92331800 | 3058 | |
9f621bb0 NC |
3059 | if (!p) |
3060 | SETs(&PL_sv_undef); | |
3061 | else if (DO_UTF8(sv)) { | |
899be101 | 3062 | SETi(utf8_length((U8*)p, (U8*)p + len)); |
92331800 NC |
3063 | } |
3064 | else | |
3065 | SETi(len); | |
656266fc | 3066 | } else if (SvOK(sv)) { |
9f621bb0 NC |
3067 | /* Neither magic nor overloaded. */ |
3068 | if (DO_UTF8(sv)) | |
3069 | SETi(sv_len_utf8(sv)); | |
3070 | else | |
3071 | SETi(sv_len(sv)); | |
656266fc NC |
3072 | } else { |
3073 | SETs(&PL_sv_undef); | |
92331800 | 3074 | } |
79072805 LW |
3075 | RETURN; |
3076 | } | |
3077 | ||
3078 | PP(pp_substr) | |
3079 | { | |
97aff369 | 3080 | dVAR; dSP; dTARGET; |
79072805 | 3081 | SV *sv; |
9c5ffd7c | 3082 | I32 len = 0; |
463ee0b2 | 3083 | STRLEN curlen; |
9402d6ed | 3084 | STRLEN utf8_curlen; |
79072805 LW |
3085 | I32 pos; |
3086 | I32 rem; | |
84902520 | 3087 | I32 fail; |
050e6362 | 3088 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
e1ec3a88 | 3089 | const char *tmps; |
fc15ae8f | 3090 | const I32 arybase = CopARYBASE_get(PL_curcop); |
9402d6ed | 3091 | SV *repl_sv = NULL; |
cbbf8932 | 3092 | const char *repl = NULL; |
7b8d334a | 3093 | STRLEN repl_len; |
050e6362 | 3094 | const int num_args = PL_op->op_private & 7; |
13e30c65 | 3095 | bool repl_need_utf8_upgrade = FALSE; |
9402d6ed | 3096 | bool repl_is_utf8 = FALSE; |
79072805 | 3097 | |
20408e3c | 3098 | SvTAINTED_off(TARG); /* decontaminate */ |
7e2040f0 | 3099 | SvUTF8_off(TARG); /* decontaminate */ |
78f9721b SM |
3100 | if (num_args > 2) { |
3101 | if (num_args > 3) { | |
9402d6ed | 3102 | repl_sv = POPs; |
83003860 | 3103 | repl = SvPV_const(repl_sv, repl_len); |
9402d6ed | 3104 | repl_is_utf8 = DO_UTF8(repl_sv) && SvCUR(repl_sv); |
7b8d334a | 3105 | } |
79072805 | 3106 | len = POPi; |
5d82c453 | 3107 | } |
84902520 | 3108 | pos = POPi; |
79072805 | 3109 | sv = POPs; |
849ca7ee | 3110 | PUTBACK; |
9402d6ed JH |
3111 | if (repl_sv) { |
3112 | if (repl_is_utf8) { | |
3113 | if (!DO_UTF8(sv)) | |
3114 | sv_utf8_upgrade(sv); | |
3115 | } | |
13e30c65 JH |
3116 | else if (DO_UTF8(sv)) |
3117 | repl_need_utf8_upgrade = TRUE; | |
9402d6ed | 3118 | } |
5c144d81 | 3119 | tmps = SvPV_const(sv, curlen); |
7e2040f0 | 3120 | if (DO_UTF8(sv)) { |
9402d6ed JH |
3121 | utf8_curlen = sv_len_utf8(sv); |
3122 | if (utf8_curlen == curlen) | |
3123 | utf8_curlen = 0; | |
a0ed51b3 | 3124 | else |
9402d6ed | 3125 | curlen = utf8_curlen; |
a0ed51b3 | 3126 | } |
d1c2b58a | 3127 | else |
9402d6ed | 3128 | utf8_curlen = 0; |
a0ed51b3 | 3129 | |
84902520 TB |
3130 | if (pos >= arybase) { |
3131 | pos -= arybase; | |
3132 | rem = curlen-pos; | |
3133 | fail = rem; | |
78f9721b | 3134 | if (num_args > 2) { |
5d82c453 GA |
3135 | if (len < 0) { |
3136 | rem += len; | |
3137 | if (rem < 0) | |
3138 | rem = 0; | |
3139 | } | |
3140 | else if (rem > len) | |
3141 | rem = len; | |
3142 | } | |
68dc0745 | 3143 | } |
84902520 | 3144 | else { |
5d82c453 | 3145 | pos += curlen; |
78f9721b | 3146 | if (num_args < 3) |
5d82c453 GA |
3147 | rem = curlen; |
3148 | else if (len >= 0) { | |
3149 | rem = pos+len; | |
3150 | if (rem > (I32)curlen) | |
3151 | rem = curlen; | |
3152 | } | |
3153 | else { | |
3154 | rem = curlen+len; | |
3155 | if (rem < pos) | |
3156 | rem = pos; | |
3157 | } | |
3158 | if (pos < 0) | |
3159 | pos = 0; | |
3160 | fail = rem; | |
3161 | rem -= pos; | |
84902520 TB |
3162 | } |
3163 | if (fail < 0) { | |
e476b1b5 GS |
3164 | if (lvalue || repl) |
3165 | Perl_croak(aTHX_ "substr outside of string"); | |
a2a5de95 | 3166 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), "substr outside of string"); |
2304df62 AD |
3167 | RETPUSHUNDEF; |
3168 | } | |
79072805 | 3169 | else { |
1b6737cc AL |
3170 | const I32 upos = pos; |
3171 | const I32 urem = rem; | |
9402d6ed | 3172 | if (utf8_curlen) |
a0ed51b3 | 3173 | sv_pos_u2b(sv, &pos, &rem); |
79072805 | 3174 | tmps += pos; |
781e7547 DM |
3175 | /* we either return a PV or an LV. If the TARG hasn't been used |
3176 | * before, or is of that type, reuse it; otherwise use a mortal | |
3177 | * instead. Note that LVs can have an extended lifetime, so also | |
3178 | * dont reuse if refcount > 1 (bug #20933) */ | |
3179 | if (SvTYPE(TARG) > SVt_NULL) { | |
3180 | if ( (SvTYPE(TARG) == SVt_PVLV) | |
3181 | ? (!lvalue || SvREFCNT(TARG) > 1) | |
3182 | : lvalue) | |
3183 | { | |
3184 | TARG = sv_newmortal(); | |
3185 | } | |
3186 | } | |
3187 | ||
050e6362 | 3188 | sv_setpvn(TARG, tmps, rem); |
12aa1545 | 3189 | #ifdef USE_LOCALE_COLLATE |
14befaf4 | 3190 | sv_unmagic(TARG, PERL_MAGIC_collxfrm); |
12aa1545 | 3191 | #endif |
9402d6ed | 3192 | if (utf8_curlen) |
7f66633b | 3193 | SvUTF8_on(TARG); |
f7928d6c | 3194 | if (repl) { |
13e30c65 JH |
3195 | SV* repl_sv_copy = NULL; |
3196 | ||
3197 | if (repl_need_utf8_upgrade) { | |
3198 | repl_sv_copy = newSVsv(repl_sv); | |
3199 | sv_utf8_upgrade(repl_sv_copy); | |
349d4f2f | 3200 | repl = SvPV_const(repl_sv_copy, repl_len); |
13e30c65 JH |
3201 | repl_is_utf8 = DO_UTF8(repl_sv_copy) && SvCUR(sv); |
3202 | } | |
502d9230 VP |
3203 | if (!SvOK(sv)) |
3204 | sv_setpvs(sv, ""); | |
c0dd94a0 | 3205 | sv_insert_flags(sv, pos, rem, repl, repl_len, 0); |
9402d6ed | 3206 | if (repl_is_utf8) |
f7928d6c | 3207 | SvUTF8_on(sv); |
ef8d46e8 | 3208 | SvREFCNT_dec(repl_sv_copy); |
f7928d6c | 3209 | } |
c8faf1c5 | 3210 | else if (lvalue) { /* it's an lvalue! */ |
dedeecda | 3211 | if (!SvGMAGICAL(sv)) { |
3212 | if (SvROK(sv)) { | |
13c5b33c | 3213 | SvPV_force_nolen(sv); |
a2a5de95 NC |
3214 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), |
3215 | "Attempt to use reference as lvalue in substr"); | |
dedeecda | 3216 | } |
f7877b28 NC |
3217 | if (isGV_with_GP(sv)) |
3218 | SvPV_force_nolen(sv); | |
3219 | else if (SvOK(sv)) /* is it defined ? */ | |
7f66633b | 3220 | (void)SvPOK_only_UTF8(sv); |
dedeecda | 3221 | else |
523f125d | 3222 | sv_setpvs(sv, ""); /* avoid lexical reincarnation */ |
dedeecda | 3223 | } |
5f05dabc | 3224 | |
a0d0e21e LW |
3225 | if (SvTYPE(TARG) < SVt_PVLV) { |
3226 | sv_upgrade(TARG, SVt_PVLV); | |
c445ea15 | 3227 | sv_magic(TARG, NULL, PERL_MAGIC_substr, NULL, 0); |
ed6116ce | 3228 | } |
a0d0e21e | 3229 | |
5f05dabc | 3230 | LvTYPE(TARG) = 'x'; |
6ff81951 | 3231 | if (LvTARG(TARG) != sv) { |
cb39f75f | 3232 | SvREFCNT_dec(LvTARG(TARG)); |
b37c2d43 | 3233 | LvTARG(TARG) = SvREFCNT_inc_simple(sv); |
6ff81951 | 3234 | } |
9aa983d2 JH |
3235 | LvTARGOFF(TARG) = upos; |
3236 | LvTARGLEN(TARG) = urem; | |
79072805 LW |
3237 | } |
3238 | } | |
849ca7ee | 3239 | SPAGAIN; |
79072805 LW |
3240 | PUSHs(TARG); /* avoid SvSETMAGIC here */ |
3241 | RETURN; | |
3242 | } | |
3243 | ||
3244 | PP(pp_vec) | |
3245 | { | |
97aff369 | 3246 | dVAR; dSP; dTARGET; |
1b6737cc AL |
3247 | register const IV size = POPi; |
3248 | register const IV offset = POPi; | |
3249 | register SV * const src = POPs; | |
3250 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; | |
a0d0e21e | 3251 | |
81e118e0 JH |
3252 | SvTAINTED_off(TARG); /* decontaminate */ |
3253 | if (lvalue) { /* it's an lvalue! */ | |
24aef97f HS |
3254 | if (SvREFCNT(TARG) > 1) /* don't share the TARG (#20933) */ |
3255 | TARG = sv_newmortal(); | |
81e118e0 JH |
3256 | if (SvTYPE(TARG) < SVt_PVLV) { |
3257 | sv_upgrade(TARG, SVt_PVLV); | |
c445ea15 | 3258 | sv_magic(TARG, NULL, PERL_MAGIC_vec, NULL, 0); |
79072805 | 3259 | } |
81e118e0 JH |
3260 | LvTYPE(TARG) = 'v'; |
3261 | if (LvTARG(TARG) != src) { | |
cb39f75f | 3262 | SvREFCNT_dec(LvTARG(TARG)); |
b37c2d43 | 3263 | LvTARG(TARG) = SvREFCNT_inc_simple(src); |
79072805 | 3264 | } |
81e118e0 JH |
3265 | LvTARGOFF(TARG) = offset; |
3266 | LvTARGLEN(TARG) = size; | |
79072805 LW |
3267 | } |
3268 | ||
81e118e0 | 3269 | sv_setuv(TARG, do_vecget(src, offset, size)); |
79072805 LW |
3270 | PUSHs(TARG); |
3271 | RETURN; | |
3272 | } | |
3273 | ||
3274 | PP(pp_index) | |
3275 | { | |
97aff369 | 3276 | dVAR; dSP; dTARGET; |
79072805 LW |
3277 | SV *big; |
3278 | SV *little; | |
c445ea15 | 3279 | SV *temp = NULL; |
ad66a58c | 3280 | STRLEN biglen; |
2723d216 | 3281 | STRLEN llen = 0; |
79072805 LW |
3282 | I32 offset; |
3283 | I32 retval; | |
73ee8be2 NC |
3284 | const char *big_p; |
3285 | const char *little_p; | |
fc15ae8f | 3286 | const I32 arybase = CopARYBASE_get(PL_curcop); |
2f040f7f NC |
3287 | bool big_utf8; |
3288 | bool little_utf8; | |
2723d216 | 3289 | const bool is_index = PL_op->op_type == OP_INDEX; |
79072805 | 3290 | |
2723d216 NC |
3291 | if (MAXARG >= 3) { |
3292 | /* arybase is in characters, like offset, so combine prior to the | |
3293 | UTF-8 to bytes calculation. */ | |
79072805 | 3294 | offset = POPi - arybase; |
2723d216 | 3295 | } |
79072805 LW |
3296 | little = POPs; |
3297 | big = POPs; | |
73ee8be2 NC |
3298 | big_p = SvPV_const(big, biglen); |
3299 | little_p = SvPV_const(little, llen); | |
3300 | ||
e609e586 NC |
3301 | big_utf8 = DO_UTF8(big); |
3302 | little_utf8 = DO_UTF8(little); | |
3303 | if (big_utf8 ^ little_utf8) { | |
3304 | /* One needs to be upgraded. */ | |
2f040f7f NC |
3305 | if (little_utf8 && !PL_encoding) { |
3306 | /* Well, maybe instead we might be able to downgrade the small | |
3307 | string? */ | |
1eced8f8 | 3308 | char * const pv = (char*)bytes_from_utf8((U8 *)little_p, &llen, |
2f040f7f NC |
3309 | &little_utf8); |
3310 | if (little_utf8) { | |
3311 | /* If the large string is ISO-8859-1, and it's not possible to | |
3312 | convert the small string to ISO-8859-1, then there is no | |
3313 | way that it could be found anywhere by index. */ | |
3314 | retval = -1; | |
3315 | goto fail; | |
3316 | } | |
e609e586 | 3317 | |
2f040f7f NC |
3318 | /* At this point, pv is a malloc()ed string. So donate it to temp |
3319 | to ensure it will get free()d */ | |
3320 | little = temp = newSV(0); | |
73ee8be2 NC |
3321 | sv_usepvn(temp, pv, llen); |
3322 | little_p = SvPVX(little); | |
e609e586 | 3323 | } else { |
73ee8be2 NC |
3324 | temp = little_utf8 |
3325 | ? newSVpvn(big_p, biglen) : newSVpvn(little_p, llen); | |
2f040f7f NC |
3326 | |
3327 | if (PL_encoding) { | |
3328 | sv_recode_to_utf8(temp, PL_encoding); | |
3329 | } else { | |
3330 | sv_utf8_upgrade(temp); | |
3331 | } | |
3332 | if (little_utf8) { | |
3333 | big = temp; | |
3334 | big_utf8 = TRUE; | |
73ee8be2 | 3335 | big_p = SvPV_const(big, biglen); |
2f040f7f NC |
3336 | } else { |
3337 | little = temp; | |
73ee8be2 | 3338 | little_p = SvPV_const(little, llen); |
2f040f7f | 3339 | } |
e609e586 NC |
3340 | } |
3341 | } | |
73ee8be2 NC |
3342 | if (SvGAMAGIC(big)) { |
3343 | /* Life just becomes a lot easier if I use a temporary here. | |
3344 | Otherwise I need to avoid calls to sv_pos_u2b(), which (dangerously) | |
3345 | will trigger magic and overloading again, as will fbm_instr() | |
3346 | */ | |
59cd0e26 NC |
3347 | big = newSVpvn_flags(big_p, biglen, |
3348 | SVs_TEMP | (big_utf8 ? SVf_UTF8 : 0)); | |
73ee8be2 NC |
3349 | big_p = SvPVX(big); |
3350 | } | |
e4e44778 | 3351 | if (SvGAMAGIC(little) || (is_index && !SvOK(little))) { |
73ee8be2 NC |
3352 | /* index && SvOK() is a hack. fbm_instr() calls SvPV_const, which will |
3353 | warn on undef, and we've already triggered a warning with the | |
3354 | SvPV_const some lines above. We can't remove that, as we need to | |
3355 | call some SvPV to trigger overloading early and find out if the | |
3356 | string is UTF-8. | |
3357 | This is all getting to messy. The API isn't quite clean enough, | |
3358 | because data access has side effects. | |
3359 | */ | |
59cd0e26 NC |
3360 | little = newSVpvn_flags(little_p, llen, |
3361 | SVs_TEMP | (little_utf8 ? SVf_UTF8 : 0)); | |
73ee8be2 NC |
3362 | little_p = SvPVX(little); |
3363 | } | |
e609e586 | 3364 | |
79072805 | 3365 | if (MAXARG < 3) |
2723d216 | 3366 | offset = is_index ? 0 : biglen; |
a0ed51b3 | 3367 | else { |
ad66a58c | 3368 | if (big_utf8 && offset > 0) |
a0ed51b3 | 3369 | sv_pos_u2b(big, &offset, 0); |
73ee8be2 NC |
3370 | if (!is_index) |
3371 | offset += llen; | |
a0ed51b3 | 3372 | } |
79072805 LW |
3373 | if (offset < 0) |
3374 | offset = 0; | |
ad66a58c NC |
3375 | else if (offset > (I32)biglen) |
3376 | offset = biglen; | |
73ee8be2 NC |
3377 | if (!(little_p = is_index |
3378 | ? fbm_instr((unsigned char*)big_p + offset, | |
3379 | (unsigned char*)big_p + biglen, little, 0) | |
3380 | : rninstr(big_p, big_p + offset, | |
3381 | little_p, little_p + llen))) | |
a0ed51b3 | 3382 | retval = -1; |
ad66a58c | 3383 | else { |
73ee8be2 | 3384 | retval = little_p - big_p; |
ad66a58c NC |
3385 | if (retval > 0 && big_utf8) |
3386 | sv_pos_b2u(big, &retval); | |
3387 | } | |
ef8d46e8 | 3388 | SvREFCNT_dec(temp); |
2723d216 | 3389 | fail: |
a0ed51b3 | 3390 | PUSHi(retval + arybase); |
79072805 LW |
3391 | RETURN; |
3392 | } | |
3393 | ||
3394 | PP(pp_sprintf) | |
3395 | { | |
97aff369 | 3396 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
20ee07fb RGS |
3397 | if (SvTAINTED(MARK[1])) |
3398 | TAINT_PROPER("sprintf"); | |
79072805 | 3399 | do_sprintf(TARG, SP-MARK, MARK+1); |
bbce6d69 | 3400 | TAINT_IF(SvTAINTED(TARG)); |
79072805 LW |
3401 | SP = ORIGMARK; |
3402 | PUSHTARG; | |
3403 | RETURN; | |
3404 | } | |
3405 | ||
79072805 LW |
3406 | PP(pp_ord) |
3407 | { | |
97aff369 | 3408 | dVAR; dSP; dTARGET; |
1eced8f8 | 3409 | |
7df053ec | 3410 | SV *argsv = POPs; |
ba210ebe | 3411 | STRLEN len; |
349d4f2f | 3412 | const U8 *s = (U8*)SvPV_const(argsv, len); |
121910a4 | 3413 | |
799ef3cb | 3414 | if (PL_encoding && SvPOK(argsv) && !DO_UTF8(argsv)) { |
1eced8f8 | 3415 | SV * const tmpsv = sv_2mortal(newSVsv(argsv)); |
799ef3cb | 3416 | s = (U8*)sv_recode_to_utf8(tmpsv, PL_encoding); |
121910a4 JH |
