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 | 31 | #include "reentr.h" |
685289b5 | 32 | #include "regcharclass.h" |
a4af207c | 33 | |
dfe9444c AD |
34 | /* XXX I can't imagine anyone who doesn't have this actually _needs_ |
35 | it, since pid_t is an integral type. | |
36 | --AD 2/20/1998 | |
37 | */ | |
38 | #ifdef NEED_GETPID_PROTO | |
39 | extern Pid_t getpid (void); | |
8ac85365 NIS |
40 | #endif |
41 | ||
0630166f SP |
42 | /* |
43 | * Some BSDs and Cygwin default to POSIX math instead of IEEE. | |
44 | * This switches them over to IEEE. | |
45 | */ | |
46 | #if defined(LIBM_LIB_VERSION) | |
47 | _LIB_VERSION_TYPE _LIB_VERSION = _IEEE_; | |
48 | #endif | |
49 | ||
a78bc3c6 KW |
50 | static const STRLEN small_mu_len = sizeof(GREEK_SMALL_LETTER_MU_UTF8) - 1; |
51 | static const STRLEN capital_iota_len = sizeof(GREEK_CAPITAL_LETTER_IOTA_UTF8) - 1; | |
52 | ||
13017935 SM |
53 | /* variations on pp_null */ |
54 | ||
93a17b20 LW |
55 | PP(pp_stub) |
56 | { | |
39644a26 | 57 | dSP; |
54310121 | 58 | if (GIMME_V == G_SCALAR) |
3280af22 | 59 | XPUSHs(&PL_sv_undef); |
93a17b20 LW |
60 | RETURN; |
61 | } | |
62 | ||
79072805 LW |
63 | /* Pushy stuff. */ |
64 | ||
bdaf10a5 | 65 | /* This is also called directly by pp_lvavref. */ |
93a17b20 LW |
66 | PP(pp_padav) |
67 | { | |
20b7effb | 68 | dSP; dTARGET; |
1c23e2bd | 69 | U8 gimme; |
e190e9b4 | 70 | assert(SvTYPE(TARG) == SVt_PVAV); |
3dbcc5e0 S |
71 | if (UNLIKELY( PL_op->op_private & OPpLVAL_INTRO )) |
72 | if (LIKELY( !(PL_op->op_private & OPpPAD_STATE) )) | |
a5911867 | 73 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); |
85e6fe83 | 74 | EXTEND(SP, 1); |
a46a7b6e | 75 | |
533c011a | 76 | if (PL_op->op_flags & OPf_REF) { |
85e6fe83 | 77 | PUSHs(TARG); |
93a17b20 | 78 | RETURN; |
a46a7b6e DM |
79 | } |
80 | else if (PL_op->op_private & OPpMAYBE_LVSUB) { | |
81 | const I32 flags = is_lvalue_sub(); | |
82 | if (flags && !(flags & OPpENTERSUB_INARGS)) { | |
83 | if (GIMME_V == G_SCALAR) | |
84 | /* diag_listed_as: Can't return %s to lvalue scalar context */ | |
85 | Perl_croak(aTHX_ "Can't return array to lvalue scalar context"); | |
86 | PUSHs(TARG); | |
87 | RETURN; | |
40c94d11 | 88 | } |
85e6fe83 | 89 | } |
a46a7b6e | 90 | |
13017935 SM |
91 | gimme = GIMME_V; |
92 | if (gimme == G_ARRAY) { | |
d5524600 | 93 | /* XXX see also S_pushav in pp_hot.c */ |
052a7c76 | 94 | const SSize_t maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
85e6fe83 | 95 | EXTEND(SP, maxarg); |
93965878 | 96 | if (SvMAGICAL(TARG)) { |
052a7c76 | 97 | SSize_t i; |
c70927a6 | 98 | for (i=0; i < maxarg; i++) { |
502c6561 | 99 | SV * const * const svp = av_fetch(MUTABLE_AV(TARG), i, FALSE); |
3280af22 | 100 | SP[i+1] = (svp) ? *svp : &PL_sv_undef; |
93965878 NIS |
101 | } |
102 | } | |
103 | else { | |
052a7c76 DM |
104 | SSize_t i; |
105 | for (i=0; i < maxarg; i++) { | |
428ccf1e FC |
106 | SV * const sv = AvARRAY((const AV *)TARG)[i]; |
107 | SP[i+1] = sv ? sv : &PL_sv_undef; | |
108 | } | |
93965878 | 109 | } |
85e6fe83 LW |
110 | SP += maxarg; |
111 | } | |
13017935 | 112 | else if (gimme == G_SCALAR) { |
1b6737cc | 113 | SV* const sv = sv_newmortal(); |
c70927a6 | 114 | const SSize_t maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
85e6fe83 LW |
115 | sv_setiv(sv, maxarg); |
116 | PUSHs(sv); | |
117 | } | |
118 | RETURN; | |
93a17b20 LW |
119 | } |
120 | ||
121 | PP(pp_padhv) | |
122 | { | |
20b7effb | 123 | dSP; dTARGET; |
1c23e2bd | 124 | U8 gimme; |
54310121 | 125 | |
e190e9b4 | 126 | assert(SvTYPE(TARG) == SVt_PVHV); |
93a17b20 | 127 | XPUSHs(TARG); |
3dbcc5e0 S |
128 | if (UNLIKELY( PL_op->op_private & OPpLVAL_INTRO )) |
129 | if (LIKELY( !(PL_op->op_private & OPpPAD_STATE) )) | |
a5911867 | 130 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); |
a46a7b6e | 131 | |
533c011a | 132 | if (PL_op->op_flags & OPf_REF) |
93a17b20 | 133 | RETURN; |
40c94d11 | 134 | else if (PL_op->op_private & OPpMAYBE_LVSUB) { |
a46a7b6e DM |
135 | const I32 flags = is_lvalue_sub(); |
136 | if (flags && !(flags & OPpENTERSUB_INARGS)) { | |
137 | if (GIMME_V == G_SCALAR) | |
138 | /* diag_listed_as: Can't return %s to lvalue scalar context */ | |
139 | Perl_croak(aTHX_ "Can't return hash to lvalue scalar context"); | |
140 | RETURN; | |
141 | } | |
78f9721b | 142 | } |
a46a7b6e | 143 | |
54310121 | 144 | gimme = GIMME_V; |
145 | if (gimme == G_ARRAY) { | |
981b7185 | 146 | RETURNOP(Perl_do_kv(aTHX)); |
85e6fe83 | 147 | } |
c8fe3bdf | 148 | else if ((PL_op->op_private & OPpTRUEBOOL |
adc42c31 | 149 | || ( PL_op->op_private & OPpMAYBE_TRUEBOOL |
c8fe3bdf | 150 | && block_gimme() == G_VOID )) |
a46a7b6e DM |
151 | && (!SvRMAGICAL(TARG) || !mg_find(TARG, PERL_MAGIC_tied)) |
152 | ) | |
b243b193 | 153 | SETs(HvUSEDKEYS(TARG) ? &PL_sv_yes : &PL_sv_no); |
54310121 | 154 | else if (gimme == G_SCALAR) { |
85fbaab2 | 155 | SV* const sv = Perl_hv_scalar(aTHX_ MUTABLE_HV(TARG)); |
85e6fe83 | 156 | SETs(sv); |
85e6fe83 | 157 | } |
54310121 | 158 | RETURN; |
93a17b20 LW |
159 | } |
160 | ||
ac217057 FC |
161 | PP(pp_padcv) |
162 | { | |
20b7effb | 163 | dSP; dTARGET; |
97b03d64 FC |
164 | assert(SvTYPE(TARG) == SVt_PVCV); |
165 | XPUSHs(TARG); | |
166 | RETURN; | |
ac217057 FC |
167 | } |
168 | ||
ecf9c8b7 FC |
169 | PP(pp_introcv) |
170 | { | |
20b7effb | 171 | dTARGET; |
6d5c2147 FC |
172 | SvPADSTALE_off(TARG); |
173 | return NORMAL; | |
ecf9c8b7 FC |
174 | } |
175 | ||
13f89586 FC |
176 | PP(pp_clonecv) |
177 | { | |
20b7effb | 178 | dTARGET; |
0f94cb1f FC |
179 | CV * const protocv = PadnamePROTOCV( |
180 | PadlistNAMESARRAY(CvPADLIST(find_runcv(NULL)))[ARGTARG] | |
181 | ); | |
6d5c2147 | 182 | assert(SvTYPE(TARG) == SVt_PVCV); |
0f94cb1f FC |
183 | assert(protocv); |
184 | if (CvISXSUB(protocv)) { /* constant */ | |
6d5c2147 | 185 | /* XXX Should we clone it here? */ |
6d5c2147 FC |
186 | /* If this changes to use SAVECLEARSV, we can move the SAVECLEARSV |
187 | to introcv and remove the SvPADSTALE_off. */ | |
188 | SAVEPADSVANDMORTALIZE(ARGTARG); | |
0f94cb1f | 189 | PAD_SVl(ARGTARG) = SvREFCNT_inc_simple_NN(protocv); |
6d5c2147 FC |
190 | } |
191 | else { | |
0f94cb1f FC |
192 | if (CvROOT(protocv)) { |
193 | assert(CvCLONE(protocv)); | |
194 | assert(!CvCLONED(protocv)); | |
6d5c2147 | 195 | } |
0f94cb1f | 196 | cv_clone_into(protocv,(CV *)TARG); |
6d5c2147 FC |
197 | SAVECLEARSV(PAD_SVl(ARGTARG)); |
198 | } | |
199 | return NORMAL; | |
13f89586 FC |
200 | } |
201 | ||
79072805 LW |
202 | /* Translations. */ |
203 | ||
6f7909da FC |
204 | /* In some cases this function inspects PL_op. If this function is called |
205 | for new op types, more bool parameters may need to be added in place of | |
206 | the checks. | |
207 | ||
208 | When noinit is true, the absence of a gv will cause a retval of undef. | |
209 | This is unrelated to the cv-to-gv assignment case. | |
6f7909da FC |
210 | */ |
211 | ||
212 | static SV * | |
213 | S_rv2gv(pTHX_ SV *sv, const bool vivify_sv, const bool strict, | |
214 | const bool noinit) | |
215 | { | |
f64c9ac5 | 216 | if (!isGV(sv) || SvFAKE(sv)) SvGETMAGIC(sv); |
ed6116ce | 217 | if (SvROK(sv)) { |
93d7320b DM |
218 | if (SvAMAGIC(sv)) { |
219 | sv = amagic_deref_call(sv, to_gv_amg); | |
93d7320b | 220 | } |
e4a1664f | 221 | wasref: |
ed6116ce | 222 | sv = SvRV(sv); |
b1dadf13 | 223 | if (SvTYPE(sv) == SVt_PVIO) { |
159b6efe | 224 | GV * const gv = MUTABLE_GV(sv_newmortal()); |
885f468a | 225 | gv_init(gv, 0, "__ANONIO__", 10, 0); |
a45c7426 | 226 | GvIOp(gv) = MUTABLE_IO(sv); |
b37c2d43 | 227 | SvREFCNT_inc_void_NN(sv); |
ad64d0ec | 228 | sv = MUTABLE_SV(gv); |
ef54e1a4 | 229 | } |
81d52ecd JH |
230 | else if (!isGV_with_GP(sv)) { |
231 | Perl_die(aTHX_ "Not a GLOB reference"); | |
232 | } | |
79072805 LW |
233 | } |
234 | else { | |
6e592b3a | 235 | if (!isGV_with_GP(sv)) { |
f132ae69 | 236 | if (!SvOK(sv)) { |
b13b2135 | 237 | /* If this is a 'my' scalar and flag is set then vivify |
853846ea | 238 | * NI-S 1999/05/07 |
b13b2135 | 239 | */ |
f132ae69 | 240 | if (vivify_sv && sv != &PL_sv_undef) { |
2c8ac474 | 241 | GV *gv; |
ce74145d | 242 | if (SvREADONLY(sv)) |
cb077ed2 | 243 | Perl_croak_no_modify(); |
2c8ac474 | 244 | if (cUNOP->op_targ) { |
0bd48802 | 245 | SV * const namesv = PAD_SV(cUNOP->op_targ); |
94e7eb6f FC |
246 | HV *stash = CopSTASH(PL_curcop); |
247 | if (SvTYPE(stash) != SVt_PVHV) stash = NULL; | |
159b6efe | 248 | gv = MUTABLE_GV(newSV(0)); |
94e7eb6f | 249 | gv_init_sv(gv, stash, namesv, 0); |
2c8ac474 GS |
250 | } |
251 | else { | |
0bd48802 | 252 | const char * const name = CopSTASHPV(PL_curcop); |
6b10071b | 253 | gv = newGVgen_flags(name, |
d14578b8 | 254 | HvNAMEUTF8(CopSTASH(PL_curcop)) ? SVf_UTF8 : 0 ); |
7bdb4ff0 | 255 | SvREFCNT_inc_simple_void_NN(gv); |
1d8d4d2a | 256 | } |
43230e26 | 257 | prepare_SV_for_RV(sv); |
ad64d0ec | 258 | SvRV_set(sv, MUTABLE_SV(gv)); |
853846ea | 259 | SvROK_on(sv); |
1d8d4d2a | 260 | SvSETMAGIC(sv); |
853846ea | 261 | goto wasref; |
2c8ac474 | 262 | } |
81d52ecd JH |
263 | if (PL_op->op_flags & OPf_REF || strict) { |
264 | Perl_die(aTHX_ PL_no_usym, "a symbol"); | |
265 | } | |
599cee73 | 266 | if (ckWARN(WARN_UNINITIALIZED)) |
29489e7c | 267 | report_uninit(sv); |
6f7909da | 268 | return &PL_sv_undef; |
a0d0e21e | 269 | } |
6f7909da | 270 | if (noinit) |
35cd451c | 271 | { |
77cb3b01 FC |
272 | if (!(sv = MUTABLE_SV(gv_fetchsv_nomg( |
273 | sv, GV_ADDMG, SVt_PVGV | |
23496c6e | 274 | )))) |
6f7909da | 275 | return &PL_sv_undef; |
35cd451c GS |
276 | } |
277 | else { | |
81d52ecd JH |
278 | if (strict) { |
279 | Perl_die(aTHX_ | |
fedf30e1 | 280 | PL_no_symref_sv, |
81d52ecd JH |
281 | sv, |
282 | (SvPOKp(sv) && SvCUR(sv)>32 ? "..." : ""), | |
283 | "a symbol" | |
284 | ); | |
285 | } | |
e26df76a NC |
286 | if ((PL_op->op_private & (OPpLVAL_INTRO|OPpDONT_INIT_GV)) |
287 | == OPpDONT_INIT_GV) { | |
288 | /* We are the target of a coderef assignment. Return | |
289 | the scalar unchanged, and let pp_sasssign deal with | |
290 | things. */ | |
6f7909da | 291 | return sv; |
e26df76a | 292 | } |
77cb3b01 | 293 | sv = MUTABLE_SV(gv_fetchsv_nomg(sv, GV_ADD, SVt_PVGV)); |
35cd451c | 294 | } |
2acc3314 | 295 | /* FAKE globs in the symbol table cause weird bugs (#77810) */ |
96293f45 | 296 | SvFAKE_off(sv); |
93a17b20 | 297 | } |
79072805 | 298 | } |
8dc99089 | 299 | if (SvFAKE(sv) && !(PL_op->op_private & OPpALLOW_FAKE)) { |
2acc3314 | 300 | SV *newsv = sv_newmortal(); |
5cf4b255 | 301 | sv_setsv_flags(newsv, sv, 0); |
2acc3314 | 302 | SvFAKE_off(newsv); |
d8906c05 | 303 | sv = newsv; |
2acc3314 | 304 | } |
6f7909da FC |
305 | return sv; |
306 | } | |
307 | ||
308 | PP(pp_rv2gv) | |
309 | { | |
20b7effb | 310 | dSP; dTOPss; |
6f7909da FC |
311 | |
312 | sv = S_rv2gv(aTHX_ | |
313 | sv, PL_op->op_private & OPpDEREF, | |
314 | PL_op->op_private & HINT_STRICT_REFS, | |
315 | ((PL_op->op_flags & OPf_SPECIAL) && !(PL_op->op_flags & OPf_MOD)) | |
316 | || PL_op->op_type == OP_READLINE | |
317 | ); | |
d8906c05 FC |
318 | if (PL_op->op_private & OPpLVAL_INTRO) |
319 | save_gp(MUTABLE_GV(sv), !(PL_op->op_flags & OPf_SPECIAL)); | |
320 | SETs(sv); | |
79072805 LW |
321 | RETURN; |
322 | } | |
323 | ||
dc3c76f8 NC |
324 | /* Helper function for pp_rv2sv and pp_rv2av */ |
325 | GV * | |
fe9845cc RB |
326 | Perl_softref2xv(pTHX_ SV *const sv, const char *const what, |
327 | const svtype type, SV ***spp) | |
dc3c76f8 | 328 | { |
dc3c76f8 NC |
329 | GV *gv; |
330 | ||
7918f24d NC |
331 | PERL_ARGS_ASSERT_SOFTREF2XV; |
332 | ||
dc3c76f8 NC |
333 | if (PL_op->op_private & HINT_STRICT_REFS) { |
334 | if (SvOK(sv)) | |
fedf30e1 | 335 | Perl_die(aTHX_ PL_no_symref_sv, sv, |
bf3d870f | 336 | (SvPOKp(sv) && SvCUR(sv)>32 ? "..." : ""), what); |
dc3c76f8 NC |
337 | else |
338 | Perl_die(aTHX_ PL_no_usym, what); | |
339 | } | |
340 | if (!SvOK(sv)) { | |
fd1d9b5c | 341 | if ( |
c8fe3bdf | 342 | PL_op->op_flags & OPf_REF |
fd1d9b5c | 343 | ) |
dc3c76f8 NC |
344 | Perl_die(aTHX_ PL_no_usym, what); |
345 | if (ckWARN(WARN_UNINITIALIZED)) | |
346 | report_uninit(sv); | |
347 | if (type != SVt_PV && GIMME_V == G_ARRAY) { | |
348 | (*spp)--; | |
349 | return NULL; | |
350 | } | |
351 | **spp = &PL_sv_undef; | |
352 | return NULL; | |
353 | } | |
354 | if ((PL_op->op_flags & OPf_SPECIAL) && | |
355 | !(PL_op->op_flags & OPf_MOD)) | |
356 | { | |
77cb3b01 | 357 | if (!(gv = gv_fetchsv_nomg(sv, GV_ADDMG, type))) |
dc3c76f8 NC |
358 | { |
359 | **spp = &PL_sv_undef; | |
360 | return NULL; | |
361 | } | |
362 | } | |
363 | else { | |
77cb3b01 | 364 | gv = gv_fetchsv_nomg(sv, GV_ADD, type); |
dc3c76f8 NC |
365 | } |
366 | return gv; | |
367 | } | |
368 | ||
79072805 LW |
369 | PP(pp_rv2sv) |
370 | { | |
20b7effb | 371 | dSP; dTOPss; |
c445ea15 | 372 | GV *gv = NULL; |
79072805 | 373 | |
9026059d | 374 | SvGETMAGIC(sv); |
ed6116ce | 375 | if (SvROK(sv)) { |
93d7320b DM |
376 | if (SvAMAGIC(sv)) { |
377 | sv = amagic_deref_call(sv, to_sv_amg); | |
93d7320b | 378 | } |
f5284f61 | 379 | |
ed6116ce | 380 | sv = SvRV(sv); |
69f00f67 | 381 | if (SvTYPE(sv) >= SVt_PVAV) |
cea2e8a9 | 382 | DIE(aTHX_ "Not a SCALAR reference"); |
79072805 LW |
383 | } |
384 | else { | |
159b6efe | 385 | gv = MUTABLE_GV(sv); |
748a9306 | 386 | |
6e592b3a | 387 | if (!isGV_with_GP(gv)) { |
dc3c76f8 NC |
388 | gv = Perl_softref2xv(aTHX_ sv, "a SCALAR", SVt_PV, &sp); |
389 | if (!gv) | |
390 | RETURN; | |
463ee0b2 | 391 | } |
29c711a3 | 392 | sv = GvSVn(gv); |
a0d0e21e | 393 | } |
533c011a | 394 | if (PL_op->op_flags & OPf_MOD) { |
82d03984 RGS |
395 | if (PL_op->op_private & OPpLVAL_INTRO) { |
396 | if (cUNOP->op_first->op_type == OP_NULL) | |
159b6efe | 397 | sv = save_scalar(MUTABLE_GV(TOPs)); |
82d03984 RGS |
398 | else if (gv) |
399 | sv = save_scalar(gv); | |
400 | else | |
f1f66076 | 401 | Perl_croak(aTHX_ "%s", PL_no_localize_ref); |
82d03984 | 402 | } |
533c011a | 403 | else if (PL_op->op_private & OPpDEREF) |
9026059d | 404 | sv = vivify_ref(sv, PL_op->op_private & OPpDEREF); |
79072805 | 405 | } |
655f5b26 | 406 | SPAGAIN; /* in case chasing soft refs reallocated the stack */ |
a0d0e21e | 407 | SETs(sv); |
79072805 LW |
408 | RETURN; |
409 | } | |
410 | ||
411 | PP(pp_av2arylen) | |
412 | { | |
20b7effb | 413 | dSP; |
502c6561 | 414 | AV * const av = MUTABLE_AV(TOPs); |
02d85cc3 EB |
415 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
416 | if (lvalue) { | |
8160c8f5 DM |
417 | SV ** const svp = Perl_av_arylen_p(aTHX_ MUTABLE_AV(av)); |
418 | if (!*svp) { | |
419 | *svp = newSV_type(SVt_PVMG); | |
420 | sv_magic(*svp, MUTABLE_SV(av), PERL_MAGIC_arylen, NULL, 0); | |
02d85cc3 | 421 | } |
8160c8f5 | 422 | SETs(*svp); |
02d85cc3 | 423 | } else { |
e1dccc0d | 424 | SETs(sv_2mortal(newSViv(AvFILL(MUTABLE_AV(av))))); |
79072805 | 425 | } |
79072805 LW |
426 | RETURN; |
427 | } | |
428 | ||
a0d0e21e LW |
429 | PP(pp_pos) |
430 | { | |
27a8dde8 | 431 | dSP; dTOPss; |
8ec5e241 | 432 | |
78f9721b | 433 | if (PL_op->op_flags & OPf_MOD || LVRET) { |
d14578b8 | 434 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV));/* Not TARG RT#67838 */ |
16eb5365 FC |
435 | sv_magic(ret, NULL, PERL_MAGIC_pos, NULL, 0); |
436 | LvTYPE(ret) = '.'; | |
437 | LvTARG(ret) = SvREFCNT_inc_simple(sv); | |
27a8dde8 | 438 | SETs(ret); /* no SvSETMAGIC */ |
a0d0e21e LW |
439 | } |
440 | else { | |
96c2a8ff | 441 | const MAGIC * const mg = mg_find_mglob(sv); |
6174b39a | 442 | if (mg && mg->mg_len != -1) { |
2154eca7 | 443 | dTARGET; |
6174b39a | 444 | STRLEN i = mg->mg_len; |
25fdce4a | 445 | if (mg->mg_flags & MGf_BYTES && DO_UTF8(sv)) |
6174b39a | 446 | i = sv_pos_b2u_flags(sv, i, SV_GMAGIC|SV_CONST_RETURN); |
27a8dde8 FC |
447 | SETu(i); |
448 | return NORMAL; | |
a0d0e21e | 449 | } |
27a8dde8 | 450 | SETs(&PL_sv_undef); |
a0d0e21e | 451 | } |
27a8dde8 | 452 | return NORMAL; |
a0d0e21e LW |
453 | } |
454 | ||
79072805 LW |
455 | PP(pp_rv2cv) |
456 | { | |
20b7effb | 457 | dSP; |
79072805 | 458 | GV *gv; |
1eced8f8 | 459 | HV *stash_unused; |
c445ea15 | 460 | const I32 flags = (PL_op->op_flags & OPf_SPECIAL) |
9da346da | 461 | ? GV_ADDMG |
d14578b8 KW |
462 | : ((PL_op->op_private & (OPpLVAL_INTRO|OPpMAY_RETURN_CONSTANT)) |
463 | == OPpMAY_RETURN_CONSTANT) | |
c445ea15 AL |
464 | ? GV_ADD|GV_NOEXPAND |
465 | : GV_ADD; | |
4633a7c4 LW |
466 | /* We usually try to add a non-existent subroutine in case of AUTOLOAD. */ |
467 | /* (But not in defined().) */ | |
e26df76a | 468 | |
1eced8f8 | 469 | CV *cv = sv_2cv(TOPs, &stash_unused, &gv, flags); |
5a20ba3d | 470 | if (cv) NOOP; |
e26df76a | 471 | else if ((flags == (GV_ADD|GV_NOEXPAND)) && gv && SvROK(gv)) { |
2eaf799e FC |
472 | cv = SvTYPE(SvRV(gv)) == SVt_PVCV |
473 | ? MUTABLE_CV(SvRV(gv)) | |
474 | : MUTABLE_CV(gv); | |
e26df76a | 475 | } |
07055b4c | 476 | else |
ea726b52 | 477 | cv = MUTABLE_CV(&PL_sv_undef); |
ad64d0ec | 478 | SETs(MUTABLE_SV(cv)); |
3d79e3ee | 479 | return NORMAL; |
79072805 LW |
480 | } |
481 | ||
c07a80fd | 482 | PP(pp_prototype) |
483 | { | |
20b7effb | 484 | dSP; |
c07a80fd | 485 | CV *cv; |
486 | HV *stash; | |
487 | GV *gv; | |
fabdb6c0 | 488 | SV *ret = &PL_sv_undef; |
c07a80fd | 489 | |
6954f42f | 490 | if (SvGMAGICAL(TOPs)) SETs(sv_mortalcopy(TOPs)); |
b6c543e3 | 491 | if (SvPOK(TOPs) && SvCUR(TOPs) >= 7) { |
e3f73d4e | 492 | const char * s = SvPVX_const(TOPs); |
b6c543e3 | 493 | if (strnEQ(s, "CORE::", 6)) { |
be1b855b | 494 | const int code = keyword(s + 6, SvCUR(TOPs) - 6, 1); |
a96df643 | 495 | if (!code) |
147e3846 | 496 | DIE(aTHX_ "Can't find an opnumber for \"%" UTF8f "\"", |
b17a0679 | 497 | UTF8fARG(SvFLAGS(TOPs) & SVf_UTF8, SvCUR(TOPs)-6, s+6)); |
4e338c21 | 498 | { |
b66130dd FC |
499 | SV * const sv = core_prototype(NULL, s + 6, code, NULL); |
500 | if (sv) ret = sv; | |
501 | } | |
b8c38f0a | 502 | goto set; |
b6c543e3 IZ |
503 | } |
504 | } | |
f2c0649b | 505 | cv = sv_2cv(TOPs, &stash, &gv, 0); |
5f05dabc | 506 | if (cv && SvPOK(cv)) |
8fa6a409 FC |
507 | ret = newSVpvn_flags( |
508 | CvPROTO(cv), CvPROTOLEN(cv), SVs_TEMP | SvUTF8(cv) | |
509 | ); | |
b6c543e3 | 510 | set: |
c07a80fd | 511 | SETs(ret); |
512 | RETURN; | |
513 | } | |
514 | ||
a0d0e21e LW |
515 | PP(pp_anoncode) |
516 | { | |
20b7effb | 517 | dSP; |
ea726b52 | 518 | CV *cv = MUTABLE_CV(PAD_SV(PL_op->op_targ)); |
a5f75d66 | 519 | if (CvCLONE(cv)) |
ad64d0ec | 520 | cv = MUTABLE_CV(sv_2mortal(MUTABLE_SV(cv_clone(cv)))); |
5f05dabc | 521 | EXTEND(SP,1); |
ad64d0ec | 522 | PUSHs(MUTABLE_SV(cv)); |
a0d0e21e LW |
523 | RETURN; |
524 | } | |
525 | ||
526 | PP(pp_srefgen) | |
79072805 | 527 | { |
20b7effb | 528 | dSP; |
71be2cbc | 529 | *SP = refto(*SP); |
3ed34c76 | 530 | return NORMAL; |
8ec5e241 | 531 | } |
a0d0e21e LW |
532 | |
533 | PP(pp_refgen) | |
534 | { | |
20b7effb | 535 | dSP; dMARK; |
82334630 | 536 | if (GIMME_V != G_ARRAY) { |
5f0b1d4e GS |
537 | if (++MARK <= SP) |
538 | *MARK = *SP; | |
539 | else | |
1d51ab6c FC |
540 | { |
541 | MEXTEND(SP, 1); | |
3280af22 | 542 | *MARK = &PL_sv_undef; |
1d51ab6c | 543 | } |
5f0b1d4e GS |
544 | *MARK = refto(*MARK); |
545 | SP = MARK; | |
546 | RETURN; | |
a0d0e21e | 547 | } |
bbce6d69 | 548 | EXTEND_MORTAL(SP - MARK); |
71be2cbc | 549 | while (++MARK <= SP) |
550 | *MARK = refto(*MARK); | |
a0d0e21e | 551 | RETURN; |
79072805 LW |
552 | } |
553 | ||
76e3520e | 554 | STATIC SV* |
cea2e8a9 | 555 | S_refto(pTHX_ SV *sv) |
71be2cbc | 556 | { |
557 | SV* rv; | |
558 | ||
7918f24d NC |
559 | PERL_ARGS_ASSERT_REFTO; |
560 | ||
71be2cbc | 561 | if (SvTYPE(sv) == SVt_PVLV && LvTYPE(sv) == 'y') { |
562 | if (LvTARGLEN(sv)) | |
68dc0745 | 563 | vivify_defelem(sv); |
564 | if (!(sv = LvTARG(sv))) | |
3280af22 | 565 | sv = &PL_sv_undef; |
0dd88869 | 566 | else |
b37c2d43 | 567 | SvREFCNT_inc_void_NN(sv); |
71be2cbc | 568 | } |
d8b46c1b | 569 | else if (SvTYPE(sv) == SVt_PVAV) { |
502c6561 NC |
570 | if (!AvREAL((const AV *)sv) && AvREIFY((const AV *)sv)) |
571 | av_reify(MUTABLE_AV(sv)); | |
d8b46c1b | 572 | SvTEMP_off(sv); |
b37c2d43 | 573 | SvREFCNT_inc_void_NN(sv); |
d8b46c1b | 574 | } |
60779a30 | 575 | else if (SvPADTMP(sv)) { |
f2933f5f | 576 | sv = newSVsv(sv); |
60779a30 | 577 | } |
71be2cbc | 578 | else { |
579 | SvTEMP_off(sv); | |
b37c2d43 | 580 | SvREFCNT_inc_void_NN(sv); |
71be2cbc | 581 | } |
582 | rv = sv_newmortal(); | |
4df7f6af | 583 | sv_upgrade(rv, SVt_IV); |
b162af07 | 584 | SvRV_set(rv, sv); |
71be2cbc | 585 | SvROK_on(rv); |
586 | return rv; | |
587 | } | |
588 | ||
79072805 LW |
589 | PP(pp_ref) |
590 | { | |
3c1e67ac DD |
591 | dSP; |
592 | SV * const sv = TOPs; | |
f12c7020 | 593 | |
511ddbdf | 594 | SvGETMAGIC(sv); |
ba75e9a4 | 595 | if (!SvROK(sv)) { |
3c1e67ac | 596 | SETs(&PL_sv_no); |
ba75e9a4 DM |
597 | return NORMAL; |
598 | } | |
599 | ||
600 | /* op is in boolean context? */ | |
601 | if ( (PL_op->op_private & OPpTRUEBOOL) | |
602 | || ( (PL_op->op_private & OPpMAYBE_TRUEBOOL) | |
603 | && block_gimme() == G_VOID)) | |
604 | { | |
605 | /* refs are always true - unless it's to an object blessed into a | |
606 | * class with a false name, i.e. "0". So we have to check for | |
607 | * that remote possibility. The following is is basically an | |
608 | * unrolled SvTRUE(sv_reftype(rv)) */ | |
609 | SV * const rv = SvRV(sv); | |
610 | if (SvOBJECT(rv)) { | |
611 | HV *stash = SvSTASH(rv); | |
612 | HEK *hek = HvNAME_HEK(stash); | |
613 | if (hek) { | |
614 | I32 len = HEK_LEN(hek); | |
615 | /* bail out and do it the hard way? */ | |
616 | if (UNLIKELY( | |
617 | len == HEf_SVKEY | |
618 | || (len == 1 && HEK_KEY(hek)[0] == '0') | |
619 | )) | |
620 | goto do_sv_ref; | |
621 | } | |
622 | } | |
623 | SETs(&PL_sv_yes); | |
624 | return NORMAL; | |
625 | } | |
626 | ||
627 | do_sv_ref: | |
628 | { | |
3c1e67ac DD |
629 | dTARGET; |
630 | SETs(TARG); | |
ba75e9a4 | 631 | sv_ref(TARG, SvRV(sv), TRUE); |
a10e04b5 | 632 | SvSETMAGIC(TARG); |
ba75e9a4 | 633 | return NORMAL; |
3c1e67ac | 634 | } |
79072805 | 635 | |
79072805 LW |
636 | } |
637 | ||
ba75e9a4 | 638 | |
79072805 LW |
639 | PP(pp_bless) |
640 | { | |
20b7effb | 641 | dSP; |
463ee0b2 | 642 | HV *stash; |
79072805 | 643 | |
463ee0b2 | 644 | if (MAXARG == 1) |
dcdfe746 | 645 | { |
c2f922f1 | 646 | curstash: |
11faa288 | 647 | stash = CopSTASH(PL_curcop); |
dcdfe746 FC |
648 | if (SvTYPE(stash) != SVt_PVHV) |
649 | Perl_croak(aTHX_ "Attempt to bless into a freed package"); | |
650 | } | |
7b8d334a | 651 | else { |
1b6737cc | 652 | SV * const ssv = POPs; |
7b8d334a | 653 | STRLEN len; |
e1ec3a88 | 654 | const char *ptr; |
81689caa | 655 | |
c2f922f1 | 656 | if (!ssv) goto curstash; |
8d9dd4b9 | 657 | SvGETMAGIC(ssv); |
c7ea825d FC |
658 | if (SvROK(ssv)) { |
659 | if (!SvAMAGIC(ssv)) { | |
660 | frog: | |
81689caa | 661 | Perl_croak(aTHX_ "Attempt to bless into a reference"); |
c7ea825d FC |
662 | } |
663 | /* SvAMAGIC is on here, but it only means potentially overloaded, | |
664 | so after stringification: */ | |
665 | ptr = SvPV_nomg_const(ssv,len); | |
666 | /* We need to check the flag again: */ | |
667 | if (!SvAMAGIC(ssv)) goto frog; | |
668 | } | |
669 | else ptr = SvPV_nomg_const(ssv,len); | |
a2a5de95 NC |
670 | if (len == 0) |
671 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), | |
672 | "Explicit blessing to '' (assuming package main)"); | |
e69c50fe | 673 | stash = gv_stashpvn(ptr, len, GV_ADD|SvUTF8(ssv)); |
7b8d334a | 674 | } |
a0d0e21e | 675 | |
5d3fdfeb | 676 | (void)sv_bless(TOPs, stash); |
79072805 LW |
677 | RETURN; |
678 | } | |
679 | ||
fb73857a | 680 | PP(pp_gelem) |
681 | { | |
20b7effb | 682 | dSP; |
b13b2135 | 683 | |
1b6737cc | 684 | SV *sv = POPs; |
a180b31a BF |
685 | STRLEN len; |
686 | const char * const elem = SvPV_const(sv, len); | |
5695161e | 687 | GV * const gv = MUTABLE_GV(TOPs); |
c445ea15 | 688 | SV * tmpRef = NULL; |
1b6737cc | 689 | |
c445ea15 | 690 | sv = NULL; |
c4ba80c3 NC |
691 | if (elem) { |
692 | /* elem will always be NUL terminated. */ | |
c4ba80c3 NC |
693 | switch (*elem) { |
694 | case 'A': | |
500f3e18 | 695 | if (memEQs(elem, len, "ARRAY")) |
e14698d8 | 696 | { |
ad64d0ec | 697 | tmpRef = MUTABLE_SV(GvAV(gv)); |
e14698d8 FC |
698 | if (tmpRef && !AvREAL((const AV *)tmpRef) |
699 | && AvREIFY((const AV *)tmpRef)) | |
700 | av_reify(MUTABLE_AV(tmpRef)); | |
701 | } | |
c4ba80c3 NC |
702 | break; |
703 | case 'C': | |
500f3e18 | 704 | if (memEQs(elem, len, "CODE")) |
ad64d0ec | 705 | tmpRef = MUTABLE_SV(GvCVu(gv)); |
c4ba80c3 NC |
706 | break; |
707 | case 'F': | |
500f3e18 | 708 | if (memEQs(elem, len, "FILEHANDLE")) { |
ad64d0ec | 709 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
c4ba80c3 NC |
710 | } |
711 | else | |
500f3e18 | 712 | if (memEQs(elem, len, "FORMAT")) |
ad64d0ec | 713 | tmpRef = MUTABLE_SV(GvFORM(gv)); |
c4ba80c3 NC |
714 | break; |
715 | case 'G': | |
500f3e18 | 716 | if (memEQs(elem, len, "GLOB")) |
ad64d0ec | 717 | tmpRef = MUTABLE_SV(gv); |
c4ba80c3 NC |
718 | break; |
719 | case 'H': | |
500f3e18 | 720 | if (memEQs(elem, len, "HASH")) |
ad64d0ec | 721 | tmpRef = MUTABLE_SV(GvHV(gv)); |
c4ba80c3 NC |
722 | break; |
723 | case 'I': | |
500f3e18 | 724 | if (memEQs(elem, len, "IO")) |
ad64d0ec | 725 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
c4ba80c3 NC |
726 | break; |
727 | case 'N': | |
500f3e18 | 728 | if (memEQs(elem, len, "NAME")) |
a663657d | 729 | sv = newSVhek(GvNAME_HEK(gv)); |
c4ba80c3 NC |
730 | break; |
731 | case 'P': | |
500f3e18 | 732 | if (memEQs(elem, len, "PACKAGE")) { |
7fa3a4ab NC |
733 | const HV * const stash = GvSTASH(gv); |
734 | const HEK * const hek = stash ? HvNAME_HEK(stash) : NULL; | |
396482e1 | 735 | sv = hek ? newSVhek(hek) : newSVpvs("__ANON__"); |
c4ba80c3 NC |
736 | } |
737 | break; | |
738 | case 'S': | |
500f3e18 | 739 | if (memEQs(elem, len, "SCALAR")) |
f9d52e31 | 740 | tmpRef = GvSVn(gv); |
c4ba80c3 | 741 | break; |
39b99f21 | 742 | } |
fb73857a | 743 | } |
76e3520e GS |
744 | if (tmpRef) |
745 | sv = newRV(tmpRef); | |
fb73857a | 746 | if (sv) |
747 | sv_2mortal(sv); | |
748 | else | |
3280af22 | 749 | sv = &PL_sv_undef; |
5695161e | 750 | SETs(sv); |
fb73857a | 751 | RETURN; |
752 | } | |
753 | ||
a0d0e21e | 754 | /* Pattern matching */ |
79072805 | 755 | |
a0d0e21e | 756 | PP(pp_study) |
79072805 | 757 | { |
add3e777 | 758 | dSP; dTOPss; |
a0d0e21e LW |
759 | STRLEN len; |
760 | ||
1fa930f2 | 761 | (void)SvPV(sv, len); |
bc9a5256 | 762 | if (len == 0 || len > I32_MAX || !SvPOK(sv) || SvUTF8(sv) || SvVALID(sv)) { |
32f0ea87 | 763 | /* Historically, study was skipped in these cases. */ |
add3e777 FC |
764 | SETs(&PL_sv_no); |
765 | return NORMAL; | |
a4f4e906 NC |
766 | } |
767 | ||
a58a85fa | 768 | /* Make study a no-op. It's no longer useful and its existence |
32f0ea87 | 769 | complicates matters elsewhere. */ |
add3e777 FC |
770 | SETs(&PL_sv_yes); |
771 | return NORMAL; | |
79072805 LW |
772 | } |
773 | ||
b1c05ba5 DM |
774 | |
775 | /* also used for: pp_transr() */ | |
776 | ||
a0d0e21e | 777 | PP(pp_trans) |
79072805 | 778 | { |
6442877a | 779 | dSP; |
a0d0e21e LW |
780 | SV *sv; |
781 | ||
533c011a | 782 | if (PL_op->op_flags & OPf_STACKED) |
a0d0e21e | 783 | sv = POPs; |
79072805 | 784 | else { |
a0d0e21e | 785 | EXTEND(SP,1); |
f605e527 | 786 | if (ARGTARG) |
6442877a | 787 | sv = PAD_SV(ARGTARG); |
f605e527 FC |
788 | else { |
789 | sv = DEFSV; | |
790 | } | |
79072805 | 791 | } |
bb16bae8 | 792 | if(PL_op->op_type == OP_TRANSR) { |
290797f7 FC |
793 | STRLEN len; |
794 | const char * const pv = SvPV(sv,len); | |
795 | SV * const newsv = newSVpvn_flags(pv, len, SVs_TEMP|SvUTF8(sv)); | |
bb16bae8 | 796 | do_trans(newsv); |
290797f7 | 797 | PUSHs(newsv); |
bb16bae8 | 798 | } |
5bbe7184 | 799 | else { |
bcb10b84 VP |
800 | I32 i = do_trans(sv); |
801 | mPUSHi(i); | |
5bbe7184 | 802 | } |
a0d0e21e | 803 | RETURN; |
79072805 LW |
804 | } |
805 | ||
a0d0e21e | 806 | /* Lvalue operators. */ |
79072805 | 807 | |
f595e19f | 808 | static size_t |
81745e4e NC |
809 | S_do_chomp(pTHX_ SV *retval, SV *sv, bool chomping) |
810 | { | |
81745e4e NC |
811 | STRLEN len; |
812 | char *s; | |
f595e19f | 813 | size_t count = 0; |
81745e4e NC |
814 | |
815 | PERL_ARGS_ASSERT_DO_CHOMP; | |
816 | ||
817 | if (chomping && (RsSNARF(PL_rs) || RsRECORD(PL_rs))) | |
f595e19f | 818 | return 0; |
81745e4e NC |
819 | if (SvTYPE(sv) == SVt_PVAV) { |
820 | I32 i; | |
821 | AV *const av = MUTABLE_AV(sv); | |
822 | const I32 max = AvFILL(av); | |
823 | ||
824 | for (i = 0; i <= max; i++) { | |
825 | sv = MUTABLE_SV(av_fetch(av, i, FALSE)); | |
826 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) | |
f595e19f | 827 | count += do_chomp(retval, sv, chomping); |
81745e4e | 828 | } |
f595e19f | 829 | return count; |
81745e4e NC |
830 | } |
831 | else if (SvTYPE(sv) == SVt_PVHV) { | |
832 | HV* const hv = MUTABLE_HV(sv); | |
833 | HE* entry; | |
834 | (void)hv_iterinit(hv); | |
835 | while ((entry = hv_iternext(hv))) | |
f595e19f FC |
836 | count += do_chomp(retval, hv_iterval(hv,entry), chomping); |
837 | return count; | |
81745e4e NC |
838 | } |
839 | else if (SvREADONLY(sv)) { | |
cb077ed2 | 840 | Perl_croak_no_modify(); |
81745e4e NC |
841 | } |
842 | ||
81745e4e NC |
843 | s = SvPV(sv, len); |
844 | if (chomping) { | |
81745e4e | 845 | if (s && len) { |
997c424a DD |
846 | char *temp_buffer = NULL; |
847 | SV *svrecode = NULL; | |
81745e4e NC |
848 | s += --len; |
849 | if (RsPARA(PL_rs)) { | |
850 | if (*s != '\n') | |
997c424a | 851 | goto nope_free_nothing; |
f595e19f | 852 | ++count; |
81745e4e NC |
853 | while (len && s[-1] == '\n') { |
854 | --len; | |
855 | --s; | |
f595e19f | 856 | ++count; |
81745e4e NC |
857 | } |
858 | } | |
859 | else { | |
860 | STRLEN rslen, rs_charlen; | |
861 | const char *rsptr = SvPV_const(PL_rs, rslen); | |
862 | ||
863 | rs_charlen = SvUTF8(PL_rs) | |
864 | ? sv_len_utf8(PL_rs) | |
865 | : rslen; | |
866 | ||
867 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { | |
868 | /* Assumption is that rs is shorter than the scalar. */ | |
869 | if (SvUTF8(PL_rs)) { | |
870 | /* RS is utf8, scalar is 8 bit. */ | |
871 | bool is_utf8 = TRUE; | |
872 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, | |
873 | &rslen, &is_utf8); | |
874 | if (is_utf8) { | |
997c424a DD |
875 | /* Cannot downgrade, therefore cannot possibly match. |
876 | At this point, temp_buffer is not alloced, and | |
877 | is the buffer inside PL_rs, so dont free it. | |
81745e4e NC |
878 | */ |
879 | assert (temp_buffer == rsptr); | |
997c424a | 880 | goto nope_free_sv; |
81745e4e NC |
881 | } |
882 | rsptr = temp_buffer; | |
883 | } | |
81745e4e NC |
884 | else { |
885 | /* RS is 8 bit, scalar is utf8. */ | |
886 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); | |
887 | rsptr = temp_buffer; | |
888 | } | |
889 | } | |
890 | if (rslen == 1) { | |
891 | if (*s != *rsptr) | |
997c424a | 892 | goto nope_free_all; |
f595e19f | 893 | ++count; |
81745e4e NC |
894 | } |
895 | else { | |
896 | if (len < rslen - 1) | |
997c424a | 897 | goto nope_free_all; |
81745e4e NC |
898 | len -= rslen - 1; |
899 | s -= rslen - 1; | |
900 | if (memNE(s, rsptr, rslen)) | |
997c424a | 901 | goto nope_free_all; |
f595e19f | 902 | count += rs_charlen; |
81745e4e NC |
903 | } |
904 | } | |
3b7ded39 | 905 | SvPV_force_nomg_nolen(sv); |
81745e4e NC |
906 | SvCUR_set(sv, len); |
