https://perl5.git.perl.org / perl5.git / blame
| Commit | Line | Data |
|---|---|---|
| a0d0e21e | 1 | /* pp.c |
| 79072805 | 2 | * |
| 1129b882 NC |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, |
| 4 | * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others | |
| 79072805 | 5 | * |
| a0d0e21e LW |
6 | * You may distribute under the terms of either the GNU General Public |
| 7 | * License or the Artistic License, as specified in the README file. | |
| 79072805 | 8 | * |
| a0d0e21e LW |
9 | */ |
| 10 | ||
| 11 | /* | |
| 4ac71550 TC |
12 | * 'It's a big house this, and very peculiar. Always a bit more |
| 13 | * to discover, and no knowing what you'll find round a corner. | |
| 14 | * And Elves, sir!' --Samwise Gamgee | |
| 15 | * | |
| 16 | * [p.225 of _The Lord of the Rings_, II/i: "Many Meetings"] | |
| a0d0e21e | 17 | */ |
| 79072805 | 18 | |
| 166f8a29 DM |
19 | /* This file contains general pp ("push/pop") functions that execute the |
| 20 | * opcodes that make up a perl program. A typical pp function expects to | |
| 21 | * find its arguments on the stack, and usually pushes its results onto | |
| 22 | * the stack, hence the 'pp' terminology. Each OP structure contains | |
| 23 | * a pointer to the relevant pp_foo() function. | |
| 24 | */ | |
| 25 | ||
| 79072805 | 26 | #include "EXTERN.h" |
| 864dbfa3 | 27 | #define PERL_IN_PP_C |
| 79072805 | 28 | #include "perl.h" |
| 77bc9082 | 29 | #include "keywords.h" |
| 79072805 | 30 | |
| a4af207c JH |
31 | #include "reentr.h" |
| 32 | ||
| dfe9444c AD |
33 | /* XXX I can't imagine anyone who doesn't have this actually _needs_ |
| 34 | it, since pid_t is an integral type. | |
| 35 | --AD 2/20/1998 | |
| 36 | */ | |
| 37 | #ifdef NEED_GETPID_PROTO | |
| 38 | extern Pid_t getpid (void); | |
| 8ac85365 NIS |
39 | #endif |
| 40 | ||
| 0630166f SP |
41 | /* |
| 42 | * Some BSDs and Cygwin default to POSIX math instead of IEEE. | |
| 43 | * This switches them over to IEEE. | |
| 44 | */ | |
| 45 | #if defined(LIBM_LIB_VERSION) | |
| 46 | _LIB_VERSION_TYPE _LIB_VERSION = _IEEE_; | |
| 47 | #endif | |
| 48 | ||
| 13017935 SM |
49 | /* variations on pp_null */ |
| 50 | ||
| 93a17b20 LW |
51 | PP(pp_stub) |
| 52 | { | |
| 97aff369 | 53 | dVAR; |
| 39644a26 | 54 | dSP; |
| 54310121 | 55 | if (GIMME_V == G_SCALAR) |
| 3280af22 | 56 | XPUSHs(&PL_sv_undef); |
| 93a17b20 LW |
57 | RETURN; |
| 58 | } | |
| 59 | ||
| 79072805 LW |
60 | /* Pushy stuff. */ |
| 61 | ||
| 93a17b20 LW |
62 | PP(pp_padav) |
| 63 | { | |
| 97aff369 | 64 | dVAR; dSP; dTARGET; |
| 13017935 | 65 | I32 gimme; |
| e190e9b4 | 66 | assert(SvTYPE(TARG) == SVt_PVAV); |
| 533c011a | 67 | if (PL_op->op_private & OPpLVAL_INTRO) |
| a5911867 RGS |
68 | if (!(PL_op->op_private & OPpPAD_STATE)) |
| 69 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); | |
| 85e6fe83 | 70 | EXTEND(SP, 1); |
| 533c011a | 71 | if (PL_op->op_flags & OPf_REF) { |
| 85e6fe83 | 72 | PUSHs(TARG); |
| 93a17b20 | 73 | RETURN; |
| 78f9721b SM |
74 | } else if (LVRET) { |
| 75 | if (GIMME == G_SCALAR) | |
| 76 | Perl_croak(aTHX_ "Can't return array to lvalue scalar context"); | |
| 77 | PUSHs(TARG); | |
| 78 | RETURN; | |
| 85e6fe83 | 79 | } |
| 13017935 SM |
80 | gimme = GIMME_V; |
| 81 | if (gimme == G_ARRAY) { | |
| 502c6561 | 82 | const I32 maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
| 85e6fe83 | 83 | EXTEND(SP, maxarg); |
| 93965878 NIS |
84 | if (SvMAGICAL(TARG)) { |
| 85 | U32 i; | |
| eb160463 | 86 | for (i=0; i < (U32)maxarg; i++) { |
| 502c6561 | 87 | SV * const * const svp = av_fetch(MUTABLE_AV(TARG), i, FALSE); |
| 3280af22 | 88 | SP[i+1] = (svp) ? *svp : &PL_sv_undef; |
| 93965878 NIS |
89 | } |
| 90 | } | |
| 91 | else { | |
| 502c6561 | 92 | Copy(AvARRAY((const AV *)TARG), SP+1, maxarg, SV*); |
| 93965878 | 93 | } |
| 85e6fe83 LW |
94 | SP += maxarg; |
| 95 | } | |
| 13017935 | 96 | else if (gimme == G_SCALAR) { |
| 1b6737cc | 97 | SV* const sv = sv_newmortal(); |
| 502c6561 | 98 | const I32 maxarg = AvFILL(MUTABLE_AV(TARG)) + 1; |
| 85e6fe83 LW |
99 | sv_setiv(sv, maxarg); |
| 100 | PUSHs(sv); | |
| 101 | } | |
| 102 | RETURN; | |
| 93a17b20 LW |
103 | } |
| 104 | ||
| 105 | PP(pp_padhv) | |
| 106 | { | |
| 97aff369 | 107 | dVAR; dSP; dTARGET; |
| 54310121 | 108 | I32 gimme; |
| 109 | ||
| e190e9b4 | 110 | assert(SvTYPE(TARG) == SVt_PVHV); |
| 93a17b20 | 111 | XPUSHs(TARG); |
| 533c011a | 112 | if (PL_op->op_private & OPpLVAL_INTRO) |
| a5911867 RGS |
113 | if (!(PL_op->op_private & OPpPAD_STATE)) |
| 114 | SAVECLEARSV(PAD_SVl(PL_op->op_targ)); | |
| 533c011a | 115 | if (PL_op->op_flags & OPf_REF) |
| 93a17b20 | 116 | RETURN; |
| 78f9721b SM |
117 | else if (LVRET) { |
| 118 | if (GIMME == G_SCALAR) | |
| 119 | Perl_croak(aTHX_ "Can't return hash to lvalue scalar context"); | |
| 120 | RETURN; | |
| 121 | } | |
| 54310121 | 122 | gimme = GIMME_V; |
| 123 | if (gimme == G_ARRAY) { | |
| 981b7185 | 124 | RETURNOP(Perl_do_kv(aTHX)); |
| 85e6fe83 | 125 | } |
| 54310121 | 126 | else if (gimme == G_SCALAR) { |
| 85fbaab2 | 127 | SV* const sv = Perl_hv_scalar(aTHX_ MUTABLE_HV(TARG)); |
| 85e6fe83 | 128 | SETs(sv); |
| 85e6fe83 | 129 | } |
| 54310121 | 130 | RETURN; |
| 93a17b20 LW |
131 | } |
| 132 | ||
| 79072805 LW |
133 | /* Translations. */ |
| 134 | ||
| 4bdf8368 | 135 | static const char S_no_symref_sv[] = |
| def89bff NC |
136 | "Can't use string (\"%" SVf32 "\"%s) as %s ref while \"strict refs\" in use"; |
| 137 | ||
| 79072805 LW |
138 | PP(pp_rv2gv) |
| 139 | { | |
| 97aff369 | 140 | dVAR; dSP; dTOPss; |
| 8ec5e241 | 141 | |
| f64c9ac5 | 142 | if (!isGV(sv) || SvFAKE(sv)) SvGETMAGIC(sv); |
| ed6116ce | 143 | if (SvROK(sv)) { |
| a0d0e21e | 144 | wasref: |
| 93d7320b DM |
145 | if (SvAMAGIC(sv)) { |
| 146 | sv = amagic_deref_call(sv, to_gv_amg); | |
| 147 | SPAGAIN; | |
| 148 | } | |
| ed6116ce | 149 | sv = SvRV(sv); |
| b1dadf13 | 150 | if (SvTYPE(sv) == SVt_PVIO) { |
| 159b6efe | 151 | GV * const gv = MUTABLE_GV(sv_newmortal()); |
| b1dadf13 | 152 | gv_init(gv, 0, "", 0, 0); |
| a45c7426 | 153 | GvIOp(gv) = MUTABLE_IO(sv); |
| b37c2d43 | 154 | SvREFCNT_inc_void_NN(sv); |
| ad64d0ec | 155 | sv = MUTABLE_SV(gv); |
| ef54e1a4 | 156 | } |
| 6e592b3a | 157 | else if (!isGV_with_GP(sv)) |
| cea2e8a9 | 158 | DIE(aTHX_ "Not a GLOB reference"); |
| 79072805 LW |
159 | } |
| 160 | else { | |
| 6e592b3a | 161 | if (!isGV_with_GP(sv)) { |
| afd1915d | 162 | if (!SvOK(sv) && sv != &PL_sv_undef) { |
| b13b2135 | 163 | /* If this is a 'my' scalar and flag is set then vivify |
| 853846ea | 164 | * NI-S 1999/05/07 |
| b13b2135 | 165 | */ |
| ac53db4c | 166 | if (SvREADONLY(sv)) |
| 6ad8f254 | 167 | Perl_croak_no_modify(aTHX); |
| 1d8d4d2a | 168 | if (PL_op->op_private & OPpDEREF) { |
| 2c8ac474 GS |
169 | GV *gv; |
| 170 | if (cUNOP->op_targ) { | |
| 171 | STRLEN len; | |
| 0bd48802 AL |
172 | SV * const namesv = PAD_SV(cUNOP->op_targ); |
| 173 | const char * const name = SvPV(namesv, len); | |
| 159b6efe | 174 | gv = MUTABLE_GV(newSV(0)); |
| 2c8ac474 GS |
175 | gv_init(gv, CopSTASH(PL_curcop), name, len, 0); |
| 176 | } | |
| 177 | else { | |
| 0bd48802 | 178 | const char * const name = CopSTASHPV(PL_curcop); |
| 2c8ac474 | 179 | gv = newGVgen(name); |
| 1d8d4d2a | 180 | } |
| 43230e26 | 181 | prepare_SV_for_RV(sv); |
| ad64d0ec | 182 | SvRV_set(sv, MUTABLE_SV(gv)); |
| 853846ea | 183 | SvROK_on(sv); |
| 1d8d4d2a | 184 | SvSETMAGIC(sv); |
| 853846ea | 185 | goto wasref; |
| 2c8ac474 | 186 | } |
| 533c011a NIS |
187 | if (PL_op->op_flags & OPf_REF || |
| 188 | PL_op->op_private & HINT_STRICT_REFS) | |
| cea2e8a9 | 189 | DIE(aTHX_ PL_no_usym, "a symbol"); |
| 599cee73 | 190 | if (ckWARN(WARN_UNINITIALIZED)) |
| 29489e7c | 191 | report_uninit(sv); |
| a0d0e21e LW |
192 | RETSETUNDEF; |
| 193 | } | |
| 35cd451c GS |
194 | if ((PL_op->op_flags & OPf_SPECIAL) && |
| 195 | !(PL_op->op_flags & OPf_MOD)) | |
| 196 | { | |
| ad64d0ec | 197 | SV * const temp = MUTABLE_SV(gv_fetchsv(sv, 0, SVt_PVGV)); |
| 7a5fd60d NC |
198 | if (!temp |
| 199 | && (!is_gv_magical_sv(sv,0) | |
| ad64d0ec NC |
200 | || !(sv = MUTABLE_SV(gv_fetchsv(sv, GV_ADD, |
| 201 | SVt_PVGV))))) { | |
| 35cd451c | 202 | RETSETUNDEF; |
| c9d5ac95 | 203 | } |
| 7a5fd60d | 204 | sv = temp; |
| 35cd451c GS |
205 | } |
| 206 | else { | |
| 207 | if (PL_op->op_private & HINT_STRICT_REFS) | |
| 10b53e54 | 208 | DIE(aTHX_ S_no_symref_sv, sv, (SvPOK(sv) && SvCUR(sv)>32 ? "..." : ""), "a symbol"); |
| e26df76a NC |
209 | if ((PL_op->op_private & (OPpLVAL_INTRO|OPpDONT_INIT_GV)) |
| 210 | == OPpDONT_INIT_GV) { | |
| 211 | /* We are the target of a coderef assignment. Return | |
| 212 | the scalar unchanged, and let pp_sasssign deal with | |
| 213 | things. */ | |
| 214 | RETURN; | |
| 215 | } | |
| ad64d0ec | 216 | sv = MUTABLE_SV(gv_fetchsv(sv, GV_ADD, SVt_PVGV)); |
| 35cd451c | 217 | } |
| 2acc3314 FC |
218 | /* FAKE globs in the symbol table cause weird bugs (#77810) */ |
| 219 | if (sv) SvFAKE_off(sv); | |
| 93a17b20 | 220 | } |
| 79072805 | 221 | } |
| 2acc3314 FC |
222 | if (sv && SvFAKE(sv)) { |
| 223 | SV *newsv = sv_newmortal(); | |
| 5cf4b255 | 224 | sv_setsv_flags(newsv, sv, 0); |
| 2acc3314 | 225 | SvFAKE_off(newsv); |
| d8906c05 | 226 | sv = newsv; |
| 2acc3314 | 227 | } |
| d8906c05 FC |
228 | if (PL_op->op_private & OPpLVAL_INTRO) |
| 229 | save_gp(MUTABLE_GV(sv), !(PL_op->op_flags & OPf_SPECIAL)); | |
| 230 | SETs(sv); | |
| 79072805 LW |
231 | RETURN; |
| 232 | } | |
| 233 | ||
| dc3c76f8 NC |
234 | /* Helper function for pp_rv2sv and pp_rv2av */ |
| 235 | GV * | |
| fe9845cc RB |
236 | Perl_softref2xv(pTHX_ SV *const sv, const char *const what, |
| 237 | const svtype type, SV ***spp) | |
| dc3c76f8 NC |
238 | { |
| 239 | dVAR; | |
| 240 | GV *gv; | |
| 241 | ||
| 7918f24d NC |
242 | PERL_ARGS_ASSERT_SOFTREF2XV; |
| 243 | ||
| dc3c76f8 NC |
244 | if (PL_op->op_private & HINT_STRICT_REFS) { |
| 245 | if (SvOK(sv)) | |
| 10b53e54 | 246 | Perl_die(aTHX_ S_no_symref_sv, sv, (SvPOK(sv) && SvCUR(sv)>32 ? "..." : ""), what); |
| dc3c76f8 NC |
247 | else |
| 248 | Perl_die(aTHX_ PL_no_usym, what); | |
| 249 | } | |
| 250 | if (!SvOK(sv)) { | |
| fd1d9b5c FC |
251 | if ( |
| 252 | PL_op->op_flags & OPf_REF && | |
| 253 | PL_op->op_next->op_type != OP_BOOLKEYS | |
| 254 | ) | |
| dc3c76f8 NC |
255 | Perl_die(aTHX_ PL_no_usym, what); |
| 256 | if (ckWARN(WARN_UNINITIALIZED)) | |
| 257 | report_uninit(sv); | |
| 258 | if (type != SVt_PV && GIMME_V == G_ARRAY) { | |
| 259 | (*spp)--; | |
| 260 | return NULL; | |
| 261 | } | |
| 262 | **spp = &PL_sv_undef; | |
| 263 | return NULL; | |
| 264 | } | |
| 265 | if ((PL_op->op_flags & OPf_SPECIAL) && | |
| 266 | !(PL_op->op_flags & OPf_MOD)) | |
| 267 | { | |
| 81e3fc25 | 268 | gv = gv_fetchsv(sv, 0, type); |
| dc3c76f8 NC |
269 | if (!gv |
| 270 | && (!is_gv_magical_sv(sv,0) | |
| 81e3fc25 | 271 | || !(gv = gv_fetchsv(sv, GV_ADD, type)))) |
| dc3c76f8 NC |
272 | { |
| 273 | **spp = &PL_sv_undef; | |
| 274 | return NULL; | |
| 275 | } | |
| 276 | } | |
| 277 | else { | |
| 81e3fc25 | 278 | gv = gv_fetchsv(sv, GV_ADD, type); |
| dc3c76f8 NC |
279 | } |
| 280 | return gv; | |
| 281 | } | |
| 282 | ||
| 79072805 LW |
283 | PP(pp_rv2sv) |
| 284 | { | |
| 97aff369 | 285 | dVAR; dSP; dTOPss; |
| c445ea15 | 286 | GV *gv = NULL; |
| 79072805 | 287 | |
| 0824d667 DM |
288 | if (!(PL_op->op_private & OPpDEREFed)) |
| 289 | SvGETMAGIC(sv); | |
| ed6116ce | 290 | if (SvROK(sv)) { |
| 93d7320b DM |
291 | if (SvAMAGIC(sv)) { |
| 292 | sv = amagic_deref_call(sv, to_sv_amg); | |
| 293 | SPAGAIN; | |
| 294 | } | |
| f5284f61 | 295 | |
| ed6116ce | 296 | sv = SvRV(sv); |
| 79072805 LW |
297 | switch (SvTYPE(sv)) { |
| 298 | case SVt_PVAV: | |
| 299 | case SVt_PVHV: | |
| 300 | case SVt_PVCV: | |
| cbae9b9f YST |
301 | case SVt_PVFM: |
| 302 | case SVt_PVIO: | |
| cea2e8a9 | 303 | DIE(aTHX_ "Not a SCALAR reference"); |
| 42d0e0b7 | 304 | default: NOOP; |
| 79072805 LW |
305 | } |
| 306 | } | |
| 307 | else { | |
| 159b6efe | 308 | gv = MUTABLE_GV(sv); |
| 748a9306 | 309 | |
| 6e592b3a | 310 | if (!isGV_with_GP(gv)) { |
| dc3c76f8 NC |
311 | gv = Perl_softref2xv(aTHX_ sv, "a SCALAR", SVt_PV, &sp); |
| 312 | if (!gv) | |
| 313 | RETURN; | |
| 463ee0b2 | 314 | } |
| 29c711a3 | 315 | sv = GvSVn(gv); |
| a0d0e21e | 316 | } |
| 533c011a | 317 | if (PL_op->op_flags & OPf_MOD) { |
| 82d03984 RGS |
318 | if (PL_op->op_private & OPpLVAL_INTRO) { |
| 319 | if (cUNOP->op_first->op_type == OP_NULL) | |
| 159b6efe | 320 | sv = save_scalar(MUTABLE_GV(TOPs)); |
| 82d03984 RGS |
321 | else if (gv) |
| 322 | sv = save_scalar(gv); | |
| 323 | else | |
| f1f66076 | 324 | Perl_croak(aTHX_ "%s", PL_no_localize_ref); |
| 82d03984 | 325 | } |
| 533c011a NIS |
326 | else if (PL_op->op_private & OPpDEREF) |
| 327 | vivify_ref(sv, PL_op->op_private & OPpDEREF); | |
| 79072805 | 328 | } |
| a0d0e21e | 329 | SETs(sv); |
| 79072805 LW |
330 | RETURN; |
| 331 | } | |
| 332 | ||
| 333 | PP(pp_av2arylen) | |
| 334 | { | |
| 97aff369 | 335 | dVAR; dSP; |
| 502c6561 | 336 | AV * const av = MUTABLE_AV(TOPs); |
| 02d85cc3 EB |
337 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
| 338 | if (lvalue) { | |
| 339 | SV ** const sv = Perl_av_arylen_p(aTHX_ MUTABLE_AV(av)); | |
| 340 | if (!*sv) { | |
| 341 | *sv = newSV_type(SVt_PVMG); | |
| 342 | sv_magic(*sv, MUTABLE_SV(av), PERL_MAGIC_arylen, NULL, 0); | |
| 343 | } | |
| 344 | SETs(*sv); | |
| 345 | } else { | |
| 346 | SETs(sv_2mortal(newSViv( | |
| 347 | AvFILL(MUTABLE_AV(av)) + CopARYBASE_get(PL_curcop) | |
| 348 | ))); | |
| 79072805 | 349 | } |
| 79072805 LW |
350 | RETURN; |
| 351 | } | |
| 352 | ||
| a0d0e21e LW |
353 | PP(pp_pos) |
| 354 | { | |
| 2154eca7 | 355 | dVAR; dSP; dPOPss; |
| 8ec5e241 | 356 | |
| 78f9721b | 357 | if (PL_op->op_flags & OPf_MOD || LVRET) { |
| 2154eca7 EB |
358 | SV * const ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
| 359 | sv_magic(ret, NULL, PERL_MAGIC_pos, NULL, 0); | |
| 360 | LvTYPE(ret) = '.'; | |
| 361 | LvTARG(ret) = SvREFCNT_inc_simple(sv); | |
| 362 | PUSHs(ret); /* no SvSETMAGIC */ | |
| a0d0e21e LW |
363 | RETURN; |
| 364 | } | |
| 365 | else { | |
| a0d0e21e | 366 | if (SvTYPE(sv) >= SVt_PVMG && SvMAGIC(sv)) { |
| 1b6737cc | 367 | const MAGIC * const mg = mg_find(sv, PERL_MAGIC_regex_global); |
| 565764a8 | 368 | if (mg && mg->mg_len >= 0) { |
| 2154eca7 | 369 | dTARGET; |
| a0ed51b3 | 370 | I32 i = mg->mg_len; |
| 7e2040f0 | 371 | if (DO_UTF8(sv)) |
| a0ed51b3 | 372 | sv_pos_b2u(sv, &i); |
| fc15ae8f | 373 | PUSHi(i + CopARYBASE_get(PL_curcop)); |
| a0d0e21e LW |
374 | RETURN; |
| 375 | } | |
| 376 | } | |
| 377 | RETPUSHUNDEF; | |
| 378 | } | |
| 379 | } | |
| 380 | ||
| 79072805 LW |
381 | PP(pp_rv2cv) |
| 382 | { | |
| 97aff369 | 383 | dVAR; dSP; |
| 79072805 | 384 | GV *gv; |
| 1eced8f8 | 385 | HV *stash_unused; |
| c445ea15 AL |
386 | const I32 flags = (PL_op->op_flags & OPf_SPECIAL) |
| 387 | ? 0 | |
| 388 | : ((PL_op->op_private & (OPpLVAL_INTRO|OPpMAY_RETURN_CONSTANT)) == OPpMAY_RETURN_CONSTANT) | |
| 389 | ? GV_ADD|GV_NOEXPAND | |
| 390 | : GV_ADD; | |
| 4633a7c4 LW |
391 | /* We usually try to add a non-existent subroutine in case of AUTOLOAD. */ |
| 392 | /* (But not in defined().) */ | |
| e26df76a | 393 | |
| 1eced8f8 | 394 | CV *cv = sv_2cv(TOPs, &stash_unused, &gv, flags); |
| 07055b4c CS |
395 | if (cv) { |
| 396 | if (CvCLONE(cv)) | |
| ad64d0ec | 397 | cv = MUTABLE_CV(sv_2mortal(MUTABLE_SV(cv_clone(cv)))); |
| d32f2495 SC |
398 | if ((PL_op->op_private & OPpLVAL_INTRO)) { |
| 399 | if (gv && GvCV(gv) == cv && (gv = gv_autoload4(GvSTASH(gv), GvNAME(gv), GvNAMELEN(gv), FALSE))) | |
| 400 | cv = GvCV(gv); | |
| 401 | if (!CvLVALUE(cv)) | |
| 402 | DIE(aTHX_ "Can't modify non-lvalue subroutine call"); | |
| 403 | } | |
| 07055b4c | 404 | } |
| e26df76a | 405 | else if ((flags == (GV_ADD|GV_NOEXPAND)) && gv && SvROK(gv)) { |
| ea726b52 | 406 | cv = MUTABLE_CV(gv); |
| e26df76a | 407 | } |
| 07055b4c | 408 | else |
| ea726b52 | 409 | cv = MUTABLE_CV(&PL_sv_undef); |
| ad64d0ec | 410 | SETs(MUTABLE_SV(cv)); |
| 79072805 LW |
411 | RETURN; |
| 412 | } | |
| 413 | ||
| c07a80fd | 414 | PP(pp_prototype) |
| 415 | { | |
| 97aff369 | 416 | dVAR; dSP; |
| c07a80fd | 417 | CV *cv; |
| 418 | HV *stash; | |
| 419 | GV *gv; | |
| fabdb6c0 | 420 | SV *ret = &PL_sv_undef; |
| c07a80fd | 421 | |
| b6c543e3 | 422 | if (SvPOK(TOPs) && SvCUR(TOPs) >= 7) { |
| e3f73d4e | 423 | const char * s = SvPVX_const(TOPs); |
| b6c543e3 | 424 | if (strnEQ(s, "CORE::", 6)) { |
| 5458a98a | 425 | const int code = keyword(s + 6, SvCUR(TOPs) - 6, 1); |
| b6c543e3 IZ |
426 | if (code < 0) { /* Overridable. */ |
| 427 | #define MAX_ARGS_OP ((sizeof(I32) - 1) * 2) | |
| 59b085e1 | 428 | int i = 0, n = 0, seen_question = 0, defgv = 0; |
| b6c543e3 IZ |
429 | I32 oa; |
| 430 | char str[ MAX_ARGS_OP * 2 + 2 ]; /* One ';', one '\0' */ | |
| 431 | ||
| bdf1bb36 | 432 | if (code == -KEY_chop || code == -KEY_chomp |
| f23102e2 | 433 | || code == -KEY_exec || code == -KEY_system) |
| 77bc9082 | 434 | goto set; |
| d116c547 | 435 | if (code == -KEY_mkdir) { |
| 84bafc02 | 436 | ret = newSVpvs_flags("_;$", SVs_TEMP); |
| d116c547 RGS |
437 | goto set; |
| 438 | } | |
| 7c8178a1 | 439 | if (code == -KEY_keys || code == -KEY_values || code == -KEY_each) { |
| cba5a3b0 DG |
440 | ret = newSVpvs_flags("+", SVs_TEMP); |
| 441 | goto set; | |
| 442 | } | |
| 443 | if (code == -KEY_push || code == -KEY_unshift) { | |
| 444 | ret = newSVpvs_flags("+@", SVs_TEMP); | |
| 445 | goto set; | |
| 446 | } | |
| 447 | if (code == -KEY_pop || code == -KEY_shift) { | |
| 448 | ret = newSVpvs_flags(";+", SVs_TEMP); | |
| 449 | goto set; | |
| 450 | } | |
| 451 | if (code == -KEY_splice) { | |
| 452 | ret = newSVpvs_flags("+;$$@", SVs_TEMP); | |
| 1db4d195 FC |
453 | goto set; |
| 454 | } | |
| 455 | if (code == -KEY_tied || code == -KEY_untie) { | |
| 456 | ret = newSVpvs_flags("\\[$@%*]", SVs_TEMP); | |
| 457 | goto set; | |
| 458 | } | |
| 459 | if (code == -KEY_tie) { | |
| 460 | ret = newSVpvs_flags("\\[$@%*]$@", SVs_TEMP); | |
| 7c8178a1 RGS |
461 | goto set; |
| 462 | } | |
| e3f73d4e RGS |
463 | if (code == -KEY_readpipe) { |
| 464 | s = "CORE::backtick"; | |
| 465 | } | |
| b6c543e3 | 466 | while (i < MAXO) { /* The slow way. */ |
| 22c35a8c GS |
467 | if (strEQ(s + 6, PL_op_name[i]) |
| 468 | || strEQ(s + 6, PL_op_desc[i])) | |
| 469 | { | |
| b6c543e3 | 470 | goto found; |
| 22c35a8c | 471 | } |
| b6c543e3 IZ |
472 | i++; |
| 473 | } | |
| 474 | goto nonesuch; /* Should not happen... */ | |
| 475 | found: | |
| 59b085e1 | 476 | defgv = PL_opargs[i] & OA_DEFGV; |
| 22c35a8c | 477 | oa = PL_opargs[i] >> OASHIFT; |
| b6c543e3 | 478 | while (oa) { |
| 59b085e1 | 479 | if (oa & OA_OPTIONAL && !seen_question && !defgv) { |
| b6c543e3 IZ |
480 | seen_question = 1; |
| 481 | str[n++] = ';'; | |
| ef54e1a4 | 482 | } |
| b13b2135 | 483 | if ((oa & (OA_OPTIONAL - 1)) >= OA_AVREF |
| 6e97e420 SC |
484 | && (oa & (OA_OPTIONAL - 1)) <= OA_SCALARREF |
| 485 | /* But globs are already references (kinda) */ | |
| 486 | && (oa & (OA_OPTIONAL - 1)) != OA_FILEREF | |
| 487 | ) { | |
| b6c543e3 IZ |
488 | str[n++] = '\\'; |
| 489 | } | |
| b6c543e3 IZ |
490 | str[n++] = ("?$@@%&*$")[oa & (OA_OPTIONAL - 1)]; |
| 491 | oa = oa >> 4; | |
| 492 | } | |
| 59b085e1 RGS |
493 | if (defgv && str[n - 1] == '$') |
| 494 | str[n - 1] = '_'; | |
| b6c543e3 | 495 | str[n++] = '\0'; |
| 59cd0e26 | 496 | ret = newSVpvn_flags(str, n - 1, SVs_TEMP); |
| ef54e1a4 JH |
497 | } |
| 498 | else if (code) /* Non-Overridable */ | |
| b6c543e3 IZ |
499 | goto set; |
| 500 | else { /* None such */ | |
| 501 | nonesuch: | |
| d470f89e | 502 | DIE(aTHX_ "Can't find an opnumber for \"%s\"", s+6); |
| b6c543e3 IZ |
503 | } |
| 504 | } | |
| 505 | } | |
| f2c0649b | 506 | cv = sv_2cv(TOPs, &stash, &gv, 0); |
| 5f05dabc | 507 | if (cv && SvPOK(cv)) |
| 59cd0e26 | 508 | ret = newSVpvn_flags(SvPVX_const(cv), SvCUR(cv), SVs_TEMP); |
| b6c543e3 | 509 | set: |
| c07a80fd | 510 | SETs(ret); |
| 511 | RETURN; | |
| 512 | } | |
| 513 | ||
| a0d0e21e LW |
514 | PP(pp_anoncode) |
| 515 | { | |
| 97aff369 | 516 | dVAR; dSP; |
| ea726b52 | 517 | CV *cv = MUTABLE_CV(PAD_SV(PL_op->op_targ)); |
| a5f75d66 | 518 | if (CvCLONE(cv)) |
| ad64d0ec | 519 | cv = MUTABLE_CV(sv_2mortal(MUTABLE_SV(cv_clone(cv)))); |
| 5f05dabc | 520 | EXTEND(SP,1); |
| ad64d0ec | 521 | PUSHs(MUTABLE_SV(cv)); |
| a0d0e21e LW |
522 | RETURN; |
| 523 | } | |
| 524 | ||
| 525 | PP(pp_srefgen) | |
| 79072805 | 526 | { |
| 97aff369 | 527 | dVAR; dSP; |
| 71be2cbc | 528 | *SP = refto(*SP); |
| 79072805 | 529 | RETURN; |
| 8ec5e241 | 530 | } |
| a0d0e21e LW |
531 | |
| 532 | PP(pp_refgen) | |
| 533 | { | |
| 97aff369 | 534 | dVAR; dSP; dMARK; |
| a0d0e21e | 535 | if (GIMME != G_ARRAY) { |
| 5f0b1d4e GS |
536 | if (++MARK <= SP) |
| 537 | *MARK = *SP; | |
| 538 | else | |
| 3280af22 | 539 | *MARK = &PL_sv_undef; |
| 5f0b1d4e GS |
540 | *MARK = refto(*MARK); |
| 541 | SP = MARK; | |
| 542 | RETURN; | |
| a0d0e21e | 543 | } |
| bbce6d69 | 544 | EXTEND_MORTAL(SP - MARK); |
| 71be2cbc | 545 | while (++MARK <= SP) |
| 546 | *MARK = refto(*MARK); | |
| a0d0e21e | 547 | RETURN; |
| 79072805 LW |
548 | } |
| 549 | ||
| 76e3520e | 550 | STATIC SV* |
| cea2e8a9 | 551 | S_refto(pTHX_ SV *sv) |
| 71be2cbc | 552 | { |
| 97aff369 | 553 | dVAR; |
| 71be2cbc | 554 | SV* rv; |
| 555 | ||
| 7918f24d NC |
556 | PERL_ARGS_ASSERT_REFTO; |
| 557 | ||
| 71be2cbc | 558 | if (SvTYPE(sv) == SVt_PVLV && LvTYPE(sv) == 'y') { |
| 559 | if (LvTARGLEN(sv)) | |
| 68dc0745 | 560 | vivify_defelem(sv); |
| 561 | if (!(sv = LvTARG(sv))) | |
| 3280af22 | 562 | sv = &PL_sv_undef; |
| 0dd88869 | 563 | else |
| b37c2d43 | 564 | SvREFCNT_inc_void_NN(sv); |
| 71be2cbc | 565 | } |
| d8b46c1b | 566 | else if (SvTYPE(sv) == SVt_PVAV) { |
| 502c6561 NC |
567 | if (!AvREAL((const AV *)sv) && AvREIFY((const AV *)sv)) |
| 568 | av_reify(MUTABLE_AV(sv)); | |
| d8b46c1b | 569 | SvTEMP_off(sv); |
| b37c2d43 | 570 | SvREFCNT_inc_void_NN(sv); |
| d8b46c1b | 571 | } |
| f2933f5f DM |
572 | else if (SvPADTMP(sv) && !IS_PADGV(sv)) |
| 573 | sv = newSVsv(sv); | |
| 71be2cbc | 574 | else { |
| 575 | SvTEMP_off(sv); | |
| b37c2d43 | 576 | SvREFCNT_inc_void_NN(sv); |
| 71be2cbc | 577 | } |
| 578 | rv = sv_newmortal(); | |
| 4df7f6af | 579 | sv_upgrade(rv, SVt_IV); |
| b162af07 | 580 | SvRV_set(rv, sv); |
| 71be2cbc | 581 | SvROK_on(rv); |
| 582 | return rv; | |
| 583 | } | |
| 584 | ||
| 79072805 LW |
585 | PP(pp_ref) |
| 586 | { | |
| 97aff369 | 587 | dVAR; dSP; dTARGET; |
| e1ec3a88 | 588 | const char *pv; |
| 1b6737cc | 589 | SV * const sv = POPs; |
| f12c7020 | 590 | |
| 5b295bef RD |
591 | if (sv) |
| 592 | SvGETMAGIC(sv); | |
| f12c7020 | 593 | |
| a0d0e21e | 594 | if (!sv || !SvROK(sv)) |
| 4633a7c4 | 595 | RETPUSHNO; |
| 79072805 | 596 | |
| cba0b539 FR |
597 | pv = sv_reftype(SvRV(sv),TRUE); |
| 598 | PUSHp(pv, strlen(pv)); | |
| 79072805 LW |
599 | RETURN; |
| 600 | } | |
| 601 | ||
| 602 | PP(pp_bless) | |
| 603 | { | |
| 97aff369 | 604 | dVAR; dSP; |
| 463ee0b2 | 605 | HV *stash; |
| 79072805 | 606 | |
| 463ee0b2 | 607 | if (MAXARG == 1) |
| 11faa288 | 608 | stash = CopSTASH(PL_curcop); |
| 7b8d334a | 609 | else { |
| 1b6737cc | 610 | SV * const ssv = POPs; |
| 7b8d334a | 611 | STRLEN len; |
| e1ec3a88 | 612 | const char *ptr; |
| 81689caa | 613 | |
| 016a42f3 | 614 | if (ssv && !SvGMAGICAL(ssv) && !SvAMAGIC(ssv) && SvROK(ssv)) |
| 81689caa | 615 | Perl_croak(aTHX_ "Attempt to bless into a reference"); |
| 5c144d81 | 616 | ptr = SvPV_const(ssv,len); |
| a2a5de95 NC |
617 | if (len == 0) |
| 618 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), | |
| 619 | "Explicit blessing to '' (assuming package main)"); | |
| da51bb9b | 620 | stash = gv_stashpvn(ptr, len, GV_ADD); |
| 7b8d334a | 621 | } |
| a0d0e21e | 622 | |
| 5d3fdfeb | 623 | (void)sv_bless(TOPs, stash); |
| 79072805 LW |
624 | RETURN; |
| 625 | } | |
| 626 | ||
| fb73857a | 627 | PP(pp_gelem) |
| 628 | { | |
| 97aff369 | 629 | dVAR; dSP; |
| b13b2135 | 630 | |
| 1b6737cc AL |
631 | SV *sv = POPs; |
| 632 | const char * const elem = SvPV_nolen_const(sv); | |
| 159b6efe | 633 | GV * const gv = MUTABLE_GV(POPs); |
| c445ea15 | 634 | SV * tmpRef = NULL; |
| 1b6737cc | 635 | |
| c445ea15 | 636 | sv = NULL; |
| c4ba80c3 NC |
637 | if (elem) { |
| 638 | /* elem will always be NUL terminated. */ | |
| 1b6737cc | 639 | const char * const second_letter = elem + 1; |
| c4ba80c3 NC |
640 | switch (*elem) { |
| 641 | case 'A': | |
| 1b6737cc | 642 | if (strEQ(second_letter, "RRAY")) |
| ad64d0ec | 643 | tmpRef = MUTABLE_SV(GvAV(gv)); |
| c4ba80c3 NC |
644 | break; |
| 645 | case 'C': | |
| 1b6737cc | 646 | if (strEQ(second_letter, "ODE")) |
| ad64d0ec | 647 | tmpRef = MUTABLE_SV(GvCVu(gv)); |
| c4ba80c3 NC |
648 | break; |
| 649 | case 'F': | |
| 1b6737cc | 650 | if (strEQ(second_letter, "ILEHANDLE")) { |
| c4ba80c3 NC |
651 | /* finally deprecated in 5.8.0 */ |
| 652 | deprecate("*glob{FILEHANDLE}"); | |
| ad64d0ec | 653 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
| c4ba80c3 NC |
654 | } |
| 655 | else | |
| 1b6737cc | 656 | if (strEQ(second_letter, "ORMAT")) |
| ad64d0ec | 657 | tmpRef = MUTABLE_SV(GvFORM(gv)); |
| c4ba80c3 NC |
658 | break; |
| 659 | case 'G': | |
| 1b6737cc | 660 | if (strEQ(second_letter, "LOB")) |
| ad64d0ec | 661 | tmpRef = MUTABLE_SV(gv); |
| c4ba80c3 NC |
662 | break; |
| 663 | case 'H': | |
| 1b6737cc | 664 | if (strEQ(second_letter, "ASH")) |
| ad64d0ec | 665 | tmpRef = MUTABLE_SV(GvHV(gv)); |
| c4ba80c3 NC |
666 | break; |
| 667 | case 'I': | |
| 1b6737cc | 668 | if (*second_letter == 'O' && !elem[2]) |
| ad64d0ec | 669 | tmpRef = MUTABLE_SV(GvIOp(gv)); |
| c4ba80c3 NC |
670 | break; |
| 671 | case 'N': | |
| 1b6737cc | 672 | if (strEQ(second_letter, "AME")) |
| a663657d | 673 | sv = newSVhek(GvNAME_HEK(gv)); |
| c4ba80c3 NC |
674 | break; |
| 675 | case 'P': | |
| 1b6737cc | 676 | if (strEQ(second_letter, "ACKAGE")) { |
| 7fa3a4ab NC |
677 | const HV * const stash = GvSTASH(gv); |
| 678 | const HEK * const hek = stash ? HvNAME_HEK(stash) : NULL; | |
| 396482e1 | 679 | sv = hek ? newSVhek(hek) : newSVpvs("__ANON__"); |
| c4ba80c3 NC |
680 | } |
| 681 | break; | |
| 682 | case 'S': | |
| 1b6737cc | 683 | if (strEQ(second_letter, "CALAR")) |
| f9d52e31 | 684 | tmpRef = GvSVn(gv); |
| c4ba80c3 | 685 | break; |
| 39b99f21 | 686 | } |
| fb73857a | 687 | } |
| 76e3520e GS |
688 | if (tmpRef) |
| 689 | sv = newRV(tmpRef); | |
| fb73857a | 690 | if (sv) |
| 691 | sv_2mortal(sv); | |
| 692 | else | |
| 3280af22 | 693 | sv = &PL_sv_undef; |
| fb73857a | 694 | XPUSHs(sv); |
| 695 | RETURN; | |
| 696 | } | |
| 697 | ||
| a0d0e21e | 698 | /* Pattern matching */ |
| 79072805 | 699 | |
| a0d0e21e | 700 | PP(pp_study) |
| 79072805 | 701 | { |
| 97aff369 | 702 | dVAR; dSP; dPOPss; |
| a0d0e21e LW |
703 | register unsigned char *s; |
| 704 | register I32 pos; | |
| 705 | register I32 ch; | |
| 706 | register I32 *sfirst; | |
| 707 | register I32 *snext; | |
| a0d0e21e LW |
708 | STRLEN len; |
| 709 | ||
| 3280af22 | 710 | if (sv == PL_lastscream) { |
| 1e422769 | 711 | if (SvSCREAM(sv)) |
| 712 | RETPUSHYES; | |
| 713 | } | |
| a4f4e906 NC |
714 | s = (unsigned char*)(SvPV(sv, len)); |
| 715 | pos = len; | |
| c9b9f909 | 716 | if (pos <= 0 || !SvPOK(sv) || SvUTF8(sv)) { |
| a4f4e906 NC |
717 | /* No point in studying a zero length string, and not safe to study |
| 718 | anything that doesn't appear to be a simple scalar (and hence might | |
| 719 | change between now and when the regexp engine runs without our set | |
| bd473224 | 720 | magic ever running) such as a reference to an object with overloaded |
| a4f4e906 NC |
721 | stringification. */ |
| 722 | RETPUSHNO; | |
| 723 | } | |
| 724 | ||
| 725 | if (PL_lastscream) { | |
| 726 | SvSCREAM_off(PL_lastscream); | |
| 727 | SvREFCNT_dec(PL_lastscream); | |
| c07a80fd | 728 | } |
| b37c2d43 | 729 | PL_lastscream = SvREFCNT_inc_simple(sv); |
| 1e422769 | 730 | |
| 731 | s = (unsigned char*)(SvPV(sv, len)); | |
| 732 | pos = len; | |
| 733 | if (pos <= 0) | |
| 734 | RETPUSHNO; | |
| 3280af22 NIS |
735 | if (pos > PL_maxscream) { |
| 736 | if (PL_maxscream < 0) { | |
| 737 | PL_maxscream = pos + 80; | |
| a02a5408 JC |
738 | Newx(PL_screamfirst, 256, I32); |
| 739 | Newx(PL_screamnext, PL_maxscream, I32); | |
| 79072805 LW |
740 | } |
| 741 | else { | |
| 3280af22 NIS |
742 | PL_maxscream = pos + pos / 4; |
| 743 | Renew(PL_screamnext, PL_maxscream, I32); | |
| 79072805 | 744 | } |
| 79072805 | 745 | } |
| a0d0e21e | 746 | |
| 3280af22 NIS |
747 | sfirst = PL_screamfirst; |
| 748 | snext = PL_screamnext; | |
| a0d0e21e LW |
749 | |
| 750 | if (!sfirst || !snext) | |
| cea2e8a9 | 751 | DIE(aTHX_ "do_study: out of memory"); |
| a0d0e21e LW |
752 | |
| 753 | for (ch = 256; ch; --ch) | |
| 754 | *sfirst++ = -1; | |
| 755 | sfirst -= 256; | |
| 756 | ||
| 757 | while (--pos >= 0) { | |
| 1b6737cc | 758 | register const I32 ch = s[pos]; |
| a0d0e21e LW |
759 | if (sfirst[ch] >= 0) |
| 760 | snext[pos] = sfirst[ch] - pos; | |
| 761 | else | |
| 762 | snext[pos] = -pos; | |
| 763 | sfirst[ch] = pos; | |
| 79072805 LW |
764 | } |
| 765 | ||
| c07a80fd | 766 | SvSCREAM_on(sv); |
| 14befaf4 | 767 | /* piggyback on m//g magic */ |
| c445ea15 | 768 | sv_magic(sv, NULL, PERL_MAGIC_regex_global, NULL, 0); |
| 1e422769 | 769 | RETPUSHYES; |
| 79072805 LW |
770 | } |
| 771 | ||
| a0d0e21e | 772 | PP(pp_trans) |
| 79072805 | 773 | { |
| 97aff369 | 774 | dVAR; dSP; dTARG; |
| a0d0e21e LW |
775 | SV *sv; |
| 776 | ||
| 533c011a | 777 | if (PL_op->op_flags & OPf_STACKED) |
| a0d0e21e | 778 | sv = POPs; |
| 59f00321 RGS |
779 | else if (PL_op->op_private & OPpTARGET_MY) |
| 780 | sv = GETTARGET; | |
| 79072805 | 781 | else { |
| 54b9620d | 782 | sv = DEFSV; |
| a0d0e21e | 783 | EXTEND(SP,1); |
| 79072805 | 784 | } |
| adbc6bb1 | 785 | TARG = sv_newmortal(); |
| bb16bae8 FC |
786 | if(PL_op->op_type == OP_TRANSR) { |
| 787 | SV * const newsv = newSVsv(sv); | |
| 788 | do_trans(newsv); | |
| 789 | mPUSHs(newsv); | |
| 790 | } | |
| 791 | else PUSHi(do_trans(sv)); | |
| a0d0e21e | 792 | RETURN; |
| 79072805 LW |
793 | } |
| 794 | ||
| a0d0e21e | 795 | /* Lvalue operators. */ |
| 79072805 | 796 | |
| 81745e4e NC |
797 | static void |
| 798 | S_do_chomp(pTHX_ SV *retval, SV *sv, bool chomping) | |
| 799 | { | |
| 800 | dVAR; | |
| 801 | STRLEN len; | |
| 802 | char *s; | |
| 803 | ||
| 804 | PERL_ARGS_ASSERT_DO_CHOMP; | |
| 805 | ||