3417 | argsv = tmpsv; |
3418 | } | |
79072805 | 3419 | |
872c91ae | 3420 | XPUSHu(DO_UTF8(argsv) ? |
89ebb4a3 | 3421 | utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, UTF8_ALLOW_ANYUV) : |
5fc32dea | 3422 | (UV)(*s & 0xff)); |
68795e93 | 3423 | |
79072805 LW |
3424 | RETURN; |
3425 | } | |
3426 | ||
463ee0b2 LW |
3427 | PP(pp_chr) |
3428 | { | |
97aff369 | 3429 | dVAR; dSP; dTARGET; |
463ee0b2 | 3430 | char *tmps; |
8a064bd6 JH |
3431 | UV value; |
3432 | ||
3433 | if (((SvIOK_notUV(TOPs) && SvIV(TOPs) < 0) | |
3434 | || | |
3435 | (SvNOK(TOPs) && SvNV(TOPs) < 0.0))) { | |
3436 | if (IN_BYTES) { | |
3437 | value = POPu; /* chr(-1) eq chr(0xff), etc. */ | |
3438 | } else { | |
3439 | (void) POPs; /* Ignore the argument value. */ | |
3440 | value = UNICODE_REPLACEMENT; | |
3441 | } | |
3442 | } else { | |
3443 | value = POPu; | |
3444 | } | |
463ee0b2 | 3445 | |
862a34c6 | 3446 | SvUPGRADE(TARG,SVt_PV); |
a0ed51b3 | 3447 | |
0064a8a9 | 3448 | if (value > 255 && !IN_BYTES) { |
eb160463 | 3449 | SvGROW(TARG, (STRLEN)UNISKIP(value)+1); |
62961d2e | 3450 | tmps = (char*)uvchr_to_utf8_flags((U8*)SvPVX(TARG), value, 0); |
349d4f2f | 3451 | SvCUR_set(TARG, tmps - SvPVX_const(TARG)); |
a0ed51b3 LW |
3452 | *tmps = '\0'; |
3453 | (void)SvPOK_only(TARG); | |
aa6ffa16 | 3454 | SvUTF8_on(TARG); |
a0ed51b3 LW |
3455 | XPUSHs(TARG); |
3456 | RETURN; | |
3457 | } | |
3458 | ||
748a9306 | 3459 | SvGROW(TARG,2); |
463ee0b2 LW |
3460 | SvCUR_set(TARG, 1); |
3461 | tmps = SvPVX(TARG); | |
eb160463 | 3462 | *tmps++ = (char)value; |
748a9306 | 3463 | *tmps = '\0'; |
a0d0e21e | 3464 | (void)SvPOK_only(TARG); |
4c5ed6e2 | 3465 | |
88632417 | 3466 | if (PL_encoding && !IN_BYTES) { |
799ef3cb | 3467 | sv_recode_to_utf8(TARG, PL_encoding); |
88632417 JH |
3468 | tmps = SvPVX(TARG); |
3469 | if (SvCUR(TARG) == 0 || !is_utf8_string((U8*)tmps, SvCUR(TARG)) || | |
4c5ed6e2 TS |
3470 | UNICODE_IS_REPLACEMENT(utf8_to_uvchr((U8*)tmps, NULL))) { |
3471 | SvGROW(TARG, 2); | |
d5a15ac2 | 3472 | tmps = SvPVX(TARG); |
4c5ed6e2 TS |
3473 | SvCUR_set(TARG, 1); |
3474 | *tmps++ = (char)value; | |
88632417 | 3475 | *tmps = '\0'; |
4c5ed6e2 | 3476 | SvUTF8_off(TARG); |
88632417 JH |
3477 | } |
3478 | } | |
4c5ed6e2 | 3479 | |
463ee0b2 LW |
3480 | XPUSHs(TARG); |
3481 | RETURN; | |
3482 | } | |
3483 | ||
79072805 LW |
3484 | PP(pp_crypt) |
3485 | { | |
79072805 | 3486 | #ifdef HAS_CRYPT |
97aff369 | 3487 | dVAR; dSP; dTARGET; |
5f74f29c | 3488 | dPOPTOPssrl; |
85c16d83 | 3489 | STRLEN len; |
10516c54 | 3490 | const char *tmps = SvPV_const(left, len); |
2bc69dc4 | 3491 | |
85c16d83 | 3492 | if (DO_UTF8(left)) { |
2bc69dc4 | 3493 | /* If Unicode, try to downgrade. |
f2791508 JH |
3494 | * If not possible, croak. |
3495 | * Yes, we made this up. */ | |
1b6737cc | 3496 | SV* const tsv = sv_2mortal(newSVsv(left)); |
2bc69dc4 | 3497 | |
f2791508 | 3498 | SvUTF8_on(tsv); |
2bc69dc4 | 3499 | sv_utf8_downgrade(tsv, FALSE); |
349d4f2f | 3500 | tmps = SvPV_const(tsv, len); |
85c16d83 | 3501 | } |
05404ffe JH |
3502 | # ifdef USE_ITHREADS |
3503 | # ifdef HAS_CRYPT_R | |
3504 | if (!PL_reentrant_buffer->_crypt_struct_buffer) { | |
3505 | /* This should be threadsafe because in ithreads there is only | |
3506 | * one thread per interpreter. If this would not be true, | |
3507 | * we would need a mutex to protect this malloc. */ | |
3508 | PL_reentrant_buffer->_crypt_struct_buffer = | |
3509 | (struct crypt_data *)safemalloc(sizeof(struct crypt_data)); | |
3510 | #if defined(__GLIBC__) || defined(__EMX__) | |
3511 | if (PL_reentrant_buffer->_crypt_struct_buffer) { | |
3512 | PL_reentrant_buffer->_crypt_struct_buffer->initialized = 0; | |
3513 | /* work around glibc-2.2.5 bug */ | |
3514 | PL_reentrant_buffer->_crypt_struct_buffer->current_saltbits = 0; | |
3515 | } | |
05404ffe | 3516 | #endif |
6ab58e4d | 3517 | } |
05404ffe JH |
3518 | # endif /* HAS_CRYPT_R */ |
3519 | # endif /* USE_ITHREADS */ | |
5f74f29c | 3520 | # ifdef FCRYPT |
83003860 | 3521 | sv_setpv(TARG, fcrypt(tmps, SvPV_nolen_const(right))); |
5f74f29c | 3522 | # else |
83003860 | 3523 | sv_setpv(TARG, PerlProc_crypt(tmps, SvPV_nolen_const(right))); |
5f74f29c | 3524 | # endif |
ec93b65f | 3525 | SETTARG; |
4808266b | 3526 | RETURN; |
79072805 | 3527 | #else |
b13b2135 | 3528 | DIE(aTHX_ |
79072805 | 3529 | "The crypt() function is unimplemented due to excessive paranoia."); |
9e95c635 | 3530 | return NORMAL; |
79072805 | 3531 | #endif |
79072805 LW |
3532 | } |
3533 | ||
00f254e2 KW |
3534 | /* Generally UTF-8 and UTF-EBCDIC are indistinguishable at this level. So |
3535 | * most comments below say UTF-8, when in fact they mean UTF-EBCDIC as well */ | |
3536 | ||
3537 | /* Both the characters below can be stored in two UTF-8 bytes. In UTF-8 the max | |
3538 | * character that 2 bytes can hold is U+07FF, and in UTF-EBCDIC it is U+03FF. | |
3539 | * See http://www.unicode.org/unicode/reports/tr16 */ | |
3540 | #define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */ | |
3541 | #define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */ | |
3542 | ||
3543 | /* Below are several macros that generate code */ | |
3544 | /* Generates code to store a unicode codepoint c that is known to occupy | |
3545 | * exactly two UTF-8 and UTF-EBCDIC bytes; it is stored into p and p+1. */ | |
3546 | #define STORE_UNI_TO_UTF8_TWO_BYTE(p, c) \ | |
3547 | STMT_START { \ | |
3548 | *(p) = UTF8_TWO_BYTE_HI(c); \ | |
3549 | *((p)+1) = UTF8_TWO_BYTE_LO(c); \ | |
3550 | } STMT_END | |
3551 | ||
3552 | /* Like STORE_UNI_TO_UTF8_TWO_BYTE, but advances p to point to the next | |
3553 | * available byte after the two bytes */ | |
3554 | #define CAT_UNI_TO_UTF8_TWO_BYTE(p, c) \ | |
3555 | STMT_START { \ | |
3556 | *(p)++ = UTF8_TWO_BYTE_HI(c); \ | |
3557 | *((p)++) = UTF8_TWO_BYTE_LO(c); \ | |
3558 | } STMT_END | |
3559 | ||
3560 | /* Generates code to store the upper case of latin1 character l which is known | |
3561 | * to have its upper case be non-latin1 into the two bytes p and p+1. There | |
3562 | * are only two characters that fit this description, and this macro knows | |
3563 | * about them, and that the upper case values fit into two UTF-8 or UTF-EBCDIC | |
3564 | * bytes */ | |
3565 | #define STORE_NON_LATIN1_UC(p, l) \ | |
3566 | STMT_START { \ | |
3567 | if ((l) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
3568 | STORE_UNI_TO_UTF8_TWO_BYTE((p), LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS); \ | |
3569 | } else { /* Must be the following letter */ \ | |
3570 | STORE_UNI_TO_UTF8_TWO_BYTE((p), GREEK_CAPITAL_LETTER_MU); \ | |
3571 | } \ | |
3572 | } STMT_END | |
3573 | ||
3574 | /* Like STORE_NON_LATIN1_UC, but advances p to point to the next available byte | |
3575 | * after the character stored */ | |
3576 | #define CAT_NON_LATIN1_UC(p, l) \ | |
3577 | STMT_START { \ | |
3578 | if ((l) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
3579 | CAT_UNI_TO_UTF8_TWO_BYTE((p), LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS); \ | |
3580 | } else { \ | |
3581 | CAT_UNI_TO_UTF8_TWO_BYTE((p), GREEK_CAPITAL_LETTER_MU); \ | |
3582 | } \ | |
3583 | } STMT_END | |
3584 | ||
3585 | /* Generates code to add the two UTF-8 bytes (probably u) that are the upper | |
3586 | * case of l into p and p+1. u must be the result of toUPPER_LATIN1_MOD(l), | |
3587 | * and must require two bytes to store it. Advances p to point to the next | |
3588 | * available position */ | |
3589 | #define CAT_TWO_BYTE_UNI_UPPER_MOD(p, l, u) \ | |
3590 | STMT_START { \ | |
3591 | if ((u) != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
3592 | CAT_UNI_TO_UTF8_TWO_BYTE((p), (u)); /* not special, just save it */ \ | |
3593 | } else if (l == LATIN_SMALL_LETTER_SHARP_S) { \ | |
3594 | *(p)++ = 'S'; *(p)++ = 'S'; /* upper case is 'SS' */ \ | |
3595 | } else {/* else is one of the other two special cases */ \ | |
3596 | CAT_NON_LATIN1_UC((p), (l)); \ | |
3597 | } \ | |
3598 | } STMT_END | |
3599 | ||
79072805 LW |
3600 | PP(pp_ucfirst) |
3601 | { | |
00f254e2 KW |
3602 | /* Actually is both lcfirst() and ucfirst(). Only the first character |
3603 | * changes. This means that possibly we can change in-place, ie., just | |
3604 | * take the source and change that one character and store it back, but not | |
3605 | * if read-only etc, or if the length changes */ | |
3606 | ||
97aff369 | 3607 | dVAR; |
39644a26 | 3608 | dSP; |
d54190f6 | 3609 | SV *source = TOPs; |
00f254e2 | 3610 | STRLEN slen; /* slen is the byte length of the whole SV. */ |
d54190f6 NC |
3611 | STRLEN need; |
3612 | SV *dest; | |
00f254e2 KW |
3613 | bool inplace; /* ? Convert first char only, in-place */ |
3614 | bool doing_utf8 = FALSE; /* ? using utf8 */ | |
3615 | bool convert_source_to_utf8 = FALSE; /* ? need to convert */ | |
12e9c124 | 3616 | const int op_type = PL_op->op_type; |
d54190f6 NC |
3617 | const U8 *s; |
3618 | U8 *d; | |
3619 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
00f254e2 KW |
3620 | STRLEN ulen; /* ulen is the byte length of the original Unicode character |
3621 | * stored as UTF-8 at s. */ | |
3622 | STRLEN tculen; /* tculen is the byte length of the freshly titlecased (or | |
3623 | * lowercased) character stored in tmpbuf. May be either | |
3624 | * UTF-8 or not, but in either case is the number of bytes */ | |
d54190f6 NC |
3625 | |
3626 | SvGETMAGIC(source); | |
3627 | if (SvOK(source)) { | |
3628 | s = (const U8*)SvPV_nomg_const(source, slen); | |
3629 | } else { | |
0a0ffbce RGS |
3630 | if (ckWARN(WARN_UNINITIALIZED)) |
3631 | report_uninit(source); | |
1eced8f8 | 3632 | s = (const U8*)""; |
d54190f6 NC |
3633 | slen = 0; |
3634 | } | |
a0ed51b3 | 3635 | |
00f254e2 KW |
3636 | /* We may be able to get away with changing only the first character, in |
3637 | * place, but not if read-only, etc. Later we may discover more reasons to | |
3638 | * not convert in-place. */ | |
3639 | inplace = SvPADTMP(source) && !SvREADONLY(source) && SvTEMP(source); | |
3640 | ||
3641 | /* First calculate what the changed first character should be. This affects | |
3642 | * whether we can just swap it out, leaving the rest of the string unchanged, | |
3643 | * or even if have to convert the dest to UTF-8 when the source isn't */ | |
3644 | ||
3645 | if (! slen) { /* If empty */ | |
3646 | need = 1; /* still need a trailing NUL */ | |
3647 | } | |
3648 | else if (DO_UTF8(source)) { /* Is the source utf8? */ | |
d54190f6 | 3649 | doing_utf8 = TRUE; |
00f254e2 KW |
3650 | |
3651 | /* TODO: This is #ifdefd out because it has hard-coded the standard mappings, | |
3652 | * and doesn't allow for the user to specify their own. When code is added to | |
3653 | * detect if there is a user-defined mapping in force here, and if so to use | |
3654 | * that, then the code below can be compiled. The detection would be a good | |
3655 | * thing anyway, as currently the user-defined mappings only work on utf8 | |
3656 | * strings, and thus depend on the chosen internal storage method, which is a | |
3657 | * bad thing */ | |
3658 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
3659 | if (UTF8_IS_INVARIANT(*s)) { | |
3660 | ||
3661 | /* An invariant source character is either ASCII or, in EBCDIC, an | |
3662 | * ASCII equivalent or a caseless C1 control. In both these cases, | |
3663 | * the lower and upper cases of any character are also invariants | |
3664 | * (and title case is the same as upper case). So it is safe to | |
3665 | * use the simple case change macros which avoid the overhead of | |
3666 | * the general functions. Note that if perl were to be extended to | |
3667 | * do locale handling in UTF-8 strings, this wouldn't be true in, | |
3668 | * for example, Lithuanian or Turkic. */ | |
3669 | *tmpbuf = (op_type == OP_LCFIRST) ? toLOWER(*s) : toUPPER(*s); | |
3670 | tculen = ulen = 1; | |
3671 | need = slen + 1; | |
12e9c124 | 3672 | } |
00f254e2 KW |
3673 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { |
3674 | U8 chr; | |
3675 | ||
3676 | /* Similarly, if the source character isn't invariant but is in the | |
3677 | * latin1 range (or EBCDIC equivalent thereof), we have the case | |
3678 | * changes compiled into perl, and can avoid the overhead of the | |
3679 | * general functions. In this range, the characters are stored as | |
3680 | * two UTF-8 bytes, and it so happens that any changed-case version | |
3681 | * is also two bytes (in both ASCIIish and EBCDIC machines). */ | |
3682 | tculen = ulen = 2; | |
3683 | need = slen + 1; | |
3684 | ||
3685 | /* Convert the two source bytes to a single Unicode code point | |
3686 | * value, change case and save for below */ | |
3687 | chr = UTF8_ACCUMULATE(*s, *(s+1)); | |
3688 | if (op_type == OP_LCFIRST) { /* lower casing is easy */ | |
3689 | U8 lower = toLOWER_LATIN1(chr); | |
3690 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, lower); | |
3691 | } | |
3692 | else { /* ucfirst */ | |
3693 | U8 upper = toUPPER_LATIN1_MOD(chr); | |
3694 | ||
3695 | /* Most of the latin1 range characters are well-behaved. Their | |
3696 | * title and upper cases are the same, and are also in the | |
3697 | * latin1 range. The macro above returns their upper (hence | |
3698 | * title) case, and all that need be done is to save the result | |
3699 | * for below. However, several characters are problematic, and | |
3700 | * have to be handled specially. The MOD in the macro name | |
3701 | * above means that these tricky characters all get mapped to | |
3702 | * the single character LATIN_SMALL_LETTER_Y_WITH_DIAERESIS. | |
3703 | * This mapping saves some tests for the majority of the | |
3704 | * characters */ | |
3705 | ||