907 | *SvEND(sv) = '\0'; | |
908 | SvNIOK_off(sv); | |
909 | SvSETMAGIC(sv); | |
81745e4e | 910 | |
997c424a DD |
911 | nope_free_all: |
912 | Safefree(temp_buffer); | |
913 | nope_free_sv: | |
914 | SvREFCNT_dec(svrecode); | |
915 | nope_free_nothing: ; | |
916 | } | |
81745e4e | 917 | } else { |
f8c80a8e | 918 | if (len && (!SvPOK(sv) || SvIsCOW(sv))) |
81745e4e NC |
919 | s = SvPV_force_nomg(sv, len); |
920 | if (DO_UTF8(sv)) { | |
921 | if (s && len) { | |
922 | char * const send = s + len; | |
923 | char * const start = s; | |
924 | s = send - 1; | |
925 | while (s > start && UTF8_IS_CONTINUATION(*s)) | |
926 | s--; | |
927 | if (is_utf8_string((U8*)s, send - s)) { | |
928 | sv_setpvn(retval, s, send - s); | |
929 | *s = '\0'; | |
930 | SvCUR_set(sv, s - start); | |
931 | SvNIOK_off(sv); | |
932 | SvUTF8_on(retval); | |
933 | } | |
934 | } | |
935 | else | |
500f3e18 | 936 | SvPVCLEAR(retval); |
81745e4e NC |
937 | } |
938 | else if (s && len) { | |
939 | s += --len; | |
940 | sv_setpvn(retval, s, 1); | |
941 | *s = '\0'; | |
942 | SvCUR_set(sv, len); | |
943 | SvUTF8_off(sv); | |
944 | SvNIOK_off(sv); | |
945 | } | |
946 | else | |
500f3e18 | 947 | SvPVCLEAR(retval); |
81745e4e NC |
948 | SvSETMAGIC(sv); |
949 | } | |
f595e19f | 950 | return count; |
81745e4e NC |
951 | } |
952 | ||
b1c05ba5 DM |
953 | |
954 | /* also used for: pp_schomp() */ | |
955 | ||
a0d0e21e LW |
956 | PP(pp_schop) |
957 | { | |
20b7effb | 958 | dSP; dTARGET; |
fa54efae NC |
959 | const bool chomping = PL_op->op_type == OP_SCHOMP; |
960 | ||
f595e19f | 961 | const size_t count = do_chomp(TARG, TOPs, chomping); |
fa54efae | 962 | if (chomping) |
f595e19f | 963 | sv_setiv(TARG, count); |
a0d0e21e | 964 | SETTARG; |
ee41d8c7 | 965 | return NORMAL; |
79072805 LW |
966 | } |
967 | ||
b1c05ba5 DM |
968 | |
969 | /* also used for: pp_chomp() */ | |
970 | ||
a0d0e21e | 971 | PP(pp_chop) |
79072805 | 972 | { |
20b7effb | 973 | dSP; dMARK; dTARGET; dORIGMARK; |
fa54efae | 974 | const bool chomping = PL_op->op_type == OP_CHOMP; |
f595e19f | 975 | size_t count = 0; |
8ec5e241 | 976 | |
20cf1f79 | 977 | while (MARK < SP) |
f595e19f FC |
978 | count += do_chomp(TARG, *++MARK, chomping); |
979 | if (chomping) | |
980 | sv_setiv(TARG, count); | |
20cf1f79 NC |
981 | SP = ORIGMARK; |
982 | XPUSHTARG; | |
a0d0e21e | 983 | RETURN; |
79072805 LW |
984 | } |
985 | ||
a0d0e21e LW |
986 | PP(pp_undef) |
987 | { | |
20b7effb | 988 | dSP; |
a0d0e21e LW |
989 | SV *sv; |
990 | ||
533c011a | 991 | if (!PL_op->op_private) { |
774d564b | 992 | EXTEND(SP, 1); |
a0d0e21e | 993 | RETPUSHUNDEF; |
774d564b | 994 | } |
79072805 | 995 | |
821f14b0 | 996 | sv = TOPs; |
a0d0e21e | 997 | if (!sv) |
821f14b0 FC |
998 | { |
999 | SETs(&PL_sv_undef); | |
1000 | return NORMAL; | |
1001 | } | |
85e6fe83 | 1002 | |
4dda930b FC |
1003 | if (SvTHINKFIRST(sv)) |
1004 | sv_force_normal_flags(sv, SV_COW_DROP_PV|SV_IMMEDIATE_UNREF); | |
85e6fe83 | 1005 | |
a0d0e21e LW |
1006 | switch (SvTYPE(sv)) { |
1007 | case SVt_NULL: | |
1008 | break; | |
1009 | case SVt_PVAV: | |
60edcf09 | 1010 | av_undef(MUTABLE_AV(sv)); |
a0d0e21e LW |
1011 | break; |
1012 | case SVt_PVHV: | |
60edcf09 | 1013 | hv_undef(MUTABLE_HV(sv)); |
a0d0e21e LW |
1014 | break; |
1015 | case SVt_PVCV: | |
a2a5de95 | 1016 | if (cv_const_sv((const CV *)sv)) |
714cd18f | 1017 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), |
147e3846 | 1018 | "Constant subroutine %" SVf " undefined", |
714cd18f BF |
1019 | SVfARG(CvANON((const CV *)sv) |
1020 | ? newSVpvs_flags("(anonymous)", SVs_TEMP) | |
bdbfc51a FC |
1021 | : sv_2mortal(newSVhek( |
1022 | CvNAMED(sv) | |
1023 | ? CvNAME_HEK((CV *)sv) | |
1024 | : GvENAME_HEK(CvGV((const CV *)sv)) | |
1025 | )) | |
1026 | )); | |
5f66b61c | 1027 | /* FALLTHROUGH */ |
9607fc9c | 1028 | case SVt_PVFM: |
6fc92669 | 1029 | /* let user-undef'd sub keep its identity */ |
b7acb0a3 | 1030 | cv_undef_flags(MUTABLE_CV(sv), CV_UNDEF_KEEP_NAME); |
a0d0e21e | 1031 | break; |
8e07c86e | 1032 | case SVt_PVGV: |
bc1df6c2 FC |
1033 | assert(isGV_with_GP(sv)); |
1034 | assert(!SvFAKE(sv)); | |
1035 | { | |
20408e3c | 1036 | GP *gp; |
dd69841b BB |
1037 | HV *stash; |
1038 | ||
dd69841b | 1039 | /* undef *Pkg::meth_name ... */ |
e530fb81 FC |
1040 | bool method_changed |
1041 | = GvCVu((const GV *)sv) && (stash = GvSTASH((const GV *)sv)) | |
1042 | && HvENAME_get(stash); | |
1043 | /* undef *Foo:: */ | |
1044 | if((stash = GvHV((const GV *)sv))) { | |
1045 | if(HvENAME_get(stash)) | |
1046 | SvREFCNT_inc_simple_void_NN(sv_2mortal((SV *)stash)); | |
1047 | else stash = NULL; | |
1048 | } | |
dd69841b | 1049 | |
795eb8c8 | 1050 | SvREFCNT_inc_simple_void_NN(sv_2mortal(sv)); |
159b6efe | 1051 | gp_free(MUTABLE_GV(sv)); |
a02a5408 | 1052 | Newxz(gp, 1, GP); |
c43ae56f | 1053 | GvGP_set(sv, gp_ref(gp)); |
2e3295e3 | 1054 | #ifndef PERL_DONT_CREATE_GVSV |
561b68a9 | 1055 | GvSV(sv) = newSV(0); |
2e3295e3 | 1056 | #endif |
57843af0 | 1057 | GvLINE(sv) = CopLINE(PL_curcop); |
159b6efe | 1058 | GvEGV(sv) = MUTABLE_GV(sv); |
20408e3c | 1059 | GvMULTI_on(sv); |
e530fb81 FC |
1060 | |
1061 | if(stash) | |
afdbe55d | 1062 | mro_package_moved(NULL, stash, (const GV *)sv, 0); |
e530fb81 FC |
1063 | stash = NULL; |
1064 | /* undef *Foo::ISA */ | |
1065 | if( strEQ(GvNAME((const GV *)sv), "ISA") | |
1066 | && (stash = GvSTASH((const GV *)sv)) | |
1067 | && (method_changed || HvENAME(stash)) ) | |
1068 | mro_isa_changed_in(stash); | |
1069 | else if(method_changed) | |
1070 | mro_method_changed_in( | |
da9043f5 | 1071 | GvSTASH((const GV *)sv) |
e530fb81 FC |
1072 | ); |
1073 | ||
6e592b3a | 1074 | break; |
20408e3c | 1075 | } |
a0d0e21e | 1076 | default: |
b15aece3 | 1077 | if (SvTYPE(sv) >= SVt_PV && SvPVX_const(sv) && SvLEN(sv)) { |
8bd4d4c5 | 1078 | SvPV_free(sv); |
c445ea15 | 1079 | SvPV_set(sv, NULL); |
4633a7c4 | 1080 | SvLEN_set(sv, 0); |
a0d0e21e | 1081 | } |
0c34ef67 | 1082 | SvOK_off(sv); |
4633a7c4 | 1083 | SvSETMAGIC(sv); |
79072805 | 1084 | } |
a0d0e21e | 1085 | |
821f14b0 FC |
1086 | SETs(&PL_sv_undef); |
1087 | return NORMAL; | |
79072805 LW |
1088 | } |
1089 | ||
b1c05ba5 | 1090 | |
20e96431 | 1091 | /* common "slow" code for pp_postinc and pp_postdec */ |
b1c05ba5 | 1092 | |
20e96431 DM |
1093 | static OP * |
1094 | S_postincdec_common(pTHX_ SV *sv, SV *targ) | |
a0d0e21e | 1095 | { |
20e96431 | 1096 | dSP; |
c22c99bc FC |
1097 | const bool inc = |
1098 | PL_op->op_type == OP_POSTINC || PL_op->op_type == OP_I_POSTINC; | |
20e96431 DM |
1099 | |
1100 | if (SvROK(sv)) | |
7dcb9b98 | 1101 | TARG = sv_newmortal(); |
20e96431 DM |
1102 | sv_setsv(TARG, sv); |
1103 | if (inc) | |
1104 | sv_inc_nomg(sv); | |
1105 | else | |
1106 | sv_dec_nomg(sv); | |
1107 | SvSETMAGIC(sv); | |
1e54a23f | 1108 | /* special case for undef: see thread at 2003-03/msg00536.html in archive */ |
c22c99bc | 1109 | if (inc && !SvOK(TARG)) |
a0d0e21e | 1110 | sv_setiv(TARG, 0); |
e87de4ab | 1111 | SETTARG; |
a0d0e21e LW |
1112 | return NORMAL; |
1113 | } | |
79072805 | 1114 | |
20e96431 DM |
1115 | |
1116 | /* also used for: pp_i_postinc() */ | |
1117 | ||
1118 | PP(pp_postinc) | |
1119 | { | |
1120 | dSP; dTARGET; | |
1121 | SV *sv = TOPs; | |
1122 | ||
1123 | /* special-case sv being a simple integer */ | |
1124 | if (LIKELY(((sv->sv_flags & | |
1125 | (SVf_THINKFIRST|SVs_GMG|SVf_IVisUV| | |
1126 | SVf_IOK|SVf_NOK|SVf_POK|SVp_NOK|SVp_POK|SVf_ROK)) | |
1127 | == SVf_IOK)) | |
1128 | && SvIVX(sv) != IV_MAX) | |
1129 | { | |
1130 | IV iv = SvIVX(sv); | |
1131 | SvIV_set(sv, iv + 1); | |
1132 | TARGi(iv, 0); /* arg not GMG, so can't be tainted */ | |
1133 | SETs(TARG); | |
1134 | return NORMAL; | |
1135 | } | |
1136 | ||
1137 | return S_postincdec_common(aTHX_ sv, TARG); | |
1138 | } | |
1139 | ||
1140 | ||
1141 | /* also used for: pp_i_postdec() */ | |
1142 | ||
1143 | PP(pp_postdec) | |
1144 | { | |
1145 | dSP; dTARGET; | |
1146 | SV *sv = TOPs; | |
1147 | ||
1148 | /* special-case sv being a simple integer */ | |
1149 | if (LIKELY(((sv->sv_flags & | |
1150 | (SVf_THINKFIRST|SVs_GMG|SVf_IVisUV| | |
1151 | SVf_IOK|SVf_NOK|SVf_POK|SVp_NOK|SVp_POK|SVf_ROK)) | |
1152 | == SVf_IOK)) | |
1153 | && SvIVX(sv) != IV_MIN) | |
1154 | { | |
1155 | IV iv = SvIVX(sv); | |
1156 | SvIV_set(sv, iv - 1); | |
1157 | TARGi(iv, 0); /* arg not GMG, so can't be tainted */ | |
1158 | SETs(TARG); | |
1159 | return NORMAL; | |
1160 | } | |
1161 | ||
1162 | return S_postincdec_common(aTHX_ sv, TARG); | |
1163 | } | |
1164 | ||
1165 | ||
a0d0e21e LW |
1166 | /* Ordinary operators. */ |
1167 | ||
1168 | PP(pp_pow) | |
1169 | { | |
20b7effb | 1170 | dSP; dATARGET; SV *svl, *svr; |
58d76dfd | 1171 | #ifdef PERL_PRESERVE_IVUV |
52a96ae6 HS |
1172 | bool is_int = 0; |
1173 | #endif | |
6f1401dc DM |
1174 | tryAMAGICbin_MG(pow_amg, AMGf_assign|AMGf_numeric); |
1175 | svr = TOPs; | |
1176 | svl = TOPm1s; | |
52a96ae6 HS |
1177 | #ifdef PERL_PRESERVE_IVUV |
1178 | /* For integer to integer power, we do the calculation by hand wherever | |
1179 | we're sure it is safe; otherwise we call pow() and try to convert to | |
1180 | integer afterwards. */ | |
01f91bf2 | 1181 | if (SvIV_please_nomg(svr) && SvIV_please_nomg(svl)) { |
900658e3 PF |
1182 | UV power; |
1183 | bool baseuok; | |
1184 | UV baseuv; | |
1185 | ||
800401ee JH |
1186 | if (SvUOK(svr)) { |
1187 | power = SvUVX(svr); | |
900658e3 | 1188 | } else { |
800401ee | 1189 | const IV iv = SvIVX(svr); |
900658e3 PF |
1190 | if (iv >= 0) { |
1191 | power = iv; | |
1192 | } else { | |
1193 | goto float_it; /* Can't do negative powers this way. */ | |
1194 | } | |
1195 | } | |
1196 | ||
800401ee | 1197 | baseuok = SvUOK(svl); |
900658e3 | 1198 | if (baseuok) { |
800401ee | 1199 | baseuv = SvUVX(svl); |
900658e3 | 1200 | } else { |
800401ee | 1201 | const IV iv = SvIVX(svl); |
900658e3 PF |
1202 | if (iv >= 0) { |
1203 | baseuv = iv; | |
1204 | baseuok = TRUE; /* effectively it's a UV now */ | |
1205 | } else { | |
1206 | baseuv = -iv; /* abs, baseuok == false records sign */ | |
1207 | } | |
1208 | } | |
52a96ae6 HS |
1209 | /* now we have integer ** positive integer. */ |
1210 | is_int = 1; | |
1211 | ||
1212 | /* foo & (foo - 1) is zero only for a power of 2. */ | |
58d76dfd | 1213 | if (!(baseuv & (baseuv - 1))) { |
52a96ae6 | 1214 | /* We are raising power-of-2 to a positive integer. |
58d76dfd JH |
1215 | The logic here will work for any base (even non-integer |
1216 | bases) but it can be less accurate than | |
1217 | pow (base,power) or exp (power * log (base)) when the | |
1218 | intermediate values start to spill out of the mantissa. | |
1219 | With powers of 2 we know this can't happen. | |
1220 | And powers of 2 are the favourite thing for perl | |
1221 | programmers to notice ** not doing what they mean. */ | |
1222 | NV result = 1.0; | |
1223 | NV base = baseuok ? baseuv : -(NV)baseuv; | |
900658e3 PF |
1224 | |
1225 | if (power & 1) { | |
1226 | result *= base; | |
1227 | } | |
1228 | while (power >>= 1) { | |
1229 | base *= base; | |
1230 | if (power & 1) { | |
1231 | result *= base; | |
1232 | } | |
1233 | } | |
58d76dfd JH |
1234 | SP--; |
1235 | SETn( result ); | |
6f1401dc | 1236 | SvIV_please_nomg(svr); |
58d76dfd | 1237 | RETURN; |
52a96ae6 | 1238 | } else { |
eb578fdb KW |
1239 | unsigned int highbit = 8 * sizeof(UV); |
1240 | unsigned int diff = 8 * sizeof(UV); | |
900658e3 PF |
1241 | while (diff >>= 1) { |
1242 | highbit -= diff; | |
1243 | if (baseuv >> highbit) { | |
1244 | highbit += diff; | |
1245 | } | |
52a96ae6 HS |
1246 | } |
1247 | /* we now have baseuv < 2 ** highbit */ | |
1248 | if (power * highbit <= 8 * sizeof(UV)) { | |
1249 | /* result will definitely fit in UV, so use UV math | |
1250 | on same algorithm as above */ | |
eb578fdb KW |
1251 | UV result = 1; |
1252 | UV base = baseuv; | |
f2338a2e | 1253 | const bool odd_power = cBOOL(power & 1); |
900658e3 PF |
1254 | if (odd_power) { |
1255 | result *= base; | |
1256 | } | |
1257 | while (power >>= 1) { | |
1258 | base *= base; | |
1259 | if (power & 1) { | |
52a96ae6 | 1260 | result *= base; |
52a96ae6 HS |
1261 | } |
1262 | } | |
1263 | SP--; | |
0615a994 | 1264 | if (baseuok || !odd_power) |
52a96ae6 HS |
1265 | /* answer is positive */ |
1266 | SETu( result ); | |
1267 | else if (result <= (UV)IV_MAX) | |
1268 | /* answer negative, fits in IV */ | |
1269 | SETi( -(IV)result ); | |
1270 | else if (result == (UV)IV_MIN) | |
1271 | /* 2's complement assumption: special case IV_MIN */ | |
1272 | SETi( IV_MIN ); | |
1273 | else | |
1274 | /* answer negative, doesn't fit */ | |
1275 | SETn( -(NV)result ); | |
1276 | RETURN; | |
1277 | } | |
1278 | } | |
58d76dfd | 1279 | } |
52a96ae6 | 1280 | float_it: |
58d76dfd | 1281 | #endif |
a0d0e21e | 1282 | { |
6f1401dc DM |
1283 | NV right = SvNV_nomg(svr); |
1284 | NV left = SvNV_nomg(svl); | |
4efa5a16 | 1285 | (void)POPs; |
3aaeb624 JA |
1286 | |
1287 | #if defined(USE_LONG_DOUBLE) && defined(HAS_AIX_POWL_NEG_BASE_BUG) | |
1288 | /* | |
1289 | We are building perl with long double support and are on an AIX OS | |
1290 | afflicted with a powl() function that wrongly returns NaNQ for any | |
1291 | negative base. This was reported to IBM as PMR #23047-379 on | |
1292 | 03/06/2006. The problem exists in at least the following versions | |
1293 | of AIX and the libm fileset, and no doubt others as well: | |
1294 | ||
1295 | AIX 4.3.3-ML10 bos.adt.libm 4.3.3.50 | |
1296 | AIX 5.1.0-ML04 bos.adt.libm 5.1.0.29 | |
1297 | AIX 5.2.0 bos.adt.libm 5.2.0.85 | |
1298 | ||
1299 | So, until IBM fixes powl(), we provide the following workaround to | |
1300 | handle the problem ourselves. Our logic is as follows: for | |
1301 | negative bases (left), we use fmod(right, 2) to check if the | |
1302 | exponent is an odd or even integer: | |
1303 | ||
1304 | - if odd, powl(left, right) == -powl(-left, right) | |
1305 | - if even, powl(left, right) == powl(-left, right) | |
1306 | ||
1307 | If the exponent is not an integer, the result is rightly NaNQ, so | |
1308 | we just return that (as NV_NAN). | |
1309 | */ | |
1310 | ||
1311 | if (left < 0.0) { | |
1312 | NV mod2 = Perl_fmod( right, 2.0 ); | |
1313 | if (mod2 == 1.0 || mod2 == -1.0) { /* odd integer */ | |
1314 | SETn( -Perl_pow( -left, right) ); | |
1315 | } else if (mod2 == 0.0) { /* even integer */ | |
1316 | SETn( Perl_pow( -left, right) ); | |
1317 | } else { /* fractional power */ | |
1318 | SETn( NV_NAN ); | |
1319 | } | |
1320 | } else { | |
1321 | SETn( Perl_pow( left, right) ); | |
1322 | } | |
1323 | #else | |
52a96ae6 | 1324 | SETn( Perl_pow( left, right) ); |
3aaeb624 JA |
1325 | #endif /* HAS_AIX_POWL_NEG_BASE_BUG */ |
1326 | ||
52a96ae6 HS |
1327 | #ifdef PERL_PRESERVE_IVUV |
1328 | if (is_int) | |
6f1401dc | 1329 | SvIV_please_nomg(svr); |
52a96ae6 HS |
1330 | #endif |
1331 | RETURN; | |
93a17b20 | 1332 | } |
a0d0e21e LW |
1333 | } |
1334 | ||
1335 | PP(pp_multiply) | |
1336 | { | |
20b7effb | 1337 | dSP; dATARGET; SV *svl, *svr; |
6f1401dc DM |
1338 | tryAMAGICbin_MG(mult_amg, AMGf_assign|AMGf_numeric); |
1339 | svr = TOPs; | |
1340 | svl = TOPm1s; | |
230ee21f | 1341 | |
28e5dec8 | 1342 | #ifdef PERL_PRESERVE_IVUV |
230ee21f DM |
1343 | |
1344 | /* special-case some simple common cases */ | |
1345 | if (!((svl->sv_flags|svr->sv_flags) & (SVf_IVisUV|SVs_GMG))) { | |
1346 | IV il, ir; | |
1347 | U32 flags = (svl->sv_flags & svr->sv_flags); | |
1348 | if (flags & SVf_IOK) { | |
1349 | /* both args are simple IVs */ | |
1350 | UV topl, topr; | |
1351 | il = SvIVX(svl); | |
1352 | ir = SvIVX(svr); | |
1353 | do_iv: | |
1354 | topl = ((UV)il) >> (UVSIZE * 4 - 1); | |
1355 | topr = ((UV)ir) >> (UVSIZE * 4 - 1); | |
1356 | ||
1357 | /* if both are in a range that can't under/overflow, do a | |
1358 | * simple integer multiply: if the top halves(*) of both numbers | |
1359 | * are 00...00 or 11...11, then it's safe. | |
1360 | * (*) for 32-bits, the "top half" is the top 17 bits, | |
1361 | * for 64-bits, its 33 bits */ | |
1362 | if (!( | |
1363 | ((topl+1) | (topr+1)) | |
1364 | & ( (((UV)1) << (UVSIZE * 4 + 1)) - 2) /* 11..110 */ | |
1365 | )) { | |
1366 | SP--; | |
1367 | TARGi(il * ir, 0); /* args not GMG, so can't be tainted */ | |
1368 | SETs(TARG); | |
1369 | RETURN; | |
1370 | } | |
1371 | goto generic; | |
1372 | } | |
1373 | else if (flags & SVf_NOK) { | |
1374 | /* both args are NVs */ | |
1375 | NV nl = SvNVX(svl); | |
1376 | NV nr = SvNVX(svr); | |
1377 | NV result; | |
1378 | ||
3336af0b DD |
1379 | if ( |
1380 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) | |
1381 | !Perl_isnan(nl) && nl == (NV)(il = (IV)nl) | |
1382 | && !Perl_isnan(nr) && nr == (NV)(ir = (IV)nr) | |
1383 | #else | |
1384 | nl == (NV)(il = (IV)nl) && nr == (NV)(ir = (IV)nr) | |
1385 | #endif | |
1386 | ) | |
230ee21f DM |
1387 | /* nothing was lost by converting to IVs */ |
1388 | goto do_iv; | |
1389 | SP--; | |
1390 | result = nl * nr; | |
1f02ab1d | 1391 | # if defined(__sgi) && defined(USE_LONG_DOUBLE) && LONG_DOUBLEKIND == LONG_DOUBLE_IS_DOUBLEDOUBLE_128_BIT_BE_BE && NVSIZE == 16 |
230ee21f DM |
1392 | if (Perl_isinf(result)) { |
1393 | Zero((U8*)&result + 8, 8, U8); | |
1394 | } | |
1395 | # endif | |
1396 | TARGn(result, 0); /* args not GMG, so can't be tainted */ | |
1397 | SETs(TARG); | |
1398 | RETURN; | |
1399 | } | |
1400 | } | |
1401 | ||
1402 | generic: | |
1403 | ||
01f91bf2 | 1404 | if (SvIV_please_nomg(svr)) { |
28e5dec8 JH |
1405 | /* Unless the left argument is integer in range we are going to have to |
1406 | use NV maths. Hence only attempt to coerce the right argument if | |
1407 | we know the left is integer. */ | |
1408 | /* Left operand is defined, so is it IV? */ | |
01f91bf2 | 1409 | if (SvIV_please_nomg(svl)) { |
800401ee JH |
1410 | bool auvok = SvUOK(svl); |
1411 | bool buvok = SvUOK(svr); | |
28e5dec8 JH |
1412 | const UV topmask = (~ (UV)0) << (4 * sizeof (UV)); |
1413 | const UV botmask = ~((~ (UV)0) << (4 * sizeof (UV))); | |
1414 | UV alow; | |
1415 | UV ahigh; | |
1416 | UV blow; | |
1417 | UV bhigh; | |
1418 | ||
1419 | if (auvok) { | |
800401ee | 1420 | alow = SvUVX(svl); |
28e5dec8 | 1421 | } else { |
800401ee | 1422 | const IV aiv = SvIVX(svl); |
28e5dec8 JH |
1423 | if (aiv >= 0) { |
1424 | alow = aiv; | |
1425 | auvok = TRUE; /* effectively it's a UV now */ | |
1426 | } else { | |
53e2bfb7 DM |
1427 | /* abs, auvok == false records sign */ |
1428 | alow = (aiv == IV_MIN) ? (UV)aiv : (UV)(-aiv); | |
28e5dec8 JH |
1429 | } |
1430 | } | |
1431 | if (buvok) { | |
800401ee | 1432 | blow = SvUVX(svr); |
28e5dec8 | 1433 | } else { |
800401ee | 1434 | const IV biv = SvIVX(svr); |
28e5dec8 JH |
1435 | if (biv >= 0) { |
1436 | blow = biv; | |
1437 | buvok = TRUE; /* effectively it's a UV now */ | |
1438 | } else { | |
53e2bfb7 DM |
1439 | /* abs, buvok == false records sign */ |
1440 | blow = (biv == IV_MIN) ? (UV)biv : (UV)(-biv); | |
28e5dec8 JH |
1441 | } |
1442 | } | |
1443 | ||
1444 | /* If this does sign extension on unsigned it's time for plan B */ | |
1445 | ahigh = alow >> (4 * sizeof (UV)); | |
1446 | alow &= botmask; | |
1447 | bhigh = blow >> (4 * sizeof (UV)); | |
1448 | blow &= botmask; | |
1449 | if (ahigh && bhigh) { | |
6f207bd3 | 1450 | NOOP; |
28e5dec8 JH |
1451 | /* eg 32 bit is at least 0x10000 * 0x10000 == 0x100000000 |
1452 | which is overflow. Drop to NVs below. */ | |
1453 | } else if (!ahigh && !bhigh) { | |
1454 | /* eg 32 bit is at most 0xFFFF * 0xFFFF == 0xFFFE0001 | |
1455 | so the unsigned multiply cannot overflow. */ | |
c445ea15 | 1456 | const UV product = alow * blow; |
28e5dec8 JH |
1457 | if (auvok == buvok) { |
1458 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
1459 | SP--; | |
1460 | SETu( product ); | |
1461 | RETURN; | |
1462 | } else if (product <= (UV)IV_MIN) { | |
1463 | /* 2s complement assumption that (UV)-IV_MIN is correct. */ | |
1464 | /* -ve result, which could overflow an IV */ | |
1465 | SP--; | |
02b08bbc DM |
1466 | /* can't negate IV_MIN, but there are aren't two |
1467 | * integers such that !ahigh && !bhigh, where the | |
1468 | * product equals 0x800....000 */ | |
1469 | assert(product != (UV)IV_MIN); | |
25716404 | 1470 | SETi( -(IV)product ); |
28e5dec8 JH |
1471 | RETURN; |
1472 | } /* else drop to NVs below. */ | |
1473 | } else { | |
1474 | /* One operand is large, 1 small */ | |
1475 | UV product_middle; | |
1476 | if (bhigh) { | |
1477 | /* swap the operands */ | |
1478 | ahigh = bhigh; | |
1479 | bhigh = blow; /* bhigh now the temp var for the swap */ | |
1480 | blow = alow; | |
1481 | alow = bhigh; | |
1482 | } | |
1483 | /* now, ((ahigh * blow) << half_UV_len) + (alow * blow) | |
1484 | multiplies can't overflow. shift can, add can, -ve can. */ | |
1485 | product_middle = ahigh * blow; | |
1486 | if (!(product_middle & topmask)) { | |
1487 | /* OK, (ahigh * blow) won't lose bits when we shift it. */ | |
1488 | UV product_low; | |
1489 | product_middle <<= (4 * sizeof (UV)); | |
1490 | product_low = alow * blow; | |
1491 | ||
1492 | /* as for pp_add, UV + something mustn't get smaller. | |
1493 | IIRC ANSI mandates this wrapping *behaviour* for | |
1494 | unsigned whatever the actual representation*/ | |
1495 | product_low += product_middle; | |
1496 | if (product_low >= product_middle) { | |
1497 | /* didn't overflow */ | |
1498 | if (auvok == buvok) { | |
1499 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
1500 | SP--; | |
1501 | SETu( product_low ); | |
1502 | RETURN; | |
1503 | } else if (product_low <= (UV)IV_MIN) { | |
1504 | /* 2s complement assumption again */ | |
1505 | /* -ve result, which could overflow an IV */ | |
1506 | SP--; | |
53e2bfb7 DM |
1507 | SETi(product_low == (UV)IV_MIN |
1508 | ? IV_MIN : -(IV)product_low); | |
28e5dec8 JH |
1509 | RETURN; |
1510 | } /* else drop to NVs below. */ | |
1511 | } | |
1512 | } /* product_middle too large */ | |
1513 | } /* ahigh && bhigh */ | |
800401ee JH |
1514 | } /* SvIOK(svl) */ |
1515 | } /* SvIOK(svr) */ | |
28e5dec8 | 1516 | #endif |
a0d0e21e | 1517 | { |
6f1401dc DM |
1518 | NV right = SvNV_nomg(svr); |
1519 | NV left = SvNV_nomg(svl); | |
230ee21f DM |
1520 | NV result = left * right; |
1521 | ||
4efa5a16 | 1522 | (void)POPs; |
1f02ab1d | 1523 | #if defined(__sgi) && defined(USE_LONG_DOUBLE) && LONG_DOUBLEKIND == LONG_DOUBLE_IS_DOUBLEDOUBLE_128_BIT_BE_BE && NVSIZE == 16 |
230ee21f DM |
1524 | if (Perl_isinf(result)) { |
1525 | Zero((U8*)&result + 8, 8, U8); | |
3ec400f5 | 1526 | } |
3ec400f5 | 1527 | #endif |
230ee21f | 1528 | SETn(result); |
a0d0e21e | 1529 | RETURN; |
79072805 | 1530 | } |
a0d0e21e LW |
1531 | } |
1532 | ||
1533 | PP(pp_divide) | |
1534 | { | |
20b7effb | 1535 | dSP; dATARGET; SV *svl, *svr; |
6f1401dc DM |
1536 | tryAMAGICbin_MG(div_amg, AMGf_assign|AMGf_numeric); |
1537 | svr = TOPs; | |
1538 | svl = TOPm1s; | |
5479d192 | 1539 | /* Only try to do UV divide first |
68795e93 | 1540 | if ((SLOPPYDIVIDE is true) or |
5479d192 NC |
1541 | (PERL_PRESERVE_IVUV is true and one or both SV is a UV too large |
1542 | to preserve)) | |
1543 | The assumption is that it is better to use floating point divide | |
1544 | whenever possible, only doing integer divide first if we can't be sure. | |
1545 | If NV_PRESERVES_UV is true then we know at compile time that no UV | |
1546 | can be too large to preserve, so don't need to compile the code to | |
1547 | test the size of UVs. */ | |
1548 | ||
a0d0e21e | 1549 | #ifdef SLOPPYDIVIDE |
5479d192 NC |
1550 | # define PERL_TRY_UV_DIVIDE |
1551 | /* ensure that 20./5. == 4. */ | |
a0d0e21e | 1552 | #else |
5479d192 NC |
1553 | # ifdef PERL_PRESERVE_IVUV |
1554 | # ifndef NV_PRESERVES_UV | |
1555 | # define PERL_TRY_UV_DIVIDE | |
1556 | # endif | |
1557 | # endif | |
a0d0e21e | 1558 | #endif |
5479d192 NC |
1559 | |
1560 | #ifdef PERL_TRY_UV_DIVIDE | |
01f91bf2 | 1561 | if (SvIV_please_nomg(svr) && SvIV_please_nomg(svl)) { |
800401ee JH |
1562 | bool left_non_neg = SvUOK(svl); |
1563 | bool right_non_neg = SvUOK(svr); | |
5479d192 NC |
1564 | UV left; |
1565 | UV right; | |
1566 | ||
1567 | if (right_non_neg) { | |
800401ee | 1568 | right = SvUVX(svr); |
5479d192 NC |
1569 | } |
1570 | else { | |
800401ee | 1571 | const IV biv = SvIVX(svr); |
5479d192 NC |
1572 | if (biv >= 0) { |
1573 | right = biv; | |
1574 | right_non_neg = TRUE; /* effectively it's a UV now */ | |
1575 | } | |
1576 | else { | |
02b08bbc | 1577 | right = (biv == IV_MIN) ? (UV)biv : (UV)(-biv); |
5479d192 NC |
1578 | } |
1579 | } | |
1580 | /* historically undef()/0 gives a "Use of uninitialized value" | |
1581 | warning before dieing, hence this test goes here. | |
1582 | If it were immediately before the second SvIV_please, then | |
1583 | DIE() would be invoked before left was even inspected, so | |
486ec47a | 1584 | no inspection would give no warning. */ |
5479d192 NC |
1585 | if (right == 0) |
1586 | DIE(aTHX_ "Illegal division by zero"); | |
1587 | ||
1588 | if (left_non_neg) { | |
800401ee | 1589 | left = SvUVX(svl); |
5479d192 NC |
1590 | } |
1591 | else { | |
800401ee | 1592 | const IV aiv = SvIVX(svl); |
5479d192 NC |
1593 | if (aiv >= 0) { |
1594 | left = aiv; | |
1595 | left_non_neg = TRUE; /* effectively it's a UV now */ | |
1596 | } | |
1597 | else { | |
02b08bbc | 1598 | left = (aiv == IV_MIN) ? (UV)aiv : (UV)(-aiv); |
5479d192 NC |
1599 | } |
1600 | } | |
1601 | ||
1602 | if (left >= right | |
1603 | #ifdef SLOPPYDIVIDE | |
1604 | /* For sloppy divide we always attempt integer division. */ | |
1605 | #else | |
1606 | /* Otherwise we only attempt it if either or both operands | |
1607 | would not be preserved by an NV. If both fit in NVs | |
0c2ee62a NC |
1608 | we fall through to the NV divide code below. However, |
1609 | as left >= right to ensure integer result here, we know that | |
1610 | we can skip the test on the right operand - right big | |
1611 | enough not to be preserved can't get here unless left is | |
1612 | also too big. */ | |
1613 | ||
1614 | && (left > ((UV)1 << NV_PRESERVES_UV_BITS)) | |
5479d192 NC |
1615 | #endif |
1616 | ) { | |
1617 | /* Integer division can't overflow, but it can be imprecise. */ | |
1b6737cc | 1618 | const UV result = left / right; |
5479d192 NC |
1619 | if (result * right == left) { |
1620 | SP--; /* result is valid */ | |
1621 | if (left_non_neg == right_non_neg) { | |
1622 | /* signs identical, result is positive. */ | |
1623 | SETu( result ); | |
1624 | RETURN; | |
1625 | } | |
1626 | /* 2s complement assumption */ | |
1627 | if (result <= (UV)IV_MIN) | |
02b08bbc | 1628 | SETi(result == (UV)IV_MIN ? IV_MIN : -(IV)result); |
5479d192 NC |
1629 | else { |
1630 | /* It's exact but too negative for IV. */ | |
1631 | SETn( -(NV)result ); | |
1632 | } | |
1633 | RETURN; | |
1634 | } /* tried integer divide but it was not an integer result */ | |
32fdb065 | 1635 | } /* else (PERL_ABS(result) < 1.0) or (both UVs in range for NV) */ |
01f91bf2 | 1636 | } /* one operand wasn't SvIOK */ |
5479d192 NC |
1637 | #endif /* PERL_TRY_UV_DIVIDE */ |
1638 | { | |
6f1401dc DM |
1639 | NV right = SvNV_nomg(svr); |
1640 | NV left = SvNV_nomg(svl); | |
4efa5a16 | 1641 | (void)POPs;(void)POPs; |
ebc6a117 PD |
1642 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
1643 | if (! Perl_isnan(right) && right == 0.0) | |
1644 | #else | |
659c4b96 | 1645 | if (right == 0.0) |
ebc6a117 | 1646 | #endif |
5479d192 NC |
1647 | DIE(aTHX_ "Illegal division by zero"); |
1648 | PUSHn( left / right ); | |
1649 | RETURN; | |
79072805 | 1650 | } |
a0d0e21e LW |
1651 | } |
1652 | ||
1653 | PP(pp_modulo) | |
1654 | { | |
20b7effb | 1655 | dSP; dATARGET; |
6f1401dc | 1656 | tryAMAGICbin_MG(modulo_amg, AMGf_assign|AMGf_numeric); |
a0d0e21e | 1657 | { |
9c5ffd7c JH |
1658 | UV left = 0; |
1659 | UV right = 0; | |
dc656993 JH |
1660 | bool left_neg = FALSE; |
1661 | bool right_neg = FALSE; | |
e2c88acc NC |
1662 | bool use_double = FALSE; |
1663 | bool dright_valid = FALSE; | |
9c5ffd7c JH |
1664 | NV dright = 0.0; |
1665 | NV dleft = 0.0; | |
6f1401dc DM |
1666 | SV * const svr = TOPs; |
1667 | SV * const svl = TOPm1s; | |
01f91bf2 | 1668 | if (SvIV_please_nomg(svr)) { |
800401ee | 1669 | right_neg = !SvUOK(svr); |
e2c88acc | 1670 | if (!right_neg) { |
800401ee | 1671 | right = SvUVX(svr); |
e2c88acc | 1672 | } else { |
800401ee | 1673 | const IV biv = SvIVX(svr); |
e2c88acc NC |
1674 | if (biv >= 0) { |
1675 | right = biv; | |
1676 | right_neg = FALSE; /* effectively it's a UV now */ | |
1677 | } else { | |
02b08bbc | 1678 | right = (biv == IV_MIN) ? (UV)biv : (UV)(-biv); |
e2c88acc NC |
1679 | } |
1680 | } | |
1681 | } | |
1682 | else { | |
6f1401dc | 1683 | dright = SvNV_nomg(svr); |
787eafbd IZ |
1684 | right_neg = dright < 0; |
1685 | if (right_neg) | |
1686 | dright = -dright; | |
e2c88acc NC |
1687 | if (dright < UV_MAX_P1) { |
1688 | right = U_V(dright); | |
1689 | dright_valid = TRUE; /* In case we need to use double below. */ | |
1690 | } else { | |
1691 | use_double = TRUE; | |
1692 | } | |
787eafbd | 1693 | } |
a0d0e21e | 1694 | |
e2c88acc NC |
1695 | /* At this point use_double is only true if right is out of range for |
1696 | a UV. In range NV has been rounded down to nearest UV and | |
1697 | use_double false. */ | |
01f91bf2 | 1698 | if (!use_double && SvIV_please_nomg(svl)) { |
800401ee | 1699 | left_neg = !SvUOK(svl); |
e2c88acc | 1700 | if (!left_neg) { |
800401ee | 1701 | left = SvUVX(svl); |
e2c88acc | 1702 | } else { |
800401ee | 1703 | const IV aiv = SvIVX(svl); |
e2c88acc NC |
1704 | if (aiv >= 0) { |
1705 | left = aiv; | |
1706 | left_neg = FALSE; /* effectively it's a UV now */ | |
1707 | } else { | |
02b08bbc | 1708 | left = (aiv == IV_MIN) ? (UV)aiv : (UV)(-aiv); |
e2c88acc NC |
1709 | } |
1710 | } | |
e2c88acc | 1711 | } |
787eafbd | 1712 | else { |
6f1401dc | 1713 | dleft = SvNV_nomg(svl); |
787eafbd IZ |
1714 | left_neg = dleft < 0; |
1715 | if (left_neg) | |
1716 | dleft = -dleft; | |
68dc0745 | 1717 | |
e2c88acc NC |
1718 | /* This should be exactly the 5.6 behaviour - if left and right are |
1719 | both in range for UV then use U_V() rather than floor. */ | |
1720 | if (!use_double) { | |
1721 | if (dleft < UV_MAX_P1) { | |
1722 | /* right was in range, so is dleft, so use UVs not double. | |
1723 | */ | |
1724 | left = U_V(dleft); | |
1725 | } | |
1726 | /* left is out of range for UV, right was in range, so promote | |
1727 | right (back) to double. */ | |
1728 | else { | |
1729 | /* The +0.5 is used in 5.6 even though it is not strictly | |
1730 | consistent with the implicit +0 floor in the U_V() | |
1731 | inside the #if 1. */ | |
1732 | dleft = Perl_floor(dleft + 0.5); | |
1733 | use_double = TRUE; | |
1734 | if (dright_valid) | |
1735 | dright = Perl_floor(dright + 0.5); | |
1736 | else | |
1737 | dright = right; | |
1738 | } | |
1739 | } | |
1740 | } | |
6f1401dc | 1741 | sp -= 2; |
787eafbd | 1742 | if (use_double) { |
65202027 | 1743 | NV dans; |
787eafbd | 1744 | |
659c4b96 | 1745 | if (!dright) |
cea2e8a9 | 1746 | DIE(aTHX_ "Illegal modulus zero"); |
787eafbd | 1747 | |
65202027 | 1748 | dans = Perl_fmod(dleft, dright); |
659c4b96 | 1749 | if ((left_neg != right_neg) && dans) |
787eafbd IZ |
1750 | dans = dright - dans; |
1751 | if (right_neg) | |
1752 | dans = -dans; | |