| 806 | if (chomping && (RsSNARF(PL_rs) || RsRECORD(PL_rs))) | |
| 807 | return; | |
| 808 | if (SvTYPE(sv) == SVt_PVAV) { | |
| 809 | I32 i; | |
| 810 | AV *const av = MUTABLE_AV(sv); | |
| 811 | const I32 max = AvFILL(av); | |
| 812 | ||
| 813 | for (i = 0; i <= max; i++) { | |
| 814 | sv = MUTABLE_SV(av_fetch(av, i, FALSE)); | |
| 815 | if (sv && ((sv = *(SV**)sv), sv != &PL_sv_undef)) | |
| 816 | do_chomp(retval, sv, chomping); | |
| 817 | } | |
| 818 | return; | |
| 819 | } | |
| 820 | else if (SvTYPE(sv) == SVt_PVHV) { | |
| 821 | HV* const hv = MUTABLE_HV(sv); | |
| 822 | HE* entry; | |
| 823 | (void)hv_iterinit(hv); | |
| 824 | while ((entry = hv_iternext(hv))) | |
| 825 | do_chomp(retval, hv_iterval(hv,entry), chomping); | |
| 826 | return; | |
| 827 | } | |
| 828 | else if (SvREADONLY(sv)) { | |
| 829 | if (SvFAKE(sv)) { | |
| 830 | /* SV is copy-on-write */ | |
| 831 | sv_force_normal_flags(sv, 0); | |
| 832 | } | |
| 833 | if (SvREADONLY(sv)) | |
| 834 | Perl_croak_no_modify(aTHX); | |
| 835 | } | |
| 836 | ||
| 837 | if (PL_encoding) { | |
| 838 | if (!SvUTF8(sv)) { | |
| 839 | /* XXX, here sv is utf8-ized as a side-effect! | |
| 840 | If encoding.pm is used properly, almost string-generating | |
| 841 | operations, including literal strings, chr(), input data, etc. | |
| 842 | should have been utf8-ized already, right? | |
| 843 | */ | |
| 844 | sv_recode_to_utf8(sv, PL_encoding); | |
| 845 | } | |
| 846 | } | |
| 847 | ||
| 848 | s = SvPV(sv, len); | |
| 849 | if (chomping) { | |
| 850 | char *temp_buffer = NULL; | |
| 851 | SV *svrecode = NULL; | |
| 852 | ||
| 853 | if (s && len) { | |
| 854 | s += --len; | |
| 855 | if (RsPARA(PL_rs)) { | |
| 856 | if (*s != '\n') | |
| 857 | goto nope; | |
| 858 | ++SvIVX(retval); | |
| 859 | while (len && s[-1] == '\n') { | |
| 860 | --len; | |
| 861 | --s; | |
| 862 | ++SvIVX(retval); | |
| 863 | } | |
| 864 | } | |
| 865 | else { | |
| 866 | STRLEN rslen, rs_charlen; | |
| 867 | const char *rsptr = SvPV_const(PL_rs, rslen); | |
| 868 | ||
| 869 | rs_charlen = SvUTF8(PL_rs) | |
| 870 | ? sv_len_utf8(PL_rs) | |
| 871 | : rslen; | |
| 872 | ||
| 873 | if (SvUTF8(PL_rs) != SvUTF8(sv)) { | |
| 874 | /* Assumption is that rs is shorter than the scalar. */ | |
| 875 | if (SvUTF8(PL_rs)) { | |
| 876 | /* RS is utf8, scalar is 8 bit. */ | |
| 877 | bool is_utf8 = TRUE; | |
| 878 | temp_buffer = (char*)bytes_from_utf8((U8*)rsptr, | |
| 879 | &rslen, &is_utf8); | |
| 880 | if (is_utf8) { | |
| 881 | /* Cannot downgrade, therefore cannot possibly match | |
| 882 | */ | |
| 883 | assert (temp_buffer == rsptr); | |
| 884 | temp_buffer = NULL; | |
| 885 | goto nope; | |
| 886 | } | |
| 887 | rsptr = temp_buffer; | |
| 888 | } | |
| 889 | else if (PL_encoding) { | |
| 890 | /* RS is 8 bit, encoding.pm is used. | |
| 891 | * Do not recode PL_rs as a side-effect. */ | |
| 892 | svrecode = newSVpvn(rsptr, rslen); | |
| 893 | sv_recode_to_utf8(svrecode, PL_encoding); | |
| 894 | rsptr = SvPV_const(svrecode, rslen); | |
| 895 | rs_charlen = sv_len_utf8(svrecode); | |
| 896 | } | |
| 897 | else { | |
| 898 | /* RS is 8 bit, scalar is utf8. */ | |
| 899 | temp_buffer = (char*)bytes_to_utf8((U8*)rsptr, &rslen); | |
| 900 | rsptr = temp_buffer; | |
| 901 | } | |
| 902 | } | |
| 903 | if (rslen == 1) { | |
| 904 | if (*s != *rsptr) | |
| 905 | goto nope; | |
| 906 | ++SvIVX(retval); | |
| 907 | } | |
| 908 | else { | |
| 909 | if (len < rslen - 1) | |
| 910 | goto nope; | |
| 911 | len -= rslen - 1; | |
| 912 | s -= rslen - 1; | |
| 913 | if (memNE(s, rsptr, rslen)) | |
| 914 | goto nope; | |
| 915 | SvIVX(retval) += rs_charlen; | |
| 916 | } | |
| 917 | } | |
| 918 | s = SvPV_force_nolen(sv); | |
| 919 | SvCUR_set(sv, len); | |
| 920 | *SvEND(sv) = '\0'; | |
| 921 | SvNIOK_off(sv); | |
| 922 | SvSETMAGIC(sv); | |
| 923 | } | |
| 924 | nope: | |
| 925 | ||
| 926 | SvREFCNT_dec(svrecode); | |
| 927 | ||
| 928 | Safefree(temp_buffer); | |
| 929 | } else { | |
| 930 | if (len && !SvPOK(sv)) | |
| 931 | s = SvPV_force_nomg(sv, len); | |
| 932 | if (DO_UTF8(sv)) { | |
| 933 | if (s && len) { | |
| 934 | char * const send = s + len; | |
| 935 | char * const start = s; | |
| 936 | s = send - 1; | |
| 937 | while (s > start && UTF8_IS_CONTINUATION(*s)) | |
| 938 | s--; | |
| 939 | if (is_utf8_string((U8*)s, send - s)) { | |
| 940 | sv_setpvn(retval, s, send - s); | |
| 941 | *s = '\0'; | |
| 942 | SvCUR_set(sv, s - start); | |
| 943 | SvNIOK_off(sv); | |
| 944 | SvUTF8_on(retval); | |
| 945 | } | |
| 946 | } | |
| 947 | else | |
| 948 | sv_setpvs(retval, ""); | |
| 949 | } | |
| 950 | else if (s && len) { | |
| 951 | s += --len; | |
| 952 | sv_setpvn(retval, s, 1); | |
| 953 | *s = '\0'; | |
| 954 | SvCUR_set(sv, len); | |
| 955 | SvUTF8_off(sv); | |
| 956 | SvNIOK_off(sv); | |
| 957 | } | |
| 958 | else | |
| 959 | sv_setpvs(retval, ""); | |
| 960 | SvSETMAGIC(sv); | |
| 961 | } | |
| 962 | } | |
| 963 | ||
| a0d0e21e LW |
964 | PP(pp_schop) |
| 965 | { | |
| 97aff369 | 966 | dVAR; dSP; dTARGET; |
| fa54efae NC |
967 | const bool chomping = PL_op->op_type == OP_SCHOMP; |
| 968 | ||
| 969 | if (chomping) | |
| 970 | sv_setiv(TARG, 0); | |
| 971 | do_chomp(TARG, TOPs, chomping); | |
| a0d0e21e LW |
972 | SETTARG; |
| 973 | RETURN; | |
| 79072805 LW |
974 | } |
| 975 | ||
| a0d0e21e | 976 | PP(pp_chop) |
| 79072805 | 977 | { |
| 97aff369 | 978 | dVAR; dSP; dMARK; dTARGET; dORIGMARK; |
| fa54efae | 979 | const bool chomping = PL_op->op_type == OP_CHOMP; |
| 8ec5e241 | 980 | |
| fa54efae NC |
981 | if (chomping) |
| 982 | sv_setiv(TARG, 0); | |
| 20cf1f79 | 983 | while (MARK < SP) |
| fa54efae | 984 | do_chomp(TARG, *++MARK, chomping); |
| 20cf1f79 NC |
985 | SP = ORIGMARK; |
| 986 | XPUSHTARG; | |
| a0d0e21e | 987 | RETURN; |
| 79072805 LW |
988 | } |
| 989 | ||
| a0d0e21e LW |
990 | PP(pp_undef) |
| 991 | { | |
| 97aff369 | 992 | dVAR; dSP; |
| a0d0e21e LW |
993 | SV *sv; |
| 994 | ||
| 533c011a | 995 | if (!PL_op->op_private) { |
| 774d564b | 996 | EXTEND(SP, 1); |
| a0d0e21e | 997 | RETPUSHUNDEF; |
| 774d564b | 998 | } |
| 79072805 | 999 | |
| a0d0e21e LW |
1000 | sv = POPs; |
| 1001 | if (!sv) | |
| 1002 | RETPUSHUNDEF; | |
| 85e6fe83 | 1003 | |
| 765f542d | 1004 | SV_CHECK_THINKFIRST_COW_DROP(sv); |
| 85e6fe83 | 1005 | |
| a0d0e21e LW |
1006 | switch (SvTYPE(sv)) { |
| 1007 | case SVt_NULL: | |
| 1008 | break; | |
| 1009 | case SVt_PVAV: | |
| 502c6561 | 1010 | av_undef(MUTABLE_AV(sv)); |
| a0d0e21e LW |
1011 | break; |
| 1012 | case SVt_PVHV: | |
| 85fbaab2 | 1013 | hv_undef(MUTABLE_HV(sv)); |
| a0d0e21e LW |
1014 | break; |
| 1015 | case SVt_PVCV: | |
| a2a5de95 NC |
1016 | if (cv_const_sv((const CV *)sv)) |
| 1017 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "Constant subroutine %s undefined", | |
| 1018 | CvANON((const CV *)sv) ? "(anonymous)" | |
| 1019 | : GvENAME(CvGV((const CV *)sv))); | |
| 5f66b61c | 1020 | /* FALLTHROUGH */ |
| 9607fc9c | 1021 | case SVt_PVFM: |
| 6fc92669 GS |
1022 | { |
| 1023 | /* let user-undef'd sub keep its identity */ | |
| ea726b52 NC |
1024 | GV* const gv = CvGV((const CV *)sv); |
| 1025 | cv_undef(MUTABLE_CV(sv)); | |
| b3f91e91 | 1026 | CvGV_set(MUTABLE_CV(sv), gv); |
| 6fc92669 | 1027 | } |
| a0d0e21e | 1028 | break; |
| 8e07c86e | 1029 | case SVt_PVGV: |
| 6e592b3a | 1030 | if (SvFAKE(sv)) { |
| 3280af22 | 1031 | SvSetMagicSV(sv, &PL_sv_undef); |
| 6e592b3a BM |
1032 | break; |
| 1033 | } | |
| 1034 | else if (isGV_with_GP(sv)) { | |
| 20408e3c | 1035 | GP *gp; |
| dd69841b BB |
1036 | HV *stash; |
| 1037 | ||
| dd69841b | 1038 | /* undef *Pkg::meth_name ... */ |
| e530fb81 FC |
1039 | bool method_changed |
| 1040 | = GvCVu((const GV *)sv) && (stash = GvSTASH((const GV *)sv)) | |
| 1041 | && HvENAME_get(stash); | |
| 1042 | /* undef *Foo:: */ | |
| 1043 | if((stash = GvHV((const GV *)sv))) { | |
| 1044 | if(HvENAME_get(stash)) | |
| 1045 | SvREFCNT_inc_simple_void_NN(sv_2mortal((SV *)stash)); | |
| 1046 | else stash = NULL; | |
| 1047 | } | |
| dd69841b | 1048 | |
| 159b6efe | 1049 | gp_free(MUTABLE_GV(sv)); |
| a02a5408 | 1050 | Newxz(gp, 1, GP); |
| c43ae56f | 1051 | GvGP_set(sv, gp_ref(gp)); |
| 561b68a9 | 1052 | GvSV(sv) = newSV(0); |
| 57843af0 | 1053 | GvLINE(sv) = CopLINE(PL_curcop); |
| 159b6efe | 1054 | GvEGV(sv) = MUTABLE_GV(sv); |
| 20408e3c | 1055 | GvMULTI_on(sv); |
| e530fb81 FC |
1056 | |
| 1057 | if(stash) | |
| afdbe55d | 1058 | mro_package_moved(NULL, stash, (const GV *)sv, 0); |
| e530fb81 FC |
1059 | stash = NULL; |
| 1060 | /* undef *Foo::ISA */ | |
| 1061 | if( strEQ(GvNAME((const GV *)sv), "ISA") | |
| 1062 | && (stash = GvSTASH((const GV *)sv)) | |
| 1063 | && (method_changed || HvENAME(stash)) ) | |
| 1064 | mro_isa_changed_in(stash); | |
| 1065 | else if(method_changed) | |
| 1066 | mro_method_changed_in( | |
| da9043f5 | 1067 | GvSTASH((const GV *)sv) |
| e530fb81 FC |
1068 | ); |
| 1069 | ||
| 6e592b3a | 1070 | break; |
| 20408e3c | 1071 | } |
| 6e592b3a | 1072 | /* FALL THROUGH */ |
| a0d0e21e | 1073 | default: |
| b15aece3 | 1074 | if (SvTYPE(sv) >= SVt_PV && SvPVX_const(sv) && SvLEN(sv)) { |
| 8bd4d4c5 | 1075 | SvPV_free(sv); |
| c445ea15 | 1076 | SvPV_set(sv, NULL); |
| 4633a7c4 | 1077 | SvLEN_set(sv, 0); |
| a0d0e21e | 1078 | } |
| 0c34ef67 | 1079 | SvOK_off(sv); |
| 4633a7c4 | 1080 | SvSETMAGIC(sv); |
| 79072805 | 1081 | } |
| a0d0e21e LW |
1082 | |
| 1083 | RETPUSHUNDEF; | |
| 79072805 LW |
1084 | } |
| 1085 | ||
| a0d0e21e | 1086 | PP(pp_predec) |
| 79072805 | 1087 | { |
| 97aff369 | 1088 | dVAR; dSP; |
| 6e592b3a | 1089 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
| 6ad8f254 | 1090 | Perl_croak_no_modify(aTHX); |
| 3510b4a1 NC |
1091 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
| 1092 | && SvIVX(TOPs) != IV_MIN) | |
| 55497cff | 1093 | { |
| 45977657 | 1094 | SvIV_set(TOPs, SvIVX(TOPs) - 1); |
| 55497cff | 1095 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
| 748a9306 LW |
1096 | } |
| 1097 | else | |
| 1098 | sv_dec(TOPs); | |
| a0d0e21e LW |
1099 | SvSETMAGIC(TOPs); |
| 1100 | return NORMAL; | |
| 1101 | } | |
| 79072805 | 1102 | |
| a0d0e21e LW |
1103 | PP(pp_postinc) |
| 1104 | { | |
| 97aff369 | 1105 | dVAR; dSP; dTARGET; |
| 6e592b3a | 1106 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
| 6ad8f254 | 1107 | Perl_croak_no_modify(aTHX); |
| 7dcb9b98 DM |
1108 | if (SvROK(TOPs)) |
| 1109 | TARG = sv_newmortal(); | |
| a0d0e21e | 1110 | sv_setsv(TARG, TOPs); |
| 3510b4a1 NC |
1111 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
| 1112 | && SvIVX(TOPs) != IV_MAX) | |
| 55497cff | 1113 | { |
| 45977657 | 1114 | SvIV_set(TOPs, SvIVX(TOPs) + 1); |
| 55497cff | 1115 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
| 748a9306 LW |
1116 | } |
| 1117 | else | |
| 6f1401dc | 1118 | sv_inc_nomg(TOPs); |
| a0d0e21e | 1119 | SvSETMAGIC(TOPs); |
| 1e54a23f | 1120 | /* special case for undef: see thread at 2003-03/msg00536.html in archive */ |
| a0d0e21e LW |
1121 | if (!SvOK(TARG)) |
| 1122 | sv_setiv(TARG, 0); | |
| 1123 | SETs(TARG); | |
| 1124 | return NORMAL; | |
| 1125 | } | |
| 79072805 | 1126 | |
| a0d0e21e LW |
1127 | PP(pp_postdec) |
| 1128 | { | |
| 97aff369 | 1129 | dVAR; dSP; dTARGET; |
| 6e592b3a | 1130 | if (SvTYPE(TOPs) >= SVt_PVAV || isGV_with_GP(TOPs)) |
| 6ad8f254 | 1131 | Perl_croak_no_modify(aTHX); |
| 7dcb9b98 DM |
1132 | if (SvROK(TOPs)) |
| 1133 | TARG = sv_newmortal(); | |
| a0d0e21e | 1134 | sv_setsv(TARG, TOPs); |
| 3510b4a1 NC |
1135 | if (!SvREADONLY(TOPs) && SvIOK_notUV(TOPs) && !SvNOK(TOPs) && !SvPOK(TOPs) |
| 1136 | && SvIVX(TOPs) != IV_MIN) | |
| 55497cff | 1137 | { |
| 45977657 | 1138 | SvIV_set(TOPs, SvIVX(TOPs) - 1); |
| 55497cff | 1139 | SvFLAGS(TOPs) &= ~(SVp_NOK|SVp_POK); |
| 748a9306 LW |
1140 | } |
| 1141 | else | |
| 6f1401dc | 1142 | sv_dec_nomg(TOPs); |
| a0d0e21e LW |
1143 | SvSETMAGIC(TOPs); |
| 1144 | SETs(TARG); | |
| 1145 | return NORMAL; | |
| 1146 | } | |
| 79072805 | 1147 | |
| a0d0e21e LW |
1148 | /* Ordinary operators. */ |
| 1149 | ||
| 1150 | PP(pp_pow) | |
| 1151 | { | |
| 800401ee | 1152 | dVAR; dSP; dATARGET; SV *svl, *svr; |
| 58d76dfd | 1153 | #ifdef PERL_PRESERVE_IVUV |
| 52a96ae6 HS |
1154 | bool is_int = 0; |
| 1155 | #endif | |
| 6f1401dc DM |
1156 | tryAMAGICbin_MG(pow_amg, AMGf_assign|AMGf_numeric); |
| 1157 | svr = TOPs; | |
| 1158 | svl = TOPm1s; | |
| 52a96ae6 HS |
1159 | #ifdef PERL_PRESERVE_IVUV |
| 1160 | /* For integer to integer power, we do the calculation by hand wherever | |
| 1161 | we're sure it is safe; otherwise we call pow() and try to convert to | |
| 1162 | integer afterwards. */ | |
| 58d76dfd | 1163 | { |
| 6f1401dc | 1164 | SvIV_please_nomg(svr); |
| 800401ee | 1165 | if (SvIOK(svr)) { |
| 6f1401dc | 1166 | SvIV_please_nomg(svl); |
| 800401ee | 1167 | if (SvIOK(svl)) { |
| 900658e3 PF |
1168 | UV power; |
| 1169 | bool baseuok; | |
| 1170 | UV baseuv; | |
| 1171 | ||
| 800401ee JH |
1172 | if (SvUOK(svr)) { |
| 1173 | power = SvUVX(svr); | |
| 900658e3 | 1174 | } else { |
| 800401ee | 1175 | const IV iv = SvIVX(svr); |
| 900658e3 PF |
1176 | if (iv >= 0) { |
| 1177 | power = iv; | |
| 1178 | } else { | |
| 1179 | goto float_it; /* Can't do negative powers this way. */ | |
| 1180 | } | |
| 1181 | } | |
| 1182 | ||
| 800401ee | 1183 | baseuok = SvUOK(svl); |
| 900658e3 | 1184 | if (baseuok) { |
| 800401ee | 1185 | baseuv = SvUVX(svl); |
| 900658e3 | 1186 | } else { |
| 800401ee | 1187 | const IV iv = SvIVX(svl); |
| 900658e3 PF |
1188 | if (iv >= 0) { |
| 1189 | baseuv = iv; | |
| 1190 | baseuok = TRUE; /* effectively it's a UV now */ | |
| 1191 | } else { | |
| 1192 | baseuv = -iv; /* abs, baseuok == false records sign */ | |
| 1193 | } | |
| 1194 | } | |
| 52a96ae6 HS |
1195 | /* now we have integer ** positive integer. */ |
| 1196 | is_int = 1; | |
| 1197 | ||
| 1198 | /* foo & (foo - 1) is zero only for a power of 2. */ | |
| 58d76dfd | 1199 | if (!(baseuv & (baseuv - 1))) { |
| 52a96ae6 | 1200 | /* We are raising power-of-2 to a positive integer. |
| 58d76dfd JH |
1201 | The logic here will work for any base (even non-integer |
| 1202 | bases) but it can be less accurate than | |
| 1203 | pow (base,power) or exp (power * log (base)) when the | |
| 1204 | intermediate values start to spill out of the mantissa. | |
| 1205 | With powers of 2 we know this can't happen. | |
| 1206 | And powers of 2 are the favourite thing for perl | |
| 1207 | programmers to notice ** not doing what they mean. */ | |
| 1208 | NV result = 1.0; | |
| 1209 | NV base = baseuok ? baseuv : -(NV)baseuv; | |
| 900658e3 PF |
1210 | |
| 1211 | if (power & 1) { | |
| 1212 | result *= base; | |
| 1213 | } | |
| 1214 | while (power >>= 1) { | |
| 1215 | base *= base; | |
| 1216 | if (power & 1) { | |
| 1217 | result *= base; | |
| 1218 | } | |
| 1219 | } | |
| 58d76dfd JH |
1220 | SP--; |
| 1221 | SETn( result ); | |
| 6f1401dc | 1222 | SvIV_please_nomg(svr); |
| 58d76dfd | 1223 | RETURN; |
| 52a96ae6 HS |
1224 | } else { |
| 1225 | register unsigned int highbit = 8 * sizeof(UV); | |
| 900658e3 PF |
1226 | register unsigned int diff = 8 * sizeof(UV); |
| 1227 | while (diff >>= 1) { | |
| 1228 | highbit -= diff; | |
| 1229 | if (baseuv >> highbit) { | |
| 1230 | highbit += diff; | |
| 1231 | } | |
| 52a96ae6 HS |
1232 | } |
| 1233 | /* we now have baseuv < 2 ** highbit */ | |
| 1234 | if (power * highbit <= 8 * sizeof(UV)) { | |
| 1235 | /* result will definitely fit in UV, so use UV math | |
| 1236 | on same algorithm as above */ | |
| 1237 | register UV result = 1; | |
| 1238 | register UV base = baseuv; | |
| f2338a2e | 1239 | const bool odd_power = cBOOL(power & 1); |
| 900658e3 PF |
1240 | if (odd_power) { |
| 1241 | result *= base; | |
| 1242 | } | |
| 1243 | while (power >>= 1) { | |
| 1244 | base *= base; | |
| 1245 | if (power & 1) { | |
| 52a96ae6 | 1246 | result *= base; |
| 52a96ae6 HS |
1247 | } |
| 1248 | } | |
| 1249 | SP--; | |
| 0615a994 | 1250 | if (baseuok || !odd_power) |
| 52a96ae6 HS |
1251 | /* answer is positive */ |
| 1252 | SETu( result ); | |
| 1253 | else if (result <= (UV)IV_MAX) | |
| 1254 | /* answer negative, fits in IV */ | |
| 1255 | SETi( -(IV)result ); | |
| 1256 | else if (result == (UV)IV_MIN) | |
| 1257 | /* 2's complement assumption: special case IV_MIN */ | |
| 1258 | SETi( IV_MIN ); | |
| 1259 | else | |
| 1260 | /* answer negative, doesn't fit */ | |
| 1261 | SETn( -(NV)result ); | |
| 1262 | RETURN; | |
| 1263 | } | |
| 1264 | } | |
| 1265 | } | |
| 1266 | } | |
| 58d76dfd | 1267 | } |
| 52a96ae6 | 1268 | float_it: |
| 58d76dfd | 1269 | #endif |
| a0d0e21e | 1270 | { |
| 6f1401dc DM |
1271 | NV right = SvNV_nomg(svr); |
| 1272 | NV left = SvNV_nomg(svl); | |
| 4efa5a16 | 1273 | (void)POPs; |
| 3aaeb624 JA |
1274 | |
| 1275 | #if defined(USE_LONG_DOUBLE) && defined(HAS_AIX_POWL_NEG_BASE_BUG) | |
| 1276 | /* | |
| 1277 | We are building perl with long double support and are on an AIX OS | |
| 1278 | afflicted with a powl() function that wrongly returns NaNQ for any | |
| 1279 | negative base. This was reported to IBM as PMR #23047-379 on | |
| 1280 | 03/06/2006. The problem exists in at least the following versions | |
| 1281 | of AIX and the libm fileset, and no doubt others as well: | |
| 1282 | ||
| 1283 | AIX 4.3.3-ML10 bos.adt.libm 4.3.3.50 | |
| 1284 | AIX 5.1.0-ML04 bos.adt.libm 5.1.0.29 | |
| 1285 | AIX 5.2.0 bos.adt.libm 5.2.0.85 | |
| 1286 | ||
| 1287 | So, until IBM fixes powl(), we provide the following workaround to | |
| 1288 | handle the problem ourselves. Our logic is as follows: for | |
| 1289 | negative bases (left), we use fmod(right, 2) to check if the | |
| 1290 | exponent is an odd or even integer: | |
| 1291 | ||
| 1292 | - if odd, powl(left, right) == -powl(-left, right) | |
| 1293 | - if even, powl(left, right) == powl(-left, right) | |
| 1294 | ||
| 1295 | If the exponent is not an integer, the result is rightly NaNQ, so | |
| 1296 | we just return that (as NV_NAN). | |
| 1297 | */ | |
| 1298 | ||
| 1299 | if (left < 0.0) { | |
| 1300 | NV mod2 = Perl_fmod( right, 2.0 ); | |
| 1301 | if (mod2 == 1.0 || mod2 == -1.0) { /* odd integer */ | |
| 1302 | SETn( -Perl_pow( -left, right) ); | |
| 1303 | } else if (mod2 == 0.0) { /* even integer */ | |
| 1304 | SETn( Perl_pow( -left, right) ); | |
| 1305 | } else { /* fractional power */ | |
| 1306 | SETn( NV_NAN ); | |
| 1307 | } | |
| 1308 | } else { | |
| 1309 | SETn( Perl_pow( left, right) ); | |
| 1310 | } | |
| 1311 | #else | |
| 52a96ae6 | 1312 | SETn( Perl_pow( left, right) ); |
| 3aaeb624 JA |
1313 | #endif /* HAS_AIX_POWL_NEG_BASE_BUG */ |
| 1314 | ||
| 52a96ae6 HS |
1315 | #ifdef PERL_PRESERVE_IVUV |
| 1316 | if (is_int) | |
| 6f1401dc | 1317 | SvIV_please_nomg(svr); |
| 52a96ae6 HS |
1318 | #endif |
| 1319 | RETURN; | |
| 93a17b20 | 1320 | } |
| a0d0e21e LW |
1321 | } |
| 1322 | ||
| 1323 | PP(pp_multiply) | |
| 1324 | { | |
| 800401ee | 1325 | dVAR; dSP; dATARGET; SV *svl, *svr; |
| 6f1401dc DM |
1326 | tryAMAGICbin_MG(mult_amg, AMGf_assign|AMGf_numeric); |
| 1327 | svr = TOPs; | |
| 1328 | svl = TOPm1s; | |
| 28e5dec8 | 1329 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 1330 | SvIV_please_nomg(svr); |
| 800401ee | 1331 | if (SvIOK(svr)) { |
| 28e5dec8 JH |
1332 | /* Unless the left argument is integer in range we are going to have to |
| 1333 | use NV maths. Hence only attempt to coerce the right argument if | |
| 1334 | we know the left is integer. */ | |
| 1335 | /* Left operand is defined, so is it IV? */ | |
| 6f1401dc | 1336 | SvIV_please_nomg(svl); |
| 800401ee JH |
1337 | if (SvIOK(svl)) { |
| 1338 | bool auvok = SvUOK(svl); | |
| 1339 | bool buvok = SvUOK(svr); | |
| 28e5dec8 JH |
1340 | const UV topmask = (~ (UV)0) << (4 * sizeof (UV)); |
| 1341 | const UV botmask = ~((~ (UV)0) << (4 * sizeof (UV))); | |
| 1342 | UV alow; | |
| 1343 | UV ahigh; | |
| 1344 | UV blow; | |
| 1345 | UV bhigh; | |
| 1346 | ||
| 1347 | if (auvok) { | |
| 800401ee | 1348 | alow = SvUVX(svl); |
| 28e5dec8 | 1349 | } else { |
| 800401ee | 1350 | const IV aiv = SvIVX(svl); |
| 28e5dec8 JH |
1351 | if (aiv >= 0) { |
| 1352 | alow = aiv; | |
| 1353 | auvok = TRUE; /* effectively it's a UV now */ | |
| 1354 | } else { | |
| 1355 | alow = -aiv; /* abs, auvok == false records sign */ | |
| 1356 | } | |
| 1357 | } | |
| 1358 | if (buvok) { | |
| 800401ee | 1359 | blow = SvUVX(svr); |
| 28e5dec8 | 1360 | } else { |
| 800401ee | 1361 | const IV biv = SvIVX(svr); |
| 28e5dec8 JH |
1362 | if (biv >= 0) { |
| 1363 | blow = biv; | |
| 1364 | buvok = TRUE; /* effectively it's a UV now */ | |
| 1365 | } else { | |
| 1366 | blow = -biv; /* abs, buvok == false records sign */ | |
| 1367 | } | |
| 1368 | } | |
| 1369 | ||
| 1370 | /* If this does sign extension on unsigned it's time for plan B */ | |
| 1371 | ahigh = alow >> (4 * sizeof (UV)); | |
| 1372 | alow &= botmask; | |
| 1373 | bhigh = blow >> (4 * sizeof (UV)); | |
| 1374 | blow &= botmask; | |
| 1375 | if (ahigh && bhigh) { | |
| 6f207bd3 | 1376 | NOOP; |
| 28e5dec8 JH |
1377 | /* eg 32 bit is at least 0x10000 * 0x10000 == 0x100000000 |
| 1378 | which is overflow. Drop to NVs below. */ | |
| 1379 | } else if (!ahigh && !bhigh) { | |
| 1380 | /* eg 32 bit is at most 0xFFFF * 0xFFFF == 0xFFFE0001 | |
| 1381 | so the unsigned multiply cannot overflow. */ | |
| c445ea15 | 1382 | const UV product = alow * blow; |
| 28e5dec8 JH |
1383 | if (auvok == buvok) { |
| 1384 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
| 1385 | SP--; | |
| 1386 | SETu( product ); | |
| 1387 | RETURN; | |
| 1388 | } else if (product <= (UV)IV_MIN) { | |
| 1389 | /* 2s complement assumption that (UV)-IV_MIN is correct. */ | |
| 1390 | /* -ve result, which could overflow an IV */ | |
| 1391 | SP--; | |
| 25716404 | 1392 | SETi( -(IV)product ); |
| 28e5dec8 JH |
1393 | RETURN; |
| 1394 | } /* else drop to NVs below. */ | |
| 1395 | } else { | |
| 1396 | /* One operand is large, 1 small */ | |
| 1397 | UV product_middle; | |
| 1398 | if (bhigh) { | |
| 1399 | /* swap the operands */ | |
| 1400 | ahigh = bhigh; | |
| 1401 | bhigh = blow; /* bhigh now the temp var for the swap */ | |
| 1402 | blow = alow; | |
| 1403 | alow = bhigh; | |
| 1404 | } | |
| 1405 | /* now, ((ahigh * blow) << half_UV_len) + (alow * blow) | |
| 1406 | multiplies can't overflow. shift can, add can, -ve can. */ | |
| 1407 | product_middle = ahigh * blow; | |
| 1408 | if (!(product_middle & topmask)) { | |
| 1409 | /* OK, (ahigh * blow) won't lose bits when we shift it. */ | |
| 1410 | UV product_low; | |
| 1411 | product_middle <<= (4 * sizeof (UV)); | |
| 1412 | product_low = alow * blow; | |
| 1413 | ||
| 1414 | /* as for pp_add, UV + something mustn't get smaller. | |
| 1415 | IIRC ANSI mandates this wrapping *behaviour* for | |
| 1416 | unsigned whatever the actual representation*/ | |
| 1417 | product_low += product_middle; | |
| 1418 | if (product_low >= product_middle) { | |
| 1419 | /* didn't overflow */ | |
| 1420 | if (auvok == buvok) { | |
| 1421 | /* -ve * -ve or +ve * +ve gives a +ve result. */ | |
| 1422 | SP--; | |
| 1423 | SETu( product_low ); | |
| 1424 | RETURN; | |
| 1425 | } else if (product_low <= (UV)IV_MIN) { | |
| 1426 | /* 2s complement assumption again */ | |
| 1427 | /* -ve result, which could overflow an IV */ | |
| 1428 | SP--; | |
| 25716404 | 1429 | SETi( -(IV)product_low ); |
| 28e5dec8 JH |
1430 | RETURN; |
| 1431 | } /* else drop to NVs below. */ | |
| 1432 | } | |
| 1433 | } /* product_middle too large */ | |
| 1434 | } /* ahigh && bhigh */ | |
| 800401ee JH |
1435 | } /* SvIOK(svl) */ |
| 1436 | } /* SvIOK(svr) */ | |
| 28e5dec8 | 1437 | #endif |
| a0d0e21e | 1438 | { |
| 6f1401dc DM |
1439 | NV right = SvNV_nomg(svr); |
| 1440 | NV left = SvNV_nomg(svl); | |
| 4efa5a16 | 1441 | (void)POPs; |
| a0d0e21e LW |
1442 | SETn( left * right ); |
| 1443 | RETURN; | |
| 79072805 | 1444 | } |
| a0d0e21e LW |
1445 | } |
| 1446 | ||
| 1447 | PP(pp_divide) | |
| 1448 | { | |
| 800401ee | 1449 | dVAR; dSP; dATARGET; SV *svl, *svr; |
| 6f1401dc DM |
1450 | tryAMAGICbin_MG(div_amg, AMGf_assign|AMGf_numeric); |
| 1451 | svr = TOPs; | |
| 1452 | svl = TOPm1s; | |
| 5479d192 | 1453 | /* Only try to do UV divide first |
| 68795e93 | 1454 | if ((SLOPPYDIVIDE is true) or |
| 5479d192 NC |
1455 | (PERL_PRESERVE_IVUV is true and one or both SV is a UV too large |
| 1456 | to preserve)) | |
| 1457 | The assumption is that it is better to use floating point divide | |
| 1458 | whenever possible, only doing integer divide first if we can't be sure. | |
| 1459 | If NV_PRESERVES_UV is true then we know at compile time that no UV | |
| 1460 | can be too large to preserve, so don't need to compile the code to | |
| 1461 | test the size of UVs. */ | |
| 1462 | ||
| a0d0e21e | 1463 | #ifdef SLOPPYDIVIDE |
| 5479d192 NC |
1464 | # define PERL_TRY_UV_DIVIDE |
| 1465 | /* ensure that 20./5. == 4. */ | |
| a0d0e21e | 1466 | #else |
| 5479d192 NC |
1467 | # ifdef PERL_PRESERVE_IVUV |
| 1468 | # ifndef NV_PRESERVES_UV | |
| 1469 | # define PERL_TRY_UV_DIVIDE | |
| 1470 | # endif | |
| 1471 | # endif | |
| a0d0e21e | 1472 | #endif |
| 5479d192 NC |
1473 | |
| 1474 | #ifdef PERL_TRY_UV_DIVIDE | |
| 6f1401dc | 1475 | SvIV_please_nomg(svr); |
| 800401ee | 1476 | if (SvIOK(svr)) { |
| 6f1401dc | 1477 | SvIV_please_nomg(svl); |
| 800401ee JH |
1478 | if (SvIOK(svl)) { |
| 1479 | bool left_non_neg = SvUOK(svl); | |
| 1480 | bool right_non_neg = SvUOK(svr); | |
| 5479d192 NC |
1481 | UV left; |
| 1482 | UV right; | |
| 1483 | ||
| 1484 | if (right_non_neg) { | |
| 800401ee | 1485 | right = SvUVX(svr); |
| 5479d192 NC |
1486 | } |
| 1487 | else { | |
| 800401ee | 1488 | const IV biv = SvIVX(svr); |
| 5479d192 NC |
1489 | if (biv >= 0) { |
| 1490 | right = biv; | |
| 1491 | right_non_neg = TRUE; /* effectively it's a UV now */ | |
| 1492 | } | |
| 1493 | else { | |
| 1494 | right = -biv; | |
| 1495 | } | |
| 1496 | } | |
| 1497 | /* historically undef()/0 gives a "Use of uninitialized value" | |
| 1498 | warning before dieing, hence this test goes here. | |
| 1499 | If it were immediately before the second SvIV_please, then | |
| 1500 | DIE() would be invoked before left was even inspected, so | |
| 486ec47a | 1501 | no inspection would give no warning. */ |
| 5479d192 NC |
1502 | if (right == 0) |
| 1503 | DIE(aTHX_ "Illegal division by zero"); | |
| 1504 | ||
| 1505 | if (left_non_neg) { | |
| 800401ee | 1506 | left = SvUVX(svl); |
| 5479d192 NC |
1507 | } |
| 1508 | else { | |
| 800401ee | 1509 | const IV aiv = SvIVX(svl); |
| 5479d192 NC |
1510 | if (aiv >= 0) { |
| 1511 | left = aiv; | |
| 1512 | left_non_neg = TRUE; /* effectively it's a UV now */ | |
| 1513 | } | |
| 1514 | else { | |
| 1515 | left = -aiv; | |
| 1516 | } | |
| 1517 | } | |
| 1518 | ||
| 1519 | if (left >= right | |
| 1520 | #ifdef SLOPPYDIVIDE | |
| 1521 | /* For sloppy divide we always attempt integer division. */ | |
| 1522 | #else | |
| 1523 | /* Otherwise we only attempt it if either or both operands | |
| 1524 | would not be preserved by an NV. If both fit in NVs | |
| 0c2ee62a NC |
1525 | we fall through to the NV divide code below. However, |
| 1526 | as left >= right to ensure integer result here, we know that | |
| 1527 | we can skip the test on the right operand - right big | |
| 1528 | enough not to be preserved can't get here unless left is | |
| 1529 | also too big. */ | |
| 1530 | ||
| 1531 | && (left > ((UV)1 << NV_PRESERVES_UV_BITS)) | |
| 5479d192 NC |
1532 | #endif |
| 1533 | ) { | |
| 1534 | /* Integer division can't overflow, but it can be imprecise. */ | |
| 1b6737cc | 1535 | const UV result = left / right; |
| 5479d192 NC |
1536 | if (result * right == left) { |
| 1537 | SP--; /* result is valid */ | |
| 1538 | if (left_non_neg == right_non_neg) { | |
| 1539 | /* signs identical, result is positive. */ | |
| 1540 | SETu( result ); | |
| 1541 | RETURN; | |
| 1542 | } | |
| 1543 | /* 2s complement assumption */ | |
| 1544 | if (result <= (UV)IV_MIN) | |
| 91f3b821 | 1545 | SETi( -(IV)result ); |
| 5479d192 NC |
1546 | else { |
| 1547 | /* It's exact but too negative for IV. */ | |
| 1548 | SETn( -(NV)result ); | |
| 1549 | } | |
| 1550 | RETURN; | |
| 1551 | } /* tried integer divide but it was not an integer result */ | |
| 32fdb065 | 1552 | } /* else (PERL_ABS(result) < 1.0) or (both UVs in range for NV) */ |
| 5479d192 NC |
1553 | } /* left wasn't SvIOK */ |
| 1554 | } /* right wasn't SvIOK */ | |
| 1555 | #endif /* PERL_TRY_UV_DIVIDE */ | |
| 1556 | { | |
| 6f1401dc DM |
1557 | NV right = SvNV_nomg(svr); |
| 1558 | NV left = SvNV_nomg(svl); | |
| 4efa5a16 | 1559 | (void)POPs;(void)POPs; |
| ebc6a117 PD |
1560 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 1561 | if (! Perl_isnan(right) && right == 0.0) | |
| 1562 | #else | |
| 5479d192 | 1563 | if (right == 0.0) |
| ebc6a117 | 1564 | #endif |
| 5479d192 NC |
1565 | DIE(aTHX_ "Illegal division by zero"); |
| 1566 | PUSHn( left / right ); | |
| 1567 | RETURN; | |
| 79072805 | 1568 | } |
| a0d0e21e LW |
1569 | } |
| 1570 | ||
| 1571 | PP(pp_modulo) | |
| 1572 | { | |
| 6f1401dc DM |
1573 | dVAR; dSP; dATARGET; |
| 1574 | tryAMAGICbin_MG(modulo_amg, AMGf_assign|AMGf_numeric); | |
| a0d0e21e | 1575 | { |
| 9c5ffd7c JH |
1576 | UV left = 0; |
| 1577 | UV right = 0; | |
| dc656993 JH |
1578 | bool left_neg = FALSE; |
| 1579 | bool right_neg = FALSE; | |
| e2c88acc NC |
1580 | bool use_double = FALSE; |
| 1581 | bool dright_valid = FALSE; | |
| 9c5ffd7c JH |
1582 | NV dright = 0.0; |
| 1583 | NV dleft = 0.0; | |
| 6f1401dc DM |
1584 | SV * const svr = TOPs; |
| 1585 | SV * const svl = TOPm1s; | |
| 1586 | SvIV_please_nomg(svr); | |
| 800401ee JH |
1587 | if (SvIOK(svr)) { |
| 1588 | right_neg = !SvUOK(svr); | |
| e2c88acc | 1589 | if (!right_neg) { |
| 800401ee | 1590 | right = SvUVX(svr); |
| e2c88acc | 1591 | } else { |
| 800401ee | 1592 | const IV biv = SvIVX(svr); |
| e2c88acc NC |
1593 | if (biv >= 0) { |
| 1594 | right = biv; | |
| 1595 | right_neg = FALSE; /* effectively it's a UV now */ | |
| 1596 | } else { | |
| 1597 | right = -biv; | |
| 1598 | } | |
| 1599 | } | |
| 1600 | } | |
| 1601 | else { | |
| 6f1401dc | 1602 | dright = SvNV_nomg(svr); |
| 787eafbd IZ |
1603 | right_neg = dright < 0; |
| 1604 | if (right_neg) | |
| 1605 | dright = -dright; | |
| e2c88acc NC |
1606 | if (dright < UV_MAX_P1) { |
| 1607 | right = U_V(dright); | |
| 1608 | dright_valid = TRUE; /* In case we need to use double below. */ | |
| 1609 | } else { | |
| 1610 | use_double = TRUE; | |
| 1611 | } | |
| 787eafbd | 1612 | } |