3706 | if (upper != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { | |
3707 | ||
3708 | /* Not tricky. Just save it. */ | |
3709 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, upper); | |
3710 | } | |
3711 | else if (chr == LATIN_SMALL_LETTER_SHARP_S) { | |
3712 | ||
3713 | /* This one is tricky because it is two characters long, | |
3714 | * though the UTF-8 is still two bytes, so the stored | |
3715 | * length doesn't change */ | |
3716 | *tmpbuf = 'S'; /* The UTF-8 is 'Ss' */ | |
3717 | *(tmpbuf + 1) = 's'; | |
3718 | } | |
3719 | else { | |
3720 | ||
3721 | /* The other two have their title and upper cases the same, | |
3722 | * but are tricky because the changed-case characters | |
3723 | * aren't in the latin1 range. They, however, do fit into | |
3724 | * two UTF-8 bytes */ | |
3725 | STORE_NON_LATIN1_UC(tmpbuf, chr); | |
3726 | } | |
3727 | } | |
3728 | } | |
3729 | else { | |
3730 | #endif /* end of dont want to break user-defined casing */ | |
3731 | ||
3732 | /* Here, can't short-cut the general case */ | |
3733 | ||
3734 | utf8_to_uvchr(s, &ulen); | |
3735 | if (op_type == OP_UCFIRST) toTITLE_utf8(s, tmpbuf, &tculen); | |
3736 | else toLOWER_utf8(s, tmpbuf, &tculen); | |
3737 | ||
3738 | /* we can't do in-place if the length changes. */ | |
3739 | if (ulen != tculen) inplace = FALSE; | |
3740 | need = slen + 1 - ulen + tculen; | |
3741 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
3742 | } | |
3743 | #endif | |
d54190f6 | 3744 | } |
00f254e2 KW |
3745 | else { /* Non-zero length, non-UTF-8, Need to consider locale and if |
3746 | * latin1 is treated as caseless. Note that a locale takes | |
3747 | * precedence */ | |
3748 | tculen = 1; /* Most characters will require one byte, but this will | |
3749 | * need to be overridden for the tricky ones */ | |
3750 | need = slen + 1; | |
3751 | ||
3752 | if (op_type == OP_LCFIRST) { | |
d54190f6 | 3753 | |
00f254e2 KW |
3754 | /* lower case the first letter: no trickiness for any character */ |
3755 | *tmpbuf = (IN_LOCALE_RUNTIME) ? toLOWER_LC(*s) : | |
3756 | ((IN_UNI_8_BIT) ? toLOWER_LATIN1(*s) : toLOWER(*s)); | |
3757 | } | |
3758 | /* is ucfirst() */ | |
3759 | else if (IN_LOCALE_RUNTIME) { | |
3760 | *tmpbuf = toUPPER_LC(*s); /* This would be a bug if any locales | |
3761 | * have upper and title case different | |
3762 | */ | |
3763 | } | |
3764 | else if (! IN_UNI_8_BIT) { | |
3765 | *tmpbuf = toUPPER(*s); /* Returns caseless for non-ascii, or | |
3766 | * on EBCDIC machines whatever the | |
3767 | * native function does */ | |
3768 | } | |
3769 | else { /* is ucfirst non-UTF-8, not in locale, and cased latin1 */ | |
3770 | *tmpbuf = toUPPER_LATIN1_MOD(*s); | |
3771 | ||
3772 | /* tmpbuf now has the correct title case for all latin1 characters | |
3773 | * except for the several ones that have tricky handling. All | |
3774 | * of these are mapped by the MOD to the letter below. */ | |
3775 | if (*tmpbuf == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { | |
3776 | ||
3777 | /* The length is going to change, with all three of these, so | |
3778 | * can't replace just the first character */ | |
3779 | inplace = FALSE; | |
3780 | ||
3781 | /* We use the original to distinguish between these tricky | |
3782 | * cases */ | |
3783 | if (*s == LATIN_SMALL_LETTER_SHARP_S) { | |
3784 | /* Two character title case 'Ss', but can remain non-UTF-8 */ | |
3785 | need = slen + 2; | |
3786 | *tmpbuf = 'S'; | |
3787 | *(tmpbuf + 1) = 's'; /* Assert: length(tmpbuf) >= 2 */ | |
3788 | tculen = 2; | |
3789 | } | |
3790 | else { | |
d54190f6 | 3791 | |
00f254e2 KW |
3792 | /* The other two tricky ones have their title case outside |
3793 | * latin1. It is the same as their upper case. */ | |
3794 | doing_utf8 = TRUE; | |
3795 | STORE_NON_LATIN1_UC(tmpbuf, *s); | |
3796 | ||
3797 | /* The UTF-8 and UTF-EBCDIC lengths of both these characters | |
3798 | * and their upper cases is 2. */ | |
3799 | tculen = ulen = 2; | |
3800 | ||
3801 | /* The entire result will have to be in UTF-8. Assume worst | |
3802 | * case sizing in conversion. (all latin1 characters occupy | |
3803 | * at most two bytes in utf8) */ | |
3804 | convert_source_to_utf8 = TRUE; | |
3805 | need = slen * 2 + 1; | |
3806 | } | |
3807 | } /* End of is one of the three special chars */ | |
3808 | } /* End of use Unicode (Latin1) semantics */ | |
3809 | } /* End of changing the case of the first character */ | |
3810 | ||
3811 | /* Here, have the first character's changed case stored in tmpbuf. Ready to | |
3812 | * generate the result */ | |
3813 | if (inplace) { | |
3814 | ||
3815 | /* We can convert in place. This means we change just the first | |
3816 | * character without disturbing the rest; no need to grow */ | |
d54190f6 NC |
3817 | dest = source; |
3818 | s = d = (U8*)SvPV_force_nomg(source, slen); | |
3819 | } else { | |
3820 | dTARGET; | |
3821 | ||
3822 | dest = TARG; | |
3823 | ||
00f254e2 KW |
3824 | /* Here, we can't convert in place; we earlier calculated how much |
3825 | * space we will need, so grow to accommodate that */ | |
d54190f6 | 3826 | SvUPGRADE(dest, SVt_PV); |
3b416f41 | 3827 | d = (U8*)SvGROW(dest, need); |
d54190f6 NC |
3828 | (void)SvPOK_only(dest); |
3829 | ||
3830 | SETs(dest); | |
d54190f6 | 3831 | } |
44bc797b | 3832 | |
d54190f6 | 3833 | if (doing_utf8) { |
00f254e2 KW |
3834 | if (! inplace) { |
3835 | if (! convert_source_to_utf8) { | |
3836 | ||
3837 | /* Here both source and dest are in UTF-8, but have to create | |
3838 | * the entire output. We initialize the result to be the | |
3839 | * title/lower cased first character, and then append the rest | |
3840 | * of the string. */ | |
3841 | sv_setpvn(dest, (char*)tmpbuf, tculen); | |
3842 | if (slen > ulen) { | |
3843 | sv_catpvn(dest, (char*)(s + ulen), slen - ulen); | |
3844 | } | |
3845 | } | |
3846 | else { | |
3847 | const U8 *const send = s + slen; | |
3848 | ||
3849 | /* Here the dest needs to be in UTF-8, but the source isn't, | |
3850 | * except we earlier UTF-8'd the first character of the source | |
3851 | * into tmpbuf. First put that into dest, and then append the | |
3852 | * rest of the source, converting it to UTF-8 as we go. */ | |
3853 | ||
3854 | /* Assert tculen is 2 here because the only two characters that | |
3855 | * get to this part of the code have 2-byte UTF-8 equivalents */ | |
3856 | *d++ = *tmpbuf; | |
3857 | *d++ = *(tmpbuf + 1); | |
3858 | s++; /* We have just processed the 1st char */ | |
3859 | ||
3860 | for (; s < send; s++) { | |
3861 | d = uvchr_to_utf8(d, *s); | |
3862 | } | |
3863 | *d = '\0'; | |
3864 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
3865 | } | |
d54190f6 | 3866 | SvUTF8_on(dest); |
a0ed51b3 | 3867 | } |
00f254e2 | 3868 | else { /* in-place UTF-8. Just overwrite the first character */ |
d54190f6 NC |
3869 | Copy(tmpbuf, d, tculen, U8); |
3870 | SvCUR_set(dest, need - 1); | |
a0ed51b3 | 3871 | } |
a0ed51b3 | 3872 | } |
00f254e2 KW |
3873 | else { /* Neither source nor dest are in or need to be UTF-8 */ |
3874 | if (slen) { | |
2de3dbcc | 3875 | if (IN_LOCALE_RUNTIME) { |
31351b04 | 3876 | TAINT; |
d54190f6 | 3877 | SvTAINTED_on(dest); |
31351b04 | 3878 | } |
00f254e2 KW |
3879 | if (inplace) { /* in-place, only need to change the 1st char */ |
3880 | *d = *tmpbuf; | |
3881 | } | |
3882 | else { /* Not in-place */ | |
3883 | ||
3884 | /* Copy the case-changed character(s) from tmpbuf */ | |
3885 | Copy(tmpbuf, d, tculen, U8); | |
3886 | d += tculen - 1; /* Code below expects d to point to final | |
3887 | * character stored */ | |
3888 | } | |
3889 | } | |
3890 | else { /* empty source */ | |
3891 | /* See bug #39028: Don't taint if empty */ | |
d54190f6 NC |
3892 | *d = *s; |
3893 | } | |
3894 | ||
00f254e2 KW |
3895 | /* In a "use bytes" we don't treat the source as UTF-8, but, still want |
3896 | * the destination to retain that flag */ | |
d54190f6 NC |
3897 | if (SvUTF8(source)) |
3898 | SvUTF8_on(dest); | |
3899 | ||
00f254e2 | 3900 | if (!inplace) { /* Finish the rest of the string, unchanged */ |
d54190f6 NC |
3901 | /* This will copy the trailing NUL */ |
3902 | Copy(s + 1, d + 1, slen, U8); | |
3903 | SvCUR_set(dest, need - 1); | |
bbce6d69 | 3904 | } |
bbce6d69 | 3905 | } |
d54190f6 | 3906 | SvSETMAGIC(dest); |
79072805 LW |
3907 | RETURN; |
3908 | } | |
3909 | ||
67306194 NC |
3910 | /* There's so much setup/teardown code common between uc and lc, I wonder if |
3911 | it would be worth merging the two, and just having a switch outside each | |
00f254e2 | 3912 | of the three tight loops. There is less and less commonality though */ |
79072805 LW |
3913 | PP(pp_uc) |
3914 | { | |
97aff369 | 3915 | dVAR; |
39644a26 | 3916 | dSP; |
67306194 | 3917 | SV *source = TOPs; |
463ee0b2 | 3918 | STRLEN len; |
67306194 NC |
3919 | STRLEN min; |
3920 | SV *dest; | |
3921 | const U8 *s; | |
3922 | U8 *d; | |
79072805 | 3923 | |
67306194 NC |
3924 | SvGETMAGIC(source); |
3925 | ||
3926 | if (SvPADTMP(source) && !SvREADONLY(source) && !SvAMAGIC(source) | |
00f254e2 KW |
3927 | && SvTEMP(source) && !DO_UTF8(source) |
3928 | && (IN_LOCALE_RUNTIME || ! IN_UNI_8_BIT)) { | |
3929 | ||
3930 | /* We can convert in place. The reason we can't if in UNI_8_BIT is to | |
3931 | * make the loop tight, so we overwrite the source with the dest before | |
3932 | * looking at it, and we need to look at the original source | |
3933 | * afterwards. There would also need to be code added to handle | |
3934 | * switching to not in-place in midstream if we run into characters | |
3935 | * that change the length. | |
3936 | */ | |
67306194 NC |
3937 | dest = source; |
3938 | s = d = (U8*)SvPV_force_nomg(source, len); | |
3939 | min = len + 1; | |
3940 | } else { | |
a0ed51b3 | 3941 | dTARGET; |
a0ed51b3 | 3942 | |
67306194 | 3943 | dest = TARG; |
128c9517 | 3944 | |
67306194 NC |
3945 | /* The old implementation would copy source into TARG at this point. |
3946 | This had the side effect that if source was undef, TARG was now | |
3947 | an undefined SV with PADTMP set, and they don't warn inside | |
3948 | sv_2pv_flags(). However, we're now getting the PV direct from | |
3949 | source, which doesn't have PADTMP set, so it would warn. Hence the | |
3950 | little games. */ | |
3951 | ||
3952 | if (SvOK(source)) { | |
3953 | s = (const U8*)SvPV_nomg_const(source, len); | |
3954 | } else { | |
0a0ffbce RGS |
3955 | if (ckWARN(WARN_UNINITIALIZED)) |
3956 | report_uninit(source); | |
1eced8f8 | 3957 | s = (const U8*)""; |
67306194 | 3958 | len = 0; |
a0ed51b3 | 3959 | } |
67306194 NC |
3960 | min = len + 1; |
3961 | ||
3962 | SvUPGRADE(dest, SVt_PV); | |
3b416f41 | 3963 | d = (U8*)SvGROW(dest, min); |
67306194 NC |
3964 | (void)SvPOK_only(dest); |
3965 | ||
3966 | SETs(dest); | |
a0ed51b3 | 3967 | } |
31351b04 | 3968 | |
67306194 NC |
3969 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need |
3970 | to check DO_UTF8 again here. */ | |
3971 | ||
3972 | if (DO_UTF8(source)) { | |
3973 | const U8 *const send = s + len; | |
3974 | U8 tmpbuf[UTF8_MAXBYTES+1]; | |
3975 | ||
00f254e2 KW |
3976 | /* This is ifdefd out because it needs more work and thought. It isn't clear |
3977 | * that we should do it. These are hard-coded rules from the Unicode standard, | |
3978 | * and may change. 5.2 gives new guidance on the iota subscript, for example, | |
3979 | * which has not been checked against this; and secondly it may be that we are | |
3980 | * passed a subset of the context, via a \U...\E, for example, and its not | |
3981 | * clear what the best approach is to that */ | |
3982 | #ifdef CONTEXT_DEPENDENT_CASING | |
3983 | bool in_iota_subscript = FALSE; | |
3984 | #endif | |
3985 | ||
67306194 | 3986 | while (s < send) { |
00f254e2 KW |
3987 | #ifdef CONTEXT_DEPENDENT_CASING |
3988 | if (in_iota_subscript && ! is_utf8_mark(s)) { | |
3989 | /* A non-mark. Time to output the iota subscript */ | |
3990 | #define GREEK_CAPITAL_LETTER_IOTA 0x0399 | |
3991 | #define COMBINING_GREEK_YPOGEGRAMMENI 0x0345 | |
3992 | ||
3993 | CAT_UNI_TO_UTF8_TWO_BYTE(d, GREEK_CAPITAL_LETTER_IOTA); | |
3994 | in_iota_subscript = FALSE; | |
3995 | } | |
3996 | #endif | |
3997 | ||
3998 | ||
3999 | /* See comments at the first instance in this file of this ifdef */ | |
4000 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
67306194 | 4001 | |
00f254e2 KW |
4002 | /* If the UTF-8 character is invariant, then it is in the range |
4003 | * known by the standard macro; result is only one byte long */ | |
4004 | if (UTF8_IS_INVARIANT(*s)) { | |
4005 | *d++ = toUPPER(*s); | |
4006 | s++; | |
4007 | } | |
4008 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { | |
4009 | ||
4010 | /* Likewise, if it fits in a byte, its case change is in our | |
4011 | * table */ | |
4012 | U8 orig = UTF8_ACCUMULATE(*s, *(s+1)); | |
4013 | U8 upper = toUPPER_LATIN1_MOD(orig); | |
4014 | CAT_TWO_BYTE_UNI_UPPER_MOD(d, orig, upper); | |
4015 | s += 2; | |
4016 | } | |
4017 | else { | |
4018 | #else | |
4019 | { | |
4020 | #endif | |
4021 | ||
4022 | /* Otherwise, need the general UTF-8 case. Get the changed | |
4023 | * case value and copy it to the output buffer */ | |
4024 | ||
4025 | const STRLEN u = UTF8SKIP(s); | |
4026 | STRLEN ulen; | |
67306194 | 4027 | |
00f254e2 KW |
4028 | #ifndef CONTEXT_DEPENDENT_CASING |
4029 | toUPPER_utf8(s, tmpbuf, &ulen); | |
4030 | #else | |
4031 | const UV uv = toUPPER_utf8(s, tmpbuf, &ulen); | |
4032 | if (uv == GREEK_CAPITAL_LETTER_IOTA && utf8_to_uvchr(s, 0) == COMBINING_GREEK_YPOGEGRAMMENI) { | |
4033 | in_iota_subscript = TRUE; | |
4034 | } | |
4035 | else { | |
4036 | #endif | |
4037 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { | |
4038 | /* If the eventually required minimum size outgrows | |
4039 | * the available space, we need to grow. */ | |
4040 | const UV o = d - (U8*)SvPVX_const(dest); | |
4041 | ||
4042 | /* If someone uppercases one million U+03B0s we | |
4043 | * SvGROW() one million times. Or we could try | |
4044 | * guessing how much to allocate without allocating too | |
4045 | * much. Such is life. See corresponding comment in lc code | |
4046 | * for another option */ | |
4047 | SvGROW(dest, min); | |
4048 | d = (U8*)SvPVX(dest) + o; | |
4049 | } | |
4050 | Copy(tmpbuf, d, ulen, U8); | |
4051 | d += ulen; | |
4052 | #ifdef CONTEXT_DEPENDENT_CASING | |
4053 | } | |
4054 | #endif | |
4055 | s += u; | |
67306194 | 4056 | } |
67306194 | 4057 | } |
00f254e2 KW |
4058 | #ifdef CONTEXT_DEPENDENT_CASING |
4059 | if (in_iota_subscript) CAT_UNI_TO_UTF8_TWO_BYTE(d, GREEK_CAPITAL_LETTER_IOTA); | |
4060 | #endif | |
67306194 NC |
4061 | SvUTF8_on(dest); |
4062 | *d = '\0'; | |
4063 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
00f254e2 | 4064 | } else { /* Not UTF-8 */ |
67306194 NC |
4065 | if (len) { |
4066 | const U8 *const send = s + len; | |
00f254e2 KW |
4067 | |
4068 | /* Use locale casing if in locale; regular style if not treating | |
4069 | * latin1 as having case; otherwise the latin1 casing. Do the | |
4070 | * whole thing in a tight loop, for speed, */ | |
2de3dbcc | 4071 | if (IN_LOCALE_RUNTIME) { |
31351b04 | 4072 | TAINT; |
67306194 NC |
4073 | SvTAINTED_on(dest); |
4074 | for (; s < send; d++, s++) | |
4075 | *d = toUPPER_LC(*s); | |
31351b04 | 4076 | } |
00f254e2 KW |
4077 | else if (! IN_UNI_8_BIT) { |
4078 | for (; s < send; d++, s++) { | |
67306194 | 4079 | *d = toUPPER(*s); |
00f254e2 | 4080 | } |
31351b04 | 4081 | } |
00f254e2 KW |
4082 | else { |
4083 | for (; s < send; d++, s++) { | |
4084 | *d = toUPPER_LATIN1_MOD(*s); | |
4085 | if (*d != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) continue; | |
4086 | ||
4087 | /* The mainstream case is the tight loop above. To avoid | |
4088 | * extra tests in that, all three characters that require | |
4089 | * special handling are mapped by the MOD to the one tested | |
4090 | * just above. | |
4091 | * Use the source to distinguish between the three cases */ | |
4092 | ||
4093 | if (*s == LATIN_SMALL_LETTER_SHARP_S) { | |
4094 | ||
4095 | /* uc() of this requires 2 characters, but they are | |
4096 | * ASCII. If not enough room, grow the string */ | |
4097 | if (SvLEN(dest) < ++min) { | |
4098 | const UV o = d - (U8*)SvPVX_const(dest); | |
4099 | SvGROW(dest, min); | |
4100 | d = (U8*)SvPVX(dest) + o; | |
4101 | } | |
4102 | *d++ = 'S'; *d = 'S'; /* upper case is 'SS' */ | |
4103 | continue; /* Back to the tight loop; still in ASCII */ | |
4104 | } | |
4105 | ||
4106 | /* The other two special handling characters have their | |
4107 | * upper cases outside the latin1 range, hence need to be | |
4108 | * in UTF-8, so the whole result needs to be in UTF-8. So, | |
4109 | * here we are somewhere in the middle of processing a | |
4110 | * non-UTF-8 string, and realize that we will have to convert | |
4111 | * the whole thing to UTF-8. What to do? There are | |
4112 | * several possibilities. The simplest to code is to | |
4113 | * convert what we have so far, set a flag, and continue on | |
4114 | * in the loop. The flag would be tested each time through | |
4115 | * the loop, and if set, the next character would be | |
4116 | * converted to UTF-8 and stored. But, I (khw) didn't want | |
4117 | * to slow down the mainstream case at all for this fairly | |
4118 | * rare case, so I didn't want to add a test that didn't | |
4119 | * absolutely have to be there in the loop, besides the | |
4120 | * possibility that it would get too complicated for | |
4121 | * optimizers to deal with. Another possibility is to just | |
4122 | * give up, convert the source to UTF-8, and restart the | |
4123 | * function that way. Another possibility is to convert | |
4124 | * both what has already been processed and what is yet to | |
4125 | * come separately to UTF-8, then jump into the loop that | |
4126 | * handles UTF-8. But the most efficient time-wise of the | |
4127 | * ones I could think of is what follows, and turned out to | |
4128 | * not require much extra code. */ | |
4129 | ||
4130 | /* Convert what we have so far into UTF-8, telling the | |
4131 | * function that we know it should be converted, and to | |
4132 | * allow extra space for what we haven't processed yet. | |
4133 | * Assume the worst case space requirements for converting | |
4134 | * what we haven't processed so far: that it will require | |
4135 | * two bytes for each remaining source character, plus the | |
4136 | * NUL at the end. This may cause the string pointer to | |
4137 | * move, so re-find it. */ | |
4138 | ||
4139 | len = d - (U8*)SvPVX_const(dest); | |
4140 | SvCUR_set(dest, len); | |
4141 | len = sv_utf8_upgrade_flags_grow(dest, | |
4142 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
4143 | (send -s) * 2 + 1); | |
4144 | d = (U8*)SvPVX(dest) + len; | |
4145 | ||
4146 | /* And append the current character's upper case in UTF-8 */ | |
4147 | CAT_NON_LATIN1_UC(d, *s); | |
4148 | ||
4149 | /* Now process the remainder of the source, converting to | |
4150 | * upper and UTF-8. If a resulting byte is invariant in | |
4151 | * UTF-8, output it as-is, otherwise convert to UTF-8 and | |
4152 | * append it to the output. */ | |
4153 | ||
4154 | s++; | |
4155 | for (; s < send; s++) { | |
4156 | U8 upper = toUPPER_LATIN1_MOD(*s); | |
4157 | if UTF8_IS_INVARIANT(upper) { | |
4158 | *d++ = upper; | |
4159 | } | |
4160 | else { | |
4161 | CAT_TWO_BYTE_UNI_UPPER_MOD(d, *s, upper); | |
4162 | } | |
4163 | } | |
4164 | ||
4165 | /* Here have processed the whole source; no need to continue | |
4166 | * with the outer loop. Each character has been converted | |
4167 | * to upper case and converted to UTF-8 */ | |
4168 | ||
4169 | break; | |
4170 | } /* End of processing all latin1-style chars */ | |
4171 | } /* End of processing all chars */ | |
4172 | } /* End of source is not empty */ | |
4173 | ||
67306194 | 4174 | if (source != dest) { |
00f254e2 | 4175 | *d = '\0'; /* Here d points to 1 after last char, add NUL */ |
67306194 NC |
4176 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); |
4177 | } | |
00f254e2 | 4178 | } /* End of isn't utf8 */ |
67306194 | 4179 | SvSETMAGIC(dest); |
79072805 LW |
4180 | RETURN; |
4181 | } | |
4182 | ||
4183 | PP(pp_lc) | |
4184 | { | |
97aff369 | 4185 | dVAR; |
39644a26 | 4186 | dSP; |
ec9af7d4 | 4187 | SV *source = TOPs; |
463ee0b2 | 4188 | STRLEN len; |
ec9af7d4 NC |
4189 | STRLEN min; |
4190 | SV *dest; | |
4191 | const U8 *s; | |
4192 | U8 *d; | |
79072805 | 4193 | |
ec9af7d4 NC |
4194 | SvGETMAGIC(source); |
4195 | ||
4196 | if (SvPADTMP(source) && !SvREADONLY(source) && !SvAMAGIC(source) | |
17fa0776 | 4197 | && SvTEMP(source) && !DO_UTF8(source)) { |
ec9af7d4 | 4198 | |
00f254e2 KW |
4199 | /* We can convert in place, as lowercasing anything in the latin1 range |
4200 | * (or else DO_UTF8 would have been on) doesn't lengthen it */ | |
ec9af7d4 NC |
4201 | dest = source; |
4202 | s = d = (U8*)SvPV_force_nomg(source, len); | |
4203 | min = len + 1; | |
4204 | } else { | |
a0ed51b3 | 4205 | dTARGET; |
a0ed51b3 | 4206 | |
ec9af7d4 NC |
4207 | dest = TARG; |
4208 | ||
4209 | /* The old implementation would copy source into TARG at this point. | |
4210 | This had the side effect that if source was undef, TARG was now | |
4211 | an undefined SV with PADTMP set, and they don't warn inside | |
4212 | sv_2pv_flags(). However, we're now getting the PV direct from | |
4213 | source, which doesn't have PADTMP set, so it would warn. Hence the | |
4214 | little games. */ | |
4215 | ||
4216 | if (SvOK(source)) { | |
4217 | s = (const U8*)SvPV_nomg_const(source, len); | |
4218 | } else { | |
0a0ffbce RGS |
4219 | if (ckWARN(WARN_UNINITIALIZED)) |
4220 | report_uninit(source); | |
1eced8f8 | 4221 | s = (const U8*)""; |
ec9af7d4 | 4222 | len = 0; |
a0ed51b3 | 4223 | } |
ec9af7d4 | 4224 | min = len + 1; |
128c9517 | 4225 | |
ec9af7d4 | 4226 | SvUPGRADE(dest, SVt_PV); |
3b416f41 | 4227 | d = (U8*)SvGROW(dest, min); |
ec9af7d4 NC |
4228 | (void)SvPOK_only(dest); |
4229 | ||
4230 | SETs(dest); | |
4231 | } | |
4232 | ||
4233 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need | |
4234 | to check DO_UTF8 again here. */ | |
4235 | ||
4236 | if (DO_UTF8(source)) { | |
4237 | const U8 *const send = s + len; | |
4238 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
4239 | ||
4240 | while (s < send) { | |
00f254e2 KW |
4241 | /* See comments at the first instance in this file of this ifdef */ |
4242 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
4243 | if (UTF8_IS_INVARIANT(*s)) { | |
89ebb4a3 | 4244 | |
00f254e2 | 4245 | /* Invariant characters use the standard mappings compiled in. |
ec9af7d4 | 4246 | */ |
00f254e2 KW |
4247 | *d++ = toLOWER(*s); |
4248 | s++; | |
ec9af7d4 | 4249 | } |
00f254e2 | 4250 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { |
89ebb4a3 | 4251 | |
00f254e2 KW |
4252 | /* As do the ones in the Latin1 range */ |
4253 | U8 lower = toLOWER_LATIN1(UTF8_ACCUMULATE(*s, *(s+1))); | |
4254 | CAT_UNI_TO_UTF8_TWO_BYTE(d, lower); | |
4255 | s += 2; | |
a0ed51b3 | 4256 | } |
00f254e2 KW |
4257 | else { |
4258 | #endif | |
4259 | /* Here, is utf8 not in Latin-1 range, have to go out and get | |
4260 | * the mappings from the tables. */ | |
4261 | ||
4262 | const STRLEN u = UTF8SKIP(s); | |
4263 | STRLEN ulen; | |
4264 | ||
4265 | /* See comments at the first instance in this file of this ifdef */ | |
4266 | #ifndef CONTEXT_DEPENDENT_CASING | |
4267 | toLOWER_utf8(s, tmpbuf, &ulen); | |
4268 | #else | |
4269 | /* Here is context dependent casing, not compiled in currently; | |
4270 | * needs more thought and work */ | |
4271 | ||
4272 | const UV uv = toLOWER_utf8(s, tmpbuf, &ulen); | |
4273 | ||
4274 | /* If the lower case is a small sigma, it may be that we need | |
4275 | * to change it to a final sigma. This happens at the end of | |
4276 | * a word that contains more than just this character, and only | |
4277 | * when we started with a capital sigma. */ | |
4278 | if (uv == UNICODE_GREEK_SMALL_LETTER_SIGMA && | |
4279 | s > send - len && /* Makes sure not the first letter */ | |
4280 | utf8_to_uvchr(s, 0) == UNICODE_GREEK_CAPITAL_LETTER_SIGMA | |
4281 | ) { | |
4282 | ||
4283 | /* We use the algorithm in: | |
4284 | * http://www.unicode.org/versions/Unicode5.0.0/ch03.pdf (C | |
4285 | * is a CAPITAL SIGMA): If C is preceded by a sequence | |
4286 | * consisting of a cased letter and a case-ignorable | |
4287 | * sequence, and C is not followed by a sequence consisting | |
4288 | * of a case ignorable sequence and then a cased letter, | |
4289 | * then when lowercasing C, C becomes a final sigma */ | |
4290 | ||
4291 | /* To determine if this is the end of a word, need to peek | |
4292 | * ahead. Look at the next character */ | |
4293 | const U8 *peek = s + u; | |
4294 | ||
4295 | /* Skip any case ignorable characters */ | |
4296 | while (peek < send && is_utf8_case_ignorable(peek)) { | |
4297 | peek += UTF8SKIP(peek); | |
4298 | } | |
4299 | ||
4300 | /* If we reached the end of the string without finding any | |
4301 | * non-case ignorable characters, or if the next such one | |
4302 | * is not-cased, then we have met the conditions for it | |
4303 | * being a final sigma with regards to peek ahead, and so | |
4304 | * must do peek behind for the remaining conditions. (We | |
4305 | * know there is stuff behind to look at since we tested | |
4306 | * above that this isn't the first letter) */ | |
4307 | if (peek >= send || ! is_utf8_cased(peek)) { | |
4308 | peek = utf8_hop(s, -1); | |
4309 | ||
4310 | /* Here are at the beginning of the first character | |
4311 | * before the original upper case sigma. Keep backing | |
4312 | * up, skipping any case ignorable characters */ | |
4313 | while (is_utf8_case_ignorable(peek)) { | |
4314 | peek = utf8_hop(peek, -1); | |
4315 | } | |
4316 | ||
4317 | /* Here peek points to the first byte of the closest | |
4318 | * non-case-ignorable character before the capital | |
4319 | * sigma. If it is cased, then by the Unicode | |
4320 | * algorithm, we should use a small final sigma instead | |
4321 | * of what we have */ | |
4322 | if (is_utf8_cased(peek)) { | |
4323 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, | |
4324 | UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA); | |
4325 | } | |
4326 | } | |
4327 | } | |
4328 | else { /* Not a context sensitive mapping */ | |
4329 | #endif /* End of commented out context sensitive */ | |
4330 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { | |
4331 | ||
4332 | /* If the eventually required minimum size outgrows | |
4333 | * the available space, we need to grow. */ | |
4334 | const UV o = d - (U8*)SvPVX_const(dest); | |
4335 | ||
4336 | /* If someone lowercases one million U+0130s we | |
4337 | * SvGROW() one million times. Or we could try | |
4338 | * guessing how much to allocate without allocating too | |
4339 | * much. Such is life. Another option would be to | |
4340 | * grow an extra byte or two more each time we need to | |
4341 | * grow, which would cut down the million to 500K, with | |
4342 | * little waste */ | |
4343 | SvGROW(dest, min); | |
4344 | d = (U8*)SvPVX(dest) + o; | |
4345 | } | |
4346 | #ifdef CONTEXT_DEPENDENT_CASING | |
4347 | } | |
4348 | #endif | |
4349 | /* Copy the newly lowercased letter to the output buffer we're | |
4350 | * building */ | |