1753 | sv_setnv(TARG, dans); | |
1754 | } | |
1755 | else { | |
1756 | UV ans; | |
1757 | ||
787eafbd | 1758 | if (!right) |
cea2e8a9 | 1759 | DIE(aTHX_ "Illegal modulus zero"); |
787eafbd IZ |
1760 | |
1761 | ans = left % right; | |
1762 | if ((left_neg != right_neg) && ans) | |
1763 | ans = right - ans; | |
1764 | if (right_neg) { | |
1765 | /* XXX may warn: unary minus operator applied to unsigned type */ | |
1766 | /* could change -foo to be (~foo)+1 instead */ | |
1767 | if (ans <= ~((UV)IV_MAX)+1) | |
1768 | sv_setiv(TARG, ~ans+1); | |
1769 | else | |
65202027 | 1770 | sv_setnv(TARG, -(NV)ans); |
787eafbd IZ |
1771 | } |
1772 | else | |
1773 | sv_setuv(TARG, ans); | |
1774 | } | |
1775 | PUSHTARG; | |
1776 | RETURN; | |
79072805 | 1777 | } |
a0d0e21e | 1778 | } |
79072805 | 1779 | |
a0d0e21e LW |
1780 | PP(pp_repeat) |
1781 | { | |
20b7effb | 1782 | dSP; dATARGET; |
eb578fdb | 1783 | IV count; |
6f1401dc | 1784 | SV *sv; |
02a7a248 | 1785 | bool infnan = FALSE; |
6f1401dc | 1786 | |
82334630 | 1787 | if (GIMME_V == G_ARRAY && PL_op->op_private & OPpREPEAT_DOLIST) { |
6f1401dc DM |
1788 | /* TODO: think of some way of doing list-repeat overloading ??? */ |
1789 | sv = POPs; | |
1790 | SvGETMAGIC(sv); | |
1791 | } | |
1792 | else { | |
3a100dab FC |
1793 | if (UNLIKELY(PL_op->op_private & OPpREPEAT_DOLIST)) { |
1794 | /* The parser saw this as a list repeat, and there | |
1795 | are probably several items on the stack. But we're | |
1796 | in scalar/void context, and there's no pp_list to save us | |
1797 | now. So drop the rest of the items -- robin@kitsite.com | |
1798 | */ | |
1799 | dMARK; | |
1800 | if (MARK + 1 < SP) { | |
1801 | MARK[1] = TOPm1s; | |
1802 | MARK[2] = TOPs; | |
1803 | } | |
1804 | else { | |
1805 | dTOPss; | |
1806 | ASSUME(MARK + 1 == SP); | |
1807 | XPUSHs(sv); | |
1808 | MARK[1] = &PL_sv_undef; | |
1809 | } | |
1810 | SP = MARK + 2; | |
1811 | } | |
6f1401dc DM |
1812 | tryAMAGICbin_MG(repeat_amg, AMGf_assign); |
1813 | sv = POPs; | |
1814 | } | |
1815 | ||
2b573ace JH |
1816 | if (SvIOKp(sv)) { |
1817 | if (SvUOK(sv)) { | |
6f1401dc | 1818 | const UV uv = SvUV_nomg(sv); |
2b573ace JH |
1819 | if (uv > IV_MAX) |
1820 | count = IV_MAX; /* The best we can do? */ | |
1821 | else | |
1822 | count = uv; | |
1823 | } else { | |
b3211734 | 1824 | count = SvIV_nomg(sv); |
2b573ace JH |
1825 | } |
1826 | } | |
1827 | else if (SvNOKp(sv)) { | |
02a7a248 JH |
1828 | const NV nv = SvNV_nomg(sv); |
1829 | infnan = Perl_isinfnan(nv); | |
1830 | if (UNLIKELY(infnan)) { | |
1831 | count = 0; | |
1832 | } else { | |
1833 | if (nv < 0.0) | |
1834 | count = -1; /* An arbitrary negative integer */ | |
1835 | else | |
1836 | count = (IV)nv; | |
1837 | } | |
2b573ace JH |
1838 | } |
1839 | else | |
02a7a248 | 1840 | count = SvIV_nomg(sv); |
6f1401dc | 1841 | |
02a7a248 JH |
1842 | if (infnan) { |
1843 | Perl_ck_warner(aTHX_ packWARN(WARN_NUMERIC), | |
1844 | "Non-finite repeat count does nothing"); | |
1845 | } else if (count < 0) { | |
b3211734 KW |
1846 | count = 0; |
1847 | Perl_ck_warner(aTHX_ packWARN(WARN_NUMERIC), | |
02a7a248 | 1848 | "Negative repeat count does nothing"); |
b3211734 KW |
1849 | } |
1850 | ||
82334630 | 1851 | if (GIMME_V == G_ARRAY && PL_op->op_private & OPpREPEAT_DOLIST) { |
a0d0e21e | 1852 | dMARK; |
052a7c76 | 1853 | const SSize_t items = SP - MARK; |
da9e430b | 1854 | const U8 mod = PL_op->op_flags & OPf_MOD; |
79072805 | 1855 | |
a0d0e21e | 1856 | if (count > 1) { |
052a7c76 | 1857 | SSize_t max; |
b3b27d01 | 1858 | |
052a7c76 DM |
1859 | if ( items > SSize_t_MAX / count /* max would overflow */ |
1860 | /* repeatcpy would overflow */ | |
1861 | || items > I32_MAX / (I32)sizeof(SV *) | |
b3b27d01 DM |
1862 | ) |
1863 | Perl_croak(aTHX_ "%s","Out of memory during list extend"); | |
1864 | max = items * count; | |
1865 | MEXTEND(MARK, max); | |
1866 | ||
a0d0e21e | 1867 | while (SP > MARK) { |
60779a30 DM |
1868 | if (*SP) { |
1869 | if (mod && SvPADTMP(*SP)) { | |
da9e430b | 1870 | *SP = sv_mortalcopy(*SP); |
60779a30 | 1871 | } |
976c8a39 | 1872 | SvTEMP_off((*SP)); |
da9e430b | 1873 | } |
a0d0e21e | 1874 | SP--; |
79072805 | 1875 | } |
a0d0e21e LW |
1876 | MARK++; |
1877 | repeatcpy((char*)(MARK + items), (char*)MARK, | |
ad64d0ec | 1878 | items * sizeof(const SV *), count - 1); |
a0d0e21e | 1879 | SP += max; |
79072805 | 1880 | } |
a0d0e21e | 1881 | else if (count <= 0) |
052a7c76 | 1882 | SP = MARK; |
79072805 | 1883 | } |
a0d0e21e | 1884 | else { /* Note: mark already snarfed by pp_list */ |
0bd48802 | 1885 | SV * const tmpstr = POPs; |
a0d0e21e | 1886 | STRLEN len; |
9b877dbb | 1887 | bool isutf; |
a0d0e21e | 1888 | |
6f1401dc DM |
1889 | if (TARG != tmpstr) |
1890 | sv_setsv_nomg(TARG, tmpstr); | |
1891 | SvPV_force_nomg(TARG, len); | |
9b877dbb | 1892 | isutf = DO_UTF8(TARG); |
8ebc5c01 | 1893 | if (count != 1) { |
1894 | if (count < 1) | |
1895 | SvCUR_set(TARG, 0); | |
1896 | else { | |
b3b27d01 DM |
1897 | STRLEN max; |
1898 | ||
1899 | if ( len > (MEM_SIZE_MAX-1) / (UV)count /* max would overflow */ | |
1900 | || len > (U32)I32_MAX /* repeatcpy would overflow */ | |
1901 | ) | |
1902 | Perl_croak(aTHX_ "%s", | |
1903 | "Out of memory during string extend"); | |
1904 | max = (UV)count * len + 1; | |
1905 | SvGROW(TARG, max); | |
1906 | ||
a0d0e21e | 1907 | repeatcpy(SvPVX(TARG) + len, SvPVX(TARG), len, count - 1); |
b162af07 | 1908 | SvCUR_set(TARG, SvCUR(TARG) * count); |
7a4c00b4 | 1909 | } |
a0d0e21e | 1910 | *SvEND(TARG) = '\0'; |
a0d0e21e | 1911 | } |
dfcb284a GS |
1912 | if (isutf) |
1913 | (void)SvPOK_only_UTF8(TARG); | |
1914 | else | |
1915 | (void)SvPOK_only(TARG); | |
b80b6069 | 1916 | |
a0d0e21e | 1917 | PUSHTARG; |
79072805 | 1918 | } |
a0d0e21e LW |
1919 | RETURN; |
1920 | } | |
79072805 | 1921 | |
a0d0e21e LW |
1922 | PP(pp_subtract) |
1923 | { | |
20b7effb | 1924 | dSP; dATARGET; bool useleft; SV *svl, *svr; |
6f1401dc DM |
1925 | tryAMAGICbin_MG(subtr_amg, AMGf_assign|AMGf_numeric); |
1926 | svr = TOPs; | |
1927 | svl = TOPm1s; | |
230ee21f | 1928 | |
28e5dec8 | 1929 | #ifdef PERL_PRESERVE_IVUV |
230ee21f DM |
1930 | |
1931 | /* special-case some simple common cases */ | |
1932 | if (!((svl->sv_flags|svr->sv_flags) & (SVf_IVisUV|SVs_GMG))) { | |
1933 | IV il, ir; | |
1934 | U32 flags = (svl->sv_flags & svr->sv_flags); | |
1935 | if (flags & SVf_IOK) { | |
1936 | /* both args are simple IVs */ | |
1937 | UV topl, topr; | |
1938 | il = SvIVX(svl); | |
1939 | ir = SvIVX(svr); | |
1940 | do_iv: | |
1941 | topl = ((UV)il) >> (UVSIZE * 8 - 2); | |
1942 | topr = ((UV)ir) >> (UVSIZE * 8 - 2); | |
1943 | ||
1944 | /* if both are in a range that can't under/overflow, do a | |
1945 | * simple integer subtract: if the top of both numbers | |
1946 | * are 00 or 11, then it's safe */ | |
1947 | if (!( ((topl+1) | (topr+1)) & 2)) { | |
1948 | SP--; | |
1949 | TARGi(il - ir, 0); /* args not GMG, so can't be tainted */ | |
1950 | SETs(TARG); | |
1951 | RETURN; | |
1952 | } | |
1953 | goto generic; | |
1954 | } | |
1955 | else if (flags & SVf_NOK) { | |
1956 | /* both args are NVs */ | |
1957 | NV nl = SvNVX(svl); | |
1958 | NV nr = SvNVX(svr); | |
1959 | ||
3336af0b DD |
1960 | if ( |
1961 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) | |
1962 | !Perl_isnan(nl) && nl == (NV)(il = (IV)nl) | |
1963 | && !Perl_isnan(nr) && nr == (NV)(ir = (IV)nr) | |
1964 | #else | |
1965 | nl == (NV)(il = (IV)nl) && nr == (NV)(ir = (IV)nr) | |
1966 | #endif | |
1967 | ) | |
230ee21f DM |
1968 | /* nothing was lost by converting to IVs */ |
1969 | goto do_iv; | |
1970 | SP--; | |
1971 | TARGn(nl - nr, 0); /* args not GMG, so can't be tainted */ | |
1972 | SETs(TARG); | |
1973 | RETURN; | |
1974 | } | |
1975 | } | |
1976 | ||
1977 | generic: | |
1978 | ||
1979 | useleft = USE_LEFT(svl); | |
7dca457a NC |
1980 | /* See comments in pp_add (in pp_hot.c) about Overflow, and how |
1981 | "bad things" happen if you rely on signed integers wrapping. */ | |
01f91bf2 | 1982 | if (SvIV_please_nomg(svr)) { |
28e5dec8 JH |
1983 | /* Unless the left argument is integer in range we are going to have to |
1984 | use NV maths. Hence only attempt to coerce the right argument if | |
1985 | we know the left is integer. */ | |
eb578fdb | 1986 | UV auv = 0; |
9c5ffd7c | 1987 | bool auvok = FALSE; |
7dca457a NC |
1988 | bool a_valid = 0; |
1989 | ||
28e5dec8 | 1990 | if (!useleft) { |
7dca457a NC |
1991 | auv = 0; |
1992 | a_valid = auvok = 1; | |
1993 | /* left operand is undef, treat as zero. */ | |
28e5dec8 JH |
1994 | } else { |
1995 | /* Left operand is defined, so is it IV? */ | |
01f91bf2 | 1996 | if (SvIV_please_nomg(svl)) { |
800401ee JH |
1997 | if ((auvok = SvUOK(svl))) |
1998 | auv = SvUVX(svl); | |
7dca457a | 1999 | else { |
eb578fdb | 2000 | const IV aiv = SvIVX(svl); |
7dca457a NC |
2001 | if (aiv >= 0) { |
2002 | auv = aiv; | |
2003 | auvok = 1; /* Now acting as a sign flag. */ | |
2004 | } else { /* 2s complement assumption for IV_MIN */ | |
53e2bfb7 | 2005 | auv = (aiv == IV_MIN) ? (UV)aiv : (UV)-aiv; |
28e5dec8 | 2006 | } |
7dca457a NC |
2007 | } |
2008 | a_valid = 1; | |
2009 | } | |
2010 | } | |
2011 | if (a_valid) { | |
2012 | bool result_good = 0; | |
2013 | UV result; | |
eb578fdb | 2014 | UV buv; |
800401ee | 2015 | bool buvok = SvUOK(svr); |
9041c2e3 | 2016 | |
7dca457a | 2017 | if (buvok) |
800401ee | 2018 | buv = SvUVX(svr); |
7dca457a | 2019 | else { |
eb578fdb | 2020 | const IV biv = SvIVX(svr); |
7dca457a NC |
2021 | if (biv >= 0) { |
2022 | buv = biv; | |
2023 | buvok = 1; | |
2024 | } else | |
53e2bfb7 | 2025 | buv = (biv == IV_MIN) ? (UV)biv : (UV)-biv; |
7dca457a NC |
2026 | } |
2027 | /* ?uvok if value is >= 0. basically, flagged as UV if it's +ve, | |
602f51c4 | 2028 | else "IV" now, independent of how it came in. |
7dca457a NC |
2029 | if a, b represents positive, A, B negative, a maps to -A etc |
2030 | a - b => (a - b) | |
2031 | A - b => -(a + b) | |
2032 | a - B => (a + b) | |
2033 | A - B => -(a - b) | |
2034 | all UV maths. negate result if A negative. | |
2035 | subtract if signs same, add if signs differ. */ | |
2036 | ||
2037 | if (auvok ^ buvok) { | |
2038 | /* Signs differ. */ | |
2039 | result = auv + buv; | |
2040 | if (result >= auv) | |
2041 | result_good = 1; | |
2042 | } else { | |
2043 | /* Signs same */ | |
2044 | if (auv >= buv) { | |
2045 | result = auv - buv; | |
2046 | /* Must get smaller */ | |
2047 | if (result <= auv) | |
2048 | result_good = 1; | |
2049 | } else { | |
2050 | result = buv - auv; | |
2051 | if (result <= buv) { | |
2052 | /* result really should be -(auv-buv). as its negation | |
2053 | of true value, need to swap our result flag */ | |
2054 | auvok = !auvok; | |
2055 | result_good = 1; | |
28e5dec8 | 2056 | } |
28e5dec8 JH |
2057 | } |
2058 | } | |
7dca457a NC |
2059 | if (result_good) { |
2060 | SP--; | |
2061 | if (auvok) | |
2062 | SETu( result ); | |
2063 | else { | |
2064 | /* Negate result */ | |
2065 | if (result <= (UV)IV_MIN) | |
53e2bfb7 DM |
2066 | SETi(result == (UV)IV_MIN |
2067 | ? IV_MIN : -(IV)result); | |
7dca457a NC |
2068 | else { |
2069 | /* result valid, but out of range for IV. */ | |
2070 | SETn( -(NV)result ); | |
2071 | } | |
2072 | } | |
2073 | RETURN; | |
2074 | } /* Overflow, drop through to NVs. */ | |
28e5dec8 JH |
2075 | } |
2076 | } | |
230ee21f DM |
2077 | #else |
2078 | useleft = USE_LEFT(svl); | |
28e5dec8 | 2079 | #endif |
a0d0e21e | 2080 | { |
6f1401dc | 2081 | NV value = SvNV_nomg(svr); |
4efa5a16 RD |
2082 | (void)POPs; |
2083 | ||
28e5dec8 JH |
2084 | if (!useleft) { |
2085 | /* left operand is undef, treat as zero - value */ | |
2086 | SETn(-value); | |
2087 | RETURN; | |
2088 | } | |
6f1401dc | 2089 | SETn( SvNV_nomg(svl) - value ); |
28e5dec8 | 2090 | RETURN; |
79072805 | 2091 | } |
a0d0e21e | 2092 | } |
79072805 | 2093 | |
b3498293 JH |
2094 | #define IV_BITS (IVSIZE * 8) |
2095 | ||
2096 | static UV S_uv_shift(UV uv, int shift, bool left) | |
2097 | { | |
2098 | if (shift < 0) { | |
2099 | shift = -shift; | |
2100 | left = !left; | |
2101 | } | |
2102 | if (shift >= IV_BITS) { | |
2103 | return 0; | |
2104 | } | |
2105 | return left ? uv << shift : uv >> shift; | |
2106 | } | |
2107 | ||
2108 | static IV S_iv_shift(IV iv, int shift, bool left) | |
2109 | { | |
2110 | if (shift < 0) { | |
2111 | shift = -shift; | |
2112 | left = !left; | |
2113 | } | |
2114 | if (shift >= IV_BITS) { | |
b69687e7 | 2115 | return iv < 0 && !left ? -1 : 0; |
b3498293 JH |
2116 | } |
2117 | return left ? iv << shift : iv >> shift; | |
2118 | } | |
2119 | ||
2120 | #define UV_LEFT_SHIFT(uv, shift) S_uv_shift(uv, shift, TRUE) | |
2121 | #define UV_RIGHT_SHIFT(uv, shift) S_uv_shift(uv, shift, FALSE) | |
2122 | #define IV_LEFT_SHIFT(iv, shift) S_iv_shift(iv, shift, TRUE) | |
2123 | #define IV_RIGHT_SHIFT(iv, shift) S_iv_shift(iv, shift, FALSE) | |
2124 | ||
a0d0e21e LW |
2125 | PP(pp_left_shift) |
2126 | { | |
20b7effb | 2127 | dSP; dATARGET; SV *svl, *svr; |
a42d0242 | 2128 | tryAMAGICbin_MG(lshift_amg, AMGf_assign|AMGf_numeric); |
6f1401dc DM |
2129 | svr = POPs; |
2130 | svl = TOPs; | |
a0d0e21e | 2131 | { |
6f1401dc | 2132 | const IV shift = SvIV_nomg(svr); |
d0ba1bd2 | 2133 | if (PL_op->op_private & HINT_INTEGER) { |
b3498293 | 2134 | SETi(IV_LEFT_SHIFT(SvIV_nomg(svl), shift)); |
d0ba1bd2 JH |
2135 | } |
2136 | else { | |
b3498293 | 2137 | SETu(UV_LEFT_SHIFT(SvUV_nomg(svl), shift)); |
d0ba1bd2 | 2138 | } |
55497cff | 2139 | RETURN; |
79072805 | 2140 | } |
a0d0e21e | 2141 | } |
79072805 | 2142 | |
a0d0e21e LW |
2143 | PP(pp_right_shift) |
2144 | { | |
20b7effb | 2145 | dSP; dATARGET; SV *svl, *svr; |
a42d0242 | 2146 | tryAMAGICbin_MG(rshift_amg, AMGf_assign|AMGf_numeric); |
6f1401dc DM |
2147 | svr = POPs; |
2148 | svl = TOPs; | |
a0d0e21e | 2149 | { |
6f1401dc | 2150 | const IV shift = SvIV_nomg(svr); |
d0ba1bd2 | 2151 | if (PL_op->op_private & HINT_INTEGER) { |
b3498293 | 2152 | SETi(IV_RIGHT_SHIFT(SvIV_nomg(svl), shift)); |
d0ba1bd2 JH |
2153 | } |
2154 | else { | |
b3498293 | 2155 | SETu(UV_RIGHT_SHIFT(SvUV_nomg(svl), shift)); |
d0ba1bd2 | 2156 | } |
a0d0e21e | 2157 | RETURN; |
93a17b20 | 2158 | } |
79072805 LW |
2159 | } |
2160 | ||
a0d0e21e | 2161 | PP(pp_lt) |
79072805 | 2162 | { |
20b7effb | 2163 | dSP; |
33efebe6 DM |
2164 | SV *left, *right; |
2165 | ||
a42d0242 | 2166 | tryAMAGICbin_MG(lt_amg, AMGf_set|AMGf_numeric); |
33efebe6 DM |
2167 | right = POPs; |
2168 | left = TOPs; | |
2169 | SETs(boolSV( | |
2170 | (SvIOK_notUV(left) && SvIOK_notUV(right)) | |
2171 | ? (SvIVX(left) < SvIVX(right)) | |
2172 | : (do_ncmp(left, right) == -1) | |
2173 | )); | |
2174 | RETURN; | |
a0d0e21e | 2175 | } |
79072805 | 2176 | |
a0d0e21e LW |
2177 | PP(pp_gt) |
2178 | { | |
20b7effb | 2179 | dSP; |
33efebe6 | 2180 | SV *left, *right; |
1b6737cc | 2181 | |
33efebe6 DM |
2182 | tryAMAGICbin_MG(gt_amg, AMGf_set|AMGf_numeric); |
2183 | right = POPs; | |
2184 | left = TOPs; | |
2185 | SETs(boolSV( | |
2186 | (SvIOK_notUV(left) && SvIOK_notUV(right)) | |
2187 | ? (SvIVX(left) > SvIVX(right)) | |
2188 | : (do_ncmp(left, right) == 1) | |
2189 | )); | |
2190 | RETURN; | |
a0d0e21e LW |
2191 | } |
2192 | ||
2193 | PP(pp_le) | |
2194 | { | |
20b7effb | 2195 | dSP; |
33efebe6 | 2196 | SV *left, *right; |
1b6737cc | 2197 | |
33efebe6 DM |
2198 | tryAMAGICbin_MG(le_amg, AMGf_set|AMGf_numeric); |
2199 | right = POPs; | |
2200 | left = TOPs; | |
2201 | SETs(boolSV( | |
2202 | (SvIOK_notUV(left) && SvIOK_notUV(right)) | |
2203 | ? (SvIVX(left) <= SvIVX(right)) | |
2204 | : (do_ncmp(left, right) <= 0) | |
2205 | )); | |
2206 | RETURN; | |
a0d0e21e LW |
2207 | } |
2208 | ||
2209 | PP(pp_ge) | |
2210 | { | |
20b7effb | 2211 | dSP; |
33efebe6 DM |
2212 | SV *left, *right; |
2213 | ||
2214 | tryAMAGICbin_MG(ge_amg, AMGf_set|AMGf_numeric); | |
2215 | right = POPs; | |
2216 | left = TOPs; | |
2217 | SETs(boolSV( | |
2218 | (SvIOK_notUV(left) && SvIOK_notUV(right)) | |
2219 | ? (SvIVX(left) >= SvIVX(right)) | |
2220 | : ( (do_ncmp(left, right) & 2) == 0) | |
2221 | )); | |
2222 | RETURN; | |
2223 | } | |
1b6737cc | 2224 | |
33efebe6 DM |
2225 | PP(pp_ne) |
2226 | { | |
20b7effb | 2227 | dSP; |
33efebe6 DM |
2228 | SV *left, *right; |
2229 | ||
2230 | tryAMAGICbin_MG(ne_amg, AMGf_set|AMGf_numeric); | |
2231 | right = POPs; | |
2232 | left = TOPs; | |
2233 | SETs(boolSV( | |
2234 | (SvIOK_notUV(left) && SvIOK_notUV(right)) | |
2235 | ? (SvIVX(left) != SvIVX(right)) | |
2236 | : (do_ncmp(left, right) != 0) | |
2237 | )); | |
2238 | RETURN; | |
2239 | } | |
1b6737cc | 2240 | |
33efebe6 DM |
2241 | /* compare left and right SVs. Returns: |
2242 | * -1: < | |
2243 | * 0: == | |
2244 | * 1: > | |
2245 | * 2: left or right was a NaN | |
2246 | */ | |
2247 | I32 | |
2248 | Perl_do_ncmp(pTHX_ SV* const left, SV * const right) | |
2249 | { | |
33efebe6 DM |
2250 | PERL_ARGS_ASSERT_DO_NCMP; |
2251 | #ifdef PERL_PRESERVE_IVUV | |
33efebe6 | 2252 | /* Fortunately it seems NaN isn't IOK */ |
01f91bf2 | 2253 | if (SvIV_please_nomg(right) && SvIV_please_nomg(left)) { |
33efebe6 DM |
2254 | if (!SvUOK(left)) { |
2255 | const IV leftiv = SvIVX(left); | |
2256 | if (!SvUOK(right)) { | |
2257 | /* ## IV <=> IV ## */ | |
2258 | const IV rightiv = SvIVX(right); | |
2259 | return (leftiv > rightiv) - (leftiv < rightiv); | |
28e5dec8 | 2260 | } |
33efebe6 DM |
2261 | /* ## IV <=> UV ## */ |
2262 | if (leftiv < 0) | |
2263 | /* As (b) is a UV, it's >=0, so it must be < */ | |
2264 | return -1; | |
2265 | { | |
2266 | const UV rightuv = SvUVX(right); | |
2267 | return ((UV)leftiv > rightuv) - ((UV)leftiv < rightuv); | |
28e5dec8 | 2268 | } |
28e5dec8 | 2269 | } |
79072805 | 2270 | |
33efebe6 DM |
2271 | if (SvUOK(right)) { |
2272 | /* ## UV <=> UV ## */ | |
2273 | const UV leftuv = SvUVX(left); | |
2274 | const UV rightuv = SvUVX(right); | |
2275 | return (leftuv > rightuv) - (leftuv < rightuv); | |
28e5dec8 | 2276 | } |
33efebe6 DM |
2277 | /* ## UV <=> IV ## */ |
2278 | { | |
2279 | const IV rightiv = SvIVX(right); | |
2280 | if (rightiv < 0) | |
2281 | /* As (a) is a UV, it's >=0, so it cannot be < */ | |
2282 | return 1; | |
2283 | { | |
2284 | const UV leftuv = SvUVX(left); | |
2285 | return (leftuv > (UV)rightiv) - (leftuv < (UV)rightiv); | |
28e5dec8 | 2286 | } |
28e5dec8 | 2287 | } |
e5964223 | 2288 | NOT_REACHED; /* NOTREACHED */ |
28e5dec8 JH |
2289 | } |
2290 | #endif | |
a0d0e21e | 2291 | { |
33efebe6 DM |
2292 | NV const rnv = SvNV_nomg(right); |
2293 | NV const lnv = SvNV_nomg(left); | |
2294 | ||
cab190d4 | 2295 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
33efebe6 DM |
2296 | if (Perl_isnan(lnv) || Perl_isnan(rnv)) { |
2297 | return 2; | |
2298 | } | |
2299 | return (lnv > rnv) - (lnv < rnv); | |
cab190d4 | 2300 | #else |
33efebe6 DM |
2301 | if (lnv < rnv) |
2302 | return -1; | |
2303 | if (lnv > rnv) | |
2304 | return 1; | |
659c4b96 | 2305 | if (lnv == rnv) |
33efebe6 DM |
2306 | return 0; |
2307 | return 2; | |
cab190d4 | 2308 | #endif |
a0d0e21e | 2309 | } |
79072805 LW |
2310 | } |
2311 | ||
33efebe6 | 2312 | |
a0d0e21e | 2313 | PP(pp_ncmp) |
79072805 | 2314 | { |
20b7effb | 2315 | dSP; |
33efebe6 DM |
2316 | SV *left, *right; |
2317 | I32 value; | |
a42d0242 | 2318 | tryAMAGICbin_MG(ncmp_amg, AMGf_numeric); |
33efebe6 DM |
2319 | right = POPs; |
2320 | left = TOPs; | |
2321 | value = do_ncmp(left, right); | |
2322 | if (value == 2) { | |
3280af22 | 2323 | SETs(&PL_sv_undef); |
79072805 | 2324 | } |
33efebe6 DM |
2325 | else { |
2326 | dTARGET; | |
2327 | SETi(value); | |
2328 | } | |
2329 | RETURN; | |
a0d0e21e | 2330 | } |
79072805 | 2331 | |
b1c05ba5 DM |
2332 | |
2333 | /* also used for: pp_sge() pp_sgt() pp_slt() */ | |
2334 | ||
afd9910b | 2335 | PP(pp_sle) |
a0d0e21e | 2336 | { |
20b7effb | 2337 | dSP; |
79072805 | 2338 | |
afd9910b NC |
2339 | int amg_type = sle_amg; |
2340 | int multiplier = 1; | |
2341 | int rhs = 1; | |
79072805 | 2342 | |
afd9910b NC |
2343 | switch (PL_op->op_type) { |
2344 | case OP_SLT: | |
2345 | amg_type = slt_amg; | |
2346 | /* cmp < 0 */ | |
2347 | rhs = 0; | |
2348 | break; | |
2349 | case OP_SGT: | |
2350 | amg_type = sgt_amg; | |
2351 | /* cmp > 0 */ | |
2352 | multiplier = -1; | |
2353 | rhs = 0; | |
2354 | break; | |
2355 | case OP_SGE: | |
2356 | amg_type = sge_amg; | |
2357 | /* cmp >= 0 */ | |
2358 | multiplier = -1; | |
2359 | break; | |
79072805 | 2360 | } |
79072805 | 2361 | |
6f1401dc | 2362 | tryAMAGICbin_MG(amg_type, AMGf_set); |
a0d0e21e LW |
2363 | { |
2364 | dPOPTOPssrl; | |
130c5df3 | 2365 | const int cmp = |
5778acb6 | 2366 | #ifdef USE_LOCALE_COLLATE |
130c5df3 KW |
2367 | (IN_LC_RUNTIME(LC_COLLATE)) |
2368 | ? sv_cmp_locale_flags(left, right, 0) | |
2369 | : | |
2370 | #endif | |
2371 | sv_cmp_flags(left, right, 0); | |
afd9910b | 2372 | SETs(boolSV(cmp * multiplier < rhs)); |
a0d0e21e LW |
2373 | RETURN; |
2374 | } | |
2375 | } | |
79072805 | 2376 | |
36477c24 | 2377 | PP(pp_seq) |
2378 | { | |
20b7effb | 2379 | dSP; |
6f1401dc | 2380 | tryAMAGICbin_MG(seq_amg, AMGf_set); |
36477c24 | 2381 | { |
2382 | dPOPTOPssrl; | |
078504b2 | 2383 | SETs(boolSV(sv_eq_flags(left, right, 0))); |
a0d0e21e LW |
2384 | RETURN; |
2385 | } | |
2386 | } | |
79072805 | 2387 | |
a0d0e21e | 2388 | PP(pp_sne) |
79072805 | 2389 | { |
20b7effb | 2390 | dSP; |
6f1401dc | 2391 | tryAMAGICbin_MG(sne_amg, AMGf_set); |
a0d0e21e LW |
2392 | { |
2393 | dPOPTOPssrl; | |
078504b2 | 2394 | SETs(boolSV(!sv_eq_flags(left, right, 0))); |
a0d0e21e | 2395 | RETURN; |
463ee0b2 | 2396 | } |
79072805 LW |
2397 | } |
2398 | ||
a0d0e21e | 2399 | PP(pp_scmp) |
79072805 | 2400 | { |
20b7effb | 2401 | dSP; dTARGET; |
6f1401dc | 2402 | tryAMAGICbin_MG(scmp_amg, 0); |
a0d0e21e LW |
2403 | { |
2404 | dPOPTOPssrl; | |
130c5df3 | 2405 | const int cmp = |
5778acb6 | 2406 | #ifdef USE_LOCALE_COLLATE |
130c5df3 KW |
2407 | (IN_LC_RUNTIME(LC_COLLATE)) |
2408 | ? sv_cmp_locale_flags(left, right, 0) | |
2409 | : | |
2410 | #endif | |
2411 | sv_cmp_flags(left, right, 0); | |
bbce6d69 | 2412 | SETi( cmp ); |
a0d0e21e LW |
2413 | RETURN; |
2414 | } | |
2415 | } | |
79072805 | 2416 | |
55497cff | 2417 | PP(pp_bit_and) |
2418 | { | |
20b7effb | 2419 | dSP; dATARGET; |
6f1401dc | 2420 | tryAMAGICbin_MG(band_amg, AMGf_assign); |
a0d0e21e LW |
2421 | { |
2422 | dPOPTOPssrl; | |
4633a7c4 | 2423 | if (SvNIOKp(left) || SvNIOKp(right)) { |
b20c4ee1 FC |
2424 | const bool left_ro_nonnum = !SvNIOKp(left) && SvREADONLY(left); |
2425 | const bool right_ro_nonnum = !SvNIOKp(right) && SvREADONLY(right); | |
d0ba1bd2 | 2426 | if (PL_op->op_private & HINT_INTEGER) { |
1b6737cc | 2427 | const IV i = SvIV_nomg(left) & SvIV_nomg(right); |
972b05a9 | 2428 | SETi(i); |
d0ba1bd2 JH |
2429 | } |
2430 | else { | |
1b6737cc | 2431 | const UV u = SvUV_nomg(left) & SvUV_nomg(right); |
972b05a9 | 2432 | SETu(u); |
d0ba1bd2 | 2433 | } |
5ee80e13 | 2434 | if (left_ro_nonnum && left != TARG) SvNIOK_off(left); |
b20c4ee1 | 2435 | if (right_ro_nonnum) SvNIOK_off(right); |
a0d0e21e LW |
2436 | } |
2437 | else { | |
533c011a | 2438 | do_vop(PL_op->op_type, TARG, left, right); |
a0d0e21e LW |
2439 | SETTARG; |
2440 | } | |
2441 | RETURN; | |
2442 | } | |
2443 | } | |
79072805 | 2444 | |
5d01050a FC |
2445 | PP(pp_nbit_and) |
2446 | { | |
2447 | dSP; | |
636ac8fc | 2448 | tryAMAGICbin_MG(band_amg, AMGf_assign|AMGf_numarg); |
5d01050a FC |
2449 | { |
2450 | dATARGET; dPOPTOPssrl; | |
2451 | if (PL_op->op_private & HINT_INTEGER) { | |
2452 | const IV i = SvIV_nomg(left) & SvIV_nomg(right); | |
2453 | SETi(i); | |
2454 | } | |
2455 | else { | |
2456 | const UV u = SvUV_nomg(left) & SvUV_nomg(right); | |
2457 | SETu(u); | |
2458 | } | |
2459 | } | |
2460 | RETURN; | |
2461 | } | |
2462 | ||
2463 | PP(pp_sbit_and) | |
2464 | { | |
2465 | dSP; | |
2466 | tryAMAGICbin_MG(sband_amg, AMGf_assign); | |
2467 | { | |
2468 | dATARGET; dPOPTOPssrl; | |
2469 | do_vop(OP_BIT_AND, TARG, left, right); | |
2470 | RETSETTARG; | |
2471 | } | |
2472 | } | |
b1c05ba5 DM |
2473 | |
2474 | /* also used for: pp_bit_xor() */ | |
2475 | ||
a0d0e21e LW |
2476 | PP(pp_bit_or) |
2477 | { | |
20b7effb | 2478 | dSP; dATARGET; |
3658c1f1 NC |
2479 | const int op_type = PL_op->op_type; |
2480 | ||
6f1401dc | 2481 | tryAMAGICbin_MG((op_type == OP_BIT_OR ? bor_amg : bxor_amg), AMGf_assign); |
a0d0e21e LW |
2482 | { |
2483 | dPOPTOPssrl; | |
4633a7c4 | 2484 | if (SvNIOKp(left) || SvNIOKp(right)) { |
b20c4ee1 FC |
2485 | const bool left_ro_nonnum = !SvNIOKp(left) && SvREADONLY(left); |
2486 | const bool right_ro_nonnum = !SvNIOKp(right) && SvREADONLY(right); | |
d0ba1bd2 | 2487 | if (PL_op->op_private & HINT_INTEGER) { |
3658c1f1 NC |
2488 | const IV l = (USE_LEFT(left) ? SvIV_nomg(left) : 0); |
2489 | const IV r = SvIV_nomg(right); | |
2490 | const IV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
2491 | SETi(result); | |
d0ba1bd2 JH |
2492 | } |
2493 | else { | |
3658c1f1 NC |
2494 | const UV l = (USE_LEFT(left) ? SvUV_nomg(left) : 0); |
2495 | const UV r = SvUV_nomg(right); | |
2496 | const UV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
2497 | SETu(result); | |
d0ba1bd2 | 2498 | } |
5ee80e13 | 2499 | if (left_ro_nonnum && left != TARG) SvNIOK_off(left); |
b20c4ee1 | 2500 | if (right_ro_nonnum) SvNIOK_off(right); |
a0d0e21e LW |
2501 | } |
2502 | else { | |
3658c1f1 | 2503 | do_vop(op_type, TARG, left, right); |
a0d0e21e LW |
2504 | SETTARG; |
2505 | } | |
2506 | RETURN; | |
79072805 | 2507 | } |
a0d0e21e | 2508 | } |
79072805 | 2509 | |
5d01050a FC |
2510 | /* also used for: pp_nbit_xor() */ |
2511 | ||
2512 | PP(pp_nbit_or) | |
2513 | { | |
2514 | dSP; | |
2515 | const int op_type = PL_op->op_type; | |
2516 | ||
2517 | tryAMAGICbin_MG((op_type == OP_NBIT_OR ? bor_amg : bxor_amg), | |
636ac8fc | 2518 | AMGf_assign|AMGf_numarg); |
5d01050a FC |
2519 | { |
2520 | dATARGET; dPOPTOPssrl; | |
2521 | if (PL_op->op_private & HINT_INTEGER) { | |
2522 | const IV l = (USE_LEFT(left) ? SvIV_nomg(left) : 0); | |
2523 | const IV r = SvIV_nomg(right); | |
2524 | const IV result = op_type == OP_NBIT_OR ? (l | r) : (l ^ r); | |
2525 | SETi(result); | |
2526 | } | |
2527 | else { | |
2528 | const UV l = (USE_LEFT(left) ? SvUV_nomg(left) : 0); | |
2529 | const UV r = SvUV_nomg(right); | |
2530 | const UV result = op_type == OP_NBIT_OR ? (l | r) : (l ^ r); | |
2531 | SETu(result); | |
2532 | } | |
2533 | } | |
2534 | RETURN; | |
2535 | } | |
2536 | ||
2537 | /* also used for: pp_sbit_xor() */ | |
2538 | ||
2539 | PP(pp_sbit_or) | |
2540 | { | |
2541 | dSP; | |
2542 | const int op_type = PL_op->op_type; | |
2543 | ||
2544 | tryAMAGICbin_MG((op_type == OP_SBIT_OR ? sbor_amg : sbxor_amg), | |
2545 | AMGf_assign); | |
2546 | { | |
2547 | dATARGET; dPOPTOPssrl; | |
2548 | do_vop(op_type == OP_SBIT_OR ? OP_BIT_OR : OP_BIT_XOR, TARG, left, | |
2549 | right); | |
2550 | RETSETTARG; | |
2551 | } | |
2552 | } | |
2553 | ||
1c2b3fd6 FC |
2554 | PERL_STATIC_INLINE bool |
2555 | S_negate_string(pTHX) | |
2556 | { | |
2557 | dTARGET; dSP; | |
2558 | STRLEN len; | |
2559 | const char *s; | |
2560 | SV * const sv = TOPs; | |
2561 | if (!SvPOKp(sv) || SvNIOK(sv) || (!SvPOK(sv) && SvNIOKp(sv))) | |
2562 | return FALSE; | |
2563 | s = SvPV_nomg_const(sv, len); | |
2564 | if (isIDFIRST(*s)) { | |
2565 | sv_setpvs(TARG, "-"); | |
2566 | sv_catsv(TARG, sv); | |
2567 | } | |
2568 | else if (*s == '+' || (*s == '-' && !looks_like_number(sv))) { | |
2569 | sv_setsv_nomg(TARG, sv); | |
2570 | *SvPV_force_nomg(TARG, len) = *s == '-' ? '+' : '-'; | |
2571 | } | |
2572 | else return FALSE; | |
245d035e | 2573 | SETTARG; |
1c2b3fd6 FC |
2574 | return TRUE; |
2575 | } | |
2576 | ||
a0d0e21e LW |
2577 | PP(pp_negate) |
2578 | { | |
20b7effb | 2579 | dSP; dTARGET; |
6f1401dc | 2580 | tryAMAGICun_MG(neg_amg, AMGf_numeric); |
1c2b3fd6 | 2581 | if (S_negate_string(aTHX)) return NORMAL; |
a0d0e21e | 2582 | { |
6f1401dc | 2583 | SV * const sv = TOPs; |
a5b92898 | 2584 | |
d96ab1b5 | 2585 | if (SvIOK(sv)) { |
7dbe3150 | 2586 | /* It's publicly an integer */ |
28e5dec8 | 2587 | oops_its_an_int: |
9b0e499b GS |
2588 | if (SvIsUV(sv)) { |
2589 | if (SvIVX(sv) == IV_MIN) { | |
28e5dec8 | 2590 | /* 2s complement assumption. */ |
d14578b8 KW |
2591 | SETi(SvIVX(sv)); /* special case: -((UV)IV_MAX+1) == |
2592 | IV_MIN */ | |
245d035e | 2593 | return NORMAL; |
9b0e499b GS |
2594 | } |
2595 | else if (SvUVX(sv) <= IV_MAX) { | |
beccb14c | 2596 | SETi(-SvIVX(sv)); |
245d035e | 2597 | return NORMAL; |
9b0e499b GS |
2598 | } |
2599 | } | |
2600 | else if (SvIVX(sv) != IV_MIN) { | |
2601 | SETi(-SvIVX(sv)); | |
245d035e | 2602 | return NORMAL; |
9b0e499b | 2603 | } |
28e5dec8 JH |
2604 | #ifdef PERL_PRESERVE_IVUV |
2605 | else { | |
2606 | SETu((UV)IV_MIN); | |
245d035e | 2607 | return NORMAL; |
28e5dec8 JH |
2608 | } |
2609 | #endif | |
9b0e499b | 2610 | } |
8a5decd8 | 2611 | if (SvNIOKp(sv) && (SvNIOK(sv) || !SvPOK(sv))) |
6f1401dc | 2612 | SETn(-SvNV_nomg(sv)); |
1c2b3fd6 | 2613 | else if (SvPOKp(sv) && SvIV_please_nomg(sv)) |
8eb28a70 | 2614 | goto oops_its_an_int; |
4633a7c4 | 2615 | else |
6f1401dc | 2616 | SETn(-SvNV_nomg(sv)); |
79072805 | 2617 | } |
245d035e | 2618 | return NORMAL; |
79072805 LW |
2619 | } |
2620 | ||
a0d0e21e | 2621 | PP(pp_not) |
79072805 | 2622 | { |
20b7effb | 2623 | dSP; |
6f1401dc | 2624 | tryAMAGICun_MG(not_amg, AMGf_set); |
06c841cf | 2625 | *PL_stack_sp = boolSV(!SvTRUE_nomg(*PL_stack_sp)); |
a0d0e21e | 2626 | return NORMAL; |
79072805 LW |
2627 | } |
2628 | ||
5d01050a FC |
2629 | static void |
2630 | S_scomplement(pTHX_ SV *targ, SV *sv) | |
79072805 | 2631 | { |
eb578fdb KW |
2632 | U8 *tmps; |
2633 | I32 anum; | |
a0d0e21e LW |
2634 | STRLEN len; |
2635 | ||
85b0ee6e FC |
2636 | sv_copypv_nomg(TARG, sv); |
2637 | tmps = (U8*)SvPV_nomg(TARG, len); | |
08b6664b | 2638 | |
1d68d6cd | 2639 | if (SvUTF8(TARG)) { |
08b6664b KW |
2640 | if (len && ! utf8_to_bytes(tmps, &len)) { |
2641 | Perl_croak(aTHX_ fatal_above_ff_msg, PL_op_desc[PL_op->op_type]); | |
2642 | } | |
2643 | SvCUR(TARG) = len; | |
2644 | SvUTF8_off(TARG); | |
2645 | } | |
2646 | ||
2647 | anum = len; | |
1d68d6cd | 2648 | |
a0d0e21e | 2649 | #ifdef LIBERAL |
51723571 | 2650 | { |
eb578fdb | 2651 | long *tmpl; |
51723571 JH |
2652 | for ( ; anum && (unsigned long)tmps % sizeof(long); anum--, tmps++) |
2653 | *tmps = ~*tmps; | |
2654 | tmpl = (long*)tmps; | |
bb7a0f54 | 2655 | for ( ; anum >= (I32)sizeof(long); anum -= (I32)sizeof(long), tmpl++) |
51723571 JH |
2656 | *tmpl = ~*tmpl; |
2657 | tmps = (U8*)tmpl; | |
2658 | } | |
a0d0e21e LW |
2659 | #endif |
2660 | for ( ; anum > 0; anum--, tmps++) | |
2661 | *tmps = ~*tmps; | |
5d01050a FC |
2662 | } |
2663 | ||
2664 | PP(pp_complement) | |
2665 | { | |
2666 | dSP; dTARGET; | |
2667 | tryAMAGICun_MG(compl_amg, AMGf_numeric); | |
2668 | { | |
2669 | dTOPss; | |
2670 | if (SvNIOKp(sv)) { | |
2671 | if (PL_op->op_private & HINT_INTEGER) { | |
2672 | const IV i = ~SvIV_nomg(sv); | |
2673 | SETi(i); | |
2674 | } | |
2675 | else { | |
2676 | const UV u = ~SvUV_nomg(sv); | |
2677 | SETu(u); | |
2678 | } | |
2679 | } | |
2680 | else { | |
2681 | S_scomplement(aTHX_ TARG, sv); | |
ec93b65f | 2682 | SETTARG; |
a0d0e21e | 2683 | } |
24840750 | 2684 | return NORMAL; |
5d01050a FC |
2685 | } |