| a0d0e21e | 1613 | |
| e2c88acc NC |
1614 | /* At this point use_double is only true if right is out of range for |
| 1615 | a UV. In range NV has been rounded down to nearest UV and | |
| 1616 | use_double false. */ | |
| 6f1401dc | 1617 | SvIV_please_nomg(svl); |
| 800401ee JH |
1618 | if (!use_double && SvIOK(svl)) { |
| 1619 | if (SvIOK(svl)) { | |
| 1620 | left_neg = !SvUOK(svl); | |
| e2c88acc | 1621 | if (!left_neg) { |
| 800401ee | 1622 | left = SvUVX(svl); |
| e2c88acc | 1623 | } else { |
| 800401ee | 1624 | const IV aiv = SvIVX(svl); |
| e2c88acc NC |
1625 | if (aiv >= 0) { |
| 1626 | left = aiv; | |
| 1627 | left_neg = FALSE; /* effectively it's a UV now */ | |
| 1628 | } else { | |
| 1629 | left = -aiv; | |
| 1630 | } | |
| 1631 | } | |
| 1632 | } | |
| 1633 | } | |
| 787eafbd | 1634 | else { |
| 6f1401dc | 1635 | dleft = SvNV_nomg(svl); |
| 787eafbd IZ |
1636 | left_neg = dleft < 0; |
| 1637 | if (left_neg) | |
| 1638 | dleft = -dleft; | |
| 68dc0745 | 1639 | |
| e2c88acc NC |
1640 | /* This should be exactly the 5.6 behaviour - if left and right are |
| 1641 | both in range for UV then use U_V() rather than floor. */ | |
| 1642 | if (!use_double) { | |
| 1643 | if (dleft < UV_MAX_P1) { | |
| 1644 | /* right was in range, so is dleft, so use UVs not double. | |
| 1645 | */ | |
| 1646 | left = U_V(dleft); | |
| 1647 | } | |
| 1648 | /* left is out of range for UV, right was in range, so promote | |
| 1649 | right (back) to double. */ | |
| 1650 | else { | |
| 1651 | /* The +0.5 is used in 5.6 even though it is not strictly | |
| 1652 | consistent with the implicit +0 floor in the U_V() | |
| 1653 | inside the #if 1. */ | |
| 1654 | dleft = Perl_floor(dleft + 0.5); | |
| 1655 | use_double = TRUE; | |
| 1656 | if (dright_valid) | |
| 1657 | dright = Perl_floor(dright + 0.5); | |
| 1658 | else | |
| 1659 | dright = right; | |
| 1660 | } | |
| 1661 | } | |
| 1662 | } | |
| 6f1401dc | 1663 | sp -= 2; |
| 787eafbd | 1664 | if (use_double) { |
| 65202027 | 1665 | NV dans; |
| 787eafbd | 1666 | |
| 787eafbd | 1667 | if (!dright) |
| cea2e8a9 | 1668 | DIE(aTHX_ "Illegal modulus zero"); |
| 787eafbd | 1669 | |
| 65202027 | 1670 | dans = Perl_fmod(dleft, dright); |
| 787eafbd IZ |
1671 | if ((left_neg != right_neg) && dans) |
| 1672 | dans = dright - dans; | |
| 1673 | if (right_neg) | |
| 1674 | dans = -dans; | |
| 1675 | sv_setnv(TARG, dans); | |
| 1676 | } | |
| 1677 | else { | |
| 1678 | UV ans; | |
| 1679 | ||
| 787eafbd | 1680 | if (!right) |
| cea2e8a9 | 1681 | DIE(aTHX_ "Illegal modulus zero"); |
| 787eafbd IZ |
1682 | |
| 1683 | ans = left % right; | |
| 1684 | if ((left_neg != right_neg) && ans) | |
| 1685 | ans = right - ans; | |
| 1686 | if (right_neg) { | |
| 1687 | /* XXX may warn: unary minus operator applied to unsigned type */ | |
| 1688 | /* could change -foo to be (~foo)+1 instead */ | |
| 1689 | if (ans <= ~((UV)IV_MAX)+1) | |
| 1690 | sv_setiv(TARG, ~ans+1); | |
| 1691 | else | |
| 65202027 | 1692 | sv_setnv(TARG, -(NV)ans); |
| 787eafbd IZ |
1693 | } |
| 1694 | else | |
| 1695 | sv_setuv(TARG, ans); | |
| 1696 | } | |
| 1697 | PUSHTARG; | |
| 1698 | RETURN; | |
| 79072805 | 1699 | } |
| a0d0e21e | 1700 | } |
| 79072805 | 1701 | |
| a0d0e21e LW |
1702 | PP(pp_repeat) |
| 1703 | { | |
| 6f1401dc | 1704 | dVAR; dSP; dATARGET; |
| 2b573ace | 1705 | register IV count; |
| 6f1401dc DM |
1706 | SV *sv; |
| 1707 | ||
| 1708 | if (GIMME == G_ARRAY && PL_op->op_private & OPpREPEAT_DOLIST) { | |
| 1709 | /* TODO: think of some way of doing list-repeat overloading ??? */ | |
| 1710 | sv = POPs; | |
| 1711 | SvGETMAGIC(sv); | |
| 1712 | } | |
| 1713 | else { | |
| 1714 | tryAMAGICbin_MG(repeat_amg, AMGf_assign); | |
| 1715 | sv = POPs; | |
| 1716 | } | |
| 1717 | ||
| 2b573ace JH |
1718 | if (SvIOKp(sv)) { |
| 1719 | if (SvUOK(sv)) { | |
| 6f1401dc | 1720 | const UV uv = SvUV_nomg(sv); |
| 2b573ace JH |
1721 | if (uv > IV_MAX) |
| 1722 | count = IV_MAX; /* The best we can do? */ | |
| 1723 | else | |
| 1724 | count = uv; | |
| 1725 | } else { | |
| 6f1401dc | 1726 | const IV iv = SvIV_nomg(sv); |
| 2b573ace JH |
1727 | if (iv < 0) |
| 1728 | count = 0; | |
| 1729 | else | |
| 1730 | count = iv; | |
| 1731 | } | |
| 1732 | } | |
| 1733 | else if (SvNOKp(sv)) { | |
| 6f1401dc | 1734 | const NV nv = SvNV_nomg(sv); |
| 2b573ace JH |
1735 | if (nv < 0.0) |
| 1736 | count = 0; | |
| 1737 | else | |
| 1738 | count = (IV)nv; | |
| 1739 | } | |
| 1740 | else | |
| 6f1401dc DM |
1741 | count = SvIV_nomg(sv); |
| 1742 | ||
| 533c011a | 1743 | if (GIMME == G_ARRAY && PL_op->op_private & OPpREPEAT_DOLIST) { |
| a0d0e21e | 1744 | dMARK; |
| 0bd48802 AL |
1745 | static const char oom_list_extend[] = "Out of memory during list extend"; |
| 1746 | const I32 items = SP - MARK; | |
| 1747 | const I32 max = items * count; | |
| 79072805 | 1748 | |
| 2b573ace JH |
1749 | MEM_WRAP_CHECK_1(max, SV*, oom_list_extend); |
| 1750 | /* Did the max computation overflow? */ | |
| 27d5b266 | 1751 | if (items > 0 && max > 0 && (max < items || max < count)) |
| 2b573ace | 1752 | Perl_croak(aTHX_ oom_list_extend); |
| a0d0e21e LW |
1753 | MEXTEND(MARK, max); |
| 1754 | if (count > 1) { | |
| 1755 | while (SP > MARK) { | |
| 976c8a39 JH |
1756 | #if 0 |
| 1757 | /* This code was intended to fix 20010809.028: | |
| 1758 | ||
| 1759 | $x = 'abcd'; | |
| 1760 | for (($x =~ /./g) x 2) { | |
| 1761 | print chop; # "abcdabcd" expected as output. | |
| 1762 | } | |
| 1763 | ||
| 1764 | * but that change (#11635) broke this code: | |
| 1765 | ||
| 1766 | $x = [("foo")x2]; # only one "foo" ended up in the anonlist. | |
| 1767 | ||
| 1768 | * I can't think of a better fix that doesn't introduce | |
| 1769 | * an efficiency hit by copying the SVs. The stack isn't | |
| 1770 | * refcounted, and mortalisation obviously doesn't | |
| 1771 | * Do The Right Thing when the stack has more than | |
| 1772 | * one pointer to the same mortal value. | |
| 1773 | * .robin. | |
| 1774 | */ | |
| e30acc16 RH |
1775 | if (*SP) { |
| 1776 | *SP = sv_2mortal(newSVsv(*SP)); | |
| 1777 | SvREADONLY_on(*SP); | |
| 1778 | } | |
| 976c8a39 JH |
1779 | #else |
| 1780 | if (*SP) | |
| 1781 | SvTEMP_off((*SP)); | |
| 1782 | #endif | |
| a0d0e21e | 1783 | SP--; |
| 79072805 | 1784 | } |
| a0d0e21e LW |
1785 | MARK++; |
| 1786 | repeatcpy((char*)(MARK + items), (char*)MARK, | |
| ad64d0ec | 1787 | items * sizeof(const SV *), count - 1); |
| a0d0e21e | 1788 | SP += max; |
| 79072805 | 1789 | } |
| a0d0e21e LW |
1790 | else if (count <= 0) |
| 1791 | SP -= items; | |
| 79072805 | 1792 | } |
| a0d0e21e | 1793 | else { /* Note: mark already snarfed by pp_list */ |
| 0bd48802 | 1794 | SV * const tmpstr = POPs; |
| a0d0e21e | 1795 | STRLEN len; |
| 9b877dbb | 1796 | bool isutf; |
| 2b573ace JH |
1797 | static const char oom_string_extend[] = |
| 1798 | "Out of memory during string extend"; | |
| a0d0e21e | 1799 | |
| 6f1401dc DM |
1800 | if (TARG != tmpstr) |
| 1801 | sv_setsv_nomg(TARG, tmpstr); | |
| 1802 | SvPV_force_nomg(TARG, len); | |
| 9b877dbb | 1803 | isutf = DO_UTF8(TARG); |
| 8ebc5c01 | 1804 | if (count != 1) { |
| 1805 | if (count < 1) | |
| 1806 | SvCUR_set(TARG, 0); | |
| 1807 | else { | |
| c445ea15 | 1808 | const STRLEN max = (UV)count * len; |
| 19a94d75 | 1809 | if (len > MEM_SIZE_MAX / count) |
| 2b573ace JH |
1810 | Perl_croak(aTHX_ oom_string_extend); |
| 1811 | MEM_WRAP_CHECK_1(max, char, oom_string_extend); | |
| 8569b9dc | 1812 | SvGROW(TARG, max + 1); |
| a0d0e21e | 1813 | repeatcpy(SvPVX(TARG) + len, SvPVX(TARG), len, count - 1); |
| b162af07 | 1814 | SvCUR_set(TARG, SvCUR(TARG) * count); |
| 7a4c00b4 | 1815 | } |
| a0d0e21e | 1816 | *SvEND(TARG) = '\0'; |
| a0d0e21e | 1817 | } |
| dfcb284a GS |
1818 | if (isutf) |
| 1819 | (void)SvPOK_only_UTF8(TARG); | |
| 1820 | else | |
| 1821 | (void)SvPOK_only(TARG); | |
| b80b6069 RH |
1822 | |
| 1823 | if (PL_op->op_private & OPpREPEAT_DOLIST) { | |
| 1824 | /* The parser saw this as a list repeat, and there | |
| 1825 | are probably several items on the stack. But we're | |
| 1826 | in scalar context, and there's no pp_list to save us | |
| 1827 | now. So drop the rest of the items -- robin@kitsite.com | |
| 1828 | */ | |
| 1829 | dMARK; | |
| 1830 | SP = MARK; | |
| 1831 | } | |
| a0d0e21e | 1832 | PUSHTARG; |
| 79072805 | 1833 | } |
| a0d0e21e LW |
1834 | RETURN; |
| 1835 | } | |
| 79072805 | 1836 | |
| a0d0e21e LW |
1837 | PP(pp_subtract) |
| 1838 | { | |
| 800401ee | 1839 | dVAR; dSP; dATARGET; bool useleft; SV *svl, *svr; |
| 6f1401dc DM |
1840 | tryAMAGICbin_MG(subtr_amg, AMGf_assign|AMGf_numeric); |
| 1841 | svr = TOPs; | |
| 1842 | svl = TOPm1s; | |
| 800401ee | 1843 | useleft = USE_LEFT(svl); |
| 28e5dec8 | 1844 | #ifdef PERL_PRESERVE_IVUV |
| 7dca457a NC |
1845 | /* See comments in pp_add (in pp_hot.c) about Overflow, and how |
| 1846 | "bad things" happen if you rely on signed integers wrapping. */ | |
| 6f1401dc | 1847 | SvIV_please_nomg(svr); |
| 800401ee | 1848 | if (SvIOK(svr)) { |
| 28e5dec8 JH |
1849 | /* Unless the left argument is integer in range we are going to have to |
| 1850 | use NV maths. Hence only attempt to coerce the right argument if | |
| 1851 | we know the left is integer. */ | |
| 9c5ffd7c JH |
1852 | register UV auv = 0; |
| 1853 | bool auvok = FALSE; | |
| 7dca457a NC |
1854 | bool a_valid = 0; |
| 1855 | ||
| 28e5dec8 | 1856 | if (!useleft) { |
| 7dca457a NC |
1857 | auv = 0; |
| 1858 | a_valid = auvok = 1; | |
| 1859 | /* left operand is undef, treat as zero. */ | |
| 28e5dec8 JH |
1860 | } else { |
| 1861 | /* Left operand is defined, so is it IV? */ | |
| 6f1401dc | 1862 | SvIV_please_nomg(svl); |
| 800401ee JH |
1863 | if (SvIOK(svl)) { |
| 1864 | if ((auvok = SvUOK(svl))) | |
| 1865 | auv = SvUVX(svl); | |
| 7dca457a | 1866 | else { |
| 800401ee | 1867 | register const IV aiv = SvIVX(svl); |
| 7dca457a NC |
1868 | if (aiv >= 0) { |
| 1869 | auv = aiv; | |
| 1870 | auvok = 1; /* Now acting as a sign flag. */ | |
| 1871 | } else { /* 2s complement assumption for IV_MIN */ | |
| 1872 | auv = (UV)-aiv; | |
| 28e5dec8 | 1873 | } |
| 7dca457a NC |
1874 | } |
| 1875 | a_valid = 1; | |
| 1876 | } | |
| 1877 | } | |
| 1878 | if (a_valid) { | |
| 1879 | bool result_good = 0; | |
| 1880 | UV result; | |
| 1881 | register UV buv; | |
| 800401ee | 1882 | bool buvok = SvUOK(svr); |
| 9041c2e3 | 1883 | |
| 7dca457a | 1884 | if (buvok) |
| 800401ee | 1885 | buv = SvUVX(svr); |
| 7dca457a | 1886 | else { |
| 800401ee | 1887 | register const IV biv = SvIVX(svr); |
| 7dca457a NC |
1888 | if (biv >= 0) { |
| 1889 | buv = biv; | |
| 1890 | buvok = 1; | |
| 1891 | } else | |
| 1892 | buv = (UV)-biv; | |
| 1893 | } | |
| 1894 | /* ?uvok if value is >= 0. basically, flagged as UV if it's +ve, | |
| 602f51c4 | 1895 | else "IV" now, independent of how it came in. |
| 7dca457a NC |
1896 | if a, b represents positive, A, B negative, a maps to -A etc |
| 1897 | a - b => (a - b) | |
| 1898 | A - b => -(a + b) | |
| 1899 | a - B => (a + b) | |
| 1900 | A - B => -(a - b) | |
| 1901 | all UV maths. negate result if A negative. | |
| 1902 | subtract if signs same, add if signs differ. */ | |
| 1903 | ||
| 1904 | if (auvok ^ buvok) { | |
| 1905 | /* Signs differ. */ | |
| 1906 | result = auv + buv; | |
| 1907 | if (result >= auv) | |
| 1908 | result_good = 1; | |
| 1909 | } else { | |
| 1910 | /* Signs same */ | |
| 1911 | if (auv >= buv) { | |
| 1912 | result = auv - buv; | |
| 1913 | /* Must get smaller */ | |
| 1914 | if (result <= auv) | |
| 1915 | result_good = 1; | |
| 1916 | } else { | |
| 1917 | result = buv - auv; | |
| 1918 | if (result <= buv) { | |
| 1919 | /* result really should be -(auv-buv). as its negation | |
| 1920 | of true value, need to swap our result flag */ | |
| 1921 | auvok = !auvok; | |
| 1922 | result_good = 1; | |
| 28e5dec8 | 1923 | } |
| 28e5dec8 JH |
1924 | } |
| 1925 | } | |
| 7dca457a NC |
1926 | if (result_good) { |
| 1927 | SP--; | |
| 1928 | if (auvok) | |
| 1929 | SETu( result ); | |
| 1930 | else { | |
| 1931 | /* Negate result */ | |
| 1932 | if (result <= (UV)IV_MIN) | |
| 1933 | SETi( -(IV)result ); | |
| 1934 | else { | |
| 1935 | /* result valid, but out of range for IV. */ | |
| 1936 | SETn( -(NV)result ); | |
| 1937 | } | |
| 1938 | } | |
| 1939 | RETURN; | |
| 1940 | } /* Overflow, drop through to NVs. */ | |
| 28e5dec8 JH |
1941 | } |
| 1942 | } | |
| 1943 | #endif | |
| a0d0e21e | 1944 | { |
| 6f1401dc | 1945 | NV value = SvNV_nomg(svr); |
| 4efa5a16 RD |
1946 | (void)POPs; |
| 1947 | ||
| 28e5dec8 JH |
1948 | if (!useleft) { |
| 1949 | /* left operand is undef, treat as zero - value */ | |
| 1950 | SETn(-value); | |
| 1951 | RETURN; | |
| 1952 | } | |
| 6f1401dc | 1953 | SETn( SvNV_nomg(svl) - value ); |
| 28e5dec8 | 1954 | RETURN; |
| 79072805 | 1955 | } |
| a0d0e21e | 1956 | } |
| 79072805 | 1957 | |
| a0d0e21e LW |
1958 | PP(pp_left_shift) |
| 1959 | { | |
| 6f1401dc | 1960 | dVAR; dSP; dATARGET; SV *svl, *svr; |
| a42d0242 | 1961 | tryAMAGICbin_MG(lshift_amg, AMGf_assign|AMGf_numeric); |
| 6f1401dc DM |
1962 | svr = POPs; |
| 1963 | svl = TOPs; | |
| a0d0e21e | 1964 | { |
| 6f1401dc | 1965 | const IV shift = SvIV_nomg(svr); |
| d0ba1bd2 | 1966 | if (PL_op->op_private & HINT_INTEGER) { |
| 6f1401dc | 1967 | const IV i = SvIV_nomg(svl); |
| 972b05a9 | 1968 | SETi(i << shift); |
| d0ba1bd2 JH |
1969 | } |
| 1970 | else { | |
| 6f1401dc | 1971 | const UV u = SvUV_nomg(svl); |
| 972b05a9 | 1972 | SETu(u << shift); |
| d0ba1bd2 | 1973 | } |
| 55497cff | 1974 | RETURN; |
| 79072805 | 1975 | } |
| a0d0e21e | 1976 | } |
| 79072805 | 1977 | |
| a0d0e21e LW |
1978 | PP(pp_right_shift) |
| 1979 | { | |
| 6f1401dc | 1980 | dVAR; dSP; dATARGET; SV *svl, *svr; |
| a42d0242 | 1981 | tryAMAGICbin_MG(rshift_amg, AMGf_assign|AMGf_numeric); |
| 6f1401dc DM |
1982 | svr = POPs; |
| 1983 | svl = TOPs; | |
| a0d0e21e | 1984 | { |
| 6f1401dc | 1985 | const IV shift = SvIV_nomg(svr); |
| d0ba1bd2 | 1986 | if (PL_op->op_private & HINT_INTEGER) { |
| 6f1401dc | 1987 | const IV i = SvIV_nomg(svl); |
| 972b05a9 | 1988 | SETi(i >> shift); |
| d0ba1bd2 JH |
1989 | } |
| 1990 | else { | |
| 6f1401dc | 1991 | const UV u = SvUV_nomg(svl); |
| 972b05a9 | 1992 | SETu(u >> shift); |
| d0ba1bd2 | 1993 | } |
| a0d0e21e | 1994 | RETURN; |
| 93a17b20 | 1995 | } |
| 79072805 LW |
1996 | } |
| 1997 | ||
| a0d0e21e | 1998 | PP(pp_lt) |
| 79072805 | 1999 | { |
| 6f1401dc | 2000 | dVAR; dSP; |
| a42d0242 | 2001 | tryAMAGICbin_MG(lt_amg, AMGf_set|AMGf_numeric); |
| 28e5dec8 | 2002 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 2003 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2004 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2005 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 JH |
2006 | if (SvIOK(TOPm1s)) { |
| 2007 | bool auvok = SvUOK(TOPm1s); | |
| 2008 | bool buvok = SvUOK(TOPs); | |
| a227d84d | 2009 | |
| 28e5dec8 | 2010 | if (!auvok && !buvok) { /* ## IV < IV ## */ |
| 1b6737cc AL |
2011 | const IV aiv = SvIVX(TOPm1s); |
| 2012 | const IV biv = SvIVX(TOPs); | |
| 28e5dec8 JH |
2013 | |
| 2014 | SP--; | |
| 2015 | SETs(boolSV(aiv < biv)); | |
| 2016 | RETURN; | |
| 2017 | } | |
| 2018 | if (auvok && buvok) { /* ## UV < UV ## */ | |
| 1b6737cc AL |
2019 | const UV auv = SvUVX(TOPm1s); |
| 2020 | const UV buv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2021 | |
| 2022 | SP--; | |
| 2023 | SETs(boolSV(auv < buv)); | |
| 2024 | RETURN; | |
| 2025 | } | |
| 2026 | if (auvok) { /* ## UV < IV ## */ | |
| 2027 | UV auv; | |
| 1b6737cc | 2028 | const IV biv = SvIVX(TOPs); |
| 28e5dec8 JH |
2029 | SP--; |
| 2030 | if (biv < 0) { | |
| 2031 | /* As (a) is a UV, it's >=0, so it cannot be < */ | |
| 2032 | SETs(&PL_sv_no); | |
| 2033 | RETURN; | |
| 2034 | } | |
| 2035 | auv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2036 | SETs(boolSV(auv < (UV)biv)); |
| 2037 | RETURN; | |
| 2038 | } | |
| 2039 | { /* ## IV < UV ## */ | |
| 1b6737cc | 2040 | const IV aiv = SvIVX(TOPm1s); |
| 28e5dec8 JH |
2041 | UV buv; |
| 2042 | ||
| 28e5dec8 JH |
2043 | if (aiv < 0) { |
| 2044 | /* As (b) is a UV, it's >=0, so it must be < */ | |
| 2045 | SP--; | |
| 2046 | SETs(&PL_sv_yes); | |
| 2047 | RETURN; | |
| 2048 | } | |
| 2049 | buv = SvUVX(TOPs); | |
| 2050 | SP--; | |
| 28e5dec8 JH |
2051 | SETs(boolSV((UV)aiv < buv)); |
| 2052 | RETURN; | |
| 2053 | } | |
| 2054 | } | |
| 2055 | } | |
| 2056 | #endif | |
| 30de85b6 | 2057 | #ifndef NV_PRESERVES_UV |
| 50fb3111 NC |
2058 | #ifdef PERL_PRESERVE_IVUV |
| 2059 | else | |
| 2060 | #endif | |
| ed3b9b3c | 2061 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| 0bdaccee NC |
2062 | SP--; |
| 2063 | SETs(boolSV(SvRV(TOPs) < SvRV(TOPp1s))); | |
| 2064 | RETURN; | |
| 2065 | } | |
| 30de85b6 | 2066 | #endif |
| a0d0e21e | 2067 | { |
| cab190d4 | 2068 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 6f1401dc | 2069 | dPOPTOPnnrl_nomg; |
| cab190d4 JD |
2070 | if (Perl_isnan(left) || Perl_isnan(right)) |
| 2071 | RETSETNO; | |
| 2072 | SETs(boolSV(left < right)); | |
| 2073 | #else | |
| 6f1401dc DM |
2074 | dPOPnv_nomg; |
| 2075 | SETs(boolSV(SvNV_nomg(TOPs) < value)); | |
| cab190d4 | 2076 | #endif |
| a0d0e21e | 2077 | RETURN; |
| 79072805 | 2078 | } |
| a0d0e21e | 2079 | } |
| 79072805 | 2080 | |
| a0d0e21e LW |
2081 | PP(pp_gt) |
| 2082 | { | |
| 6f1401dc | 2083 | dVAR; dSP; |
| a42d0242 | 2084 | tryAMAGICbin_MG(gt_amg, AMGf_set|AMGf_numeric); |
| 28e5dec8 | 2085 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 2086 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2087 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2088 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 JH |
2089 | if (SvIOK(TOPm1s)) { |
| 2090 | bool auvok = SvUOK(TOPm1s); | |
| 2091 | bool buvok = SvUOK(TOPs); | |
| a227d84d | 2092 | |
| 28e5dec8 | 2093 | if (!auvok && !buvok) { /* ## IV > IV ## */ |
| 1b6737cc AL |
2094 | const IV aiv = SvIVX(TOPm1s); |
| 2095 | const IV biv = SvIVX(TOPs); | |
| 2096 | ||
| 28e5dec8 JH |
2097 | SP--; |
| 2098 | SETs(boolSV(aiv > biv)); | |
| 2099 | RETURN; | |
| 2100 | } | |
| 2101 | if (auvok && buvok) { /* ## UV > UV ## */ | |
| 1b6737cc AL |
2102 | const UV auv = SvUVX(TOPm1s); |
| 2103 | const UV buv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2104 | |
| 2105 | SP--; | |
| 2106 | SETs(boolSV(auv > buv)); | |
| 2107 | RETURN; | |
| 2108 | } | |
| 2109 | if (auvok) { /* ## UV > IV ## */ | |
| 2110 | UV auv; | |
| 1b6737cc AL |
2111 | const IV biv = SvIVX(TOPs); |
| 2112 | ||
| 28e5dec8 JH |
2113 | SP--; |
| 2114 | if (biv < 0) { | |
| 2115 | /* As (a) is a UV, it's >=0, so it must be > */ | |
| 2116 | SETs(&PL_sv_yes); | |
| 2117 | RETURN; | |
| 2118 | } | |
| 2119 | auv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2120 | SETs(boolSV(auv > (UV)biv)); |
| 2121 | RETURN; | |
| 2122 | } | |
| 2123 | { /* ## IV > UV ## */ | |
| 1b6737cc | 2124 | const IV aiv = SvIVX(TOPm1s); |
| 28e5dec8 JH |
2125 | UV buv; |
| 2126 | ||
| 28e5dec8 JH |
2127 | if (aiv < 0) { |
| 2128 | /* As (b) is a UV, it's >=0, so it cannot be > */ | |
| 2129 | SP--; | |
| 2130 | SETs(&PL_sv_no); | |
| 2131 | RETURN; | |
| 2132 | } | |
| 2133 | buv = SvUVX(TOPs); | |
| 2134 | SP--; | |
| 28e5dec8 JH |
2135 | SETs(boolSV((UV)aiv > buv)); |
| 2136 | RETURN; | |
| 2137 | } | |
| 2138 | } | |
| 2139 | } | |
| 2140 | #endif | |
| 30de85b6 | 2141 | #ifndef NV_PRESERVES_UV |
| 50fb3111 NC |
2142 | #ifdef PERL_PRESERVE_IVUV |
| 2143 | else | |
| 2144 | #endif | |
| ed3b9b3c | 2145 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| 30de85b6 NC |
2146 | SP--; |
| 2147 | SETs(boolSV(SvRV(TOPs) > SvRV(TOPp1s))); | |
| 2148 | RETURN; | |
| 2149 | } | |
| 2150 | #endif | |
| a0d0e21e | 2151 | { |
| cab190d4 | 2152 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 6f1401dc | 2153 | dPOPTOPnnrl_nomg; |
| cab190d4 JD |
2154 | if (Perl_isnan(left) || Perl_isnan(right)) |
| 2155 | RETSETNO; | |
| 2156 | SETs(boolSV(left > right)); | |
| 2157 | #else | |
| 6f1401dc DM |
2158 | dPOPnv_nomg; |
| 2159 | SETs(boolSV(SvNV_nomg(TOPs) > value)); | |
| cab190d4 | 2160 | #endif |
| a0d0e21e | 2161 | RETURN; |
| 79072805 | 2162 | } |
| a0d0e21e LW |
2163 | } |
| 2164 | ||
| 2165 | PP(pp_le) | |
| 2166 | { | |
| 6f1401dc | 2167 | dVAR; dSP; |
| a42d0242 | 2168 | tryAMAGICbin_MG(le_amg, AMGf_set|AMGf_numeric); |
| 28e5dec8 | 2169 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 2170 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2171 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2172 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 JH |
2173 | if (SvIOK(TOPm1s)) { |
| 2174 | bool auvok = SvUOK(TOPm1s); | |
| 2175 | bool buvok = SvUOK(TOPs); | |
| a227d84d | 2176 | |
| 28e5dec8 | 2177 | if (!auvok && !buvok) { /* ## IV <= IV ## */ |
| 1b6737cc AL |
2178 | const IV aiv = SvIVX(TOPm1s); |
| 2179 | const IV biv = SvIVX(TOPs); | |
| 28e5dec8 JH |
2180 | |
| 2181 | SP--; | |
| 2182 | SETs(boolSV(aiv <= biv)); | |
| 2183 | RETURN; | |
| 2184 | } | |
| 2185 | if (auvok && buvok) { /* ## UV <= UV ## */ | |
| 2186 | UV auv = SvUVX(TOPm1s); | |
| 2187 | UV buv = SvUVX(TOPs); | |
| 2188 | ||
| 2189 | SP--; | |
| 2190 | SETs(boolSV(auv <= buv)); | |
| 2191 | RETURN; | |
| 2192 | } | |
| 2193 | if (auvok) { /* ## UV <= IV ## */ | |
| 2194 | UV auv; | |
| 1b6737cc AL |
2195 | const IV biv = SvIVX(TOPs); |
| 2196 | ||
| 28e5dec8 JH |
2197 | SP--; |
| 2198 | if (biv < 0) { | |
| 2199 | /* As (a) is a UV, it's >=0, so a cannot be <= */ | |
| 2200 | SETs(&PL_sv_no); | |
| 2201 | RETURN; | |
| 2202 | } | |
| 2203 | auv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2204 | SETs(boolSV(auv <= (UV)biv)); |
| 2205 | RETURN; | |
| 2206 | } | |
| 2207 | { /* ## IV <= UV ## */ | |
| 1b6737cc | 2208 | const IV aiv = SvIVX(TOPm1s); |
| 28e5dec8 | 2209 | UV buv; |
| 1b6737cc | 2210 | |
| 28e5dec8 JH |
2211 | if (aiv < 0) { |
| 2212 | /* As (b) is a UV, it's >=0, so a must be <= */ | |
| 2213 | SP--; | |
| 2214 | SETs(&PL_sv_yes); | |
| 2215 | RETURN; | |
| 2216 | } | |
| 2217 | buv = SvUVX(TOPs); | |
| 2218 | SP--; | |
| 28e5dec8 JH |
2219 | SETs(boolSV((UV)aiv <= buv)); |
| 2220 | RETURN; | |
| 2221 | } | |
| 2222 | } | |
| 2223 | } | |
| 2224 | #endif | |
| 30de85b6 | 2225 | #ifndef NV_PRESERVES_UV |
| 50fb3111 NC |
2226 | #ifdef PERL_PRESERVE_IVUV |
| 2227 | else | |
| 2228 | #endif | |
| ed3b9b3c | 2229 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| 30de85b6 NC |
2230 | SP--; |
| 2231 | SETs(boolSV(SvRV(TOPs) <= SvRV(TOPp1s))); | |
| 2232 | RETURN; | |
| 2233 | } | |
| 2234 | #endif | |
| a0d0e21e | 2235 | { |
| cab190d4 | 2236 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 6f1401dc | 2237 | dPOPTOPnnrl_nomg; |
| cab190d4 JD |
2238 | if (Perl_isnan(left) || Perl_isnan(right)) |
| 2239 | RETSETNO; | |
| 2240 | SETs(boolSV(left <= right)); | |
| 2241 | #else | |
| 6f1401dc DM |
2242 | dPOPnv_nomg; |
| 2243 | SETs(boolSV(SvNV_nomg(TOPs) <= value)); | |
| cab190d4 | 2244 | #endif |
| a0d0e21e | 2245 | RETURN; |
| 79072805 | 2246 | } |
| a0d0e21e LW |
2247 | } |
| 2248 | ||
| 2249 | PP(pp_ge) | |
| 2250 | { | |
| 6f1401dc | 2251 | dVAR; dSP; |
| a42d0242 | 2252 | tryAMAGICbin_MG(ge_amg,AMGf_set|AMGf_numeric); |
| 28e5dec8 | 2253 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 2254 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2255 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2256 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 JH |
2257 | if (SvIOK(TOPm1s)) { |
| 2258 | bool auvok = SvUOK(TOPm1s); | |
| 2259 | bool buvok = SvUOK(TOPs); | |
| a227d84d | 2260 | |
| 28e5dec8 | 2261 | if (!auvok && !buvok) { /* ## IV >= IV ## */ |
| 1b6737cc AL |
2262 | const IV aiv = SvIVX(TOPm1s); |
| 2263 | const IV biv = SvIVX(TOPs); | |
| 2264 | ||
| 28e5dec8 JH |
2265 | SP--; |
| 2266 | SETs(boolSV(aiv >= biv)); | |
| 2267 | RETURN; | |
| 2268 | } | |
| 2269 | if (auvok && buvok) { /* ## UV >= UV ## */ | |
| 1b6737cc AL |
2270 | const UV auv = SvUVX(TOPm1s); |
| 2271 | const UV buv = SvUVX(TOPs); | |
| 2272 | ||
| 28e5dec8 JH |
2273 | SP--; |
| 2274 | SETs(boolSV(auv >= buv)); | |
| 2275 | RETURN; | |
| 2276 | } | |
| 2277 | if (auvok) { /* ## UV >= IV ## */ | |
| 2278 | UV auv; | |
| 1b6737cc AL |
2279 | const IV biv = SvIVX(TOPs); |
| 2280 | ||
| 28e5dec8 JH |
2281 | SP--; |
| 2282 | if (biv < 0) { | |
| 2283 | /* As (a) is a UV, it's >=0, so it must be >= */ | |
| 2284 | SETs(&PL_sv_yes); | |
| 2285 | RETURN; | |
| 2286 | } | |
| 2287 | auv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2288 | SETs(boolSV(auv >= (UV)biv)); |
| 2289 | RETURN; | |
| 2290 | } | |
| 2291 | { /* ## IV >= UV ## */ | |
| 1b6737cc | 2292 | const IV aiv = SvIVX(TOPm1s); |
| 28e5dec8 | 2293 | UV buv; |
| 1b6737cc | 2294 | |
| 28e5dec8 JH |
2295 | if (aiv < 0) { |
| 2296 | /* As (b) is a UV, it's >=0, so a cannot be >= */ | |
| 2297 | SP--; | |
| 2298 | SETs(&PL_sv_no); | |
| 2299 | RETURN; | |
| 2300 | } | |
| 2301 | buv = SvUVX(TOPs); | |
| 2302 | SP--; | |
| 28e5dec8 JH |
2303 | SETs(boolSV((UV)aiv >= buv)); |
| 2304 | RETURN; | |
| 2305 | } | |
| 2306 | } | |
| 2307 | } | |
| 2308 | #endif | |
| 30de85b6 | 2309 | #ifndef NV_PRESERVES_UV |
| 50fb3111 NC |
2310 | #ifdef PERL_PRESERVE_IVUV |
| 2311 | else | |
| 2312 | #endif | |
| ed3b9b3c | 2313 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| 30de85b6 NC |
2314 | SP--; |
| 2315 | SETs(boolSV(SvRV(TOPs) >= SvRV(TOPp1s))); | |
| 2316 | RETURN; | |
| 2317 | } | |
| 2318 | #endif | |
| a0d0e21e | 2319 | { |
| cab190d4 | 2320 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 6f1401dc | 2321 | dPOPTOPnnrl_nomg; |
| cab190d4 JD |
2322 | if (Perl_isnan(left) || Perl_isnan(right)) |
| 2323 | RETSETNO; | |
| 2324 | SETs(boolSV(left >= right)); | |
| 2325 | #else | |
| 6f1401dc DM |
2326 | dPOPnv_nomg; |
| 2327 | SETs(boolSV(SvNV_nomg(TOPs) >= value)); | |
| cab190d4 | 2328 | #endif |
| a0d0e21e | 2329 | RETURN; |
| 79072805 | 2330 | } |
| a0d0e21e | 2331 | } |
| 79072805 | 2332 | |
| a0d0e21e LW |
2333 | PP(pp_ne) |
| 2334 | { | |
| 6f1401dc | 2335 | dVAR; dSP; |
| a42d0242 | 2336 | tryAMAGICbin_MG(ne_amg,AMGf_set|AMGf_numeric); |
| 3bb2c415 | 2337 | #ifndef NV_PRESERVES_UV |
| ed3b9b3c | 2338 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| e61d22ef NC |
2339 | SP--; |
| 2340 | SETs(boolSV(SvRV(TOPs) != SvRV(TOPp1s))); | |
| 3bb2c415 JH |
2341 | RETURN; |
| 2342 | } | |
| 2343 | #endif | |
| 28e5dec8 | 2344 | #ifdef PERL_PRESERVE_IVUV |
| 6f1401dc | 2345 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2346 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2347 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 | 2348 | if (SvIOK(TOPm1s)) { |
| 0bd48802 AL |
2349 | const bool auvok = SvUOK(TOPm1s); |
| 2350 | const bool buvok = SvUOK(TOPs); | |
| a227d84d | 2351 | |
| 30de85b6 NC |
2352 | if (auvok == buvok) { /* ## IV == IV or UV == UV ## */ |
| 2353 | /* Casting IV to UV before comparison isn't going to matter | |
| 2354 | on 2s complement. On 1s complement or sign&magnitude | |
| 2355 | (if we have any of them) it could make negative zero | |
| 2356 | differ from normal zero. As I understand it. (Need to | |
| 2357 | check - is negative zero implementation defined behaviour | |
| 2358 | anyway?). NWC */ | |
| 1b6737cc AL |
2359 | const UV buv = SvUVX(POPs); |
| 2360 | const UV auv = SvUVX(TOPs); | |
| 2361 | ||
| 28e5dec8 JH |
2362 | SETs(boolSV(auv != buv)); |
| 2363 | RETURN; | |
| 2364 | } | |
| 2365 | { /* ## Mixed IV,UV ## */ | |
| 2366 | IV iv; | |
| 2367 | UV uv; | |
| 2368 | ||
| 2369 | /* != is commutative so swap if needed (save code) */ | |
| 2370 | if (auvok) { | |
| 2371 | /* swap. top of stack (b) is the iv */ | |
| 2372 | iv = SvIVX(TOPs); | |
| 2373 | SP--; | |
| 2374 | if (iv < 0) { | |
| 2375 | /* As (a) is a UV, it's >0, so it cannot be == */ | |
| 2376 | SETs(&PL_sv_yes); | |
| 2377 | RETURN; | |
| 2378 | } | |
| 2379 | uv = SvUVX(TOPs); | |
| 2380 | } else { | |
| 2381 | iv = SvIVX(TOPm1s); | |
| 2382 | SP--; | |
| 2383 | if (iv < 0) { | |
| 2384 | /* As (b) is a UV, it's >0, so it cannot be == */ | |
| 2385 | SETs(&PL_sv_yes); | |
| 2386 | RETURN; | |
| 2387 | } | |
| 2388 | uv = SvUVX(*(SP+1)); /* Do I want TOPp1s() ? */ | |
| 2389 | } | |
| 28e5dec8 JH |
2390 | SETs(boolSV((UV)iv != uv)); |
| 2391 | RETURN; | |
| 2392 | } | |
| 2393 | } | |
| 2394 | } | |
| 2395 | #endif | |
| a0d0e21e | 2396 | { |
| cab190d4 | 2397 | #if defined(NAN_COMPARE_BROKEN) && defined(Perl_isnan) |
| 6f1401dc | 2398 | dPOPTOPnnrl_nomg; |
| cab190d4 JD |
2399 | if (Perl_isnan(left) || Perl_isnan(right)) |
| 2400 | RETSETYES; | |
| 2401 | SETs(boolSV(left != right)); | |
| 2402 | #else | |
| 6f1401dc DM |
2403 | dPOPnv_nomg; |
| 2404 | SETs(boolSV(SvNV_nomg(TOPs) != value)); | |
| cab190d4 | 2405 | #endif |
| a0d0e21e LW |
2406 | RETURN; |
| 2407 | } | |
| 79072805 LW |
2408 | } |
| 2409 | ||
| a0d0e21e | 2410 | PP(pp_ncmp) |
| 79072805 | 2411 | { |
| 6f1401dc | 2412 | dVAR; dSP; dTARGET; |
| a42d0242 | 2413 | tryAMAGICbin_MG(ncmp_amg, AMGf_numeric); |
| d8c7644e | 2414 | #ifndef NV_PRESERVES_UV |
| ed3b9b3c | 2415 | if (SvROK(TOPs) && !SvAMAGIC(TOPs) && SvROK(TOPm1s) && !SvAMAGIC(TOPm1s)) { |
| 0bd48802 AL |
2416 | const UV right = PTR2UV(SvRV(POPs)); |
| 2417 | const UV left = PTR2UV(SvRV(TOPs)); | |
| e61d22ef | 2418 | SETi((left > right) - (left < right)); |
| d8c7644e JH |
2419 | RETURN; |
| 2420 | } | |
| 2421 | #endif | |
| 28e5dec8 JH |
2422 | #ifdef PERL_PRESERVE_IVUV |
| 2423 | /* Fortunately it seems NaN isn't IOK */ | |
| 6f1401dc | 2424 | SvIV_please_nomg(TOPs); |
| 28e5dec8 | 2425 | if (SvIOK(TOPs)) { |
| 6f1401dc | 2426 | SvIV_please_nomg(TOPm1s); |
| 28e5dec8 | 2427 | if (SvIOK(TOPm1s)) { |
| 1b6737cc AL |
2428 | const bool leftuvok = SvUOK(TOPm1s); |
| 2429 | const bool rightuvok = SvUOK(TOPs); | |
| 28e5dec8 JH |
2430 | I32 value; |
| 2431 | if (!leftuvok && !rightuvok) { /* ## IV <=> IV ## */ | |
| 1b6737cc AL |
2432 | const IV leftiv = SvIVX(TOPm1s); |
| 2433 | const IV rightiv = SvIVX(TOPs); | |
| 28e5dec8 JH |
2434 | |
| 2435 | if (leftiv > rightiv) | |
| 2436 | value = 1; | |
| 2437 | else if (leftiv < rightiv) | |
| 2438 | value = -1; | |
| 2439 | else | |
| 2440 | value = 0; | |
| 2441 | } else if (leftuvok && rightuvok) { /* ## UV <=> UV ## */ | |
| 1b6737cc AL |
2442 | const UV leftuv = SvUVX(TOPm1s); |
| 2443 | const UV rightuv = SvUVX(TOPs); | |
| 28e5dec8 JH |
2444 | |
| 2445 | if (leftuv > rightuv) | |
| 2446 | value = 1; | |
