1 /* handy.h 2 * 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000, 4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others 5 * 6 * You may distribute under the terms of either the GNU General Public 7 * License or the Artistic License, as specified in the README file. 8 * 9 */ 10 11 /* IMPORTANT NOTE: Everything whose name begins with an underscore is for 12 * internal core Perl use only. */ 13 14 #ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */ 15 #define PERL_HANDY_H_ 16 17 #ifndef PERL_CORE 18 # define Null(type) ((type)NULL) 19 20 /* 21 =for apidoc_section $string 22 =for apidoc AmnU||Nullch 23 Null character pointer. (No longer available when C<PERL_CORE> is 24 defined.) 25 26 =for apidoc_section $SV 27 =for apidoc AmnU||Nullsv 28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.) 29 30 =cut 31 32 Below are signatures of functions from config.h which can't easily be gleaned 33 from it, and are very unlikely to change 34 35 =for apidoc_section $signals 36 =for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs 37 =for apidoc Am|void|Siglongjmp|jmp_buf env|int val 38 39 =for apidoc_section $filesystem 40 =for apidoc Am|void *|FILE_ptr|FILE * f 41 =for apidoc Am|Size_t|FILE_cnt|FILE * f 42 =for apidoc Am|void *|FILE_base|FILE * f 43 =for apidoc Am|Size_t|FILE_bufsiz|FILE *f 44 45 =for apidoc_section $string 46 =for apidoc Amu|token|CAT2|token x|token y 47 =for apidoc Amu|string|STRINGIFY|token x 48 49 =for apidoc_section $numeric 50 =for apidoc Am|double|Drand01 51 =for apidoc Am|void|seedDrand01|Rand_seed_t x 52 =for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b 53 54 =cut 55 */ 56 57 # define Nullch Null(char*) 58 # define Nullfp Null(PerlIO*) 59 # define Nullsv Null(SV*) 60 #endif 61 62 #ifdef TRUE 63 #undef TRUE 64 #endif 65 #ifdef FALSE 66 #undef FALSE 67 #endif 68 #define TRUE (1) 69 #define FALSE (0) 70 71 /* 72 =for apidoc_section $SV 73 =for apidoc Am |AV * |MUTABLE_AV |AV * p 74 =for apidoc_item |CV * |MUTABLE_CV |CV * p 75 =for apidoc_item |GV * |MUTABLE_GV |GV * p 76 =for apidoc_item |HV * |MUTABLE_HV |HV * p 77 =for apidoc_item |IO * |MUTABLE_IO |IO * p 78 =for apidoc_item |void *|MUTABLE_PTR|void * p 79 =for apidoc_item |SV * |MUTABLE_SV |SV * p 80 81 The C<MUTABLE_I<*>>() macros cast pointers to the types shown, in such a way 82 (compiler permitting) that casting away const-ness will give a warning; 83 e.g.: 84 85 const SV *sv = ...; 86 AV *av1 = (AV*)sv; <== BAD: the const has been silently 87 cast away 88 AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn 89 90 C<MUTABLE_PTR> is the base macro used to derive new casts. The other 91 already-built-in ones return pointers to what their names indicate. 92 93 =cut 94 95 The brace group version will raise a diagnostic if 'p' is const; the other 96 blindly casts away const. 97 */ 98 #if defined(PERL_USE_GCC_BRACE_GROUPS) 99 # define MUTABLE_PTR(p) ({ void *p_ = (p); p_; }) 100 #else 101 # define MUTABLE_PTR(p) ((void *) (p)) 102 #endif 103 104 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p)) 105 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p)) 106 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p)) 107 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p)) 108 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p)) 109 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p)) 110 111 /* 112 =for apidoc_section $SV 113 =for apidoc Am |AV *|AV_FROM_REF|SV * ref 114 =for apidoc_item |CV *|CV_FROM_REF|SV * ref 115 =for apidoc_item |HV *|HV_FROM_REF|SV * ref 116 117 The C<I<*>V_FROM_REF> macros extract the C<SvRV()> from a given reference SV 118 and return a suitably-cast to pointer to the referenced SV. When running 119 under C<-DDEBUGGING>, assertions are also applied that check that I<ref> is 120 definitely a reference SV that refers to an SV of the right type. 121 122 =cut 123 */ 124 125 #if defined(DEBUGGING) && defined(PERL_USE_GCC_BRACE_GROUPS) 