2686 | } | |
2687 | ||
2688 | PP(pp_ncomplement) | |
2689 | { | |
2690 | dSP; | |
636ac8fc | 2691 | tryAMAGICun_MG(compl_amg, AMGf_numeric|AMGf_numarg); |
5d01050a FC |
2692 | { |
2693 | dTARGET; dTOPss; | |
2694 | if (PL_op->op_private & HINT_INTEGER) { | |
2695 | const IV i = ~SvIV_nomg(sv); | |
2696 | SETi(i); | |
2697 | } | |
2698 | else { | |
2699 | const UV u = ~SvUV_nomg(sv); | |
2700 | SETu(u); | |
2701 | } | |
2702 | } | |
2703 | return NORMAL; | |
2704 | } | |
2705 | ||
2706 | PP(pp_scomplement) | |
2707 | { | |
2708 | dSP; | |
2709 | tryAMAGICun_MG(scompl_amg, AMGf_numeric); | |
2710 | { | |
2711 | dTARGET; dTOPss; | |
2712 | S_scomplement(aTHX_ TARG, sv); | |
2713 | SETTARG; | |
2714 | return NORMAL; | |
a0d0e21e | 2715 | } |
79072805 LW |
2716 | } |
2717 | ||
a0d0e21e LW |
2718 | /* integer versions of some of the above */ |
2719 | ||
a0d0e21e | 2720 | PP(pp_i_multiply) |
79072805 | 2721 | { |
20b7effb | 2722 | dSP; dATARGET; |
6f1401dc | 2723 | tryAMAGICbin_MG(mult_amg, AMGf_assign); |
a0d0e21e | 2724 | { |
6f1401dc | 2725 | dPOPTOPiirl_nomg; |
a0d0e21e LW |
2726 | SETi( left * right ); |
2727 | RETURN; | |
2728 | } | |
79072805 LW |
2729 | } |
2730 | ||
a0d0e21e | 2731 | PP(pp_i_divide) |
79072805 | 2732 | { |
85935d8e | 2733 | IV num; |
20b7effb | 2734 | dSP; dATARGET; |
6f1401dc | 2735 | tryAMAGICbin_MG(div_amg, AMGf_assign); |
a0d0e21e | 2736 | { |
6f1401dc | 2737 | dPOPTOPssrl; |
85935d8e | 2738 | IV value = SvIV_nomg(right); |
a0d0e21e | 2739 | if (value == 0) |
ece1bcef | 2740 | DIE(aTHX_ "Illegal division by zero"); |
85935d8e | 2741 | num = SvIV_nomg(left); |
a0cec769 YST |
2742 | |
2743 | /* avoid FPE_INTOVF on some platforms when num is IV_MIN */ | |
2744 | if (value == -1) | |
2745 | value = - num; | |
2746 | else | |
2747 | value = num / value; | |
6f1401dc | 2748 | SETi(value); |
a0d0e21e LW |
2749 | RETURN; |
2750 | } | |
79072805 LW |
2751 | } |
2752 | ||
befad5d1 | 2753 | PP(pp_i_modulo) |
224ec323 JH |
2754 | { |
2755 | /* This is the vanilla old i_modulo. */ | |
20b7effb | 2756 | dSP; dATARGET; |
6f1401dc | 2757 | tryAMAGICbin_MG(modulo_amg, AMGf_assign); |
224ec323 | 2758 | { |
6f1401dc | 2759 | dPOPTOPiirl_nomg; |
224ec323 JH |
2760 | if (!right) |
2761 | DIE(aTHX_ "Illegal modulus zero"); | |
a0cec769 YST |
2762 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
2763 | if (right == -1) | |
2764 | SETi( 0 ); | |
2765 | else | |
2766 | SETi( left % right ); | |
224ec323 JH |
2767 | RETURN; |
2768 | } | |
2769 | } | |
2770 | ||
0927ade0 | 2771 | #if defined(__GLIBC__) && IVSIZE == 8 \ |
bf3d06aa | 2772 | && ( __GLIBC__ < 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ < 8)) |
befad5d1 | 2773 | |
0927ade0 | 2774 | PP(pp_i_modulo_glibc_bugfix) |
224ec323 | 2775 | { |
224ec323 | 2776 | /* This is the i_modulo with the workaround for the _moddi3 bug |
fce2b89e | 2777 | * in (at least) glibc 2.2.5 (the PERL_ABS() the workaround). |
224ec323 | 2778 | * See below for pp_i_modulo. */ |
20b7effb | 2779 | dSP; dATARGET; |
6f1401dc | 2780 | tryAMAGICbin_MG(modulo_amg, AMGf_assign); |
224ec323 | 2781 | { |
6f1401dc | 2782 | dPOPTOPiirl_nomg; |
224ec323 JH |
2783 | if (!right) |
2784 | DIE(aTHX_ "Illegal modulus zero"); | |
a0cec769 YST |
2785 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
2786 | if (right == -1) | |
2787 | SETi( 0 ); | |
2788 | else | |
2789 | SETi( left % PERL_ABS(right) ); | |
224ec323 JH |
2790 | RETURN; |
2791 | } | |
224ec323 | 2792 | } |
befad5d1 | 2793 | #endif |
79072805 | 2794 | |
a0d0e21e | 2795 | PP(pp_i_add) |
79072805 | 2796 | { |
20b7effb | 2797 | dSP; dATARGET; |
6f1401dc | 2798 | tryAMAGICbin_MG(add_amg, AMGf_assign); |
a0d0e21e | 2799 | { |
6f1401dc | 2800 | dPOPTOPiirl_ul_nomg; |
a0d0e21e LW |
2801 | SETi( left + right ); |
2802 | RETURN; | |
79072805 | 2803 | } |
79072805 LW |
2804 | } |
2805 | ||
a0d0e21e | 2806 | PP(pp_i_subtract) |
79072805 | 2807 | { |
20b7effb | 2808 | dSP; dATARGET; |
6f1401dc | 2809 | tryAMAGICbin_MG(subtr_amg, AMGf_assign); |
a0d0e21e | 2810 | { |
6f1401dc | 2811 | dPOPTOPiirl_ul_nomg; |
a0d0e21e LW |
2812 | SETi( left - right ); |
2813 | RETURN; | |
79072805 | 2814 | } |
79072805 LW |
2815 | } |
2816 | ||
a0d0e21e | 2817 | PP(pp_i_lt) |
79072805 | 2818 | { |
20b7effb | 2819 | dSP; |
6f1401dc | 2820 | tryAMAGICbin_MG(lt_amg, AMGf_set); |
a0d0e21e | 2821 | { |
96b6b87f | 2822 | dPOPTOPiirl_nomg; |
54310121 | 2823 | SETs(boolSV(left < right)); |
a0d0e21e LW |
2824 | RETURN; |
2825 | } | |
79072805 LW |
2826 | } |
2827 | ||
a0d0e21e | 2828 | PP(pp_i_gt) |
79072805 | 2829 | { |
20b7effb | 2830 | dSP; |
6f1401dc | 2831 | tryAMAGICbin_MG(gt_amg, AMGf_set); |
a0d0e21e | 2832 | { |
96b6b87f | 2833 | dPOPTOPiirl_nomg; |
54310121 | 2834 | SETs(boolSV(left > right)); |
a0d0e21e LW |
2835 | RETURN; |
2836 | } | |
79072805 LW |
2837 | } |
2838 | ||
a0d0e21e | 2839 | PP(pp_i_le) |
79072805 | 2840 | { |
20b7effb | 2841 | dSP; |
6f1401dc | 2842 | tryAMAGICbin_MG(le_amg, AMGf_set); |
a0d0e21e | 2843 | { |
96b6b87f | 2844 | dPOPTOPiirl_nomg; |
54310121 | 2845 | SETs(boolSV(left <= right)); |
a0d0e21e | 2846 | RETURN; |
85e6fe83 | 2847 | } |
79072805 LW |
2848 | } |
2849 | ||
a0d0e21e | 2850 | PP(pp_i_ge) |
79072805 | 2851 | { |
20b7effb | 2852 | dSP; |
6f1401dc | 2853 | tryAMAGICbin_MG(ge_amg, AMGf_set); |
a0d0e21e | 2854 | { |
96b6b87f | 2855 | dPOPTOPiirl_nomg; |
54310121 | 2856 | SETs(boolSV(left >= right)); |
a0d0e21e LW |
2857 | RETURN; |
2858 | } | |
79072805 LW |
2859 | } |
2860 | ||
a0d0e21e | 2861 | PP(pp_i_eq) |
79072805 | 2862 | { |
20b7effb | 2863 | dSP; |
6f1401dc | 2864 | tryAMAGICbin_MG(eq_amg, AMGf_set); |
a0d0e21e | 2865 | { |
96b6b87f | 2866 | dPOPTOPiirl_nomg; |
54310121 | 2867 | SETs(boolSV(left == right)); |
a0d0e21e LW |
2868 | RETURN; |
2869 | } | |
79072805 LW |
2870 | } |
2871 | ||
a0d0e21e | 2872 | PP(pp_i_ne) |
79072805 | 2873 | { |
20b7effb | 2874 | dSP; |
6f1401dc | 2875 | tryAMAGICbin_MG(ne_amg, AMGf_set); |
a0d0e21e | 2876 | { |
96b6b87f | 2877 | dPOPTOPiirl_nomg; |
54310121 | 2878 | SETs(boolSV(left != right)); |
a0d0e21e LW |
2879 | RETURN; |
2880 | } | |
79072805 LW |
2881 | } |
2882 | ||
a0d0e21e | 2883 | PP(pp_i_ncmp) |
79072805 | 2884 | { |
20b7effb | 2885 | dSP; dTARGET; |
6f1401dc | 2886 | tryAMAGICbin_MG(ncmp_amg, 0); |
a0d0e21e | 2887 | { |
96b6b87f | 2888 | dPOPTOPiirl_nomg; |
a0d0e21e | 2889 | I32 value; |
79072805 | 2890 | |
a0d0e21e | 2891 | if (left > right) |
79072805 | 2892 | value = 1; |
a0d0e21e | 2893 | else if (left < right) |
79072805 | 2894 | value = -1; |
a0d0e21e | 2895 | else |
79072805 | 2896 | value = 0; |
a0d0e21e LW |
2897 | SETi(value); |
2898 | RETURN; | |
79072805 | 2899 | } |
85e6fe83 LW |
2900 | } |
2901 | ||
2902 | PP(pp_i_negate) | |
2903 | { | |
20b7effb | 2904 | dSP; dTARGET; |
6f1401dc | 2905 | tryAMAGICun_MG(neg_amg, 0); |
1c2b3fd6 | 2906 | if (S_negate_string(aTHX)) return NORMAL; |
6f1401dc DM |
2907 | { |
2908 | SV * const sv = TOPs; | |
2909 | IV const i = SvIV_nomg(sv); | |
2910 | SETi(-i); | |
ae642386 | 2911 | return NORMAL; |
6f1401dc | 2912 | } |
85e6fe83 LW |
2913 | } |
2914 | ||
79072805 LW |
2915 | /* High falutin' math. */ |
2916 | ||
2917 | PP(pp_atan2) | |
2918 | { | |
20b7effb | 2919 | dSP; dTARGET; |
6f1401dc | 2920 | tryAMAGICbin_MG(atan2_amg, 0); |
a0d0e21e | 2921 | { |
096c060c | 2922 | dPOPTOPnnrl_nomg; |
a1021d57 | 2923 | SETn(Perl_atan2(left, right)); |
a0d0e21e LW |
2924 | RETURN; |
2925 | } | |
79072805 LW |
2926 | } |
2927 | ||
b1c05ba5 DM |
2928 | |
2929 | /* also used for: pp_cos() pp_exp() pp_log() pp_sqrt() */ | |
2930 | ||
79072805 LW |
2931 | PP(pp_sin) |
2932 | { | |
20b7effb | 2933 | dSP; dTARGET; |
af71714e | 2934 | int amg_type = fallback_amg; |
71302fe3 | 2935 | const char *neg_report = NULL; |
71302fe3 NC |
2936 | const int op_type = PL_op->op_type; |
2937 | ||
2938 | switch (op_type) { | |
af71714e JH |
2939 | case OP_SIN: amg_type = sin_amg; break; |
2940 | case OP_COS: amg_type = cos_amg; break; | |
2941 | case OP_EXP: amg_type = exp_amg; break; | |
2942 | case OP_LOG: amg_type = log_amg; neg_report = "log"; break; | |
2943 | case OP_SQRT: amg_type = sqrt_amg; neg_report = "sqrt"; break; | |
a0d0e21e | 2944 | } |
79072805 | 2945 | |
af71714e | 2946 | assert(amg_type != fallback_amg); |
6f1401dc DM |
2947 | |
2948 | tryAMAGICun_MG(amg_type, 0); | |
a0d0e21e | 2949 | { |
8c78ed36 | 2950 | SV * const arg = TOPs; |
6f1401dc | 2951 | const NV value = SvNV_nomg(arg); |
a5dc2484 | 2952 | #ifdef NV_NAN |
f256868e | 2953 | NV result = NV_NAN; |
a5dc2484 JH |
2954 | #else |
2955 | NV result = 0.0; | |
2956 | #endif | |
af71714e | 2957 | if (neg_report) { /* log or sqrt */ |
a3463d96 DD |
2958 | if ( |
2959 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) | |
2960 | ! Perl_isnan(value) && | |
2961 | #endif | |
2962 | (op_type == OP_LOG ? (value <= 0.0) : (value < 0.0))) { | |
71302fe3 | 2963 | SET_NUMERIC_STANDARD(); |
dcbac5bb | 2964 | /* diag_listed_as: Can't take log of %g */ |
147e3846 | 2965 | DIE(aTHX_ "Can't take %s of %" NVgf, neg_report, value); |
71302fe3 NC |
2966 | } |
2967 | } | |
af71714e | 2968 | switch (op_type) { |
f256868e | 2969 | default: |
af71714e JH |
2970 | case OP_SIN: result = Perl_sin(value); break; |
2971 | case OP_COS: result = Perl_cos(value); break; | |
2972 | case OP_EXP: result = Perl_exp(value); break; | |
2973 | case OP_LOG: result = Perl_log(value); break; | |
2974 | case OP_SQRT: result = Perl_sqrt(value); break; | |
2975 | } | |
8c78ed36 FC |
2976 | SETn(result); |
2977 | return NORMAL; | |
a0d0e21e | 2978 | } |
79072805 LW |
2979 | } |
2980 | ||
56cb0a1c AD |
2981 | /* Support Configure command-line overrides for rand() functions. |
2982 | After 5.005, perhaps we should replace this by Configure support | |
2983 | for drand48(), random(), or rand(). For 5.005, though, maintain | |
2984 | compatibility by calling rand() but allow the user to override it. | |
2985 | See INSTALL for details. --Andy Dougherty 15 July 1998 | |
2986 | */ | |
85ab1d1d JH |
2987 | /* Now it's after 5.005, and Configure supports drand48() and random(), |
2988 | in addition to rand(). So the overrides should not be needed any more. | |
2989 | --Jarkko Hietaniemi 27 September 1998 | |
2990 | */ | |
2991 | ||
79072805 LW |
2992 | PP(pp_rand) |
2993 | { | |
80252599 | 2994 | if (!PL_srand_called) { |
85ab1d1d | 2995 | (void)seedDrand01((Rand_seed_t)seed()); |
80252599 | 2996 | PL_srand_called = TRUE; |
93dc8474 | 2997 | } |
fdf4dddd DD |
2998 | { |
2999 | dSP; | |
3000 | NV value; | |
fdf4dddd DD |
3001 | |
3002 | if (MAXARG < 1) | |
7e9044f9 FC |
3003 | { |
3004 | EXTEND(SP, 1); | |
fdf4dddd | 3005 | value = 1.0; |
7e9044f9 | 3006 | } |
fdf4dddd DD |
3007 | else { |
3008 | SV * const sv = POPs; | |
3009 | if(!sv) | |
3010 | value = 1.0; | |
3011 | else | |
3012 | value = SvNV(sv); | |
3013 | } | |
3014 | /* 1 of 2 things can be carried through SvNV, SP or TARG, SP was carried */ | |
a3463d96 DD |
3015 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
3016 | if (! Perl_isnan(value) && value == 0.0) | |
3017 | #else | |
659c4b96 | 3018 | if (value == 0.0) |
a3463d96 | 3019 | #endif |
fdf4dddd DD |
3020 | value = 1.0; |
3021 | { | |
3022 | dTARGET; | |
3023 | PUSHs(TARG); | |
3024 | PUTBACK; | |
3025 | value *= Drand01(); | |
3026 | sv_setnv_mg(TARG, value); | |
3027 | } | |
3028 | } | |
3029 | return NORMAL; | |
79072805 LW |
3030 | } |
3031 | ||
3032 | PP(pp_srand) | |
3033 | { | |
20b7effb | 3034 | dSP; dTARGET; |
f914a682 JL |
3035 | UV anum; |
3036 | ||
0a5f3363 | 3037 | if (MAXARG >= 1 && (TOPs || POPs)) { |
f914a682 JL |
3038 | SV *top; |
3039 | char *pv; | |
3040 | STRLEN len; | |
3041 | int flags; | |
3042 | ||
3043 | top = POPs; | |
3044 | pv = SvPV(top, len); | |
3045 | flags = grok_number(pv, len, &anum); | |
3046 | ||
3047 | if (!(flags & IS_NUMBER_IN_UV)) { | |
3048 | Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW), | |
3049 | "Integer overflow in srand"); | |
3050 | anum = UV_MAX; | |
3051 | } | |
3052 | } | |
3053 | else { | |
3054 | anum = seed(); | |
3055 | } | |
3056 | ||
85ab1d1d | 3057 | (void)seedDrand01((Rand_seed_t)anum); |
80252599 | 3058 | PL_srand_called = TRUE; |
da1010ec NC |
3059 | if (anum) |
3060 | XPUSHu(anum); | |
3061 | else { | |
3062 | /* Historically srand always returned true. We can avoid breaking | |
3063 | that like this: */ | |
3064 | sv_setpvs(TARG, "0 but true"); | |
3065 | XPUSHTARG; | |
3066 | } | |
83832992 | 3067 | RETURN; |
79072805 LW |
3068 | } |
3069 | ||
79072805 LW |
3070 | PP(pp_int) |
3071 | { | |
20b7effb | 3072 | dSP; dTARGET; |
6f1401dc | 3073 | tryAMAGICun_MG(int_amg, AMGf_numeric); |
774d564b | 3074 | { |
6f1401dc DM |
3075 | SV * const sv = TOPs; |
3076 | const IV iv = SvIV_nomg(sv); | |
28e5dec8 JH |
3077 | /* XXX it's arguable that compiler casting to IV might be subtly |
3078 | different from modf (for numbers inside (IV_MIN,UV_MAX)) in which | |
3079 | else preferring IV has introduced a subtle behaviour change bug. OTOH | |
3080 | relying on floating point to be accurate is a bug. */ | |
3081 | ||
c781a409 | 3082 | if (!SvOK(sv)) { |
922c4365 | 3083 | SETu(0); |
c781a409 RD |
3084 | } |
3085 | else if (SvIOK(sv)) { | |
3086 | if (SvIsUV(sv)) | |
6f1401dc | 3087 | SETu(SvUV_nomg(sv)); |
c781a409 | 3088 | else |
28e5dec8 | 3089 | SETi(iv); |
c781a409 | 3090 | } |
c781a409 | 3091 | else { |
6f1401dc | 3092 | const NV value = SvNV_nomg(sv); |
b9d05018 FC |
3093 | if (UNLIKELY(Perl_isinfnan(value))) |
3094 | SETn(value); | |
5bf8b78e | 3095 | else if (value >= 0.0) { |
28e5dec8 JH |
3096 | if (value < (NV)UV_MAX + 0.5) { |
3097 | SETu(U_V(value)); | |
3098 | } else { | |
059a1014 | 3099 | SETn(Perl_floor(value)); |
28e5dec8 | 3100 | } |
1048ea30 | 3101 | } |
28e5dec8 JH |
3102 | else { |
3103 | if (value > (NV)IV_MIN - 0.5) { | |
3104 | SETi(I_V(value)); | |
3105 | } else { | |
1bbae031 | 3106 | SETn(Perl_ceil(value)); |
28e5dec8 JH |
3107 | } |
3108 | } | |
774d564b | 3109 | } |
79072805 | 3110 | } |
699e9491 | 3111 | return NORMAL; |
79072805 LW |
3112 | } |
3113 | ||
463ee0b2 LW |
3114 | PP(pp_abs) |
3115 | { | |
20b7effb | 3116 | dSP; dTARGET; |
6f1401dc | 3117 | tryAMAGICun_MG(abs_amg, AMGf_numeric); |
a0d0e21e | 3118 | { |
6f1401dc | 3119 | SV * const sv = TOPs; |
28e5dec8 | 3120 | /* This will cache the NV value if string isn't actually integer */ |
6f1401dc | 3121 | const IV iv = SvIV_nomg(sv); |
a227d84d | 3122 | |
800401ee | 3123 | if (!SvOK(sv)) { |
922c4365 | 3124 | SETu(0); |
800401ee JH |
3125 | } |
3126 | else if (SvIOK(sv)) { | |
28e5dec8 | 3127 | /* IVX is precise */ |
800401ee | 3128 | if (SvIsUV(sv)) { |
6f1401dc | 3129 | SETu(SvUV_nomg(sv)); /* force it to be numeric only */ |
28e5dec8 JH |
3130 | } else { |
3131 | if (iv >= 0) { | |
3132 | SETi(iv); | |
3133 | } else { | |
3134 | if (iv != IV_MIN) { | |
3135 | SETi(-iv); | |
3136 | } else { | |
3137 | /* 2s complement assumption. Also, not really needed as | |
3138 | IV_MIN and -IV_MIN should both be %100...00 and NV-able */ | |
b396d0d8 | 3139 | SETu((UV)IV_MIN); |
28e5dec8 | 3140 | } |
a227d84d | 3141 | } |
28e5dec8 JH |
3142 | } |
3143 | } else{ | |
6f1401dc | 3144 | const NV value = SvNV_nomg(sv); |
774d564b | 3145 | if (value < 0.0) |
1b6737cc | 3146 | SETn(-value); |
a4474c9e DD |
3147 | else |
3148 | SETn(value); | |
774d564b | 3149 | } |
a0d0e21e | 3150 | } |
067b7929 | 3151 | return NORMAL; |
463ee0b2 LW |
3152 | } |
3153 | ||
b1c05ba5 DM |
3154 | |
3155 | /* also used for: pp_hex() */ | |
3156 | ||
79072805 LW |
3157 | PP(pp_oct) |
3158 | { | |
20b7effb | 3159 | dSP; dTARGET; |
5c144d81 | 3160 | const char *tmps; |
53305cf1 | 3161 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES; |
6f894ead | 3162 | STRLEN len; |
53305cf1 NC |
3163 | NV result_nv; |
3164 | UV result_uv; | |
4e51bcca | 3165 | SV* const sv = TOPs; |
79072805 | 3166 | |
349d4f2f | 3167 | tmps = (SvPV_const(sv, len)); |
2bc69dc4 NIS |
3168 | if (DO_UTF8(sv)) { |
3169 | /* If Unicode, try to downgrade | |
3170 | * If not possible, croak. */ | |
1b6737cc | 3171 | SV* const tsv = sv_2mortal(newSVsv(sv)); |
2bc69dc4 NIS |
3172 | |
3173 | SvUTF8_on(tsv); | |
3174 | sv_utf8_downgrade(tsv, FALSE); | |
349d4f2f | 3175 | tmps = SvPV_const(tsv, len); |
2bc69dc4 | 3176 | } |
daa2adfd NC |
3177 | if (PL_op->op_type == OP_HEX) |
3178 | goto hex; | |
3179 | ||
6f894ead | 3180 | while (*tmps && len && isSPACE(*tmps)) |
53305cf1 | 3181 | tmps++, len--; |
9e24b6e2 | 3182 | if (*tmps == '0') |
53305cf1 | 3183 | tmps++, len--; |
305b8651 | 3184 | if (isALPHA_FOLD_EQ(*tmps, 'x')) { |
daa2adfd | 3185 | hex: |
53305cf1 | 3186 | result_uv = grok_hex (tmps, &len, &flags, &result_nv); |
daa2adfd | 3187 | } |
305b8651 | 3188 | else if (isALPHA_FOLD_EQ(*tmps, 'b')) |
53305cf1 | 3189 | result_uv = grok_bin (tmps, &len, &flags, &result_nv); |
464e2e8a | 3190 | else |
53305cf1 NC |
3191 | result_uv = grok_oct (tmps, &len, &flags, &result_nv); |
3192 | ||
3193 | if (flags & PERL_SCAN_GREATER_THAN_UV_MAX) { | |
4e51bcca | 3194 | SETn(result_nv); |
53305cf1 NC |
3195 | } |
3196 | else { | |
4e51bcca | 3197 | SETu(result_uv); |
53305cf1 | 3198 | } |
4e51bcca | 3199 | return NORMAL; |
79072805 LW |
3200 | } |
3201 | ||
3202 | /* String stuff. */ | |
3203 | ||
3204 | PP(pp_length) | |
3205 | { | |
20b7effb | 3206 | dSP; dTARGET; |
0bd48802 | 3207 | SV * const sv = TOPs; |
a0ed51b3 | 3208 | |
7776003e DD |
3209 | U32 in_bytes = IN_BYTES; |
3210 | /* simplest case shortcut */ | |
3211 | /* turn off SVf_UTF8 in tmp flags if HINT_BYTES on*/ | |
3212 | U32 svflags = (SvFLAGS(sv) ^ (in_bytes << 26)) & (SVf_POK|SVs_GMG|SVf_UTF8); | |
6d59e610 | 3213 | STATIC_ASSERT_STMT(HINT_BYTES == 0x00000008 && SVf_UTF8 == 0x20000000 && (SVf_UTF8 == HINT_BYTES << 26)); |
7776003e DD |
3214 | SETs(TARG); |
3215 | ||
3216 | if(LIKELY(svflags == SVf_POK)) | |
3217 | goto simple_pv; | |
3218 | if(svflags & SVs_GMG) | |
3219 | mg_get(sv); | |
0f43fd57 | 3220 | if (SvOK(sv)) { |
7776003e DD |
3221 | if (!IN_BYTES) /* reread to avoid using an C auto/register */ |
3222 | sv_setiv(TARG, (IV)sv_len_utf8_nomg(sv)); | |
9f621bb0 | 3223 | else |
0f43fd57 FC |
3224 | { |
3225 | STRLEN len; | |
7776003e DD |
3226 | /* unrolled SvPV_nomg_const(sv,len) */ |
3227 | if(SvPOK_nog(sv)){ | |
3228 | simple_pv: | |
3229 | len = SvCUR(sv); | |
3230 | } else { | |
3231 | (void)sv_2pv_flags(sv, &len, 0|SV_CONST_RETURN); | |
3232 | } | |
3233 | sv_setiv(TARG, (IV)(len)); | |
0f43fd57 | 3234 | } |
656266fc | 3235 | } else { |
9407f9c1 | 3236 | if (!SvPADTMP(TARG)) { |
e03e82a0 | 3237 | sv_set_undef(TARG); |
7776003e DD |
3238 | } else { /* TARG is on stack at this point and is overwriten by SETs. |
3239 | This branch is the odd one out, so put TARG by default on | |
3240 | stack earlier to let local SP go out of liveness sooner */ | |
3241 | SETs(&PL_sv_undef); | |
3242 | goto no_set_magic; | |
3243 | } | |
92331800 | 3244 | } |
7776003e DD |
3245 | SvSETMAGIC(TARG); |
3246 | no_set_magic: | |
3247 | return NORMAL; /* no putback, SP didn't move in this opcode */ | |
79072805 LW |
3248 | } |
3249 | ||
83f78d1a FC |
3250 | /* Returns false if substring is completely outside original string. |
3251 | No length is indicated by len_iv = 0 and len_is_uv = 0. len_is_uv must | |
3252 | always be true for an explicit 0. | |
3253 | */ | |
3254 | bool | |
ddeaf645 DD |
3255 | Perl_translate_substr_offsets( STRLEN curlen, IV pos1_iv, |
3256 | bool pos1_is_uv, IV len_iv, | |
3257 | bool len_is_uv, STRLEN *posp, | |
3258 | STRLEN *lenp) | |
83f78d1a FC |
3259 | { |
3260 | IV pos2_iv; | |
3261 | int pos2_is_uv; | |
3262 | ||
3263 | PERL_ARGS_ASSERT_TRANSLATE_SUBSTR_OFFSETS; | |
3264 | ||
3265 | if (!pos1_is_uv && pos1_iv < 0 && curlen) { | |
3266 | pos1_is_uv = curlen-1 > ~(UV)pos1_iv; | |
3267 | pos1_iv += curlen; | |
3268 | } | |
3269 | if ((pos1_is_uv || pos1_iv > 0) && (UV)pos1_iv > curlen) | |
3270 | return FALSE; | |
3271 | ||
3272 | if (len_iv || len_is_uv) { | |
3273 | if (!len_is_uv && len_iv < 0) { | |
3274 | pos2_iv = curlen + len_iv; | |
3275 | if (curlen) | |
3276 | pos2_is_uv = curlen-1 > ~(UV)len_iv; | |
3277 | else | |
3278 | pos2_is_uv = 0; | |
3279 | } else { /* len_iv >= 0 */ | |
3280 | if (!pos1_is_uv && pos1_iv < 0) { | |
3281 | pos2_iv = pos1_iv + len_iv; | |
3282 | pos2_is_uv = (UV)len_iv > (UV)IV_MAX; | |
3283 | } else { | |
3284 | if ((UV)len_iv > curlen-(UV)pos1_iv) | |
3285 | pos2_iv = curlen; | |
3286 | else | |
3287 | pos2_iv = pos1_iv+len_iv; | |
3288 | pos2_is_uv = 1; | |
3289 | } | |
3290 | } | |
3291 | } | |
3292 | else { | |
3293 | pos2_iv = curlen; | |
3294 | pos2_is_uv = 1; | |
3295 | } | |
3296 | ||
3297 | if (!pos2_is_uv && pos2_iv < 0) { | |
3298 | if (!pos1_is_uv && pos1_iv < 0) | |
3299 | return FALSE; | |
3300 | pos2_iv = 0; | |
3301 | } | |
3302 | else if (!pos1_is_uv && pos1_iv < 0) | |
3303 | pos1_iv = 0; | |
3304 | ||
3305 | if ((UV)pos2_iv < (UV)pos1_iv) | |
3306 | pos2_iv = pos1_iv; | |
3307 | if ((UV)pos2_iv > curlen) | |
3308 | pos2_iv = curlen; | |
3309 | ||
3310 | /* pos1_iv and pos2_iv both in 0..curlen, so the cast is safe */ | |
3311 | *posp = (STRLEN)( (UV)pos1_iv ); | |
3312 | *lenp = (STRLEN)( (UV)pos2_iv - (UV)pos1_iv ); | |
3313 | ||
3314 | return TRUE; | |
3315 | } | |
3316 | ||
79072805 LW |
3317 | PP(pp_substr) |
3318 | { | |
20b7effb | 3319 | dSP; dTARGET; |
79072805 | 3320 | SV *sv; |
463ee0b2 | 3321 | STRLEN curlen; |
9402d6ed | 3322 | STRLEN utf8_curlen; |
777f7c56 EB |
3323 | SV * pos_sv; |
3324 | IV pos1_iv; | |
3325 | int pos1_is_uv; | |
777f7c56 EB |
3326 | SV * len_sv; |
3327 | IV len_iv = 0; | |
83f78d1a | 3328 | int len_is_uv = 0; |
24fcb59f | 3329 | I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
bbddc9e0 | 3330 | const bool rvalue = (GIMME_V != G_VOID); |
e1ec3a88 | 3331 | const char *tmps; |
9402d6ed | 3332 | SV *repl_sv = NULL; |
cbbf8932 | 3333 | const char *repl = NULL; |
7b8d334a | 3334 | STRLEN repl_len; |
7bc95ae1 | 3335 | int num_args = PL_op->op_private & 7; |
13e30c65 | 3336 | bool repl_need_utf8_upgrade = FALSE; |
79072805 | 3337 | |
78f9721b SM |
3338 | if (num_args > 2) { |
3339 | if (num_args > 3) { | |
24fcb59f | 3340 | if(!(repl_sv = POPs)) num_args--; |
7bc95ae1 FC |
3341 | } |
3342 | if ((len_sv = POPs)) { | |
3343 | len_iv = SvIV(len_sv); | |
83f78d1a | 3344 | len_is_uv = len_iv ? SvIOK_UV(len_sv) : 1; |
7b8d334a | 3345 | } |
7bc95ae1 | 3346 | else num_args--; |
5d82c453 | 3347 | } |
777f7c56 EB |
3348 | pos_sv = POPs; |
3349 | pos1_iv = SvIV(pos_sv); | |
3350 | pos1_is_uv = SvIOK_UV(pos_sv); | |
79072805 | 3351 | sv = POPs; |
24fcb59f FC |
3352 | if (PL_op->op_private & OPpSUBSTR_REPL_FIRST) { |
3353 | assert(!repl_sv); | |
3354 | repl_sv = POPs; | |
3355 | } | |
6582db62 | 3356 | if (lvalue && !repl_sv) { |
83f78d1a FC |
3357 | SV * ret; |
3358 | ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ | |
3359 | sv_magic(ret, NULL, PERL_MAGIC_substr, NULL, 0); | |
3360 | LvTYPE(ret) = 'x'; | |
3361 | LvTARG(ret) = SvREFCNT_inc_simple(sv); | |
3362 | LvTARGOFF(ret) = | |
3363 | pos1_is_uv || pos1_iv >= 0 | |
3364 | ? (STRLEN)(UV)pos1_iv | |
b063b0a8 | 3365 | : (LvFLAGS(ret) |= LVf_NEG_OFF, (STRLEN)(UV)-pos1_iv); |
83f78d1a FC |
3366 | LvTARGLEN(ret) = |
3367 | len_is_uv || len_iv > 0 | |
3368 | ? (STRLEN)(UV)len_iv | |
b063b0a8 | 3369 | : (LvFLAGS(ret) |= LVf_NEG_LEN, (STRLEN)(UV)-len_iv); |
83f78d1a | 3370 | |
83f78d1a FC |
3371 | PUSHs(ret); /* avoid SvSETMAGIC here */ |
3372 | RETURN; | |
a74fb2cd | 3373 | } |
6582db62 FC |
3374 | if (repl_sv) { |
3375 | repl = SvPV_const(repl_sv, repl_len); | |
3376 | SvGETMAGIC(sv); | |
3377 | if (SvROK(sv)) | |
3378 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), | |
3379 | "Attempt to use reference as lvalue in substr" | |
3380 | ); | |
3381 | tmps = SvPV_force_nomg(sv, curlen); | |
3382 | if (DO_UTF8(repl_sv) && repl_len) { | |
3383 | if (!DO_UTF8(sv)) { | |
41b1e858 AC |
3384 | /* Upgrade the dest, and recalculate tmps in case the buffer |
3385 | * got reallocated; curlen may also have been changed */ | |
01680ee9 | 3386 | sv_utf8_upgrade_nomg(sv); |
41b1e858 | 3387 | tmps = SvPV_nomg(sv, curlen); |
6582db62 FC |
3388 | } |
3389 | } | |
3390 | else if (DO_UTF8(sv)) | |
3391 | repl_need_utf8_upgrade = TRUE; | |
3392 | } | |
3393 | else tmps = SvPV_const(sv, curlen); | |
7e2040f0 | 3394 | if (DO_UTF8(sv)) { |
0d788f38 | 3395 | utf8_curlen = sv_or_pv_len_utf8(sv, tmps, curlen); |
9402d6ed JH |
3396 | if (utf8_curlen == curlen) |
3397 | utf8_curlen = 0; | |
a0ed51b3 | 3398 | else |
9402d6ed | 3399 | curlen = utf8_curlen; |
a0ed51b3 | 3400 | } |
d1c2b58a | 3401 | else |
9402d6ed | 3402 | utf8_curlen = 0; |
a0ed51b3 | 3403 | |
83f78d1a FC |
3404 | { |
3405 | STRLEN pos, len, byte_len, byte_pos; | |
777f7c56 | 3406 | |
83f78d1a FC |
3407 | if (!translate_substr_offsets( |
3408 | curlen, pos1_iv, pos1_is_uv, len_iv, len_is_uv, &pos, &len | |
3409 | )) goto bound_fail; | |
777f7c56 | 3410 | |
83f78d1a FC |
3411 | byte_len = len; |
3412 | byte_pos = utf8_curlen | |
0d788f38 | 3413 | ? sv_or_pv_pos_u2b(sv, tmps, pos, &byte_len) : pos; |
d931b1be | 3414 | |
2154eca7 | 3415 | tmps += byte_pos; |
bbddc9e0 CS |
3416 | |
3417 | if (rvalue) { | |
3418 | SvTAINTED_off(TARG); /* decontaminate */ | |
3419 | SvUTF8_off(TARG); /* decontaminate */ | |
3420 | sv_setpvn(TARG, tmps, byte_len); | |
12aa1545 | 3421 | #ifdef USE_LOCALE_COLLATE |
bbddc9e0 | 3422 | sv_unmagic(TARG, PERL_MAGIC_collxfrm); |
12aa1545 | 3423 | #endif |
bbddc9e0 CS |
3424 | if (utf8_curlen) |
3425 | SvUTF8_on(TARG); | |
3426 | } | |
2154eca7 | 3427 | |
f7928d6c | 3428 | if (repl) { |
13e30c65 JH |
3429 | SV* repl_sv_copy = NULL; |
3430 | ||
3431 | if (repl_need_utf8_upgrade) { | |
3432 | repl_sv_copy = newSVsv(repl_sv); | |
3433 | sv_utf8_upgrade(repl_sv_copy); | |
349d4f2f | 3434 | repl = SvPV_const(repl_sv_copy, repl_len); |
13e30c65 | 3435 | } |
502d9230 | 3436 | if (!SvOK(sv)) |
500f3e18 | 3437 | SvPVCLEAR(sv); |
777f7c56 | 3438 | sv_insert_flags(sv, byte_pos, byte_len, repl, repl_len, 0); |
ef8d46e8 | 3439 | SvREFCNT_dec(repl_sv_copy); |
f7928d6c | 3440 | } |
79072805 | 3441 | } |
6a9665b0 FC |
3442 | if (PL_op->op_private & OPpSUBSTR_REPL_FIRST) |
3443 | SP++; | |
3444 | else if (rvalue) { | |
bbddc9e0 CS |
3445 | SvSETMAGIC(TARG); |
3446 | PUSHs(TARG); | |
3447 | } | |
79072805 | 3448 | RETURN; |
777f7c56 | 3449 | |
7b52d656 | 3450 | bound_fail: |
83f78d1a | 3451 | if (repl) |
777f7c56 EB |
3452 | Perl_croak(aTHX_ "substr outside of string"); |
3453 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), "substr outside of string"); | |
3454 | RETPUSHUNDEF; | |
79072805 LW |
3455 | } |
3456 | ||
3457 | PP(pp_vec) | |
3458 | { | |
20b7effb | 3459 | dSP; |
eb578fdb | 3460 | const IV size = POPi; |
d69c4304 | 3461 | SV* offsetsv = POPs; |
eb578fdb | 3462 | SV * const src = POPs; |
1b6737cc | 3463 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
2154eca7 | 3464 | SV * ret; |
1b92e694 DM |
3465 | UV retuv; |
3466 | STRLEN offset = 0; | |
3467 | char errflags = 0; | |
d69c4304 DM |
3468 | |
3469 | /* extract a STRLEN-ranged integer value from offsetsv into offset, | |
1b92e694 | 3470 | * or flag that its out of range */ |
d69c4304 DM |
3471 | { |
3472 | IV iv = SvIV(offsetsv); | |
3473 | ||
3474 | /* avoid a large UV being wrapped to a negative value */ | |
1b92e694 | 3475 | if (SvIOK_UV(offsetsv) && SvUVX(offsetsv) > (UV)IV_MAX) |
b063b0a8 | 3476 | errflags = LVf_OUT_OF_RANGE; |
1b92e694 | 3477 | else if (iv < 0) |
b063b0a8 | 3478 | errflags = (LVf_NEG_OFF|LVf_OUT_OF_RANGE); |
d69c4304 | 3479 | #if PTRSIZE < IVSIZE |
1b92e694 | 3480 | else if (iv > Size_t_MAX) |
b063b0a8 | 3481 | errflags = LVf_OUT_OF_RANGE; |
d69c4304 | 3482 | #endif |
1b92e694 DM |
3483 | else |
3484 | offset = (STRLEN)iv; | |
d69c4304 DM |
3485 | } |
3486 | ||
1b92e694 | 3487 | retuv = errflags ? 0 : do_vecget(src, offset, size); |
a0d0e21e | 3488 | |
81e118e0 | 3489 | if (lvalue) { /* it's an lvalue! */ |
2154eca7 EB |
3490 | ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
3491 | sv_magic(ret, NULL, PERL_MAGIC_vec, NULL, 0); | |
3492 | LvTYPE(ret) = 'v'; | |
3493 | LvTARG(ret) = SvREFCNT_inc_simple(src); | |
3494 | LvTARGOFF(ret) = offset; | |
3495 | LvTARGLEN(ret) = size; | |
1b92e694 | 3496 | LvFLAGS(ret) = errflags; |
2154eca7 EB |
3497 | } |
3498 | else { | |
3499 | dTARGET; | |
3500 | SvTAINTED_off(TARG); /* decontaminate */ | |
3501 | ret = TARG; | |
79072805 LW |
3502 | } |
3503 | ||
d69c4304 | 3504 | sv_setuv(ret, retuv); |
f9e95907 FC |
3505 | if (!lvalue) |
3506 | SvSETMAGIC(ret); | |
2154eca7 | 3507 | PUSHs(ret); |
79072805 LW |
3508 | RETURN; |
3509 | } | |
3510 | ||
b1c05ba5 DM |
3511 | |
3512 | /* also used for: pp_rindex() */ | |
3513 | ||
79072805 LW |
3514 | PP(pp_index) |
3515 | { | |
20b7effb | 3516 | dSP; dTARGET; |
79072805 LW |
3517 | SV *big; |
3518 | SV *little; | |
c445ea15 | 3519 | SV *temp = NULL; |
ad66a58c | 3520 | STRLEN biglen; |
2723d216 | 3521 | STRLEN llen = 0; |
b464e2b7 TC |
3522 | SSize_t offset = 0; |
3523 | SSize_t retval; | |
73ee8be2 NC |
3524 | const char *big_p; |
3525 | const char *little_p; | |
2f040f7f NC |
3526 | bool big_utf8; |
3527 | bool little_utf8; | |
2723d216 | 3528 | const bool is_index = PL_op->op_type == OP_INDEX; |
d3e26383 | 3529 | const bool threeargs = MAXARG >= 3 && (TOPs || ((void)POPs,0)); |
79072805 | 3530 | |
e1dccc0d Z |
3531 | if (threeargs) |
3532 | offset = POPi; | |
79072805 LW |
3533 | little = POPs; |
3534 | big = POPs; | |
73ee8be2 NC |
3535 | big_p = SvPV_const(big, biglen); |
3536 | little_p = SvPV_const(little, llen); | |
3537 | ||
e609e586 NC |
3538 | big_utf8 = DO_UTF8(big); |
3539 | little_utf8 = DO_UTF8(little); | |
3540 | if (big_utf8 ^ little_utf8) { | |
3541 | /* One needs to be upgraded. */ | |
8df0e7a2 | 3542 | if (little_utf8) { |
2f040f7f NC |
3543 | /* Well, maybe instead we might be able to downgrade the small |
3544 | string? */ | |
1eced8f8 | 3545 | char * const pv = (char*)bytes_from_utf8((U8 *)little_p, &llen, |
2f040f7f NC |
3546 | &little_utf8); |
3547 | if (little_utf8) { | |
3548 | /* If the large string is ISO-8859-1, and it's not possible to | |
3549 | convert the small string to ISO-8859-1, then there is no | |
3550 | way that it could be found anywhere by index. */ | |
3551 | retval = -1; | |
3552 | goto fail; | |
3553 | } | |
e609e586 | 3554 | |
2f040f7f NC |
3555 | /* At this point, pv is a malloc()ed string. So donate it to temp |
3556 | to ensure it will get free()d */ | |
3557 | little = temp = newSV(0); | |
73ee8be2 NC |
3558 | sv_usepvn(temp, pv, llen); |
3559 | little_p = SvPVX(little); | |
e609e586 | 3560 | } else { |
20e67ba1 | 3561 | temp = newSVpvn(little_p, llen); |
2f040f7f | 3562 | |
8df0e7a2 | 3563 | sv_utf8_upgrade(temp); |
20e67ba1 FC |
3564 | little = temp; |
3565 | little_p = SvPV_const(little, llen); | |
e609e586 NC |
3566 | } |
3567 | } | |
73ee8be2 NC |
3568 | if (SvGAMAGIC(big)) { |
3569 | /* Life just becomes a lot easier if I use a temporary here. | |
3570 | Otherwise I need to avoid calls to sv_pos_u2b(), which (dangerously) | |
3571 | will trigger magic and overloading again, as will fbm_instr() | |
3572 | */ | |
59cd0e26 NC |
3573 | big = newSVpvn_flags(big_p, biglen, |