4351 | Copy(tmpbuf, d, ulen, U8); | |
4352 | d += ulen; | |
4353 | s += u; | |
4354 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
4355 | } | |
4356 | #endif | |
4357 | } /* End of looping through the source string */ | |
ec9af7d4 NC |
4358 | SvUTF8_on(dest); |
4359 | *d = '\0'; | |
4360 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
00f254e2 | 4361 | } else { /* Not utf8 */ |
31351b04 | 4362 | if (len) { |
ec9af7d4 | 4363 | const U8 *const send = s + len; |
00f254e2 KW |
4364 | |
4365 | /* Use locale casing if in locale; regular style if not treating | |
4366 | * latin1 as having case; otherwise the latin1 casing. Do the | |
4367 | * whole thing in a tight loop, for speed, */ | |
2de3dbcc | 4368 | if (IN_LOCALE_RUNTIME) { |
31351b04 | 4369 | TAINT; |
ec9af7d4 NC |
4370 | SvTAINTED_on(dest); |
4371 | for (; s < send; d++, s++) | |
4372 | *d = toLOWER_LC(*s); | |
31351b04 | 4373 | } |
00f254e2 KW |
4374 | else if (! IN_UNI_8_BIT) { |
4375 | for (; s < send; d++, s++) { | |
ec9af7d4 | 4376 | *d = toLOWER(*s); |
00f254e2 KW |
4377 | } |
4378 | } | |
4379 | else { | |
4380 | for (; s < send; d++, s++) { | |
4381 | *d = toLOWER_LATIN1(*s); | |
4382 | } | |
31351b04 | 4383 | } |
bbce6d69 | 4384 | } |
ec9af7d4 NC |
4385 | if (source != dest) { |
4386 | *d = '\0'; | |
4387 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
4388 | } | |
79072805 | 4389 | } |
ec9af7d4 | 4390 | SvSETMAGIC(dest); |
79072805 LW |
4391 | RETURN; |
4392 | } | |
4393 | ||
a0d0e21e | 4394 | PP(pp_quotemeta) |
79072805 | 4395 | { |
97aff369 | 4396 | dVAR; dSP; dTARGET; |
1b6737cc | 4397 | SV * const sv = TOPs; |
a0d0e21e | 4398 | STRLEN len; |
0d46e09a | 4399 | register const char *s = SvPV_const(sv,len); |
79072805 | 4400 | |
7e2040f0 | 4401 | SvUTF8_off(TARG); /* decontaminate */ |
a0d0e21e | 4402 | if (len) { |
1b6737cc | 4403 | register char *d; |
862a34c6 | 4404 | SvUPGRADE(TARG, SVt_PV); |
c07a80fd | 4405 | SvGROW(TARG, (len * 2) + 1); |
a0d0e21e | 4406 | d = SvPVX(TARG); |
7e2040f0 | 4407 | if (DO_UTF8(sv)) { |
0dd2cdef | 4408 | while (len) { |
fd400ab9 | 4409 | if (UTF8_IS_CONTINUED(*s)) { |
0dd2cdef LW |
4410 | STRLEN ulen = UTF8SKIP(s); |
4411 | if (ulen > len) | |
4412 | ulen = len; | |
4413 | len -= ulen; | |
4414 | while (ulen--) | |
4415 | *d++ = *s++; | |
4416 | } | |
4417 | else { | |
4418 | if (!isALNUM(*s)) | |
4419 | *d++ = '\\'; | |
4420 | *d++ = *s++; | |
4421 | len--; | |
4422 | } | |
4423 | } | |
7e2040f0 | 4424 | SvUTF8_on(TARG); |
0dd2cdef LW |
4425 | } |
4426 | else { | |
4427 | while (len--) { | |
4428 | if (!isALNUM(*s)) | |
4429 | *d++ = '\\'; | |
4430 | *d++ = *s++; | |
4431 | } | |
79072805 | 4432 | } |
a0d0e21e | 4433 | *d = '\0'; |
349d4f2f | 4434 | SvCUR_set(TARG, d - SvPVX_const(TARG)); |
3aa33fe5 | 4435 | (void)SvPOK_only_UTF8(TARG); |
79072805 | 4436 | } |
a0d0e21e LW |
4437 | else |
4438 | sv_setpvn(TARG, s, len); | |
ec93b65f | 4439 | SETTARG; |
79072805 LW |
4440 | RETURN; |
4441 | } | |
4442 | ||
a0d0e21e | 4443 | /* Arrays. */ |
79072805 | 4444 | |
a0d0e21e | 4445 | PP(pp_aslice) |
79072805 | 4446 | { |
97aff369 | 4447 | dVAR; dSP; dMARK; dORIGMARK; |
502c6561 | 4448 | register AV *const av = MUTABLE_AV(POPs); |
1b6737cc | 4449 | register const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); |
79072805 | 4450 | |
a0d0e21e | 4451 | if (SvTYPE(av) == SVt_PVAV) { |
fc15ae8f | 4452 | const I32 arybase = CopARYBASE_get(PL_curcop); |
4ad10a0b VP |
4453 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; |
4454 | bool can_preserve = FALSE; | |
4455 | ||
4456 | if (localizing) { | |
4457 | MAGIC *mg; | |
4458 | HV *stash; | |
4459 | ||
4460 | can_preserve = SvCANEXISTDELETE(av); | |
4461 | } | |
4462 | ||
4463 | if (lval && localizing) { | |
1b6737cc | 4464 | register SV **svp; |
748a9306 | 4465 | I32 max = -1; |
924508f0 | 4466 | for (svp = MARK + 1; svp <= SP; svp++) { |
4ea561bc | 4467 | const I32 elem = SvIV(*svp); |
748a9306 LW |
4468 | if (elem > max) |
4469 | max = elem; | |
4470 | } | |
4471 | if (max > AvMAX(av)) | |
4472 | av_extend(av, max); | |
4473 | } | |
4ad10a0b | 4474 | |
a0d0e21e | 4475 | while (++MARK <= SP) { |
1b6737cc | 4476 | register SV **svp; |
4ea561bc | 4477 | I32 elem = SvIV(*MARK); |
4ad10a0b | 4478 | bool preeminent = TRUE; |
a0d0e21e | 4479 | |
748a9306 LW |
4480 | if (elem > 0) |
4481 | elem -= arybase; | |
4ad10a0b VP |
4482 | if (localizing && can_preserve) { |
4483 | /* If we can determine whether the element exist, | |
4484 | * Try to preserve the existenceness of a tied array | |
4485 | * element by using EXISTS and DELETE if possible. | |
4486 | * Fallback to FETCH and STORE otherwise. */ | |
4487 | preeminent = av_exists(av, elem); | |
4488 | } | |
4489 | ||
a0d0e21e LW |
4490 | svp = av_fetch(av, elem, lval); |
4491 | if (lval) { | |
3280af22 | 4492 | if (!svp || *svp == &PL_sv_undef) |
cea2e8a9 | 4493 | DIE(aTHX_ PL_no_aelem, elem); |
4ad10a0b VP |
4494 | if (localizing) { |
4495 | if (preeminent) | |
4496 | save_aelem(av, elem, svp); | |
4497 | else | |
4498 | SAVEADELETE(av, elem); | |
4499 | } | |
79072805 | 4500 | } |
3280af22 | 4501 | *MARK = svp ? *svp : &PL_sv_undef; |
79072805 LW |
4502 | } |
4503 | } | |
748a9306 | 4504 | if (GIMME != G_ARRAY) { |
a0d0e21e | 4505 | MARK = ORIGMARK; |
04ab2c87 | 4506 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
a0d0e21e LW |
4507 | SP = MARK; |
4508 | } | |
79072805 LW |
4509 | RETURN; |
4510 | } | |
4511 | ||
878d132a NC |
4512 | PP(pp_aeach) |
4513 | { | |
4514 | dVAR; | |
4515 | dSP; | |
502c6561 | 4516 | AV *array = MUTABLE_AV(POPs); |
878d132a | 4517 | const I32 gimme = GIMME_V; |
453d94a9 | 4518 | IV *iterp = Perl_av_iter_p(aTHX_ array); |
878d132a NC |
4519 | const IV current = (*iterp)++; |
4520 | ||
4521 | if (current > av_len(array)) { | |
4522 | *iterp = 0; | |
4523 | if (gimme == G_SCALAR) | |
4524 | RETPUSHUNDEF; | |
4525 | else | |
4526 | RETURN; | |
4527 | } | |
4528 | ||
4529 | EXTEND(SP, 2); | |
4530 | mPUSHi(CopARYBASE_get(PL_curcop) + current); | |
4531 | if (gimme == G_ARRAY) { | |
4532 | SV **const element = av_fetch(array, current, 0); | |
4533 | PUSHs(element ? *element : &PL_sv_undef); | |
4534 | } | |
4535 | RETURN; | |
4536 | } | |
4537 | ||
4538 | PP(pp_akeys) | |
4539 | { | |
4540 | dVAR; | |
4541 | dSP; | |
502c6561 | 4542 | AV *array = MUTABLE_AV(POPs); |
878d132a NC |
4543 | const I32 gimme = GIMME_V; |
4544 | ||
4545 | *Perl_av_iter_p(aTHX_ array) = 0; | |
4546 | ||
4547 | if (gimme == G_SCALAR) { | |
4548 | dTARGET; | |
4549 | PUSHi(av_len(array) + 1); | |
4550 | } | |
4551 | else if (gimme == G_ARRAY) { | |
4552 | IV n = Perl_av_len(aTHX_ array); | |
4553 | IV i = CopARYBASE_get(PL_curcop); | |
4554 | ||
4555 | EXTEND(SP, n + 1); | |
4556 | ||
4557 | if (PL_op->op_type == OP_AKEYS) { | |
4558 | n += i; | |
4559 | for (; i <= n; i++) { | |
4560 | mPUSHi(i); | |
4561 | } | |
4562 | } | |
4563 | else { | |
4564 | for (i = 0; i <= n; i++) { | |
4565 | SV *const *const elem = Perl_av_fetch(aTHX_ array, i, 0); | |
4566 | PUSHs(elem ? *elem : &PL_sv_undef); | |
4567 | } | |
4568 | } | |
4569 | } | |
4570 | RETURN; | |
4571 | } | |
4572 | ||
79072805 LW |
4573 | /* Associative arrays. */ |
4574 | ||
4575 | PP(pp_each) | |
4576 | { | |
97aff369 | 4577 | dVAR; |
39644a26 | 4578 | dSP; |
85fbaab2 | 4579 | HV * hash = MUTABLE_HV(POPs); |
c07a80fd | 4580 | HE *entry; |
f54cb97a | 4581 | const I32 gimme = GIMME_V; |
8ec5e241 | 4582 | |
c07a80fd | 4583 | PUTBACK; |
c750a3ec | 4584 | /* might clobber stack_sp */ |
6d822dc4 | 4585 | entry = hv_iternext(hash); |
c07a80fd | 4586 | SPAGAIN; |
79072805 | 4587 | |
79072805 LW |
4588 | EXTEND(SP, 2); |
4589 | if (entry) { | |
1b6737cc | 4590 | SV* const sv = hv_iterkeysv(entry); |
574c8022 | 4591 | PUSHs(sv); /* won't clobber stack_sp */ |
54310121 | 4592 | if (gimme == G_ARRAY) { |
59af0135 | 4593 | SV *val; |
c07a80fd | 4594 | PUTBACK; |
c750a3ec | 4595 | /* might clobber stack_sp */ |
6d822dc4 | 4596 | val = hv_iterval(hash, entry); |
c07a80fd | 4597 | SPAGAIN; |
59af0135 | 4598 | PUSHs(val); |
79072805 | 4599 | } |
79072805 | 4600 | } |
54310121 | 4601 | else if (gimme == G_SCALAR) |
79072805 LW |
4602 | RETPUSHUNDEF; |
4603 | ||
4604 | RETURN; | |
4605 | } | |
4606 | ||
7332a6c4 VP |
4607 | STATIC OP * |
4608 | S_do_delete_local(pTHX) | |
79072805 | 4609 | { |
97aff369 | 4610 | dVAR; |
39644a26 | 4611 | dSP; |
f54cb97a | 4612 | const I32 gimme = GIMME_V; |
7332a6c4 VP |
4613 | const MAGIC *mg; |
4614 | HV *stash; | |
4615 | ||
4616 | if (PL_op->op_private & OPpSLICE) { | |
4617 | dMARK; dORIGMARK; | |
4618 | SV * const osv = POPs; | |
4619 | const bool tied = SvRMAGICAL(osv) | |
4620 | && mg_find((const SV *)osv, PERL_MAGIC_tied); | |
4621 | const bool can_preserve = SvCANEXISTDELETE(osv) | |
4622 | || mg_find((const SV *)osv, PERL_MAGIC_env); | |
4623 | const U32 type = SvTYPE(osv); | |
4624 | if (type == SVt_PVHV) { /* hash element */ | |
4625 | HV * const hv = MUTABLE_HV(osv); | |
4626 | while (++MARK <= SP) { | |
4627 | SV * const keysv = *MARK; | |
4628 | SV *sv = NULL; | |
4629 | bool preeminent = TRUE; | |
4630 | if (can_preserve) | |
4631 | preeminent = hv_exists_ent(hv, keysv, 0); | |
4632 | if (tied) { | |
4633 | HE *he = hv_fetch_ent(hv, keysv, 1, 0); | |
4634 | if (he) | |
4635 | sv = HeVAL(he); | |
4636 | else | |
4637 | preeminent = FALSE; | |
4638 | } | |
4639 | else { | |
4640 | sv = hv_delete_ent(hv, keysv, 0, 0); | |
4641 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
4642 | } | |
4643 | if (preeminent) { | |
4644 | save_helem_flags(hv, keysv, &sv, SAVEf_KEEPOLDELEM); | |
4645 | if (tied) { | |
4646 | *MARK = sv_mortalcopy(sv); | |
4647 | mg_clear(sv); | |
4648 | } else | |
4649 | *MARK = sv; | |
4650 | } | |
4651 | else { | |
4652 | SAVEHDELETE(hv, keysv); | |
4653 | *MARK = &PL_sv_undef; | |
4654 | } | |
4655 | } | |
4656 | } | |
4657 | else if (type == SVt_PVAV) { /* array element */ | |
4658 | if (PL_op->op_flags & OPf_SPECIAL) { | |
4659 | AV * const av = MUTABLE_AV(osv); | |
4660 | while (++MARK <= SP) { | |
4661 | I32 idx = SvIV(*MARK); | |
4662 | SV *sv = NULL; | |
4663 | bool preeminent = TRUE; | |
4664 | if (can_preserve) | |
4665 | preeminent = av_exists(av, idx); | |
4666 | if (tied) { | |
4667 | SV **svp = av_fetch(av, idx, 1); | |
4668 | if (svp) | |
4669 | sv = *svp; | |
4670 | else | |
4671 | preeminent = FALSE; | |
4672 | } | |
4673 | else { | |
4674 | sv = av_delete(av, idx, 0); | |
4675 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
4676 | } | |
4677 | if (preeminent) { | |
4678 | save_aelem_flags(av, idx, &sv, SAVEf_KEEPOLDELEM); | |
4679 | if (tied) { | |
4680 | *MARK = sv_mortalcopy(sv); | |
4681 | mg_clear(sv); | |
4682 | } else | |
4683 | *MARK = sv; | |
4684 | } | |
4685 | else { | |
4686 | SAVEADELETE(av, idx); | |
4687 | *MARK = &PL_sv_undef; | |
4688 | } | |
4689 | } | |
4690 | } | |
4691 | } | |
4692 | else | |
4693 | DIE(aTHX_ "Not a HASH reference"); | |
4694 | if (gimme == G_VOID) | |
4695 | SP = ORIGMARK; | |
4696 | else if (gimme == G_SCALAR) { | |
4697 | MARK = ORIGMARK; | |
4698 | if (SP > MARK) | |
4699 | *++MARK = *SP; | |
4700 | else | |
4701 | *++MARK = &PL_sv_undef; | |
4702 | SP = MARK; | |
4703 | } | |
4704 | } | |
4705 | else { | |
4706 | SV * const keysv = POPs; | |
4707 | SV * const osv = POPs; | |
4708 | const bool tied = SvRMAGICAL(osv) | |
4709 | && mg_find((const SV *)osv, PERL_MAGIC_tied); | |
4710 | const bool can_preserve = SvCANEXISTDELETE(osv) | |
4711 | || mg_find((const SV *)osv, PERL_MAGIC_env); | |
4712 | const U32 type = SvTYPE(osv); | |
4713 | SV *sv = NULL; | |
4714 | if (type == SVt_PVHV) { | |
4715 | HV * const hv = MUTABLE_HV(osv); | |
4716 | bool preeminent = TRUE; | |
4717 | if (can_preserve) | |
4718 | preeminent = hv_exists_ent(hv, keysv, 0); | |
4719 | if (tied) { | |
4720 | HE *he = hv_fetch_ent(hv, keysv, 1, 0); | |
4721 | if (he) | |
4722 | sv = HeVAL(he); | |
4723 | else | |
4724 | preeminent = FALSE; | |
4725 | } | |
4726 | else { | |
4727 | sv = hv_delete_ent(hv, keysv, 0, 0); | |
4728 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
4729 | } | |
4730 | if (preeminent) { | |
4731 | save_helem_flags(hv, keysv, &sv, SAVEf_KEEPOLDELEM); | |
4732 | if (tied) { | |
4733 | SV *nsv = sv_mortalcopy(sv); | |
4734 | mg_clear(sv); | |
4735 | sv = nsv; | |
4736 | } | |
4737 | } | |
4738 | else | |
4739 | SAVEHDELETE(hv, keysv); | |
4740 | } | |
4741 | else if (type == SVt_PVAV) { | |
4742 | if (PL_op->op_flags & OPf_SPECIAL) { | |
4743 | AV * const av = MUTABLE_AV(osv); | |
4744 | I32 idx = SvIV(keysv); | |
4745 | bool preeminent = TRUE; | |
4746 | if (can_preserve) | |
4747 | preeminent = av_exists(av, idx); | |
4748 | if (tied) { | |
4749 | SV **svp = av_fetch(av, idx, 1); | |
4750 | if (svp) | |
4751 | sv = *svp; | |
4752 | else | |
4753 | preeminent = FALSE; | |
4754 | } | |
4755 | else { | |
4756 | sv = av_delete(av, idx, 0); | |
4757 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
4758 | } | |
4759 | if (preeminent) { | |
4760 | save_aelem_flags(av, idx, &sv, SAVEf_KEEPOLDELEM); | |
4761 | if (tied) { | |
4762 | SV *nsv = sv_mortalcopy(sv); | |
4763 | mg_clear(sv); | |
4764 | sv = nsv; | |
4765 | } | |
4766 | } | |
4767 | else | |
4768 | SAVEADELETE(av, idx); | |
4769 | } | |
4770 | else | |
4771 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
4772 | } | |
4773 | else | |
4774 | DIE(aTHX_ "Not a HASH reference"); | |
4775 | if (!sv) | |
4776 | sv = &PL_sv_undef; | |
4777 | if (gimme != G_VOID) | |
4778 | PUSHs(sv); | |
4779 | } | |
4780 | ||
4781 | RETURN; | |
4782 | } | |
4783 | ||
4784 | PP(pp_delete) | |
4785 | { | |
4786 | dVAR; | |
4787 | dSP; | |
4788 | I32 gimme; | |