126 # define xV_FROM_REF(XV, ref) \ 127 ({ SV *_ref = ref; \ 128 assert(SvROK(_ref)); \ 129 assert(SvTYPE(SvRV(_ref)) == SVt_PV ## XV); \ 130 (XV *)(SvRV(_ref)); }) 131 #else 132 # define xV_FROM_REF(XV, ref) ((XV *)(SvRV(ref))) 133 #endif 134 135 #define AV_FROM_REF(ref) xV_FROM_REF(AV, ref) 136 #define CV_FROM_REF(ref) xV_FROM_REF(CV, ref) 137 #define HV_FROM_REF(ref) xV_FROM_REF(HV, ref) 138 139 #ifndef __cplusplus 140 # include <stdbool.h> 141 #endif 142 143 /* 144 =for apidoc_section $casting 145 =for apidoc Am|bool|cBOOL|bool expr 146 147 Cast-to-bool. When Perl was able to be compiled on pre-C99 compilers, a 148 C<(bool)> cast didn't necessarily do the right thing, so this macro was 149 created (and made somewhat complicated to work around bugs in old 150 compilers). Now, many years later, and C99 is used, this is no longer 151 required, but is kept for backwards compatibility. 152 153 =cut 154 */ 155 #define cBOOL(cbool) ((bool) (cbool)) 156 157 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any. 158 * XXX Should really be a Configure probe, with HAS__FUNCTION__ 159 * and FUNCTION__ as results. 160 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */ 161 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */ 162 # define FUNCTION__ __func__ 163 # define SAFE_FUNCTION__ __func__ 164 #elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tru64, -c99 not known, only -std1). */ 165 # define FUNCTION__ ("") 166 # define SAFE_FUNCTION__ ("UNKNOWN") 167 #else 168 # define FUNCTION__ __FUNCTION__ /* Common extension. */ 169 # define SAFE_FUNCTION__ __FUNCTION__ /* Common extension. */ 170 #endif 171 172 /* XXX A note on the perl source internal type system. The 173 original intent was that I32 be *exactly* 32 bits. 174 175 Currently, we only guarantee that I32 is *at least* 32 bits. 176 Specifically, if int is 64 bits, then so is I32. (This is the case 177 for the Cray.) This has the advantage of meshing nicely with 178 standard library calls (where we pass an I32 and the library is 179 expecting an int), but the disadvantage that an I32 is not 32 bits. 180 Andy Dougherty August 1996 181 182 There is no guarantee that there is *any* integral type with 183 exactly 32 bits. It is perfectly legal for a system to have 184 sizeof(short) == sizeof(int) == sizeof(long) == 8. 185 186 Similarly, there is no guarantee that I16 and U16 have exactly 16 187 bits. 188 189 For dealing with issues that may arise from various 32/64-bit 190 systems, we will ask Configure to check out 191 192 SHORTSIZE == sizeof(short) 193 INTSIZE == sizeof(int) 194 LONGSIZE == sizeof(long) 195 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG) 196 PTRSIZE == sizeof(void *) 197 DOUBLESIZE == sizeof(double) 198 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE). 199 200 */ 201 202 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */ 203 # include <inttypes.h> 204 # ifdef INT32_MIN_BROKEN 205 # undef INT32_MIN 206 # define INT32_MIN (-2147483647-1) 207 # endif 208 # ifdef INT64_MIN_BROKEN 209 # undef INT64_MIN 210 # define INT64_MIN (-9223372036854775807LL-1) 211 # endif 212 #endif 213 214 typedef I8TYPE I8; 215 typedef U8TYPE U8; 216 typedef I16TYPE I16; 217 typedef U16TYPE U16; 218 typedef I32TYPE I32; 219 typedef U32TYPE U32; 220 221 #ifdef QUADKIND 222 typedef I64TYPE I64; 223 typedef U64TYPE U64; 224 #endif 225 226 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type. 227 Please search CHAR_MAX in perl.h for further details. */ 228 #ifdef UINT8_MAX 229 # define U8_MAX UINT8_MAX 230 #else 231 # define U8_MAX PERL_UCHAR_MAX 232 #endif 233 #ifdef UINT8_MIN 234 # define U8_MIN UINT8_MIN 235 #else 236 # define U8_MIN PERL_UCHAR_MIN 237 #endif 238 239 #ifdef INT16_MAX 240 # define I16_MAX INT16_MAX 241 #else 242 # define I16_MAX PERL_SHORT_MAX 243 #endif 244 #ifdef INT16_MIN 245 # define I16_MIN INT16_MIN 246 #else 247 # define I16_MIN