3574 | SVs_TEMP | (big_utf8 ? SVf_UTF8 : 0)); | |
73ee8be2 NC |
3575 | big_p = SvPVX(big); |
3576 | } | |
e4e44778 | 3577 | if (SvGAMAGIC(little) || (is_index && !SvOK(little))) { |
73ee8be2 NC |
3578 | /* index && SvOK() is a hack. fbm_instr() calls SvPV_const, which will |
3579 | warn on undef, and we've already triggered a warning with the | |
3580 | SvPV_const some lines above. We can't remove that, as we need to | |
3581 | call some SvPV to trigger overloading early and find out if the | |
3582 | string is UTF-8. | |
8bd97c0c | 3583 | This is all getting too messy. The API isn't quite clean enough, |
73ee8be2 NC |
3584 | because data access has side effects. |
3585 | */ | |
59cd0e26 NC |
3586 | little = newSVpvn_flags(little_p, llen, |
3587 | SVs_TEMP | (little_utf8 ? SVf_UTF8 : 0)); | |
73ee8be2 NC |
3588 | little_p = SvPVX(little); |
3589 | } | |
e609e586 | 3590 | |
d3e26383 | 3591 | if (!threeargs) |
2723d216 | 3592 | offset = is_index ? 0 : biglen; |
a0ed51b3 | 3593 | else { |
ad66a58c | 3594 | if (big_utf8 && offset > 0) |
b464e2b7 | 3595 | offset = sv_pos_u2b_flags(big, offset, 0, SV_CONST_RETURN); |
73ee8be2 NC |
3596 | if (!is_index) |
3597 | offset += llen; | |
a0ed51b3 | 3598 | } |
79072805 LW |
3599 | if (offset < 0) |
3600 | offset = 0; | |
b464e2b7 | 3601 | else if (offset > (SSize_t)biglen) |
ad66a58c | 3602 | offset = biglen; |
73ee8be2 NC |
3603 | if (!(little_p = is_index |
3604 | ? fbm_instr((unsigned char*)big_p + offset, | |
3605 | (unsigned char*)big_p + biglen, little, 0) | |
3606 | : rninstr(big_p, big_p + offset, | |
3607 | little_p, little_p + llen))) | |
a0ed51b3 | 3608 | retval = -1; |
ad66a58c | 3609 | else { |
73ee8be2 | 3610 | retval = little_p - big_p; |
15c41403 | 3611 | if (retval > 1 && big_utf8) |
b464e2b7 | 3612 | retval = sv_pos_b2u_flags(big, retval, SV_CONST_RETURN); |
ad66a58c | 3613 | } |
ef8d46e8 | 3614 | SvREFCNT_dec(temp); |
2723d216 | 3615 | fail: |
e1dccc0d | 3616 | PUSHi(retval); |
79072805 LW |
3617 | RETURN; |
3618 | } | |
3619 | ||
3620 | PP(pp_sprintf) | |
3621 | { | |
20b7effb | 3622 | dSP; dMARK; dORIGMARK; dTARGET; |
3e6bd4bf | 3623 | SvTAINTED_off(TARG); |
79072805 | 3624 | do_sprintf(TARG, SP-MARK, MARK+1); |
bbce6d69 | 3625 | TAINT_IF(SvTAINTED(TARG)); |
79072805 LW |
3626 | SP = ORIGMARK; |
3627 | PUSHTARG; | |
3628 | RETURN; | |
3629 | } | |
3630 | ||
79072805 LW |
3631 | PP(pp_ord) |
3632 | { | |
20b7effb | 3633 | dSP; dTARGET; |
1eced8f8 | 3634 | |
6ba92227 | 3635 | SV *argsv = TOPs; |
ba210ebe | 3636 | STRLEN len; |
349d4f2f | 3637 | const U8 *s = (U8*)SvPV_const(argsv, len); |
121910a4 | 3638 | |
6ba92227 | 3639 | SETu(DO_UTF8(argsv) |
aee9b917 | 3640 | ? (len ? utf8n_to_uvchr(s, len, 0, UTF8_ALLOW_ANYUV) : 0) |
f3943cf2 | 3641 | : (UV)(*s)); |
68795e93 | 3642 | |
6ba92227 | 3643 | return NORMAL; |
79072805 LW |
3644 | } |
3645 | ||
463ee0b2 LW |
3646 | PP(pp_chr) |
3647 | { | |
20b7effb | 3648 | dSP; dTARGET; |
463ee0b2 | 3649 | char *tmps; |
8a064bd6 | 3650 | UV value; |
d3261b99 | 3651 | SV *top = TOPs; |
8a064bd6 | 3652 | |
71739502 | 3653 | SvGETMAGIC(top); |
9911fc4e FC |
3654 | if (UNLIKELY(SvAMAGIC(top))) |
3655 | top = sv_2num(top); | |
99f450cc | 3656 | if (UNLIKELY(isinfnansv(top))) |
147e3846 | 3657 | Perl_croak(aTHX_ "Cannot chr %" NVgf, SvNV(top)); |
1cd88304 JH |
3658 | else { |
3659 | if (!IN_BYTES /* under bytes, chr(-1) eq chr(0xff), etc. */ | |
3660 | && ((SvIOKp(top) && !SvIsUV(top) && SvIV_nomg(top) < 0) | |
3661 | || | |
3662 | ((SvNOKp(top) || (SvOK(top) && !SvIsUV(top))) | |
2cc2a5a0 KW |
3663 | && SvNV_nomg(top) < 0.0))) |
3664 | { | |
b3fe8680 FC |
3665 | if (ckWARN(WARN_UTF8)) { |
3666 | if (SvGMAGICAL(top)) { | |
3667 | SV *top2 = sv_newmortal(); | |
3668 | sv_setsv_nomg(top2, top); | |
3669 | top = top2; | |
3670 | } | |
1cd88304 | 3671 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
147e3846 | 3672 | "Invalid negative number (%" SVf ") in chr", SVfARG(top)); |
1cd88304 JH |
3673 | } |
3674 | value = UNICODE_REPLACEMENT; | |
3675 | } else { | |
3676 | value = SvUV_nomg(top); | |
3677 | } | |
8a064bd6 | 3678 | } |
463ee0b2 | 3679 | |
862a34c6 | 3680 | SvUPGRADE(TARG,SVt_PV); |
a0ed51b3 | 3681 | |
0064a8a9 | 3682 | if (value > 255 && !IN_BYTES) { |
5f560d8a | 3683 | SvGROW(TARG, (STRLEN)UVCHR_SKIP(value)+1); |
62961d2e | 3684 | tmps = (char*)uvchr_to_utf8_flags((U8*)SvPVX(TARG), value, 0); |
349d4f2f | 3685 | SvCUR_set(TARG, tmps - SvPVX_const(TARG)); |
a0ed51b3 LW |
3686 | *tmps = '\0'; |
3687 | (void)SvPOK_only(TARG); | |
aa6ffa16 | 3688 | SvUTF8_on(TARG); |
d3261b99 FC |
3689 | SETTARG; |
3690 | return NORMAL; | |
a0ed51b3 LW |
3691 | } |
3692 | ||
748a9306 | 3693 | SvGROW(TARG,2); |
463ee0b2 LW |
3694 | SvCUR_set(TARG, 1); |
3695 | tmps = SvPVX(TARG); | |
eb160463 | 3696 | *tmps++ = (char)value; |
748a9306 | 3697 | *tmps = '\0'; |
a0d0e21e | 3698 | (void)SvPOK_only(TARG); |
4c5ed6e2 | 3699 | |
d3261b99 FC |
3700 | SETTARG; |
3701 | return NORMAL; | |
463ee0b2 LW |
3702 | } |
3703 | ||
79072805 LW |
3704 | PP(pp_crypt) |
3705 | { | |
79072805 | 3706 | #ifdef HAS_CRYPT |
20b7effb | 3707 | dSP; dTARGET; |
5f74f29c | 3708 | dPOPTOPssrl; |
85c16d83 | 3709 | STRLEN len; |
10516c54 | 3710 | const char *tmps = SvPV_const(left, len); |
2bc69dc4 | 3711 | |
85c16d83 | 3712 | if (DO_UTF8(left)) { |
2bc69dc4 | 3713 | /* If Unicode, try to downgrade. |
f2791508 JH |
3714 | * If not possible, croak. |
3715 | * Yes, we made this up. */ | |
659fbb76 | 3716 | SV* const tsv = newSVpvn_flags(tmps, len, SVf_UTF8|SVs_TEMP); |
2bc69dc4 | 3717 | |
2bc69dc4 | 3718 | sv_utf8_downgrade(tsv, FALSE); |
349d4f2f | 3719 | tmps = SvPV_const(tsv, len); |
85c16d83 | 3720 | } |
05404ffe JH |
3721 | # ifdef USE_ITHREADS |
3722 | # ifdef HAS_CRYPT_R | |
3723 | if (!PL_reentrant_buffer->_crypt_struct_buffer) { | |
3724 | /* This should be threadsafe because in ithreads there is only | |
3725 | * one thread per interpreter. If this would not be true, | |
3726 | * we would need a mutex to protect this malloc. */ | |
3727 | PL_reentrant_buffer->_crypt_struct_buffer = | |
3728 | (struct crypt_data *)safemalloc(sizeof(struct crypt_data)); | |
3729 | #if defined(__GLIBC__) || defined(__EMX__) | |
3730 | if (PL_reentrant_buffer->_crypt_struct_buffer) { | |
3731 | PL_reentrant_buffer->_crypt_struct_buffer->initialized = 0; | |
3732 | /* work around glibc-2.2.5 bug */ | |
3733 | PL_reentrant_buffer->_crypt_struct_buffer->current_saltbits = 0; | |
3734 | } | |
05404ffe | 3735 | #endif |
6ab58e4d | 3736 | } |
05404ffe JH |
3737 | # endif /* HAS_CRYPT_R */ |
3738 | # endif /* USE_ITHREADS */ | |
5f74f29c | 3739 | # ifdef FCRYPT |
83003860 | 3740 | sv_setpv(TARG, fcrypt(tmps, SvPV_nolen_const(right))); |
5f74f29c | 3741 | # else |
83003860 | 3742 | sv_setpv(TARG, PerlProc_crypt(tmps, SvPV_nolen_const(right))); |
5f74f29c | 3743 | # endif |
fbc76eb3 | 3744 | SvUTF8_off(TARG); |
ec93b65f | 3745 | SETTARG; |
4808266b | 3746 | RETURN; |
79072805 | 3747 | #else |
b13b2135 | 3748 | DIE(aTHX_ |
79072805 LW |
3749 | "The crypt() function is unimplemented due to excessive paranoia."); |
3750 | #endif | |
79072805 LW |
3751 | } |
3752 | ||
00f254e2 KW |
3753 | /* Generally UTF-8 and UTF-EBCDIC are indistinguishable at this level. So |
3754 | * most comments below say UTF-8, when in fact they mean UTF-EBCDIC as well */ | |
3755 | ||
b1c05ba5 DM |
3756 | |
3757 | /* also used for: pp_lcfirst() */ | |
3758 | ||
79072805 LW |
3759 | PP(pp_ucfirst) |
3760 | { | |
00f254e2 KW |
3761 | /* Actually is both lcfirst() and ucfirst(). Only the first character |
3762 | * changes. This means that possibly we can change in-place, ie., just | |
3763 | * take the source and change that one character and store it back, but not | |
3764 | * if read-only etc, or if the length changes */ | |
3765 | ||
39644a26 | 3766 | dSP; |
d54190f6 | 3767 | SV *source = TOPs; |
00f254e2 | 3768 | STRLEN slen; /* slen is the byte length of the whole SV. */ |
d54190f6 NC |
3769 | STRLEN need; |
3770 | SV *dest; | |
00f254e2 KW |
3771 | bool inplace; /* ? Convert first char only, in-place */ |
3772 | bool doing_utf8 = FALSE; /* ? using utf8 */ | |
3773 | bool convert_source_to_utf8 = FALSE; /* ? need to convert */ | |
12e9c124 | 3774 | const int op_type = PL_op->op_type; |
d54190f6 NC |
3775 | const U8 *s; |
3776 | U8 *d; | |
3777 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
00f254e2 KW |
3778 | STRLEN ulen; /* ulen is the byte length of the original Unicode character |
3779 | * stored as UTF-8 at s. */ | |
3780 | STRLEN tculen; /* tculen is the byte length of the freshly titlecased (or | |
3781 | * lowercased) character stored in tmpbuf. May be either | |
3782 | * UTF-8 or not, but in either case is the number of bytes */ | |
d54190f6 | 3783 | |
841a5e18 | 3784 | s = (const U8*)SvPV_const(source, slen); |
a0ed51b3 | 3785 | |
00f254e2 KW |
3786 | /* We may be able to get away with changing only the first character, in |
3787 | * place, but not if read-only, etc. Later we may discover more reasons to | |
3788 | * not convert in-place. */ | |
1921e031 | 3789 | inplace = !SvREADONLY(source) && SvPADTMP(source); |
00f254e2 KW |
3790 | |
3791 | /* First calculate what the changed first character should be. This affects | |
3792 | * whether we can just swap it out, leaving the rest of the string unchanged, | |
3793 | * or even if have to convert the dest to UTF-8 when the source isn't */ | |
3794 | ||
3795 | if (! slen) { /* If empty */ | |
3796 | need = 1; /* still need a trailing NUL */ | |
b7576bcb | 3797 | ulen = 0; |
00f254e2 KW |
3798 | } |
3799 | else if (DO_UTF8(source)) { /* Is the source utf8? */ | |
d54190f6 | 3800 | doing_utf8 = TRUE; |
17e95c9d | 3801 | ulen = UTF8SKIP(s); |
094a2f8c | 3802 | if (op_type == OP_UCFIRST) { |
130c5df3 | 3803 | #ifdef USE_LOCALE_CTYPE |
a1a5ec35 | 3804 | _toTITLE_utf8_flags(s, s +slen, tmpbuf, &tculen, IN_LC_RUNTIME(LC_CTYPE)); |
130c5df3 | 3805 | #else |
a1a5ec35 | 3806 | _toTITLE_utf8_flags(s, s +slen, tmpbuf, &tculen, 0); |
130c5df3 | 3807 | #endif |
094a2f8c KW |
3808 | } |
3809 | else { | |
130c5df3 | 3810 | #ifdef USE_LOCALE_CTYPE |
a1a5ec35 | 3811 | _toLOWER_utf8_flags(s, s + slen, tmpbuf, &tculen, IN_LC_RUNTIME(LC_CTYPE)); |
130c5df3 | 3812 | #else |
a1a5ec35 | 3813 | _toLOWER_utf8_flags(s, s + slen, tmpbuf, &tculen, 0); |
130c5df3 | 3814 | #endif |
094a2f8c | 3815 | } |
00f254e2 | 3816 | |
17e95c9d KW |
3817 | /* we can't do in-place if the length changes. */ |
3818 | if (ulen != tculen) inplace = FALSE; | |
3819 | need = slen + 1 - ulen + tculen; | |
d54190f6 | 3820 | } |
00f254e2 KW |
3821 | else { /* Non-zero length, non-UTF-8, Need to consider locale and if |
3822 | * latin1 is treated as caseless. Note that a locale takes | |
3823 | * precedence */ | |
167d19f2 | 3824 | ulen = 1; /* Original character is 1 byte */ |
00f254e2 KW |
3825 | tculen = 1; /* Most characters will require one byte, but this will |
3826 | * need to be overridden for the tricky ones */ | |
3827 | need = slen + 1; | |
3828 | ||
3829 | if (op_type == OP_LCFIRST) { | |
d54190f6 | 3830 | |
00f254e2 | 3831 | /* lower case the first letter: no trickiness for any character */ |
130c5df3 | 3832 | #ifdef USE_LOCALE_CTYPE |
780fcc9f KW |
3833 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
3834 | _CHECK_AND_WARN_PROBLEMATIC_LOCALE; | |
3835 | *tmpbuf = toLOWER_LC(*s); | |
3836 | } | |
3837 | else | |
130c5df3 | 3838 | #endif |
780fcc9f KW |
3839 | { |
3840 | *tmpbuf = (IN_UNI_8_BIT) | |
3841 | ? toLOWER_LATIN1(*s) | |
3842 | : toLOWER(*s); | |
3843 | } | |
00f254e2 | 3844 | } |
130c5df3 | 3845 | #ifdef USE_LOCALE_CTYPE |
780fcc9f | 3846 | /* is ucfirst() */ |
d6ded950 | 3847 | else if (IN_LC_RUNTIME(LC_CTYPE)) { |
31f05a37 KW |
3848 | if (IN_UTF8_CTYPE_LOCALE) { |
3849 | goto do_uni_rules; | |
3850 | } | |
3851 | ||
780fcc9f | 3852 | _CHECK_AND_WARN_PROBLEMATIC_LOCALE; |
31f05a37 KW |
3853 | *tmpbuf = (U8) toUPPER_LC(*s); /* This would be a bug if any |
3854 | locales have upper and title case | |
3855 | different */ | |
00f254e2 | 3856 | } |
130c5df3 | 3857 | #endif |
00f254e2 KW |
3858 | else if (! IN_UNI_8_BIT) { |
3859 | *tmpbuf = toUPPER(*s); /* Returns caseless for non-ascii, or | |
3860 | * on EBCDIC machines whatever the | |
3861 | * native function does */ | |
3862 | } | |
31f05a37 KW |
3863 | else { |
3864 | /* Here, is ucfirst non-UTF-8, not in locale (unless that locale is | |
3865 | * UTF-8, which we treat as not in locale), and cased latin1 */ | |
3866 | UV title_ord; | |
91191cf7 | 3867 | #ifdef USE_LOCALE_CTYPE |
31f05a37 | 3868 | do_uni_rules: |
91191cf7 | 3869 | #endif |
31f05a37 KW |
3870 | |
3871 | title_ord = _to_upper_title_latin1(*s, tmpbuf, &tculen, 's'); | |
167d19f2 KW |
3872 | if (tculen > 1) { |
3873 | assert(tculen == 2); | |
3874 | ||
3875 | /* If the result is an upper Latin1-range character, it can | |
3876 | * still be represented in one byte, which is its ordinal */ | |
3877 | if (UTF8_IS_DOWNGRADEABLE_START(*tmpbuf)) { | |
3878 | *tmpbuf = (U8) title_ord; | |
3879 | tculen = 1; | |
00f254e2 KW |
3880 | } |
3881 | else { | |
167d19f2 KW |
3882 | /* Otherwise it became more than one ASCII character (in |
3883 | * the case of LATIN_SMALL_LETTER_SHARP_S) or changed to | |
3884 | * beyond Latin1, so the number of bytes changed, so can't | |
3885 | * replace just the first character in place. */ | |
3886 | inplace = FALSE; | |
3887 | ||
d14578b8 KW |
3888 | /* If the result won't fit in a byte, the entire result |
3889 | * will have to be in UTF-8. Assume worst case sizing in | |
3890 | * conversion. (all latin1 characters occupy at most two | |
3891 | * bytes in utf8) */ | |
167d19f2 KW |
3892 | if (title_ord > 255) { |
3893 | doing_utf8 = TRUE; | |
3894 | convert_source_to_utf8 = TRUE; | |
3895 | need = slen * 2 + 1; | |
3896 | ||
3897 | /* The (converted) UTF-8 and UTF-EBCDIC lengths of all | |
3898 | * (both) characters whose title case is above 255 is | |
3899 | * 2. */ | |
3900 | ulen = 2; | |
3901 | } | |
3902 | else { /* LATIN_SMALL_LETTER_SHARP_S expands by 1 byte */ | |
3903 | need = slen + 1 + 1; | |
3904 | } | |
00f254e2 | 3905 | } |
167d19f2 | 3906 | } |
00f254e2 KW |
3907 | } /* End of use Unicode (Latin1) semantics */ |
3908 | } /* End of changing the case of the first character */ | |
3909 | ||
3910 | /* Here, have the first character's changed case stored in tmpbuf. Ready to | |
3911 | * generate the result */ | |
3912 | if (inplace) { | |
3913 | ||
3914 | /* We can convert in place. This means we change just the first | |
3915 | * character without disturbing the rest; no need to grow */ | |
d54190f6 NC |
3916 | dest = source; |
3917 | s = d = (U8*)SvPV_force_nomg(source, slen); | |
3918 | } else { | |
3919 | dTARGET; | |
3920 | ||
3921 | dest = TARG; | |
3922 | ||
00f254e2 KW |
3923 | /* Here, we can't convert in place; we earlier calculated how much |
3924 | * space we will need, so grow to accommodate that */ | |
d54190f6 | 3925 | SvUPGRADE(dest, SVt_PV); |
3b416f41 | 3926 | d = (U8*)SvGROW(dest, need); |
d54190f6 NC |
3927 | (void)SvPOK_only(dest); |
3928 | ||
3929 | SETs(dest); | |
d54190f6 | 3930 | } |
44bc797b | 3931 | |
d54190f6 | 3932 | if (doing_utf8) { |
00f254e2 KW |
3933 | if (! inplace) { |
3934 | if (! convert_source_to_utf8) { | |
3935 | ||
3936 | /* Here both source and dest are in UTF-8, but have to create | |
3937 | * the entire output. We initialize the result to be the | |
3938 | * title/lower cased first character, and then append the rest | |
3939 | * of the string. */ | |
3940 | sv_setpvn(dest, (char*)tmpbuf, tculen); | |
3941 | if (slen > ulen) { | |
3942 | sv_catpvn(dest, (char*)(s + ulen), slen - ulen); | |
3943 | } | |
3944 | } | |
3945 | else { | |
3946 | const U8 *const send = s + slen; | |
3947 | ||
3948 | /* Here the dest needs to be in UTF-8, but the source isn't, | |
3949 | * except we earlier UTF-8'd the first character of the source | |
3950 | * into tmpbuf. First put that into dest, and then append the | |
3951 | * rest of the source, converting it to UTF-8 as we go. */ | |
3952 | ||
3953 | /* Assert tculen is 2 here because the only two characters that | |
3954 | * get to this part of the code have 2-byte UTF-8 equivalents */ | |
3955 | *d++ = *tmpbuf; | |
3956 | *d++ = *(tmpbuf + 1); | |
3957 | s++; /* We have just processed the 1st char */ | |
3958 | ||
3959 | for (; s < send; s++) { | |
3960 | d = uvchr_to_utf8(d, *s); | |
3961 | } | |
3962 | *d = '\0'; | |
3963 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
3964 | } | |
d54190f6 | 3965 | SvUTF8_on(dest); |
a0ed51b3 | 3966 | } |
00f254e2 | 3967 | else { /* in-place UTF-8. Just overwrite the first character */ |
d54190f6 NC |
3968 | Copy(tmpbuf, d, tculen, U8); |
3969 | SvCUR_set(dest, need - 1); | |
a0ed51b3 | 3970 | } |
094a2f8c | 3971 | |
a0ed51b3 | 3972 | } |
00f254e2 KW |
3973 | else { /* Neither source nor dest are in or need to be UTF-8 */ |
3974 | if (slen) { | |
00f254e2 KW |
3975 | if (inplace) { /* in-place, only need to change the 1st char */ |
3976 | *d = *tmpbuf; | |
3977 | } | |
3978 | else { /* Not in-place */ | |
3979 | ||
3980 | /* Copy the case-changed character(s) from tmpbuf */ | |
3981 | Copy(tmpbuf, d, tculen, U8); | |
3982 | d += tculen - 1; /* Code below expects d to point to final | |
3983 | * character stored */ | |
3984 | } | |
3985 | } | |
3986 | else { /* empty source */ | |
3987 | /* See bug #39028: Don't taint if empty */ | |
d54190f6 NC |
3988 | *d = *s; |
3989 | } | |
3990 | ||
00f254e2 KW |
3991 | /* In a "use bytes" we don't treat the source as UTF-8, but, still want |
3992 | * the destination to retain that flag */ | |
93e088e8 | 3993 | if (SvUTF8(source) && ! IN_BYTES) |
d54190f6 NC |
3994 | SvUTF8_on(dest); |
3995 | ||
00f254e2 | 3996 | if (!inplace) { /* Finish the rest of the string, unchanged */ |
d54190f6 NC |
3997 | /* This will copy the trailing NUL */ |
3998 | Copy(s + 1, d + 1, slen, U8); | |
3999 | SvCUR_set(dest, need - 1); | |
bbce6d69 | 4000 | } |
bbce6d69 | 4001 | } |
130c5df3 | 4002 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4003 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
445bf929 KW |
4004 | TAINT; |
4005 | SvTAINTED_on(dest); | |
4006 | } | |
130c5df3 | 4007 | #endif |
539689e7 FC |
4008 | if (dest != source && SvTAINTED(source)) |
4009 | SvTAINT(dest); | |
d54190f6 | 4010 | SvSETMAGIC(dest); |
3cb4e04f | 4011 | return NORMAL; |
79072805 LW |
4012 | } |
4013 | ||
67306194 NC |
4014 | /* There's so much setup/teardown code common between uc and lc, I wonder if |
4015 | it would be worth merging the two, and just having a switch outside each | |
00f254e2 | 4016 | of the three tight loops. There is less and less commonality though */ |
79072805 LW |
4017 | PP(pp_uc) |
4018 | { | |
39644a26 | 4019 | dSP; |
67306194 | 4020 | SV *source = TOPs; |
463ee0b2 | 4021 | STRLEN len; |
67306194 NC |
4022 | STRLEN min; |
4023 | SV *dest; | |
4024 | const U8 *s; | |
4025 | U8 *d; | |
79072805 | 4026 | |
67306194 NC |
4027 | SvGETMAGIC(source); |
4028 | ||
1921e031 | 4029 | if ( SvPADTMP(source) |
5cd5e2d6 FC |
4030 | && !SvREADONLY(source) && SvPOK(source) |
4031 | && !DO_UTF8(source) | |
130c5df3 KW |
4032 | && ( |
4033 | #ifdef USE_LOCALE_CTYPE | |
4034 | (IN_LC_RUNTIME(LC_CTYPE)) | |
31f05a37 | 4035 | ? ! IN_UTF8_CTYPE_LOCALE |
130c5df3 KW |
4036 | : |
4037 | #endif | |
4038 | ! IN_UNI_8_BIT)) | |
31f05a37 KW |
4039 | { |
4040 | ||
4041 | /* We can convert in place. The reason we can't if in UNI_8_BIT is to | |
4042 | * make the loop tight, so we overwrite the source with the dest before | |
4043 | * looking at it, and we need to look at the original source | |
4044 | * afterwards. There would also need to be code added to handle | |
4045 | * switching to not in-place in midstream if we run into characters | |
4046 | * that change the length. Since being in locale overrides UNI_8_BIT, | |
4047 | * that latter becomes irrelevant in the above test; instead for | |
4048 | * locale, the size can't normally change, except if the locale is a | |
4049 | * UTF-8 one */ | |
67306194 NC |
4050 | dest = source; |
4051 | s = d = (U8*)SvPV_force_nomg(source, len); | |
4052 | min = len + 1; | |
4053 | } else { | |
a0ed51b3 | 4054 | dTARGET; |
a0ed51b3 | 4055 | |
67306194 | 4056 | dest = TARG; |
128c9517 | 4057 | |
841a5e18 | 4058 | s = (const U8*)SvPV_nomg_const(source, len); |
67306194 NC |
4059 | min = len + 1; |
4060 | ||
4061 | SvUPGRADE(dest, SVt_PV); | |
3b416f41 | 4062 | d = (U8*)SvGROW(dest, min); |
67306194 NC |
4063 | (void)SvPOK_only(dest); |
4064 | ||
4065 | SETs(dest); | |
a0ed51b3 | 4066 | } |
31351b04 | 4067 | |
67306194 NC |
4068 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need |
4069 | to check DO_UTF8 again here. */ | |
4070 | ||
4071 | if (DO_UTF8(source)) { | |
4072 | const U8 *const send = s + len; | |
bfac13d4 | 4073 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
67306194 | 4074 | |
4c8a458a KW |
4075 | /* All occurrences of these are to be moved to follow any other marks. |
4076 | * This is context-dependent. We may not be passed enough context to | |
4077 | * move the iota subscript beyond all of them, but we do the best we can | |
4078 | * with what we're given. The result is always better than if we | |
4079 | * hadn't done this. And, the problem would only arise if we are | |
4080 | * passed a character without all its combining marks, which would be | |
4081 | * the caller's mistake. The information this is based on comes from a | |
4082 | * comment in Unicode SpecialCasing.txt, (and the Standard's text | |
4083 | * itself) and so can't be checked properly to see if it ever gets | |
4084 | * revised. But the likelihood of it changing is remote */ | |
00f254e2 | 4085 | bool in_iota_subscript = FALSE; |
00f254e2 | 4086 | |
67306194 | 4087 | while (s < send) { |
3e16b0e6 KW |
4088 | STRLEN u; |
4089 | STRLEN ulen; | |
4090 | UV uv; | |
7dbf68d2 | 4091 | if (in_iota_subscript && ! _is_utf8_mark(s)) { |
3e16b0e6 | 4092 | |
00f254e2 | 4093 | /* A non-mark. Time to output the iota subscript */ |
a78bc3c6 KW |
4094 | Copy(GREEK_CAPITAL_LETTER_IOTA_UTF8, d, capital_iota_len, U8); |
4095 | d += capital_iota_len; | |
00f254e2 | 4096 | in_iota_subscript = FALSE; |
8e058693 | 4097 | } |
00f254e2 | 4098 | |
8e058693 KW |
4099 | /* Then handle the current character. Get the changed case value |
4100 | * and copy it to the output buffer */ | |
00f254e2 | 4101 | |
8e058693 | 4102 | u = UTF8SKIP(s); |
130c5df3 | 4103 | #ifdef USE_LOCALE_CTYPE |
a1a5ec35 | 4104 | uv = _toUPPER_utf8_flags(s, send, tmpbuf, &ulen, IN_LC_RUNTIME(LC_CTYPE)); |
130c5df3 | 4105 | #else |
a1a5ec35 | 4106 | uv = _toUPPER_utf8_flags(s, send, tmpbuf, &ulen, 0); |
130c5df3 | 4107 | #endif |
a78bc3c6 KW |
4108 | #define GREEK_CAPITAL_LETTER_IOTA 0x0399 |
4109 | #define COMBINING_GREEK_YPOGEGRAMMENI 0x0345 | |
8e058693 | 4110 | if (uv == GREEK_CAPITAL_LETTER_IOTA |
4b88fb76 | 4111 | && utf8_to_uvchr_buf(s, send, 0) == COMBINING_GREEK_YPOGEGRAMMENI) |
8e058693 KW |
4112 | { |
4113 | in_iota_subscript = TRUE; | |
4114 | } | |
4115 | else { | |
4116 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { | |
4117 | /* If the eventually required minimum size outgrows the | |
4118 | * available space, we need to grow. */ | |
4119 | const UV o = d - (U8*)SvPVX_const(dest); | |
4120 | ||
4121 | /* If someone uppercases one million U+03B0s we SvGROW() | |
4122 | * one million times. Or we could try guessing how much to | |
4123 | * allocate without allocating too much. Such is life. | |
4124 | * See corresponding comment in lc code for another option | |
4125 | * */ | |
10656159 | 4126 | d = o + (U8*) SvGROW(dest, min); |
8e058693 KW |
4127 | } |
4128 | Copy(tmpbuf, d, ulen, U8); | |
4129 | d += ulen; | |
4130 | } | |
4131 | s += u; | |
67306194 | 4132 | } |
4c8a458a | 4133 | if (in_iota_subscript) { |
a78bc3c6 KW |
4134 | Copy(GREEK_CAPITAL_LETTER_IOTA_UTF8, d, capital_iota_len, U8); |
4135 | d += capital_iota_len; | |
4c8a458a | 4136 | } |
67306194 NC |
4137 | SvUTF8_on(dest); |
4138 | *d = '\0'; | |
094a2f8c | 4139 | |
67306194 | 4140 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); |
4c8a458a KW |
4141 | } |
4142 | else { /* Not UTF-8 */ | |
67306194 NC |
4143 | if (len) { |
4144 | const U8 *const send = s + len; | |
00f254e2 KW |
4145 | |
4146 | /* Use locale casing if in locale; regular style if not treating | |
4147 | * latin1 as having case; otherwise the latin1 casing. Do the | |
4148 | * whole thing in a tight loop, for speed, */ | |
130c5df3 | 4149 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4150 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
31f05a37 KW |
4151 | if (IN_UTF8_CTYPE_LOCALE) { |
4152 | goto do_uni_rules; | |
4153 | } | |
780fcc9f | 4154 | _CHECK_AND_WARN_PROBLEMATIC_LOCALE; |
67306194 | 4155 | for (; s < send; d++, s++) |
31f05a37 | 4156 | *d = (U8) toUPPER_LC(*s); |
31351b04 | 4157 | } |
130c5df3 KW |
4158 | else |
4159 | #endif | |
4160 | if (! IN_UNI_8_BIT) { | |
00f254e2 | 4161 | for (; s < send; d++, s++) { |
67306194 | 4162 | *d = toUPPER(*s); |
00f254e2 | 4163 | } |
31351b04 | 4164 | } |
00f254e2 | 4165 | else { |
91191cf7 | 4166 | #ifdef USE_LOCALE_CTYPE |
31f05a37 | 4167 | do_uni_rules: |
91191cf7 | 4168 | #endif |
00f254e2 KW |
4169 | for (; s < send; d++, s++) { |
4170 | *d = toUPPER_LATIN1_MOD(*s); | |
d14578b8 KW |
4171 | if (LIKELY(*d != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS)) { |
4172 | continue; | |
4173 | } | |
00f254e2 KW |
4174 | |
4175 | /* The mainstream case is the tight loop above. To avoid | |
4176 | * extra tests in that, all three characters that require | |
4177 | * special handling are mapped by the MOD to the one tested | |
4178 | * just above. | |
4179 | * Use the source to distinguish between the three cases */ | |
4180 | ||
79e064b9 KW |
4181 | #if UNICODE_MAJOR_VERSION > 2 \ |
4182 | || (UNICODE_MAJOR_VERSION == 2 && UNICODE_DOT_VERSION >= 1 \ | |
4183 | && UNICODE_DOT_DOT_VERSION >= 8) | |
00f254e2 KW |
4184 | if (*s == LATIN_SMALL_LETTER_SHARP_S) { |
4185 | ||
4186 | /* uc() of this requires 2 characters, but they are | |
4187 | * ASCII. If not enough room, grow the string */ | |
4188 | if (SvLEN(dest) < ++min) { | |
4189 | const UV o = d - (U8*)SvPVX_const(dest); | |
10656159 | 4190 | d = o + (U8*) SvGROW(dest, min); |
00f254e2 KW |
4191 | } |
4192 | *d++ = 'S'; *d = 'S'; /* upper case is 'SS' */ | |
4193 | continue; /* Back to the tight loop; still in ASCII */ | |
4194 | } | |
79e064b9 | 4195 | #endif |
00f254e2 KW |
4196 | |
4197 | /* The other two special handling characters have their | |
4198 | * upper cases outside the latin1 range, hence need to be | |
4199 | * in UTF-8, so the whole result needs to be in UTF-8. So, | |
4200 | * here we are somewhere in the middle of processing a | |
4201 | * non-UTF-8 string, and realize that we will have to convert | |
4202 | * the whole thing to UTF-8. What to do? There are | |
4203 | * several possibilities. The simplest to code is to | |
4204 | * convert what we have so far, set a flag, and continue on | |
4205 | * in the loop. The flag would be tested each time through | |
4206 | * the loop, and if set, the next character would be | |
4207 | * converted to UTF-8 and stored. But, I (khw) didn't want | |
4208 | * to slow down the mainstream case at all for this fairly | |
4209 | * rare case, so I didn't want to add a test that didn't | |
4210 | * absolutely have to be there in the loop, besides the | |
4211 | * possibility that it would get too complicated for | |
4212 | * optimizers to deal with. Another possibility is to just | |
4213 | * give up, convert the source to UTF-8, and restart the | |
4214 | * function that way. Another possibility is to convert | |
4215 | * both what has already been processed and what is yet to | |
4216 | * come separately to UTF-8, then jump into the loop that | |
4217 | * handles UTF-8. But the most efficient time-wise of the | |
4218 | * ones I could think of is what follows, and turned out to | |
4219 | * not require much extra code. */ | |
4220 | ||
4221 | /* Convert what we have so far into UTF-8, telling the | |
4222 | * function that we know it should be converted, and to | |
4223 | * allow extra space for what we haven't processed yet. | |
4224 | * Assume the worst case space requirements for converting | |
4225 | * what we haven't processed so far: that it will require | |
4226 | * two bytes for each remaining source character, plus the | |
4227 | * NUL at the end. This may cause the string pointer to | |
4228 | * move, so re-find it. */ | |
4229 | ||
4230 | len = d - (U8*)SvPVX_const(dest); | |
4231 | SvCUR_set(dest, len); | |
4232 | len = sv_utf8_upgrade_flags_grow(dest, | |
4233 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
4234 | (send -s) * 2 + 1); | |
4235 | d = (U8*)SvPVX(dest) + len; | |
4236 | ||
00f254e2 KW |
4237 | /* Now process the remainder of the source, converting to |
4238 | * upper and UTF-8. If a resulting byte is invariant in | |
4239 | * UTF-8, output it as-is, otherwise convert to UTF-8 and | |
4240 | * append it to the output. */ | |
00f254e2 | 4241 | for (; s < send; s++) { |
0ecfbd28 KW |
4242 | (void) _to_upper_title_latin1(*s, d, &len, 'S'); |
4243 | d += len; | |
00f254e2 KW |
4244 | } |
4245 | ||
4246 | /* Here have processed the whole source; no need to continue | |
4247 | * with the outer loop. Each character has been converted | |
4248 | * to upper case and converted to UTF-8 */ | |
4249 | ||
4250 | break; | |
4251 | } /* End of processing all latin1-style chars */ | |
4252 | } /* End of processing all chars */ | |
4253 | } /* End of source is not empty */ | |
4254 | ||
67306194 | 4255 | if (source != dest) { |
00f254e2 | 4256 | *d = '\0'; /* Here d points to 1 after last char, add NUL */ |
67306194 NC |
4257 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); |
4258 | } | |
00f254e2 | 4259 | } /* End of isn't utf8 */ |
130c5df3 | 4260 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4261 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
445bf929 KW |
4262 | TAINT; |
4263 | SvTAINTED_on(dest); | |
4264 | } | |
130c5df3 | 4265 | #endif |
539689e7 FC |
4266 | if (dest != source && SvTAINTED(source)) |
4267 | SvTAINT(dest); | |
67306194 | 4268 | SvSETMAGIC(dest); |
3cb4e04f | 4269 | return NORMAL; |
79072805 LW |
4270 | } |
4271 | ||
4272 | PP(pp_lc) | |
4273 | { | |
39644a26 | 4274 | dSP; |
ec9af7d4 | 4275 | SV *source = TOPs; |
463ee0b2 | 4276 | STRLEN len; |
ec9af7d4 NC |
4277 | STRLEN min; |
4278 | SV *dest; | |
4279 | const U8 *s; | |
4280 | U8 *d; | |
79072805 | 4281 | |
ec9af7d4 NC |
4282 | SvGETMAGIC(source); |
4283 | ||
1921e031 | 4284 | if ( SvPADTMP(source) |
5cd5e2d6 FC |
4285 | && !SvREADONLY(source) && SvPOK(source) |
4286 | && !DO_UTF8(source)) { | |
ec9af7d4 | 4287 | |
00f254e2 KW |
4288 | /* We can convert in place, as lowercasing anything in the latin1 range |
4289 | * (or else DO_UTF8 would have been on) doesn't lengthen it */ | |
ec9af7d4 NC |
4290 | dest = source; |
4291 | s = d = (U8*)SvPV_force_nomg(source, len); | |
4292 | min = len + 1; | |
4293 | } else { | |
a0ed51b3 | 4294 | dTARGET; |
a0ed51b3 | 4295 | |
ec9af7d4 NC |