4789 | I32 discard; | |
4790 | ||
4791 | if (PL_op->op_private & OPpLVAL_INTRO) | |
4792 | return do_delete_local(); | |
4793 | ||
4794 | gimme = GIMME_V; | |
4795 | discard = (gimme == G_VOID) ? G_DISCARD : 0; | |
5f05dabc | 4796 | |
533c011a | 4797 | if (PL_op->op_private & OPpSLICE) { |
5f05dabc | 4798 | dMARK; dORIGMARK; |
85fbaab2 | 4799 | HV * const hv = MUTABLE_HV(POPs); |
1b6737cc | 4800 | const U32 hvtype = SvTYPE(hv); |
01020589 GS |
4801 | if (hvtype == SVt_PVHV) { /* hash element */ |
4802 | while (++MARK <= SP) { | |
1b6737cc | 4803 | SV * const sv = hv_delete_ent(hv, *MARK, discard, 0); |
01020589 GS |
4804 | *MARK = sv ? sv : &PL_sv_undef; |
4805 | } | |
5f05dabc | 4806 | } |
6d822dc4 MS |
4807 | else if (hvtype == SVt_PVAV) { /* array element */ |
4808 | if (PL_op->op_flags & OPf_SPECIAL) { | |
4809 | while (++MARK <= SP) { | |
502c6561 | 4810 | SV * const sv = av_delete(MUTABLE_AV(hv), SvIV(*MARK), discard); |
6d822dc4 MS |
4811 | *MARK = sv ? sv : &PL_sv_undef; |
4812 | } | |
4813 | } | |
01020589 GS |
4814 | } |
4815 | else | |
4816 | DIE(aTHX_ "Not a HASH reference"); | |
54310121 | 4817 | if (discard) |
4818 | SP = ORIGMARK; | |
4819 | else if (gimme == G_SCALAR) { | |
5f05dabc | 4820 | MARK = ORIGMARK; |
9111c9c0 DM |
4821 | if (SP > MARK) |
4822 | *++MARK = *SP; | |
4823 | else | |
4824 | *++MARK = &PL_sv_undef; | |
5f05dabc | 4825 | SP = MARK; |
4826 | } | |
4827 | } | |
4828 | else { | |
4829 | SV *keysv = POPs; | |
85fbaab2 | 4830 | HV * const hv = MUTABLE_HV(POPs); |
295d248e | 4831 | SV *sv = NULL; |
97fcbf96 MB |
4832 | if (SvTYPE(hv) == SVt_PVHV) |
4833 | sv = hv_delete_ent(hv, keysv, discard, 0); | |
01020589 GS |
4834 | else if (SvTYPE(hv) == SVt_PVAV) { |
4835 | if (PL_op->op_flags & OPf_SPECIAL) | |
502c6561 | 4836 | sv = av_delete(MUTABLE_AV(hv), SvIV(keysv), discard); |
af288a60 HS |
4837 | else |
4838 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
01020589 | 4839 | } |
97fcbf96 | 4840 | else |
cea2e8a9 | 4841 | DIE(aTHX_ "Not a HASH reference"); |
5f05dabc | 4842 | if (!sv) |
3280af22 | 4843 | sv = &PL_sv_undef; |
54310121 | 4844 | if (!discard) |
4845 | PUSHs(sv); | |
79072805 | 4846 | } |
79072805 LW |
4847 | RETURN; |
4848 | } | |
4849 | ||
a0d0e21e | 4850 | PP(pp_exists) |
79072805 | 4851 | { |
97aff369 | 4852 | dVAR; |
39644a26 | 4853 | dSP; |
afebc493 GS |
4854 | SV *tmpsv; |
4855 | HV *hv; | |
4856 | ||
4857 | if (PL_op->op_private & OPpEXISTS_SUB) { | |
4858 | GV *gv; | |
0bd48802 | 4859 | SV * const sv = POPs; |
f2c0649b | 4860 | CV * const cv = sv_2cv(sv, &hv, &gv, 0); |
afebc493 GS |
4861 | if (cv) |
4862 | RETPUSHYES; | |
4863 | if (gv && isGV(gv) && GvCV(gv) && !GvCVGEN(gv)) | |
4864 | RETPUSHYES; | |
4865 | RETPUSHNO; | |
4866 | } | |
4867 | tmpsv = POPs; | |
85fbaab2 | 4868 | hv = MUTABLE_HV(POPs); |
c750a3ec | 4869 | if (SvTYPE(hv) == SVt_PVHV) { |
ae77835f | 4870 | if (hv_exists_ent(hv, tmpsv, 0)) |
c750a3ec | 4871 | RETPUSHYES; |
ef54e1a4 JH |
4872 | } |
4873 | else if (SvTYPE(hv) == SVt_PVAV) { | |
01020589 | 4874 | if (PL_op->op_flags & OPf_SPECIAL) { /* array element */ |
502c6561 | 4875 | if (av_exists(MUTABLE_AV(hv), SvIV(tmpsv))) |
01020589 GS |
4876 | RETPUSHYES; |
4877 | } | |
ef54e1a4 JH |
4878 | } |
4879 | else { | |
cea2e8a9 | 4880 | DIE(aTHX_ "Not a HASH reference"); |
a0d0e21e | 4881 | } |
a0d0e21e LW |
4882 | RETPUSHNO; |
4883 | } | |
79072805 | 4884 | |
a0d0e21e LW |
4885 | PP(pp_hslice) |
4886 | { | |
97aff369 | 4887 | dVAR; dSP; dMARK; dORIGMARK; |
85fbaab2 | 4888 | register HV * const hv = MUTABLE_HV(POPs); |
1b6737cc AL |
4889 | register const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); |
4890 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; | |
d30e492c | 4891 | bool can_preserve = FALSE; |
79072805 | 4892 | |
eb85dfd3 DM |
4893 | if (localizing) { |
4894 | MAGIC *mg; | |
4895 | HV *stash; | |
4896 | ||
d30e492c VP |
4897 | if (SvCANEXISTDELETE(hv) || mg_find((const SV *)hv, PERL_MAGIC_env)) |
4898 | can_preserve = TRUE; | |
eb85dfd3 DM |
4899 | } |
4900 | ||
6d822dc4 | 4901 | while (++MARK <= SP) { |
1b6737cc | 4902 | SV * const keysv = *MARK; |
6d822dc4 MS |
4903 | SV **svp; |
4904 | HE *he; | |
d30e492c VP |
4905 | bool preeminent = TRUE; |
4906 | ||
4907 | if (localizing && can_preserve) { | |
4908 | /* If we can determine whether the element exist, | |
4909 | * try to preserve the existenceness of a tied hash | |
4910 | * element by using EXISTS and DELETE if possible. | |
4911 | * Fallback to FETCH and STORE otherwise. */ | |
4912 | preeminent = hv_exists_ent(hv, keysv, 0); | |
6d822dc4 | 4913 | } |
eb85dfd3 | 4914 | |
6d822dc4 | 4915 | he = hv_fetch_ent(hv, keysv, lval, 0); |
fe5bfecd | 4916 | svp = he ? &HeVAL(he) : NULL; |
eb85dfd3 | 4917 | |
6d822dc4 MS |
4918 | if (lval) { |
4919 | if (!svp || *svp == &PL_sv_undef) { | |
be2597df | 4920 | DIE(aTHX_ PL_no_helem_sv, SVfARG(keysv)); |
6d822dc4 MS |
4921 | } |
4922 | if (localizing) { | |
7a2e501a | 4923 | if (HvNAME_get(hv) && isGV(*svp)) |
159b6efe | 4924 | save_gp(MUTABLE_GV(*svp), !(PL_op->op_flags & OPf_SPECIAL)); |
47cfc530 VP |
4925 | else if (preeminent) |
4926 | save_helem_flags(hv, keysv, svp, | |
4927 | (PL_op->op_flags & OPf_SPECIAL) ? 0 : SAVEf_SETMAGIC); | |
4928 | else | |
4929 | SAVEHDELETE(hv, keysv); | |
6d822dc4 MS |
4930 | } |
4931 | } | |
4932 | *MARK = svp ? *svp : &PL_sv_undef; | |
79072805 | 4933 | } |
a0d0e21e LW |
4934 | if (GIMME != G_ARRAY) { |
4935 | MARK = ORIGMARK; | |
04ab2c87 | 4936 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
a0d0e21e | 4937 | SP = MARK; |
79072805 | 4938 | } |
a0d0e21e LW |
4939 | RETURN; |
4940 | } | |
4941 | ||
4942 | /* List operators. */ | |
4943 | ||
4944 | PP(pp_list) | |
4945 | { | |
97aff369 | 4946 | dVAR; dSP; dMARK; |
a0d0e21e LW |
4947 | if (GIMME != G_ARRAY) { |
4948 | if (++MARK <= SP) | |
4949 | *MARK = *SP; /* unwanted list, return last item */ | |
8990e307 | 4950 | else |
3280af22 | 4951 | *MARK = &PL_sv_undef; |
a0d0e21e | 4952 | SP = MARK; |
79072805 | 4953 | } |
a0d0e21e | 4954 | RETURN; |
79072805 LW |
4955 | } |
4956 | ||
a0d0e21e | 4957 | PP(pp_lslice) |
79072805 | 4958 | { |
97aff369 | 4959 | dVAR; |
39644a26 | 4960 | dSP; |
1b6737cc AL |
4961 | SV ** const lastrelem = PL_stack_sp; |
4962 | SV ** const lastlelem = PL_stack_base + POPMARK; | |
4963 | SV ** const firstlelem = PL_stack_base + POPMARK + 1; | |
4964 | register SV ** const firstrelem = lastlelem + 1; | |
fc15ae8f | 4965 | const I32 arybase = CopARYBASE_get(PL_curcop); |
42e73ed0 | 4966 | I32 is_something_there = FALSE; |
1b6737cc AL |
4967 | |
4968 | register const I32 max = lastrelem - lastlelem; | |
a0d0e21e | 4969 | register SV **lelem; |
a0d0e21e LW |
4970 | |
4971 | if (GIMME != G_ARRAY) { | |
4ea561bc | 4972 | I32 ix = SvIV(*lastlelem); |
748a9306 LW |
4973 | if (ix < 0) |
4974 | ix += max; | |
4975 | else | |
4976 | ix -= arybase; | |
a0d0e21e | 4977 | if (ix < 0 || ix >= max) |
3280af22 | 4978 | *firstlelem = &PL_sv_undef; |
a0d0e21e LW |
4979 | else |
4980 | *firstlelem = firstrelem[ix]; | |
4981 | SP = firstlelem; | |
4982 | RETURN; | |
4983 | } | |
4984 | ||
4985 | if (max == 0) { | |
4986 | SP = firstlelem - 1; | |
4987 | RETURN; | |
4988 | } | |
4989 | ||
4990 | for (lelem = firstlelem; lelem <= lastlelem; lelem++) { | |
4ea561bc | 4991 | I32 ix = SvIV(*lelem); |
c73bf8e3 | 4992 | if (ix < 0) |
a0d0e21e | 4993 | ix += max; |
b13b2135 | 4994 | else |
748a9306 | 4995 | ix -= arybase; |
c73bf8e3 HS |
4996 | if (ix < 0 || ix >= max) |
4997 | *lelem = &PL_sv_undef; | |
4998 | else { | |
4999 | is_something_there = TRUE; | |
5000 | if (!(*lelem = firstrelem[ix])) | |
3280af22 | 5001 | *lelem = &PL_sv_undef; |
748a9306 | 5002 | } |
79072805 | 5003 | } |
4633a7c4 LW |
5004 | if (is_something_there) |
5005 | SP = lastlelem; | |
5006 | else | |
5007 | SP = firstlelem - 1; | |
79072805 LW |
5008 | RETURN; |
5009 | } | |
5010 | ||
a0d0e21e LW |
5011 | PP(pp_anonlist) |
5012 | { | |
97aff369 | 5013 | dVAR; dSP; dMARK; dORIGMARK; |
1b6737cc | 5014 | const I32 items = SP - MARK; |
ad64d0ec | 5015 | SV * const av = MUTABLE_SV(av_make(items, MARK+1)); |
44a8e56a | 5016 | SP = ORIGMARK; /* av_make() might realloc stack_sp */ |
6e449a3a MHM |
5017 | mXPUSHs((PL_op->op_flags & OPf_SPECIAL) |
5018 | ? newRV_noinc(av) : av); | |
a0d0e21e LW |
5019 | RETURN; |
5020 | } | |
5021 | ||
5022 | PP(pp_anonhash) | |
79072805 | 5023 | { |
97aff369 | 5024 | dVAR; dSP; dMARK; dORIGMARK; |
78c72037 | 5025 | HV* const hv = newHV(); |
a0d0e21e LW |
5026 | |
5027 | while (MARK < SP) { | |
1b6737cc | 5028 | SV * const key = *++MARK; |
561b68a9 | 5029 | SV * const val = newSV(0); |
a0d0e21e LW |
5030 | if (MARK < SP) |
5031 | sv_setsv(val, *++MARK); | |
a2a5de95 NC |
5032 | else |
5033 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "Odd number of elements in anonymous hash"); | |
f12c7020 | 5034 | (void)hv_store_ent(hv,key,val,0); |
79072805 | 5035 | } |
a0d0e21e | 5036 | SP = ORIGMARK; |
6e449a3a | 5037 | mXPUSHs((PL_op->op_flags & OPf_SPECIAL) |
ad64d0ec | 5038 | ? newRV_noinc(MUTABLE_SV(hv)) : MUTABLE_SV(hv)); |
79072805 LW |
5039 | RETURN; |
5040 | } | |
5041 | ||
a0d0e21e | 5042 | PP(pp_splice) |
79072805 | 5043 | { |
27da23d5 | 5044 | dVAR; dSP; dMARK; dORIGMARK; |
502c6561 | 5045 | register AV *ary = MUTABLE_AV(*++MARK); |
a0d0e21e LW |
5046 | register SV **src; |
5047 | register SV **dst; | |
5048 | register I32 i; | |
5049 | register I32 offset; | |
5050 | register I32 length; | |
5051 | I32 newlen; | |
5052 | I32 after; | |
5053 | I32 diff; | |
ad64d0ec | 5054 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
93965878 | 5055 | |
1b6737cc | 5056 | if (mg) { |
ad64d0ec | 5057 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
93965878 | 5058 | PUSHMARK(MARK); |
8ec5e241 | 5059 | PUTBACK; |
d343c3ef | 5060 | ENTER_with_name("call_SPLICE"); |
864dbfa3 | 5061 | call_method("SPLICE",GIMME_V); |
d343c3ef | 5062 | LEAVE_with_name("call_SPLICE"); |
93965878 NIS |
5063 | SPAGAIN; |
5064 | RETURN; | |
5065 | } | |
79072805 | 5066 | |
a0d0e21e | 5067 | SP++; |
79072805 | 5068 | |
a0d0e21e | 5069 | if (++MARK < SP) { |
4ea561bc | 5070 | offset = i = SvIV(*MARK); |
a0d0e21e | 5071 | if (offset < 0) |
93965878 | 5072 | offset += AvFILLp(ary) + 1; |
a0d0e21e | 5073 | else |
fc15ae8f | 5074 | offset -= CopARYBASE_get(PL_curcop); |
84902520 | 5075 | if (offset < 0) |
cea2e8a9 | 5076 | DIE(aTHX_ PL_no_aelem, i); |
a0d0e21e LW |
5077 | if (++MARK < SP) { |
5078 | length = SvIVx(*MARK++); | |
48cdf507 GA |
5079 | if (length < 0) { |
5080 | length += AvFILLp(ary) - offset + 1; | |
5081 | if (length < 0) | |
5082 | length = 0; | |
5083 | } | |
79072805 LW |
5084 | } |
5085 | else | |
a0d0e21e | 5086 | length = AvMAX(ary) + 1; /* close enough to infinity */ |
79072805 | 5087 | } |
a0d0e21e LW |
5088 | else { |
5089 | offset = 0; | |
5090 | length = AvMAX(ary) + 1; | |
5091 | } | |
8cbc2e3b | 5092 | if (offset > AvFILLp(ary) + 1) { |
a2a5de95 | 5093 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "splice() offset past end of array" ); |
93965878 | 5094 | offset = AvFILLp(ary) + 1; |
8cbc2e3b | 5095 | } |
93965878 | 5096 | after = AvFILLp(ary) + 1 - (offset + length); |
a0d0e21e LW |
5097 | if (after < 0) { /* not that much array */ |
5098 | length += after; /* offset+length now in array */ | |
5099 | after = 0; | |
5100 | if (!AvALLOC(ary)) | |
5101 | av_extend(ary, 0); | |
5102 | } | |
5103 | ||
5104 | /* At this point, MARK .. SP-1 is our new LIST */ | |
5105 | ||
5106 | newlen = SP - MARK; | |
5107 | diff = newlen - length; | |
13d7cbc1 GS |
5108 | if (newlen && !AvREAL(ary) && AvREIFY(ary)) |
5109 | av_reify(ary); | |
a0d0e21e | 5110 | |
50528de0 WL |
5111 | /* make new elements SVs now: avoid problems if they're from the array */ |
5112 | for (dst = MARK, i = newlen; i; i--) { | |
1b6737cc | 5113 | SV * const h = *dst; |
f2b990bf | 5114 | *dst++ = newSVsv(h); |
50528de0 WL |
5115 | } |
5116 | ||
a0d0e21e | 5117 | if (diff < 0) { /* shrinking the area */ |
95b63a38 | 5118 | SV **tmparyval = NULL; |
a0d0e21e | 5119 | if (newlen) { |
a02a5408 | 5120 | Newx(tmparyval, newlen, SV*); /* so remember insertion */ |
a0d0e21e | 5121 | Copy(MARK, tmparyval, newlen, SV*); |
79072805 | 5122 | } |
a0d0e21e LW |
5123 | |
5124 | MARK = ORIGMARK + 1; | |
5125 | if (GIMME == G_ARRAY) { /* copy return vals to stack */ | |
5126 | MEXTEND(MARK, length); | |
5127 | Copy(AvARRAY(ary)+offset, MARK, length, SV*); | |
5128 | if (AvREAL(ary)) { | |
bbce6d69 | 5129 | EXTEND_MORTAL(length); |
36477c24 | 5130 | for (i = length, dst = MARK; i; i--) { |
d689ffdd | 5131 | sv_2mortal(*dst); /* free them eventualy */ |
36477c24 | 5132 | dst++; |
5133 | } | |
a0d0e21e LW |
5134 | } |
5135 | MARK += length - 1; | |
79072805 | 5136 | } |
a0d0e21e LW |
5137 | else { |
5138 | *MARK = AvARRAY(ary)[offset+length-1]; | |
5139 | if (AvREAL(ary)) { | |
d689ffdd | 5140 | sv_2mortal(*MARK); |
a0d0e21e LW |
5141 | for (i = length - 1, dst = &AvARRAY(ary)[offset]; i > 0; i--) |
5142 | SvREFCNT_dec(*dst++); /* free them now */ | |
79072805 | 5143 | } |
a0d0e21e | 5144 | } |
93965878 | 5145 | AvFILLp(ary) += diff; |
a0d0e21e LW |
5146 | |
5147 | /* pull up or down? */ | |
5148 | ||
5149 | if (offset < after) { /* easier to pull up */ | |
5150 | if (offset) { /* esp. if nothing to pull */ | |
5151 | src = &AvARRAY(ary)[offset-1]; | |
5152 | dst = src - diff; /* diff is negative */ | |
5153 | for (i = offset; i > 0; i--) /* can't trust Copy */ | |
5154 | *dst-- = *src--; | |
79072805 | 5155 | } |
a0d0e21e | 5156 | dst = AvARRAY(ary); |
9c6bc640 | 5157 | AvARRAY(ary) = AvARRAY(ary) - diff; /* diff is negative */ |