| 2447 | else if (leftuv < rightuv) | |
| 2448 | value = -1; | |
| 2449 | else | |
| 2450 | value = 0; | |
| 2451 | } else if (leftuvok) { /* ## UV <=> IV ## */ | |
| 1b6737cc | 2452 | const IV rightiv = SvIVX(TOPs); |
| 28e5dec8 JH |
2453 | if (rightiv < 0) { |
| 2454 | /* As (a) is a UV, it's >=0, so it cannot be < */ | |
| 2455 | value = 1; | |
| 2456 | } else { | |
| 1b6737cc | 2457 | const UV leftuv = SvUVX(TOPm1s); |
| 83bac5dd | 2458 | if (leftuv > (UV)rightiv) { |
| 28e5dec8 JH |
2459 | value = 1; |
| 2460 | } else if (leftuv < (UV)rightiv) { | |
| 2461 | value = -1; | |
| 2462 | } else { | |
| 2463 | value = 0; | |
| 2464 | } | |
| 2465 | } | |
| 2466 | } else { /* ## IV <=> UV ## */ | |
| 1b6737cc | 2467 | const IV leftiv = SvIVX(TOPm1s); |
| 28e5dec8 JH |
2468 | if (leftiv < 0) { |
| 2469 | /* As (b) is a UV, it's >=0, so it must be < */ | |
| 2470 | value = -1; | |
| 2471 | } else { | |
| 1b6737cc | 2472 | const UV rightuv = SvUVX(TOPs); |
| 83bac5dd | 2473 | if ((UV)leftiv > rightuv) { |
| 28e5dec8 | 2474 | value = 1; |
| 83bac5dd | 2475 | } else if ((UV)leftiv < rightuv) { |
| 28e5dec8 JH |
2476 | value = -1; |
| 2477 | } else { | |
| 2478 | value = 0; | |
| 2479 | } | |
| 2480 | } | |
| 2481 | } | |
| 2482 | SP--; | |
| 2483 | SETi(value); | |
| 2484 | RETURN; | |
| 2485 | } | |
| 2486 | } | |
| 2487 | #endif | |
| a0d0e21e | 2488 | { |
| 6f1401dc | 2489 | dPOPTOPnnrl_nomg; |
| a0d0e21e | 2490 | I32 value; |
| 79072805 | 2491 | |
| a3540c92 | 2492 | #ifdef Perl_isnan |
| 1ad04cfd JH |
2493 | if (Perl_isnan(left) || Perl_isnan(right)) { |
| 2494 | SETs(&PL_sv_undef); | |
| 2495 | RETURN; | |
| 2496 | } | |
| 2497 | value = (left > right) - (left < right); | |
| 2498 | #else | |
| ff0cee69 | 2499 | if (left == right) |
| a0d0e21e | 2500 | value = 0; |
| a0d0e21e LW |
2501 | else if (left < right) |
| 2502 | value = -1; | |
| 44a8e56a | 2503 | else if (left > right) |
| 2504 | value = 1; | |
| 2505 | else { | |
| 3280af22 | 2506 | SETs(&PL_sv_undef); |
| 44a8e56a | 2507 | RETURN; |
| 2508 | } | |
| 1ad04cfd | 2509 | #endif |
| a0d0e21e LW |
2510 | SETi(value); |
| 2511 | RETURN; | |
| 79072805 | 2512 | } |
| a0d0e21e | 2513 | } |
| 79072805 | 2514 | |
| afd9910b | 2515 | PP(pp_sle) |
| a0d0e21e | 2516 | { |
| 97aff369 | 2517 | dVAR; dSP; |
| 79072805 | 2518 | |
| afd9910b NC |
2519 | int amg_type = sle_amg; |
| 2520 | int multiplier = 1; | |
| 2521 | int rhs = 1; | |
| 79072805 | 2522 | |
| afd9910b NC |
2523 | switch (PL_op->op_type) { |
| 2524 | case OP_SLT: | |
| 2525 | amg_type = slt_amg; | |
| 2526 | /* cmp < 0 */ | |
| 2527 | rhs = 0; | |
| 2528 | break; | |
| 2529 | case OP_SGT: | |
| 2530 | amg_type = sgt_amg; | |
| 2531 | /* cmp > 0 */ | |
| 2532 | multiplier = -1; | |
| 2533 | rhs = 0; | |
| 2534 | break; | |
| 2535 | case OP_SGE: | |
| 2536 | amg_type = sge_amg; | |
| 2537 | /* cmp >= 0 */ | |
| 2538 | multiplier = -1; | |
| 2539 | break; | |
| 79072805 | 2540 | } |
| 79072805 | 2541 | |
| 6f1401dc | 2542 | tryAMAGICbin_MG(amg_type, AMGf_set); |
| a0d0e21e LW |
2543 | { |
| 2544 | dPOPTOPssrl; | |
| 1b6737cc | 2545 | const int cmp = (IN_LOCALE_RUNTIME |
| 078504b2 FC |
2546 | ? sv_cmp_locale_flags(left, right, 0) |
| 2547 | : sv_cmp_flags(left, right, 0)); | |
| afd9910b | 2548 | SETs(boolSV(cmp * multiplier < rhs)); |
| a0d0e21e LW |
2549 | RETURN; |
| 2550 | } | |
| 2551 | } | |
| 79072805 | 2552 | |
| 36477c24 | 2553 | PP(pp_seq) |
| 2554 | { | |
| 6f1401dc DM |
2555 | dVAR; dSP; |
| 2556 | tryAMAGICbin_MG(seq_amg, AMGf_set); | |
| 36477c24 | 2557 | { |
| 2558 | dPOPTOPssrl; | |
| 078504b2 | 2559 | SETs(boolSV(sv_eq_flags(left, right, 0))); |
| a0d0e21e LW |
2560 | RETURN; |
| 2561 | } | |
| 2562 | } | |
| 79072805 | 2563 | |
| a0d0e21e | 2564 | PP(pp_sne) |
| 79072805 | 2565 | { |
| 6f1401dc DM |
2566 | dVAR; dSP; |
| 2567 | tryAMAGICbin_MG(sne_amg, AMGf_set); | |
| a0d0e21e LW |
2568 | { |
| 2569 | dPOPTOPssrl; | |
| 078504b2 | 2570 | SETs(boolSV(!sv_eq_flags(left, right, 0))); |
| a0d0e21e | 2571 | RETURN; |
| 463ee0b2 | 2572 | } |
| 79072805 LW |
2573 | } |
| 2574 | ||
| a0d0e21e | 2575 | PP(pp_scmp) |
| 79072805 | 2576 | { |
| 6f1401dc DM |
2577 | dVAR; dSP; dTARGET; |
| 2578 | tryAMAGICbin_MG(scmp_amg, 0); | |
| a0d0e21e LW |
2579 | { |
| 2580 | dPOPTOPssrl; | |
| 1b6737cc | 2581 | const int cmp = (IN_LOCALE_RUNTIME |
| 078504b2 FC |
2582 | ? sv_cmp_locale_flags(left, right, 0) |
| 2583 | : sv_cmp_flags(left, right, 0)); | |
| bbce6d69 | 2584 | SETi( cmp ); |
| a0d0e21e LW |
2585 | RETURN; |
| 2586 | } | |
| 2587 | } | |
| 79072805 | 2588 | |
| 55497cff | 2589 | PP(pp_bit_and) |
| 2590 | { | |
| 6f1401dc DM |
2591 | dVAR; dSP; dATARGET; |
| 2592 | tryAMAGICbin_MG(band_amg, AMGf_assign); | |
| a0d0e21e LW |
2593 | { |
| 2594 | dPOPTOPssrl; | |
| 4633a7c4 | 2595 | if (SvNIOKp(left) || SvNIOKp(right)) { |
| b20c4ee1 FC |
2596 | const bool left_ro_nonnum = !SvNIOKp(left) && SvREADONLY(left); |
| 2597 | const bool right_ro_nonnum = !SvNIOKp(right) && SvREADONLY(right); | |
| d0ba1bd2 | 2598 | if (PL_op->op_private & HINT_INTEGER) { |
| 1b6737cc | 2599 | const IV i = SvIV_nomg(left) & SvIV_nomg(right); |
| 972b05a9 | 2600 | SETi(i); |
| d0ba1bd2 JH |
2601 | } |
| 2602 | else { | |
| 1b6737cc | 2603 | const UV u = SvUV_nomg(left) & SvUV_nomg(right); |
| 972b05a9 | 2604 | SETu(u); |
| d0ba1bd2 | 2605 | } |
| b20c4ee1 FC |
2606 | if (left_ro_nonnum) SvNIOK_off(left); |
| 2607 | if (right_ro_nonnum) SvNIOK_off(right); | |
| a0d0e21e LW |
2608 | } |
| 2609 | else { | |
| 533c011a | 2610 | do_vop(PL_op->op_type, TARG, left, right); |
| a0d0e21e LW |
2611 | SETTARG; |
| 2612 | } | |
| 2613 | RETURN; | |
| 2614 | } | |
| 2615 | } | |
| 79072805 | 2616 | |
| a0d0e21e LW |
2617 | PP(pp_bit_or) |
| 2618 | { | |
| 3658c1f1 NC |
2619 | dVAR; dSP; dATARGET; |
| 2620 | const int op_type = PL_op->op_type; | |
| 2621 | ||
| 6f1401dc | 2622 | tryAMAGICbin_MG((op_type == OP_BIT_OR ? bor_amg : bxor_amg), AMGf_assign); |
| a0d0e21e LW |
2623 | { |
| 2624 | dPOPTOPssrl; | |
| 4633a7c4 | 2625 | if (SvNIOKp(left) || SvNIOKp(right)) { |
| b20c4ee1 FC |
2626 | const bool left_ro_nonnum = !SvNIOKp(left) && SvREADONLY(left); |
| 2627 | const bool right_ro_nonnum = !SvNIOKp(right) && SvREADONLY(right); | |
| d0ba1bd2 | 2628 | if (PL_op->op_private & HINT_INTEGER) { |
| 3658c1f1 NC |
2629 | const IV l = (USE_LEFT(left) ? SvIV_nomg(left) : 0); |
| 2630 | const IV r = SvIV_nomg(right); | |
| 2631 | const IV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
| 2632 | SETi(result); | |
| d0ba1bd2 JH |
2633 | } |
| 2634 | else { | |
| 3658c1f1 NC |
2635 | const UV l = (USE_LEFT(left) ? SvUV_nomg(left) : 0); |
| 2636 | const UV r = SvUV_nomg(right); | |
| 2637 | const UV result = op_type == OP_BIT_OR ? (l | r) : (l ^ r); | |
| 2638 | SETu(result); | |
| d0ba1bd2 | 2639 | } |
| b20c4ee1 FC |
2640 | if (left_ro_nonnum) SvNIOK_off(left); |
| 2641 | if (right_ro_nonnum) SvNIOK_off(right); | |
| a0d0e21e LW |
2642 | } |
| 2643 | else { | |
| 3658c1f1 | 2644 | do_vop(op_type, TARG, left, right); |
| a0d0e21e LW |
2645 | SETTARG; |
| 2646 | } | |
| 2647 | RETURN; | |
| 79072805 | 2648 | } |
| a0d0e21e | 2649 | } |
| 79072805 | 2650 | |
| a0d0e21e LW |
2651 | PP(pp_negate) |
| 2652 | { | |
| 6f1401dc DM |
2653 | dVAR; dSP; dTARGET; |
| 2654 | tryAMAGICun_MG(neg_amg, AMGf_numeric); | |
| a0d0e21e | 2655 | { |
| 6f1401dc | 2656 | SV * const sv = TOPs; |
| 1b6737cc | 2657 | const int flags = SvFLAGS(sv); |
| a5b92898 | 2658 | |
| 886a4465 | 2659 | if( !SvNIOK( sv ) && looks_like_number( sv ) ){ |
| a5b92898 RB |
2660 | SvIV_please( sv ); |
| 2661 | } | |
| 2662 | ||
| 28e5dec8 JH |
2663 | if ((flags & SVf_IOK) || ((flags & (SVp_IOK | SVp_NOK)) == SVp_IOK)) { |
| 2664 | /* It's publicly an integer, or privately an integer-not-float */ | |
| 2665 | oops_its_an_int: | |
| 9b0e499b GS |
2666 | if (SvIsUV(sv)) { |
| 2667 | if (SvIVX(sv) == IV_MIN) { | |
| 28e5dec8 | 2668 | /* 2s complement assumption. */ |
| 9b0e499b GS |
2669 | SETi(SvIVX(sv)); /* special case: -((UV)IV_MAX+1) == IV_MIN */ |
| 2670 | RETURN; | |
| 2671 | } | |
| 2672 | else if (SvUVX(sv) <= IV_MAX) { | |
| beccb14c | 2673 | SETi(-SvIVX(sv)); |
| 9b0e499b GS |
2674 | RETURN; |
| 2675 | } | |
| 2676 | } | |
| 2677 | else if (SvIVX(sv) != IV_MIN) { | |
| 2678 | SETi(-SvIVX(sv)); | |
| 2679 | RETURN; | |
| 2680 | } | |
| 28e5dec8 JH |
2681 | #ifdef PERL_PRESERVE_IVUV |
| 2682 | else { | |
| 2683 | SETu((UV)IV_MIN); | |
| 2684 | RETURN; | |
| 2685 | } | |
| 2686 | #endif | |
| 9b0e499b GS |
2687 | } |
| 2688 | if (SvNIOKp(sv)) | |
| 6f1401dc | 2689 | SETn(-SvNV_nomg(sv)); |
| 4633a7c4 | 2690 | else if (SvPOKp(sv)) { |
| a0d0e21e | 2691 | STRLEN len; |
| 6f1401dc | 2692 | const char * const s = SvPV_nomg_const(sv, len); |
| bbce6d69 | 2693 | if (isIDFIRST(*s)) { |
| 76f68e9b | 2694 | sv_setpvs(TARG, "-"); |
| a0d0e21e | 2695 | sv_catsv(TARG, sv); |
| 79072805 | 2696 | } |
| a0d0e21e | 2697 | else if (*s == '+' || *s == '-') { |
| 6f1401dc DM |
2698 | sv_setsv_nomg(TARG, sv); |
| 2699 | *SvPV_force_nomg(TARG, len) = *s == '-' ? '+' : '-'; | |
| 79072805 | 2700 | } |
| 8eb28a70 | 2701 | else if (DO_UTF8(sv)) { |
| 6f1401dc | 2702 | SvIV_please_nomg(sv); |
| 8eb28a70 JH |
2703 | if (SvIOK(sv)) |
| 2704 | goto oops_its_an_int; | |
| 2705 | if (SvNOK(sv)) | |
| 6f1401dc | 2706 | sv_setnv(TARG, -SvNV_nomg(sv)); |
| 8eb28a70 | 2707 | else { |
| 76f68e9b | 2708 | sv_setpvs(TARG, "-"); |
| 8eb28a70 JH |
2709 | sv_catsv(TARG, sv); |
| 2710 | } | |
| 834a4ddd | 2711 | } |
| 28e5dec8 | 2712 | else { |
| 6f1401dc | 2713 | SvIV_please_nomg(sv); |
| 8eb28a70 JH |
2714 | if (SvIOK(sv)) |
| 2715 | goto oops_its_an_int; | |
| 6f1401dc | 2716 | sv_setnv(TARG, -SvNV_nomg(sv)); |
| 28e5dec8 | 2717 | } |
| a0d0e21e | 2718 | SETTARG; |
| 79072805 | 2719 | } |
| 4633a7c4 | 2720 | else |
| 6f1401dc | 2721 | SETn(-SvNV_nomg(sv)); |
| 79072805 | 2722 | } |
| a0d0e21e | 2723 | RETURN; |
| 79072805 LW |
2724 | } |
| 2725 | ||
| a0d0e21e | 2726 | PP(pp_not) |
| 79072805 | 2727 | { |
| 6f1401dc DM |
2728 | dVAR; dSP; |
| 2729 | tryAMAGICun_MG(not_amg, AMGf_set); | |
| 06c841cf | 2730 | *PL_stack_sp = boolSV(!SvTRUE_nomg(*PL_stack_sp)); |
| a0d0e21e | 2731 | return NORMAL; |
| 79072805 LW |
2732 | } |
| 2733 | ||
| a0d0e21e | 2734 | PP(pp_complement) |
| 79072805 | 2735 | { |
| 6f1401dc | 2736 | dVAR; dSP; dTARGET; |
| a42d0242 | 2737 | tryAMAGICun_MG(compl_amg, AMGf_numeric); |
| a0d0e21e LW |
2738 | { |
| 2739 | dTOPss; | |
| 4633a7c4 | 2740 | if (SvNIOKp(sv)) { |
| d0ba1bd2 | 2741 | if (PL_op->op_private & HINT_INTEGER) { |
| 1b6737cc | 2742 | const IV i = ~SvIV_nomg(sv); |
| 972b05a9 | 2743 | SETi(i); |
| d0ba1bd2 JH |
2744 | } |
| 2745 | else { | |
| 1b6737cc | 2746 | const UV u = ~SvUV_nomg(sv); |
| 972b05a9 | 2747 | SETu(u); |
| d0ba1bd2 | 2748 | } |
| a0d0e21e LW |
2749 | } |
| 2750 | else { | |
| 51723571 | 2751 | register U8 *tmps; |
| 55497cff | 2752 | register I32 anum; |
| a0d0e21e LW |
2753 | STRLEN len; |
| 2754 | ||
| 10516c54 | 2755 | (void)SvPV_nomg_const(sv,len); /* force check for uninit var */ |
| 891f9566 | 2756 | sv_setsv_nomg(TARG, sv); |
| 6f1401dc | 2757 | tmps = (U8*)SvPV_force_nomg(TARG, len); |
| a0d0e21e | 2758 | anum = len; |
| 1d68d6cd | 2759 | if (SvUTF8(TARG)) { |
| a1ca4561 | 2760 | /* Calculate exact length, let's not estimate. */ |
| 1d68d6cd | 2761 | STRLEN targlen = 0; |
| ba210ebe | 2762 | STRLEN l; |
| a1ca4561 YST |
2763 | UV nchar = 0; |
| 2764 | UV nwide = 0; | |
| 01f6e806 | 2765 | U8 * const send = tmps + len; |
| 74d49cd0 TS |
2766 | U8 * const origtmps = tmps; |
| 2767 | const UV utf8flags = UTF8_ALLOW_ANYUV; | |
| 1d68d6cd | 2768 | |
| 1d68d6cd | 2769 | while (tmps < send) { |
| 74d49cd0 TS |
2770 | const UV c = utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
| 2771 | tmps += l; | |
| 5bbb0b5a | 2772 | targlen += UNISKIP(~c); |
| a1ca4561 YST |
2773 | nchar++; |
| 2774 | if (c > 0xff) | |
| 2775 | nwide++; | |
| 1d68d6cd SC |
2776 | } |
| 2777 | ||
| 2778 | /* Now rewind strings and write them. */ | |
| 74d49cd0 | 2779 | tmps = origtmps; |
| a1ca4561 YST |
2780 | |
| 2781 | if (nwide) { | |
| 01f6e806 AL |
2782 | U8 *result; |
| 2783 | U8 *p; | |
| 2784 | ||
| 74d49cd0 | 2785 | Newx(result, targlen + 1, U8); |
| 01f6e806 | 2786 | p = result; |
| a1ca4561 | 2787 | while (tmps < send) { |
| 74d49cd0 TS |
2788 | const UV c = utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
| 2789 | tmps += l; | |
| 01f6e806 | 2790 | p = uvchr_to_utf8_flags(p, ~c, UNICODE_ALLOW_ANY); |
| a1ca4561 | 2791 | } |
| 01f6e806 | 2792 | *p = '\0'; |
| c1c21316 NC |
2793 | sv_usepvn_flags(TARG, (char*)result, targlen, |
| 2794 | SV_HAS_TRAILING_NUL); | |
| a1ca4561 YST |
2795 | SvUTF8_on(TARG); |
| 2796 | } | |
| 2797 | else { | |
| 01f6e806 AL |
2798 | U8 *result; |
| 2799 | U8 *p; | |
| 2800 | ||
| 74d49cd0 | 2801 | Newx(result, nchar + 1, U8); |
| 01f6e806 | 2802 | p = result; |
| a1ca4561 | 2803 | while (tmps < send) { |
| 74d49cd0 TS |
2804 | const U8 c = (U8)utf8n_to_uvchr(tmps, send-tmps, &l, utf8flags); |
| 2805 | tmps += l; | |
| 01f6e806 | 2806 | *p++ = ~c; |
| a1ca4561 | 2807 | } |
| 01f6e806 | 2808 | *p = '\0'; |
| c1c21316 | 2809 | sv_usepvn_flags(TARG, (char*)result, nchar, SV_HAS_TRAILING_NUL); |
| d0a21e00 | 2810 | SvUTF8_off(TARG); |
| 1d68d6cd | 2811 | } |
| ec93b65f | 2812 | SETTARG; |
| 1d68d6cd SC |
2813 | RETURN; |
| 2814 | } | |
| a0d0e21e | 2815 | #ifdef LIBERAL |
| 51723571 JH |
2816 | { |
| 2817 | register long *tmpl; | |
| 2818 | for ( ; anum && (unsigned long)tmps % sizeof(long); anum--, tmps++) | |
| 2819 | *tmps = ~*tmps; | |
| 2820 | tmpl = (long*)tmps; | |
| bb7a0f54 | 2821 | for ( ; anum >= (I32)sizeof(long); anum -= (I32)sizeof(long), tmpl++) |
| 51723571 JH |
2822 | *tmpl = ~*tmpl; |
| 2823 | tmps = (U8*)tmpl; | |
| 2824 | } | |
| a0d0e21e LW |
2825 | #endif |
| 2826 | for ( ; anum > 0; anum--, tmps++) | |
| 2827 | *tmps = ~*tmps; | |
| ec93b65f | 2828 | SETTARG; |
| a0d0e21e LW |
2829 | } |
| 2830 | RETURN; | |
| 2831 | } | |
| 79072805 LW |
2832 | } |
| 2833 | ||
| a0d0e21e LW |
2834 | /* integer versions of some of the above */ |
| 2835 | ||
| a0d0e21e | 2836 | PP(pp_i_multiply) |
| 79072805 | 2837 | { |
| 6f1401dc DM |
2838 | dVAR; dSP; dATARGET; |
| 2839 | tryAMAGICbin_MG(mult_amg, AMGf_assign); | |
| a0d0e21e | 2840 | { |
| 6f1401dc | 2841 | dPOPTOPiirl_nomg; |
| a0d0e21e LW |
2842 | SETi( left * right ); |
| 2843 | RETURN; | |
| 2844 | } | |
| 79072805 LW |
2845 | } |
| 2846 | ||
| a0d0e21e | 2847 | PP(pp_i_divide) |
| 79072805 | 2848 | { |
| ece1bcef | 2849 | IV num; |
| 6f1401dc DM |
2850 | dVAR; dSP; dATARGET; |
| 2851 | tryAMAGICbin_MG(div_amg, AMGf_assign); | |
| a0d0e21e | 2852 | { |
| 6f1401dc DM |
2853 | dPOPTOPssrl; |
| 2854 | IV value = SvIV_nomg(right); | |
| a0d0e21e | 2855 | if (value == 0) |
| ece1bcef | 2856 | DIE(aTHX_ "Illegal division by zero"); |
| 6f1401dc | 2857 | num = SvIV_nomg(left); |
| a0cec769 YST |
2858 | |
| 2859 | /* avoid FPE_INTOVF on some platforms when num is IV_MIN */ | |
| 2860 | if (value == -1) | |
| 2861 | value = - num; | |
| 2862 | else | |
| 2863 | value = num / value; | |
| 6f1401dc | 2864 | SETi(value); |
| a0d0e21e LW |
2865 | RETURN; |
| 2866 | } | |
| 79072805 LW |
2867 | } |
| 2868 | ||
| befad5d1 | 2869 | #if defined(__GLIBC__) && IVSIZE == 8 |
| 224ec323 JH |
2870 | STATIC |
| 2871 | PP(pp_i_modulo_0) | |
| befad5d1 NC |
2872 | #else |
| 2873 | PP(pp_i_modulo) | |
| 2874 | #endif | |
| 224ec323 JH |
2875 | { |
| 2876 | /* This is the vanilla old i_modulo. */ | |
| 6f1401dc DM |
2877 | dVAR; dSP; dATARGET; |
| 2878 | tryAMAGICbin_MG(modulo_amg, AMGf_assign); | |
| 224ec323 | 2879 | { |
| 6f1401dc | 2880 | dPOPTOPiirl_nomg; |
| 224ec323 JH |
2881 | if (!right) |
| 2882 | DIE(aTHX_ "Illegal modulus zero"); | |
| a0cec769 YST |
2883 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
| 2884 | if (right == -1) | |
| 2885 | SETi( 0 ); | |
| 2886 | else | |
| 2887 | SETi( left % right ); | |
| 224ec323 JH |
2888 | RETURN; |
| 2889 | } | |
| 2890 | } | |
| 2891 | ||
| 11010fa3 | 2892 | #if defined(__GLIBC__) && IVSIZE == 8 |
| 224ec323 JH |
2893 | STATIC |
| 2894 | PP(pp_i_modulo_1) | |
| befad5d1 | 2895 | |
| 224ec323 | 2896 | { |
| 224ec323 | 2897 | /* This is the i_modulo with the workaround for the _moddi3 bug |
| fce2b89e | 2898 | * in (at least) glibc 2.2.5 (the PERL_ABS() the workaround). |
| 224ec323 | 2899 | * See below for pp_i_modulo. */ |
| 6f1401dc DM |
2900 | dVAR; dSP; dATARGET; |
| 2901 | tryAMAGICbin_MG(modulo_amg, AMGf_assign); | |
| 224ec323 | 2902 | { |
| 6f1401dc | 2903 | dPOPTOPiirl_nomg; |
| 224ec323 JH |
2904 | if (!right) |
| 2905 | DIE(aTHX_ "Illegal modulus zero"); | |
| a0cec769 YST |
2906 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
| 2907 | if (right == -1) | |
| 2908 | SETi( 0 ); | |
| 2909 | else | |
| 2910 | SETi( left % PERL_ABS(right) ); | |
| 224ec323 JH |
2911 | RETURN; |
| 2912 | } | |
| 224ec323 JH |
2913 | } |
| 2914 | ||
| a0d0e21e | 2915 | PP(pp_i_modulo) |
| 79072805 | 2916 | { |
| 6f1401dc DM |
2917 | dVAR; dSP; dATARGET; |
| 2918 | tryAMAGICbin_MG(modulo_amg, AMGf_assign); | |
| 224ec323 | 2919 | { |
| 6f1401dc | 2920 | dPOPTOPiirl_nomg; |
| 224ec323 JH |
2921 | if (!right) |
| 2922 | DIE(aTHX_ "Illegal modulus zero"); | |
| 2923 | /* The assumption is to use hereafter the old vanilla version... */ | |
| 2924 | PL_op->op_ppaddr = | |
| 2925 | PL_ppaddr[OP_I_MODULO] = | |
| 1c127fab | 2926 | Perl_pp_i_modulo_0; |
| 224ec323 JH |
2927 | /* .. but if we have glibc, we might have a buggy _moddi3 |
| 2928 | * (at least glicb 2.2.5 is known to have this bug), in other | |
| 2929 | * words our integer modulus with negative quad as the second | |
| 2930 | * argument might be broken. Test for this and re-patch the | |
| 2931 | * opcode dispatch table if that is the case, remembering to | |
| 2932 | * also apply the workaround so that this first round works | |
| 2933 | * right, too. See [perl #9402] for more information. */ | |
| 224ec323 JH |
2934 | { |
| 2935 | IV l = 3; | |
| 2936 | IV r = -10; | |
| 2937 | /* Cannot do this check with inlined IV constants since | |
| 2938 | * that seems to work correctly even with the buggy glibc. */ | |
| 2939 | if (l % r == -3) { | |
| 2940 | /* Yikes, we have the bug. | |
| 2941 | * Patch in the workaround version. */ | |
| 2942 | PL_op->op_ppaddr = | |
| 2943 | PL_ppaddr[OP_I_MODULO] = | |
| 2944 | &Perl_pp_i_modulo_1; | |
| 2945 | /* Make certain we work right this time, too. */ | |
| 32fdb065 | 2946 | right = PERL_ABS(right); |
| 224ec323 JH |
2947 | } |
| 2948 | } | |
| a0cec769 YST |
2949 | /* avoid FPE_INTOVF on some platforms when left is IV_MIN */ |
| 2950 | if (right == -1) | |
| 2951 | SETi( 0 ); | |
| 2952 | else | |
| 2953 | SETi( left % right ); | |
| 224ec323 JH |
2954 | RETURN; |
| 2955 | } | |
| 79072805 | 2956 | } |
| befad5d1 | 2957 | #endif |
| 79072805 | 2958 | |
| a0d0e21e | 2959 | PP(pp_i_add) |
| 79072805 | 2960 | { |
| 6f1401dc DM |
2961 | dVAR; dSP; dATARGET; |
| 2962 | tryAMAGICbin_MG(add_amg, AMGf_assign); | |
| a0d0e21e | 2963 | { |
| 6f1401dc | 2964 | dPOPTOPiirl_ul_nomg; |
| a0d0e21e LW |
2965 | SETi( left + right ); |
| 2966 | RETURN; | |
| 79072805 | 2967 | } |
| 79072805 LW |
2968 | } |
| 2969 | ||
| a0d0e21e | 2970 | PP(pp_i_subtract) |
| 79072805 | 2971 | { |
| 6f1401dc DM |
2972 | dVAR; dSP; dATARGET; |
| 2973 | tryAMAGICbin_MG(subtr_amg, AMGf_assign); | |
| a0d0e21e | 2974 | { |
| 6f1401dc | 2975 | dPOPTOPiirl_ul_nomg; |
| a0d0e21e LW |
2976 | SETi( left - right ); |
| 2977 | RETURN; | |
| 79072805 | 2978 | } |
| 79072805 LW |
2979 | } |
| 2980 | ||
| a0d0e21e | 2981 | PP(pp_i_lt) |
| 79072805 | 2982 | { |
| 6f1401dc DM |
2983 | dVAR; dSP; |
| 2984 | tryAMAGICbin_MG(lt_amg, AMGf_set); | |
| a0d0e21e | 2985 | { |
| 6f1401dc | 2986 | dPOPTOPiirl_nomg; |
| 54310121 | 2987 | SETs(boolSV(left < right)); |
| a0d0e21e LW |
2988 | RETURN; |
| 2989 | } | |
| 79072805 LW |
2990 | } |
| 2991 | ||
| a0d0e21e | 2992 | PP(pp_i_gt) |
| 79072805 | 2993 | { |
| 6f1401dc DM |
2994 | dVAR; dSP; |
| 2995 | tryAMAGICbin_MG(gt_amg, AMGf_set); | |
| a0d0e21e | 2996 | { |
| 6f1401dc | 2997 | dPOPTOPiirl_nomg; |
| 54310121 | 2998 | SETs(boolSV(left > right)); |
| a0d0e21e LW |
2999 | RETURN; |
| 3000 | } | |
| 79072805 LW |
3001 | } |
| 3002 | ||
| a0d0e21e | 3003 | PP(pp_i_le) |
| 79072805 | 3004 | { |
| 6f1401dc DM |
3005 | dVAR; dSP; |
| 3006 | tryAMAGICbin_MG(le_amg, AMGf_set); | |
| a0d0e21e | 3007 | { |
| 6f1401dc | 3008 | dPOPTOPiirl_nomg; |
| 54310121 | 3009 | SETs(boolSV(left <= right)); |
| a0d0e21e | 3010 | RETURN; |
| 85e6fe83 | 3011 | } |
| 79072805 LW |
3012 | } |
| 3013 | ||
| a0d0e21e | 3014 | PP(pp_i_ge) |
| 79072805 | 3015 | { |
| 6f1401dc DM |
3016 | dVAR; dSP; |
| 3017 | tryAMAGICbin_MG(ge_amg, AMGf_set); | |
| a0d0e21e | 3018 | { |
| 6f1401dc | 3019 | dPOPTOPiirl_nomg; |
| 54310121 | 3020 | SETs(boolSV(left >= right)); |
| a0d0e21e LW |
3021 | RETURN; |
| 3022 | } | |
| 79072805 LW |
3023 | } |
| 3024 | ||
| a0d0e21e | 3025 | PP(pp_i_eq) |
| 79072805 | 3026 | { |
| 6f1401dc DM |
3027 | dVAR; dSP; |
| 3028 | tryAMAGICbin_MG(eq_amg, AMGf_set); | |
| a0d0e21e | 3029 | { |
| 6f1401dc | 3030 | dPOPTOPiirl_nomg; |
| 54310121 | 3031 | SETs(boolSV(left == right)); |
| a0d0e21e LW |
3032 | RETURN; |
| 3033 | } | |
| 79072805 LW |
3034 | } |
| 3035 | ||
| a0d0e21e | 3036 | PP(pp_i_ne) |
| 79072805 | 3037 | { |
| 6f1401dc DM |
3038 | dVAR; dSP; |
| 3039 | tryAMAGICbin_MG(ne_amg, AMGf_set); | |
| a0d0e21e | 3040 | { |
| 6f1401dc | 3041 | dPOPTOPiirl_nomg; |
| 54310121 | 3042 | SETs(boolSV(left != right)); |
| a0d0e21e LW |
3043 | RETURN; |
| 3044 | } | |
| 79072805 LW |
3045 | } |
| 3046 | ||
| a0d0e21e | 3047 | PP(pp_i_ncmp) |
| 79072805 | 3048 | { |
| 6f1401dc DM |
3049 | dVAR; dSP; dTARGET; |
| 3050 | tryAMAGICbin_MG(ncmp_amg, 0); | |
| a0d0e21e | 3051 | { |
| 6f1401dc | 3052 | dPOPTOPiirl_nomg; |
| a0d0e21e | 3053 | I32 value; |
| 79072805 | 3054 | |
| a0d0e21e | 3055 | if (left > right) |
| 79072805 | 3056 | value = 1; |
| a0d0e21e | 3057 | else if (left < right) |
| 79072805 | 3058 | value = -1; |
| a0d0e21e | 3059 | else |
| 79072805 | 3060 | value = 0; |
| a0d0e21e LW |
3061 | SETi(value); |
| 3062 | RETURN; | |
| 79072805 | 3063 | } |
| 85e6fe83 LW |
3064 | } |
| 3065 | ||
| 3066 | PP(pp_i_negate) | |
| 3067 | { | |
| 6f1401dc DM |
3068 | dVAR; dSP; dTARGET; |
| 3069 | tryAMAGICun_MG(neg_amg, 0); | |
| 3070 | { | |
| 3071 | SV * const sv = TOPs; | |
| 3072 | IV const i = SvIV_nomg(sv); | |
| 3073 | SETi(-i); | |
| 3074 | RETURN; | |
| 3075 | } | |
| 85e6fe83 LW |
3076 | } |
| 3077 | ||
| 79072805 LW |
3078 | /* High falutin' math. */ |
| 3079 | ||
| 3080 | PP(pp_atan2) | |
| 3081 | { | |
| 6f1401dc DM |
3082 | dVAR; dSP; dTARGET; |
| 3083 | tryAMAGICbin_MG(atan2_amg, 0); | |
| a0d0e21e | 3084 | { |
| 6f1401dc | 3085 | dPOPTOPnnrl_nomg; |
| a1021d57 | 3086 | SETn(Perl_atan2(left, right)); |
| a0d0e21e LW |
3087 | RETURN; |
| 3088 | } | |
| 79072805 LW |
3089 | } |
| 3090 | ||
| 3091 | PP(pp_sin) | |
| 3092 | { | |
| 71302fe3 NC |
3093 | dVAR; dSP; dTARGET; |
| 3094 | int amg_type = sin_amg; | |
| 3095 | const char *neg_report = NULL; | |
| bc81784a | 3096 | NV (*func)(NV) = Perl_sin; |
| 71302fe3 NC |
3097 | const int op_type = PL_op->op_type; |
| 3098 | ||
| 3099 | switch (op_type) { | |
| 3100 | case OP_COS: | |
| 3101 | amg_type = cos_amg; | |
| bc81784a | 3102 | func = Perl_cos; |
| 71302fe3 NC |
3103 | break; |
| 3104 | case OP_EXP: | |
| 3105 | amg_type = exp_amg; | |
| bc81784a | 3106 | func = Perl_exp; |
| 71302fe3 NC |
3107 | break; |
| 3108 | case OP_LOG: | |
| 3109 | amg_type = log_amg; | |
| bc81784a | 3110 | func = Perl_log; |
| 71302fe3 NC |
3111 | neg_report = "log"; |
| 3112 | break; | |
| 3113 | case OP_SQRT: | |
| 3114 | amg_type = sqrt_amg; | |
| bc81784a | 3115 | func = Perl_sqrt; |
| 71302fe3 NC |
3116 | neg_report = "sqrt"; |
| 3117 | break; | |
| a0d0e21e | 3118 | } |
| 79072805 | 3119 | |
| 6f1401dc DM |
3120 | |
| 3121 | tryAMAGICun_MG(amg_type, 0); | |
| a0d0e21e | 3122 | { |
| 6f1401dc DM |
3123 | SV * const arg = POPs; |
| 3124 | const NV value = SvNV_nomg(arg); | |
| 71302fe3 NC |
3125 | if (neg_report) { |
| 3126 | if (op_type == OP_LOG ? (value <= 0.0) : (value < 0.0)) { | |
| 3127 | SET_NUMERIC_STANDARD(); | |
| 3128 | DIE(aTHX_ "Can't take %s of %"NVgf, neg_report, value); | |
| 3129 | } | |
| 3130 | } | |
| 3131 | XPUSHn(func(value)); | |
| a0d0e21e LW |
3132 | RETURN; |
| 3133 | } | |
| 79072805 LW |
3134 | } |
| 3135 | ||
| 56cb0a1c AD |
3136 | /* Support Configure command-line overrides for rand() functions. |
| 3137 | After 5.005, perhaps we should replace this by Configure support | |
| 3138 | for drand48(), random(), or rand(). For 5.005, though, maintain | |
| 3139 | compatibility by calling rand() but allow the user to override it. | |
| 3140 | See INSTALL for details. --Andy Dougherty 15 July 1998 | |
| 3141 | */ | |
| 85ab1d1d JH |
3142 | /* Now it's after 5.005, and Configure supports drand48() and random(), |
| 3143 | in addition to rand(). So the overrides should not be needed any more. | |
| 3144 | --Jarkko Hietaniemi 27 September 1998 | |
| 3145 | */ | |
| 3146 | ||
| 3147 | #ifndef HAS_DRAND48_PROTO | |
| 20ce7b12 | 3148 | extern double drand48 (void); |
| 56cb0a1c AD |
3149 | #endif |
| 3150 | ||
| 79072805 LW |
3151 | PP(pp_rand) |
| 3152 | { | |
| 97aff369 | 3153 | dVAR; dSP; dTARGET; |
| 65202027 | 3154 | NV value; |
| 79072805 LW |
3155 | if (MAXARG < 1) |
| 3156 | value = 1.0; | |
| 3157 | else | |
| 3158 | value = POPn; | |
| 3159 | if (value == 0.0) | |
| 3160 | value = 1.0; | |
| 80252599 | 3161 | if (!PL_srand_called) { |
| 85ab1d1d | 3162 | (void)seedDrand01((Rand_seed_t)seed()); |
| 80252599 | 3163 | PL_srand_called = TRUE; |
| 93dc8474 | 3164 | } |
| 85ab1d1d | 3165 | value *= Drand01(); |
| 79072805 LW |
3166 | XPUSHn(value); |
| 3167 | RETURN; | |
| 3168 | } | |
| 3169 | ||
| 3170 | PP(pp_srand) | |
| 3171 | { | |
| 83832992 | 3172 | dVAR; dSP; dTARGET; |
| 0bd48802 | 3173 | const UV anum = (MAXARG < 1) ? seed() : POPu; |
| 85ab1d1d | 3174 | (void)seedDrand01((Rand_seed_t)anum); |
| 80252599 | 3175 | PL_srand_called = TRUE; |
| da1010ec NC |
3176 | if (anum) |
| 3177 | XPUSHu(anum); | |
| 3178 | else { | |
| 3179 | /* Historically srand always returned true. We can avoid breaking | |
| 3180 | that like this: */ | |
| 3181 | sv_setpvs(TARG, "0 but true"); | |
| 3182 | XPUSHTARG; | |
| 3183 | } | |
| 83832992 | 3184 | RETURN; |
| 79072805 LW |
3185 | } |
| 3186 | ||
| 79072805 LW |
3187 | PP(pp_int) |
| 3188 | { | |
| 6f1401dc DM |
3189 | dVAR; dSP; dTARGET; |
| 3190 | tryAMAGICun_MG(int_amg, AMGf_numeric); | |
| 774d564b | 3191 | { |
| 6f1401dc DM |
3192 | SV * const sv = TOPs; |
| 3193 | const IV iv = SvIV_nomg(sv); | |
| 28e5dec8 JH |
3194 | /* XXX it's arguable that compiler casting to IV might be subtly |
| 3195 | different from modf (for numbers inside (IV_MIN,UV_MAX)) in which | |
| 3196 | else preferring IV has introduced a subtle behaviour change bug. OTOH | |
| 3197 | relying on floating point to be accurate is a bug. */ | |
| 3198 | ||
| c781a409 | 3199 | if (!SvOK(sv)) { |
| 922c4365 | 3200 | SETu(0); |
| c781a409 RD |
3201 | } |
| 3202 | else if (SvIOK(sv)) { | |
| 3203 | if (SvIsUV(sv)) | |
| 6f1401dc | 3204 | SETu(SvUV_nomg(sv)); |
| c781a409 | 3205 | else |
| 28e5dec8 | 3206 | SETi(iv); |
| c781a409 | 3207 | } |
| c781a409 | 3208 | else { |
| 6f1401dc | 3209 | const NV value = SvNV_nomg(sv); |
| 1048ea30 | 3210 | if (value >= 0.0) { |
| 28e5dec8 JH |
3211 | if (value < (NV)UV_MAX + 0.5) { |
| 3212 | SETu(U_V(value)); | |
| 3213 | } else { | |
| 059a1014 | 3214 | SETn(Perl_floor(value)); |
| 28e5dec8 | 3215 | } |
| 1048ea30 | 3216 | } |
| 28e5dec8 JH |
3217 | else { |
| 3218 | if (value > (NV)IV_MIN - 0.5) { | |
| 3219 | SETi(I_V(value)); | |
| 3220 | } else { | |
| 1bbae031 | 3221 | SETn(Perl_ceil(value)); |
| 28e5dec8 JH |
3222 | } |
| 3223 | } | |
| 774d564b | 3224 | } |
| 79072805 | 3225 | } |
| 79072805 LW |
3226 | RETURN; |
| 3227 | } | |
| 3228 | ||
| 463ee0b2 LW |
3229 | PP(pp_abs) |
| 3230 | { | |
| 6f1401dc DM |
3231 | dVAR; dSP; dTARGET; |
| 3232 | tryAMAGICun_MG(abs_amg, AMGf_numeric); | |
| a0d0e21e | 3233 | { |
| 6f1401dc | 3234 | SV * const sv = TOPs; |
| 28e5dec8 | 3235 | /* This will cache the NV value if string isn't actually integer */ |
| 6f1401dc | 3236 | const IV iv = SvIV_nomg(sv); |
| a227d84d | 3237 | |
| 800401ee | 3238 | if (!SvOK(sv)) { |
| 922c4365 | 3239 | SETu(0); |
| 800401ee JH |
3240 | } |
| 3241 | else if (SvIOK(sv)) { | |
| 28e5dec8 | 3242 | /* IVX is precise */ |
| 800401ee | 3243 | if (SvIsUV(sv)) { |
| 6f1401dc | 3244 | SETu(SvUV_nomg(sv)); /* force it to be numeric only */ |
| 28e5dec8 JH |
3245 | } else { |
| 3246 | if (iv >= 0) { | |
| 3247 | SETi(iv); | |
| 3248 | } else { | |
| 3249 | if (iv != IV_MIN) { | |
| 3250 | SETi(-iv); | |
| 3251 | } else { | |
| 3252 | /* 2s complement assumption. Also, not really needed as | |
| 3253 | IV_MIN and -IV_MIN should both be %100...00 and NV-able */ | |
| 3254 | SETu(IV_MIN); | |
| 3255 | } | |
| a227d84d | 3256 | } |
| 28e5dec8 JH |
3257 | } |
| 3258 | } else{ | |
| 6f1401dc | 3259 | const NV value = SvNV_nomg(sv); |
| 774d564b | 3260 | if (value < 0.0) |
| 1b6737cc | 3261 | SETn(-value); |
| a4474c9e DD |
3262 | else |
| 3263 | SETn(value); | |
| 774d564b | 3264 | } |
| a0d0e21e | 3265 | } |
| 774d564b | 3266 | RETURN; |
| 463ee0b2 LW |
3267 | } |
| 3268 | ||
| 79072805 LW |
3269 | PP(pp_oct) |
| 3270 | { | |
| 97aff369 | 3271 | dVAR; dSP; dTARGET; |
| 5c144d81 | 3272 | const char *tmps; |
| 53305cf1 | 3273 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES; |
| 6f894ead | 3274 | STRLEN len; |
| 53305cf1 NC |
3275 | NV result_nv; |
| 3276 | UV result_uv; | |
| 1b6737cc | 3277 | SV* const sv = POPs; |
| 79072805 | 3278 | |
| 349d4f2f | 3279 | tmps = (SvPV_const(sv, len)); |
| 2bc69dc4 NIS |
3280 | if (DO_UTF8(sv)) { |
| 3281 | /* If Unicode, try to downgrade | |
| 3282 | * If not possible, croak. */ | |
| 1b6737cc | 3283 | SV* const tsv = sv_2mortal(newSVsv(sv)); |
| 2bc69dc4 NIS |
3284 | |
| 3285 | SvUTF8_on(tsv); | |
| 3286 | sv_utf8_downgrade(tsv, FALSE); | |
| 349d4f2f | 3287 | tmps = SvPV_const(tsv, len); |
| 2bc69dc4 | 3288 | } |
| daa2adfd NC |
3289 | if (PL_op->op_type == OP_HEX) |
| 3290 | goto hex; | |
| 3291 | ||
| 6f894ead | 3292 | while (*tmps && len && isSPACE(*tmps)) |
| 53305cf1 | 3293 | tmps++, len--; |