PERL_SHORT_MIN 248 #endif 249 #ifdef UINT16_MAX 250 # define U16_MAX UINT16_MAX 251 #else 252 # define U16_MAX PERL_USHORT_MAX 253 #endif 254 #ifdef UINT16_MIN 255 # define U16_MIN UINT16_MIN 256 #else 257 # define U16_MIN PERL_USHORT_MIN 258 #endif 259 260 #ifdef INT32_MAX 261 # define I32_MAX INT32_MAX 262 #elif LONGSIZE > 4 263 # define I32_MAX PERL_INT_MAX 264 #else 265 # define I32_MAX PERL_LONG_MAX 266 #endif 267 #ifdef INT32_MIN 268 # define I32_MIN INT32_MIN 269 #elif LONGSIZE > 4 270 # define I32_MIN PERL_INT_MIN 271 #else 272 # define I32_MIN PERL_LONG_MIN 273 #endif 274 #ifdef UINT32_MAX 275 # ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */ 276 # define U32_MAX UINT_MAX 277 # else 278 # define U32_MAX 4294967295U 279 # endif 280 #elif LONGSIZE > 4 281 # define U32_MAX PERL_UINT_MAX 282 #else 283 # define U32_MAX PERL_ULONG_MAX 284 #endif 285 #ifdef UINT32_MIN 286 # define U32_MIN UINT32_MIN 287 #elif LONGSIZE > 4 288 # define U32_MIN PERL_UINT_MIN 289 #else 290 # define U32_MIN PERL_ULONG_MIN 291 #endif 292 293 /* 294 =for apidoc_section $integer 295 =for apidoc Ay|| PERL_INT_FAST8_T 296 =for apidoc_item PERL_INT_FAST16_T 297 =for apidoc_item PERL_UINT_FAST8_T 298 =for apidoc_item PERL_UINT_FAST16_T 299 300 These are equivalent to the correspondingly-named C99 typedefs on platforms 301 that have those; they evaluate to C<int> and C<unsigned int> on platforms that 302 don't, so that you can portably take advantage of this C99 feature. 303 304 =cut 305 */ 306 #ifdef I_STDINT 307 typedef int_fast8_t PERL_INT_FAST8_T; 308 typedef uint_fast8_t PERL_UINT_FAST8_T; 309 typedef int_fast16_t PERL_INT_FAST16_T; 310 typedef uint_fast16_t PERL_UINT_FAST16_T; 311 #else 312 typedef int PERL_INT_FAST8_T; 313 typedef unsigned int PERL_UINT_FAST8_T; 314 typedef int PERL_INT_FAST16_T; 315 typedef unsigned int PERL_UINT_FAST16_T; 316 #endif 317 318 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case 319 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits 320 * required to represent any possible unsigned number containing N bits. 321 * TYPE_DIGITS gives the number of decimal digits required to represent any 322 * possible unsigned number of type T. */ 323 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */ 324 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8) 325 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */ 326 327 /* Unused by core; should be deprecated */ 328 #define Ctl(ch) ((ch) & 037) 329 330 #if defined(PERL_CORE) || defined(PERL_EXT) 331 # ifndef MIN 332 # define MIN(a,b) ((a) < (b) ? (a) : (b)) 333 # endif 334 # ifndef MAX 335 # define MAX(a,b) ((a) > (b) ? (a) : (b)) 336 # endif 337 #endif 338 339 /* Returns a boolean as to whether the input unsigned number is a power of 2 340 * (2**0, 2**1, etc). In other words if it has just a single bit set. 341 * If not, subtracting 1 would leave the uppermost bit set, so the & would 342 * yield non-zero */ 343 #if defined(PERL_CORE) || defined(PERL_EXT) 344 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0) 345 #endif 346 347 /* Returns a mask with the lowest n bits set */ 348 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1) 349 350 /* The largest unsigned number that will fit into n bits */ 351 #define nBIT_UMAX(n) nBIT_MASK(n) 352 353 /* 354 =for apidoc_section $directives 355 =for apidoc Am||__ASSERT_|bool expr 356 357 This is a helper macro to avoid preprocessor issues, replaced by nothing 358 unless under DEBUGGING, where it expands to an assert of its argument, 359 followed by a comma (hence the comma operator). If we just used a straight 360 assert(), we would get a comma with nothing before it when not DEBUGGING. 