a0d0e21e LW |
5158 | AvMAX(ary) += diff; |
5159 | } | |
5160 | else { | |
5161 | if (after) { /* anything to pull down? */ | |
5162 | src = AvARRAY(ary) + offset + length; | |
5163 | dst = src + diff; /* diff is negative */ | |
5164 | Move(src, dst, after, SV*); | |
79072805 | 5165 | } |
93965878 | 5166 | dst = &AvARRAY(ary)[AvFILLp(ary)+1]; |
a0d0e21e LW |
5167 | /* avoid later double free */ |
5168 | } | |
5169 | i = -diff; | |
5170 | while (i) | |
3280af22 | 5171 | dst[--i] = &PL_sv_undef; |
a0d0e21e LW |
5172 | |
5173 | if (newlen) { | |
50528de0 | 5174 | Copy( tmparyval, AvARRAY(ary) + offset, newlen, SV* ); |
a0d0e21e LW |
5175 | Safefree(tmparyval); |
5176 | } | |
5177 | } | |
5178 | else { /* no, expanding (or same) */ | |
d3961450 | 5179 | SV** tmparyval = NULL; |
a0d0e21e | 5180 | if (length) { |
a02a5408 | 5181 | Newx(tmparyval, length, SV*); /* so remember deletion */ |
a0d0e21e LW |
5182 | Copy(AvARRAY(ary)+offset, tmparyval, length, SV*); |
5183 | } | |
5184 | ||
5185 | if (diff > 0) { /* expanding */ | |
a0d0e21e | 5186 | /* push up or down? */ |
a0d0e21e LW |
5187 | if (offset < after && diff <= AvARRAY(ary) - AvALLOC(ary)) { |
5188 | if (offset) { | |
5189 | src = AvARRAY(ary); | |
5190 | dst = src - diff; | |
5191 | Move(src, dst, offset, SV*); | |
79072805 | 5192 | } |
9c6bc640 | 5193 | AvARRAY(ary) = AvARRAY(ary) - diff;/* diff is positive */ |
a0d0e21e | 5194 | AvMAX(ary) += diff; |
93965878 | 5195 | AvFILLp(ary) += diff; |
79072805 LW |
5196 | } |
5197 | else { | |
93965878 NIS |
5198 | if (AvFILLp(ary) + diff >= AvMAX(ary)) /* oh, well */ |
5199 | av_extend(ary, AvFILLp(ary) + diff); | |
5200 | AvFILLp(ary) += diff; | |
a0d0e21e LW |
5201 | |
5202 | if (after) { | |
93965878 | 5203 | dst = AvARRAY(ary) + AvFILLp(ary); |
a0d0e21e LW |
5204 | src = dst - diff; |
5205 | for (i = after; i; i--) { | |
5206 | *dst-- = *src--; | |
5207 | } | |
79072805 LW |
5208 | } |
5209 | } | |
a0d0e21e LW |
5210 | } |
5211 | ||
50528de0 WL |
5212 | if (newlen) { |
5213 | Copy( MARK, AvARRAY(ary) + offset, newlen, SV* ); | |
a0d0e21e | 5214 | } |
50528de0 | 5215 | |
a0d0e21e LW |
5216 | MARK = ORIGMARK + 1; |
5217 | if (GIMME == G_ARRAY) { /* copy return vals to stack */ | |
5218 | if (length) { | |
5219 | Copy(tmparyval, MARK, length, SV*); | |
5220 | if (AvREAL(ary)) { | |
bbce6d69 | 5221 | EXTEND_MORTAL(length); |
36477c24 | 5222 | for (i = length, dst = MARK; i; i--) { |
d689ffdd | 5223 | sv_2mortal(*dst); /* free them eventualy */ |
36477c24 | 5224 | dst++; |
5225 | } | |
79072805 LW |
5226 | } |
5227 | } | |
a0d0e21e LW |
5228 | MARK += length - 1; |
5229 | } | |
5230 | else if (length--) { | |
5231 | *MARK = tmparyval[length]; | |
5232 | if (AvREAL(ary)) { | |
d689ffdd | 5233 | sv_2mortal(*MARK); |
a0d0e21e LW |
5234 | while (length-- > 0) |
5235 | SvREFCNT_dec(tmparyval[length]); | |
79072805 | 5236 | } |
79072805 | 5237 | } |
a0d0e21e | 5238 | else |
3280af22 | 5239 | *MARK = &PL_sv_undef; |
d3961450 | 5240 | Safefree(tmparyval); |
79072805 | 5241 | } |
a0d0e21e | 5242 | SP = MARK; |
79072805 LW |
5243 | RETURN; |
5244 | } | |
5245 | ||
a0d0e21e | 5246 | PP(pp_push) |
79072805 | 5247 | { |
27da23d5 | 5248 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
502c6561 | 5249 | register AV * const ary = MUTABLE_AV(*++MARK); |
ad64d0ec | 5250 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
79072805 | 5251 | |
1b6737cc | 5252 | if (mg) { |
ad64d0ec | 5253 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
93965878 NIS |
5254 | PUSHMARK(MARK); |
5255 | PUTBACK; | |
d343c3ef | 5256 | ENTER_with_name("call_PUSH"); |
864dbfa3 | 5257 | call_method("PUSH",G_SCALAR|G_DISCARD); |
d343c3ef | 5258 | LEAVE_with_name("call_PUSH"); |
93965878 | 5259 | SPAGAIN; |
93965878 | 5260 | } |
a60c0954 | 5261 | else { |
89c14e2e | 5262 | PL_delaymagic = DM_DELAY; |
a60c0954 | 5263 | for (++MARK; MARK <= SP; MARK++) { |
561b68a9 | 5264 | SV * const sv = newSV(0); |
a60c0954 NIS |
5265 | if (*MARK) |
5266 | sv_setsv(sv, *MARK); | |
0a75904b | 5267 | av_store(ary, AvFILLp(ary)+1, sv); |
a60c0954 | 5268 | } |
89c14e2e | 5269 | if (PL_delaymagic & DM_ARRAY) |
ad64d0ec | 5270 | mg_set(MUTABLE_SV(ary)); |
89c14e2e BB |
5271 | |
5272 | PL_delaymagic = 0; | |
6eeabd23 VP |
5273 | } |
5274 | SP = ORIGMARK; | |
5275 | if (OP_GIMME(PL_op, 0) != G_VOID) { | |
5276 | PUSHi( AvFILL(ary) + 1 ); | |
79072805 | 5277 | } |
79072805 LW |
5278 | RETURN; |
5279 | } | |
5280 | ||
a0d0e21e | 5281 | PP(pp_shift) |
79072805 | 5282 | { |
97aff369 | 5283 | dVAR; |
39644a26 | 5284 | dSP; |
502c6561 | 5285 | AV * const av = MUTABLE_AV(POPs); |
789b4bc9 | 5286 | SV * const sv = PL_op->op_type == OP_SHIFT ? av_shift(av) : av_pop(av); |
79072805 | 5287 | EXTEND(SP, 1); |
c2b4a044 | 5288 | assert (sv); |
d689ffdd | 5289 | if (AvREAL(av)) |
a0d0e21e LW |
5290 | (void)sv_2mortal(sv); |
5291 | PUSHs(sv); | |
79072805 | 5292 | RETURN; |
79072805 LW |
5293 | } |
5294 | ||
a0d0e21e | 5295 | PP(pp_unshift) |
79072805 | 5296 | { |
27da23d5 | 5297 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
502c6561 | 5298 | register AV *ary = MUTABLE_AV(*++MARK); |
ad64d0ec | 5299 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
93965878 | 5300 | |
1b6737cc | 5301 | if (mg) { |
ad64d0ec | 5302 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
7fd66d9d | 5303 | PUSHMARK(MARK); |
93965878 | 5304 | PUTBACK; |
d343c3ef | 5305 | ENTER_with_name("call_UNSHIFT"); |
864dbfa3 | 5306 | call_method("UNSHIFT",G_SCALAR|G_DISCARD); |
d343c3ef | 5307 | LEAVE_with_name("call_UNSHIFT"); |
93965878 | 5308 | SPAGAIN; |
93965878 | 5309 | } |
a60c0954 | 5310 | else { |
1b6737cc | 5311 | register I32 i = 0; |
a60c0954 NIS |
5312 | av_unshift(ary, SP - MARK); |
5313 | while (MARK < SP) { | |
1b6737cc | 5314 | SV * const sv = newSVsv(*++MARK); |
a60c0954 NIS |
5315 | (void)av_store(ary, i++, sv); |
5316 | } | |
79072805 | 5317 | } |
a0d0e21e | 5318 | SP = ORIGMARK; |
6eeabd23 | 5319 | if (OP_GIMME(PL_op, 0) != G_VOID) { |
5658d0a9 LR |
5320 | PUSHi( AvFILL(ary) + 1 ); |
5321 | } | |
79072805 | 5322 | RETURN; |
79072805 LW |
5323 | } |
5324 | ||
a0d0e21e | 5325 | PP(pp_reverse) |
79072805 | 5326 | { |
97aff369 | 5327 | dVAR; dSP; dMARK; |
79072805 | 5328 | |
a0d0e21e | 5329 | if (GIMME == G_ARRAY) { |
484c818f VP |
5330 | if (PL_op->op_private & OPpREVERSE_INPLACE) { |
5331 | AV *av; | |
5332 | ||
5333 | /* See pp_sort() */ | |
5334 | assert( MARK+1 == SP && *SP && SvTYPE(*SP) == SVt_PVAV); | |
5335 | (void)POPMARK; /* remove mark associated with ex-OP_AASSIGN */ | |
5336 | av = MUTABLE_AV((*SP)); | |
5337 | /* In-place reversing only happens in void context for the array | |
5338 | * assignment. We don't need to push anything on the stack. */ | |
5339 | SP = MARK; | |
5340 | ||
5341 | if (SvMAGICAL(av)) { | |
5342 | I32 i, j; | |
5343 | register SV *tmp = sv_newmortal(); | |
5344 | /* For SvCANEXISTDELETE */ | |
5345 | HV *stash; | |
5346 | const MAGIC *mg; | |
5347 | bool can_preserve = SvCANEXISTDELETE(av); | |
5348 | ||
5349 | for (i = 0, j = av_len(av); i < j; ++i, --j) { | |
5350 | register SV *begin, *end; | |
5351 | ||
5352 | if (can_preserve) { | |
5353 | if (!av_exists(av, i)) { | |
5354 | if (av_exists(av, j)) { | |
5355 | register SV *sv = av_delete(av, j, 0); | |
5356 | begin = *av_fetch(av, i, TRUE); | |
5357 | sv_setsv_mg(begin, sv); | |
5358 | } | |
5359 | continue; | |
5360 | } | |
5361 | else if (!av_exists(av, j)) { | |
5362 | register SV *sv = av_delete(av, i, 0); | |
5363 | end = *av_fetch(av, j, TRUE); | |
5364 | sv_setsv_mg(end, sv); | |
5365 | continue; | |
5366 | } | |
5367 | } | |
5368 | ||
5369 | begin = *av_fetch(av, i, TRUE); | |
5370 | end = *av_fetch(av, j, TRUE); | |
5371 | sv_setsv(tmp, begin); | |
5372 | sv_setsv_mg(begin, end); | |
5373 | sv_setsv_mg(end, tmp); | |
5374 | } | |
5375 | } | |
5376 | else { | |
5377 | SV **begin = AvARRAY(av); | |
5378 | SV **end = begin + AvFILLp(av); | |
5379 | ||
5380 | while (begin < end) { | |
5381 | register SV * const tmp = *begin; | |
5382 | *begin++ = *end; | |
5383 | *end-- = tmp; | |
5384 | } | |
5385 | } | |
5386 | } | |
5387 | else { | |
5388 | SV **oldsp = SP; | |
5389 | MARK++; | |
5390 | while (MARK < SP) { | |
5391 | register SV * const tmp = *MARK; | |
5392 | *MARK++ = *SP; | |
5393 | *SP-- = tmp; | |
5394 | } | |
5395 | /* safe as long as stack cannot get extended in the above */ | |
5396 | SP = oldsp; | |
a0d0e21e | 5397 | } |
79072805 LW |
5398 | } |
5399 | else { | |
a0d0e21e LW |
5400 | register char *up; |
5401 | register char *down; | |
5402 | register I32 tmp; | |
5403 | dTARGET; | |
5404 | STRLEN len; | |
9f7d9405 | 5405 | PADOFFSET padoff_du; |
79072805 | 5406 | |
7e2040f0 | 5407 | SvUTF8_off(TARG); /* decontaminate */ |
a0d0e21e | 5408 | if (SP - MARK > 1) |
3280af22 | 5409 | do_join(TARG, &PL_sv_no, MARK, SP); |
1e21d011 | 5410 | else { |
e1f795dc RGS |
5411 | sv_setsv(TARG, (SP > MARK) |
5412 | ? *SP | |
29289021 | 5413 | : (padoff_du = find_rundefsvoffset(), |
00b1698f NC |
5414 | (padoff_du == NOT_IN_PAD |
5415 | || PAD_COMPNAME_FLAGS_isOUR(padoff_du)) | |
e1f795dc | 5416 | ? DEFSV : PAD_SVl(padoff_du))); |
1e21d011 B |
5417 | |
5418 | if (! SvOK(TARG) && ckWARN(WARN_UNINITIALIZED)) | |
5419 | report_uninit(TARG); | |
5420 | } | |
5421 | ||
a0d0e21e LW |
5422 | up = SvPV_force(TARG, len); |
5423 | if (len > 1) { | |
7e2040f0 | 5424 | if (DO_UTF8(TARG)) { /* first reverse each character */ |
dfe13c55 | 5425 | U8* s = (U8*)SvPVX(TARG); |
349d4f2f | 5426 | const U8* send = (U8*)(s + len); |
a0ed51b3 | 5427 | while (s < send) { |
d742c382 | 5428 | if (UTF8_IS_INVARIANT(*s)) { |
a0ed51b3 LW |
5429 | s++; |
5430 | continue; | |
5431 | } | |
5432 | else { | |
9041c2e3 | 5433 | if (!utf8_to_uvchr(s, 0)) |
a0dbb045 | 5434 | break; |
dfe13c55 | 5435 | up = (char*)s; |
a0ed51b3 | 5436 | s += UTF8SKIP(s); |
dfe13c55 | 5437 | down = (char*)(s - 1); |
a0dbb045 | 5438 | /* reverse this character */ |
a0ed51b3 LW |
5439 | while (down > up) { |
5440 | tmp = *up; | |
5441 | *up++ = *down; | |
eb160463 | 5442 | *down-- = (char)tmp; |
a0ed51b3 LW |
5443 | } |
5444 | } | |
5445 | } | |
5446 | up = SvPVX(TARG); | |
5447 | } | |
a0d0e21e LW |
5448 | down = SvPVX(TARG) + len - 1; |
5449 | while (down > up) { | |
5450 | tmp = *up; | |
5451 | *up++ = *down; | |
eb160463 | 5452 | *down-- = (char)tmp; |
a0d0e21e | 5453 | } |
3aa33fe5 | 5454 | (void)SvPOK_only_UTF8(TARG); |
79072805 | 5455 | } |
a0d0e21e LW |
5456 | SP = MARK + 1; |
5457 | SETTARG; | |
79072805 | 5458 | } |
a0d0e21e | 5459 | RETURN; |
79072805 LW |
5460 | } |
5461 | ||
a0d0e21e | 5462 | PP(pp_split) |
79072805 | 5463 | { |
27da23d5 | 5464 | dVAR; dSP; dTARG; |
a0d0e21e | 5465 | AV *ary; |
467f0320 | 5466 | register IV limit = POPi; /* note, negative is forever */ |
1b6737cc | 5467 | SV * const sv = POPs; |
a0d0e21e | 5468 | STRLEN len; |
727b7506 | 5469 | register const char *s = SvPV_const(sv, len); |
1b6737cc | 5470 | const bool do_utf8 = DO_UTF8(sv); |
727b7506 | 5471 | const char *strend = s + len; |
44a8e56a | 5472 | register PMOP *pm; |
d9f97599 | 5473 | register REGEXP *rx; |
a0d0e21e | 5474 | register SV *dstr; |
727b7506 | 5475 | register const char *m; |
a0d0e21e | 5476 | I32 iters = 0; |
bb7a0f54 | 5477 | const STRLEN slen = do_utf8 ? utf8_length((U8*)s, (U8*)strend) : (STRLEN)(strend - s); |
792b2c16 | 5478 | I32 maxiters = slen + 10; |
c1a7495a | 5479 | I32 trailing_empty = 0; |
727b7506 | 5480 | const char *orig; |
1b6737cc | 5481 | const I32 origlimit = limit; |
a0d0e21e LW |
5482 | I32 realarray = 0; |
5483 | I32 base; | |
f54cb97a | 5484 | const I32 gimme = GIMME_V; |
941446f6 | 5485 | bool gimme_scalar; |
f54cb97a | 5486 | const I32 oldsave = PL_savestack_ix; |
437d3b4e | 5487 | U32 make_mortal = SVs_TEMP; |
7fba1cd6 | 5488 | bool multiline = 0; |
b37c2d43 | 5489 | MAGIC *mg = NULL; |
79072805 | 5490 | |
44a8e56a | 5491 | #ifdef DEBUGGING |
5492 | Copy(&LvTARGOFF(POPs), &pm, 1, PMOP*); | |
5493 | #else | |
5494 | pm = (PMOP*)POPs; | |
5495 | #endif | |
a0d0e21e | 5496 | if (!pm || !s) |
2269b42e | 5497 | DIE(aTHX_ "panic: pp_split"); |
aaa362c4 | 5498 | rx = PM_GETRE(pm); |
bbce6d69 | 5499 | |
07bc277f NC |
5500 | TAINT_IF((RX_EXTFLAGS(rx) & RXf_PMf_LOCALE) && |
5501 | (RX_EXTFLAGS(rx) & (RXf_WHITE | RXf_SKIPWHITE))); | |
bbce6d69 | 5502 | |
a30b2f1f | 5503 | RX_MATCH_UTF8_set(rx, do_utf8); |
d9f424b2 | 5504 | |
971a9dd3 | 5505 | #ifdef USE_ITHREADS |
20e98b0f | 5506 | if (pm->op_pmreplrootu.op_pmtargetoff) { |
159b6efe | 5507 | ary = GvAVn(MUTABLE_GV(PAD_SVl(pm->op_pmreplrootu.op_pmtargetoff))); |
20e98b0f | 5508 | } |
971a9dd3 | 5509 | #else |
20e98b0f NC |
5510 | if (pm->op_pmreplrootu.op_pmtargetgv) { |
5511 | ary = GvAVn(pm->op_pmreplrootu.op_pmtargetgv); | |
971a9dd3 | 5512 | } |
20e98b0f | 5513 | #endif |
79072805 | 5514 | else |
7d49f689 | 5515 | ary = NULL; |
a0d0e21e LW |
5516 | if (ary && (gimme != G_ARRAY || (pm->op_pmflags & PMf_ONCE))) { |
5517 | realarray = 1; | |
8ec5e241 | 5518 | PUTBACK; |
a0d0e21e LW |
5519 | av_extend(ary,0); |
5520 | av_clear(ary); | |
8ec5e241 | 5521 | SPAGAIN; |
ad64d0ec | 5522 | if ((mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied))) { |
8ec5e241 | 5523 | PUSHMARK(SP); |
ad64d0ec | 5524 | XPUSHs(SvTIED_obj(MUTABLE_SV(ary), mg)); |
8ec5e241 NIS |
5525 | } |
5526 | else { | |
1c0b011c | 5527 | if (!AvREAL(ary)) { |
1b6737cc | 5528 | I32 i; |
1c0b011c | 5529 | AvREAL_on(ary); |
abff13bb | 5530 | AvREIFY_off(ary); |
1c0b011c | 5531 | for (i = AvFILLp(ary); i >= 0; i--) |
3280af22 | 5532 | AvARRAY(ary)[i] = &PL_sv_undef; /* don't free mere refs */ |
1c0b011c NIS |
5533 | } |
5534 | /* temporarily switch stacks */ | |