| 9e24b6e2 | 3294 | if (*tmps == '0') |
| 53305cf1 | 3295 | tmps++, len--; |
| a674e8db | 3296 | if (*tmps == 'x' || *tmps == 'X') { |
| daa2adfd | 3297 | hex: |
| 53305cf1 | 3298 | result_uv = grok_hex (tmps, &len, &flags, &result_nv); |
| daa2adfd | 3299 | } |
| a674e8db | 3300 | else if (*tmps == 'b' || *tmps == 'B') |
| 53305cf1 | 3301 | result_uv = grok_bin (tmps, &len, &flags, &result_nv); |
| 464e2e8a | 3302 | else |
| 53305cf1 NC |
3303 | result_uv = grok_oct (tmps, &len, &flags, &result_nv); |
| 3304 | ||
| 3305 | if (flags & PERL_SCAN_GREATER_THAN_UV_MAX) { | |
| 3306 | XPUSHn(result_nv); | |
| 3307 | } | |
| 3308 | else { | |
| 3309 | XPUSHu(result_uv); | |
| 3310 | } | |
| 79072805 LW |
3311 | RETURN; |
| 3312 | } | |
| 3313 | ||
| 3314 | /* String stuff. */ | |
| 3315 | ||
| 3316 | PP(pp_length) | |
| 3317 | { | |
| 97aff369 | 3318 | dVAR; dSP; dTARGET; |
| 0bd48802 | 3319 | SV * const sv = TOPs; |
| a0ed51b3 | 3320 | |
| 656266fc | 3321 | if (SvGAMAGIC(sv)) { |
| 9f621bb0 NC |
3322 | /* For an overloaded or magic scalar, we can't know in advance if |
| 3323 | it's going to be UTF-8 or not. Also, we can't call sv_len_utf8 as | |
| 3324 | it likes to cache the length. Maybe that should be a documented | |
| 3325 | feature of it. | |
| 92331800 NC |
3326 | */ |
| 3327 | STRLEN len; | |
| 9f621bb0 NC |
3328 | const char *const p |
| 3329 | = sv_2pv_flags(sv, &len, | |
| 3330 | SV_UNDEF_RETURNS_NULL|SV_CONST_RETURN|SV_GMAGIC); | |
| 92331800 | 3331 | |
| d88e091f | 3332 | if (!p) { |
| 9407f9c1 DL |
3333 | if (!SvPADTMP(TARG)) { |
| 3334 | sv_setsv(TARG, &PL_sv_undef); | |
| 3335 | SETTARG; | |
| 3336 | } | |
| 3337 | SETs(&PL_sv_undef); | |
| d88e091f | 3338 | } |
| 9f621bb0 | 3339 | else if (DO_UTF8(sv)) { |
| 899be101 | 3340 | SETi(utf8_length((U8*)p, (U8*)p + len)); |
| 92331800 NC |
3341 | } |
| 3342 | else | |
| 3343 | SETi(len); | |
| 656266fc | 3344 | } else if (SvOK(sv)) { |
| 9f621bb0 NC |
3345 | /* Neither magic nor overloaded. */ |
| 3346 | if (DO_UTF8(sv)) | |
| 3347 | SETi(sv_len_utf8(sv)); | |
| 3348 | else | |
| 3349 | SETi(sv_len(sv)); | |
| 656266fc | 3350 | } else { |
| 9407f9c1 DL |
3351 | if (!SvPADTMP(TARG)) { |
| 3352 | sv_setsv_nomg(TARG, &PL_sv_undef); | |
| 3353 | SETTARG; | |
| 3354 | } | |
| 3355 | SETs(&PL_sv_undef); | |
| 92331800 | 3356 | } |
| 79072805 LW |
3357 | RETURN; |
| 3358 | } | |
| 3359 | ||
| 3360 | PP(pp_substr) | |
| 3361 | { | |
| 97aff369 | 3362 | dVAR; dSP; dTARGET; |
| 79072805 | 3363 | SV *sv; |
| 463ee0b2 | 3364 | STRLEN curlen; |
| 9402d6ed | 3365 | STRLEN utf8_curlen; |
| 777f7c56 EB |
3366 | SV * pos_sv; |
| 3367 | IV pos1_iv; | |
| 3368 | int pos1_is_uv; | |
| 3369 | IV pos2_iv; | |
| 3370 | int pos2_is_uv; | |
| 3371 | SV * len_sv; | |
| 3372 | IV len_iv = 0; | |
| 3373 | int len_is_uv = 1; | |
| 050e6362 | 3374 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; |
| e1ec3a88 | 3375 | const char *tmps; |
| 777f7c56 | 3376 | const IV arybase = CopARYBASE_get(PL_curcop); |
| 9402d6ed | 3377 | SV *repl_sv = NULL; |
| cbbf8932 | 3378 | const char *repl = NULL; |
| 7b8d334a | 3379 | STRLEN repl_len; |
| 050e6362 | 3380 | const int num_args = PL_op->op_private & 7; |
| 13e30c65 | 3381 | bool repl_need_utf8_upgrade = FALSE; |
| 9402d6ed | 3382 | bool repl_is_utf8 = FALSE; |
| 79072805 | 3383 | |
| 78f9721b SM |
3384 | if (num_args > 2) { |
| 3385 | if (num_args > 3) { | |
| 9402d6ed | 3386 | repl_sv = POPs; |
| 83003860 | 3387 | repl = SvPV_const(repl_sv, repl_len); |
| 9402d6ed | 3388 | repl_is_utf8 = DO_UTF8(repl_sv) && SvCUR(repl_sv); |
| 7b8d334a | 3389 | } |
| 777f7c56 EB |
3390 | len_sv = POPs; |
| 3391 | len_iv = SvIV(len_sv); | |
| 3392 | len_is_uv = SvIOK_UV(len_sv); | |
| 5d82c453 | 3393 | } |
| 777f7c56 EB |
3394 | pos_sv = POPs; |
| 3395 | pos1_iv = SvIV(pos_sv); | |
| 3396 | pos1_is_uv = SvIOK_UV(pos_sv); | |
| 79072805 | 3397 | sv = POPs; |
| 849ca7ee | 3398 | PUTBACK; |
| 9402d6ed JH |
3399 | if (repl_sv) { |
| 3400 | if (repl_is_utf8) { | |
| 3401 | if (!DO_UTF8(sv)) | |
| 3402 | sv_utf8_upgrade(sv); | |
| 3403 | } | |
| 13e30c65 JH |
3404 | else if (DO_UTF8(sv)) |
| 3405 | repl_need_utf8_upgrade = TRUE; | |
| 9402d6ed | 3406 | } |
| 5c144d81 | 3407 | tmps = SvPV_const(sv, curlen); |
| 7e2040f0 | 3408 | if (DO_UTF8(sv)) { |
| 9402d6ed JH |
3409 | utf8_curlen = sv_len_utf8(sv); |
| 3410 | if (utf8_curlen == curlen) | |
| 3411 | utf8_curlen = 0; | |
| a0ed51b3 | 3412 | else |
| 9402d6ed | 3413 | curlen = utf8_curlen; |
| a0ed51b3 | 3414 | } |
| d1c2b58a | 3415 | else |
| 9402d6ed | 3416 | utf8_curlen = 0; |
| a0ed51b3 | 3417 | |
| 777f7c56 EB |
3418 | if ( (pos1_is_uv && arybase < 0) || (pos1_iv >= arybase) ) { /* pos >= $[ */ |
| 3419 | UV pos1_uv = pos1_iv-arybase; | |
| 3420 | /* Overflow can occur when $[ < 0 */ | |
| 3421 | if (arybase < 0 && pos1_uv < (UV)pos1_iv) | |
| 1c900557 | 3422 | goto bound_fail; |
| 777f7c56 EB |
3423 | pos1_iv = pos1_uv; |
| 3424 | pos1_is_uv = 1; | |
| 3425 | } | |
| 3426 | else if (pos1_is_uv ? (UV)pos1_iv > 0 : pos1_iv > 0) { | |
| 1c900557 | 3427 | goto bound_fail; /* $[=3; substr($_,2,...) */ |
| 777f7c56 EB |
3428 | } |
| 3429 | else { /* pos < $[ */ | |
| 3430 | if (pos1_iv == 0) { /* $[=1; substr($_,0,...) */ | |
| 3431 | pos1_iv = curlen; | |
| 3432 | pos1_is_uv = 1; | |
| 3433 | } else { | |
| 3434 | if (curlen) { | |
| 3435 | pos1_is_uv = curlen-1 > ~(UV)pos1_iv; | |
| 3436 | pos1_iv += curlen; | |
| 3437 | } | |
| 5d82c453 | 3438 | } |
| 68dc0745 | 3439 | } |
| 777f7c56 EB |
3440 | if (pos1_is_uv || pos1_iv > 0) { |
| 3441 | if ((UV)pos1_iv > curlen) | |
| 1c900557 | 3442 | goto bound_fail; |
| 777f7c56 EB |
3443 | } |
| 3444 | ||
| 3445 | if (num_args > 2) { | |
| 3446 | if (!len_is_uv && len_iv < 0) { | |
| 3447 | pos2_iv = curlen + len_iv; | |
| 3448 | if (curlen) | |
| 3449 | pos2_is_uv = curlen-1 > ~(UV)len_iv; | |
| 3450 | else | |
| 3451 | pos2_is_uv = 0; | |
| 3452 | } else { /* len_iv >= 0 */ | |
| 3453 | if (!pos1_is_uv && pos1_iv < 0) { | |
| 3454 | pos2_iv = pos1_iv + len_iv; | |
| 3455 | pos2_is_uv = (UV)len_iv > (UV)IV_MAX; | |
| 3456 | } else { | |
| 3457 | if ((UV)len_iv > curlen-(UV)pos1_iv) | |
| 3458 | pos2_iv = curlen; | |
| 3459 | else | |
| 3460 | pos2_iv = pos1_iv+len_iv; | |
| 3461 | pos2_is_uv = 1; | |
| 3462 | } | |
| 5d82c453 | 3463 | } |
| 2304df62 | 3464 | } |
| 79072805 | 3465 | else { |
| 777f7c56 EB |
3466 | pos2_iv = curlen; |
| 3467 | pos2_is_uv = 1; | |
| 3468 | } | |
| 3469 | ||
| 3470 | if (!pos2_is_uv && pos2_iv < 0) { | |
| 3471 | if (!pos1_is_uv && pos1_iv < 0) | |
| 1c900557 | 3472 | goto bound_fail; |
| 777f7c56 EB |
3473 | pos2_iv = 0; |
| 3474 | } | |
| 3475 | else if (!pos1_is_uv && pos1_iv < 0) | |
| 3476 | pos1_iv = 0; | |
| 3477 | ||
| 3478 | if ((UV)pos2_iv < (UV)pos1_iv) | |
| 3479 | pos2_iv = pos1_iv; | |
| 3480 | if ((UV)pos2_iv > curlen) | |
| 3481 | pos2_iv = curlen; | |
| 3482 | ||
| 3483 | { | |
| 3484 | /* pos1_iv and pos2_iv both in 0..curlen, so the cast is safe */ | |
| 3485 | const STRLEN pos = (STRLEN)( (UV)pos1_iv ); | |
| 3486 | const STRLEN len = (STRLEN)( (UV)pos2_iv - (UV)pos1_iv ); | |
| 777f7c56 | 3487 | STRLEN byte_len = len; |
| d931b1be NC |
3488 | STRLEN byte_pos = utf8_curlen |
| 3489 | ? sv_pos_u2b_flags(sv, pos, &byte_len, SV_CONST_RETURN) : pos; | |
| 3490 | ||
| 2154eca7 EB |
3491 | if (lvalue && !repl) { |
| 3492 | SV * ret; | |
| 3493 | ||
| 3494 | if (!SvGMAGICAL(sv)) { | |
| 3495 | if (SvROK(sv)) { | |
| 3496 | SvPV_force_nolen(sv); | |
| 3497 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), | |
| 3498 | "Attempt to use reference as lvalue in substr"); | |
| 3499 | } | |
| 3500 | if (isGV_with_GP(sv)) | |
| 3501 | SvPV_force_nolen(sv); | |
| 3502 | else if (SvOK(sv)) /* is it defined ? */ | |
| 3503 | (void)SvPOK_only_UTF8(sv); | |
| 3504 | else | |
| 3505 | sv_setpvs(sv, ""); /* avoid lexical reincarnation */ | |
| 781e7547 | 3506 | } |
| 2154eca7 EB |
3507 | |
| 3508 | ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ | |
| 3509 | sv_magic(ret, NULL, PERL_MAGIC_substr, NULL, 0); | |
| 3510 | LvTYPE(ret) = 'x'; | |
| 3511 | LvTARG(ret) = SvREFCNT_inc_simple(sv); | |
| 3512 | LvTARGOFF(ret) = pos; | |
| 3513 | LvTARGLEN(ret) = len; | |
| 3514 | ||
| 3515 | SPAGAIN; | |
| 3516 | PUSHs(ret); /* avoid SvSETMAGIC here */ | |
| 3517 | RETURN; | |
| 781e7547 DM |
3518 | } |
| 3519 | ||
| 2154eca7 EB |
3520 | SvTAINTED_off(TARG); /* decontaminate */ |
| 3521 | SvUTF8_off(TARG); /* decontaminate */ | |
| 3522 | ||
| 3523 | tmps += byte_pos; | |
| 777f7c56 | 3524 | sv_setpvn(TARG, tmps, byte_len); |
| 12aa1545 | 3525 | #ifdef USE_LOCALE_COLLATE |
| 14befaf4 | 3526 | sv_unmagic(TARG, PERL_MAGIC_collxfrm); |
| 12aa1545 | 3527 | #endif |
| 9402d6ed | 3528 | if (utf8_curlen) |
| 7f66633b | 3529 | SvUTF8_on(TARG); |
| 2154eca7 | 3530 | |
| f7928d6c | 3531 | if (repl) { |
| 13e30c65 JH |
3532 | SV* repl_sv_copy = NULL; |
| 3533 | ||
| 3534 | if (repl_need_utf8_upgrade) { | |
| 3535 | repl_sv_copy = newSVsv(repl_sv); | |
| 3536 | sv_utf8_upgrade(repl_sv_copy); | |
| 349d4f2f | 3537 | repl = SvPV_const(repl_sv_copy, repl_len); |
| 13e30c65 JH |
3538 | repl_is_utf8 = DO_UTF8(repl_sv_copy) && SvCUR(sv); |
| 3539 | } | |
| 502d9230 VP |
3540 | if (!SvOK(sv)) |
| 3541 | sv_setpvs(sv, ""); | |
| 777f7c56 | 3542 | sv_insert_flags(sv, byte_pos, byte_len, repl, repl_len, 0); |
| 9402d6ed | 3543 | if (repl_is_utf8) |
| f7928d6c | 3544 | SvUTF8_on(sv); |
| ef8d46e8 | 3545 | SvREFCNT_dec(repl_sv_copy); |
| f7928d6c | 3546 | } |
| 79072805 | 3547 | } |
| 849ca7ee | 3548 | SPAGAIN; |
| e27c778f FC |
3549 | SvSETMAGIC(TARG); |
| 3550 | PUSHs(TARG); | |
| 79072805 | 3551 | RETURN; |
| 777f7c56 | 3552 | |
| 1c900557 | 3553 | bound_fail: |
| 777f7c56 EB |
3554 | if (lvalue || repl) |
| 3555 | Perl_croak(aTHX_ "substr outside of string"); | |
| 3556 | Perl_ck_warner(aTHX_ packWARN(WARN_SUBSTR), "substr outside of string"); | |
| 3557 | RETPUSHUNDEF; | |
| 79072805 LW |
3558 | } |
| 3559 | ||
| 3560 | PP(pp_vec) | |
| 3561 | { | |
| 2154eca7 | 3562 | dVAR; dSP; |
| 1b6737cc AL |
3563 | register const IV size = POPi; |
| 3564 | register const IV offset = POPi; | |
| 3565 | register SV * const src = POPs; | |
| 3566 | const I32 lvalue = PL_op->op_flags & OPf_MOD || LVRET; | |
| 2154eca7 | 3567 | SV * ret; |
| a0d0e21e | 3568 | |
| 81e118e0 | 3569 | if (lvalue) { /* it's an lvalue! */ |
| 2154eca7 EB |
3570 | ret = sv_2mortal(newSV_type(SVt_PVLV)); /* Not TARG RT#67838 */ |
| 3571 | sv_magic(ret, NULL, PERL_MAGIC_vec, NULL, 0); | |
| 3572 | LvTYPE(ret) = 'v'; | |
| 3573 | LvTARG(ret) = SvREFCNT_inc_simple(src); | |
| 3574 | LvTARGOFF(ret) = offset; | |
| 3575 | LvTARGLEN(ret) = size; | |
| 3576 | } | |
| 3577 | else { | |
| 3578 | dTARGET; | |
| 3579 | SvTAINTED_off(TARG); /* decontaminate */ | |
| 3580 | ret = TARG; | |
| 79072805 LW |
3581 | } |
| 3582 | ||
| 2154eca7 EB |
3583 | sv_setuv(ret, do_vecget(src, offset, size)); |
| 3584 | PUSHs(ret); | |
| 79072805 LW |
3585 | RETURN; |
| 3586 | } | |
| 3587 | ||
| 3588 | PP(pp_index) | |
| 3589 | { | |
| 97aff369 | 3590 | dVAR; dSP; dTARGET; |
| 79072805 LW |
3591 | SV *big; |
| 3592 | SV *little; | |
| c445ea15 | 3593 | SV *temp = NULL; |
| ad66a58c | 3594 | STRLEN biglen; |
| 2723d216 | 3595 | STRLEN llen = 0; |
| 79072805 LW |
3596 | I32 offset; |
| 3597 | I32 retval; | |
| 73ee8be2 NC |
3598 | const char *big_p; |
| 3599 | const char *little_p; | |
| fc15ae8f | 3600 | const I32 arybase = CopARYBASE_get(PL_curcop); |
| 2f040f7f NC |
3601 | bool big_utf8; |
| 3602 | bool little_utf8; | |
| 2723d216 | 3603 | const bool is_index = PL_op->op_type == OP_INDEX; |
| 79072805 | 3604 | |
| 2723d216 NC |
3605 | if (MAXARG >= 3) { |
| 3606 | /* arybase is in characters, like offset, so combine prior to the | |
| 3607 | UTF-8 to bytes calculation. */ | |
| 79072805 | 3608 | offset = POPi - arybase; |
| 2723d216 | 3609 | } |
| 79072805 LW |
3610 | little = POPs; |
| 3611 | big = POPs; | |
| 73ee8be2 NC |
3612 | big_p = SvPV_const(big, biglen); |
| 3613 | little_p = SvPV_const(little, llen); | |
| 3614 | ||
| e609e586 NC |
3615 | big_utf8 = DO_UTF8(big); |
| 3616 | little_utf8 = DO_UTF8(little); | |
| 3617 | if (big_utf8 ^ little_utf8) { | |
| 3618 | /* One needs to be upgraded. */ | |
| 2f040f7f NC |
3619 | if (little_utf8 && !PL_encoding) { |
| 3620 | /* Well, maybe instead we might be able to downgrade the small | |
| 3621 | string? */ | |
| 1eced8f8 | 3622 | char * const pv = (char*)bytes_from_utf8((U8 *)little_p, &llen, |
| 2f040f7f NC |
3623 | &little_utf8); |
| 3624 | if (little_utf8) { | |
| 3625 | /* If the large string is ISO-8859-1, and it's not possible to | |
| 3626 | convert the small string to ISO-8859-1, then there is no | |
| 3627 | way that it could be found anywhere by index. */ | |
| 3628 | retval = -1; | |
| 3629 | goto fail; | |
| 3630 | } | |
| e609e586 | 3631 | |
| 2f040f7f NC |
3632 | /* At this point, pv is a malloc()ed string. So donate it to temp |
| 3633 | to ensure it will get free()d */ | |
| 3634 | little = temp = newSV(0); | |
| 73ee8be2 NC |
3635 | sv_usepvn(temp, pv, llen); |
| 3636 | little_p = SvPVX(little); | |
| e609e586 | 3637 | } else { |
| 73ee8be2 NC |
3638 | temp = little_utf8 |
| 3639 | ? newSVpvn(big_p, biglen) : newSVpvn(little_p, llen); | |
| 2f040f7f NC |
3640 | |
| 3641 | if (PL_encoding) { | |
| 3642 | sv_recode_to_utf8(temp, PL_encoding); | |
| 3643 | } else { | |
| 3644 | sv_utf8_upgrade(temp); | |
| 3645 | } | |
| 3646 | if (little_utf8) { | |
| 3647 | big = temp; | |
| 3648 | big_utf8 = TRUE; | |
| 73ee8be2 | 3649 | big_p = SvPV_const(big, biglen); |
| 2f040f7f NC |
3650 | } else { |
| 3651 | little = temp; | |
| 73ee8be2 | 3652 | little_p = SvPV_const(little, llen); |
| 2f040f7f | 3653 | } |
| e609e586 NC |
3654 | } |
| 3655 | } | |
| 73ee8be2 NC |
3656 | if (SvGAMAGIC(big)) { |
| 3657 | /* Life just becomes a lot easier if I use a temporary here. | |
| 3658 | Otherwise I need to avoid calls to sv_pos_u2b(), which (dangerously) | |
| 3659 | will trigger magic and overloading again, as will fbm_instr() | |
| 3660 | */ | |
| 59cd0e26 NC |
3661 | big = newSVpvn_flags(big_p, biglen, |
| 3662 | SVs_TEMP | (big_utf8 ? SVf_UTF8 : 0)); | |
| 73ee8be2 NC |
3663 | big_p = SvPVX(big); |
| 3664 | } | |
| e4e44778 | 3665 | if (SvGAMAGIC(little) || (is_index && !SvOK(little))) { |
| 73ee8be2 NC |
3666 | /* index && SvOK() is a hack. fbm_instr() calls SvPV_const, which will |
| 3667 | warn on undef, and we've already triggered a warning with the | |
| 3668 | SvPV_const some lines above. We can't remove that, as we need to | |
| 3669 | call some SvPV to trigger overloading early and find out if the | |
| 3670 | string is UTF-8. | |
| 3671 | This is all getting to messy. The API isn't quite clean enough, | |
| 3672 | because data access has side effects. | |
| 3673 | */ | |
| 59cd0e26 NC |
3674 | little = newSVpvn_flags(little_p, llen, |
| 3675 | SVs_TEMP | (little_utf8 ? SVf_UTF8 : 0)); | |
| 73ee8be2 NC |
3676 | little_p = SvPVX(little); |
| 3677 | } | |
| e609e586 | 3678 | |
| 79072805 | 3679 | if (MAXARG < 3) |
| 2723d216 | 3680 | offset = is_index ? 0 : biglen; |
| a0ed51b3 | 3681 | else { |
| ad66a58c | 3682 | if (big_utf8 && offset > 0) |
| a0ed51b3 | 3683 | sv_pos_u2b(big, &offset, 0); |
| 73ee8be2 NC |
3684 | if (!is_index) |
| 3685 | offset += llen; | |
| a0ed51b3 | 3686 | } |
| 79072805 LW |
3687 | if (offset < 0) |
| 3688 | offset = 0; | |
| ad66a58c NC |
3689 | else if (offset > (I32)biglen) |
| 3690 | offset = biglen; | |
| 73ee8be2 NC |
3691 | if (!(little_p = is_index |
| 3692 | ? fbm_instr((unsigned char*)big_p + offset, | |
| 3693 | (unsigned char*)big_p + biglen, little, 0) | |
| 3694 | : rninstr(big_p, big_p + offset, | |
| 3695 | little_p, little_p + llen))) | |
| a0ed51b3 | 3696 | retval = -1; |
| ad66a58c | 3697 | else { |
| 73ee8be2 | 3698 | retval = little_p - big_p; |
| ad66a58c NC |
3699 | if (retval > 0 && big_utf8) |
| 3700 | sv_pos_b2u(big, &retval); | |
| 3701 | } | |
| ef8d46e8 | 3702 | SvREFCNT_dec(temp); |
| 2723d216 | 3703 | fail: |
| a0ed51b3 | 3704 | PUSHi(retval + arybase); |
| 79072805 LW |
3705 | RETURN; |
| 3706 | } | |
| 3707 | ||
| 3708 | PP(pp_sprintf) | |
| 3709 | { | |
| 97aff369 | 3710 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
| 3e6bd4bf | 3711 | SvTAINTED_off(TARG); |
| 79072805 | 3712 | do_sprintf(TARG, SP-MARK, MARK+1); |
| bbce6d69 | 3713 | TAINT_IF(SvTAINTED(TARG)); |
| 79072805 LW |
3714 | SP = ORIGMARK; |
| 3715 | PUSHTARG; | |
| 3716 | RETURN; | |
| 3717 | } | |
| 3718 | ||
| 79072805 LW |
3719 | PP(pp_ord) |
| 3720 | { | |
| 97aff369 | 3721 | dVAR; dSP; dTARGET; |
| 1eced8f8 | 3722 | |
| 7df053ec | 3723 | SV *argsv = POPs; |
| ba210ebe | 3724 | STRLEN len; |
| 349d4f2f | 3725 | const U8 *s = (U8*)SvPV_const(argsv, len); |
| 121910a4 | 3726 | |
| 799ef3cb | 3727 | if (PL_encoding && SvPOK(argsv) && !DO_UTF8(argsv)) { |
| 1eced8f8 | 3728 | SV * const tmpsv = sv_2mortal(newSVsv(argsv)); |
| 799ef3cb | 3729 | s = (U8*)sv_recode_to_utf8(tmpsv, PL_encoding); |
| 121910a4 JH |
3730 | argsv = tmpsv; |
| 3731 | } | |
| 79072805 | 3732 | |
| 872c91ae | 3733 | XPUSHu(DO_UTF8(argsv) ? |
| 89ebb4a3 | 3734 | utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, UTF8_ALLOW_ANYUV) : |
| 5fc32dea | 3735 | (UV)(*s & 0xff)); |
| 68795e93 | 3736 | |
| 79072805 LW |
3737 | RETURN; |
| 3738 | } | |
| 3739 | ||
| 463ee0b2 LW |
3740 | PP(pp_chr) |
| 3741 | { | |
| 97aff369 | 3742 | dVAR; dSP; dTARGET; |
| 463ee0b2 | 3743 | char *tmps; |
| 8a064bd6 JH |
3744 | UV value; |
| 3745 | ||
| 3746 | if (((SvIOK_notUV(TOPs) && SvIV(TOPs) < 0) | |
| 3747 | || | |
| 3748 | (SvNOK(TOPs) && SvNV(TOPs) < 0.0))) { | |
| 3749 | if (IN_BYTES) { | |
| 3750 | value = POPu; /* chr(-1) eq chr(0xff), etc. */ | |
| 3751 | } else { | |
| 3752 | (void) POPs; /* Ignore the argument value. */ | |
| 3753 | value = UNICODE_REPLACEMENT; | |
| 3754 | } | |
| 3755 | } else { | |
| 3756 | value = POPu; | |
| 3757 | } | |
| 463ee0b2 | 3758 | |
| 862a34c6 | 3759 | SvUPGRADE(TARG,SVt_PV); |
| a0ed51b3 | 3760 | |
| 0064a8a9 | 3761 | if (value > 255 && !IN_BYTES) { |
| eb160463 | 3762 | SvGROW(TARG, (STRLEN)UNISKIP(value)+1); |
| 62961d2e | 3763 | tmps = (char*)uvchr_to_utf8_flags((U8*)SvPVX(TARG), value, 0); |
| 349d4f2f | 3764 | SvCUR_set(TARG, tmps - SvPVX_const(TARG)); |
| a0ed51b3 LW |
3765 | *tmps = '\0'; |
| 3766 | (void)SvPOK_only(TARG); | |
| aa6ffa16 | 3767 | SvUTF8_on(TARG); |
| a0ed51b3 LW |
3768 | XPUSHs(TARG); |
| 3769 | RETURN; | |
| 3770 | } | |
| 3771 | ||
| 748a9306 | 3772 | SvGROW(TARG,2); |
| 463ee0b2 LW |
3773 | SvCUR_set(TARG, 1); |
| 3774 | tmps = SvPVX(TARG); | |
| eb160463 | 3775 | *tmps++ = (char)value; |
| 748a9306 | 3776 | *tmps = '\0'; |
| a0d0e21e | 3777 | (void)SvPOK_only(TARG); |
| 4c5ed6e2 | 3778 | |
| 88632417 | 3779 | if (PL_encoding && !IN_BYTES) { |
| 799ef3cb | 3780 | sv_recode_to_utf8(TARG, PL_encoding); |
| 88632417 JH |
3781 | tmps = SvPVX(TARG); |
| 3782 | if (SvCUR(TARG) == 0 || !is_utf8_string((U8*)tmps, SvCUR(TARG)) || | |
| 4c5ed6e2 TS |
3783 | UNICODE_IS_REPLACEMENT(utf8_to_uvchr((U8*)tmps, NULL))) { |
| 3784 | SvGROW(TARG, 2); | |
| d5a15ac2 | 3785 | tmps = SvPVX(TARG); |
| 4c5ed6e2 TS |
3786 | SvCUR_set(TARG, 1); |
| 3787 | *tmps++ = (char)value; | |
| 88632417 | 3788 | *tmps = '\0'; |
| 4c5ed6e2 | 3789 | SvUTF8_off(TARG); |
| 88632417 JH |
3790 | } |
| 3791 | } | |
| 4c5ed6e2 | 3792 | |
| 463ee0b2 LW |
3793 | XPUSHs(TARG); |
| 3794 | RETURN; | |
| 3795 | } | |
| 3796 | ||
| 79072805 LW |
3797 | PP(pp_crypt) |
| 3798 | { | |
| 79072805 | 3799 | #ifdef HAS_CRYPT |
| 97aff369 | 3800 | dVAR; dSP; dTARGET; |
| 5f74f29c | 3801 | dPOPTOPssrl; |
| 85c16d83 | 3802 | STRLEN len; |
| 10516c54 | 3803 | const char *tmps = SvPV_const(left, len); |
| 2bc69dc4 | 3804 | |
| 85c16d83 | 3805 | if (DO_UTF8(left)) { |
| 2bc69dc4 | 3806 | /* If Unicode, try to downgrade. |
| f2791508 JH |
3807 | * If not possible, croak. |
| 3808 | * Yes, we made this up. */ | |
| 1b6737cc | 3809 | SV* const tsv = sv_2mortal(newSVsv(left)); |
| 2bc69dc4 | 3810 | |
| f2791508 | 3811 | SvUTF8_on(tsv); |
| 2bc69dc4 | 3812 | sv_utf8_downgrade(tsv, FALSE); |
| 349d4f2f | 3813 | tmps = SvPV_const(tsv, len); |
| 85c16d83 | 3814 | } |
| 05404ffe JH |
3815 | # ifdef USE_ITHREADS |
| 3816 | # ifdef HAS_CRYPT_R | |
| 3817 | if (!PL_reentrant_buffer->_crypt_struct_buffer) { | |
| 3818 | /* This should be threadsafe because in ithreads there is only | |
| 3819 | * one thread per interpreter. If this would not be true, | |
| 3820 | * we would need a mutex to protect this malloc. */ | |
| 3821 | PL_reentrant_buffer->_crypt_struct_buffer = | |
| 3822 | (struct crypt_data *)safemalloc(sizeof(struct crypt_data)); | |
| 3823 | #if defined(__GLIBC__) || defined(__EMX__) | |
| 3824 | if (PL_reentrant_buffer->_crypt_struct_buffer) { | |
| 3825 | PL_reentrant_buffer->_crypt_struct_buffer->initialized = 0; | |
| 3826 | /* work around glibc-2.2.5 bug */ | |
| 3827 | PL_reentrant_buffer->_crypt_struct_buffer->current_saltbits = 0; | |
| 3828 | } | |
| 05404ffe | 3829 | #endif |
| 6ab58e4d | 3830 | } |
| 05404ffe JH |
3831 | # endif /* HAS_CRYPT_R */ |
| 3832 | # endif /* USE_ITHREADS */ | |
| 5f74f29c | 3833 | # ifdef FCRYPT |
| 83003860 | 3834 | sv_setpv(TARG, fcrypt(tmps, SvPV_nolen_const(right))); |
| 5f74f29c | 3835 | # else |
| 83003860 | 3836 | sv_setpv(TARG, PerlProc_crypt(tmps, SvPV_nolen_const(right))); |
| 5f74f29c | 3837 | # endif |
| ec93b65f | 3838 | SETTARG; |
| 4808266b | 3839 | RETURN; |
| 79072805 | 3840 | #else |
| b13b2135 | 3841 | DIE(aTHX_ |
| 79072805 LW |
3842 | "The crypt() function is unimplemented due to excessive paranoia."); |
| 3843 | #endif | |
| 79072805 LW |
3844 | } |
| 3845 | ||
| 00f254e2 KW |
3846 | /* Generally UTF-8 and UTF-EBCDIC are indistinguishable at this level. So |
| 3847 | * most comments below say UTF-8, when in fact they mean UTF-EBCDIC as well */ | |
| 3848 | ||
| 00f254e2 KW |
3849 | /* Below are several macros that generate code */ |
| 3850 | /* Generates code to store a unicode codepoint c that is known to occupy | |
| 3851 | * exactly two UTF-8 and UTF-EBCDIC bytes; it is stored into p and p+1. */ | |
| 3852 | #define STORE_UNI_TO_UTF8_TWO_BYTE(p, c) \ | |
| 3853 | STMT_START { \ | |
| 3854 | *(p) = UTF8_TWO_BYTE_HI(c); \ | |
| 3855 | *((p)+1) = UTF8_TWO_BYTE_LO(c); \ | |
| 3856 | } STMT_END | |
| 3857 | ||
| 3858 | /* Like STORE_UNI_TO_UTF8_TWO_BYTE, but advances p to point to the next | |
| 3859 | * available byte after the two bytes */ | |
| 3860 | #define CAT_UNI_TO_UTF8_TWO_BYTE(p, c) \ | |
| 3861 | STMT_START { \ | |
| 3862 | *(p)++ = UTF8_TWO_BYTE_HI(c); \ | |
| 3863 | *((p)++) = UTF8_TWO_BYTE_LO(c); \ | |
| 3864 | } STMT_END | |
| 3865 | ||
| 3866 | /* Generates code to store the upper case of latin1 character l which is known | |
| 3867 | * to have its upper case be non-latin1 into the two bytes p and p+1. There | |
| 3868 | * are only two characters that fit this description, and this macro knows | |
| 3869 | * about them, and that the upper case values fit into two UTF-8 or UTF-EBCDIC | |
| 3870 | * bytes */ | |
| 3871 | #define STORE_NON_LATIN1_UC(p, l) \ | |
| 3872 | STMT_START { \ | |
| 3873 | if ((l) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
| 3874 | STORE_UNI_TO_UTF8_TWO_BYTE((p), LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS); \ | |
| 3875 | } else { /* Must be the following letter */ \ | |
| 3876 | STORE_UNI_TO_UTF8_TWO_BYTE((p), GREEK_CAPITAL_LETTER_MU); \ | |
| 3877 | } \ | |
| 3878 | } STMT_END | |
| 3879 | ||
| 3880 | /* Like STORE_NON_LATIN1_UC, but advances p to point to the next available byte | |
| 3881 | * after the character stored */ | |
| 3882 | #define CAT_NON_LATIN1_UC(p, l) \ | |
| 3883 | STMT_START { \ | |
| 3884 | if ((l) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
| 3885 | CAT_UNI_TO_UTF8_TWO_BYTE((p), LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS); \ | |
| 3886 | } else { \ | |
| 3887 | CAT_UNI_TO_UTF8_TWO_BYTE((p), GREEK_CAPITAL_LETTER_MU); \ | |
| 3888 | } \ | |
| 3889 | } STMT_END | |
| 3890 | ||
| 3891 | /* Generates code to add the two UTF-8 bytes (probably u) that are the upper | |
| 3892 | * case of l into p and p+1. u must be the result of toUPPER_LATIN1_MOD(l), | |
| 3893 | * and must require two bytes to store it. Advances p to point to the next | |
| 3894 | * available position */ | |
| 3895 | #define CAT_TWO_BYTE_UNI_UPPER_MOD(p, l, u) \ | |
| 3896 | STMT_START { \ | |
| 3897 | if ((u) != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { \ | |
| 3898 | CAT_UNI_TO_UTF8_TWO_BYTE((p), (u)); /* not special, just save it */ \ | |
| 3899 | } else if (l == LATIN_SMALL_LETTER_SHARP_S) { \ | |
| 3900 | *(p)++ = 'S'; *(p)++ = 'S'; /* upper case is 'SS' */ \ | |
| 3901 | } else {/* else is one of the other two special cases */ \ | |
| 3902 | CAT_NON_LATIN1_UC((p), (l)); \ | |
| 3903 | } \ | |
| 3904 | } STMT_END | |
| 3905 | ||
| 79072805 LW |
3906 | PP(pp_ucfirst) |
| 3907 | { | |
| 00f254e2 KW |
3908 | /* Actually is both lcfirst() and ucfirst(). Only the first character |
| 3909 | * changes. This means that possibly we can change in-place, ie., just | |
| 3910 | * take the source and change that one character and store it back, but not | |
| 3911 | * if read-only etc, or if the length changes */ | |
| 3912 | ||
| 97aff369 | 3913 | dVAR; |
| 39644a26 | 3914 | dSP; |
| d54190f6 | 3915 | SV *source = TOPs; |
| 00f254e2 | 3916 | STRLEN slen; /* slen is the byte length of the whole SV. */ |
| d54190f6 NC |
3917 | STRLEN need; |
| 3918 | SV *dest; | |
| 00f254e2 KW |
3919 | bool inplace; /* ? Convert first char only, in-place */ |
| 3920 | bool doing_utf8 = FALSE; /* ? using utf8 */ | |
| 3921 | bool convert_source_to_utf8 = FALSE; /* ? need to convert */ | |
| 12e9c124 | 3922 | const int op_type = PL_op->op_type; |
| d54190f6 NC |
3923 | const U8 *s; |
| 3924 | U8 *d; | |
| 3925 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
| 00f254e2 KW |
3926 | STRLEN ulen; /* ulen is the byte length of the original Unicode character |
| 3927 | * stored as UTF-8 at s. */ | |
| 3928 | STRLEN tculen; /* tculen is the byte length of the freshly titlecased (or | |
| 3929 | * lowercased) character stored in tmpbuf. May be either | |
| 3930 | * UTF-8 or not, but in either case is the number of bytes */ | |
| d54190f6 NC |
3931 | |
| 3932 | SvGETMAGIC(source); | |
| 3933 | if (SvOK(source)) { | |
| 3934 | s = (const U8*)SvPV_nomg_const(source, slen); | |
| 3935 | } else { | |
| 0a0ffbce RGS |
3936 | if (ckWARN(WARN_UNINITIALIZED)) |
| 3937 | report_uninit(source); | |
| 1eced8f8 | 3938 | s = (const U8*)""; |
| d54190f6 NC |
3939 | slen = 0; |
| 3940 | } | |
| a0ed51b3 | 3941 | |
| 00f254e2 KW |
3942 | /* We may be able to get away with changing only the first character, in |
| 3943 | * place, but not if read-only, etc. Later we may discover more reasons to | |
| 3944 | * not convert in-place. */ | |
| 3945 | inplace = SvPADTMP(source) && !SvREADONLY(source) && SvTEMP(source); | |
| 3946 | ||
| 3947 | /* First calculate what the changed first character should be. This affects | |
| 3948 | * whether we can just swap it out, leaving the rest of the string unchanged, | |
| 3949 | * or even if have to convert the dest to UTF-8 when the source isn't */ | |
| 3950 | ||
| 3951 | if (! slen) { /* If empty */ | |
| 3952 | need = 1; /* still need a trailing NUL */ | |
| 3953 | } | |
| 3954 | else if (DO_UTF8(source)) { /* Is the source utf8? */ | |
| d54190f6 | 3955 | doing_utf8 = TRUE; |
| 00f254e2 KW |
3956 | |
| 3957 | /* TODO: This is #ifdefd out because it has hard-coded the standard mappings, | |
| 3958 | * and doesn't allow for the user to specify their own. When code is added to | |
| 3959 | * detect if there is a user-defined mapping in force here, and if so to use | |
| 3960 | * that, then the code below can be compiled. The detection would be a good | |
| 3961 | * thing anyway, as currently the user-defined mappings only work on utf8 | |
| 3962 | * strings, and thus depend on the chosen internal storage method, which is a | |
| 3963 | * bad thing */ | |
| 3964 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
| 3965 | if (UTF8_IS_INVARIANT(*s)) { | |
| 3966 | ||
| 3967 | /* An invariant source character is either ASCII or, in EBCDIC, an | |
| 3968 | * ASCII equivalent or a caseless C1 control. In both these cases, | |
| 3969 | * the lower and upper cases of any character are also invariants | |
| 3970 | * (and title case is the same as upper case). So it is safe to | |
| 3971 | * use the simple case change macros which avoid the overhead of | |
| 3972 | * the general functions. Note that if perl were to be extended to | |
| 3973 | * do locale handling in UTF-8 strings, this wouldn't be true in, | |
| 3974 | * for example, Lithuanian or Turkic. */ | |
| 3975 | *tmpbuf = (op_type == OP_LCFIRST) ? toLOWER(*s) : toUPPER(*s); | |
| 3976 | tculen = ulen = 1; | |
| 3977 | need = slen + 1; | |
| 12e9c124 | 3978 | } |
| 00f254e2 KW |
3979 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { |
| 3980 | U8 chr; | |
| 3981 | ||
| 3982 | /* Similarly, if the source character isn't invariant but is in the | |
| 3983 | * latin1 range (or EBCDIC equivalent thereof), we have the case | |
| 3984 | * changes compiled into perl, and can avoid the overhead of the | |
| 3985 | * general functions. In this range, the characters are stored as | |
| 3986 | * two UTF-8 bytes, and it so happens that any changed-case version | |
| 3987 | * is also two bytes (in both ASCIIish and EBCDIC machines). */ | |
| 3988 | tculen = ulen = 2; | |
| 3989 | need = slen + 1; | |
| 3990 | ||
| 3991 | /* Convert the two source bytes to a single Unicode code point | |
| 3992 | * value, change case and save for below */ | |
| 356979f4 | 3993 | chr = TWO_BYTE_UTF8_TO_UNI(*s, *(s+1)); |
| 00f254e2 KW |
3994 | if (op_type == OP_LCFIRST) { /* lower casing is easy */ |
| 3995 | U8 lower = toLOWER_LATIN1(chr); | |
| 3996 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, lower); | |
| 3997 | } | |
| 3998 | else { /* ucfirst */ | |
| 3999 | U8 upper = toUPPER_LATIN1_MOD(chr); | |
| 4000 | ||
| 4001 | /* Most of the latin1 range characters are well-behaved. Their | |
| 4002 | * title and upper cases are the same, and are also in the | |
| 4003 | * latin1 range. The macro above returns their upper (hence | |
| 4004 | * title) case, and all that need be done is to save the result | |