361 362 =cut 363 364 We also use empty definition under Coverity since the __ASSERT_ 365 checks often check for things that Really Cannot Happen, and Coverity 366 detects that and gets all excited. */ 367 368 #if defined(DEBUGGING) && !defined(__COVERITY__) \ 369 && ! defined(PERL_SMALL_MACRO_BUFFER) 370 # define __ASSERT_(statement) assert(statement), 371 #else 372 # define __ASSERT_(statement) 373 #endif 374 375 /* 376 =for apidoc_section $SV 377 378 =for apidoc Ama|SV*|newSVpvs|"literal string" 379 Like C<newSVpvn>, but takes a literal string instead of a 380 string/length pair. 381 382 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags 383 Like C<newSVpvn_flags>, but takes a literal string instead of 384 a string/length pair. 385 386 =for apidoc Ama|SV*|newSVpvs_share|"literal string" 387 Like C<newSVpvn_share>, but takes a literal string instead of 388 a string/length pair and omits the hash parameter. 389 390 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags 391 Like C<sv_catpvn_flags>, but takes a literal string instead 392 of a string/length pair. 393 394 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string" 395 Like C<sv_catpvn_nomg>, but takes a literal string instead of 396 a string/length pair. 397 398 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string" 399 Like C<sv_catpvn>, but takes a literal string instead of a 400 string/length pair. 401 402 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string" 403 Like C<sv_catpvn_mg>, but takes a literal string instead of a 404 string/length pair. 405 406 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string" 407 Like C<sv_setref_pvn>, but takes a literal string instead of 408 a string/length pair. 409 410 =for apidoc_section $string 411 412 =for apidoc Ama|char*|savepvs|"literal string" 413 Like C<savepvn>, but takes a literal string instead of a 414 string/length pair. 415 416 =for apidoc Ama|char*|savesharedpvs|"literal string" 417 A version of C<savepvs()> which allocates the duplicate string in memory 418 which is shared between threads. 419 420 =for apidoc_section $GV 421 422 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create 423 Like C<gv_stashpvn>, but takes a literal string instead of a 424 string/length pair. 425 426 =for apidoc_section $HV 427 428 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval 429 Like C<hv_fetch>, but takes a literal string instead of a 430 string/length pair. 431 =for apidoc_section $lexer 432 433 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags 434 435 Like L</lex_stuff_pvn>, but takes a literal string instead of 436 a string/length pair. 437 438 =cut 439 */ 440 441 #define ASSERT_IS_LITERAL(s) ("" s "") 442 443 /* 444 =for apidoc_section $string 445 446 =for apidoc Amu|pair|STR_WITH_LEN|"literal string" 447 448 Returns two comma separated tokens of the input literal string, and its length. 449 This is convenience macro which helps out in some API calls. 450 Note that it can't be used as an argument to macros or functions that under 451 some configurations might be macros, which means that it requires the full 452 Perl_xxx(aTHX_ ...) form for any API calls where it's used. 453 454 =cut 455 */ 456 457 #define STR_WITH_LEN(s) ASSERT_IS_LITERAL(s), (sizeof(s)-1) 458 459 /* STR_WITH_LEN() shortcuts */ 460 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str)) 461 #define newSVpvs_flags(str,flags) \ 462 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags) 463 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0) 464 #define sv_catpvs_flags(sv, str, flags) \ 465 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags) 466 #define sv_catpvs_nomg(sv, str) \ 467 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0) 468 #define sv_catpvs(sv, str) \ 469 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC) 470 #define sv_catpvs_mg(sv, str) \ 471 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC) 472 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str)) 473 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str)) 474 #define sv_setref_pvs(rv, classname, str) \ 475 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str)) 