8b7059b1 | 5535 | SAVESWITCHSTACK(PL_curstack, ary); |
8ec5e241 | 5536 | make_mortal = 0; |
1c0b011c | 5537 | } |
79072805 | 5538 | } |
3280af22 | 5539 | base = SP - PL_stack_base; |
a0d0e21e | 5540 | orig = s; |
07bc277f | 5541 | if (RX_EXTFLAGS(rx) & RXf_SKIPWHITE) { |
613f191e TS |
5542 | if (do_utf8) { |
5543 | while (*s == ' ' || is_utf8_space((U8*)s)) | |
5544 | s += UTF8SKIP(s); | |
5545 | } | |
07bc277f | 5546 | else if (RX_EXTFLAGS(rx) & RXf_PMf_LOCALE) { |
bbce6d69 | 5547 | while (isSPACE_LC(*s)) |
5548 | s++; | |
5549 | } | |
5550 | else { | |
5551 | while (isSPACE(*s)) | |
5552 | s++; | |
5553 | } | |
a0d0e21e | 5554 | } |
07bc277f | 5555 | if (RX_EXTFLAGS(rx) & PMf_MULTILINE) { |
7fba1cd6 | 5556 | multiline = 1; |
c07a80fd | 5557 | } |
5558 | ||
941446f6 FC |
5559 | gimme_scalar = gimme == G_SCALAR && !ary; |
5560 | ||
a0d0e21e LW |
5561 | if (!limit) |
5562 | limit = maxiters + 2; | |
07bc277f | 5563 | if (RX_EXTFLAGS(rx) & RXf_WHITE) { |
a0d0e21e | 5564 | while (--limit) { |
bbce6d69 | 5565 | m = s; |
8727f688 YO |
5566 | /* this one uses 'm' and is a negative test */ |
5567 | if (do_utf8) { | |
613f191e TS |
5568 | while (m < strend && !( *m == ' ' || is_utf8_space((U8*)m) )) { |
5569 | const int t = UTF8SKIP(m); | |
5570 | /* is_utf8_space returns FALSE for malform utf8 */ | |
5571 | if (strend - m < t) | |
5572 | m = strend; | |
5573 | else | |
5574 | m += t; | |
5575 | } | |
07bc277f | 5576 | } else if (RX_EXTFLAGS(rx) & RXf_PMf_LOCALE) { |
8727f688 YO |
5577 | while (m < strend && !isSPACE_LC(*m)) |
5578 | ++m; | |
5579 | } else { | |
5580 | while (m < strend && !isSPACE(*m)) | |
5581 | ++m; | |
5582 | } | |
a0d0e21e LW |
5583 | if (m >= strend) |
5584 | break; | |
bbce6d69 | 5585 | |
c1a7495a BB |
5586 | if (gimme_scalar) { |
5587 | iters++; | |
5588 | if (m-s == 0) | |
5589 | trailing_empty++; | |
5590 | else | |
5591 | trailing_empty = 0; | |
5592 | } else { | |
5593 | dstr = newSVpvn_flags(s, m-s, | |
5594 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5595 | XPUSHs(dstr); | |
5596 | } | |
bbce6d69 | 5597 | |
613f191e TS |
5598 | /* skip the whitespace found last */ |
5599 | if (do_utf8) | |
5600 | s = m + UTF8SKIP(m); | |
5601 | else | |
5602 | s = m + 1; | |
5603 | ||
8727f688 YO |
5604 | /* this one uses 's' and is a positive test */ |
5605 | if (do_utf8) { | |
613f191e | 5606 | while (s < strend && ( *s == ' ' || is_utf8_space((U8*)s) )) |
8727f688 | 5607 | s += UTF8SKIP(s); |
07bc277f | 5608 | } else if (RX_EXTFLAGS(rx) & RXf_PMf_LOCALE) { |
8727f688 YO |
5609 | while (s < strend && isSPACE_LC(*s)) |
5610 | ++s; | |
5611 | } else { | |
5612 | while (s < strend && isSPACE(*s)) | |
5613 | ++s; | |
5614 | } | |
79072805 LW |
5615 | } |
5616 | } | |
07bc277f | 5617 | else if (RX_EXTFLAGS(rx) & RXf_START_ONLY) { |
a0d0e21e | 5618 | while (--limit) { |
a6e20a40 AL |
5619 | for (m = s; m < strend && *m != '\n'; m++) |
5620 | ; | |
a0d0e21e LW |
5621 | m++; |
5622 | if (m >= strend) | |
5623 | break; | |
c1a7495a BB |
5624 | |
5625 | if (gimme_scalar) { | |
5626 | iters++; | |
5627 | if (m-s == 0) | |
5628 | trailing_empty++; | |
5629 | else | |
5630 | trailing_empty = 0; | |
5631 | } else { | |
5632 | dstr = newSVpvn_flags(s, m-s, | |
5633 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5634 | XPUSHs(dstr); | |
5635 | } | |
a0d0e21e LW |
5636 | s = m; |
5637 | } | |
5638 | } | |
07bc277f | 5639 | else if (RX_EXTFLAGS(rx) & RXf_NULL && !(s >= strend)) { |
640f820d AB |
5640 | /* |
5641 | Pre-extend the stack, either the number of bytes or | |
5642 | characters in the string or a limited amount, triggered by: | |
5643 | ||
5644 | my ($x, $y) = split //, $str; | |
5645 | or | |
5646 | split //, $str, $i; | |
5647 | */ | |
c1a7495a BB |
5648 | if (!gimme_scalar) { |
5649 | const U32 items = limit - 1; | |
5650 | if (items < slen) | |
5651 | EXTEND(SP, items); | |
5652 | else | |
5653 | EXTEND(SP, slen); | |
5654 | } | |
640f820d | 5655 | |
e9515b0f AB |
5656 | if (do_utf8) { |
5657 | while (--limit) { | |
5658 | /* keep track of how many bytes we skip over */ | |
5659 | m = s; | |
640f820d | 5660 | s += UTF8SKIP(s); |
c1a7495a BB |
5661 | if (gimme_scalar) { |
5662 | iters++; | |
5663 | if (s-m == 0) | |
5664 | trailing_empty++; | |
5665 | else | |
5666 | trailing_empty = 0; | |
5667 | } else { | |
5668 | dstr = newSVpvn_flags(m, s-m, SVf_UTF8 | make_mortal); | |
640f820d | 5669 | |
c1a7495a BB |
5670 | PUSHs(dstr); |
5671 | } | |
640f820d | 5672 | |
e9515b0f AB |
5673 | if (s >= strend) |
5674 | break; | |
5675 | } | |
5676 | } else { | |
5677 | while (--limit) { | |
c1a7495a BB |
5678 | if (gimme_scalar) { |
5679 | iters++; | |
5680 | } else { | |
5681 | dstr = newSVpvn(s, 1); | |
e9515b0f | 5682 | |
e9515b0f | 5683 | |
c1a7495a BB |
5684 | if (make_mortal) |
5685 | sv_2mortal(dstr); | |
640f820d | 5686 | |
c1a7495a BB |
5687 | PUSHs(dstr); |
5688 | } | |
5689 | ||
5690 | s++; | |
e9515b0f AB |
5691 | |
5692 | if (s >= strend) | |
5693 | break; | |
5694 | } | |
640f820d AB |
5695 | } |
5696 | } | |
3c8556c3 | 5697 | else if (do_utf8 == (RX_UTF8(rx) != 0) && |
07bc277f NC |
5698 | (RX_EXTFLAGS(rx) & RXf_USE_INTUIT) && !RX_NPARENS(rx) |
5699 | && (RX_EXTFLAGS(rx) & RXf_CHECK_ALL) | |
5700 | && !(RX_EXTFLAGS(rx) & RXf_ANCH)) { | |
5701 | const int tail = (RX_EXTFLAGS(rx) & RXf_INTUIT_TAIL); | |
f9f4320a | 5702 | SV * const csv = CALLREG_INTUIT_STRING(rx); |
cf93c79d | 5703 | |
07bc277f | 5704 | len = RX_MINLENRET(rx); |
3c8556c3 | 5705 | if (len == 1 && !RX_UTF8(rx) && !tail) { |
1b6737cc | 5706 | const char c = *SvPV_nolen_const(csv); |
a0d0e21e | 5707 | while (--limit) { |
a6e20a40 AL |
5708 | for (m = s; m < strend && *m != c; m++) |
5709 | ; | |
a0d0e21e LW |
5710 | if (m >= strend) |
5711 | break; | |
c1a7495a BB |
5712 | if (gimme_scalar) { |
5713 | iters++; | |
5714 | if (m-s == 0) | |
5715 | trailing_empty++; | |
5716 | else | |
5717 | trailing_empty = 0; | |
5718 | } else { | |
5719 | dstr = newSVpvn_flags(s, m-s, | |
5720 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5721 | XPUSHs(dstr); | |
5722 | } | |
93f04dac JH |
5723 | /* The rx->minlen is in characters but we want to step |
5724 | * s ahead by bytes. */ | |
1aa99e6b IH |
5725 | if (do_utf8) |
5726 | s = (char*)utf8_hop((U8*)m, len); | |
5727 | else | |
5728 | s = m + len; /* Fake \n at the end */ | |
a0d0e21e LW |
5729 | } |
5730 | } | |
5731 | else { | |
a0d0e21e | 5732 | while (s < strend && --limit && |
f722798b | 5733 | (m = fbm_instr((unsigned char*)s, (unsigned char*)strend, |
7fba1cd6 | 5734 | csv, multiline ? FBMrf_MULTILINE : 0)) ) |
a0d0e21e | 5735 | { |
c1a7495a BB |
5736 | if (gimme_scalar) { |
5737 | iters++; | |
5738 | if (m-s == 0) | |
5739 | trailing_empty++; | |
5740 | else | |
5741 | trailing_empty = 0; | |
5742 | } else { | |
5743 | dstr = newSVpvn_flags(s, m-s, | |
5744 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5745 | XPUSHs(dstr); | |
5746 | } | |
93f04dac JH |
5747 | /* The rx->minlen is in characters but we want to step |
5748 | * s ahead by bytes. */ | |
1aa99e6b IH |
5749 | if (do_utf8) |
5750 | s = (char*)utf8_hop((U8*)m, len); | |
5751 | else | |
5752 | s = m + len; /* Fake \n at the end */ | |
a0d0e21e | 5753 | } |
463ee0b2 | 5754 | } |
463ee0b2 | 5755 | } |
a0d0e21e | 5756 | else { |
07bc277f | 5757 | maxiters += slen * RX_NPARENS(rx); |
080c2dec | 5758 | while (s < strend && --limit) |
bbce6d69 | 5759 | { |
1b6737cc | 5760 | I32 rex_return; |
080c2dec | 5761 | PUTBACK; |
f9f4320a | 5762 | rex_return = CALLREGEXEC(rx, (char*)s, (char*)strend, (char*)orig, 1 , |
727b7506 | 5763 | sv, NULL, 0); |
080c2dec | 5764 | SPAGAIN; |
1b6737cc | 5765 | if (rex_return == 0) |
080c2dec | 5766 | break; |
d9f97599 | 5767 | TAINT_IF(RX_MATCH_TAINTED(rx)); |
07bc277f | 5768 | if (RX_MATCH_COPIED(rx) && RX_SUBBEG(rx) != orig) { |
a0d0e21e LW |
5769 | m = s; |
5770 | s = orig; | |
07bc277f | 5771 | orig = RX_SUBBEG(rx); |
a0d0e21e LW |
5772 | s = orig + (m - s); |
5773 | strend = s + (strend - m); | |
5774 | } | |
07bc277f | 5775 | m = RX_OFFS(rx)[0].start + orig; |
c1a7495a BB |
5776 | |
5777 | if (gimme_scalar) { | |
5778 | iters++; | |
5779 | if (m-s == 0) | |
5780 | trailing_empty++; | |
5781 | else | |
5782 | trailing_empty = 0; | |
5783 | } else { | |
5784 | dstr = newSVpvn_flags(s, m-s, | |
5785 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5786 | XPUSHs(dstr); | |
5787 | } | |
07bc277f | 5788 | if (RX_NPARENS(rx)) { |
1b6737cc | 5789 | I32 i; |
07bc277f NC |
5790 | for (i = 1; i <= (I32)RX_NPARENS(rx); i++) { |
5791 | s = RX_OFFS(rx)[i].start + orig; | |
5792 | m = RX_OFFS(rx)[i].end + orig; | |
6de67870 JP |
5793 | |
5794 | /* japhy (07/27/01) -- the (m && s) test doesn't catch | |
5795 | parens that didn't match -- they should be set to | |
5796 | undef, not the empty string */ | |
c1a7495a BB |
5797 | if (gimme_scalar) { |
5798 | iters++; | |
5799 | if (m-s == 0) | |
5800 | trailing_empty++; | |
5801 | else | |
5802 | trailing_empty = 0; | |
5803 | } else { | |
5804 | if (m >= orig && s >= orig) { | |
5805 | dstr = newSVpvn_flags(s, m-s, | |
5806 | (do_utf8 ? SVf_UTF8 : 0) | |
5807 | | make_mortal); | |
5808 | } | |
5809 | else | |
5810 | dstr = &PL_sv_undef; /* undef, not "" */ | |
5811 | XPUSHs(dstr); | |
748a9306 | 5812 | } |
c1a7495a | 5813 | |
a0d0e21e LW |
5814 | } |
5815 | } | |
07bc277f | 5816 | s = RX_OFFS(rx)[0].end + orig; |
a0d0e21e | 5817 | } |
79072805 | 5818 | } |
8ec5e241 | 5819 | |
c1a7495a BB |
5820 | if (!gimme_scalar) { |
5821 | iters = (SP - PL_stack_base) - base; | |
5822 | } | |
a0d0e21e | 5823 | if (iters > maxiters) |
cea2e8a9 | 5824 | DIE(aTHX_ "Split loop"); |
8ec5e241 | 5825 | |
a0d0e21e LW |
5826 | /* keep field after final delim? */ |
5827 | if (s < strend || (iters && origlimit)) { | |
c1a7495a BB |
5828 | if (!gimme_scalar) { |
5829 | const STRLEN l = strend - s; | |
5830 | dstr = newSVpvn_flags(s, l, (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5831 | XPUSHs(dstr); | |
5832 | } | |
a0d0e21e | 5833 | iters++; |
79072805 | 5834 | } |
a0d0e21e | 5835 | else if (!origlimit) { |
c1a7495a BB |
5836 | if (gimme_scalar) { |
5837 | iters -= trailing_empty; | |
5838 | } else { | |
5839 | while (iters > 0 && (!TOPs || !SvANY(TOPs) || SvCUR(TOPs) == 0)) { | |
5840 | if (TOPs && !make_mortal) | |
5841 | sv_2mortal(TOPs); | |
5842 | *SP-- = &PL_sv_undef; | |
5843 | iters--; | |
5844 | } | |
89900bd3 | 5845 | } |
a0d0e21e | 5846 | } |
8ec5e241 | 5847 | |
8b7059b1 DM |
5848 | PUTBACK; |
5849 | LEAVE_SCOPE(oldsave); /* may undo an earlier SWITCHSTACK */ | |
5850 | SPAGAIN; | |
a0d0e21e | 5851 | if (realarray) { |
8ec5e241 | 5852 | if (!mg) { |
1c0b011c NIS |
5853 | if (SvSMAGICAL(ary)) { |
5854 | PUTBACK; | |
ad64d0ec | 5855 | mg_set(MUTABLE_SV(ary)); |
1c0b011c NIS |
5856 | SPAGAIN; |
5857 | } | |
5858 | if (gimme == G_ARRAY) { | |
5859 | EXTEND(SP, iters); | |
5860 | Copy(AvARRAY(ary), SP + 1, iters, SV*); | |
5861 | SP += iters; | |
5862 | RETURN; | |
5863 | } | |
8ec5e241 | 5864 | } |
1c0b011c | 5865 | else { |
fb73857a | 5866 | PUTBACK; |
d343c3ef | 5867 | ENTER_with_name("call_PUSH"); |
864dbfa3 | 5868 | call_method("PUSH",G_SCALAR|G_DISCARD); |
d343c3ef | 5869 | LEAVE_with_name("call_PUSH"); |
fb73857a | 5870 | SPAGAIN; |
8ec5e241 | 5871 | if (gimme == G_ARRAY) { |
1b6737cc | 5872 | I32 i; |
8ec5e241 NIS |
5873 | /* EXTEND should not be needed - we just popped them */ |
5874 | EXTEND(SP, iters); | |
5875 | for (i=0; i < iters; i++) { | |
5876 | SV **svp = av_fetch(ary, i, FALSE); | |
3280af22 | 5877 | PUSHs((svp) ? *svp : &PL_sv_undef); |
8ec5e241 | 5878 | } |
1c0b011c NIS |
5879 | RETURN; |
5880 | } | |
a0d0e21e LW |
5881 | } |
5882 | } | |
5883 | else { | |
5884 | if (gimme == G_ARRAY) | |
5885 | RETURN; | |
5886 | } | |
7f18b612 YST |
5887 | |
5888 | GETTARGET; | |
5889 | PUSHi(iters); | |
5890 | RETURN; | |
79072805 | 5891 | } |
85e6fe83 | 5892 | |
c5917253 NC |
5893 | PP(pp_once) |
5894 | { | |
5895 | dSP; | |
5896 | SV *const sv = PAD_SVl(PL_op->op_targ); | |
5897 | ||
5898 | if (SvPADSTALE(sv)) { | |
5899 | /* First time. */ | |
5900 | SvPADSTALE_off(sv); | |
5901 | RETURNOP(cLOGOP->op_other); | |
5902 | } | |
5903 | RETURNOP(cLOGOP->op_next); | |
5904 | } | |
5905 | ||
c0329465 MB |
5906 | PP(pp_lock) |
5907 | { | |
97aff369 | 5908 | dVAR; |
39644a26 | 5909 | dSP; |
c0329465 | 5910 | dTOPss; |
e55aaa0e | 5911 | SV *retsv = sv; |
076a2a80 | 5912 | assert(SvTYPE(retsv) != SVt_PVCV); |
68795e93 | 5913 | SvLOCK(sv); |
076a2a80 | 5914 | if (SvTYPE(retsv) == SVt_PVAV || SvTYPE(retsv) == SVt_PVHV) { |
e55aaa0e MB |
5915 | retsv = refto(retsv); |
5916 | } | |
5917 | SETs(retsv); | |
c0329465 MB |
5918 | RETURN; |
5919 | } | |
a863c7d1 | 5920 | |
65bca31a NC |
5921 | |
5922 | PP(unimplemented_op) | |
5923 | { | |
97aff369 | 5924 | dVAR; |
65bca31a NC |
5925 | DIE(aTHX_ "panic: unimplemented op %s (#%d) called", OP_NAME(PL_op), |
5926 | PL_op->op_type); | |
bb4c52e0 | 5927 | return NORMAL; |
65bca31a NC |
5928 | } |
5929 | ||
867fa1e2 YO |
5930 | PP(pp_boolkeys) |
5931 | { | |
5932 | dVAR; | |
5933 | dSP; | |
5934 | HV * const hv = (HV*)POPs; | |
5935 | ||
5936 | if (SvRMAGICAL(hv)) { | |
5937 | MAGIC * const mg = mg_find((SV*)hv, PERL_MAGIC_tied); | |
5938 | if (mg) { | |
5939 | XPUSHs(magic_scalarpack(hv, mg)); | |
5940 | RETURN; | |
5941 | } | |
5942 | } | |
5943 | ||
5944 | XPUSHs(boolSV(HvKEYS(hv) != 0)); | |
5945 | RETURN; | |
5946 | } | |
5947 | ||
e609e586 NC |
5948 | /* |
5949 | * Local variables: | |
5950 | * c-indentation-style: bsd | |
5951 | * c-basic-offset: 4 | |
5952 | * indent-tabs-mode: t | |
5953 | * End: | |
5954 | * | |
37442d52 RGS |
5955 | * ex: set ts=8 sts=4 sw=4 noet: |
5956 | */ |