| 4005 | * for below. However, several characters are problematic, and | |
| 4006 | * have to be handled specially. The MOD in the macro name | |
| 4007 | * above means that these tricky characters all get mapped to | |
| 4008 | * the single character LATIN_SMALL_LETTER_Y_WITH_DIAERESIS. | |
| 4009 | * This mapping saves some tests for the majority of the | |
| 4010 | * characters */ | |
| 4011 | ||
| 4012 | if (upper != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { | |
| 4013 | ||
| 4014 | /* Not tricky. Just save it. */ | |
| 4015 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, upper); | |
| 4016 | } | |
| 4017 | else if (chr == LATIN_SMALL_LETTER_SHARP_S) { | |
| 4018 | ||
| 4019 | /* This one is tricky because it is two characters long, | |
| 4020 | * though the UTF-8 is still two bytes, so the stored | |
| 4021 | * length doesn't change */ | |
| 4022 | *tmpbuf = 'S'; /* The UTF-8 is 'Ss' */ | |
| 4023 | *(tmpbuf + 1) = 's'; | |
| 4024 | } | |
| 4025 | else { | |
| 4026 | ||
| 4027 | /* The other two have their title and upper cases the same, | |
| 4028 | * but are tricky because the changed-case characters | |
| 4029 | * aren't in the latin1 range. They, however, do fit into | |
| 4030 | * two UTF-8 bytes */ | |
| 4031 | STORE_NON_LATIN1_UC(tmpbuf, chr); | |
| 4032 | } | |
| 4033 | } | |
| 4034 | } | |
| 4035 | else { | |
| 4036 | #endif /* end of dont want to break user-defined casing */ | |
| 4037 | ||
| 4038 | /* Here, can't short-cut the general case */ | |
| 4039 | ||
| 4040 | utf8_to_uvchr(s, &ulen); | |
| 4041 | if (op_type == OP_UCFIRST) toTITLE_utf8(s, tmpbuf, &tculen); | |
| 4042 | else toLOWER_utf8(s, tmpbuf, &tculen); | |
| 4043 | ||
| 4044 | /* we can't do in-place if the length changes. */ | |
| 4045 | if (ulen != tculen) inplace = FALSE; | |
| 4046 | need = slen + 1 - ulen + tculen; | |
| 4047 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
| 4048 | } | |
| 4049 | #endif | |
| d54190f6 | 4050 | } |
| 00f254e2 KW |
4051 | else { /* Non-zero length, non-UTF-8, Need to consider locale and if |
| 4052 | * latin1 is treated as caseless. Note that a locale takes | |
| 4053 | * precedence */ | |
| 4054 | tculen = 1; /* Most characters will require one byte, but this will | |
| 4055 | * need to be overridden for the tricky ones */ | |
| 4056 | need = slen + 1; | |
| 4057 | ||
| 4058 | if (op_type == OP_LCFIRST) { | |
| d54190f6 | 4059 | |
| 00f254e2 KW |
4060 | /* lower case the first letter: no trickiness for any character */ |
| 4061 | *tmpbuf = (IN_LOCALE_RUNTIME) ? toLOWER_LC(*s) : | |
| 4062 | ((IN_UNI_8_BIT) ? toLOWER_LATIN1(*s) : toLOWER(*s)); | |
| 4063 | } | |
| 4064 | /* is ucfirst() */ | |
| 4065 | else if (IN_LOCALE_RUNTIME) { | |
| 4066 | *tmpbuf = toUPPER_LC(*s); /* This would be a bug if any locales | |
| 4067 | * have upper and title case different | |
| 4068 | */ | |
| 4069 | } | |
| 4070 | else if (! IN_UNI_8_BIT) { | |
| 4071 | *tmpbuf = toUPPER(*s); /* Returns caseless for non-ascii, or | |
| 4072 | * on EBCDIC machines whatever the | |
| 4073 | * native function does */ | |
| 4074 | } | |
| 4075 | else { /* is ucfirst non-UTF-8, not in locale, and cased latin1 */ | |
| 4076 | *tmpbuf = toUPPER_LATIN1_MOD(*s); | |
| 4077 | ||
| 4078 | /* tmpbuf now has the correct title case for all latin1 characters | |
| 4079 | * except for the several ones that have tricky handling. All | |
| 4080 | * of these are mapped by the MOD to the letter below. */ | |
| 4081 | if (*tmpbuf == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) { | |
| 4082 | ||
| 4083 | /* The length is going to change, with all three of these, so | |
| 4084 | * can't replace just the first character */ | |
| 4085 | inplace = FALSE; | |
| 4086 | ||
| 4087 | /* We use the original to distinguish between these tricky | |
| 4088 | * cases */ | |
| 4089 | if (*s == LATIN_SMALL_LETTER_SHARP_S) { | |
| 4090 | /* Two character title case 'Ss', but can remain non-UTF-8 */ | |
| 4091 | need = slen + 2; | |
| 4092 | *tmpbuf = 'S'; | |
| 4093 | *(tmpbuf + 1) = 's'; /* Assert: length(tmpbuf) >= 2 */ | |
| 4094 | tculen = 2; | |
| 4095 | } | |
| 4096 | else { | |
| d54190f6 | 4097 | |
| 00f254e2 KW |
4098 | /* The other two tricky ones have their title case outside |
| 4099 | * latin1. It is the same as their upper case. */ | |
| 4100 | doing_utf8 = TRUE; | |
| 4101 | STORE_NON_LATIN1_UC(tmpbuf, *s); | |
| 4102 | ||
| 4103 | /* The UTF-8 and UTF-EBCDIC lengths of both these characters | |
| 4104 | * and their upper cases is 2. */ | |
| 4105 | tculen = ulen = 2; | |
| 4106 | ||
| 4107 | /* The entire result will have to be in UTF-8. Assume worst | |
| 4108 | * case sizing in conversion. (all latin1 characters occupy | |
| 4109 | * at most two bytes in utf8) */ | |
| 4110 | convert_source_to_utf8 = TRUE; | |
| 4111 | need = slen * 2 + 1; | |
| 4112 | } | |
| 4113 | } /* End of is one of the three special chars */ | |
| 4114 | } /* End of use Unicode (Latin1) semantics */ | |
| 4115 | } /* End of changing the case of the first character */ | |
| 4116 | ||
| 4117 | /* Here, have the first character's changed case stored in tmpbuf. Ready to | |
| 4118 | * generate the result */ | |
| 4119 | if (inplace) { | |
| 4120 | ||
| 4121 | /* We can convert in place. This means we change just the first | |
| 4122 | * character without disturbing the rest; no need to grow */ | |
| d54190f6 NC |
4123 | dest = source; |
| 4124 | s = d = (U8*)SvPV_force_nomg(source, slen); | |
| 4125 | } else { | |
| 4126 | dTARGET; | |
| 4127 | ||
| 4128 | dest = TARG; | |
| 4129 | ||
| 00f254e2 KW |
4130 | /* Here, we can't convert in place; we earlier calculated how much |
| 4131 | * space we will need, so grow to accommodate that */ | |
| d54190f6 | 4132 | SvUPGRADE(dest, SVt_PV); |
| 3b416f41 | 4133 | d = (U8*)SvGROW(dest, need); |
| d54190f6 NC |
4134 | (void)SvPOK_only(dest); |
| 4135 | ||
| 4136 | SETs(dest); | |
| d54190f6 | 4137 | } |
| 44bc797b | 4138 | |
| d54190f6 | 4139 | if (doing_utf8) { |
| 00f254e2 KW |
4140 | if (! inplace) { |
| 4141 | if (! convert_source_to_utf8) { | |
| 4142 | ||
| 4143 | /* Here both source and dest are in UTF-8, but have to create | |
| 4144 | * the entire output. We initialize the result to be the | |
| 4145 | * title/lower cased first character, and then append the rest | |
| 4146 | * of the string. */ | |
| 4147 | sv_setpvn(dest, (char*)tmpbuf, tculen); | |
| 4148 | if (slen > ulen) { | |
| 4149 | sv_catpvn(dest, (char*)(s + ulen), slen - ulen); | |
| 4150 | } | |
| 4151 | } | |
| 4152 | else { | |
| 4153 | const U8 *const send = s + slen; | |
| 4154 | ||
| 4155 | /* Here the dest needs to be in UTF-8, but the source isn't, | |
| 4156 | * except we earlier UTF-8'd the first character of the source | |
| 4157 | * into tmpbuf. First put that into dest, and then append the | |
| 4158 | * rest of the source, converting it to UTF-8 as we go. */ | |
| 4159 | ||
| 4160 | /* Assert tculen is 2 here because the only two characters that | |
| 4161 | * get to this part of the code have 2-byte UTF-8 equivalents */ | |
| 4162 | *d++ = *tmpbuf; | |
| 4163 | *d++ = *(tmpbuf + 1); | |
| 4164 | s++; /* We have just processed the 1st char */ | |
| 4165 | ||
| 4166 | for (; s < send; s++) { | |
| 4167 | d = uvchr_to_utf8(d, *s); | |
| 4168 | } | |
| 4169 | *d = '\0'; | |
| 4170 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
| 4171 | } | |
| d54190f6 | 4172 | SvUTF8_on(dest); |
| a0ed51b3 | 4173 | } |
| 00f254e2 | 4174 | else { /* in-place UTF-8. Just overwrite the first character */ |
| d54190f6 NC |
4175 | Copy(tmpbuf, d, tculen, U8); |
| 4176 | SvCUR_set(dest, need - 1); | |
| a0ed51b3 | 4177 | } |
| a0ed51b3 | 4178 | } |
| 00f254e2 KW |
4179 | else { /* Neither source nor dest are in or need to be UTF-8 */ |
| 4180 | if (slen) { | |
| 2de3dbcc | 4181 | if (IN_LOCALE_RUNTIME) { |
| 31351b04 | 4182 | TAINT; |
| d54190f6 | 4183 | SvTAINTED_on(dest); |
| 31351b04 | 4184 | } |
| 00f254e2 KW |
4185 | if (inplace) { /* in-place, only need to change the 1st char */ |
| 4186 | *d = *tmpbuf; | |
| 4187 | } | |
| 4188 | else { /* Not in-place */ | |
| 4189 | ||
| 4190 | /* Copy the case-changed character(s) from tmpbuf */ | |
| 4191 | Copy(tmpbuf, d, tculen, U8); | |
| 4192 | d += tculen - 1; /* Code below expects d to point to final | |
| 4193 | * character stored */ | |
| 4194 | } | |
| 4195 | } | |
| 4196 | else { /* empty source */ | |
| 4197 | /* See bug #39028: Don't taint if empty */ | |
| d54190f6 NC |
4198 | *d = *s; |
| 4199 | } | |
| 4200 | ||
| 00f254e2 KW |
4201 | /* In a "use bytes" we don't treat the source as UTF-8, but, still want |
| 4202 | * the destination to retain that flag */ | |
| d54190f6 NC |
4203 | if (SvUTF8(source)) |
| 4204 | SvUTF8_on(dest); | |
| 4205 | ||
| 00f254e2 | 4206 | if (!inplace) { /* Finish the rest of the string, unchanged */ |
| d54190f6 NC |
4207 | /* This will copy the trailing NUL */ |
| 4208 | Copy(s + 1, d + 1, slen, U8); | |
| 4209 | SvCUR_set(dest, need - 1); | |
| bbce6d69 | 4210 | } |
| bbce6d69 | 4211 | } |
| d54190f6 | 4212 | SvSETMAGIC(dest); |
| 79072805 LW |
4213 | RETURN; |
| 4214 | } | |
| 4215 | ||
| 67306194 NC |
4216 | /* There's so much setup/teardown code common between uc and lc, I wonder if |
| 4217 | it would be worth merging the two, and just having a switch outside each | |
| 00f254e2 | 4218 | of the three tight loops. There is less and less commonality though */ |
| 79072805 LW |
4219 | PP(pp_uc) |
| 4220 | { | |
| 97aff369 | 4221 | dVAR; |
| 39644a26 | 4222 | dSP; |
| 67306194 | 4223 | SV *source = TOPs; |
| 463ee0b2 | 4224 | STRLEN len; |
| 67306194 NC |
4225 | STRLEN min; |
| 4226 | SV *dest; | |
| 4227 | const U8 *s; | |
| 4228 | U8 *d; | |
| 79072805 | 4229 | |
| 67306194 NC |
4230 | SvGETMAGIC(source); |
| 4231 | ||
| 4232 | if (SvPADTMP(source) && !SvREADONLY(source) && !SvAMAGIC(source) | |
| 00f254e2 KW |
4233 | && SvTEMP(source) && !DO_UTF8(source) |
| 4234 | && (IN_LOCALE_RUNTIME || ! IN_UNI_8_BIT)) { | |
| 4235 | ||
| 4236 | /* We can convert in place. The reason we can't if in UNI_8_BIT is to | |
| 4237 | * make the loop tight, so we overwrite the source with the dest before | |
| 4238 | * looking at it, and we need to look at the original source | |
| 4239 | * afterwards. There would also need to be code added to handle | |
| 4240 | * switching to not in-place in midstream if we run into characters | |
| 4241 | * that change the length. | |
| 4242 | */ | |
| 67306194 NC |
4243 | dest = source; |
| 4244 | s = d = (U8*)SvPV_force_nomg(source, len); | |
| 4245 | min = len + 1; | |
| 4246 | } else { | |
| a0ed51b3 | 4247 | dTARGET; |
| a0ed51b3 | 4248 | |
| 67306194 | 4249 | dest = TARG; |
| 128c9517 | 4250 | |
| 67306194 NC |
4251 | /* The old implementation would copy source into TARG at this point. |
| 4252 | This had the side effect that if source was undef, TARG was now | |
| 4253 | an undefined SV with PADTMP set, and they don't warn inside | |
| 4254 | sv_2pv_flags(). However, we're now getting the PV direct from | |
| 4255 | source, which doesn't have PADTMP set, so it would warn. Hence the | |
| 4256 | little games. */ | |
| 4257 | ||
| 4258 | if (SvOK(source)) { | |
| 4259 | s = (const U8*)SvPV_nomg_const(source, len); | |
| 4260 | } else { | |
| 0a0ffbce RGS |
4261 | if (ckWARN(WARN_UNINITIALIZED)) |
| 4262 | report_uninit(source); | |
| 1eced8f8 | 4263 | s = (const U8*)""; |
| 67306194 | 4264 | len = 0; |
| a0ed51b3 | 4265 | } |
| 67306194 NC |
4266 | min = len + 1; |
| 4267 | ||
| 4268 | SvUPGRADE(dest, SVt_PV); | |
| 3b416f41 | 4269 | d = (U8*)SvGROW(dest, min); |
| 67306194 NC |
4270 | (void)SvPOK_only(dest); |
| 4271 | ||
| 4272 | SETs(dest); | |
| a0ed51b3 | 4273 | } |
| 31351b04 | 4274 | |
| 67306194 NC |
4275 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need |
| 4276 | to check DO_UTF8 again here. */ | |
| 4277 | ||
| 4278 | if (DO_UTF8(source)) { | |
| 4279 | const U8 *const send = s + len; | |
| 4280 | U8 tmpbuf[UTF8_MAXBYTES+1]; | |
| 4281 | ||
| 4c8a458a KW |
4282 | /* All occurrences of these are to be moved to follow any other marks. |
| 4283 | * This is context-dependent. We may not be passed enough context to | |
| 4284 | * move the iota subscript beyond all of them, but we do the best we can | |
| 4285 | * with what we're given. The result is always better than if we | |
| 4286 | * hadn't done this. And, the problem would only arise if we are | |
| 4287 | * passed a character without all its combining marks, which would be | |
| 4288 | * the caller's mistake. The information this is based on comes from a | |
| 4289 | * comment in Unicode SpecialCasing.txt, (and the Standard's text | |
| 4290 | * itself) and so can't be checked properly to see if it ever gets | |
| 4291 | * revised. But the likelihood of it changing is remote */ | |
| 00f254e2 | 4292 | bool in_iota_subscript = FALSE; |
| 00f254e2 | 4293 | |
| 67306194 | 4294 | while (s < send) { |
| 00f254e2 KW |
4295 | if (in_iota_subscript && ! is_utf8_mark(s)) { |
| 4296 | /* A non-mark. Time to output the iota subscript */ | |
| 4297 | #define GREEK_CAPITAL_LETTER_IOTA 0x0399 | |
| 4298 | #define COMBINING_GREEK_YPOGEGRAMMENI 0x0345 | |
| 4299 | ||
| 4300 | CAT_UNI_TO_UTF8_TWO_BYTE(d, GREEK_CAPITAL_LETTER_IOTA); | |
| 4301 | in_iota_subscript = FALSE; | |
| 4302 | } | |
| 00f254e2 KW |
4303 | |
| 4304 | ||
| 4305 | /* See comments at the first instance in this file of this ifdef */ | |
| 4306 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
| 67306194 | 4307 | |
| 00f254e2 KW |
4308 | /* If the UTF-8 character is invariant, then it is in the range |
| 4309 | * known by the standard macro; result is only one byte long */ | |
| 4310 | if (UTF8_IS_INVARIANT(*s)) { | |
| 4311 | *d++ = toUPPER(*s); | |
| 4312 | s++; | |
| 4313 | } | |
| 4314 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { | |
| 4315 | ||
| 4316 | /* Likewise, if it fits in a byte, its case change is in our | |
| 4317 | * table */ | |
| 81367fea | 4318 | U8 orig = TWO_BYTE_UTF8_TO_UNI(*s, *s++); |
| 00f254e2 KW |
4319 | U8 upper = toUPPER_LATIN1_MOD(orig); |
| 4320 | CAT_TWO_BYTE_UNI_UPPER_MOD(d, orig, upper); | |
| 81367fea | 4321 | s++; |
| 00f254e2 KW |
4322 | } |
| 4323 | else { | |
| 4324 | #else | |
| 4325 | { | |
| 4326 | #endif | |
| 4327 | ||
| 4328 | /* Otherwise, need the general UTF-8 case. Get the changed | |
| 4329 | * case value and copy it to the output buffer */ | |
| 4330 | ||
| 4331 | const STRLEN u = UTF8SKIP(s); | |
| 4332 | STRLEN ulen; | |
| 67306194 | 4333 | |
| 00f254e2 | 4334 | const UV uv = toUPPER_utf8(s, tmpbuf, &ulen); |
| 4c8a458a KW |
4335 | if (uv == GREEK_CAPITAL_LETTER_IOTA |
| 4336 | && utf8_to_uvchr(s, 0) == COMBINING_GREEK_YPOGEGRAMMENI) | |
| 4337 | { | |
| 00f254e2 KW |
4338 | in_iota_subscript = TRUE; |
| 4339 | } | |
| 4340 | else { | |
| 00f254e2 KW |
4341 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { |
| 4342 | /* If the eventually required minimum size outgrows | |
| 4343 | * the available space, we need to grow. */ | |
| 4344 | const UV o = d - (U8*)SvPVX_const(dest); | |
| 4345 | ||
| 4346 | /* If someone uppercases one million U+03B0s we | |
| 4347 | * SvGROW() one million times. Or we could try | |
| 4348 | * guessing how much to allocate without allocating too | |
| 4c8a458a KW |
4349 | * much. Such is life. See corresponding comment in |
| 4350 | * lc code for another option */ | |
| 00f254e2 KW |
4351 | SvGROW(dest, min); |
| 4352 | d = (U8*)SvPVX(dest) + o; | |
| 4353 | } | |
| 4354 | Copy(tmpbuf, d, ulen, U8); | |
| 4355 | d += ulen; | |
| 00f254e2 | 4356 | } |
| 00f254e2 | 4357 | s += u; |
| 67306194 | 4358 | } |
| 67306194 | 4359 | } |
| 4c8a458a KW |
4360 | if (in_iota_subscript) { |
| 4361 | CAT_UNI_TO_UTF8_TWO_BYTE(d, GREEK_CAPITAL_LETTER_IOTA); | |
| 4362 | } | |
| 67306194 NC |
4363 | SvUTF8_on(dest); |
| 4364 | *d = '\0'; | |
| 4365 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
| 4c8a458a KW |
4366 | } |
| 4367 | else { /* Not UTF-8 */ | |
| 67306194 NC |
4368 | if (len) { |
| 4369 | const U8 *const send = s + len; | |
| 00f254e2 KW |
4370 | |
| 4371 | /* Use locale casing if in locale; regular style if not treating | |
| 4372 | * latin1 as having case; otherwise the latin1 casing. Do the | |
| 4373 | * whole thing in a tight loop, for speed, */ | |
| 2de3dbcc | 4374 | if (IN_LOCALE_RUNTIME) { |
| 31351b04 | 4375 | TAINT; |
| 67306194 NC |
4376 | SvTAINTED_on(dest); |
| 4377 | for (; s < send; d++, s++) | |
| 4378 | *d = toUPPER_LC(*s); | |
| 31351b04 | 4379 | } |
| 00f254e2 KW |
4380 | else if (! IN_UNI_8_BIT) { |
| 4381 | for (; s < send; d++, s++) { | |
| 67306194 | 4382 | *d = toUPPER(*s); |
| 00f254e2 | 4383 | } |
| 31351b04 | 4384 | } |
| 00f254e2 KW |
4385 | else { |
| 4386 | for (; s < send; d++, s++) { | |
| 4387 | *d = toUPPER_LATIN1_MOD(*s); | |
| 4388 | if (*d != LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) continue; | |
| 4389 | ||
| 4390 | /* The mainstream case is the tight loop above. To avoid | |
| 4391 | * extra tests in that, all three characters that require | |
| 4392 | * special handling are mapped by the MOD to the one tested | |
| 4393 | * just above. | |
| 4394 | * Use the source to distinguish between the three cases */ | |
| 4395 | ||
| 4396 | if (*s == LATIN_SMALL_LETTER_SHARP_S) { | |
| 4397 | ||
| 4398 | /* uc() of this requires 2 characters, but they are | |
| 4399 | * ASCII. If not enough room, grow the string */ | |
| 4400 | if (SvLEN(dest) < ++min) { | |
| 4401 | const UV o = d - (U8*)SvPVX_const(dest); | |
| 4402 | SvGROW(dest, min); | |
| 4403 | d = (U8*)SvPVX(dest) + o; | |
| 4404 | } | |
| 4405 | *d++ = 'S'; *d = 'S'; /* upper case is 'SS' */ | |
| 4406 | continue; /* Back to the tight loop; still in ASCII */ | |
| 4407 | } | |
| 4408 | ||
| 4409 | /* The other two special handling characters have their | |
| 4410 | * upper cases outside the latin1 range, hence need to be | |
| 4411 | * in UTF-8, so the whole result needs to be in UTF-8. So, | |
| 4412 | * here we are somewhere in the middle of processing a | |
| 4413 | * non-UTF-8 string, and realize that we will have to convert | |
| 4414 | * the whole thing to UTF-8. What to do? There are | |
| 4415 | * several possibilities. The simplest to code is to | |
| 4416 | * convert what we have so far, set a flag, and continue on | |
| 4417 | * in the loop. The flag would be tested each time through | |
| 4418 | * the loop, and if set, the next character would be | |
| 4419 | * converted to UTF-8 and stored. But, I (khw) didn't want | |
| 4420 | * to slow down the mainstream case at all for this fairly | |
| 4421 | * rare case, so I didn't want to add a test that didn't | |
| 4422 | * absolutely have to be there in the loop, besides the | |
| 4423 | * possibility that it would get too complicated for | |
| 4424 | * optimizers to deal with. Another possibility is to just | |
| 4425 | * give up, convert the source to UTF-8, and restart the | |
| 4426 | * function that way. Another possibility is to convert | |
| 4427 | * both what has already been processed and what is yet to | |
| 4428 | * come separately to UTF-8, then jump into the loop that | |
| 4429 | * handles UTF-8. But the most efficient time-wise of the | |
| 4430 | * ones I could think of is what follows, and turned out to | |
| 4431 | * not require much extra code. */ | |
| 4432 | ||
| 4433 | /* Convert what we have so far into UTF-8, telling the | |
| 4434 | * function that we know it should be converted, and to | |
| 4435 | * allow extra space for what we haven't processed yet. | |
| 4436 | * Assume the worst case space requirements for converting | |
| 4437 | * what we haven't processed so far: that it will require | |
| 4438 | * two bytes for each remaining source character, plus the | |
| 4439 | * NUL at the end. This may cause the string pointer to | |
| 4440 | * move, so re-find it. */ | |
| 4441 | ||
| 4442 | len = d - (U8*)SvPVX_const(dest); | |
| 4443 | SvCUR_set(dest, len); | |
| 4444 | len = sv_utf8_upgrade_flags_grow(dest, | |
| 4445 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
| 4446 | (send -s) * 2 + 1); | |
| 4447 | d = (U8*)SvPVX(dest) + len; | |
| 4448 | ||
| 4449 | /* And append the current character's upper case in UTF-8 */ | |
| 4450 | CAT_NON_LATIN1_UC(d, *s); | |
| 4451 | ||
| 4452 | /* Now process the remainder of the source, converting to | |
| 4453 | * upper and UTF-8. If a resulting byte is invariant in | |
| 4454 | * UTF-8, output it as-is, otherwise convert to UTF-8 and | |
| 4455 | * append it to the output. */ | |
| 4456 | ||
| 4457 | s++; | |
| 4458 | for (; s < send; s++) { | |
| 4459 | U8 upper = toUPPER_LATIN1_MOD(*s); | |
| 4460 | if UTF8_IS_INVARIANT(upper) { | |
| 4461 | *d++ = upper; | |
| 4462 | } | |
| 4463 | else { | |
| 4464 | CAT_TWO_BYTE_UNI_UPPER_MOD(d, *s, upper); | |
| 4465 | } | |
| 4466 | } | |
| 4467 | ||
| 4468 | /* Here have processed the whole source; no need to continue | |
| 4469 | * with the outer loop. Each character has been converted | |
| 4470 | * to upper case and converted to UTF-8 */ | |
| 4471 | ||
| 4472 | break; | |
| 4473 | } /* End of processing all latin1-style chars */ | |
| 4474 | } /* End of processing all chars */ | |
| 4475 | } /* End of source is not empty */ | |
| 4476 | ||
| 67306194 | 4477 | if (source != dest) { |
| 00f254e2 | 4478 | *d = '\0'; /* Here d points to 1 after last char, add NUL */ |
| 67306194 NC |
4479 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); |
| 4480 | } | |
| 00f254e2 | 4481 | } /* End of isn't utf8 */ |
| 67306194 | 4482 | SvSETMAGIC(dest); |
| 79072805 LW |
4483 | RETURN; |
| 4484 | } | |
| 4485 | ||
| 4486 | PP(pp_lc) | |
| 4487 | { | |
| 97aff369 | 4488 | dVAR; |
| 39644a26 | 4489 | dSP; |
| ec9af7d4 | 4490 | SV *source = TOPs; |
| 463ee0b2 | 4491 | STRLEN len; |
| ec9af7d4 NC |
4492 | STRLEN min; |
| 4493 | SV *dest; | |
| 4494 | const U8 *s; | |
| 4495 | U8 *d; | |
| 79072805 | 4496 | |
| ec9af7d4 NC |
4497 | SvGETMAGIC(source); |
| 4498 | ||
| 4499 | if (SvPADTMP(source) && !SvREADONLY(source) && !SvAMAGIC(source) | |
| 17fa0776 | 4500 | && SvTEMP(source) && !DO_UTF8(source)) { |
| ec9af7d4 | 4501 | |
| 00f254e2 KW |
4502 | /* We can convert in place, as lowercasing anything in the latin1 range |
| 4503 | * (or else DO_UTF8 would have been on) doesn't lengthen it */ | |
| ec9af7d4 NC |
4504 | dest = source; |
| 4505 | s = d = (U8*)SvPV_force_nomg(source, len); | |
| 4506 | min = len + 1; | |
| 4507 | } else { | |
| a0ed51b3 | 4508 | dTARGET; |
| a0ed51b3 | 4509 | |
| ec9af7d4 NC |
4510 | dest = TARG; |
| 4511 | ||
| 4512 | /* The old implementation would copy source into TARG at this point. | |
| 4513 | This had the side effect that if source was undef, TARG was now | |
| 4514 | an undefined SV with PADTMP set, and they don't warn inside | |
| 4515 | sv_2pv_flags(). However, we're now getting the PV direct from | |
| 4516 | source, which doesn't have PADTMP set, so it would warn. Hence the | |
| 4517 | little games. */ | |
| 4518 | ||
| 4519 | if (SvOK(source)) { | |
| 4520 | s = (const U8*)SvPV_nomg_const(source, len); | |
| 4521 | } else { | |
| 0a0ffbce RGS |
4522 | if (ckWARN(WARN_UNINITIALIZED)) |
| 4523 | report_uninit(source); | |
| 1eced8f8 | 4524 | s = (const U8*)""; |
| ec9af7d4 | 4525 | len = 0; |
| a0ed51b3 | 4526 | } |
| ec9af7d4 | 4527 | min = len + 1; |
| 128c9517 | 4528 | |
| ec9af7d4 | 4529 | SvUPGRADE(dest, SVt_PV); |
| 3b416f41 | 4530 | d = (U8*)SvGROW(dest, min); |
| ec9af7d4 NC |
4531 | (void)SvPOK_only(dest); |
| 4532 | ||
| 4533 | SETs(dest); | |
| 4534 | } | |
| 4535 | ||
| 4536 | /* Overloaded values may have toggled the UTF-8 flag on source, so we need | |
| 4537 | to check DO_UTF8 again here. */ | |
| 4538 | ||
| 4539 | if (DO_UTF8(source)) { | |
| 4540 | const U8 *const send = s + len; | |
| 4541 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; | |
| 4542 | ||
| 4543 | while (s < send) { | |
| 00f254e2 KW |
4544 | /* See comments at the first instance in this file of this ifdef */ |
| 4545 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
| 4546 | if (UTF8_IS_INVARIANT(*s)) { | |
| 89ebb4a3 | 4547 | |
| 00f254e2 | 4548 | /* Invariant characters use the standard mappings compiled in. |
| ec9af7d4 | 4549 | */ |
| 00f254e2 KW |
4550 | *d++ = toLOWER(*s); |
| 4551 | s++; | |
| ec9af7d4 | 4552 | } |
| 00f254e2 | 4553 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { |
| 89ebb4a3 | 4554 | |
| 00f254e2 | 4555 | /* As do the ones in the Latin1 range */ |
| 81367fea | 4556 | U8 lower = toLOWER_LATIN1(TWO_BYTE_UTF8_TO_UNI(*s, *s++)); |
| 00f254e2 | 4557 | CAT_UNI_TO_UTF8_TWO_BYTE(d, lower); |
| 81367fea | 4558 | s++; |
| a0ed51b3 | 4559 | } |
| 00f254e2 KW |
4560 | else { |
| 4561 | #endif | |
| 4562 | /* Here, is utf8 not in Latin-1 range, have to go out and get | |
| 4563 | * the mappings from the tables. */ | |
| 4564 | ||
| 4565 | const STRLEN u = UTF8SKIP(s); | |
| 4566 | STRLEN ulen; | |
| 4567 | ||
| 00f254e2 KW |
4568 | #ifndef CONTEXT_DEPENDENT_CASING |
| 4569 | toLOWER_utf8(s, tmpbuf, &ulen); | |
| 4570 | #else | |
| 4c8a458a KW |
4571 | /* This is ifdefd out because it needs more work and thought. It isn't clear |
| 4572 | * that we should do it. | |
| 4573 | * A minor objection is that this is based on a hard-coded rule from the | |
| 4574 | * Unicode standard, and may change, but this is not very likely at all. | |
| 4575 | * mktables should check and warn if it does. | |
| 4576 | * More importantly, if the sigma occurs at the end of the string, we don't | |
| 4577 | * have enough context to know whether it is part of a larger string or going | |
| 4578 | * to be or not. It may be that we are passed a subset of the context, via | |
| 4579 | * a \U...\E, for example, and we could conceivably know the larger context if | |
| 4580 | * code were changed to pass that in. But, if the string passed in is an | |
| 4581 | * intermediate result, and the user concatenates two strings together | |
| 4582 | * after we have made a final sigma, that would be wrong. If the final sigma | |
| 4583 | * occurs in the middle of the string we are working on, then we know that it | |
| 4584 | * should be a final sigma, but otherwise we can't be sure. */ | |
| 00f254e2 KW |
4585 | |
| 4586 | const UV uv = toLOWER_utf8(s, tmpbuf, &ulen); | |
| 4587 | ||
| 4588 | /* If the lower case is a small sigma, it may be that we need | |
| 4589 | * to change it to a final sigma. This happens at the end of | |
| 4590 | * a word that contains more than just this character, and only | |
| 4591 | * when we started with a capital sigma. */ | |
| 4592 | if (uv == UNICODE_GREEK_SMALL_LETTER_SIGMA && | |
| 4593 | s > send - len && /* Makes sure not the first letter */ | |
| 4594 | utf8_to_uvchr(s, 0) == UNICODE_GREEK_CAPITAL_LETTER_SIGMA | |
| 4595 | ) { | |
| 4596 | ||
| 4597 | /* We use the algorithm in: | |
| 4598 | * http://www.unicode.org/versions/Unicode5.0.0/ch03.pdf (C | |
| 4599 | * is a CAPITAL SIGMA): If C is preceded by a sequence | |
| 4600 | * consisting of a cased letter and a case-ignorable | |
| 4601 | * sequence, and C is not followed by a sequence consisting | |
| 4602 | * of a case ignorable sequence and then a cased letter, | |
| 4603 | * then when lowercasing C, C becomes a final sigma */ | |
| 4604 | ||
| 4605 | /* To determine if this is the end of a word, need to peek | |
| 4606 | * ahead. Look at the next character */ | |
| 4607 | const U8 *peek = s + u; | |
| 4608 | ||
| 4609 | /* Skip any case ignorable characters */ | |
| 4610 | while (peek < send && is_utf8_case_ignorable(peek)) { | |
| 4611 | peek += UTF8SKIP(peek); | |
| 4612 | } | |
| 4613 | ||
| 4614 | /* If we reached the end of the string without finding any | |
| 4615 | * non-case ignorable characters, or if the next such one | |
| 4616 | * is not-cased, then we have met the conditions for it | |
| 4617 | * being a final sigma with regards to peek ahead, and so | |
| 4618 | * must do peek behind for the remaining conditions. (We | |
| 4619 | * know there is stuff behind to look at since we tested | |
| 4620 | * above that this isn't the first letter) */ | |
| 4621 | if (peek >= send || ! is_utf8_cased(peek)) { | |
| 4622 | peek = utf8_hop(s, -1); | |
| 4623 | ||
| 4624 | /* Here are at the beginning of the first character | |
| 4625 | * before the original upper case sigma. Keep backing | |
| 4626 | * up, skipping any case ignorable characters */ | |
| 4627 | while (is_utf8_case_ignorable(peek)) { | |
| 4628 | peek = utf8_hop(peek, -1); | |
| 4629 | } | |
| 4630 | ||
| 4631 | /* Here peek points to the first byte of the closest | |
| 4632 | * non-case-ignorable character before the capital | |
| 4633 | * sigma. If it is cased, then by the Unicode | |
| 4634 | * algorithm, we should use a small final sigma instead | |
| 4635 | * of what we have */ | |
| 4636 | if (is_utf8_cased(peek)) { | |
| 4637 | STORE_UNI_TO_UTF8_TWO_BYTE(tmpbuf, | |
| 4638 | UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA); | |
| 4639 | } | |
| 4640 | } | |
| 4641 | } | |
| 4642 | else { /* Not a context sensitive mapping */ | |
| 4643 | #endif /* End of commented out context sensitive */ | |
| 4644 | if (ulen > u && (SvLEN(dest) < (min += ulen - u))) { | |
| 4645 | ||
| 4646 | /* If the eventually required minimum size outgrows | |
| 4647 | * the available space, we need to grow. */ | |
| 4648 | const UV o = d - (U8*)SvPVX_const(dest); | |
| 4649 | ||
| 4650 | /* If someone lowercases one million U+0130s we | |
| 4651 | * SvGROW() one million times. Or we could try | |
| 4652 | * guessing how much to allocate without allocating too | |
| 4653 | * much. Such is life. Another option would be to | |
| 4654 | * grow an extra byte or two more each time we need to | |
| 4655 | * grow, which would cut down the million to 500K, with | |
| 4656 | * little waste */ | |
| 4657 | SvGROW(dest, min); | |
| 4658 | d = (U8*)SvPVX(dest) + o; | |
| 4659 | } | |
| 4660 | #ifdef CONTEXT_DEPENDENT_CASING | |
| 4661 | } | |
| 4662 | #endif | |
| 4663 | /* Copy the newly lowercased letter to the output buffer we're | |
| 4664 | * building */ | |
| 4665 | Copy(tmpbuf, d, ulen, U8); | |
| 4666 | d += ulen; | |
| 4667 | s += u; | |
| 4668 | #ifdef GO_AHEAD_AND_BREAK_USER_DEFINED_CASE_MAPPINGS | |
| 4669 | } | |
| 4670 | #endif | |
| 4671 | } /* End of looping through the source string */ | |
| ec9af7d4 NC |
4672 | SvUTF8_on(dest); |
| 4673 | *d = '\0'; | |
| 4674 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