4296 | dest = TARG; |
4297 | ||
841a5e18 | 4298 | s = (const U8*)SvPV_nomg_const(source, len); |
ec9af7d4 | 4299 | min = len + 1; |
128c9517 | 4300 | |
ec9af7d4 | 4301 | SvUPGRADE(dest, SVt_PV); |
3b416f41 | 4302 | d = (U8*)SvGROW(dest, min); |
ec9af7d4 NC |
4303 | (void)SvPOK_only(dest); |
4304 | ||
4305 | SETs(dest); | |
4306 | } | |
4307 | ||
4308 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need | |
4309 | to check DO_UTF8 again here. */ | |
4310 | ||
4311 | if (DO_UTF8(source)) { | |
4312 | const U8 *const send = s + len; | |
4313 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
4314 | ||
4315 | while (s < send) { | |
06b5486a KW |
4316 | const STRLEN u = UTF8SKIP(s); |
4317 | STRLEN ulen; | |
00f254e2 | 4318 | |
130c5df3 | 4319 | #ifdef USE_LOCALE_CTYPE |
a1a5ec35 | 4320 | _toLOWER_utf8_flags(s, send, tmpbuf, &ulen, IN_LC_RUNTIME(LC_CTYPE)); |
130c5df3 | 4321 | #else |
a1a5ec35 | 4322 | _toLOWER_utf8_flags(s, send, tmpbuf, &ulen, 0); |
130c5df3 | 4323 | #endif |
00f254e2 | 4324 | |
06b5486a | 4325 | /* Here is where we would do context-sensitive actions. See the |
6006ebd0 | 4326 | * commit message for 86510fb15 for why there isn't any */ |
00f254e2 | 4327 | |
06b5486a | 4328 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { |
fdb34c52 | 4329 | |
06b5486a KW |
4330 | /* If the eventually required minimum size outgrows the |
4331 | * available space, we need to grow. */ | |
4332 | const UV o = d - (U8*)SvPVX_const(dest); | |
fdb34c52 | 4333 | |
06b5486a KW |
4334 | /* If someone lowercases one million U+0130s we SvGROW() one |
4335 | * million times. Or we could try guessing how much to | |
4336 | * allocate without allocating too much. Such is life. | |
4337 | * Another option would be to grow an extra byte or two more | |
4338 | * each time we need to grow, which would cut down the million | |
4339 | * to 500K, with little waste */ | |
10656159 | 4340 | d = o + (U8*) SvGROW(dest, min); |
06b5486a | 4341 | } |
86510fb1 | 4342 | |
06b5486a KW |
4343 | /* Copy the newly lowercased letter to the output buffer we're |
4344 | * building */ | |
4345 | Copy(tmpbuf, d, ulen, U8); | |
4346 | d += ulen; | |
4347 | s += u; | |
00f254e2 | 4348 | } /* End of looping through the source string */ |
ec9af7d4 NC |
4349 | SvUTF8_on(dest); |
4350 | *d = '\0'; | |
4351 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
00f254e2 | 4352 | } else { /* Not utf8 */ |
31351b04 | 4353 | if (len) { |
ec9af7d4 | 4354 | const U8 *const send = s + len; |
00f254e2 KW |
4355 | |
4356 | /* Use locale casing if in locale; regular style if not treating | |
4357 | * latin1 as having case; otherwise the latin1 casing. Do the | |
4358 | * whole thing in a tight loop, for speed, */ | |
130c5df3 | 4359 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4360 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
780fcc9f | 4361 | _CHECK_AND_WARN_PROBLEMATIC_LOCALE; |
ec9af7d4 NC |
4362 | for (; s < send; d++, s++) |
4363 | *d = toLOWER_LC(*s); | |
445bf929 | 4364 | } |
130c5df3 KW |
4365 | else |
4366 | #endif | |
4367 | if (! IN_UNI_8_BIT) { | |
00f254e2 | 4368 | for (; s < send; d++, s++) { |
ec9af7d4 | 4369 | *d = toLOWER(*s); |
00f254e2 KW |
4370 | } |
4371 | } | |
4372 | else { | |
4373 | for (; s < send; d++, s++) { | |
4374 | *d = toLOWER_LATIN1(*s); | |
4375 | } | |
31351b04 | 4376 | } |
bbce6d69 | 4377 | } |
ec9af7d4 NC |
4378 | if (source != dest) { |
4379 | *d = '\0'; | |
4380 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
4381 | } | |
79072805 | 4382 | } |
130c5df3 | 4383 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4384 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
445bf929 KW |
4385 | TAINT; |
4386 | SvTAINTED_on(dest); | |
4387 | } | |
130c5df3 | 4388 | #endif |
539689e7 FC |
4389 | if (dest != source && SvTAINTED(source)) |
4390 | SvTAINT(dest); | |
ec9af7d4 | 4391 | SvSETMAGIC(dest); |
3cb4e04f | 4392 | return NORMAL; |
79072805 LW |
4393 | } |
4394 | ||
a0d0e21e | 4395 | PP(pp_quotemeta) |
79072805 | 4396 | { |
20b7effb | 4397 | dSP; dTARGET; |
1b6737cc | 4398 | SV * const sv = TOPs; |
a0d0e21e | 4399 | STRLEN len; |
eb578fdb | 4400 | const char *s = SvPV_const(sv,len); |
79072805 | 4401 | |
7e2040f0 | 4402 | SvUTF8_off(TARG); /* decontaminate */ |
a0d0e21e | 4403 | if (len) { |
eb578fdb | 4404 | char *d; |
862a34c6 | 4405 | SvUPGRADE(TARG, SVt_PV); |
c07a80fd | 4406 | SvGROW(TARG, (len * 2) + 1); |
a0d0e21e | 4407 | d = SvPVX(TARG); |
7e2040f0 | 4408 | if (DO_UTF8(sv)) { |
0dd2cdef | 4409 | while (len) { |
29050de5 | 4410 | STRLEN ulen = UTF8SKIP(s); |
2e2b2571 KW |
4411 | bool to_quote = FALSE; |
4412 | ||
4413 | if (UTF8_IS_INVARIANT(*s)) { | |
4414 | if (_isQUOTEMETA(*s)) { | |
4415 | to_quote = TRUE; | |
4416 | } | |
4417 | } | |
042d9e50 | 4418 | else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, s + len)) { |
3fea7d29 | 4419 | if ( |
130c5df3 | 4420 | #ifdef USE_LOCALE_CTYPE |
20adcf7c KW |
4421 | /* In locale, we quote all non-ASCII Latin1 chars. |
4422 | * Otherwise use the quoting rules */ | |
3fea7d29 BF |
4423 | |
4424 | IN_LC_RUNTIME(LC_CTYPE) | |
4425 | || | |
4426 | #endif | |
a62b247b | 4427 | _isQUOTEMETA(EIGHT_BIT_UTF8_TO_NATIVE(*s, *(s + 1)))) |
2e2b2571 KW |
4428 | { |
4429 | to_quote = TRUE; | |
4430 | } | |
4431 | } | |
685289b5 | 4432 | else if (is_QUOTEMETA_high(s)) { |
2e2b2571 KW |
4433 | to_quote = TRUE; |
4434 | } | |
4435 | ||
4436 | if (to_quote) { | |
4437 | *d++ = '\\'; | |
4438 | } | |
29050de5 KW |
4439 | if (ulen > len) |
4440 | ulen = len; | |
4441 | len -= ulen; | |
4442 | while (ulen--) | |
4443 | *d++ = *s++; | |
0dd2cdef | 4444 | } |
7e2040f0 | 4445 | SvUTF8_on(TARG); |
0dd2cdef | 4446 | } |
2e2b2571 KW |
4447 | else if (IN_UNI_8_BIT) { |
4448 | while (len--) { | |
4449 | if (_isQUOTEMETA(*s)) | |
4450 | *d++ = '\\'; | |
4451 | *d++ = *s++; | |
4452 | } | |
4453 | } | |
0dd2cdef | 4454 | else { |
2e2b2571 KW |
4455 | /* For non UNI_8_BIT (and hence in locale) just quote all \W |
4456 | * including everything above ASCII */ | |
0dd2cdef | 4457 | while (len--) { |
adfec831 | 4458 | if (!isWORDCHAR_A(*s)) |
0dd2cdef LW |
4459 | *d++ = '\\'; |
4460 | *d++ = *s++; | |
4461 | } | |
79072805 | 4462 | } |
a0d0e21e | 4463 | *d = '\0'; |
349d4f2f | 4464 | SvCUR_set(TARG, d - SvPVX_const(TARG)); |
3aa33fe5 | 4465 | (void)SvPOK_only_UTF8(TARG); |
79072805 | 4466 | } |
a0d0e21e LW |
4467 | else |
4468 | sv_setpvn(TARG, s, len); | |
ec93b65f | 4469 | SETTARG; |
cfe40115 | 4470 | return NORMAL; |
79072805 LW |
4471 | } |
4472 | ||
838f2281 BF |
4473 | PP(pp_fc) |
4474 | { | |
838f2281 BF |
4475 | dTARGET; |
4476 | dSP; | |
4477 | SV *source = TOPs; | |
4478 | STRLEN len; | |
4479 | STRLEN min; | |
4480 | SV *dest; | |
4481 | const U8 *s; | |
4482 | const U8 *send; | |
4483 | U8 *d; | |
bfac13d4 | 4484 | U8 tmpbuf[UTF8_MAXBYTES_CASE + 1]; |
9b63e895 KW |
4485 | #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \ |
4486 | || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \ | |
4487 | || UNICODE_DOT_DOT_VERSION > 0) | |
a4b69695 KW |
4488 | const bool full_folding = TRUE; /* This variable is here so we can easily |
4489 | move to more generality later */ | |
9b63e895 KW |
4490 | #else |
4491 | const bool full_folding = FALSE; | |
4492 | #endif | |
838f2281 | 4493 | const U8 flags = ( full_folding ? FOLD_FLAGS_FULL : 0 ) |
130c5df3 KW |
4494 | #ifdef USE_LOCALE_CTYPE |
4495 | | ( IN_LC_RUNTIME(LC_CTYPE) ? FOLD_FLAGS_LOCALE : 0 ) | |
4496 | #endif | |
4497 | ; | |
838f2281 BF |
4498 | |
4499 | /* This is a facsimile of pp_lc, but with a thousand bugs thanks to me. | |
4500 | * You are welcome(?) -Hugmeir | |
4501 | */ | |
4502 | ||
4503 | SvGETMAGIC(source); | |
4504 | ||
4505 | dest = TARG; | |
4506 | ||
4507 | if (SvOK(source)) { | |
4508 | s = (const U8*)SvPV_nomg_const(source, len); | |
4509 | } else { | |
4510 | if (ckWARN(WARN_UNINITIALIZED)) | |
4511 | report_uninit(source); | |
4512 | s = (const U8*)""; | |
4513 | len = 0; | |
4514 | } | |
4515 | ||
4516 | min = len + 1; | |
4517 | ||
4518 | SvUPGRADE(dest, SVt_PV); | |
4519 | d = (U8*)SvGROW(dest, min); | |
4520 | (void)SvPOK_only(dest); | |
4521 | ||
4522 | SETs(dest); | |
4523 | ||
4524 | send = s + len; | |
4525 | if (DO_UTF8(source)) { /* UTF-8 flagged string. */ | |
838f2281 BF |
4526 | while (s < send) { |
4527 | const STRLEN u = UTF8SKIP(s); | |
4528 | STRLEN ulen; | |
4529 | ||
a1a5ec35 | 4530 | _toFOLD_utf8_flags(s, send, tmpbuf, &ulen, flags); |
838f2281 BF |
4531 | |
4532 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { | |
4533 | const UV o = d - (U8*)SvPVX_const(dest); | |
10656159 | 4534 | d = o + (U8*) SvGROW(dest, min); |
838f2281 BF |
4535 | } |
4536 | ||
4537 | Copy(tmpbuf, d, ulen, U8); | |
4538 | d += ulen; | |
4539 | s += u; | |
4540 | } | |
4541 | SvUTF8_on(dest); | |
838f2281 | 4542 | } /* Unflagged string */ |
0902dd32 | 4543 | else if (len) { |
130c5df3 | 4544 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4545 | if ( IN_LC_RUNTIME(LC_CTYPE) ) { /* Under locale */ |
31f05a37 KW |
4546 | if (IN_UTF8_CTYPE_LOCALE) { |
4547 | goto do_uni_folding; | |
4548 | } | |
780fcc9f | 4549 | _CHECK_AND_WARN_PROBLEMATIC_LOCALE; |
838f2281 | 4550 | for (; s < send; d++, s++) |
ea36a843 | 4551 | *d = (U8) toFOLD_LC(*s); |
838f2281 | 4552 | } |
130c5df3 KW |
4553 | else |
4554 | #endif | |
4555 | if ( !IN_UNI_8_BIT ) { /* Under nothing, or bytes */ | |
838f2281 | 4556 | for (; s < send; d++, s++) |
d22b930b | 4557 | *d = toFOLD(*s); |
838f2281 BF |
4558 | } |
4559 | else { | |
91191cf7 | 4560 | #ifdef USE_LOCALE_CTYPE |
31f05a37 | 4561 | do_uni_folding: |
91191cf7 | 4562 | #endif |
d14578b8 KW |
4563 | /* For ASCII and the Latin-1 range, there's only two troublesome |
4564 | * folds, \x{DF} (\N{LATIN SMALL LETTER SHARP S}), which under full | |
22e255cb | 4565 | * casefolding becomes 'ss'; and \x{B5} (\N{MICRO SIGN}), which |
d14578b8 KW |
4566 | * under any fold becomes \x{3BC} (\N{GREEK SMALL LETTER MU}) -- |
4567 | * For the rest, the casefold is their lowercase. */ | |
838f2281 BF |
4568 | for (; s < send; d++, s++) { |
4569 | if (*s == MICRO_SIGN) { | |
d14578b8 KW |
4570 | /* \N{MICRO SIGN}'s casefold is \N{GREEK SMALL LETTER MU}, |
4571 | * which is outside of the latin-1 range. There's a couple | |
4572 | * of ways to deal with this -- khw discusses them in | |
4573 | * pp_lc/uc, so go there :) What we do here is upgrade what | |
4574 | * we had already casefolded, then enter an inner loop that | |
4575 | * appends the rest of the characters as UTF-8. */ | |
838f2281 BF |
4576 | len = d - (U8*)SvPVX_const(dest); |
4577 | SvCUR_set(dest, len); | |
4578 | len = sv_utf8_upgrade_flags_grow(dest, | |
4579 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
ea4d335b KW |
4580 | /* The max expansion for latin1 |
4581 | * chars is 1 byte becomes 2 */ | |
4582 | (send -s) * 2 + 1); | |
838f2281 BF |
4583 | d = (U8*)SvPVX(dest) + len; |
4584 | ||
a78bc3c6 KW |
4585 | Copy(GREEK_SMALL_LETTER_MU_UTF8, d, small_mu_len, U8); |
4586 | d += small_mu_len; | |
838f2281 BF |
4587 | s++; |
4588 | for (; s < send; s++) { | |
4589 | STRLEN ulen; | |
4590 | UV fc = _to_uni_fold_flags(*s, tmpbuf, &ulen, flags); | |
6f2d5cbc | 4591 | if UVCHR_IS_INVARIANT(fc) { |
d14578b8 KW |
4592 | if (full_folding |
4593 | && *s == LATIN_SMALL_LETTER_SHARP_S) | |
4594 | { | |
838f2281 BF |
4595 | *d++ = 's'; |
4596 | *d++ = 's'; | |
4597 | } | |
4598 | else | |
4599 | *d++ = (U8)fc; | |
4600 | } | |
4601 | else { | |
4602 | Copy(tmpbuf, d, ulen, U8); | |
4603 | d += ulen; | |
4604 | } | |
4605 | } | |
4606 | break; | |
4607 | } | |
4608 | else if (full_folding && *s == LATIN_SMALL_LETTER_SHARP_S) { | |
d14578b8 KW |
4609 | /* Under full casefolding, LATIN SMALL LETTER SHARP S |
4610 | * becomes "ss", which may require growing the SV. */ | |
838f2281 BF |
4611 | if (SvLEN(dest) < ++min) { |
4612 | const UV o = d - (U8*)SvPVX_const(dest); | |
10656159 | 4613 | d = o + (U8*) SvGROW(dest, min); |
838f2281 BF |
4614 | } |
4615 | *(d)++ = 's'; | |
4616 | *d = 's'; | |
4617 | } | |
d14578b8 KW |
4618 | else { /* If it's not one of those two, the fold is their lower |
4619 | case */ | |
838f2281 BF |
4620 | *d = toLOWER_LATIN1(*s); |
4621 | } | |
4622 | } | |
4623 | } | |
4624 | } | |
4625 | *d = '\0'; | |
4626 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
4627 | ||
130c5df3 | 4628 | #ifdef USE_LOCALE_CTYPE |
d6ded950 | 4629 | if (IN_LC_RUNTIME(LC_CTYPE)) { |
445bf929 KW |
4630 | TAINT; |
4631 | SvTAINTED_on(dest); | |
4632 | } | |
130c5df3 | 4633 | #endif |
838f2281 BF |
4634 | if (SvTAINTED(source)) |
4635 | SvTAINT(dest); | |
4636 | SvSETMAGIC(dest); | |
4637 | RETURN; | |
4638 | } | |
4639 | ||
a0d0e21e | 4640 | /* Arrays. */ |
79072805 | 4641 | |
a0d0e21e | 4642 | PP(pp_aslice) |
79072805 | 4643 | { |
20b7effb | 4644 | dSP; dMARK; dORIGMARK; |
eb578fdb KW |
4645 | AV *const av = MUTABLE_AV(POPs); |
4646 | const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); | |
79072805 | 4647 | |
a0d0e21e | 4648 | if (SvTYPE(av) == SVt_PVAV) { |
4ad10a0b VP |
4649 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; |
4650 | bool can_preserve = FALSE; | |
4651 | ||
4652 | if (localizing) { | |
4653 | MAGIC *mg; | |
4654 | HV *stash; | |
4655 | ||
4656 | can_preserve = SvCANEXISTDELETE(av); | |
4657 | } | |
4658 | ||
4659 | if (lval && localizing) { | |
eb578fdb | 4660 | SV **svp; |
c70927a6 | 4661 | SSize_t max = -1; |
924508f0 | 4662 | for (svp = MARK + 1; svp <= SP; svp++) { |
c70927a6 | 4663 | const SSize_t elem = SvIV(*svp); |
748a9306 LW |
4664 | if (elem > max) |
4665 | max = elem; | |
4666 | } | |
4667 | if (max > AvMAX(av)) | |
4668 | av_extend(av, max); | |
4669 | } | |
4ad10a0b | 4670 | |
a0d0e21e | 4671 | while (++MARK <= SP) { |
eb578fdb | 4672 | SV **svp; |
c70927a6 | 4673 | SSize_t elem = SvIV(*MARK); |
4ad10a0b | 4674 | bool preeminent = TRUE; |
a0d0e21e | 4675 | |
4ad10a0b VP |
4676 | if (localizing && can_preserve) { |
4677 | /* If we can determine whether the element exist, | |
4678 | * Try to preserve the existenceness of a tied array | |
4679 | * element by using EXISTS and DELETE if possible. | |
4680 | * Fallback to FETCH and STORE otherwise. */ | |
4681 | preeminent = av_exists(av, elem); | |
4682 | } | |
4683 | ||
a0d0e21e LW |
4684 | svp = av_fetch(av, elem, lval); |
4685 | if (lval) { | |
ce0d59fd | 4686 | if (!svp || !*svp) |
cea2e8a9 | 4687 | DIE(aTHX_ PL_no_aelem, elem); |
4ad10a0b VP |
4688 | if (localizing) { |
4689 | if (preeminent) | |
4690 | save_aelem(av, elem, svp); | |
4691 | else | |
4692 | SAVEADELETE(av, elem); | |
4693 | } | |
79072805 | 4694 | } |
3280af22 | 4695 | *MARK = svp ? *svp : &PL_sv_undef; |
79072805 LW |
4696 | } |
4697 | } | |
82334630 | 4698 | if (GIMME_V != G_ARRAY) { |
a0d0e21e | 4699 | MARK = ORIGMARK; |
04ab2c87 | 4700 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
a0d0e21e LW |
4701 | SP = MARK; |
4702 | } | |
79072805 LW |
4703 | RETURN; |
4704 | } | |
4705 | ||
6dd3e0f2 RZ |
4706 | PP(pp_kvaslice) |
4707 | { | |
20b7effb | 4708 | dSP; dMARK; |
6dd3e0f2 RZ |
4709 | AV *const av = MUTABLE_AV(POPs); |
4710 | I32 lval = (PL_op->op_flags & OPf_MOD); | |
adad97db | 4711 | SSize_t items = SP - MARK; |
6dd3e0f2 RZ |
4712 | |
4713 | if (PL_op->op_private & OPpMAYBE_LVSUB) { | |
4714 | const I32 flags = is_lvalue_sub(); | |
4715 | if (flags) { | |
4716 | if (!(flags & OPpENTERSUB_INARGS)) | |
7aae0299 | 4717 | /* diag_listed_as: Can't modify %s in %s */ |
6dd3e0f2 RZ |
4718 | Perl_croak(aTHX_ "Can't modify index/value array slice in list assignment"); |
4719 | lval = flags; | |
4720 | } | |
4721 | } | |
4722 | ||
4723 | MEXTEND(SP,items); | |
4724 | while (items > 1) { | |
4725 | *(MARK+items*2-1) = *(MARK+items); | |
4726 | items--; | |
4727 | } | |
4728 | items = SP-MARK; | |
4729 | SP += items; | |
4730 | ||
4731 | while (++MARK <= SP) { | |
4732 | SV **svp; | |
4733 | ||
4734 | svp = av_fetch(av, SvIV(*MARK), lval); | |
4735 | if (lval) { | |
4736 | if (!svp || !*svp || *svp == &PL_sv_undef) { | |
4737 | DIE(aTHX_ PL_no_aelem, SvIV(*MARK)); | |
4738 | } | |
4739 | *MARK = sv_mortalcopy(*MARK); | |
4740 | } | |
4741 | *++MARK = svp ? *svp : &PL_sv_undef; | |
4742 | } | |
82334630 | 4743 | if (GIMME_V != G_ARRAY) { |
6dd3e0f2 RZ |
4744 | MARK = SP - items*2; |
4745 | *++MARK = items > 0 ? *SP : &PL_sv_undef; | |
4746 | SP = MARK; | |
4747 | } | |
4748 | RETURN; | |
4749 | } | |
4750 | ||
b1c05ba5 | 4751 | |
878d132a NC |
4752 | PP(pp_aeach) |
4753 | { | |
878d132a | 4754 | dSP; |
502c6561 | 4755 | AV *array = MUTABLE_AV(POPs); |
1c23e2bd | 4756 | const U8 gimme = GIMME_V; |
453d94a9 | 4757 | IV *iterp = Perl_av_iter_p(aTHX_ array); |
878d132a NC |
4758 | const IV current = (*iterp)++; |
4759 | ||
b9f2b683 | 4760 | if (current > av_tindex(array)) { |
878d132a NC |
4761 | *iterp = 0; |
4762 | if (gimme == G_SCALAR) | |
4763 | RETPUSHUNDEF; | |
4764 | else | |
4765 | RETURN; | |
4766 | } | |
4767 | ||
4768 | EXTEND(SP, 2); | |
e1dccc0d | 4769 | mPUSHi(current); |
878d132a NC |
4770 | if (gimme == G_ARRAY) { |
4771 | SV **const element = av_fetch(array, current, 0); | |
4772 | PUSHs(element ? *element : &PL_sv_undef); | |
4773 | } | |
4774 | RETURN; | |
4775 | } | |
4776 | ||
b1c05ba5 | 4777 | /* also used for: pp_avalues()*/ |
878d132a NC |
4778 | PP(pp_akeys) |
4779 | { | |
878d132a | 4780 | dSP; |
502c6561 | 4781 | AV *array = MUTABLE_AV(POPs); |
1c23e2bd | 4782 | const U8 gimme = GIMME_V; |
878d132a NC |
4783 | |
4784 | *Perl_av_iter_p(aTHX_ array) = 0; | |
4785 | ||
4786 | if (gimme == G_SCALAR) { | |
4787 | dTARGET; | |
b9f2b683 | 4788 | PUSHi(av_tindex(array) + 1); |
878d132a NC |
4789 | } |
4790 | else if (gimme == G_ARRAY) { | |
738155d2 FC |
4791 | if (UNLIKELY(PL_op->op_private & OPpMAYBE_LVSUB)) { |
4792 | const I32 flags = is_lvalue_sub(); | |
4793 | if (flags && !(flags & OPpENTERSUB_INARGS)) | |
4794 | /* diag_listed_as: Can't modify %s in %s */ | |
4795 | Perl_croak(aTHX_ | |
4796 | "Can't modify keys on array in list assignment"); | |
4797 | } | |
4798 | { | |
878d132a | 4799 | IV n = Perl_av_len(aTHX_ array); |
e1dccc0d | 4800 | IV i; |
878d132a NC |
4801 | |
4802 | EXTEND(SP, n + 1); | |
4803 | ||
73665bc4 FC |
4804 | if ( PL_op->op_type == OP_AKEYS |
4805 | || ( PL_op->op_type == OP_AVHVSWITCH | |
cd642408 | 4806 | && (PL_op->op_private & 3) + OP_AEACH == OP_AKEYS )) |
73665bc4 | 4807 | { |
e1dccc0d | 4808 | for (i = 0; i <= n; i++) { |
878d132a NC |
4809 | mPUSHi(i); |
4810 | } | |
4811 | } | |
4812 | else { | |
4813 | for (i = 0; i <= n; i++) { | |
4814 | SV *const *const elem = Perl_av_fetch(aTHX_ array, i, 0); | |
4815 | PUSHs(elem ? *elem : &PL_sv_undef); | |
4816 | } | |
4817 | } | |
738155d2 | 4818 | } |
878d132a NC |
4819 | } |
4820 | RETURN; | |
4821 | } | |
4822 | ||
79072805 LW |
4823 | /* Associative arrays. */ |
4824 | ||
4825 | PP(pp_each) | |
4826 | { | |
39644a26 | 4827 | dSP; |
85fbaab2 | 4828 | HV * hash = MUTABLE_HV(POPs); |
c07a80fd | 4829 | HE *entry; |
1c23e2bd | 4830 | const U8 gimme = GIMME_V; |
8ec5e241 | 4831 | |
6d822dc4 | 4832 | entry = hv_iternext(hash); |
79072805 | 4833 | |
79072805 LW |
4834 | EXTEND(SP, 2); |
4835 | if (entry) { | |
1b6737cc | 4836 | SV* const sv = hv_iterkeysv(entry); |
2b32fed8 | 4837 | PUSHs(sv); |
54310121 | 4838 | if (gimme == G_ARRAY) { |
59af0135 | 4839 | SV *val; |
6d822dc4 | 4840 | val = hv_iterval(hash, entry); |
59af0135 | 4841 | PUSHs(val); |
79072805 | 4842 | } |
79072805 | 4843 | } |
54310121 | 4844 | else if (gimme == G_SCALAR) |
79072805 LW |
4845 | RETPUSHUNDEF; |
4846 | ||
4847 | RETURN; | |
4848 | } | |
4849 | ||
7332a6c4 VP |
4850 | STATIC OP * |
4851 | S_do_delete_local(pTHX) | |
79072805 | 4852 | { |
39644a26 | 4853 | dSP; |
1c23e2bd | 4854 | const U8 gimme = GIMME_V; |
7332a6c4 VP |
4855 | const MAGIC *mg; |
4856 | HV *stash; | |
ca3f996a | 4857 | const bool sliced = !!(PL_op->op_private & OPpSLICE); |
626040f7 | 4858 | SV **unsliced_keysv = sliced ? NULL : sp--; |
ca3f996a | 4859 | SV * const osv = POPs; |
626040f7 | 4860 | SV **mark = sliced ? PL_stack_base + POPMARK : unsliced_keysv-1; |
ca3f996a FC |
4861 | dORIGMARK; |
4862 | const bool tied = SvRMAGICAL(osv) | |
7332a6c4 | 4863 | && mg_find((const SV *)osv, PERL_MAGIC_tied); |
ca3f996a FC |
4864 | const bool can_preserve = SvCANEXISTDELETE(osv); |
4865 | const U32 type = SvTYPE(osv); | |
626040f7 | 4866 | SV ** const end = sliced ? SP : unsliced_keysv; |
ca3f996a FC |
4867 | |
4868 | if (type == SVt_PVHV) { /* hash element */ | |
7332a6c4 | 4869 | HV * const hv = MUTABLE_HV(osv); |
ca3f996a | 4870 | while (++MARK <= end) { |
7332a6c4 VP |
4871 | SV * const keysv = *MARK; |
4872 | SV *sv = NULL; | |
4873 | bool preeminent = TRUE; | |
4874 | if (can_preserve) | |
4875 | preeminent = hv_exists_ent(hv, keysv, 0); | |
4876 | if (tied) { | |
4877 | HE *he = hv_fetch_ent(hv, keysv, 1, 0); | |
4878 | if (he) | |
4879 | sv = HeVAL(he); | |
4880 | else | |
4881 | preeminent = FALSE; | |
4882 | } | |
4883 | else { | |
4884 | sv = hv_delete_ent(hv, keysv, 0, 0); | |
9332b95f FC |
4885 | if (preeminent) |
4886 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
7332a6c4 VP |
4887 | } |
4888 | if (preeminent) { | |
be6064fd | 4889 | if (!sv) DIE(aTHX_ PL_no_helem_sv, SVfARG(keysv)); |
7332a6c4 VP |
4890 | save_helem_flags(hv, keysv, &sv, SAVEf_KEEPOLDELEM); |
4891 | if (tied) { | |
4892 | *MARK = sv_mortalcopy(sv); | |
4893 | mg_clear(sv); | |
4894 | } else | |
4895 | *MARK = sv; | |
4896 | } | |
4897 | else { | |
4898 | SAVEHDELETE(hv, keysv); | |
4899 | *MARK = &PL_sv_undef; | |
4900 | } | |
4901 | } | |
ca3f996a FC |
4902 | } |
4903 | else if (type == SVt_PVAV) { /* array element */ | |
7332a6c4 VP |
4904 | if (PL_op->op_flags & OPf_SPECIAL) { |
4905 | AV * const av = MUTABLE_AV(osv); | |
ca3f996a | 4906 | while (++MARK <= end) { |
c70927a6 | 4907 | SSize_t idx = SvIV(*MARK); |
7332a6c4 VP |
4908 | SV *sv = NULL; |
4909 | bool preeminent = TRUE; | |
4910 | if (can_preserve) | |
4911 | preeminent = av_exists(av, idx); | |
4912 | if (tied) { | |
4913 | SV **svp = av_fetch(av, idx, 1); | |
4914 | if (svp) | |
4915 | sv = *svp; | |
4916 | else | |
4917 | preeminent = FALSE; | |
4918 | } | |
4919 | else { | |
4920 | sv = av_delete(av, idx, 0); | |
9332b95f FC |
4921 | if (preeminent) |
4922 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
7332a6c4 VP |
4923 | } |
4924 | if (preeminent) { | |
4925 | save_aelem_flags(av, idx, &sv, SAVEf_KEEPOLDELEM); | |
4926 | if (tied) { | |
4927 | *MARK = sv_mortalcopy(sv); | |
4928 | mg_clear(sv); | |
4929 | } else | |
4930 | *MARK = sv; | |
4931 | } | |
4932 | else { | |
4933 | SAVEADELETE(av, idx); | |
4934 | *MARK = &PL_sv_undef; | |
4935 | } | |
4936 | } | |
4937 | } | |
ca3f996a FC |
4938 | else |
4939 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
4940 | } | |
4941 | else | |
7332a6c4 | 4942 | DIE(aTHX_ "Not a HASH reference"); |
ca3f996a | 4943 | if (sliced) { |
7332a6c4 VP |
4944 | if (gimme == G_VOID) |
4945 | SP = ORIGMARK; | |
4946 | else if (gimme == G_SCALAR) { | |
4947 | MARK = ORIGMARK; | |
4948 | if (SP > MARK) | |
4949 | *++MARK = *SP; | |
4950 | else | |
4951 | *++MARK = &PL_sv_undef; | |
4952 | SP = MARK; | |
4953 | } | |
4954 | } | |
ca3f996a | 4955 | else if (gimme != G_VOID) |
626040f7 | 4956 | PUSHs(*unsliced_keysv); |
7332a6c4 VP |
4957 | |
4958 | RETURN; | |
4959 | } | |
4960 | ||
4961 | PP(pp_delete) | |
4962 | { | |
7332a6c4 | 4963 | dSP; |
1c23e2bd | 4964 | U8 gimme; |
7332a6c4 VP |
4965 | I32 discard; |
4966 | ||
4967 | if (PL_op->op_private & OPpLVAL_INTRO) | |
4968 | return do_delete_local(); | |
4969 | ||
4970 | gimme = GIMME_V; | |
4971 | discard = (gimme == G_VOID) ? G_DISCARD : 0; | |
5f05dabc | 4972 | |
cc0776d6 | 4973 | if (PL_op->op_private & (OPpSLICE|OPpKVSLICE)) { |
5f05dabc | 4974 | dMARK; dORIGMARK; |
85fbaab2 | 4975 | HV * const hv = MUTABLE_HV(POPs); |
1b6737cc | 4976 | const U32 hvtype = SvTYPE(hv); |
cc0776d6 DIM |
4977 | int skip = 0; |
4978 | if (PL_op->op_private & OPpKVSLICE) { | |
4979 | SSize_t items = SP - MARK; | |
4980 | ||
4981 | MEXTEND(SP,items); | |
4982 | while (items > 1) { | |
4983 | *(MARK+items*2-1) = *(MARK+items); | |
4984 | items--; | |
4985 | } | |
4986 | items = SP - MARK; | |
4987 | SP += items; | |
4988 | skip = 1; | |
4989 | } | |
01020589 | 4990 | if (hvtype == SVt_PVHV) { /* hash element */ |
cc0776d6 DIM |
4991 | while ((MARK += (1+skip)) <= SP) { |
4992 | SV * const sv = hv_delete_ent(hv, *(MARK-skip), discard, 0); | |
01020589 GS |
4993 | *MARK = sv ? sv : &PL_sv_undef; |
4994 | } | |
5f05dabc | 4995 | } |
6d822dc4 MS |
4996 | else if (hvtype == SVt_PVAV) { /* array element */ |
4997 | if (PL_op->op_flags & OPf_SPECIAL) { | |
cc0776d6 DIM |
4998 | while ((MARK += (1+skip)) <= SP) { |
4999 | SV * const sv = av_delete(MUTABLE_AV(hv), SvIV(*(MARK-skip)), discard); | |
6d822dc4 MS |
5000 | *MARK = sv ? sv : &PL_sv_undef; |
5001 | } | |
5002 | } | |
01020589 GS |
5003 | } |
5004 | else | |
5005 | DIE(aTHX_ "Not a HASH reference"); | |
54310121 | 5006 | if (discard) |
5007 | SP = ORIGMARK; | |
5008 | else if (gimme == G_SCALAR) { | |
5f05dabc | 5009 | MARK = ORIGMARK; |
9111c9c0 DM |
5010 | if (SP > MARK) |
5011 | *++MARK = *SP; | |
5012 | else | |
5013 | *++MARK = &PL_sv_undef; | |
5f05dabc | 5014 | SP = MARK; |
5015 | } | |
5016 | } | |
5017 | else { | |
5018 | SV *keysv = POPs; | |
85fbaab2 | 5019 | HV * const hv = MUTABLE_HV(POPs); |
295d248e | 5020 | SV *sv = NULL; |
97fcbf96 MB |
5021 | if (SvTYPE(hv) == SVt_PVHV) |
5022 | sv = hv_delete_ent(hv, keysv, discard, 0); | |
01020589 GS |
5023 | else if (SvTYPE(hv) == SVt_PVAV) { |
5024 | if (PL_op->op_flags & OPf_SPECIAL) | |
502c6561 | 5025 | sv = av_delete(MUTABLE_AV(hv), SvIV(keysv), discard); |
af288a60 HS |
5026 | else |
5027 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
01020589 | 5028 | } |
97fcbf96 | 5029 | else |
cea2e8a9 | 5030 | DIE(aTHX_ "Not a HASH reference"); |
5f05dabc | 5031 | if (!sv) |
3280af22 | 5032 | sv = &PL_sv_undef; |
54310121 | 5033 | if (!discard) |
5034 | PUSHs(sv); | |
79072805 | 5035 | } |
79072805 LW |
5036 | RETURN; |
5037 | } | |
5038 | ||
a0d0e21e | 5039 | PP(pp_exists) |
79072805 | 5040 | { |
39644a26 | 5041 | dSP; |
afebc493 GS |
5042 | SV *tmpsv; |
5043 | HV *hv; | |
5044 | ||
c7e88ff3 | 5045 | if (UNLIKELY( PL_op->op_private & OPpEXISTS_SUB )) { |
afebc493 | 5046 | GV *gv; |
0bd48802 | 5047 | SV * const sv = POPs; |
f2c0649b | 5048 | CV * const cv = sv_2cv(sv, &hv, &gv, 0); |
afebc493 GS |
5049 | if (cv) |
5050 | RETPUSHYES; | |
5051 | if (gv && isGV(gv) && GvCV(gv) && !GvCVGEN(gv)) | |
5052 | RETPUSHYES; | |
5053 | RETPUSHNO; | |
5054 | } | |
5055 | tmpsv = POPs; | |
85fbaab2 | 5056 | hv = MUTABLE_HV(POPs); |
c7e88ff3 | 5057 | if (LIKELY( SvTYPE(hv) == SVt_PVHV )) { |
ae77835f | 5058 | if (hv_exists_ent(hv, tmpsv, 0)) |
c750a3ec | 5059 | RETPUSHYES; |
ef54e1a4 JH |
5060 | } |
5061 | else if (SvTYPE(hv) == SVt_PVAV) { | |
01020589 | 5062 | if (PL_op->op_flags & OPf_SPECIAL) { /* array element */ |
502c6561 | 5063 | if (av_exists(MUTABLE_AV(hv), SvIV(tmpsv))) |
01020589 GS |
5064 | RETPUSHYES; |
5065 | } | |
ef54e1a4 JH |
5066 | } |
5067 | else { | |
cea2e8a9 | 5068 | DIE(aTHX_ "Not a HASH reference"); |
a0d0e21e | 5069 | } |
a0d0e21e LW |
5070 | RETPUSHNO; |
5071 | } | |
79072805 | 5072 | |
a0d0e21e LW |
5073 | PP(pp_hslice) |
5074 | { | |
20b7effb | 5075 | dSP; dMARK; dORIGMARK; |
eb578fdb KW |
5076 | HV * const hv = MUTABLE_HV(POPs); |
5077 | const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); | |
1b6737cc | 5078 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; |
d30e492c | 5079 | bool can_preserve = FALSE; |
79072805 | 5080 | |
eb85dfd3 DM |
5081 | if (localizing) { |
5082 | MAGIC *mg; | |
5083 | HV *stash; | |
5084 | ||
2c5f48c2 | 5085 | if (SvCANEXISTDELETE(hv)) |
d30e492c | 5086 | can_preserve = TRUE; |
eb85dfd3 DM |
5087 | } |
5088 | ||
6d822dc4 | 5089 | while (++MARK <= SP) { |
1b6737cc | 5090 | SV * const keysv = *MARK; |
6d822dc4 MS |
5091 | SV **svp; |
5092 | HE *he; | |
d30e492c VP |
5093 | bool preeminent = TRUE; |
5094 | ||
5095 | if (localizing && can_preserve) { | |
5096 | /* If we can determine whether the element exist, | |
5097 | * try to preserve the existenceness of a tied hash | |
5098 | * element by using EXISTS and DELETE if possible. | |
5099 | * Fallback to FETCH and STORE otherwise. */ | |
5100 | preeminent = hv_exists_ent(hv, keysv, 0); | |
6d822dc4 | 5101 | } |
eb85dfd3 | 5102 | |
6d822dc4 | 5103 | he = hv_fetch_ent(hv, keysv, lval, 0); |
fe5bfecd | 5104 | svp = he ? &HeVAL(he) : NULL; |
eb85dfd3 | 5105 | |
6d822dc4 | 5106 | if (lval) { |
746f6409 | 5107 | if (!svp || !*svp || *svp == &PL_sv_undef) { |
be2597df | 5108 | DIE(aTHX_ PL_no_helem_sv, SVfARG(keysv)); |
6d822dc4 MS |
5109 | } |
5110 | if (localizing) { | |
7a2e501a | 5111 | if (HvNAME_get(hv) && isGV(*svp)) |
159b6efe | 5112 | save_gp(MUTABLE_GV(*svp), !(PL_op->op_flags & OPf_SPECIAL)); |
47cfc530 VP |
5113 | else if (preeminent) |
5114 | save_helem_flags(hv, keysv, svp, | |
5115 | (PL_op->op_flags & OPf_SPECIAL) ? 0 : SAVEf_SETMAGIC); | |
5116 | else | |
5117 | SAVEHDELETE(hv, keysv); | |
6d822dc4 MS |
5118 | } |
5119 | } | |
746f6409 | 5120 | *MARK = svp && *svp ? *svp : &PL_sv_undef; |
79072805 | 5121 | } |
82334630 | 5122 | if (GIMME_V != G_ARRAY) { |
a0d0e21e | 5123 | MARK = ORIGMARK; |
04ab2c87 | 5124 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
a0d0e21e | 5125 | SP = MARK; |
79072805 | 5126 | } |
a0d0e21e LW |
5127 | RETURN; |
5128 | } | |
5129 | ||
5cae3edb RZ |
5130 | PP(pp_kvhslice) |
5131 | { | |
20b7effb | 5132 | dSP; dMARK; |
5cae3edb RZ |
5133 | HV * const hv = MUTABLE_HV(POPs); |
5134 | I32 lval = (PL_op->op_flags & OPf_MOD); | |
adad97db | 5135 | SSize_t items = SP - MARK; |
5cae3edb RZ |
5136 | |
5137 | if (PL_op->op_private & OPpMAYBE_LVSUB) { | |
5138 | const I32 flags = is_lvalue_sub(); | |
5139 | if (flags) { | |
5140 | if (!(flags & OPpENTERSUB_INARGS)) | |
7aae0299 | 5141 | /* diag_listed_as: Can't modify %s in %s */ |
cc5f9b8a FC |
5142 | Perl_croak(aTHX_ "Can't modify key/value hash slice in %s assignment", |
5143 | GIMME_V == G_ARRAY ? "list" : "scalar"); | |
5cae3edb RZ |
5144 | lval = flags; |
5145 | } | |
5146 | } | |
5147 | ||
5148 | MEXTEND(SP,items); | |
5149 | while (items > 1) { | |
5150 | *(MARK+items*2-1) = *(MARK+items); | |
5151 | items--; | |
5152 | } | |
5153 | items = SP-MARK; | |
5154 | SP += items; | |
5155 | ||
5156 | while (++MARK <= SP) { | |
5157 | SV * const keysv = *MARK; | |
5158 | SV **svp; | |
5159 | HE *he; | |
5160 | ||
5161 | he = hv_fetch_ent(hv, keysv, lval, 0); | |
5162 | svp = he ? &HeVAL(he) : NULL; | |
5163 | ||
5164 | if (lval) { | |
5165 | if (!svp || !*svp || *svp == &PL_sv_undef) { | |
5166 | DIE(aTHX_ PL_no_helem_sv, SVfARG(keysv)); | |
5167 | } | |
5168 | *MARK = sv_mortalcopy(*MARK); | |
5169 | } | |
5170 | *++MARK = svp && *svp ? *svp : &PL_sv_undef; | |
5171 | } | |
82334630 | 5172 | if (GIMME_V != G_ARRAY) { |
5cae3edb RZ |
5173 | MARK = SP - items*2; |
5174 | *++MARK = items > 0 ? *SP : &PL_sv_undef; | |
5175 | SP = MARK; | |
5176 | } | |
5177 | RETURN; | |
5178 | } | |
5179 | ||
a0d0e21e LW |