| 00f254e2 | 4675 | } else { /* Not utf8 */ |
| 31351b04 | 4676 | if (len) { |
| ec9af7d4 | 4677 | const U8 *const send = s + len; |
| 00f254e2 KW |
4678 | |
| 4679 | /* Use locale casing if in locale; regular style if not treating | |
| 4680 | * latin1 as having case; otherwise the latin1 casing. Do the | |
| 4681 | * whole thing in a tight loop, for speed, */ | |
| 2de3dbcc | 4682 | if (IN_LOCALE_RUNTIME) { |
| 31351b04 | 4683 | TAINT; |
| ec9af7d4 NC |
4684 | SvTAINTED_on(dest); |
| 4685 | for (; s < send; d++, s++) | |
| 4686 | *d = toLOWER_LC(*s); | |
| 31351b04 | 4687 | } |
| 00f254e2 KW |
4688 | else if (! IN_UNI_8_BIT) { |
| 4689 | for (; s < send; d++, s++) { | |
| ec9af7d4 | 4690 | *d = toLOWER(*s); |
| 00f254e2 KW |
4691 | } |
| 4692 | } | |
| 4693 | else { | |
| 4694 | for (; s < send; d++, s++) { | |
| 4695 | *d = toLOWER_LATIN1(*s); | |
| 4696 | } | |
| 31351b04 | 4697 | } |
| bbce6d69 | 4698 | } |
| ec9af7d4 NC |
4699 | if (source != dest) { |
| 4700 | *d = '\0'; | |
| 4701 | SvCUR_set(dest, d - (U8*)SvPVX_const(dest)); | |
| 4702 | } | |
| 79072805 | 4703 | } |
| ec9af7d4 | 4704 | SvSETMAGIC(dest); |
| 79072805 LW |
4705 | RETURN; |
| 4706 | } | |
| 4707 | ||
| a0d0e21e | 4708 | PP(pp_quotemeta) |
| 79072805 | 4709 | { |
| 97aff369 | 4710 | dVAR; dSP; dTARGET; |
| 1b6737cc | 4711 | SV * const sv = TOPs; |
| a0d0e21e | 4712 | STRLEN len; |
| 0d46e09a | 4713 | register const char *s = SvPV_const(sv,len); |
| 79072805 | 4714 | |
| 7e2040f0 | 4715 | SvUTF8_off(TARG); /* decontaminate */ |
| a0d0e21e | 4716 | if (len) { |
| 1b6737cc | 4717 | register char *d; |
| 862a34c6 | 4718 | SvUPGRADE(TARG, SVt_PV); |
| c07a80fd | 4719 | SvGROW(TARG, (len * 2) + 1); |
| a0d0e21e | 4720 | d = SvPVX(TARG); |
| 7e2040f0 | 4721 | if (DO_UTF8(sv)) { |
| 0dd2cdef | 4722 | while (len) { |
| fd400ab9 | 4723 | if (UTF8_IS_CONTINUED(*s)) { |
| 0dd2cdef LW |
4724 | STRLEN ulen = UTF8SKIP(s); |
| 4725 | if (ulen > len) | |
| 4726 | ulen = len; | |
| 4727 | len -= ulen; | |
| 4728 | while (ulen--) | |
| 4729 | *d++ = *s++; | |
| 4730 | } | |
| 4731 | else { | |
| 4732 | if (!isALNUM(*s)) | |
| 4733 | *d++ = '\\'; | |
| 4734 | *d++ = *s++; | |
| 4735 | len--; | |
| 4736 | } | |
| 4737 | } | |
| 7e2040f0 | 4738 | SvUTF8_on(TARG); |
| 0dd2cdef LW |
4739 | } |
| 4740 | else { | |
| 4741 | while (len--) { | |
| 4742 | if (!isALNUM(*s)) | |
| 4743 | *d++ = '\\'; | |
| 4744 | *d++ = *s++; | |
| 4745 | } | |
| 79072805 | 4746 | } |
| a0d0e21e | 4747 | *d = '\0'; |
| 349d4f2f | 4748 | SvCUR_set(TARG, d - SvPVX_const(TARG)); |
| 3aa33fe5 | 4749 | (void)SvPOK_only_UTF8(TARG); |
| 79072805 | 4750 | } |
| a0d0e21e LW |
4751 | else |
| 4752 | sv_setpvn(TARG, s, len); | |
| ec93b65f | 4753 | SETTARG; |
| 79072805 LW |
4754 | RETURN; |
| 4755 | } | |
| 4756 | ||
| a0d0e21e | 4757 | /* Arrays. */ |
| 79072805 | 4758 | |
| a0d0e21e | 4759 | PP(pp_aslice) |
| 79072805 | 4760 | { |
| 97aff369 | 4761 | dVAR; dSP; dMARK; dORIGMARK; |
| 502c6561 | 4762 | register AV *const av = MUTABLE_AV(POPs); |
| 1b6737cc | 4763 | register const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); |
| 79072805 | 4764 | |
| a0d0e21e | 4765 | if (SvTYPE(av) == SVt_PVAV) { |
| fc15ae8f | 4766 | const I32 arybase = CopARYBASE_get(PL_curcop); |
| 4ad10a0b VP |
4767 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; |
| 4768 | bool can_preserve = FALSE; | |
| 4769 | ||
| 4770 | if (localizing) { | |
| 4771 | MAGIC *mg; | |
| 4772 | HV *stash; | |
| 4773 | ||
| 4774 | can_preserve = SvCANEXISTDELETE(av); | |
| 4775 | } | |
| 4776 | ||
| 4777 | if (lval && localizing) { | |
| 1b6737cc | 4778 | register SV **svp; |
| 748a9306 | 4779 | I32 max = -1; |
| 924508f0 | 4780 | for (svp = MARK + 1; svp <= SP; svp++) { |
| 4ea561bc | 4781 | const I32 elem = SvIV(*svp); |
| 748a9306 LW |
4782 | if (elem > max) |
| 4783 | max = elem; | |
| 4784 | } | |
| 4785 | if (max > AvMAX(av)) | |
| 4786 | av_extend(av, max); | |
| 4787 | } | |
| 4ad10a0b | 4788 | |
| a0d0e21e | 4789 | while (++MARK <= SP) { |
| 1b6737cc | 4790 | register SV **svp; |
| 4ea561bc | 4791 | I32 elem = SvIV(*MARK); |
| 4ad10a0b | 4792 | bool preeminent = TRUE; |
| a0d0e21e | 4793 | |
| 748a9306 LW |
4794 | if (elem > 0) |
| 4795 | elem -= arybase; | |
| 4ad10a0b VP |
4796 | if (localizing && can_preserve) { |
| 4797 | /* If we can determine whether the element exist, | |
| 4798 | * Try to preserve the existenceness of a tied array | |
| 4799 | * element by using EXISTS and DELETE if possible. | |
| 4800 | * Fallback to FETCH and STORE otherwise. */ | |
| 4801 | preeminent = av_exists(av, elem); | |
| 4802 | } | |
| 4803 | ||
| a0d0e21e LW |
4804 | svp = av_fetch(av, elem, lval); |
| 4805 | if (lval) { | |
| 3280af22 | 4806 | if (!svp || *svp == &PL_sv_undef) |
| cea2e8a9 | 4807 | DIE(aTHX_ PL_no_aelem, elem); |
| 4ad10a0b VP |
4808 | if (localizing) { |
| 4809 | if (preeminent) | |
| 4810 | save_aelem(av, elem, svp); | |
| 4811 | else | |
| 4812 | SAVEADELETE(av, elem); | |
| 4813 | } | |
| 79072805 | 4814 | } |
| 3280af22 | 4815 | *MARK = svp ? *svp : &PL_sv_undef; |
| 79072805 LW |
4816 | } |
| 4817 | } | |
| 748a9306 | 4818 | if (GIMME != G_ARRAY) { |
| a0d0e21e | 4819 | MARK = ORIGMARK; |
| 04ab2c87 | 4820 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
| a0d0e21e LW |
4821 | SP = MARK; |
| 4822 | } | |
| 79072805 LW |
4823 | RETURN; |
| 4824 | } | |
| 4825 | ||
| cba5a3b0 DG |
4826 | /* Smart dereferencing for keys, values and each */ |
| 4827 | PP(pp_rkeys) | |
| 4828 | { | |
| 4829 | dVAR; | |
| 4830 | dSP; | |
| 4831 | dPOPss; | |
| 4832 | ||
| 4833 | if (!SvOK(sv)) | |
| 4834 | RETURN; | |
| 4835 | ||
| 4836 | if (SvROK(sv)) { | |
| 4837 | SvGETMAGIC(sv); | |
| 4838 | if (SvAMAGIC(sv)) { | |
| 4839 | /* N.B.: AMG macros return sv if no overloading is found */ | |
| 31d632c3 DM |
4840 | SV *maybe_hv = AMG_CALLunary(sv, to_hv_amg); |
| 4841 | SV *maybe_av = AMG_CALLunary(sv, to_av_amg); | |
| cba5a3b0 DG |
4842 | if ( maybe_hv != sv && maybe_av != sv ) { |
| 4843 | Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS), "%s", | |
| 4844 | Perl_form(aTHX_ "Ambiguous overloaded argument to %s resolved as %%{}", | |
| 4845 | PL_op_desc[PL_op->op_type] | |
| 4846 | ) | |
| 4847 | ); | |
| 4848 | sv = maybe_hv; | |
| 4849 | } | |
| 4850 | else if ( maybe_av != sv ) { | |
| 4851 | if ( SvTYPE(SvRV(sv)) == SVt_PVHV ) { | |
| 4852 | /* @{} overload, but underlying reftype is HV */ | |
| 4853 | Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS), "%s", | |
| 4854 | Perl_form(aTHX_ "Ambiguous overloaded argument to %s resolved as @{}", | |
| 4855 | PL_op_desc[PL_op->op_type] | |
| 4856 | ) | |
| 4857 | ); | |
| 4858 | } | |
| 4859 | sv = maybe_av; | |
| 4860 | } | |
| 4861 | else if ( maybe_hv != sv ) { | |
| 4862 | if ( SvTYPE(SvRV(sv)) == SVt_PVAV ) { | |
| 4863 | /* %{} overload, but underlying reftype is AV */ | |
| 4864 | Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS), "%s", | |
| 4865 | Perl_form(aTHX_ "Ambiguous overloaded argument to %s resolved as %%{}", | |
| 4866 | PL_op_desc[PL_op->op_type] | |
| 4867 | ) | |
| 4868 | ); | |
| 4869 | } | |
| 4870 | sv = maybe_hv; | |
| 4871 | } | |
| 4872 | } | |
| 4873 | sv = SvRV(sv); | |
| 4874 | } | |
| 4875 | ||
| 4876 | if ( SvTYPE(sv) != SVt_PVHV && SvTYPE(sv) != SVt_PVAV ) { | |
| 4c540399 RGS |
4877 | DIE(aTHX_ "Type of argument to %s must be hashref or arrayref", |
| 4878 | PL_op_desc[PL_op->op_type] ); | |
| cba5a3b0 DG |
4879 | } |
| 4880 | ||
| 4881 | /* Delegate to correct function for op type */ | |
| 4882 | PUSHs(sv); | |
| 4883 | if (PL_op->op_type == OP_RKEYS || PL_op->op_type == OP_RVALUES) { | |
| 4884 | return (SvTYPE(sv) == SVt_PVHV) ? Perl_do_kv(aTHX) : Perl_pp_akeys(aTHX); | |
| 4885 | } | |
| 4886 | else { | |
| 4887 | return (SvTYPE(sv) == SVt_PVHV) ? Perl_pp_each(aTHX) : Perl_pp_aeach(aTHX); | |
| 4888 | } | |
| 4889 | } | |
| 4890 | ||
| 878d132a NC |
4891 | PP(pp_aeach) |
| 4892 | { | |
| 4893 | dVAR; | |
| 4894 | dSP; | |
| 502c6561 | 4895 | AV *array = MUTABLE_AV(POPs); |
| 878d132a | 4896 | const I32 gimme = GIMME_V; |
| 453d94a9 | 4897 | IV *iterp = Perl_av_iter_p(aTHX_ array); |
| 878d132a NC |
4898 | const IV current = (*iterp)++; |
| 4899 | ||
| 4900 | if (current > av_len(array)) { | |
| 4901 | *iterp = 0; | |
| 4902 | if (gimme == G_SCALAR) | |
| 4903 | RETPUSHUNDEF; | |
| 4904 | else | |
| 4905 | RETURN; | |
| 4906 | } | |
| 4907 | ||
| 4908 | EXTEND(SP, 2); | |
| 4909 | mPUSHi(CopARYBASE_get(PL_curcop) + current); | |
| 4910 | if (gimme == G_ARRAY) { | |
| 4911 | SV **const element = av_fetch(array, current, 0); | |
| 4912 | PUSHs(element ? *element : &PL_sv_undef); | |
| 4913 | } | |
| 4914 | RETURN; | |
| 4915 | } | |
| 4916 | ||
| 4917 | PP(pp_akeys) | |
| 4918 | { | |
| 4919 | dVAR; | |
| 4920 | dSP; | |
| 502c6561 | 4921 | AV *array = MUTABLE_AV(POPs); |
| 878d132a NC |
4922 | const I32 gimme = GIMME_V; |
| 4923 | ||
| 4924 | *Perl_av_iter_p(aTHX_ array) = 0; | |
| 4925 | ||
| 4926 | if (gimme == G_SCALAR) { | |
| 4927 | dTARGET; | |
| 4928 | PUSHi(av_len(array) + 1); | |
| 4929 | } | |
| 4930 | else if (gimme == G_ARRAY) { | |
| 4931 | IV n = Perl_av_len(aTHX_ array); | |
| 4932 | IV i = CopARYBASE_get(PL_curcop); | |
| 4933 | ||
| 4934 | EXTEND(SP, n + 1); | |
| 4935 | ||
| cba5a3b0 | 4936 | if (PL_op->op_type == OP_AKEYS || PL_op->op_type == OP_RKEYS) { |
| 878d132a NC |
4937 | n += i; |
| 4938 | for (; i <= n; i++) { | |
| 4939 | mPUSHi(i); | |
| 4940 | } | |
| 4941 | } | |
| 4942 | else { | |
| 4943 | for (i = 0; i <= n; i++) { | |
| 4944 | SV *const *const elem = Perl_av_fetch(aTHX_ array, i, 0); | |
| 4945 | PUSHs(elem ? *elem : &PL_sv_undef); | |
| 4946 | } | |
| 4947 | } | |
| 4948 | } | |
| 4949 | RETURN; | |
| 4950 | } | |
| 4951 | ||
| 79072805 LW |
4952 | /* Associative arrays. */ |
| 4953 | ||
| 4954 | PP(pp_each) | |
| 4955 | { | |
| 97aff369 | 4956 | dVAR; |
| 39644a26 | 4957 | dSP; |
| 85fbaab2 | 4958 | HV * hash = MUTABLE_HV(POPs); |
| c07a80fd | 4959 | HE *entry; |
| f54cb97a | 4960 | const I32 gimme = GIMME_V; |
| 8ec5e241 | 4961 | |
| c07a80fd | 4962 | PUTBACK; |
| c750a3ec | 4963 | /* might clobber stack_sp */ |
| 6d822dc4 | 4964 | entry = hv_iternext(hash); |
| c07a80fd | 4965 | SPAGAIN; |
| 79072805 | 4966 | |
| 79072805 LW |
4967 | EXTEND(SP, 2); |
| 4968 | if (entry) { | |
| 1b6737cc | 4969 | SV* const sv = hv_iterkeysv(entry); |
| 574c8022 | 4970 | PUSHs(sv); /* won't clobber stack_sp */ |
| 54310121 | 4971 | if (gimme == G_ARRAY) { |
| 59af0135 | 4972 | SV *val; |
| c07a80fd | 4973 | PUTBACK; |
| c750a3ec | 4974 | /* might clobber stack_sp */ |
| 6d822dc4 | 4975 | val = hv_iterval(hash, entry); |
| c07a80fd | 4976 | SPAGAIN; |
| 59af0135 | 4977 | PUSHs(val); |
| 79072805 | 4978 | } |
| 79072805 | 4979 | } |
| 54310121 | 4980 | else if (gimme == G_SCALAR) |
| 79072805 LW |
4981 | RETPUSHUNDEF; |
| 4982 | ||
| 4983 | RETURN; | |
| 4984 | } | |
| 4985 | ||
| 7332a6c4 VP |
4986 | STATIC OP * |
| 4987 | S_do_delete_local(pTHX) | |
| 79072805 | 4988 | { |
| 97aff369 | 4989 | dVAR; |
| 39644a26 | 4990 | dSP; |
| f54cb97a | 4991 | const I32 gimme = GIMME_V; |
| 7332a6c4 VP |
4992 | const MAGIC *mg; |
| 4993 | HV *stash; | |
| 4994 | ||
| 4995 | if (PL_op->op_private & OPpSLICE) { | |
| 4996 | dMARK; dORIGMARK; | |
| 4997 | SV * const osv = POPs; | |
| 4998 | const bool tied = SvRMAGICAL(osv) | |
| 4999 | && mg_find((const SV *)osv, PERL_MAGIC_tied); | |
| 5000 | const bool can_preserve = SvCANEXISTDELETE(osv) | |
| 5001 | || mg_find((const SV *)osv, PERL_MAGIC_env); | |
| 5002 | const U32 type = SvTYPE(osv); | |
| 5003 | if (type == SVt_PVHV) { /* hash element */ | |
| 5004 | HV * const hv = MUTABLE_HV(osv); | |
| 5005 | while (++MARK <= SP) { | |
| 5006 | SV * const keysv = *MARK; | |
| 5007 | SV *sv = NULL; | |
| 5008 | bool preeminent = TRUE; | |
| 5009 | if (can_preserve) | |
| 5010 | preeminent = hv_exists_ent(hv, keysv, 0); | |
| 5011 | if (tied) { | |
| 5012 | HE *he = hv_fetch_ent(hv, keysv, 1, 0); | |
| 5013 | if (he) | |
| 5014 | sv = HeVAL(he); | |
| 5015 | else | |
| 5016 | preeminent = FALSE; | |
| 5017 | } | |
| 5018 | else { | |
| 5019 | sv = hv_delete_ent(hv, keysv, 0, 0); | |
| 5020 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
| 5021 | } | |
| 5022 | if (preeminent) { | |
| 5023 | save_helem_flags(hv, keysv, &sv, SAVEf_KEEPOLDELEM); | |
| 5024 | if (tied) { | |
| 5025 | *MARK = sv_mortalcopy(sv); | |
| 5026 | mg_clear(sv); | |
| 5027 | } else | |
| 5028 | *MARK = sv; | |
| 5029 | } | |
| 5030 | else { | |
| 5031 | SAVEHDELETE(hv, keysv); | |
| 5032 | *MARK = &PL_sv_undef; | |
| 5033 | } | |
| 5034 | } | |
| 5035 | } | |
| 5036 | else if (type == SVt_PVAV) { /* array element */ | |
| 5037 | if (PL_op->op_flags & OPf_SPECIAL) { | |
| 5038 | AV * const av = MUTABLE_AV(osv); | |
| 5039 | while (++MARK <= SP) { | |
| 5040 | I32 idx = SvIV(*MARK); | |
| 5041 | SV *sv = NULL; | |
| 5042 | bool preeminent = TRUE; | |
| 5043 | if (can_preserve) | |
| 5044 | preeminent = av_exists(av, idx); | |
| 5045 | if (tied) { | |
| 5046 | SV **svp = av_fetch(av, idx, 1); | |
| 5047 | if (svp) | |
| 5048 | sv = *svp; | |
| 5049 | else | |
| 5050 | preeminent = FALSE; | |
| 5051 | } | |
| 5052 | else { | |
| 5053 | sv = av_delete(av, idx, 0); | |
| 5054 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
| 5055 | } | |
| 5056 | if (preeminent) { | |
| 5057 | save_aelem_flags(av, idx, &sv, SAVEf_KEEPOLDELEM); | |
| 5058 | if (tied) { | |
| 5059 | *MARK = sv_mortalcopy(sv); | |
| 5060 | mg_clear(sv); | |
| 5061 | } else | |
| 5062 | *MARK = sv; | |
| 5063 | } | |
| 5064 | else { | |
| 5065 | SAVEADELETE(av, idx); | |
| 5066 | *MARK = &PL_sv_undef; | |
| 5067 | } | |
| 5068 | } | |
| 5069 | } | |
| 5070 | } | |
| 5071 | else | |
| 5072 | DIE(aTHX_ "Not a HASH reference"); | |
| 5073 | if (gimme == G_VOID) | |
| 5074 | SP = ORIGMARK; | |
| 5075 | else if (gimme == G_SCALAR) { | |
| 5076 | MARK = ORIGMARK; | |
| 5077 | if (SP > MARK) | |
| 5078 | *++MARK = *SP; | |
| 5079 | else | |
| 5080 | *++MARK = &PL_sv_undef; | |
| 5081 | SP = MARK; | |
| 5082 | } | |
| 5083 | } | |
| 5084 | else { | |
| 5085 | SV * const keysv = POPs; | |
| 5086 | SV * const osv = POPs; | |
| 5087 | const bool tied = SvRMAGICAL(osv) | |
| 5088 | && mg_find((const SV *)osv, PERL_MAGIC_tied); | |
| 5089 | const bool can_preserve = SvCANEXISTDELETE(osv) | |
| 5090 | || mg_find((const SV *)osv, PERL_MAGIC_env); | |
| 5091 | const U32 type = SvTYPE(osv); | |
| 5092 | SV *sv = NULL; | |
| 5093 | if (type == SVt_PVHV) { | |
| 5094 | HV * const hv = MUTABLE_HV(osv); | |
| 5095 | bool preeminent = TRUE; | |
| 5096 | if (can_preserve) | |
| 5097 | preeminent = hv_exists_ent(hv, keysv, 0); | |
| 5098 | if (tied) { | |
| 5099 | HE *he = hv_fetch_ent(hv, keysv, 1, 0); | |
| 5100 | if (he) | |
| 5101 | sv = HeVAL(he); | |
| 5102 | else | |
| 5103 | preeminent = FALSE; | |
| 5104 | } | |
| 5105 | else { | |
| 5106 | sv = hv_delete_ent(hv, keysv, 0, 0); | |
| 5107 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
| 5108 | } | |
| 5109 | if (preeminent) { | |
| 5110 | save_helem_flags(hv, keysv, &sv, SAVEf_KEEPOLDELEM); | |
| 5111 | if (tied) { | |
| 5112 | SV *nsv = sv_mortalcopy(sv); | |
| 5113 | mg_clear(sv); | |
| 5114 | sv = nsv; | |
| 5115 | } | |
| 5116 | } | |
| 5117 | else | |
| 5118 | SAVEHDELETE(hv, keysv); | |
| 5119 | } | |
| 5120 | else if (type == SVt_PVAV) { | |
| 5121 | if (PL_op->op_flags & OPf_SPECIAL) { | |
| 5122 | AV * const av = MUTABLE_AV(osv); | |
| 5123 | I32 idx = SvIV(keysv); | |
| 5124 | bool preeminent = TRUE; | |
| 5125 | if (can_preserve) | |
| 5126 | preeminent = av_exists(av, idx); | |
| 5127 | if (tied) { | |
| 5128 | SV **svp = av_fetch(av, idx, 1); | |
| 5129 | if (svp) | |
| 5130 | sv = *svp; | |
| 5131 | else | |
| 5132 | preeminent = FALSE; | |
| 5133 | } | |
| 5134 | else { | |
| 5135 | sv = av_delete(av, idx, 0); | |
| 5136 | SvREFCNT_inc_simple_void(sv); /* De-mortalize */ | |
| 5137 | } | |
| 5138 | if (preeminent) { | |
| 5139 | save_aelem_flags(av, idx, &sv, SAVEf_KEEPOLDELEM); | |
| 5140 | if (tied) { | |
| 5141 | SV *nsv = sv_mortalcopy(sv); | |
| 5142 | mg_clear(sv); | |
| 5143 | sv = nsv; | |
| 5144 | } | |
| 5145 | } | |
| 5146 | else | |
| 5147 | SAVEADELETE(av, idx); | |
| 5148 | } | |
| 5149 | else | |
| 5150 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
| 5151 | } | |
| 5152 | else | |
| 5153 | DIE(aTHX_ "Not a HASH reference"); | |
| 5154 | if (!sv) | |
| 5155 | sv = &PL_sv_undef; | |
| 5156 | if (gimme != G_VOID) | |
| 5157 | PUSHs(sv); | |
| 5158 | } | |
| 5159 | ||
| 5160 | RETURN; | |
| 5161 | } | |
| 5162 | ||
| 5163 | PP(pp_delete) | |
| 5164 | { | |
| 5165 | dVAR; | |
| 5166 | dSP; | |
| 5167 | I32 gimme; | |
| 5168 | I32 discard; | |
| 5169 | ||
| 5170 | if (PL_op->op_private & OPpLVAL_INTRO) | |
| 5171 | return do_delete_local(); | |
| 5172 | ||
| 5173 | gimme = GIMME_V; | |
| 5174 | discard = (gimme == G_VOID) ? G_DISCARD : 0; | |
| 5f05dabc | 5175 | |
| 533c011a | 5176 | if (PL_op->op_private & OPpSLICE) { |
| 5f05dabc | 5177 | dMARK; dORIGMARK; |
| 85fbaab2 | 5178 | HV * const hv = MUTABLE_HV(POPs); |
| 1b6737cc | 5179 | const U32 hvtype = SvTYPE(hv); |
| 01020589 GS |
5180 | if (hvtype == SVt_PVHV) { /* hash element */ |
| 5181 | while (++MARK <= SP) { | |
| 1b6737cc | 5182 | SV * const sv = hv_delete_ent(hv, *MARK, discard, 0); |
| 01020589 GS |
5183 | *MARK = sv ? sv : &PL_sv_undef; |
| 5184 | } | |
| 5f05dabc | 5185 | } |
| 6d822dc4 MS |
5186 | else if (hvtype == SVt_PVAV) { /* array element */ |
| 5187 | if (PL_op->op_flags & OPf_SPECIAL) { | |
| 5188 | while (++MARK <= SP) { | |
| 502c6561 | 5189 | SV * const sv = av_delete(MUTABLE_AV(hv), SvIV(*MARK), discard); |
| 6d822dc4 MS |
5190 | *MARK = sv ? sv : &PL_sv_undef; |
| 5191 | } | |
| 5192 | } | |
| 01020589 GS |
5193 | } |
| 5194 | else | |
| 5195 | DIE(aTHX_ "Not a HASH reference"); | |
| 54310121 | 5196 | if (discard) |
| 5197 | SP = ORIGMARK; | |
| 5198 | else if (gimme == G_SCALAR) { | |
| 5f05dabc | 5199 | MARK = ORIGMARK; |
| 9111c9c0 DM |
5200 | if (SP > MARK) |
| 5201 | *++MARK = *SP; | |
| 5202 | else | |
| 5203 | *++MARK = &PL_sv_undef; | |
| 5f05dabc | 5204 | SP = MARK; |
| 5205 | } | |
| 5206 | } | |
| 5207 | else { | |
| 5208 | SV *keysv = POPs; | |
| 85fbaab2 | 5209 | HV * const hv = MUTABLE_HV(POPs); |
| 295d248e | 5210 | SV *sv = NULL; |
| 97fcbf96 MB |
5211 | if (SvTYPE(hv) == SVt_PVHV) |
| 5212 | sv = hv_delete_ent(hv, keysv, discard, 0); | |
| 01020589 GS |
5213 | else if (SvTYPE(hv) == SVt_PVAV) { |
| 5214 | if (PL_op->op_flags & OPf_SPECIAL) | |
| 502c6561 | 5215 | sv = av_delete(MUTABLE_AV(hv), SvIV(keysv), discard); |
| af288a60 HS |
5216 | else |
| 5217 | DIE(aTHX_ "panic: avhv_delete no longer supported"); | |
| 01020589 | 5218 | } |
| 97fcbf96 | 5219 | else |
| cea2e8a9 | 5220 | DIE(aTHX_ "Not a HASH reference"); |
| 5f05dabc | 5221 | if (!sv) |
| 3280af22 | 5222 | sv = &PL_sv_undef; |
| 54310121 | 5223 | if (!discard) |
| 5224 | PUSHs(sv); | |
| 79072805 | 5225 | } |
| 79072805 LW |
5226 | RETURN; |
| 5227 | } | |
| 5228 | ||
| a0d0e21e | 5229 | PP(pp_exists) |
| 79072805 | 5230 | { |
| 97aff369 | 5231 | dVAR; |
| 39644a26 | 5232 | dSP; |
| afebc493 GS |
5233 | SV *tmpsv; |
| 5234 | HV *hv; | |
| 5235 | ||
| 5236 | if (PL_op->op_private & OPpEXISTS_SUB) { | |
| 5237 | GV *gv; | |
| 0bd48802 | 5238 | SV * const sv = POPs; |
| f2c0649b | 5239 | CV * const cv = sv_2cv(sv, &hv, &gv, 0); |
| afebc493 GS |
5240 | if (cv) |
| 5241 | RETPUSHYES; | |
| 5242 | if (gv && isGV(gv) && GvCV(gv) && !GvCVGEN(gv)) | |
| 5243 | RETPUSHYES; | |
| 5244 | RETPUSHNO; | |
| 5245 | } | |
| 5246 | tmpsv = POPs; | |
| 85fbaab2 | 5247 | hv = MUTABLE_HV(POPs); |
| c750a3ec | 5248 | if (SvTYPE(hv) == SVt_PVHV) { |
| ae77835f | 5249 | if (hv_exists_ent(hv, tmpsv, 0)) |
| c750a3ec | 5250 | RETPUSHYES; |
| ef54e1a4 JH |
5251 | } |
| 5252 | else if (SvTYPE(hv) == SVt_PVAV) { | |
| 01020589 | 5253 | if (PL_op->op_flags & OPf_SPECIAL) { /* array element */ |
| 502c6561 | 5254 | if (av_exists(MUTABLE_AV(hv), SvIV(tmpsv))) |
| 01020589 GS |
5255 | RETPUSHYES; |
| 5256 | } | |
| ef54e1a4 JH |
5257 | } |
| 5258 | else { | |
| cea2e8a9 | 5259 | DIE(aTHX_ "Not a HASH reference"); |
| a0d0e21e | 5260 | } |
| a0d0e21e LW |
5261 | RETPUSHNO; |
| 5262 | } | |
| 79072805 | 5263 | |
| a0d0e21e LW |
5264 | PP(pp_hslice) |
| 5265 | { | |
| 97aff369 | 5266 | dVAR; dSP; dMARK; dORIGMARK; |
| 85fbaab2 | 5267 | register HV * const hv = MUTABLE_HV(POPs); |
| 1b6737cc AL |
5268 | register const I32 lval = (PL_op->op_flags & OPf_MOD || LVRET); |
| 5269 | const bool localizing = PL_op->op_private & OPpLVAL_INTRO; | |
| d30e492c | 5270 | bool can_preserve = FALSE; |
| 79072805 | 5271 | |
| eb85dfd3 DM |
5272 | if (localizing) { |
| 5273 | MAGIC *mg; | |
| 5274 | HV *stash; | |
| 5275 | ||
| d30e492c VP |
5276 | if (SvCANEXISTDELETE(hv) || mg_find((const SV *)hv, PERL_MAGIC_env)) |
| 5277 | can_preserve = TRUE; | |
| eb85dfd3 DM |
5278 | } |
| 5279 | ||
| 6d822dc4 | 5280 | while (++MARK <= SP) { |
| 1b6737cc | 5281 | SV * const keysv = *MARK; |
| 6d822dc4 MS |
5282 | SV **svp; |
| 5283 | HE *he; | |
| d30e492c VP |
5284 | bool preeminent = TRUE; |
| 5285 | ||
| 5286 | if (localizing && can_preserve) { | |
| 5287 | /* If we can determine whether the element exist, | |
| 5288 | * try to preserve the existenceness of a tied hash | |
| 5289 | * element by using EXISTS and DELETE if possible. | |
| 5290 | * Fallback to FETCH and STORE otherwise. */ | |
| 5291 | preeminent = hv_exists_ent(hv, keysv, 0); | |
| 6d822dc4 | 5292 | } |
| eb85dfd3 | 5293 | |
| 6d822dc4 | 5294 | he = hv_fetch_ent(hv, keysv, lval, 0); |
| fe5bfecd | 5295 | svp = he ? &HeVAL(he) : NULL; |
| eb85dfd3 | 5296 | |
| 6d822dc4 MS |
5297 | if (lval) { |
| 5298 | if (!svp || *svp == &PL_sv_undef) { | |
| be2597df | 5299 | DIE(aTHX_ PL_no_helem_sv, SVfARG(keysv)); |
| 6d822dc4 MS |
5300 | } |
| 5301 | if (localizing) { | |
| 7a2e501a | 5302 | if (HvNAME_get(hv) && isGV(*svp)) |
| 159b6efe | 5303 | save_gp(MUTABLE_GV(*svp), !(PL_op->op_flags & OPf_SPECIAL)); |
| 47cfc530 VP |
5304 | else if (preeminent) |
| 5305 | save_helem_flags(hv, keysv, svp, | |
| 5306 | (PL_op->op_flags & OPf_SPECIAL) ? 0 : SAVEf_SETMAGIC); | |
| 5307 | else | |
| 5308 | SAVEHDELETE(hv, keysv); | |
| 6d822dc4 MS |
5309 | } |
| 5310 | } | |
| 5311 | *MARK = svp ? *svp : &PL_sv_undef; | |
| 79072805 | 5312 | } |
| a0d0e21e LW |
5313 | if (GIMME != G_ARRAY) { |
| 5314 | MARK = ORIGMARK; | |
| 04ab2c87 | 5315 | *++MARK = SP > ORIGMARK ? *SP : &PL_sv_undef; |
| a0d0e21e | 5316 | SP = MARK; |
| 79072805 | 5317 | } |
| a0d0e21e LW |
5318 | RETURN; |
| 5319 | } | |
| 5320 | ||
| 5321 | /* List operators. */ | |
| 5322 | ||
| 5323 | PP(pp_list) | |
| 5324 | { | |
| 97aff369 | 5325 | dVAR; dSP; dMARK; |
| a0d0e21e LW |
5326 | if (GIMME != G_ARRAY) { |
| 5327 | if (++MARK <= SP) | |
| 5328 | *MARK = *SP; /* unwanted list, return last item */ | |
| 8990e307 | 5329 | else |
| 3280af22 | 5330 | *MARK = &PL_sv_undef; |
| a0d0e21e | 5331 | SP = MARK; |
| 79072805 | 5332 | } |
| a0d0e21e | 5333 | RETURN; |
| 79072805 LW |
5334 | } |
| 5335 | ||
| a0d0e21e | 5336 | PP(pp_lslice) |
| 79072805 | 5337 | { |
| 97aff369 | 5338 | dVAR; |
| 39644a26 | 5339 | dSP; |
| 1b6737cc AL |
5340 | SV ** const lastrelem = PL_stack_sp; |
| 5341 | SV ** const lastlelem = PL_stack_base + POPMARK; | |
| 5342 | SV ** const firstlelem = PL_stack_base + POPMARK + 1; | |
| 5343 | register SV ** const firstrelem = lastlelem + 1; | |
| fc15ae8f | 5344 | const I32 arybase = CopARYBASE_get(PL_curcop); |
| 42e73ed0 | 5345 | I32 is_something_there = FALSE; |
| 1b6737cc AL |
5346 | |
| 5347 | register const I32 max = lastrelem - lastlelem; | |
| a0d0e21e | 5348 | register SV **lelem; |
| a0d0e21e LW |
5349 | |
| 5350 | if (GIMME != G_ARRAY) { | |
| 4ea561bc | 5351 | I32 ix = SvIV(*lastlelem); |
| 748a9306 LW |
5352 | if (ix < 0) |
| 5353 | ix += max; | |
| 5354 | else | |
| 5355 | ix -= arybase; | |
| a0d0e21e | 5356 | if (ix < 0 || ix >= max) |
| 3280af22 | 5357 | *firstlelem = &PL_sv_undef; |
| a0d0e21e LW |
5358 | else |
| 5359 | *firstlelem = firstrelem[ix]; | |
| 5360 | SP = firstlelem; | |
| 5361 | RETURN; | |
| 5362 | } | |
| 5363 | ||
| 5364 | if (max == 0) { | |
| 5365 | SP = firstlelem - 1; | |
| 5366 | RETURN; | |
| 5367 | } | |
| 5368 | ||
| 5369 | for (lelem = firstlelem; lelem <= lastlelem; lelem++) { | |
| 4ea561bc | 5370 | I32 ix = SvIV(*lelem); |
| c73bf8e3 | 5371 | if (ix < 0) |
| a0d0e21e | 5372 | ix += max; |
| b13b2135 | 5373 | else |
| 748a9306 | 5374 | ix -= arybase; |
| c73bf8e3 HS |
5375 | if (ix < 0 || ix >= max) |
| 5376 | *lelem = &PL_sv_undef; | |
| 5377 | else { | |
| 5378 | is_something_there = TRUE; | |
| 5379 | if (!(*lelem = firstrelem[ix])) | |
| 3280af22 | 5380 | *lelem = &PL_sv_undef; |
| 748a9306 | 5381 | } |
| 79072805 | 5382 | } |
| 4633a7c4 LW |
5383 | if (is_something_there) |
| 5384 | SP = lastlelem; | |
| 5385 | else | |
| 5386 | SP = firstlelem - 1; | |
| 79072805 LW |
5387 | RETURN; |
| 5388 | } | |
| 5389 | ||
| a0d0e21e LW |
5390 | PP(pp_anonlist) |
| 5391 | { | |
| 97aff369 | 5392 | dVAR; dSP; dMARK; dORIGMARK; |
| 1b6737cc | 5393 | const I32 items = SP - MARK; |
| ad64d0ec | 5394 | SV * const av = MUTABLE_SV(av_make(items, MARK+1)); |
| 44a8e56a | 5395 | SP = ORIGMARK; /* av_make() might realloc stack_sp */ |
| 6e449a3a MHM |
5396 | mXPUSHs((PL_op->op_flags & OPf_SPECIAL) |
| 5397 | ? newRV_noinc(av) : av); | |
| a0d0e21e LW |
5398 | RETURN; |
| 5399 | } | |
| 5400 | ||
| 5401 | PP(pp_anonhash) | |
| 79072805 | 5402 | { |
| 97aff369 | 5403 | dVAR; dSP; dMARK; dORIGMARK; |
| 78c72037 | 5404 | HV* const hv = newHV(); |
| a0d0e21e LW |
5405 | |
| 5406 | while (MARK < SP) { | |
| 1b6737cc | 5407 | SV * const key = *++MARK; |
| 561b68a9 | 5408 | SV * const val = newSV(0); |
| a0d0e21e LW |
5409 | if (MARK < SP) |
| 5410 | sv_setsv(val, *++MARK); | |
| a2a5de95 NC |
5411 | else |
| 5412 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "Odd number of elements in anonymous hash"); | |
| f12c7020 | 5413 | (void)hv_store_ent(hv,key,val,0); |
| 79072805 | 5414 | } |
| a0d0e21e | 5415 | SP = ORIGMARK; |
| 6e449a3a | 5416 | mXPUSHs((PL_op->op_flags & OPf_SPECIAL) |
| ad64d0ec | 5417 | ? newRV_noinc(MUTABLE_SV(hv)) : MUTABLE_SV(hv)); |
| 79072805 LW |
5418 | RETURN; |
| 5419 | } | |
| 5420 | ||
| a0d0e21e | 5421 | PP(pp_splice) |
| 79072805 | 5422 | { |
| 27da23d5 | 5423 | dVAR; dSP; dMARK; dORIGMARK; |
| 502c6561 | 5424 | register AV *ary = MUTABLE_AV(*++MARK); |
| a0d0e21e LW |
5425 | register SV **src; |
| 5426 | register SV **dst; | |
| 5427 | register I32 i; | |
| 5428 | register I32 offset; | |
| 5429 | register I32 length; | |
| 5430 | I32 newlen; | |
| 5431 | I32 after; | |
| 5432 | I32 diff; | |
| ad64d0ec | 5433 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
| 93965878 | 5434 | |
| 1b6737cc | 5435 | if (mg) { |
| af71faff NC |
5436 | return Perl_tied_method(aTHX_ "SPLICE", mark - 1, MUTABLE_SV(ary), mg, |
| 5437 | GIMME_V | TIED_METHOD_ARGUMENTS_ON_STACK, | |
| 5438 | sp - mark); | |
| 93965878 | 5439 | } |
| 79072805 | 5440 | |
| a0d0e21e | 5441 | SP++; |
| 79072805 | 5442 | |
| a0d0e21e | 5443 | if (++MARK < SP) { |
| 4ea561bc | 5444 | offset = i = SvIV(*MARK); |
| a0d0e21e | 5445 | if (offset < 0) |
| 93965878 | 5446 | offset += AvFILLp(ary) + 1; |
| a0d0e21e | 5447 | else |
| fc15ae8f | 5448 | offset -= CopARYBASE_get(PL_curcop); |
| 84902520 | 5449 | if (offset < 0) |
| cea2e8a9 | 5450 | DIE(aTHX_ PL_no_aelem, i); |
| a0d0e21e LW |
5451 | if (++MARK < SP) { |
| 5452 | length = SvIVx(*MARK++); | |
| 48cdf507 GA |
5453 | if (length < 0) { |
| 5454 | length += AvFILLp(ary) - offset + 1; | |
| 5455 | if (length < 0) | |
| 5456 | length = 0; | |
| 5457 | } | |
| 79072805 LW |
5458 | } |
| 5459 | else | |
| a0d0e21e | 5460 | length = AvMAX(ary) + 1; /* close enough to infinity */ |
| 79072805 | 5461 | } |
| a0d0e21e LW |
5462 | else { |
| 5463 | offset = 0; | |
| 5464 | length = AvMAX(ary) + 1; | |
| 5465 | } | |
| 8cbc2e3b | 5466 | if (offset > AvFILLp(ary) + 1) { |
| a2a5de95 | 5467 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), "splice() offset past end of array" ); |
| 93965878 | 5468 | offset = AvFILLp(ary) + 1; |
| 8cbc2e3b | 5469 | } |
| 93965878 | 5470 | after = AvFILLp(ary) + 1 - (offset + length); |
| a0d0e21e LW |
5471 | if (after < 0) { /* not that much array */ |
| 5472 | length += after; /* offset+length now in array */ | |
| 5473 | after = 0; | |
| 5474 | if (!AvALLOC(ary)) | |
| 5475 | av_extend(ary, 0); | |
| 5476 | } | |
| 5477 | ||
| 5478 | /* At this point, MARK .. SP-1 is our new LIST */ | |
| 5479 | ||
| 5480 | newlen = SP - MARK; | |
| 5481 | diff = newlen - length; | |
| 13d7cbc1 GS |
5482 | if (newlen && !AvREAL(ary) && AvREIFY(ary)) |
| 5483 | av_reify(ary); | |
| a0d0e21e | 5484 | |
| 50528de0 WL |
5485 | /* make new elements SVs now: avoid problems if they're from the array */ |
| 5486 | for (dst = MARK, i = newlen; i; i--) { | |
| 1b6737cc | 5487 | SV * const h = *dst; |
| f2b990bf | 5488 | *dst++ = newSVsv(h); |
| 50528de0 WL |
5489 | } |
| 5490 | ||
| a0d0e21e | 5491 | if (diff < 0) { /* shrinking the area */ |
| 95b63a38 | 5492 | SV **tmparyval = NULL; |
| a0d0e21e | 5493 | if (newlen) { |
| a02a5408 | 5494 | Newx(tmparyval, newlen, SV*); /* so remember insertion */ |
| a0d0e21e | 5495 | Copy(MARK, tmparyval, newlen, SV*); |
| 79072805 | 5496 | } |
| a0d0e21e LW |
5497 | |
| 5498 | MARK = ORIGMARK + 1; | |
| 5499 | if (GIMME == G_ARRAY) { /* copy return vals to stack */ | |
| 5500 | MEXTEND(MARK, length); | |
| 5501 | Copy(AvARRAY(ary)+offset, MARK, length, SV*); | |
| 5502 | if (AvREAL(ary)) { | |
| bbce6d69 | 5503 | EXTEND_MORTAL(length); |
| 36477c24 | 5504 | for (i = length, dst = MARK; i; i--) { |
| 486ec47a | 5505 | sv_2mortal(*dst); /* free them eventually */ |
| 36477c24 | 5506 | dst++; |
| 5507 | } | |
| a0d0e21e LW |
5508 | } |
| 5509 | MARK += length - 1; | |
| 79072805 | 5510 | } |
| a0d0e21e LW |
5511 | else { |
| 5512 | *MARK = AvARRAY(ary)[offset+length-1]; | |
| 5513 | if (AvREAL(ary)) { | |
| d689ffdd | 5514 | sv_2mortal(*MARK); |
| a0d0e21e LW |
5515 | for (i = length - 1, dst = &AvARRAY(ary)[offset]; i > 0; i--) |
| 5516 | SvREFCNT_dec(*dst++); /* free them now */ | |
| 79072805 | 5517 | } |
| a0d0e21e | 5518 | } |
| 93965878 | 5519 | AvFILLp(ary) += diff; |
| a0d0e21e LW |
5520 | |
| 5521 | /* pull up or down? */ | |