5180 | /* List operators. */ |
5181 | ||
5182 | PP(pp_list) | |
5183 | { | |
4fa715fa | 5184 | I32 markidx = POPMARK; |
82334630 | 5185 | if (GIMME_V != G_ARRAY) { |
4fa715fa DD |
5186 | SV **mark = PL_stack_base + markidx; |
5187 | dSP; | |
a0d0e21e LW |
5188 | if (++MARK <= SP) |
5189 | *MARK = *SP; /* unwanted list, return last item */ | |
8990e307 | 5190 | else |
3280af22 | 5191 | *MARK = &PL_sv_undef; |
a0d0e21e | 5192 | SP = MARK; |
4fa715fa | 5193 | PUTBACK; |
79072805 | 5194 | } |
4fa715fa | 5195 | return NORMAL; |
79072805 LW |
5196 | } |
5197 | ||
a0d0e21e | 5198 | PP(pp_lslice) |
79072805 | 5199 | { |
39644a26 | 5200 | dSP; |
1b6737cc AL |
5201 | SV ** const lastrelem = PL_stack_sp; |
5202 | SV ** const lastlelem = PL_stack_base + POPMARK; | |
5203 | SV ** const firstlelem = PL_stack_base + POPMARK + 1; | |
eb578fdb | 5204 | SV ** const firstrelem = lastlelem + 1; |
706a6ebc | 5205 | const U8 mod = PL_op->op_flags & OPf_MOD; |
1b6737cc | 5206 | |
eb578fdb KW |
5207 | const I32 max = lastrelem - lastlelem; |
5208 | SV **lelem; | |
a0d0e21e | 5209 | |
82334630 | 5210 | if (GIMME_V != G_ARRAY) { |
9e59c36b TC |
5211 | if (lastlelem < firstlelem) { |
5212 | *firstlelem = &PL_sv_undef; | |
5213 | } | |
5214 | else { | |
5215 | I32 ix = SvIV(*lastlelem); | |
5216 | if (ix < 0) | |
5217 | ix += max; | |
5218 | if (ix < 0 || ix >= max) | |
5219 | *firstlelem = &PL_sv_undef; | |
5220 | else | |
5221 | *firstlelem = firstrelem[ix]; | |
5222 | } | |
5223 | SP = firstlelem; | |
5224 | RETURN; | |
a0d0e21e LW |
5225 | } |
5226 | ||
5227 | if (max == 0) { | |
5228 | SP = firstlelem - 1; | |
5229 | RETURN; | |
5230 | } | |
5231 | ||
5232 | for (lelem = firstlelem; lelem <= lastlelem; lelem++) { | |
4ea561bc | 5233 | I32 ix = SvIV(*lelem); |
c73bf8e3 | 5234 | if (ix < 0) |
a0d0e21e | 5235 | ix += max; |
c73bf8e3 HS |
5236 | if (ix < 0 || ix >= max) |
5237 | *lelem = &PL_sv_undef; | |
5238 | else { | |
c73bf8e3 | 5239 | if (!(*lelem = firstrelem[ix])) |
3280af22 | 5240 | *lelem = &PL_sv_undef; |
60779a30 | 5241 | else if (mod && SvPADTMP(*lelem)) { |
706a6ebc | 5242 | *lelem = firstrelem[ix] = sv_mortalcopy(*lelem); |
60779a30 | 5243 | } |
748a9306 | 5244 | } |
79072805 | 5245 | } |
cbce292e | 5246 | SP = lastlelem; |
79072805 LW |
5247 | RETURN; |
5248 | } | |
5249 | ||
a0d0e21e LW |
5250 | PP(pp_anonlist) |
5251 | { | |
20b7effb | 5252 | dSP; dMARK; |
1b6737cc | 5253 | const I32 items = SP - MARK; |
ad64d0ec | 5254 | SV * const av = MUTABLE_SV(av_make(items, MARK+1)); |
31476221 | 5255 | SP = MARK; |
6e449a3a MHM |
5256 | mXPUSHs((PL_op->op_flags & OPf_SPECIAL) |
5257 | ? newRV_noinc(av) : av); | |
a0d0e21e LW |
5258 | RETURN; |
5259 | } | |
5260 | ||
5261 | PP(pp_anonhash) | |
79072805 | 5262 | { |
20b7effb | 5263 | dSP; dMARK; dORIGMARK; |
67e67fd7 | 5264 | HV* const hv = newHV(); |
8d455b9f | 5265 | SV* const retval = sv_2mortal( PL_op->op_flags & OPf_SPECIAL |
67e67fd7 | 5266 | ? newRV_noinc(MUTABLE_SV(hv)) |
8d455b9f | 5267 | : MUTABLE_SV(hv) ); |
a0d0e21e LW |
5268 | |
5269 | while (MARK < SP) { | |
3ed356df FC |
5270 | SV * const key = |
5271 | (MARK++, SvGMAGICAL(*MARK) ? sv_mortalcopy(*MARK) : *MARK); | |
5272 | SV *val; | |
a0d0e21e | 5273 | if (MARK < SP) |
3ed356df FC |
5274 | { |
5275 | MARK++; | |
5276 | SvGETMAGIC(*MARK); | |
5277 | val = newSV(0); | |
d187b712 | 5278 | sv_setsv_nomg(val, *MARK); |
3ed356df | 5279 | } |
a2a5de95 | 5280 | else |
3ed356df | 5281 | { |
a2a5de95 | 5282 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "Odd number of elements in anonymous hash"); |
3ed356df FC |
5283 | val = newSV(0); |
5284 | } | |
f12c7020 | 5285 | (void)hv_store_ent(hv,key,val,0); |
79072805 | 5286 | } |
a0d0e21e | 5287 | SP = ORIGMARK; |
8d455b9f | 5288 | XPUSHs(retval); |
79072805 LW |
5289 | RETURN; |
5290 | } | |
5291 | ||
a0d0e21e | 5292 | PP(pp_splice) |
79072805 | 5293 | { |
20b7effb | 5294 | dSP; dMARK; dORIGMARK; |
5cd408a2 | 5295 | int num_args = (SP - MARK); |
00576728 | 5296 | AV *ary = MUTABLE_AV(*++MARK); |
eb578fdb KW |
5297 | SV **src; |
5298 | SV **dst; | |
c70927a6 FC |
5299 | SSize_t i; |
5300 | SSize_t offset; | |
5301 | SSize_t length; | |
5302 | SSize_t newlen; | |
5303 | SSize_t after; | |
5304 | SSize_t diff; | |
ad64d0ec | 5305 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
93965878 | 5306 | |
1b6737cc | 5307 | if (mg) { |
3e0cb5de | 5308 | return Perl_tied_method(aTHX_ SV_CONST(SPLICE), mark - 1, MUTABLE_SV(ary), mg, |
af71faff NC |
5309 | GIMME_V | TIED_METHOD_ARGUMENTS_ON_STACK, |
5310 | sp - mark); | |
93965878 | 5311 | } |
79072805 | 5312 | |
a0d0e21e | 5313 | SP++; |
79072805 | 5314 | |
a0d0e21e | 5315 | if (++MARK < SP) { |
4ea561bc | 5316 | offset = i = SvIV(*MARK); |
a0d0e21e | 5317 | if (offset < 0) |
93965878 | 5318 | offset += AvFILLp(ary) + 1; |
84902520 | 5319 | if (offset < 0) |
cea2e8a9 | 5320 | DIE(aTHX_ PL_no_aelem, i); |
a0d0e21e LW |
5321 | if (++MARK < SP) { |
5322 | length = SvIVx(*MARK++); | |
48cdf507 GA |
5323 | if (length < 0) { |
5324 | length += AvFILLp(ary) - offset + 1; | |
5325 | if (length < 0) | |
5326 | length = 0; | |
5327 | } | |
79072805 LW |
5328 | } |
5329 | else | |
a0d0e21e | 5330 | length = AvMAX(ary) + 1; /* close enough to infinity */ |
79072805 | 5331 | } |
a0d0e21e LW |
5332 | else { |
5333 | offset = 0; | |
5334 | length = AvMAX(ary) + 1; | |
5335 | } | |
8cbc2e3b | 5336 | if (offset > AvFILLp(ary) + 1) { |
5cd408a2 EB |
5337 | if (num_args > 2) |
5338 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "splice() offset past end of array" ); | |
93965878 | 5339 | offset = AvFILLp(ary) + 1; |
8cbc2e3b | 5340 | } |
93965878 | 5341 | after = AvFILLp(ary) + 1 - (offset + length); |
a0d0e21e LW |
5342 | if (after < 0) { /* not that much array */ |
5343 | length += after; /* offset+length now in array */ | |
5344 | after = 0; | |
5345 | if (!AvALLOC(ary)) | |
5346 | av_extend(ary, 0); | |
5347 | } | |
5348 | ||
5349 | /* At this point, MARK .. SP-1 is our new LIST */ | |
5350 | ||
5351 | newlen = SP - MARK; | |
5352 | diff = newlen - length; | |
13d7cbc1 GS |
5353 | if (newlen && !AvREAL(ary) && AvREIFY(ary)) |
5354 | av_reify(ary); | |
a0d0e21e | 5355 | |
50528de0 WL |
5356 | /* make new elements SVs now: avoid problems if they're from the array */ |
5357 | for (dst = MARK, i = newlen; i; i--) { | |
1b6737cc | 5358 | SV * const h = *dst; |
f2b990bf | 5359 | *dst++ = newSVsv(h); |
50528de0 WL |
5360 | } |
5361 | ||
a0d0e21e | 5362 | if (diff < 0) { /* shrinking the area */ |
95b63a38 | 5363 | SV **tmparyval = NULL; |
a0d0e21e | 5364 | if (newlen) { |
a02a5408 | 5365 | Newx(tmparyval, newlen, SV*); /* so remember insertion */ |
a0d0e21e | 5366 | Copy(MARK, tmparyval, newlen, SV*); |
79072805 | 5367 | } |
a0d0e21e LW |
5368 | |
5369 | MARK = ORIGMARK + 1; | |
82334630 | 5370 | if (GIMME_V == G_ARRAY) { /* copy return vals to stack */ |
31c61add | 5371 | const bool real = cBOOL(AvREAL(ary)); |
a0d0e21e | 5372 | MEXTEND(MARK, length); |
31c61add | 5373 | if (real) |
bbce6d69 | 5374 | EXTEND_MORTAL(length); |
31c61add FC |
5375 | for (i = 0, dst = MARK; i < length; i++) { |
5376 | if ((*dst = AvARRAY(ary)[i+offset])) { | |
5377 | if (real) | |
486ec47a | 5378 | sv_2mortal(*dst); /* free them eventually */ |
36477c24 | 5379 | } |
31c61add FC |
5380 | else |
5381 | *dst = &PL_sv_undef; | |
5382 | dst++; | |
a0d0e21e LW |
5383 | } |
5384 | MARK += length - 1; | |
79072805 | 5385 | } |
a0d0e21e LW |
5386 | else { |
5387 | *MARK = AvARRAY(ary)[offset+length-1]; | |
5388 | if (AvREAL(ary)) { | |
d689ffdd | 5389 | sv_2mortal(*MARK); |
a0d0e21e LW |
5390 | for (i = length - 1, dst = &AvARRAY(ary)[offset]; i > 0; i--) |
5391 | SvREFCNT_dec(*dst++); /* free them now */ | |
79072805 | 5392 | } |
92b69f65 FC |
5393 | if (!*MARK) |
5394 | *MARK = &PL_sv_undef; | |
a0d0e21e | 5395 | } |
93965878 | 5396 | AvFILLp(ary) += diff; |
a0d0e21e LW |
5397 | |
5398 | /* pull up or down? */ | |
5399 | ||
5400 | if (offset < after) { /* easier to pull up */ | |
5401 | if (offset) { /* esp. if nothing to pull */ | |
5402 | src = &AvARRAY(ary)[offset-1]; | |
5403 | dst = src - diff; /* diff is negative */ | |
5404 | for (i = offset; i > 0; i--) /* can't trust Copy */ | |
5405 | *dst-- = *src--; | |
79072805 | 5406 | } |
a0d0e21e | 5407 | dst = AvARRAY(ary); |
9c6bc640 | 5408 | AvARRAY(ary) = AvARRAY(ary) - diff; /* diff is negative */ |
a0d0e21e LW |
5409 | AvMAX(ary) += diff; |
5410 | } | |
5411 | else { | |
5412 | if (after) { /* anything to pull down? */ | |
5413 | src = AvARRAY(ary) + offset + length; | |
5414 | dst = src + diff; /* diff is negative */ | |
5415 | Move(src, dst, after, SV*); | |
79072805 | 5416 | } |
93965878 | 5417 | dst = &AvARRAY(ary)[AvFILLp(ary)+1]; |
a0d0e21e LW |
5418 | /* avoid later double free */ |
5419 | } | |
5420 | i = -diff; | |
5421 | while (i) | |
ce0d59fd | 5422 | dst[--i] = NULL; |
a0d0e21e LW |
5423 | |
5424 | if (newlen) { | |
50528de0 | 5425 | Copy( tmparyval, AvARRAY(ary) + offset, newlen, SV* ); |
a0d0e21e LW |
5426 | Safefree(tmparyval); |
5427 | } | |
5428 | } | |
5429 | else { /* no, expanding (or same) */ | |
d3961450 | 5430 | SV** tmparyval = NULL; |
a0d0e21e | 5431 | if (length) { |
a02a5408 | 5432 | Newx(tmparyval, length, SV*); /* so remember deletion */ |
a0d0e21e LW |
5433 | Copy(AvARRAY(ary)+offset, tmparyval, length, SV*); |
5434 | } | |
5435 | ||
5436 | if (diff > 0) { /* expanding */ | |
a0d0e21e | 5437 | /* push up or down? */ |
a0d0e21e LW |
5438 | if (offset < after && diff <= AvARRAY(ary) - AvALLOC(ary)) { |
5439 | if (offset) { | |
5440 | src = AvARRAY(ary); | |
5441 | dst = src - diff; | |
5442 | Move(src, dst, offset, SV*); | |
79072805 | 5443 | } |
9c6bc640 | 5444 | AvARRAY(ary) = AvARRAY(ary) - diff;/* diff is positive */ |
a0d0e21e | 5445 | AvMAX(ary) += diff; |
93965878 | 5446 | AvFILLp(ary) += diff; |
79072805 LW |
5447 | } |
5448 | else { | |
93965878 NIS |
5449 | if (AvFILLp(ary) + diff >= AvMAX(ary)) /* oh, well */ |
5450 | av_extend(ary, AvFILLp(ary) + diff); | |
5451 | AvFILLp(ary) += diff; | |
a0d0e21e LW |
5452 | |
5453 | if (after) { | |
93965878 | 5454 | dst = AvARRAY(ary) + AvFILLp(ary); |
a0d0e21e LW |
5455 | src = dst - diff; |
5456 | for (i = after; i; i--) { | |
5457 | *dst-- = *src--; | |
5458 | } | |
79072805 LW |
5459 | } |
5460 | } | |
a0d0e21e LW |
5461 | } |
5462 | ||
50528de0 WL |
5463 | if (newlen) { |
5464 | Copy( MARK, AvARRAY(ary) + offset, newlen, SV* ); | |
a0d0e21e | 5465 | } |
50528de0 | 5466 | |
a0d0e21e | 5467 | MARK = ORIGMARK + 1; |
82334630 | 5468 | if (GIMME_V == G_ARRAY) { /* copy return vals to stack */ |
a0d0e21e | 5469 | if (length) { |
31c61add FC |
5470 | const bool real = cBOOL(AvREAL(ary)); |
5471 | if (real) | |
bbce6d69 | 5472 | EXTEND_MORTAL(length); |
31c61add FC |
5473 | for (i = 0, dst = MARK; i < length; i++) { |
5474 | if ((*dst = tmparyval[i])) { | |
5475 | if (real) | |
486ec47a | 5476 | sv_2mortal(*dst); /* free them eventually */ |
36477c24 | 5477 | } |
31c61add FC |
5478 | else *dst = &PL_sv_undef; |
5479 | dst++; | |
79072805 LW |
5480 | } |
5481 | } | |
a0d0e21e LW |
5482 | MARK += length - 1; |
5483 | } | |
5484 | else if (length--) { | |
5485 | *MARK = tmparyval[length]; | |
5486 | if (AvREAL(ary)) { | |
d689ffdd | 5487 | sv_2mortal(*MARK); |
a0d0e21e LW |
5488 | while (length-- > 0) |
5489 | SvREFCNT_dec(tmparyval[length]); | |
79072805 | 5490 | } |
92b69f65 FC |
5491 | if (!*MARK) |
5492 | *MARK = &PL_sv_undef; | |
79072805 | 5493 | } |
a0d0e21e | 5494 | else |
3280af22 | 5495 | *MARK = &PL_sv_undef; |
d3961450 | 5496 | Safefree(tmparyval); |
79072805 | 5497 | } |
474af990 FR |
5498 | |
5499 | if (SvMAGICAL(ary)) | |
5500 | mg_set(MUTABLE_SV(ary)); | |
5501 | ||
a0d0e21e | 5502 | SP = MARK; |
79072805 LW |
5503 | RETURN; |
5504 | } | |
5505 | ||
a0d0e21e | 5506 | PP(pp_push) |
79072805 | 5507 | { |
20b7effb | 5508 | dSP; dMARK; dORIGMARK; dTARGET; |
00576728 | 5509 | AV * const ary = MUTABLE_AV(*++MARK); |
ad64d0ec | 5510 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
79072805 | 5511 | |
1b6737cc | 5512 | if (mg) { |
ad64d0ec | 5513 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
93965878 NIS |
5514 | PUSHMARK(MARK); |
5515 | PUTBACK; | |
d343c3ef | 5516 | ENTER_with_name("call_PUSH"); |
3e0cb5de | 5517 | call_sv(SV_CONST(PUSH),G_SCALAR|G_DISCARD|G_METHOD_NAMED); |
d343c3ef | 5518 | LEAVE_with_name("call_PUSH"); |
01072573 | 5519 | /* SPAGAIN; not needed: SP is assigned to immediately below */ |
93965878 | 5520 | } |
a60c0954 | 5521 | else { |
a68090fe DM |
5522 | /* PL_delaymagic is restored by JUMPENV_POP on dieing, so we |
5523 | * only need to save locally, not on the save stack */ | |
5524 | U16 old_delaymagic = PL_delaymagic; | |
5525 | ||
cb077ed2 | 5526 | if (SvREADONLY(ary) && MARK < SP) Perl_croak_no_modify(); |
89c14e2e | 5527 | PL_delaymagic = DM_DELAY; |
a60c0954 | 5528 | for (++MARK; MARK <= SP; MARK++) { |
3ed356df FC |
5529 | SV *sv; |
5530 | if (*MARK) SvGETMAGIC(*MARK); | |
5531 | sv = newSV(0); | |
a60c0954 | 5532 | if (*MARK) |
3ed356df | 5533 | sv_setsv_nomg(sv, *MARK); |
0a75904b | 5534 | av_store(ary, AvFILLp(ary)+1, sv); |
a60c0954 | 5535 | } |
354b0578 | 5536 | if (PL_delaymagic & DM_ARRAY_ISA) |
ad64d0ec | 5537 | mg_set(MUTABLE_SV(ary)); |
a68090fe | 5538 | PL_delaymagic = old_delaymagic; |
6eeabd23 VP |
5539 | } |
5540 | SP = ORIGMARK; | |
5541 | if (OP_GIMME(PL_op, 0) != G_VOID) { | |
5542 | PUSHi( AvFILL(ary) + 1 ); | |
79072805 | 5543 | } |
79072805 LW |
5544 | RETURN; |
5545 | } | |
5546 | ||
b1c05ba5 | 5547 | /* also used for: pp_pop()*/ |
a0d0e21e | 5548 | PP(pp_shift) |
79072805 | 5549 | { |
39644a26 | 5550 | dSP; |
538f5756 | 5551 | AV * const av = PL_op->op_flags & OPf_SPECIAL |
94f9945d | 5552 | ? MUTABLE_AV(GvAVn(PL_defgv)) : MUTABLE_AV(POPs); |
789b4bc9 | 5553 | SV * const sv = PL_op->op_type == OP_SHIFT ? av_shift(av) : av_pop(av); |
79072805 | 5554 | EXTEND(SP, 1); |
c2b4a044 | 5555 | assert (sv); |
d689ffdd | 5556 | if (AvREAL(av)) |
a0d0e21e LW |
5557 | (void)sv_2mortal(sv); |
5558 | PUSHs(sv); | |
79072805 | 5559 | RETURN; |
79072805 LW |
5560 | } |
5561 | ||
a0d0e21e | 5562 | PP(pp_unshift) |
79072805 | 5563 | { |
20b7effb | 5564 | dSP; dMARK; dORIGMARK; dTARGET; |
00576728 | 5565 | AV *ary = MUTABLE_AV(*++MARK); |
ad64d0ec | 5566 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
93965878 | 5567 | |
1b6737cc | 5568 | if (mg) { |
ad64d0ec | 5569 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
7fd66d9d | 5570 | PUSHMARK(MARK); |
93965878 | 5571 | PUTBACK; |
d343c3ef | 5572 | ENTER_with_name("call_UNSHIFT"); |
36925d9e | 5573 | call_sv(SV_CONST(UNSHIFT),G_SCALAR|G_DISCARD|G_METHOD_NAMED); |
d343c3ef | 5574 | LEAVE_with_name("call_UNSHIFT"); |
01072573 | 5575 | /* SPAGAIN; not needed: SP is assigned to immediately below */ |
93965878 | 5576 | } |
a60c0954 | 5577 | else { |
a68090fe DM |
5578 | /* PL_delaymagic is restored by JUMPENV_POP on dieing, so we |
5579 | * only need to save locally, not on the save stack */ | |
5580 | U16 old_delaymagic = PL_delaymagic; | |
c70927a6 | 5581 | SSize_t i = 0; |
a68090fe | 5582 | |
a60c0954 | 5583 | av_unshift(ary, SP - MARK); |
39539141 | 5584 | PL_delaymagic = DM_DELAY; |
a60c0954 | 5585 | while (MARK < SP) { |
1b6737cc | 5586 | SV * const sv = newSVsv(*++MARK); |
a60c0954 NIS |
5587 | (void)av_store(ary, i++, sv); |
5588 | } | |
39539141 DIM |
5589 | if (PL_delaymagic & DM_ARRAY_ISA) |
5590 | mg_set(MUTABLE_SV(ary)); | |
a68090fe | 5591 | PL_delaymagic = old_delaymagic; |
79072805 | 5592 | } |
a0d0e21e | 5593 | SP = ORIGMARK; |
6eeabd23 | 5594 | if (OP_GIMME(PL_op, 0) != G_VOID) { |
5658d0a9 LR |
5595 | PUSHi( AvFILL(ary) + 1 ); |
5596 | } | |
79072805 | 5597 | RETURN; |
79072805 LW |
5598 | } |
5599 | ||
a0d0e21e | 5600 | PP(pp_reverse) |
79072805 | 5601 | { |
20b7effb | 5602 | dSP; dMARK; |
79072805 | 5603 | |
82334630 | 5604 | if (GIMME_V == G_ARRAY) { |
484c818f VP |
5605 | if (PL_op->op_private & OPpREVERSE_INPLACE) { |
5606 | AV *av; | |
5607 | ||
5608 | /* See pp_sort() */ | |
5609 | assert( MARK+1 == SP && *SP && SvTYPE(*SP) == SVt_PVAV); | |
5610 | (void)POPMARK; /* remove mark associated with ex-OP_AASSIGN */ | |
5611 | av = MUTABLE_AV((*SP)); | |
5612 | /* In-place reversing only happens in void context for the array | |
5613 | * assignment. We don't need to push anything on the stack. */ | |
5614 | SP = MARK; | |
5615 | ||
5616 | if (SvMAGICAL(av)) { | |
c70927a6 | 5617 | SSize_t i, j; |
eb578fdb | 5618 | SV *tmp = sv_newmortal(); |
484c818f VP |
5619 | /* For SvCANEXISTDELETE */ |
5620 | HV *stash; | |
5621 | const MAGIC *mg; | |
5622 | bool can_preserve = SvCANEXISTDELETE(av); | |
5623 | ||
b9f2b683 | 5624 | for (i = 0, j = av_tindex(av); i < j; ++i, --j) { |
eb578fdb | 5625 | SV *begin, *end; |
484c818f VP |
5626 | |
5627 | if (can_preserve) { | |
5628 | if (!av_exists(av, i)) { | |
5629 | if (av_exists(av, j)) { | |
eb578fdb | 5630 | SV *sv = av_delete(av, j, 0); |
484c818f VP |
5631 | begin = *av_fetch(av, i, TRUE); |
5632 | sv_setsv_mg(begin, sv); | |
5633 | } | |
5634 | continue; | |
5635 | } | |
5636 | else if (!av_exists(av, j)) { | |
eb578fdb | 5637 | SV *sv = av_delete(av, i, 0); |
484c818f VP |
5638 | end = *av_fetch(av, j, TRUE); |
5639 | sv_setsv_mg(end, sv); | |
5640 | continue; | |
5641 | } | |
5642 | } | |
5643 | ||
5644 | begin = *av_fetch(av, i, TRUE); | |
5645 | end = *av_fetch(av, j, TRUE); | |
5646 | sv_setsv(tmp, begin); | |
5647 | sv_setsv_mg(begin, end); | |
5648 | sv_setsv_mg(end, tmp); | |
5649 | } | |
5650 | } | |
5651 | else { | |
5652 | SV **begin = AvARRAY(av); | |
484c818f | 5653 | |
95a26d8e VP |
5654 | if (begin) { |
5655 | SV **end = begin + AvFILLp(av); | |
5656 | ||
5657 | while (begin < end) { | |
eb578fdb | 5658 | SV * const tmp = *begin; |
95a26d8e VP |
5659 | *begin++ = *end; |
5660 | *end-- = tmp; | |
5661 | } | |
484c818f VP |
5662 | } |
5663 | } | |
5664 | } | |
5665 | else { | |
5666 | SV **oldsp = SP; | |
5667 | MARK++; | |
5668 | while (MARK < SP) { | |
eb578fdb | 5669 | SV * const tmp = *MARK; |
484c818f VP |
5670 | *MARK++ = *SP; |
5671 | *SP-- = tmp; | |
5672 | } | |
5673 | /* safe as long as stack cannot get extended in the above */ | |
5674 | SP = oldsp; | |
a0d0e21e | 5675 | } |
79072805 LW |
5676 | } |
5677 | else { | |
eb578fdb | 5678 | char *up; |
a0d0e21e LW |
5679 | dTARGET; |
5680 | STRLEN len; | |
79072805 | 5681 | |
7e2040f0 | 5682 | SvUTF8_off(TARG); /* decontaminate */ |
a0d0e21e | 5683 | if (SP - MARK > 1) |
3280af22 | 5684 | do_join(TARG, &PL_sv_no, MARK, SP); |
1e21d011 | 5685 | else { |
64d3d9fe | 5686 | sv_setsv(TARG, SP > MARK ? *SP : DEFSV); |
1e21d011 B |
5687 | } |
5688 | ||
a0d0e21e LW |
5689 | up = SvPV_force(TARG, len); |
5690 | if (len > 1) { | |
19742f39 | 5691 | char *down; |
7e2040f0 | 5692 | if (DO_UTF8(TARG)) { /* first reverse each character */ |
dfe13c55 | 5693 | U8* s = (U8*)SvPVX(TARG); |
349d4f2f | 5694 | const U8* send = (U8*)(s + len); |
a0ed51b3 | 5695 | while (s < send) { |
d742c382 | 5696 | if (UTF8_IS_INVARIANT(*s)) { |
a0ed51b3 LW |
5697 | s++; |
5698 | continue; | |
5699 | } | |
5700 | else { | |
4b88fb76 | 5701 | if (!utf8_to_uvchr_buf(s, send, 0)) |
a0dbb045 | 5702 | break; |
dfe13c55 | 5703 | up = (char*)s; |
a0ed51b3 | 5704 | s += UTF8SKIP(s); |
dfe13c55 | 5705 | down = (char*)(s - 1); |
a0dbb045 | 5706 | /* reverse this character */ |
a0ed51b3 | 5707 | while (down > up) { |
19742f39 | 5708 | const char tmp = *up; |
a0ed51b3 | 5709 | *up++ = *down; |
19742f39 | 5710 | *down-- = tmp; |
a0ed51b3 LW |
5711 | } |
5712 | } | |
5713 | } | |
5714 | up = SvPVX(TARG); | |
5715 | } | |
a0d0e21e LW |
5716 | down = SvPVX(TARG) + len - 1; |
5717 | while (down > up) { | |
19742f39 | 5718 | const char tmp = *up; |
a0d0e21e | 5719 | *up++ = *down; |
19742f39 | 5720 | *down-- = tmp; |
a0d0e21e | 5721 | } |
3aa33fe5 | 5722 | (void)SvPOK_only_UTF8(TARG); |
79072805 | 5723 | } |
a0d0e21e LW |
5724 | SP = MARK + 1; |
5725 | SETTARG; | |
79072805 | 5726 | } |
a0d0e21e | 5727 | RETURN; |
79072805 LW |
5728 | } |
5729 | ||
a0d0e21e | 5730 | PP(pp_split) |
79072805 | 5731 | { |
20b7effb | 5732 | dSP; dTARG; |
692044df DM |
5733 | AV *ary = ( (PL_op->op_private & OPpSPLIT_ASSIGN) /* @a = split */ |
5734 | && (PL_op->op_flags & OPf_STACKED)) /* @{expr} = split */ | |
5012eebe | 5735 | ? (AV *)POPs : NULL; |
eb578fdb | 5736 | IV limit = POPi; /* note, negative is forever */ |
1b6737cc | 5737 | SV * const sv = POPs; |
a0d0e21e | 5738 | STRLEN len; |
eb578fdb | 5739 | const char *s = SvPV_const(sv, len); |
1b6737cc | 5740 | const bool do_utf8 = DO_UTF8(sv); |
727b7506 | 5741 | const char *strend = s + len; |
5012eebe | 5742 | PMOP *pm = cPMOPx(PL_op); |
eb578fdb KW |
5743 | REGEXP *rx; |
5744 | SV *dstr; | |
5745 | const char *m; | |
c70927a6 | 5746 | SSize_t iters = 0; |
d14578b8 KW |
5747 | const STRLEN slen = do_utf8 |
5748 | ? utf8_length((U8*)s, (U8*)strend) | |
5749 | : (STRLEN)(strend - s); | |
c70927a6 | 5750 | SSize_t maxiters = slen + 10; |
c1a7495a | 5751 | I32 trailing_empty = 0; |
727b7506 | 5752 | const char *orig; |
052a7c76 | 5753 | const IV origlimit = limit; |
a0d0e21e LW |
5754 | I32 realarray = 0; |
5755 | I32 base; | |
1c23e2bd | 5756 | const U8 gimme = GIMME_V; |
941446f6 | 5757 | bool gimme_scalar; |
692044df | 5758 | I32 oldsave = PL_savestack_ix; |
437d3b4e | 5759 | U32 make_mortal = SVs_TEMP; |
7fba1cd6 | 5760 | bool multiline = 0; |
b37c2d43 | 5761 | MAGIC *mg = NULL; |
79072805 | 5762 | |
aaa362c4 | 5763 | rx = PM_GETRE(pm); |
bbce6d69 | 5764 | |
a62b1201 | 5765 | TAINT_IF(get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET && |
dbc200c5 | 5766 | (RX_EXTFLAGS(rx) & (RXf_WHITE | RXf_SKIPWHITE))); |
bbce6d69 | 5767 | |
692044df | 5768 | /* handle @ary = split(...) optimisation */ |
5012eebe DM |
5769 | if (PL_op->op_private & OPpSPLIT_ASSIGN) { |
5770 | if (!(PL_op->op_flags & OPf_STACKED)) { | |
692044df DM |
5771 | if (PL_op->op_private & OPpSPLIT_LEX) { |
5772 | if (PL_op->op_private & OPpLVAL_INTRO) | |
5773 | SAVECLEARSV(PAD_SVl(pm->op_pmreplrootu.op_pmtargetoff)); | |
5012eebe | 5774 | ary = (AV *)PAD_SVl(pm->op_pmreplrootu.op_pmtargetoff); |
692044df | 5775 | } |
5012eebe DM |
5776 | else { |
5777 | GV *gv = | |
971a9dd3 | 5778 | #ifdef USE_ITHREADS |
5012eebe | 5779 | MUTABLE_GV(PAD_SVl(pm->op_pmreplrootu.op_pmtargetoff)); |
971a9dd3 | 5780 | #else |
5012eebe | 5781 | pm->op_pmreplrootu.op_pmtargetgv; |
20e98b0f | 5782 | #endif |
692044df DM |
5783 | if (PL_op->op_private & OPpLVAL_INTRO) |
5784 | ary = save_ary(gv); | |
5785 | else | |
5786 | ary = GvAVn(gv); | |
5012eebe | 5787 | } |
692044df DM |
5788 | /* skip anything pushed by OPpLVAL_INTRO above */ |
5789 | oldsave = PL_savestack_ix; | |
5012eebe DM |
5790 | } |
5791 | ||
a0d0e21e | 5792 | realarray = 1; |
8ec5e241 | 5793 | PUTBACK; |
a0d0e21e | 5794 | av_extend(ary,0); |
821956c5 | 5795 | (void)sv_2mortal(SvREFCNT_inc_simple_NN(sv)); |
a0d0e21e | 5796 | av_clear(ary); |
8ec5e241 | 5797 | SPAGAIN; |
ad64d0ec | 5798 | if ((mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied))) { |
8ec5e241 | 5799 | PUSHMARK(SP); |
ad64d0ec | 5800 | XPUSHs(SvTIED_obj(MUTABLE_SV(ary), mg)); |
8ec5e241 NIS |
5801 | } |
5802 | else { | |
1c0b011c | 5803 | if (!AvREAL(ary)) { |
1b6737cc | 5804 | I32 i; |
1c0b011c | 5805 | AvREAL_on(ary); |
abff13bb | 5806 | AvREIFY_off(ary); |
1c0b011c | 5807 | for (i = AvFILLp(ary); i >= 0; i--) |
d14578b8 | 5808 | AvARRAY(ary)[i] = &PL_sv_undef; /* don't free mere refs */ |
1c0b011c NIS |
5809 | } |
5810 | /* temporarily switch stacks */ | |
8b7059b1 | 5811 | SAVESWITCHSTACK(PL_curstack, ary); |
8ec5e241 | 5812 | make_mortal = 0; |
1c0b011c | 5813 | } |
79072805 | 5814 | } |
5012eebe | 5815 | |
3280af22 | 5816 | base = SP - PL_stack_base; |
a0d0e21e | 5817 | orig = s; |
dbc200c5 | 5818 | if (RX_EXTFLAGS(rx) & RXf_SKIPWHITE) { |
613f191e | 5819 | if (do_utf8) { |
d720149d | 5820 | while (s < strend && isSPACE_utf8_safe(s, strend)) |
613f191e TS |
5821 | s += UTF8SKIP(s); |
5822 | } | |
a62b1201 | 5823 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) { |
d720149d | 5824 | while (s < strend && isSPACE_LC(*s)) |
bbce6d69 | 5825 | s++; |
5826 | } | |
5827 | else { | |
d720149d | 5828 | while (s < strend && isSPACE(*s)) |
bbce6d69 | 5829 | s++; |
5830 | } | |
a0d0e21e | 5831 | } |
73134a2e | 5832 | if (RX_EXTFLAGS(rx) & RXf_PMf_MULTILINE) { |
7fba1cd6 | 5833 | multiline = 1; |
c07a80fd | 5834 | } |
5835 | ||
941446f6 FC |
5836 | gimme_scalar = gimme == G_SCALAR && !ary; |
5837 | ||
a0d0e21e LW |
5838 | if (!limit) |
5839 | limit = maxiters + 2; | |
dbc200c5 | 5840 | if (RX_EXTFLAGS(rx) & RXf_WHITE) { |
a0d0e21e | 5841 | while (--limit) { |
bbce6d69 | 5842 | m = s; |
8727f688 YO |
5843 | /* this one uses 'm' and is a negative test */ |
5844 | if (do_utf8) { | |
7a207065 | 5845 | while (m < strend && ! isSPACE_utf8_safe(m, strend) ) { |
613f191e | 5846 | const int t = UTF8SKIP(m); |
7a207065 | 5847 | /* isSPACE_utf8_safe returns FALSE for malform utf8 */ |
613f191e TS |
5848 | if (strend - m < t) |
5849 | m = strend; | |
5850 | else | |
5851 | m += t; | |
5852 | } | |
a62b1201 | 5853 | } |
d14578b8 KW |
5854 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) |
5855 | { | |
8727f688 YO |
5856 | while (m < strend && !isSPACE_LC(*m)) |
5857 | ++m; | |
5858 | } else { | |
5859 | while (m < strend && !isSPACE(*m)) | |
5860 | ++m; | |
5861 | } | |
a0d0e21e LW |
5862 | if (m >= strend) |
5863 | break; | |
bbce6d69 | 5864 | |
c1a7495a BB |
5865 | if (gimme_scalar) { |
5866 | iters++; | |
5867 | if (m-s == 0) | |
5868 | trailing_empty++; | |
5869 | else | |
5870 | trailing_empty = 0; | |
5871 | } else { | |
5872 | dstr = newSVpvn_flags(s, m-s, | |
5873 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5874 | XPUSHs(dstr); | |
5875 | } | |
bbce6d69 | 5876 | |
613f191e TS |
5877 | /* skip the whitespace found last */ |
5878 | if (do_utf8) | |
5879 | s = m + UTF8SKIP(m); | |
5880 | else | |
5881 | s = m + 1; | |
5882 | ||
8727f688 YO |
5883 | /* this one uses 's' and is a positive test */ |
5884 | if (do_utf8) { | |
7a207065 | 5885 | while (s < strend && isSPACE_utf8_safe(s, strend) ) |
8727f688 | 5886 | s += UTF8SKIP(s); |
a62b1201 | 5887 | } |
d14578b8 KW |
5888 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) |
5889 | { | |
8727f688 YO |
5890 | while (s < strend && isSPACE_LC(*s)) |
5891 | ++s; | |
5892 | } else { | |
5893 | while (s < strend && isSPACE(*s)) | |
5894 | ++s; | |
5895 | } | |
79072805 LW |
5896 | } |
5897 | } | |
07bc277f | 5898 | else if (RX_EXTFLAGS(rx) & RXf_START_ONLY) { |
a0d0e21e | 5899 | while (--limit) { |
a6e20a40 AL |
5900 | for (m = s; m < strend && *m != '\n'; m++) |
5901 | ; | |
a0d0e21e LW |
5902 | m++; |
5903 | if (m >= strend) | |
5904 | break; | |
c1a7495a BB |
5905 | |
5906 | if (gimme_scalar) { | |
5907 | iters++; | |
5908 | if (m-s == 0) | |
5909 | trailing_empty++; | |
5910 | else | |
5911 | trailing_empty = 0; | |
5912 | } else { | |
5913 | dstr = newSVpvn_flags(s, m-s, | |
5914 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
5915 | XPUSHs(dstr); | |
5916 | } | |
a0d0e21e LW |
5917 | s = m; |
5918 | } | |
5919 | } | |
07bc277f | 5920 | else if (RX_EXTFLAGS(rx) & RXf_NULL && !(s >= strend)) { |
640f820d AB |
5921 | /* |
5922 | Pre-extend the stack, either the number of bytes or | |
5923 | characters in the string or a limited amount, triggered by: | |
5924 | ||
5925 | my ($x, $y) = split //, $str; | |
5926 | or | |
5927 | split //, $str, $i; | |
5928 | */ | |
c1a7495a | 5929 | if (!gimme_scalar) { |
052a7c76 DM |
5930 | const IV items = limit - 1; |
5931 | /* setting it to -1 will trigger a panic in EXTEND() */ | |
5932 | const SSize_t sslen = slen > SSize_t_MAX ? -1 : (SSize_t)slen; | |
5933 | if (items >=0 && items < sslen) | |
c1a7495a BB |
5934 | EXTEND(SP, items); |
5935 | else | |
052a7c76 | 5936 | EXTEND(SP, sslen); |
c1a7495a | 5937 | } |
640f820d | 5938 | |
e9515b0f AB |
5939 | if (do_utf8) { |
5940 | while (--limit) { | |
5941 | /* keep track of how many bytes we skip over */ | |
5942 | m = s; | |
640f820d | 5943 | s += UTF8SKIP(s); |
c1a7495a BB |
5944 | if (gimme_scalar) { |
5945 | iters++; | |
5946 | if (s-m == 0) | |
5947 | trailing_empty++; | |
5948 | else | |
5949 | trailing_empty = 0; | |
5950 | } else { | |
5951 | dstr = newSVpvn_flags(m, s-m, SVf_UTF8 | make_mortal); | |
640f820d | 5952 | |
c1a7495a BB |
5953 | PUSHs(dstr); |
5954 | } | |
640f820d | 5955 | |
e9515b0f AB |
5956 | if (s >= strend) |
5957 | break; | |
5958 | } | |
5959 | } else { | |
5960 | while (--limit) { | |
c1a7495a BB |
5961 | if (gimme_scalar) { |
5962 | iters++; | |
5963 | } else { | |
5964 | dstr = newSVpvn(s, 1); | |
e9515b0f | 5965 | |
e9515b0f | 5966 | |
c1a7495a BB |
5967 | if (make_mortal) |
5968 | sv_2mortal(dstr); | |
640f820d | 5969 | |
c1a7495a BB |
5970 | PUSHs(dstr); |
5971 | } | |
5972 | ||
5973 | s++; | |
e9515b0f AB |
5974 | |
5975 | if (s >= strend) | |
5976 | break; | |
5977 | } | |
640f820d AB |
5978 | } |
5979 | } | |
3c8556c3 | 5980 | else if (do_utf8 == (RX_UTF8(rx) != 0) && |
07bc277f NC |
5981 | (RX_EXTFLAGS(rx) & RXf_USE_INTUIT) && !RX_NPARENS(rx) |
5982 | && (RX_EXTFLAGS(rx) & RXf_CHECK_ALL) | |
8e1490ee | 5983 | && !(RX_EXTFLAGS(rx) & RXf_IS_ANCHORED)) { |
07bc277f | 5984 | const int tail = (RX_EXTFLAGS(rx) & RXf_INTUIT_TAIL); |
f9f4320a | 5985 | SV * const csv = CALLREG_INTUIT_STRING(rx); |
cf93c79d | 5986 | |
07bc277f | 5987 | len = RX_MINLENRET(rx); |
3c8556c3 | 5988 | if (len == 1 && !RX_UTF8(rx) && !tail) { |
1b6737cc | 5989 | const char c = *SvPV_nolen_const(csv); |
a0d0e21e | 5990 | while (--limit) { |
a6e20a40 AL |
5991 | for (m = s; m < strend && *m != c; m++) |
5992 | ; | |
a0d0e21e LW |
5993 | if (m >= strend) |
5994 | break; | |
c1a7495a BB |
5995 | if (gimme_scalar) { |
5996 | iters++; | |
5997 | if (m-s == 0) | |
5998 | trailing_empty++; | |
5999 | else | |
6000 | trailing_empty = 0; | |
6001 | } else { | |
6002 | dstr = newSVpvn_flags(s, m-s, | |
d14578b8 | 6003 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); |
c1a7495a BB |
6004 | XPUSHs(dstr); |
6005 | } | |
93f04dac JH |
6006 | /* The rx->minlen is in characters but we want to step |
6007 | * s ahead by bytes. */ | |
1aa99e6b IH |
6008 | if (do_utf8) |
6009 | s = (char*)utf8_hop((U8*)m, len); | |
6010 | else | |
6011 | s = m + len; /* Fake \n at the end */ | |
a0d0e21e LW |
6012 | } |
6013 | } | |
6014 | else { | |
a0d0e21e | 6015 | while (s < strend && --limit && |
f722798b | 6016 | (m = fbm_instr((unsigned char*)s, (unsigned char*)strend, |
7fba1cd6 | 6017 | csv, multiline ? FBMrf_MULTILINE : 0)) ) |
a0d0e21e | 6018 | { |
c1a7495a BB |
6019 | if (gimme_scalar) { |
6020 | iters++; | |
6021 | if (m-s == 0) | |
6022 | trailing_empty++; | |
6023 | else | |
6024 | trailing_empty = 0; | |
6025 | } else { | |
6026 | dstr = newSVpvn_flags(s, m-s, | |
d14578b8 | 6027 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); |
c1a7495a BB |
6028 | XPUSHs(dstr); |
6029 | } | |
93f04dac JH |
6030 | /* The rx->minlen is in characters but we want to step |
6031 | * s ahead by bytes. */ | |
1aa99e6b IH |
6032 | if (do_utf8) |
6033 | s = (char*)utf8_hop((U8*)m, len); | |
6034 | else | |