| 5522 | ||
| 5523 | if (offset < after) { /* easier to pull up */ | |
| 5524 | if (offset) { /* esp. if nothing to pull */ | |
| 5525 | src = &AvARRAY(ary)[offset-1]; | |
| 5526 | dst = src - diff; /* diff is negative */ | |
| 5527 | for (i = offset; i > 0; i--) /* can't trust Copy */ | |
| 5528 | *dst-- = *src--; | |
| 79072805 | 5529 | } |
| a0d0e21e | 5530 | dst = AvARRAY(ary); |
| 9c6bc640 | 5531 | AvARRAY(ary) = AvARRAY(ary) - diff; /* diff is negative */ |
| a0d0e21e LW |
5532 | AvMAX(ary) += diff; |
| 5533 | } | |
| 5534 | else { | |
| 5535 | if (after) { /* anything to pull down? */ | |
| 5536 | src = AvARRAY(ary) + offset + length; | |
| 5537 | dst = src + diff; /* diff is negative */ | |
| 5538 | Move(src, dst, after, SV*); | |
| 79072805 | 5539 | } |
| 93965878 | 5540 | dst = &AvARRAY(ary)[AvFILLp(ary)+1]; |
| a0d0e21e LW |
5541 | /* avoid later double free */ |
| 5542 | } | |
| 5543 | i = -diff; | |
| 5544 | while (i) | |
| 3280af22 | 5545 | dst[--i] = &PL_sv_undef; |
| a0d0e21e LW |
5546 | |
| 5547 | if (newlen) { | |
| 50528de0 | 5548 | Copy( tmparyval, AvARRAY(ary) + offset, newlen, SV* ); |
| a0d0e21e LW |
5549 | Safefree(tmparyval); |
| 5550 | } | |
| 5551 | } | |
| 5552 | else { /* no, expanding (or same) */ | |
| d3961450 | 5553 | SV** tmparyval = NULL; |
| a0d0e21e | 5554 | if (length) { |
| a02a5408 | 5555 | Newx(tmparyval, length, SV*); /* so remember deletion */ |
| a0d0e21e LW |
5556 | Copy(AvARRAY(ary)+offset, tmparyval, length, SV*); |
| 5557 | } | |
| 5558 | ||
| 5559 | if (diff > 0) { /* expanding */ | |
| a0d0e21e | 5560 | /* push up or down? */ |
| a0d0e21e LW |
5561 | if (offset < after && diff <= AvARRAY(ary) - AvALLOC(ary)) { |
| 5562 | if (offset) { | |
| 5563 | src = AvARRAY(ary); | |
| 5564 | dst = src - diff; | |
| 5565 | Move(src, dst, offset, SV*); | |
| 79072805 | 5566 | } |
| 9c6bc640 | 5567 | AvARRAY(ary) = AvARRAY(ary) - diff;/* diff is positive */ |
| a0d0e21e | 5568 | AvMAX(ary) += diff; |
| 93965878 | 5569 | AvFILLp(ary) += diff; |
| 79072805 LW |
5570 | } |
| 5571 | else { | |
| 93965878 NIS |
5572 | if (AvFILLp(ary) + diff >= AvMAX(ary)) /* oh, well */ |
| 5573 | av_extend(ary, AvFILLp(ary) + diff); | |
| 5574 | AvFILLp(ary) += diff; | |
| a0d0e21e LW |
5575 | |
| 5576 | if (after) { | |
| 93965878 | 5577 | dst = AvARRAY(ary) + AvFILLp(ary); |
| a0d0e21e LW |
5578 | src = dst - diff; |
| 5579 | for (i = after; i; i--) { | |
| 5580 | *dst-- = *src--; | |
| 5581 | } | |
| 79072805 LW |
5582 | } |
| 5583 | } | |
| a0d0e21e LW |
5584 | } |
| 5585 | ||
| 50528de0 WL |
5586 | if (newlen) { |
| 5587 | Copy( MARK, AvARRAY(ary) + offset, newlen, SV* ); | |
| a0d0e21e | 5588 | } |
| 50528de0 | 5589 | |
| a0d0e21e LW |
5590 | MARK = ORIGMARK + 1; |
| 5591 | if (GIMME == G_ARRAY) { /* copy return vals to stack */ | |
| 5592 | if (length) { | |
| 5593 | Copy(tmparyval, MARK, length, SV*); | |
| 5594 | if (AvREAL(ary)) { | |
| bbce6d69 | 5595 | EXTEND_MORTAL(length); |
| 36477c24 | 5596 | for (i = length, dst = MARK; i; i--) { |
| 486ec47a | 5597 | sv_2mortal(*dst); /* free them eventually */ |
| 36477c24 | 5598 | dst++; |
| 5599 | } | |
| 79072805 LW |
5600 | } |
| 5601 | } | |
| a0d0e21e LW |
5602 | MARK += length - 1; |
| 5603 | } | |
| 5604 | else if (length--) { | |
| 5605 | *MARK = tmparyval[length]; | |
| 5606 | if (AvREAL(ary)) { | |
| d689ffdd | 5607 | sv_2mortal(*MARK); |
| a0d0e21e LW |
5608 | while (length-- > 0) |
| 5609 | SvREFCNT_dec(tmparyval[length]); | |
| 79072805 | 5610 | } |
| 79072805 | 5611 | } |
| a0d0e21e | 5612 | else |
| 3280af22 | 5613 | *MARK = &PL_sv_undef; |
| d3961450 | 5614 | Safefree(tmparyval); |
| 79072805 | 5615 | } |
| 474af990 FR |
5616 | |
| 5617 | if (SvMAGICAL(ary)) | |
| 5618 | mg_set(MUTABLE_SV(ary)); | |
| 5619 | ||
| a0d0e21e | 5620 | SP = MARK; |
| 79072805 LW |
5621 | RETURN; |
| 5622 | } | |
| 5623 | ||
| a0d0e21e | 5624 | PP(pp_push) |
| 79072805 | 5625 | { |
| 27da23d5 | 5626 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
| 502c6561 | 5627 | register AV * const ary = MUTABLE_AV(*++MARK); |
| ad64d0ec | 5628 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
| 79072805 | 5629 | |
| 1b6737cc | 5630 | if (mg) { |
| ad64d0ec | 5631 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
| 93965878 NIS |
5632 | PUSHMARK(MARK); |
| 5633 | PUTBACK; | |
| d343c3ef | 5634 | ENTER_with_name("call_PUSH"); |
| 864dbfa3 | 5635 | call_method("PUSH",G_SCALAR|G_DISCARD); |
| d343c3ef | 5636 | LEAVE_with_name("call_PUSH"); |
| 93965878 | 5637 | SPAGAIN; |
| 93965878 | 5638 | } |
| a60c0954 | 5639 | else { |
| 89c14e2e | 5640 | PL_delaymagic = DM_DELAY; |
| a60c0954 | 5641 | for (++MARK; MARK <= SP; MARK++) { |
| 561b68a9 | 5642 | SV * const sv = newSV(0); |
| a60c0954 NIS |
5643 | if (*MARK) |
| 5644 | sv_setsv(sv, *MARK); | |
| 0a75904b | 5645 | av_store(ary, AvFILLp(ary)+1, sv); |
| a60c0954 | 5646 | } |
| 354b0578 | 5647 | if (PL_delaymagic & DM_ARRAY_ISA) |
| ad64d0ec | 5648 | mg_set(MUTABLE_SV(ary)); |
| 89c14e2e BB |
5649 | |
| 5650 | PL_delaymagic = 0; | |
| 6eeabd23 VP |
5651 | } |
| 5652 | SP = ORIGMARK; | |
| 5653 | if (OP_GIMME(PL_op, 0) != G_VOID) { | |
| 5654 | PUSHi( AvFILL(ary) + 1 ); | |
| 79072805 | 5655 | } |
| 79072805 LW |
5656 | RETURN; |
| 5657 | } | |
| 5658 | ||
| a0d0e21e | 5659 | PP(pp_shift) |
| 79072805 | 5660 | { |
| 97aff369 | 5661 | dVAR; |
| 39644a26 | 5662 | dSP; |
| 538f5756 RZ |
5663 | AV * const av = PL_op->op_flags & OPf_SPECIAL |
| 5664 | ? MUTABLE_AV(GvAV(PL_defgv)) : MUTABLE_AV(POPs); | |
| 789b4bc9 | 5665 | SV * const sv = PL_op->op_type == OP_SHIFT ? av_shift(av) : av_pop(av); |
| 79072805 | 5666 | EXTEND(SP, 1); |
| c2b4a044 | 5667 | assert (sv); |
| d689ffdd | 5668 | if (AvREAL(av)) |
| a0d0e21e LW |
5669 | (void)sv_2mortal(sv); |
| 5670 | PUSHs(sv); | |
| 79072805 | 5671 | RETURN; |
| 79072805 LW |
5672 | } |
| 5673 | ||
| a0d0e21e | 5674 | PP(pp_unshift) |
| 79072805 | 5675 | { |
| 27da23d5 | 5676 | dVAR; dSP; dMARK; dORIGMARK; dTARGET; |
| 502c6561 | 5677 | register AV *ary = MUTABLE_AV(*++MARK); |
| ad64d0ec | 5678 | const MAGIC * const mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied); |
| 93965878 | 5679 | |
| 1b6737cc | 5680 | if (mg) { |
| ad64d0ec | 5681 | *MARK-- = SvTIED_obj(MUTABLE_SV(ary), mg); |
| 7fd66d9d | 5682 | PUSHMARK(MARK); |
| 93965878 | 5683 | PUTBACK; |
| d343c3ef | 5684 | ENTER_with_name("call_UNSHIFT"); |
| 864dbfa3 | 5685 | call_method("UNSHIFT",G_SCALAR|G_DISCARD); |
| d343c3ef | 5686 | LEAVE_with_name("call_UNSHIFT"); |
| 93965878 | 5687 | SPAGAIN; |
| 93965878 | 5688 | } |
| a60c0954 | 5689 | else { |
| 1b6737cc | 5690 | register I32 i = 0; |
| a60c0954 NIS |
5691 | av_unshift(ary, SP - MARK); |
| 5692 | while (MARK < SP) { | |
| 1b6737cc | 5693 | SV * const sv = newSVsv(*++MARK); |
| a60c0954 NIS |
5694 | (void)av_store(ary, i++, sv); |
| 5695 | } | |
| 79072805 | 5696 | } |
| a0d0e21e | 5697 | SP = ORIGMARK; |
| 6eeabd23 | 5698 | if (OP_GIMME(PL_op, 0) != G_VOID) { |
| 5658d0a9 LR |
5699 | PUSHi( AvFILL(ary) + 1 ); |
| 5700 | } | |
| 79072805 | 5701 | RETURN; |
| 79072805 LW |
5702 | } |
| 5703 | ||
| a0d0e21e | 5704 | PP(pp_reverse) |
| 79072805 | 5705 | { |
| 97aff369 | 5706 | dVAR; dSP; dMARK; |
| 79072805 | 5707 | |
| a0d0e21e | 5708 | if (GIMME == G_ARRAY) { |
| 484c818f VP |
5709 | if (PL_op->op_private & OPpREVERSE_INPLACE) { |
| 5710 | AV *av; | |
| 5711 | ||
| 5712 | /* See pp_sort() */ | |
| 5713 | assert( MARK+1 == SP && *SP && SvTYPE(*SP) == SVt_PVAV); | |
| 5714 | (void)POPMARK; /* remove mark associated with ex-OP_AASSIGN */ | |
| 5715 | av = MUTABLE_AV((*SP)); | |
| 5716 | /* In-place reversing only happens in void context for the array | |
| 5717 | * assignment. We don't need to push anything on the stack. */ | |
| 5718 | SP = MARK; | |
| 5719 | ||
| 5720 | if (SvMAGICAL(av)) { | |
| 5721 | I32 i, j; | |
| 5722 | register SV *tmp = sv_newmortal(); | |
| 5723 | /* For SvCANEXISTDELETE */ | |
| 5724 | HV *stash; | |
| 5725 | const MAGIC *mg; | |
| 5726 | bool can_preserve = SvCANEXISTDELETE(av); | |
| 5727 | ||
| 5728 | for (i = 0, j = av_len(av); i < j; ++i, --j) { | |
| 5729 | register SV *begin, *end; | |
| 5730 | ||
| 5731 | if (can_preserve) { | |
| 5732 | if (!av_exists(av, i)) { | |
| 5733 | if (av_exists(av, j)) { | |
| 5734 | register SV *sv = av_delete(av, j, 0); | |
| 5735 | begin = *av_fetch(av, i, TRUE); | |
| 5736 | sv_setsv_mg(begin, sv); | |
| 5737 | } | |
| 5738 | continue; | |
| 5739 | } | |
| 5740 | else if (!av_exists(av, j)) { | |
| 5741 | register SV *sv = av_delete(av, i, 0); | |
| 5742 | end = *av_fetch(av, j, TRUE); | |
| 5743 | sv_setsv_mg(end, sv); | |
| 5744 | continue; | |
| 5745 | } | |
| 5746 | } | |
| 5747 | ||
| 5748 | begin = *av_fetch(av, i, TRUE); | |
| 5749 | end = *av_fetch(av, j, TRUE); | |
| 5750 | sv_setsv(tmp, begin); | |
| 5751 | sv_setsv_mg(begin, end); | |
| 5752 | sv_setsv_mg(end, tmp); | |
| 5753 | } | |
| 5754 | } | |
| 5755 | else { | |
| 5756 | SV **begin = AvARRAY(av); | |
| 484c818f | 5757 | |
| 95a26d8e VP |
5758 | if (begin) { |
| 5759 | SV **end = begin + AvFILLp(av); | |
| 5760 | ||
| 5761 | while (begin < end) { | |
| 5762 | register SV * const tmp = *begin; | |
| 5763 | *begin++ = *end; | |
| 5764 | *end-- = tmp; | |
| 5765 | } | |
| 484c818f VP |
5766 | } |
| 5767 | } | |
| 5768 | } | |
| 5769 | else { | |
| 5770 | SV **oldsp = SP; | |
| 5771 | MARK++; | |
| 5772 | while (MARK < SP) { | |
| 5773 | register SV * const tmp = *MARK; | |
| 5774 | *MARK++ = *SP; | |
| 5775 | *SP-- = tmp; | |
| 5776 | } | |
| 5777 | /* safe as long as stack cannot get extended in the above */ | |
| 5778 | SP = oldsp; | |
| a0d0e21e | 5779 | } |
| 79072805 LW |
5780 | } |
| 5781 | else { | |
| a0d0e21e LW |
5782 | register char *up; |
| 5783 | register char *down; | |
| 5784 | register I32 tmp; | |
| 5785 | dTARGET; | |
| 5786 | STRLEN len; | |
| 79072805 | 5787 | |
| 7e2040f0 | 5788 | SvUTF8_off(TARG); /* decontaminate */ |
| a0d0e21e | 5789 | if (SP - MARK > 1) |
| 3280af22 | 5790 | do_join(TARG, &PL_sv_no, MARK, SP); |
| 1e21d011 | 5791 | else { |
| 789bd863 | 5792 | sv_setsv(TARG, SP > MARK ? *SP : find_rundefsv()); |
| 1e21d011 B |
5793 | if (! SvOK(TARG) && ckWARN(WARN_UNINITIALIZED)) |
| 5794 | report_uninit(TARG); | |
| 5795 | } | |
| 5796 | ||
| a0d0e21e LW |
5797 | up = SvPV_force(TARG, len); |
| 5798 | if (len > 1) { | |
| 7e2040f0 | 5799 | if (DO_UTF8(TARG)) { /* first reverse each character */ |
| dfe13c55 | 5800 | U8* s = (U8*)SvPVX(TARG); |
| 349d4f2f | 5801 | const U8* send = (U8*)(s + len); |
| a0ed51b3 | 5802 | while (s < send) { |
| d742c382 | 5803 | if (UTF8_IS_INVARIANT(*s)) { |
| a0ed51b3 LW |
5804 | s++; |
| 5805 | continue; | |
| 5806 | } | |
| 5807 | else { | |
| 9041c2e3 | 5808 | if (!utf8_to_uvchr(s, 0)) |
| a0dbb045 | 5809 | break; |
| dfe13c55 | 5810 | up = (char*)s; |
| a0ed51b3 | 5811 | s += UTF8SKIP(s); |
| dfe13c55 | 5812 | down = (char*)(s - 1); |
| a0dbb045 | 5813 | /* reverse this character */ |
| a0ed51b3 LW |
5814 | while (down > up) { |
| 5815 | tmp = *up; | |
| 5816 | *up++ = *down; | |
| eb160463 | 5817 | *down-- = (char)tmp; |
| a0ed51b3 LW |
5818 | } |
| 5819 | } | |
| 5820 | } | |
| 5821 | up = SvPVX(TARG); | |
| 5822 | } | |
| a0d0e21e LW |
5823 | down = SvPVX(TARG) + len - 1; |
| 5824 | while (down > up) { | |
| 5825 | tmp = *up; | |
| 5826 | *up++ = *down; | |
| eb160463 | 5827 | *down-- = (char)tmp; |
| a0d0e21e | 5828 | } |
| 3aa33fe5 | 5829 | (void)SvPOK_only_UTF8(TARG); |
| 79072805 | 5830 | } |
| a0d0e21e LW |
5831 | SP = MARK + 1; |
| 5832 | SETTARG; | |
| 79072805 | 5833 | } |
| a0d0e21e | 5834 | RETURN; |
| 79072805 LW |
5835 | } |
| 5836 | ||
| a0d0e21e | 5837 | PP(pp_split) |
| 79072805 | 5838 | { |
| 27da23d5 | 5839 | dVAR; dSP; dTARG; |
| a0d0e21e | 5840 | AV *ary; |
| 467f0320 | 5841 | register IV limit = POPi; /* note, negative is forever */ |
| 1b6737cc | 5842 | SV * const sv = POPs; |
| a0d0e21e | 5843 | STRLEN len; |
| 727b7506 | 5844 | register const char *s = SvPV_const(sv, len); |
| 1b6737cc | 5845 | const bool do_utf8 = DO_UTF8(sv); |
| 727b7506 | 5846 | const char *strend = s + len; |
| 44a8e56a | 5847 | register PMOP *pm; |
| d9f97599 | 5848 | register REGEXP *rx; |
| a0d0e21e | 5849 | register SV *dstr; |
| 727b7506 | 5850 | register const char *m; |
| a0d0e21e | 5851 | I32 iters = 0; |
| bb7a0f54 | 5852 | const STRLEN slen = do_utf8 ? utf8_length((U8*)s, (U8*)strend) : (STRLEN)(strend - s); |
| 792b2c16 | 5853 | I32 maxiters = slen + 10; |
| c1a7495a | 5854 | I32 trailing_empty = 0; |
| 727b7506 | 5855 | const char *orig; |
| 1b6737cc | 5856 | const I32 origlimit = limit; |
| a0d0e21e LW |
5857 | I32 realarray = 0; |
| 5858 | I32 base; | |
| f54cb97a | 5859 | const I32 gimme = GIMME_V; |
| 941446f6 | 5860 | bool gimme_scalar; |
| f54cb97a | 5861 | const I32 oldsave = PL_savestack_ix; |
| 437d3b4e | 5862 | U32 make_mortal = SVs_TEMP; |
| 7fba1cd6 | 5863 | bool multiline = 0; |
| b37c2d43 | 5864 | MAGIC *mg = NULL; |
| 79072805 | 5865 | |
| 44a8e56a | 5866 | #ifdef DEBUGGING |
| 5867 | Copy(&LvTARGOFF(POPs), &pm, 1, PMOP*); | |
| 5868 | #else | |
| 5869 | pm = (PMOP*)POPs; | |
| 5870 | #endif | |
| a0d0e21e | 5871 | if (!pm || !s) |
| 2269b42e | 5872 | DIE(aTHX_ "panic: pp_split"); |
| aaa362c4 | 5873 | rx = PM_GETRE(pm); |
| bbce6d69 | 5874 | |
| a62b1201 | 5875 | TAINT_IF(get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET && |
| 07bc277f | 5876 | (RX_EXTFLAGS(rx) & (RXf_WHITE | RXf_SKIPWHITE))); |
| bbce6d69 | 5877 | |
| a30b2f1f | 5878 | RX_MATCH_UTF8_set(rx, do_utf8); |
| d9f424b2 | 5879 | |
| 971a9dd3 | 5880 | #ifdef USE_ITHREADS |
| 20e98b0f | 5881 | if (pm->op_pmreplrootu.op_pmtargetoff) { |
| 159b6efe | 5882 | ary = GvAVn(MUTABLE_GV(PAD_SVl(pm->op_pmreplrootu.op_pmtargetoff))); |
| 20e98b0f | 5883 | } |
| 971a9dd3 | 5884 | #else |
| 20e98b0f NC |
5885 | if (pm->op_pmreplrootu.op_pmtargetgv) { |
| 5886 | ary = GvAVn(pm->op_pmreplrootu.op_pmtargetgv); | |
| 971a9dd3 | 5887 | } |
| 20e98b0f | 5888 | #endif |
| 79072805 | 5889 | else |
| 7d49f689 | 5890 | ary = NULL; |
| a0d0e21e LW |
5891 | if (ary && (gimme != G_ARRAY || (pm->op_pmflags & PMf_ONCE))) { |
| 5892 | realarray = 1; | |
| 8ec5e241 | 5893 | PUTBACK; |
| a0d0e21e LW |
5894 | av_extend(ary,0); |
| 5895 | av_clear(ary); | |
| 8ec5e241 | 5896 | SPAGAIN; |
| ad64d0ec | 5897 | if ((mg = SvTIED_mg((const SV *)ary, PERL_MAGIC_tied))) { |
| 8ec5e241 | 5898 | PUSHMARK(SP); |
| ad64d0ec | 5899 | XPUSHs(SvTIED_obj(MUTABLE_SV(ary), mg)); |
| 8ec5e241 NIS |
5900 | } |
| 5901 | else { | |
| 1c0b011c | 5902 | if (!AvREAL(ary)) { |
| 1b6737cc | 5903 | I32 i; |
| 1c0b011c | 5904 | AvREAL_on(ary); |
| abff13bb | 5905 | AvREIFY_off(ary); |
| 1c0b011c | 5906 | for (i = AvFILLp(ary); i >= 0; i--) |
| 3280af22 | 5907 | AvARRAY(ary)[i] = &PL_sv_undef; /* don't free mere refs */ |
| 1c0b011c NIS |
5908 | } |
| 5909 | /* temporarily switch stacks */ | |
| 8b7059b1 | 5910 | SAVESWITCHSTACK(PL_curstack, ary); |
| 8ec5e241 | 5911 | make_mortal = 0; |
| 1c0b011c | 5912 | } |
| 79072805 | 5913 | } |
| 3280af22 | 5914 | base = SP - PL_stack_base; |
| a0d0e21e | 5915 | orig = s; |
| 07bc277f | 5916 | if (RX_EXTFLAGS(rx) & RXf_SKIPWHITE) { |
| 613f191e TS |
5917 | if (do_utf8) { |
| 5918 | while (*s == ' ' || is_utf8_space((U8*)s)) | |
| 5919 | s += UTF8SKIP(s); | |
| 5920 | } | |
| a62b1201 | 5921 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) { |
| bbce6d69 | 5922 | while (isSPACE_LC(*s)) |
| 5923 | s++; | |
| 5924 | } | |
| 5925 | else { | |
| 5926 | while (isSPACE(*s)) | |
| 5927 | s++; | |
| 5928 | } | |
| a0d0e21e | 5929 | } |
| 73134a2e | 5930 | if (RX_EXTFLAGS(rx) & RXf_PMf_MULTILINE) { |
| 7fba1cd6 | 5931 | multiline = 1; |
| c07a80fd | 5932 | } |
| 5933 | ||
| 941446f6 FC |
5934 | gimme_scalar = gimme == G_SCALAR && !ary; |
| 5935 | ||
| a0d0e21e LW |
5936 | if (!limit) |
| 5937 | limit = maxiters + 2; | |
| 07bc277f | 5938 | if (RX_EXTFLAGS(rx) & RXf_WHITE) { |
| a0d0e21e | 5939 | while (--limit) { |
| bbce6d69 | 5940 | m = s; |
| 8727f688 YO |
5941 | /* this one uses 'm' and is a negative test */ |
| 5942 | if (do_utf8) { | |
| 613f191e TS |
5943 | while (m < strend && !( *m == ' ' || is_utf8_space((U8*)m) )) { |
| 5944 | const int t = UTF8SKIP(m); | |
| 5945 | /* is_utf8_space returns FALSE for malform utf8 */ | |
| 5946 | if (strend - m < t) | |
| 5947 | m = strend; | |
| 5948 | else | |
| 5949 | m += t; | |
| 5950 | } | |
| a62b1201 KW |
5951 | } |
| 5952 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) { | |
| 8727f688 YO |
5953 | while (m < strend && !isSPACE_LC(*m)) |
| 5954 | ++m; | |
| 5955 | } else { | |
| 5956 | while (m < strend && !isSPACE(*m)) | |
| 5957 | ++m; | |
| 5958 | } | |
| a0d0e21e LW |
5959 | if (m >= strend) |
| 5960 | break; | |
| bbce6d69 | 5961 | |
| c1a7495a BB |
5962 | if (gimme_scalar) { |
| 5963 | iters++; | |
| 5964 | if (m-s == 0) | |
| 5965 | trailing_empty++; | |
| 5966 | else | |
| 5967 | trailing_empty = 0; | |
| 5968 | } else { | |
| 5969 | dstr = newSVpvn_flags(s, m-s, | |
| 5970 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 5971 | XPUSHs(dstr); | |
| 5972 | } | |
| bbce6d69 | 5973 | |
| 613f191e TS |
5974 | /* skip the whitespace found last */ |
| 5975 | if (do_utf8) | |
| 5976 | s = m + UTF8SKIP(m); | |
| 5977 | else | |
| 5978 | s = m + 1; | |
| 5979 | ||
| 8727f688 YO |
5980 | /* this one uses 's' and is a positive test */ |
| 5981 | if (do_utf8) { | |
| 613f191e | 5982 | while (s < strend && ( *s == ' ' || is_utf8_space((U8*)s) )) |
| 8727f688 | 5983 | s += UTF8SKIP(s); |
| a62b1201 KW |
5984 | } |
| 5985 | else if (get_regex_charset(RX_EXTFLAGS(rx)) == REGEX_LOCALE_CHARSET) { | |
| 8727f688 YO |
5986 | while (s < strend && isSPACE_LC(*s)) |
| 5987 | ++s; | |
| 5988 | } else { | |
| 5989 | while (s < strend && isSPACE(*s)) | |
| 5990 | ++s; | |
| 5991 | } | |
| 79072805 LW |
5992 | } |
| 5993 | } | |
| 07bc277f | 5994 | else if (RX_EXTFLAGS(rx) & RXf_START_ONLY) { |
| a0d0e21e | 5995 | while (--limit) { |
| a6e20a40 AL |
5996 | for (m = s; m < strend && *m != '\n'; m++) |
| 5997 | ; | |
| a0d0e21e LW |
5998 | m++; |
| 5999 | if (m >= strend) | |
| 6000 | break; | |
| c1a7495a BB |
6001 | |
| 6002 | if (gimme_scalar) { | |
| 6003 | iters++; | |
| 6004 | if (m-s == 0) | |
| 6005 | trailing_empty++; | |
| 6006 | else | |
| 6007 | trailing_empty = 0; | |
| 6008 | } else { | |
| 6009 | dstr = newSVpvn_flags(s, m-s, | |
| 6010 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 6011 | XPUSHs(dstr); | |
| 6012 | } | |
| a0d0e21e LW |
6013 | s = m; |
| 6014 | } | |
| 6015 | } | |
| 07bc277f | 6016 | else if (RX_EXTFLAGS(rx) & RXf_NULL && !(s >= strend)) { |
| 640f820d AB |
6017 | /* |
| 6018 | Pre-extend the stack, either the number of bytes or | |
| 6019 | characters in the string or a limited amount, triggered by: | |
| 6020 | ||
| 6021 | my ($x, $y) = split //, $str; | |
| 6022 | or | |
| 6023 | split //, $str, $i; | |
| 6024 | */ | |
| c1a7495a BB |
6025 | if (!gimme_scalar) { |
| 6026 | const U32 items = limit - 1; | |
| 6027 | if (items < slen) | |
| 6028 | EXTEND(SP, items); | |
| 6029 | else | |
| 6030 | EXTEND(SP, slen); | |
| 6031 | } | |
| 640f820d | 6032 | |
| e9515b0f AB |
6033 | if (do_utf8) { |
| 6034 | while (--limit) { | |
| 6035 | /* keep track of how many bytes we skip over */ | |
| 6036 | m = s; | |
| 640f820d | 6037 | s += UTF8SKIP(s); |
| c1a7495a BB |
6038 | if (gimme_scalar) { |
| 6039 | iters++; | |
| 6040 | if (s-m == 0) | |
| 6041 | trailing_empty++; | |
| 6042 | else | |
| 6043 | trailing_empty = 0; | |
| 6044 | } else { | |
| 6045 | dstr = newSVpvn_flags(m, s-m, SVf_UTF8 | make_mortal); | |
| 640f820d | 6046 | |
| c1a7495a BB |
6047 | PUSHs(dstr); |
| 6048 | } | |
| 640f820d | 6049 | |
| e9515b0f AB |
6050 | if (s >= strend) |
| 6051 | break; | |
| 6052 | } | |
| 6053 | } else { | |
| 6054 | while (--limit) { | |
| c1a7495a BB |
6055 | if (gimme_scalar) { |
| 6056 | iters++; | |
| 6057 | } else { | |
| 6058 | dstr = newSVpvn(s, 1); | |
| e9515b0f | 6059 | |
| e9515b0f | 6060 | |
| c1a7495a BB |
6061 | if (make_mortal) |
| 6062 | sv_2mortal(dstr); | |
| 640f820d | 6063 | |
| c1a7495a BB |
6064 | PUSHs(dstr); |
| 6065 | } | |
| 6066 | ||
| 6067 | s++; | |
| e9515b0f AB |
6068 | |
| 6069 | if (s >= strend) | |
| 6070 | break; | |
| 6071 | } | |
| 640f820d AB |
6072 | } |
| 6073 | } | |
| 3c8556c3 | 6074 | else if (do_utf8 == (RX_UTF8(rx) != 0) && |
| 07bc277f NC |
6075 | (RX_EXTFLAGS(rx) & RXf_USE_INTUIT) && !RX_NPARENS(rx) |
| 6076 | && (RX_EXTFLAGS(rx) & RXf_CHECK_ALL) | |
| 6077 | && !(RX_EXTFLAGS(rx) & RXf_ANCH)) { | |
| 6078 | const int tail = (RX_EXTFLAGS(rx) & RXf_INTUIT_TAIL); | |
| f9f4320a | 6079 | SV * const csv = CALLREG_INTUIT_STRING(rx); |
| cf93c79d | 6080 | |
| 07bc277f | 6081 | len = RX_MINLENRET(rx); |
| 3c8556c3 | 6082 | if (len == 1 && !RX_UTF8(rx) && !tail) { |
| 1b6737cc | 6083 | const char c = *SvPV_nolen_const(csv); |
| a0d0e21e | 6084 | while (--limit) { |
| a6e20a40 AL |
6085 | for (m = s; m < strend && *m != c; m++) |
| 6086 | ; | |
| a0d0e21e LW |
6087 | if (m >= strend) |
| 6088 | break; | |
| c1a7495a BB |
6089 | if (gimme_scalar) { |
| 6090 | iters++; | |
| 6091 | if (m-s == 0) | |
| 6092 | trailing_empty++; | |
| 6093 | else | |
| 6094 | trailing_empty = 0; | |
| 6095 | } else { | |
| 6096 | dstr = newSVpvn_flags(s, m-s, | |
| 6097 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 6098 | XPUSHs(dstr); | |
| 6099 | } | |
| 93f04dac JH |
6100 | /* The rx->minlen is in characters but we want to step |
| 6101 | * s ahead by bytes. */ | |
| 1aa99e6b IH |
6102 | if (do_utf8) |
| 6103 | s = (char*)utf8_hop((U8*)m, len); | |
| 6104 | else | |
| 6105 | s = m + len; /* Fake \n at the end */ | |
| a0d0e21e LW |
6106 | } |
| 6107 | } | |
| 6108 | else { | |
| a0d0e21e | 6109 | while (s < strend && --limit && |
| f722798b | 6110 | (m = fbm_instr((unsigned char*)s, (unsigned char*)strend, |
| 7fba1cd6 | 6111 | csv, multiline ? FBMrf_MULTILINE : 0)) ) |
| a0d0e21e | 6112 | { |
| c1a7495a BB |
6113 | if (gimme_scalar) { |
| 6114 | iters++; | |
| 6115 | if (m-s == 0) | |
| 6116 | trailing_empty++; | |
| 6117 | else | |
| 6118 | trailing_empty = 0; | |
| 6119 | } else { | |
| 6120 | dstr = newSVpvn_flags(s, m-s, | |
| 6121 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 6122 | XPUSHs(dstr); | |
| 6123 | } | |
| 93f04dac JH |
6124 | /* The rx->minlen is in characters but we want to step |
| 6125 | * s ahead by bytes. */ | |
| 1aa99e6b IH |
6126 | if (do_utf8) |
| 6127 | s = (char*)utf8_hop((U8*)m, len); | |
| 6128 | else | |
| 6129 | s = m + len; /* Fake \n at the end */ | |
| a0d0e21e | 6130 | } |
| 463ee0b2 | 6131 | } |
| 463ee0b2 | 6132 | } |
| a0d0e21e | 6133 | else { |
| 07bc277f | 6134 | maxiters += slen * RX_NPARENS(rx); |
| 080c2dec | 6135 | while (s < strend && --limit) |
| bbce6d69 | 6136 | { |
| 1b6737cc | 6137 | I32 rex_return; |
| 080c2dec | 6138 | PUTBACK; |
| f9f4320a | 6139 | rex_return = CALLREGEXEC(rx, (char*)s, (char*)strend, (char*)orig, 1 , |
| 727b7506 | 6140 | sv, NULL, 0); |
| 080c2dec | 6141 | SPAGAIN; |
| 1b6737cc | 6142 | if (rex_return == 0) |
| 080c2dec | 6143 | break; |
| d9f97599 | 6144 | TAINT_IF(RX_MATCH_TAINTED(rx)); |
| 07bc277f | 6145 | if (RX_MATCH_COPIED(rx) && RX_SUBBEG(rx) != orig) { |
| a0d0e21e LW |
6146 | m = s; |
| 6147 | s = orig; | |
| 07bc277f | 6148 | orig = RX_SUBBEG(rx); |
| a0d0e21e LW |
6149 | s = orig + (m - s); |
| 6150 | strend = s + (strend - m); | |
| 6151 | } | |
| 07bc277f | 6152 | m = RX_OFFS(rx)[0].start + orig; |
| c1a7495a BB |
6153 | |
| 6154 | if (gimme_scalar) { | |
| 6155 | iters++; | |
| 6156 | if (m-s == 0) | |
| 6157 | trailing_empty++; | |
| 6158 | else | |
| 6159 | trailing_empty = 0; | |
| 6160 | } else { | |
| 6161 | dstr = newSVpvn_flags(s, m-s, | |
| 6162 | (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 6163 | XPUSHs(dstr); | |
| 6164 | } | |
| 07bc277f | 6165 | if (RX_NPARENS(rx)) { |
| 1b6737cc | 6166 | I32 i; |
| 07bc277f NC |
6167 | for (i = 1; i <= (I32)RX_NPARENS(rx); i++) { |
| 6168 | s = RX_OFFS(rx)[i].start + orig; | |
| 6169 | m = RX_OFFS(rx)[i].end + orig; | |
| 6de67870 JP |
6170 | |
| 6171 | /* japhy (07/27/01) -- the (m && s) test doesn't catch | |
| 6172 | parens that didn't match -- they should be set to | |
| 6173 | undef, not the empty string */ | |
| c1a7495a BB |
6174 | if (gimme_scalar) { |
| 6175 | iters++; | |
| 6176 | if (m-s == 0) | |
| 6177 | trailing_empty++; | |
| 6178 | else | |
| 6179 | trailing_empty = 0; | |
| 6180 | } else { | |
| 6181 | if (m >= orig && s >= orig) { | |
| 6182 | dstr = newSVpvn_flags(s, m-s, | |
| 6183 | (do_utf8 ? SVf_UTF8 : 0) | |
| 6184 | | make_mortal); | |
| 6185 | } | |
| 6186 | else | |
| 6187 | dstr = &PL_sv_undef; /* undef, not "" */ | |
| 6188 | XPUSHs(dstr); | |
| 748a9306 | 6189 | } |
| c1a7495a | 6190 | |
| a0d0e21e LW |
6191 | } |
| 6192 | } | |
| 07bc277f | 6193 | s = RX_OFFS(rx)[0].end + orig; |
| a0d0e21e | 6194 | } |
| 79072805 | 6195 | } |
| 8ec5e241 | 6196 | |
| c1a7495a BB |
6197 | if (!gimme_scalar) { |
| 6198 | iters = (SP - PL_stack_base) - base; | |
| 6199 | } | |
| a0d0e21e | 6200 | if (iters > maxiters) |
| cea2e8a9 | 6201 | DIE(aTHX_ "Split loop"); |
| 8ec5e241 | 6202 | |
| a0d0e21e LW |
6203 | /* keep field after final delim? */ |
| 6204 | if (s < strend || (iters && origlimit)) { | |
| c1a7495a BB |
6205 | if (!gimme_scalar) { |
| 6206 | const STRLEN l = strend - s; | |
| 6207 | dstr = newSVpvn_flags(s, l, (do_utf8 ? SVf_UTF8 : 0) | make_mortal); | |
| 6208 | XPUSHs(dstr); | |
| 6209 | } | |
| a0d0e21e | 6210 | iters++; |
| 79072805 | 6211 | } |
| a0d0e21e | 6212 | else if (!origlimit) { |
| c1a7495a BB |
6213 | if (gimme_scalar) { |
| 6214 | iters -= trailing_empty; | |
| 6215 | } else { | |
| 6216 | while (iters > 0 && (!TOPs || !SvANY(TOPs) || SvCUR(TOPs) == 0)) { | |
| 6217 | if (TOPs && !make_mortal) | |
| 6218 | sv_2mortal(TOPs); | |
| 6219 | *SP-- = &PL_sv_undef; | |
| 6220 | iters--; | |
| 6221 | } | |
| 89900bd3 | 6222 | } |
| a0d0e21e | 6223 | } |
| 8ec5e241 | 6224 | |
| 8b7059b1 DM |
6225 | PUTBACK; |
| 6226 | LEAVE_SCOPE(oldsave); /* may undo an earlier SWITCHSTACK */ | |
| 6227 | SPAGAIN; | |
| a0d0e21e | 6228 | if (realarray) { |
| 8ec5e241 | 6229 | if (!mg) { |
| 1c0b011c NIS |
6230 | if (SvSMAGICAL(ary)) { |
| 6231 | PUTBACK; | |
| ad64d0ec | 6232 | mg_set(MUTABLE_SV(ary)); |
| 1c0b011c NIS |
6233 | SPAGAIN; |
| 6234 | } | |
| 6235 | if (gimme == G_ARRAY) { | |
| 6236 | EXTEND(SP, iters); | |
| 6237 | Copy(AvARRAY(ary), SP + 1, iters, SV*); | |
| 6238 | SP += iters; | |
| 6239 | RETURN; | |
| 6240 | } | |
| 8ec5e241 | 6241 | } |
| 1c0b011c | 6242 | else { |
| fb73857a | 6243 | PUTBACK; |
| d343c3ef | 6244 | ENTER_with_name("call_PUSH"); |
| 864dbfa3 | 6245 | call_method("PUSH",G_SCALAR|G_DISCARD); |
| d343c3ef | 6246 | LEAVE_with_name("call_PUSH"); |
| fb73857a | 6247 | SPAGAIN; |
| 8ec5e241 | 6248 | if (gimme == G_ARRAY) { |
| 1b6737cc | 6249 | I32 i; |
| 8ec5e241 NIS |
6250 | /* EXTEND should not be needed - we just popped them */ |
| 6251 | EXTEND(SP, iters); | |
| 6252 | for (i=0; i < iters; i++) { | |
| 6253 | SV **svp = av_fetch(ary, i, FALSE); | |
| 3280af22 | 6254 | PUSHs((svp) ? *svp : &PL_sv_undef); |
| 8ec5e241 | 6255 | } |
| 1c0b011c NIS |
6256 | RETURN; |
| 6257 | } | |
| a0d0e21e LW |
6258 | } |
| 6259 | } | |
| 6260 | else { | |
| 6261 | if (gimme == G_ARRAY) | |
| 6262 | RETURN; | |
| 6263 | } | |
| 7f18b612 YST |
6264 | |
| 6265 | GETTARGET; | |
| 6266 | PUSHi(iters); | |
| 6267 | RETURN; | |
| 79072805 | 6268 | } |
| 85e6fe83 | 6269 | |
| c5917253 NC |
6270 | PP(pp_once) |
| 6271 | { | |
| 6272 | dSP; | |
| 6273 | SV *const sv = PAD_SVl(PL_op->op_targ); | |
| 6274 | ||
| 6275 | if (SvPADSTALE(sv)) { | |
| 6276 | /* First time. */ | |
| 6277 | SvPADSTALE_off(sv); | |
| 6278 | RETURNOP(cLOGOP->op_other); | |
| 6279 | } | |
| 6280 | RETURNOP(cLOGOP->op_next); | |
| 6281 | } | |
| 6282 | ||
| c0329465 MB |
6283 | PP(pp_lock) |
| 6284 | { | |
| 97aff369 | 6285 | dVAR; |
| 39644a26 | 6286 | dSP; |
| c0329465 | 6287 | dTOPss; |
| e55aaa0e | 6288 | SV *retsv = sv; |
| 076a2a80 | 6289 | assert(SvTYPE(retsv) != SVt_PVCV); |
| 68795e93 | 6290 | SvLOCK(sv); |
| 076a2a80 | 6291 | if (SvTYPE(retsv) == SVt_PVAV || SvTYPE(retsv) == SVt_PVHV) { |
| e55aaa0e MB |
6292 | retsv = refto(retsv); |
| 6293 | } | |
| 6294 | SETs(retsv); | |
| c0329465 MB |
6295 | RETURN; |
| 6296 | } | |
| a863c7d1 | 6297 | |
| 65bca31a NC |
6298 | |
| 6299 | PP(unimplemented_op) | |
| 6300 | { | |
| 97aff369 | 6301 | dVAR; |
| 361ed549 NC |
6302 | const Optype op_type = PL_op->op_type; |
| 6303 | /* Using OP_NAME() isn't going to be helpful here. Firstly, it doesn't cope | |
| 6304 | with out of range op numbers - it only "special" cases op_custom. | |
| 6305 | Secondly, as the three ops we "panic" on are padmy, mapstart and custom, | |
| 6306 | if we get here for a custom op then that means that the custom op didn't | |
| 6307 | have an implementation. Given that OP_NAME() looks up the custom op | |
| 6308 | by its pp_addr, likely it will return NULL, unless someone (unhelpfully) | |
| 6309 | registers &PL_unimplemented_op as the address of their custom op. | |
| 6310 | NULL doesn't generate a useful error message. "custom" does. */ | |
| 6311 | const char *const name = op_type >= OP_max | |
| 6312 | ? "[out of range]" : PL_op_name[PL_op->op_type]; | |
| 7627e6d0 NC |
6313 | if(OP_IS_SOCKET(op_type)) |
| 6314 | DIE(aTHX_ PL_no_sock_func, name); | |
| 361ed549 | 6315 | DIE(aTHX_ "panic: unimplemented op %s (#%d) called", name, op_type); |
| 65bca31a NC |
6316 | } |
| 6317 | ||
| 867fa1e2 YO |
6318 | PP(pp_boolkeys) |
| 6319 | { | |
| 6320 | dVAR; | |
| 6321 | dSP; | |
| 6322 | HV * const hv = (HV*)POPs; | |
| 6323 | ||
| fd1d9b5c FC |
6324 | if (SvTYPE(hv) != SVt_PVHV) { XPUSHs(&PL_sv_no); RETURN; } |
| 6325 | ||
| 867fa1e2 YO |
6326 | if (SvRMAGICAL(hv)) { |
| 6327 | MAGIC * const mg = mg_find((SV*)hv, PERL_MAGIC_tied); | |
| 6328 | if (mg) { | |
| 6329 | XPUSHs(magic_scalarpack(hv, mg)); | |
| 6330 | RETURN; | |
| 6331 | } | |
| 6332 | } | |
| 6333 | ||
| 6334 | XPUSHs(boolSV(HvKEYS(hv) != 0)); | |
| 6335 | RETURN; | |
| 6336 | } | |
| 6337 | ||
| e609e586 NC |
6338 | /* |
| 6339 | * Local variables: | |
| 6340 | * c-indentation-style: bsd | |
| 6341 | * c-basic-offset: 4 | |
| 6342 | * indent-tabs-mode: t | |
| 6343 | * End: | |
| 6344 | * | |
| 37442d52 RGS |
6345 | * ex: set ts=8 sts=4 sw=4 noet: |
| 6346 | */ |