6035 | s = m + len; /* Fake \n at the end */ | |
a0d0e21e | 6036 | } |
463ee0b2 | 6037 | } |
463ee0b2 | 6038 | } |
a0d0e21e | 6039 | else { |
07bc277f | 6040 | maxiters += slen * RX_NPARENS(rx); |
080c2dec | 6041 | while (s < strend && --limit) |
bbce6d69 | 6042 | { |
1b6737cc | 6043 | I32 rex_return; |
080c2dec | 6044 | PUTBACK; |
d14578b8 | 6045 | rex_return = CALLREGEXEC(rx, (char*)s, (char*)strend, (char*)orig, 1, |
c33e64f0 | 6046 | sv, NULL, 0); |
080c2dec | 6047 | SPAGAIN; |
1b6737cc | 6048 | if (rex_return == 0) |
080c2dec | 6049 | break; |
d9f97599 | 6050 | TAINT_IF(RX_MATCH_TAINTED(rx)); |
6502e081 DM |
6051 | /* we never pass the REXEC_COPY_STR flag, so it should |
6052 | * never get copied */ | |
6053 | assert(!RX_MATCH_COPIED(rx)); | |
07bc277f | 6054 | m = RX_OFFS(rx)[0].start + orig; |
c1a7495a BB |
6055 | |
6056 | if (gimme_scalar) { | |
6057 | iters++; | |
6058 | if (m-s == 0) | |
6059 | trailing_empty++; | |
6060 | else | |
6061 | trailing_empty = 0; | |
6062 | } else { | |
6063 | dstr = newSVpvn_flags(s, m-s, | |
6064 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
6065 | XPUSHs(dstr); | |
6066 | } | |
07bc277f | 6067 | if (RX_NPARENS(rx)) { |
1b6737cc | 6068 | I32 i; |
07bc277f NC |
6069 | for (i = 1; i <= (I32)RX_NPARENS(rx); i++) { |
6070 | s = RX_OFFS(rx)[i].start + orig; | |
6071 | m = RX_OFFS(rx)[i].end + orig; | |
6de67870 JP |
6072 | |
6073 | /* japhy (07/27/01) -- the (m && s) test doesn't catch | |
6074 | parens that didn't match -- they should be set to | |
6075 | undef, not the empty string */ | |
c1a7495a BB |
6076 | if (gimme_scalar) { |
6077 | iters++; | |
6078 | if (m-s == 0) | |
6079 | trailing_empty++; | |
6080 | else | |
6081 | trailing_empty = 0; | |
6082 | } else { | |
6083 | if (m >= orig && s >= orig) { | |
6084 | dstr = newSVpvn_flags(s, m-s, | |
6085 | (do_utf8 ? SVf_UTF8 : 0) | |
6086 | | make_mortal); | |
6087 | } | |
6088 | else | |
6089 | dstr = &PL_sv_undef; /* undef, not "" */ | |
6090 | XPUSHs(dstr); | |
748a9306 | 6091 | } |
c1a7495a | 6092 | |
a0d0e21e LW |
6093 | } |
6094 | } | |
07bc277f | 6095 | s = RX_OFFS(rx)[0].end + orig; |
a0d0e21e | 6096 | } |
79072805 | 6097 | } |
8ec5e241 | 6098 | |
c1a7495a BB |
6099 | if (!gimme_scalar) { |
6100 | iters = (SP - PL_stack_base) - base; | |
6101 | } | |
a0d0e21e | 6102 | if (iters > maxiters) |
cea2e8a9 | 6103 | DIE(aTHX_ "Split loop"); |
8ec5e241 | 6104 | |
a0d0e21e LW |
6105 | /* keep field after final delim? */ |
6106 | if (s < strend || (iters && origlimit)) { | |
c1a7495a BB |
6107 | if (!gimme_scalar) { |
6108 | const STRLEN l = strend - s; | |
6109 | dstr = newSVpvn_flags(s, l, (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
6110 | XPUSHs(dstr); | |
6111 | } | |
a0d0e21e | 6112 | iters++; |
79072805 | 6113 | } |
a0d0e21e | 6114 | else if (!origlimit) { |
c1a7495a BB |
6115 | if (gimme_scalar) { |
6116 | iters -= trailing_empty; | |
6117 | } else { | |
6118 | while (iters > 0 && (!TOPs || !SvANY(TOPs) || SvCUR(TOPs) == 0)) { | |
6119 | if (TOPs && !make_mortal) | |
6120 | sv_2mortal(TOPs); | |
71ca73e5 | 6121 | *SP-- = NULL; |
c1a7495a BB |
6122 | iters--; |
6123 | } | |
89900bd3 | 6124 | } |
a0d0e21e | 6125 | } |
8ec5e241 | 6126 | |
8b7059b1 DM |
6127 | PUTBACK; |
6128 | LEAVE_SCOPE(oldsave); /* may undo an earlier SWITCHSTACK */ | |
6129 | SPAGAIN; | |
a0d0e21e | 6130 | if (realarray) { |
8ec5e241 | 6131 | if (!mg) { |
1c0b011c NIS |
6132 | if (SvSMAGICAL(ary)) { |
6133 | PUTBACK; | |
ad64d0ec | 6134 | mg_set(MUTABLE_SV(ary)); |
1c0b011c NIS |
6135 | SPAGAIN; |
6136 | } | |
6137 | if (gimme == G_ARRAY) { | |
6138 | EXTEND(SP, iters); | |
6139 | Copy(AvARRAY(ary), SP + 1, iters, SV*); | |
6140 | SP += iters; | |
6141 | RETURN; | |
6142 | } | |
8ec5e241 | 6143 | } |
1c0b011c | 6144 | else { |
fb73857a | 6145 | PUTBACK; |
d343c3ef | 6146 | ENTER_with_name("call_PUSH"); |
36925d9e | 6147 | call_sv(SV_CONST(PUSH),G_SCALAR|G_DISCARD|G_METHOD_NAMED); |
d343c3ef | 6148 | LEAVE_with_name("call_PUSH"); |
fb73857a | 6149 | SPAGAIN; |
8ec5e241 | 6150 | if (gimme == G_ARRAY) { |
c70927a6 | 6151 | SSize_t i; |
8ec5e241 NIS |
6152 | /* EXTEND should not be needed - we just popped them */ |
6153 | EXTEND(SP, iters); | |
6154 | for (i=0; i < iters; i++) { | |
6155 | SV **svp = av_fetch(ary, i, FALSE); | |
3280af22 | 6156 | PUSHs((svp) ? *svp : &PL_sv_undef); |
8ec5e241 | 6157 | } |
1c0b011c NIS |
6158 | RETURN; |
6159 | } | |
a0d0e21e LW |
6160 | } |
6161 | } | |
6162 | else { | |
6163 | if (gimme == G_ARRAY) | |
6164 | RETURN; | |
6165 | } | |
7f18b612 YST |
6166 | |
6167 | GETTARGET; | |
02c161ef | 6168 | XPUSHi(iters); |
7f18b612 | 6169 | RETURN; |
79072805 | 6170 | } |
85e6fe83 | 6171 | |
c5917253 NC |
6172 | PP(pp_once) |
6173 | { | |
6174 | dSP; | |
6175 | SV *const sv = PAD_SVl(PL_op->op_targ); | |
6176 | ||
6177 | if (SvPADSTALE(sv)) { | |
6178 | /* First time. */ | |
6179 | SvPADSTALE_off(sv); | |
6180 | RETURNOP(cLOGOP->op_other); | |
6181 | } | |
6182 | RETURNOP(cLOGOP->op_next); | |
6183 | } | |
6184 | ||
c0329465 MB |
6185 | PP(pp_lock) |
6186 | { | |
39644a26 | 6187 | dSP; |
c0329465 | 6188 | dTOPss; |
e55aaa0e | 6189 | SV *retsv = sv; |
68795e93 | 6190 | SvLOCK(sv); |
f79aa60b FC |
6191 | if (SvTYPE(retsv) == SVt_PVAV || SvTYPE(retsv) == SVt_PVHV |
6192 | || SvTYPE(retsv) == SVt_PVCV) { | |
e55aaa0e MB |
6193 | retsv = refto(retsv); |
6194 | } | |
6195 | SETs(retsv); | |
c0329465 MB |
6196 | RETURN; |
6197 | } | |
a863c7d1 | 6198 | |
65bca31a | 6199 | |
10088f56 | 6200 | /* used for: pp_padany(), pp_custom(); plus any system ops |
b1c05ba5 DM |
6201 | * that aren't implemented on a particular platform */ |
6202 | ||
65bca31a NC |
6203 | PP(unimplemented_op) |
6204 | { | |
361ed549 NC |
6205 | const Optype op_type = PL_op->op_type; |
6206 | /* Using OP_NAME() isn't going to be helpful here. Firstly, it doesn't cope | |
6207 | with out of range op numbers - it only "special" cases op_custom. | |
6208 | Secondly, as the three ops we "panic" on are padmy, mapstart and custom, | |
6209 | if we get here for a custom op then that means that the custom op didn't | |
6210 | have an implementation. Given that OP_NAME() looks up the custom op | |
6211 | by its pp_addr, likely it will return NULL, unless someone (unhelpfully) | |
6212 | registers &PL_unimplemented_op as the address of their custom op. | |
6213 | NULL doesn't generate a useful error message. "custom" does. */ | |
6214 | const char *const name = op_type >= OP_max | |
6215 | ? "[out of range]" : PL_op_name[PL_op->op_type]; | |
7627e6d0 NC |
6216 | if(OP_IS_SOCKET(op_type)) |
6217 | DIE(aTHX_ PL_no_sock_func, name); | |
361ed549 | 6218 | DIE(aTHX_ "panic: unimplemented op %s (#%d) called", name, op_type); |
65bca31a NC |
6219 | } |
6220 | ||
bea284c8 FC |
6221 | static void |
6222 | S_maybe_unwind_defav(pTHX) | |
6223 | { | |
6224 | if (CX_CUR()->cx_type & CXp_HASARGS) { | |
6225 | PERL_CONTEXT *cx = CX_CUR(); | |
6226 | ||
6227 | assert(CxHASARGS(cx)); | |
6228 | cx_popsub_args(cx); | |
6229 | cx->cx_type &= ~CXp_HASARGS; | |
6230 | } | |
6231 | } | |
6232 | ||
deb8a388 FC |
6233 | /* For sorting out arguments passed to a &CORE:: subroutine */ |
6234 | PP(pp_coreargs) | |
6235 | { | |
6236 | dSP; | |
7fa5bd9b | 6237 | int opnum = SvIOK(cSVOP_sv) ? (int)SvUV(cSVOP_sv) : 0; |
498a02d8 | 6238 | int defgv = PL_opargs[opnum] & OA_DEFGV ||opnum==OP_GLOB, whicharg = 0; |
7fa5bd9b | 6239 | AV * const at_ = GvAV(PL_defgv); |
0e80230d FC |
6240 | SV **svp = at_ ? AvARRAY(at_) : NULL; |
6241 | I32 minargs = 0, maxargs = 0, numargs = at_ ? AvFILLp(at_)+1 : 0; | |
7fa5bd9b | 6242 | I32 oa = opnum ? PL_opargs[opnum] >> OASHIFT : 0; |
46e00a91 | 6243 | bool seen_question = 0; |
7fa5bd9b | 6244 | const char *err = NULL; |
3e6568b4 | 6245 | const bool pushmark = PL_op->op_private & OPpCOREARGS_PUSHMARK; |
7fa5bd9b | 6246 | |
46e00a91 FC |
6247 | /* Count how many args there are first, to get some idea how far to |
6248 | extend the stack. */ | |
7fa5bd9b | 6249 | while (oa) { |
bf0571fd | 6250 | if ((oa & 7) == OA_LIST) { maxargs = I32_MAX; break; } |
7fa5bd9b | 6251 | maxargs++; |
46e00a91 FC |
6252 | if (oa & OA_OPTIONAL) seen_question = 1; |
6253 | if (!seen_question) minargs++; | |
7fa5bd9b FC |
6254 | oa >>= 4; |
6255 | } | |
6256 | ||
6257 | if(numargs < minargs) err = "Not enough"; | |
6258 | else if(numargs > maxargs) err = "Too many"; | |
6259 | if (err) | |
6260 | /* diag_listed_as: Too many arguments for %s */ | |
6261 | Perl_croak(aTHX_ | |
6262 | "%s arguments for %s", err, | |
2a90c7c6 | 6263 | opnum ? PL_op_desc[opnum] : SvPV_nolen_const(cSVOP_sv) |
7fa5bd9b FC |
6264 | ); |
6265 | ||
6266 | /* Reset the stack pointer. Without this, we end up returning our own | |
6267 | arguments in list context, in addition to the values we are supposed | |
6268 | to return. nextstate usually does this on sub entry, but we need | |
e1fa07e3 | 6269 | to run the next op with the caller's hints, so we cannot have a |
7fa5bd9b | 6270 | nextstate. */ |
4ebe6e95 | 6271 | SP = PL_stack_base + CX_CUR()->blk_oldsp; |
7fa5bd9b | 6272 | |
46e00a91 FC |
6273 | if(!maxargs) RETURN; |
6274 | ||
bf0571fd FC |
6275 | /* We do this here, rather than with a separate pushmark op, as it has |
6276 | to come in between two things this function does (stack reset and | |
6277 | arg pushing). This seems the easiest way to do it. */ | |
3e6568b4 | 6278 | if (pushmark) { |
bf0571fd FC |
6279 | PUTBACK; |
6280 | (void)Perl_pp_pushmark(aTHX); | |
6281 | } | |
6282 | ||
6283 | EXTEND(SP, maxargs == I32_MAX ? numargs : maxargs); | |
c931b036 | 6284 | PUTBACK; /* The code below can die in various places. */ |
46e00a91 FC |
6285 | |
6286 | oa = PL_opargs[opnum] >> OASHIFT; | |
3e6568b4 | 6287 | for (; oa&&(numargs||!pushmark); (void)(numargs&&(++svp,--numargs))) { |
c931b036 | 6288 | whicharg++; |
46e00a91 FC |
6289 | switch (oa & 7) { |
6290 | case OA_SCALAR: | |
1efec5ed | 6291 | try_defsv: |
d6d78e19 | 6292 | if (!numargs && defgv && whicharg == minargs + 1) { |
195eefec | 6293 | PUSHs(DEFSV); |
d6d78e19 FC |
6294 | } |
6295 | else PUSHs(numargs ? svp && *svp ? *svp : &PL_sv_undef : NULL); | |
46e00a91 | 6296 | break; |
bf0571fd FC |
6297 | case OA_LIST: |
6298 | while (numargs--) { | |
6299 | PUSHs(svp && *svp ? *svp : &PL_sv_undef); | |
6300 | svp++; | |
6301 | } | |
6302 | RETURN; | |
bea284c8 FC |
6303 | case OA_AVREF: |
6304 | if (!numargs) { | |
6305 | GV *gv; | |
6306 | if (CvUNIQUE(find_runcv_where(FIND_RUNCV_level_eq,1,NULL))) | |
6307 | gv = PL_argvgv; | |
6308 | else { | |
6309 | S_maybe_unwind_defav(aTHX); | |
6310 | gv = PL_defgv; | |
6311 | } | |
6312 | PUSHs((SV *)GvAVn(gv)); | |
6313 | break; | |
6314 | } | |
6315 | if (!svp || !*svp || !SvROK(*svp) | |
6316 | || SvTYPE(SvRV(*svp)) != SVt_PVAV) | |
6317 | DIE(aTHX_ | |
6318 | /* diag_listed_as: Type of arg %d to &CORE::%s must be %s*/ | |
6319 | "Type of arg %d to &CORE::%s must be array reference", | |
6320 | whicharg, PL_op_desc[opnum] | |
6321 | ); | |
6322 | PUSHs(SvRV(*svp)); | |
6323 | break; | |
19c481f4 FC |
6324 | case OA_HVREF: |
6325 | if (!svp || !*svp || !SvROK(*svp) | |
73665bc4 FC |
6326 | || ( SvTYPE(SvRV(*svp)) != SVt_PVHV |
6327 | && ( opnum == OP_DBMCLOSE || opnum == OP_DBMOPEN | |
6328 | || SvTYPE(SvRV(*svp)) != SVt_PVAV ))) | |
19c481f4 FC |
6329 | DIE(aTHX_ |
6330 | /* diag_listed_as: Type of arg %d to &CORE::%s must be %s*/ | |
73665bc4 FC |
6331 | "Type of arg %d to &CORE::%s must be hash%s reference", |
6332 | whicharg, PL_op_desc[opnum], | |
6333 | opnum == OP_DBMCLOSE || opnum == OP_DBMOPEN | |
6334 | ? "" | |
6335 | : " or array" | |
19c481f4 FC |
6336 | ); |
6337 | PUSHs(SvRV(*svp)); | |
6338 | break; | |
c931b036 | 6339 | case OA_FILEREF: |
30901a8a FC |
6340 | if (!numargs) PUSHs(NULL); |
6341 | else if(svp && *svp && SvROK(*svp) && isGV_with_GP(SvRV(*svp))) | |
c931b036 FC |
6342 | /* no magic here, as the prototype will have added an extra |
6343 | refgen and we just want what was there before that */ | |
6344 | PUSHs(SvRV(*svp)); | |
6345 | else { | |
6346 | const bool constr = PL_op->op_private & whicharg; | |
6347 | PUSHs(S_rv2gv(aTHX_ | |
6348 | svp && *svp ? *svp : &PL_sv_undef, | |
b54f893d | 6349 | constr, cBOOL(CopHINTS_get(PL_curcop) & HINT_STRICT_REFS), |
c931b036 FC |
6350 | !constr |
6351 | )); | |
6352 | } | |
6353 | break; | |
c72a5629 | 6354 | case OA_SCALARREF: |
1efec5ed FC |
6355 | if (!numargs) goto try_defsv; |
6356 | else { | |
17008668 FC |
6357 | const bool wantscalar = |
6358 | PL_op->op_private & OPpCOREARGS_SCALARMOD; | |
c72a5629 | 6359 | if (!svp || !*svp || !SvROK(*svp) |
17008668 FC |
6360 | /* We have to permit globrefs even for the \$ proto, as |
6361 | *foo is indistinguishable from ${\*foo}, and the proto- | |
6362 | type permits the latter. */ | |
6363 | || SvTYPE(SvRV(*svp)) > ( | |
efe889ae | 6364 | wantscalar ? SVt_PVLV |
46bef06f FC |
6365 | : opnum == OP_LOCK || opnum == OP_UNDEF |
6366 | ? SVt_PVCV | |
efe889ae | 6367 | : SVt_PVHV |
17008668 | 6368 | ) |
c72a5629 FC |
6369 | ) |
6370 | DIE(aTHX_ | |
17008668 | 6371 | "Type of arg %d to &CORE::%s must be %s", |
46bef06f | 6372 | whicharg, PL_op_name[opnum], |
17008668 FC |
6373 | wantscalar |
6374 | ? "scalar reference" | |
46bef06f | 6375 | : opnum == OP_LOCK || opnum == OP_UNDEF |
efe889ae FC |
6376 | ? "reference to one of [$@%&*]" |
6377 | : "reference to one of [$@%*]" | |
c72a5629 FC |
6378 | ); |
6379 | PUSHs(SvRV(*svp)); | |
bea284c8 | 6380 | if (opnum == OP_UNDEF && SvRV(*svp) == (SV *)PL_defgv) { |
88bb468b | 6381 | /* Undo @_ localisation, so that sub exit does not undo |
04e686b8 | 6382 | part of our undeffing. */ |
bea284c8 | 6383 | S_maybe_unwind_defav(aTHX); |
88bb468b | 6384 | } |
17008668 | 6385 | } |
1efec5ed | 6386 | break; |
46e00a91 | 6387 | default: |
46e00a91 FC |
6388 | DIE(aTHX_ "panic: unknown OA_*: %x", (unsigned)(oa&7)); |
6389 | } | |
6390 | oa = oa >> 4; | |
6391 | } | |
6392 | ||
deb8a388 FC |
6393 | RETURN; |
6394 | } | |
6395 | ||
88101882 FC |
6396 | PP(pp_avhvswitch) |
6397 | { | |
a73158aa | 6398 | dVAR; dSP; |
73665bc4 FC |
6399 | return PL_ppaddr[ |
6400 | (SvTYPE(TOPs) == SVt_PVAV ? OP_AEACH : OP_EACH) | |
e1e26374 | 6401 | + (PL_op->op_private & OPpAVHVSWITCH_MASK) |
73665bc4 | 6402 | ](aTHX); |
88101882 FC |
6403 | } |
6404 | ||
84ed0108 FC |
6405 | PP(pp_runcv) |
6406 | { | |
6407 | dSP; | |
6408 | CV *cv; | |
6409 | if (PL_op->op_private & OPpOFFBYONE) { | |
db4cf31d | 6410 | cv = find_runcv_where(FIND_RUNCV_level_eq, 1, NULL); |
84ed0108 FC |
6411 | } |
6412 | else cv = find_runcv(NULL); | |
e157a82b | 6413 | XPUSHs(CvEVAL(cv) ? &PL_sv_undef : sv_2mortal(newRV((SV *)cv))); |
84ed0108 FC |
6414 | RETURN; |
6415 | } | |
6416 | ||
05a34802 | 6417 | static void |
2331e434 | 6418 | S_localise_aelem_lval(pTHX_ AV * const av, SV * const keysv, |
05a34802 FC |
6419 | const bool can_preserve) |
6420 | { | |
2331e434 | 6421 | const SSize_t ix = SvIV(keysv); |
05a34802 FC |
6422 | if (can_preserve ? av_exists(av, ix) : TRUE) { |
6423 | SV ** const svp = av_fetch(av, ix, 1); | |
6424 | if (!svp || !*svp) | |
6425 | Perl_croak(aTHX_ PL_no_aelem, ix); | |
6426 | save_aelem(av, ix, svp); | |
6427 | } | |
6428 | else | |
6429 | SAVEADELETE(av, ix); | |
6430 | } | |
6431 | ||
5f94141d FC |
6432 | static void |
6433 | S_localise_helem_lval(pTHX_ HV * const hv, SV * const keysv, | |
6434 | const bool can_preserve) | |
6435 | { | |
6436 | if (can_preserve ? hv_exists_ent(hv, keysv, 0) : TRUE) { | |
6437 | HE * const he = hv_fetch_ent(hv, keysv, 1, 0); | |
6438 | SV ** const svp = he ? &HeVAL(he) : NULL; | |
6439 | if (!svp || !*svp) | |
6440 | Perl_croak(aTHX_ PL_no_helem_sv, SVfARG(keysv)); | |
6441 | save_helem_flags(hv, keysv, svp, 0); | |
6442 | } | |
6443 | else | |
6444 | SAVEHDELETE(hv, keysv); | |
6445 | } | |
6446 | ||
9782ce69 FC |
6447 | static void |
6448 | S_localise_gv_slot(pTHX_ GV *gv, U8 type) | |
6449 | { | |
6450 | if (type == OPpLVREF_SV) { | |
6451 | save_pushptrptr(gv, SvREFCNT_inc_simple(GvSV(gv)), SAVEt_GVSV); | |
6452 | GvSV(gv) = 0; | |
6453 | } | |
6454 | else if (type == OPpLVREF_AV) | |
6455 | /* XXX Inefficient, as it creates a new AV, which we are | |
6456 | about to clobber. */ | |
6457 | save_ary(gv); | |
6458 | else { | |
6459 | assert(type == OPpLVREF_HV); | |
6460 | /* XXX Likewise inefficient. */ | |
6461 | save_hash(gv); | |
6462 | } | |
6463 | } | |
6464 | ||
6465 | ||
254da51f FC |
6466 | PP(pp_refassign) |
6467 | { | |
4fec8804 | 6468 | dSP; |
6102323a | 6469 | SV * const key = PL_op->op_private & OPpLVREF_ELEM ? POPs : NULL; |
d8a875d9 | 6470 | SV * const left = PL_op->op_flags & OPf_STACKED ? POPs : NULL; |
4fec8804 | 6471 | dTOPss; |
3f114923 | 6472 | const char *bad = NULL; |
ac0da85a | 6473 | const U8 type = PL_op->op_private & OPpLVREF_TYPE; |
4fec8804 | 6474 | if (!SvROK(sv)) DIE(aTHX_ "Assigned value is not a reference"); |
ac0da85a | 6475 | switch (type) { |
3f114923 FC |
6476 | case OPpLVREF_SV: |
6477 | if (SvTYPE(SvRV(sv)) > SVt_PVLV) | |
6478 | bad = " SCALAR"; | |
6479 | break; | |
6480 | case OPpLVREF_AV: | |
6481 | if (SvTYPE(SvRV(sv)) != SVt_PVAV) | |
6482 | bad = "n ARRAY"; | |
6483 | break; | |
6484 | case OPpLVREF_HV: | |
6485 | if (SvTYPE(SvRV(sv)) != SVt_PVHV) | |
6486 | bad = " HASH"; | |
6487 | break; | |
6488 | case OPpLVREF_CV: | |
6489 | if (SvTYPE(SvRV(sv)) != SVt_PVCV) | |
6490 | bad = " CODE"; | |
6491 | } | |
6492 | if (bad) | |
1f8155a2 | 6493 | /* diag_listed_as: Assigned value is not %s reference */ |
3f114923 | 6494 | DIE(aTHX_ "Assigned value is not a%s reference", bad); |
b943805e JH |
6495 | { |
6496 | MAGIC *mg; | |
6497 | HV *stash; | |
d8a875d9 FC |
6498 | switch (left ? SvTYPE(left) : 0) { |
6499 | case 0: | |
cf5d2d91 FC |
6500 | { |
6501 | SV * const old = PAD_SV(ARGTARG); | |
d8a875d9 | 6502 | PAD_SETSV(ARGTARG, SvREFCNT_inc_NN(SvRV(sv))); |
cf5d2d91 | 6503 | SvREFCNT_dec(old); |
3ad7d304 FC |
6504 | if ((PL_op->op_private & (OPpLVAL_INTRO|OPpPAD_STATE)) |
6505 | == OPpLVAL_INTRO) | |
fc048fcf | 6506 | SAVECLEARSV(PAD_SVl(ARGTARG)); |
d8a875d9 | 6507 | break; |
cf5d2d91 | 6508 | } |
d8a875d9 | 6509 | case SVt_PVGV: |
2a57afb1 | 6510 | if (PL_op->op_private & OPpLVAL_INTRO) { |
9782ce69 | 6511 | S_localise_gv_slot(aTHX_ (GV *)left, type); |
2a57afb1 | 6512 | } |
d8a875d9 FC |
6513 | gv_setref(left, sv); |
6514 | SvSETMAGIC(left); | |
6102323a FC |
6515 | break; |
6516 | case SVt_PVAV: | |
69a23520 | 6517 | assert(key); |
40d2b828 | 6518 | if (UNLIKELY(PL_op->op_private & OPpLVAL_INTRO)) { |
2331e434 | 6519 | S_localise_aelem_lval(aTHX_ (AV *)left, key, |
05a34802 | 6520 | SvCANEXISTDELETE(left)); |
40d2b828 | 6521 | } |
6102323a FC |
6522 | av_store((AV *)left, SvIV(key), SvREFCNT_inc_simple_NN(SvRV(sv))); |
6523 | break; | |
5f94141d | 6524 | case SVt_PVHV: |
69a23520 JH |
6525 | if (UNLIKELY(PL_op->op_private & OPpLVAL_INTRO)) { |
6526 | assert(key); | |
5f94141d FC |
6527 | S_localise_helem_lval(aTHX_ (HV *)left, key, |
6528 | SvCANEXISTDELETE(left)); | |
69a23520 | 6529 | } |
7fcb36d5 | 6530 | (void)hv_store_ent((HV *)left, key, SvREFCNT_inc_simple_NN(SvRV(sv)), 0); |
d8a875d9 | 6531 | } |
4fec8804 FC |
6532 | if (PL_op->op_flags & OPf_MOD) |
6533 | SETs(sv_2mortal(newSVsv(sv))); | |
6534 | /* XXX else can weak references go stale before they are read, e.g., | |
6535 | in leavesub? */ | |
6536 | RETURN; | |
b943805e | 6537 | } |
254da51f FC |
6538 | } |
6539 | ||
4c5bab50 FC |
6540 | PP(pp_lvref) |
6541 | { | |
26a50d99 FC |
6542 | dSP; |
6543 | SV * const ret = sv_2mortal(newSV_type(SVt_PVMG)); | |
6102323a | 6544 | SV * const elem = PL_op->op_private & OPpLVREF_ELEM ? POPs : NULL; |
2a57afb1 | 6545 | SV * const arg = PL_op->op_flags & OPf_STACKED ? POPs : NULL; |
9782ce69 FC |
6546 | MAGIC * const mg = sv_magicext(ret, arg, PERL_MAGIC_lvref, |
6547 | &PL_vtbl_lvref, (char *)elem, | |
23270f96 | 6548 | elem ? HEf_SVKEY : (I32)ARGTARG); |
9782ce69 | 6549 | mg->mg_private = PL_op->op_private; |
d39c26a6 FC |
6550 | if (PL_op->op_private & OPpLVREF_ITER) |
6551 | mg->mg_flags |= MGf_PERSIST; | |
9846cd95 | 6552 | if (UNLIKELY(PL_op->op_private & OPpLVAL_INTRO)) { |
40d2b828 | 6553 | if (elem) { |
38bb0011 JH |
6554 | MAGIC *mg; |
6555 | HV *stash; | |
6556 | assert(arg); | |
6557 | { | |
6558 | const bool can_preserve = SvCANEXISTDELETE(arg); | |
6559 | if (SvTYPE(arg) == SVt_PVAV) | |
6560 | S_localise_aelem_lval(aTHX_ (AV *)arg, elem, can_preserve); | |
6561 | else | |
6562 | S_localise_helem_lval(aTHX_ (HV *)arg, elem, can_preserve); | |
6563 | } | |
40d2b828 FC |
6564 | } |
6565 | else if (arg) { | |
9782ce69 FC |
6566 | S_localise_gv_slot(aTHX_ (GV *)arg, |
6567 | PL_op->op_private & OPpLVREF_TYPE); | |
2a57afb1 | 6568 | } |
3ad7d304 | 6569 | else if (!(PL_op->op_private & OPpPAD_STATE)) |
c146a62a | 6570 | SAVECLEARSV(PAD_SVl(ARGTARG)); |
1199b01a | 6571 | } |
c146a62a FC |
6572 | XPUSHs(ret); |
6573 | RETURN; | |
4c5bab50 | 6574 | } |
84ed0108 | 6575 | |
16b99412 FC |
6576 | PP(pp_lvrefslice) |
6577 | { | |
a95dad8a | 6578 | dSP; dMARK; |
0ca7b7f7 FC |
6579 | AV * const av = (AV *)POPs; |
6580 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; | |
6581 | bool can_preserve = FALSE; | |
6582 | ||
9846cd95 | 6583 | if (UNLIKELY(localizing)) { |
0ca7b7f7 FC |
6584 | MAGIC *mg; |
6585 | HV *stash; | |
6586 | SV **svp; | |
6587 | ||
6588 | can_preserve = SvCANEXISTDELETE(av); | |
6589 | ||
6590 | if (SvTYPE(av) == SVt_PVAV) { | |
6591 | SSize_t max = -1; | |
6592 | ||
6593 | for (svp = MARK + 1; svp <= SP; svp++) { | |
6594 | const SSize_t elem = SvIV(*svp); | |
6595 | if (elem > max) | |
6596 | max = elem; | |
6597 | } | |
6598 | if (max > AvMAX(av)) | |
6599 | av_extend(av, max); | |
6600 | } | |
6601 | } | |
6602 | ||
6603 | while (++MARK <= SP) { | |
6604 | SV * const elemsv = *MARK; | |
5f94141d | 6605 | if (SvTYPE(av) == SVt_PVAV) |
2331e434 | 6606 | S_localise_aelem_lval(aTHX_ av, elemsv, can_preserve); |
5f94141d FC |
6607 | else |
6608 | S_localise_helem_lval(aTHX_ (HV *)av, elemsv, can_preserve); | |
0ca7b7f7 FC |
6609 | *MARK = sv_2mortal(newSV_type(SVt_PVMG)); |
6610 | sv_magic(*MARK,(SV *)av,PERL_MAGIC_lvref,(char *)elemsv,HEf_SVKEY); | |
6611 | } | |
6612 | RETURN; | |
16b99412 FC |
6613 | } |
6614 | ||
2882b3ff FC |
6615 | PP(pp_lvavref) |
6616 | { | |
bdaf10a5 FC |
6617 | if (PL_op->op_flags & OPf_STACKED) |
6618 | Perl_pp_rv2av(aTHX); | |
6619 | else | |
6620 | Perl_pp_padav(aTHX); | |
6621 | { | |
6622 | dSP; | |
6623 | dTOPss; | |
6624 | SETs(0); /* special alias marker that aassign recognises */ | |
6625 | XPUSHs(sv); | |
6626 | RETURN; | |
6627 | } | |
2882b3ff FC |
6628 | } |
6629 | ||
b77472f9 FC |
6630 | PP(pp_anonconst) |
6631 | { | |
6632 | dSP; | |
6633 | dTOPss; | |
6634 | SETs(sv_2mortal((SV *)newCONSTSUB(SvTYPE(CopSTASH(PL_curcop))==SVt_PVHV | |
6635 | ? CopSTASH(PL_curcop) | |
6636 | : NULL, | |
6637 | NULL, SvREFCNT_inc_simple_NN(sv)))); | |
6638 | RETURN; | |
6639 | } | |
6640 | ||
4fa06845 DM |
6641 | |
6642 | /* process one subroutine argument - typically when the sub has a signature: | |
6643 | * introduce PL_curpad[op_targ] and assign to it the value | |
6644 | * for $: (OPf_STACKED ? *sp : $_[N]) | |
6645 | * for @/%: @_[N..$#_] | |
6646 | * | |
6647 | * It's equivalent to | |
6648 | * my $foo = $_[N]; | |
6649 | * or | |
6650 | * my $foo = (value-on-stack) | |
6651 | * or | |
6652 | * my @foo = @_[N..$#_] | |
6653 | * etc | |
4fa06845 DM |
6654 | */ |
6655 | ||
6656 | PP(pp_argelem) | |
6657 | { | |
6658 | dTARG; | |
6659 | SV *val; | |
6660 | SV ** padentry; | |
6661 | OP *o = PL_op; | |
6662 | AV *defav = GvAV(PL_defgv); /* @_ */ | |
6daeaaa3 | 6663 | IV ix = PTR2IV(cUNOP_AUXo->op_aux); |
4fa06845 | 6664 | IV argc; |
4fa06845 DM |
6665 | |
6666 | /* do 'my $var, @var or %var' action */ | |
6667 | padentry = &(PAD_SVl(o->op_targ)); | |
6668 | save_clearsv(padentry); | |
6669 | targ = *padentry; | |
6670 | ||
6671 | if ((o->op_private & OPpARGELEM_MASK) == OPpARGELEM_SV) { | |
6672 | if (o->op_flags & OPf_STACKED) { | |
6673 | dSP; | |
6674 | val = POPs; | |
6675 | PUTBACK; | |
6676 | } | |
6677 | else { | |
f6ca42c7 | 6678 | SV **svp; |
4fa06845 | 6679 | /* should already have been checked */ |
f6ca42c7 | 6680 | assert(ix >= 0); |
6daeaaa3 DM |
6681 | #if IVSIZE > PTRSIZE |
6682 | assert(ix <= SSize_t_MAX); | |
6683 | #endif | |
f6ca42c7 DM |
6684 | |
6685 | svp = av_fetch(defav, ix, FALSE); | |
6686 | val = svp ? *svp : &PL_sv_undef; | |
4fa06845 DM |
6687 | } |
6688 | ||
6689 | /* $var = $val */ | |
6690 | ||
6691 | /* cargo-culted from pp_sassign */ | |
6692 | assert(TAINTING_get || !TAINT_get); | |
6693 | if (UNLIKELY(TAINT_get) && !SvTAINTED(val)) | |
6694 | TAINT_NOT; | |
6695 | ||
f6ca42c7 | 6696 | SvSetMagicSV(targ, val); |
4fa06845 DM |
6697 | return o->op_next; |
6698 | } | |
6699 | ||
6700 | /* must be AV or HV */ | |
6701 | ||
6702 | assert(!(o->op_flags & OPf_STACKED)); | |
f6ca42c7 | 6703 | argc = ((IV)AvFILL(defav) + 1) - ix; |
4fa06845 DM |
6704 | |
6705 | /* This is a copy of the relevant parts of pp_aassign(). | |
4fa06845 DM |
6706 | */ |
6707 | if ((o->op_private & OPpARGELEM_MASK) == OPpARGELEM_AV) { | |
f6ca42c7 DM |
6708 | IV i; |
6709 | ||
6710 | if (AvFILL((AV*)targ) > -1) { | |
6711 | /* target should usually be empty. If we get get | |
6712 | * here, someone's been doing some weird closure tricks. | |
6713 | * Make a copy of all args before clearing the array, | |
6714 | * to avoid the equivalent of @a = ($a[0]) prematurely freeing | |
6715 | * elements. See similar code in pp_aassign. | |
6716 | */ | |
6717 | for (i = 0; i < argc; i++) { | |
6718 | SV **svp = av_fetch(defav, ix + i, FALSE); | |
6719 | SV *newsv = newSV(0); | |
6720 | sv_setsv_flags(newsv, | |
6721 | svp ? *svp : &PL_sv_undef, | |
6722 | (SV_DO_COW_SVSETSV|SV_NOSTEAL)); | |
6723 | if (!av_store(defav, ix + i, newsv)) | |
6724 | SvREFCNT_dec_NN(newsv); | |
6725 | } | |
6726 | av_clear((AV*)targ); | |
6727 | } | |
6728 | ||
6729 | if (argc <= 0) | |
6730 | return o->op_next; | |
4fa06845 | 6731 | |
4fa06845 DM |
6732 | av_extend((AV*)targ, argc); |
6733 | ||
f6ca42c7 | 6734 | i = 0; |
4fa06845 DM |
6735 | while (argc--) { |
6736 | SV *tmpsv; | |
f6ca42c7 DM |
6737 | SV **svp = av_fetch(defav, ix + i, FALSE); |
6738 | SV *val = svp ? *svp : &PL_sv_undef; | |
4fa06845 | 6739 | tmpsv = newSV(0); |
f6ca42c7 | 6740 | sv_setsv(tmpsv, val); |
4fa06845 DM |
6741 | av_store((AV*)targ, i++, tmpsv); |
6742 | TAINT_NOT; | |
6743 | } | |
6744 | ||
6745 | } | |
6746 | else { | |
f6ca42c7 DM |
6747 | IV i; |
6748 | ||
4fa06845 DM |
6749 | assert((o->op_private & OPpARGELEM_MASK) == OPpARGELEM_HV); |
6750 | ||
f6ca42c7 DM |
6751 | if (SvRMAGICAL(targ) || HvUSEDKEYS((HV*)targ)) { |
6752 | /* see "target should usually be empty" comment above */ | |
6753 | for (i = 0; i < argc; i++) { | |
6754 | SV **svp = av_fetch(defav, ix + i, FALSE); | |
6755 | SV *newsv = newSV(0); | |
6756 | sv_setsv_flags(newsv, | |
6757 | svp ? *svp : &PL_sv_undef, | |
6758 | (SV_DO_COW_SVSETSV|SV_NOSTEAL)); | |
6759 | if (!av_store(defav, ix + i, newsv)) | |
6760 | SvREFCNT_dec_NN(newsv); | |
6761 | } | |
6762 | hv_clear((HV*)targ); | |
6763 | } | |
6764 | ||
6765 | if (argc <= 0) | |
6766 | return o->op_next; | |
4fa06845 | 6767 | assert(argc % 2 == 0); |
4fa06845 | 6768 | |
f6ca42c7 | 6769 | i = 0; |
4fa06845 DM |
6770 | while (argc) { |
6771 | SV *tmpsv; | |
f6ca42c7 DM |
6772 | SV **svp; |
6773 | SV *key; | |
6774 | SV *val; | |
6775 | ||
6776 | svp = av_fetch(defav, ix + i++, FALSE); | |
6777 | key = svp ? *svp : &PL_sv_undef; | |
6778 | svp = av_fetch(defav, ix + i++, FALSE); | |
6779 | val = svp ? *svp : &PL_sv_undef; | |
4fa06845 | 6780 | |
4fa06845 DM |
6781 | argc -= 2; |
6782 | if (UNLIKELY(SvGMAGICAL(key))) | |
6783 | key = sv_mortalcopy(key); | |
6784 | tmpsv = newSV(0); | |
6785 | sv_setsv(tmpsv, val); | |
6786 | hv_store_ent((HV*)targ, key, tmpsv, 0); | |
6787 | TAINT_NOT; | |
6788 | } | |
6789 | } | |
6790 | ||
6791 | return o->op_next; | |
6792 | } | |
6793 | ||
6794 | /* Handle a default value for one subroutine argument (typically as part | |
6795 | * of a subroutine signature). | |
6796 | * It's equivalent to | |
6797 | * @_ > op_targ ? $_[op_targ] : result_of(op_other) | |
6798 | * | |
6799 | * Intended to be used where op_next is an OP_ARGELEM | |
6800 | * | |
6801 | * We abuse the op_targ field slightly: it's an index into @_ rather than | |
6802 | * into PL_curpad. | |
6803 | */ | |
6804 | ||
6805 | PP(pp_argdefelem) | |
6806 | { | |
6807 | OP * const o = PL_op; | |
6808 | AV *defav = GvAV(PL_defgv); /* @_ */ | |
6daeaaa3 | 6809 | IV ix = (IV)o->op_targ; |
4fa06845 | 6810 | |
f6ca42c7 | 6811 | assert(ix >= 0); |
6daeaaa3 DM |
6812 | #if IVSIZE > PTRSIZE |
6813 | assert(ix <= SSize_t_MAX); | |
6814 | #endif | |
f6ca42c7 DM |
6815 | |
6816 | if (AvFILL(defav) >= ix) { | |
4fa06845 | 6817 | dSP; |
f6ca42c7 DM |
6818 | SV **svp = av_fetch(defav, ix, FALSE); |
6819 | SV *val = svp ? *svp : &PL_sv_undef; | |
6820 | XPUSHs(val); | |
4fa06845 DM |
6821 | RETURN; |
6822 | } | |
6823 | return cLOGOPo->op_other; | |
6824 | } | |
6825 | ||
6826 | ||
ac7609e4 AC |
6827 | static SV * |
6828 | S_find_runcv_name(void) | |
6829 | { | |
6830 | dTHX; | |
6831 | CV *cv; | |
6832 | GV *gv; | |
6833 | SV *sv; | |
6834 | ||
6835 | cv = find_runcv(0); | |
6836 | if (!cv) | |
6837 | return &PL_sv_no; | |
6838 | ||
6839 | gv = CvGV(cv); | |
6840 | if (!gv) | |
6841 | return &PL_sv_no; | |
6842 | ||
6843 | sv = sv_2mortal(newSV(0)); | |
6844 | gv_fullname4(sv, gv, NULL, TRUE); | |
6845 | return sv; | |
6846 | } | |
4fa06845 DM |
6847 | |
6848 | /* Check a a subs arguments - i.e. that it has the correct number of args | |
6849 | * (and anything else we might think of in future). Typically used with | |
6850 | * signatured subs. | |
6851 | */ | |
6852 | ||
6853 | PP(pp_argcheck) | |
6854 | { | |
6855 | OP * const o = PL_op; | |
6856 | UNOP_AUX_item *aux = cUNOP_AUXo->op_aux; | |
6daeaaa3 DM |
6857 | IV params = aux[0].iv; |
6858 | IV opt_params = aux[1].iv; | |
4fa06845 DM |
6859 | char slurpy = (char)(aux[2].iv); |
6860 | AV *defav = GvAV(PL_defgv); /* @_ */ | |
6daeaaa3 | 6861 | IV argc; |
4fa06845 DM |
6862 | bool too_few; |
6863 | ||
6864 | assert(!SvMAGICAL(defav)); | |
6daeaaa3 | 6865 | argc = (AvFILLp(defav) + 1); |
4fa06845 DM |
6866 | too_few = (argc < (params - opt_params)); |
6867 | ||
6868 | if (UNLIKELY(too_few || (!slurpy && argc > params))) | |
ac7609e4 AC |
6869 | /* diag_listed_as: Too few arguments for subroutine '%s' */ |
6870 | /* diag_listed_as: Too many arguments for subroutine '%s' */ | |
6871 | Perl_croak_caller("Too %s arguments for subroutine '%" SVf "'", | |
6872 | too_few ? "few" : "many", S_find_runcv_name()); | |
4fa06845 DM |
6873 | |
6874 | if (UNLIKELY(slurpy == '%' && argc > params && (argc - params) % 2)) | |
ac7609e4 AC |
6875 | /* diag_listed_as: Odd name/value argument for subroutine '%s' */ |
6876 | Perl_croak_caller("Odd name/value argument for subroutine '%" SVf "'", | |
6877 | S_find_runcv_name()); | |
4fa06845 DM |
6878 | |
6879 | return NORMAL; | |
6880 | } | |
6881 | ||
e609e586 | 6882 | /* |
14d04a33 | 6883 | * ex: set ts=8 sts=4 sw=4 et: |
37442d52 | 6884 | */ |