476 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str)) 477 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str)) 478 #define gv_stashpvs(str, create) \ 479 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create) 480 481 #define gv_fetchpvs(namebeg, flags, sv_type) \ 482 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type) 483 #define gv_fetchpvn gv_fetchpvn_flags 484 485 486 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags) 487 488 #define get_cvs(str, flags) \ 489 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags)) 490 491 /* internal helpers */ 492 /* Transitional */ 493 #ifndef PERL_VERSION_MAJOR 494 # define PERL_VERSION_MAJOR PERL_REVISION 495 #else 496 # undef PERL_REVISION /* We don't want code to be using these */ 497 #endif 498 #ifndef PERL_VERSION_MINOR 499 # define PERL_VERSION_MINOR PERL_VERSION 500 #else 501 # undef PERL_VERSION 502 #endif 503 #ifndef PERL_VERSION_PATCH 504 # define PERL_VERSION_PATCH PERL_SUBVERSION 505 #else 506 # undef PERL_SUBVERSION 507 #endif 508 509 #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \ 510 /* '10*' leaves room for things like alpha, beta, releases */ \ 511 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch)) 512 #define PERL_DECIMAL_VERSION_ \ 513 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \ 514 PERL_VERSION_PATCH) 515 516 /* 517 =for apidoc_section $versioning 518 =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch 519 =for apidoc_item PERL_VERSION_GE 520 =for apidoc_item PERL_VERSION_GT 521 =for apidoc_item PERL_VERSION_LE 522 =for apidoc_item PERL_VERSION_LT 523 =for apidoc_item PERL_VERSION_NE 524 525 Returns whether or not the perl currently being compiled has the specified 526 relationship to the perl given by the parameters. For example, 527 528 #if PERL_VERSION_GT(5,24,2) 529 code that will only be compiled on perls after v5.24.2 530 #else 531 fallback code 532 #endif 533 534 Note that this is usable in making compile-time decisions 535 536 You may use the special value '*' for the final number to mean ALL possible 537 values for it. Thus, 538 539 #if PERL_VERSION_EQ(5,31,'*') 540 541 means all perls in the 5.31 series. And 542 543 #if PERL_VERSION_NE(5,24,'*') 544 545 means all perls EXCEPT 5.24 ones. And 546 547 #if PERL_VERSION_LE(5,9,'*') 548 549 is effectively 550 551 #if PERL_VERSION_LT(5,10,0) 552 553 This means you don't have to think so much when converting from the existing 554 deprecated C<PERL_VERSION> to using this macro: 555 556 #if PERL_VERSION <= 9 557 558 becomes 559 560 #if PERL_VERSION_LE(5,9,'*') 561 562 =cut 563 */ 564 565 /* N.B. These don't work if the patch version is 42 or 92, as those are what 566 * '*' is in ASCII and EBCDIC respectively */ 567 # define PERL_VERSION_EQ(j,n,p) \ 568 (((p) == '*') \ 569 ? ( (j) == PERL_VERSION_MAJOR \ 570 && (n) == PERL_VERSION_MINOR) \ 571 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p))) 572 # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p)) 573 574 # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \ 575 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \ 576 (n), \ 577 (((p) == '*') ? 0 : p))) 578 # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p)) 579 580 # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \ 581 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \ 582 (((p) == '*') ? ((n)+1) : (n)), \ 583 (((p) == '*') ? 0 : p))) 584 # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p)) 585 586 /* 587 =for apidoc_section $string 588 589 =for apidoc Am|bool|strNE|char* s1|char* s2 590 Test two C<NUL>-terminated strings to see if they are different. Returns true 591 or false. 592 593 =for apidoc Am|bool|strEQ|char* s1|char* s2 594 Test two C<NUL>-terminated strings to see if they are equal. Returns true or 595 false. 596 597 =for apidoc Am|bool|strLT|char* s1|char* s2 598 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the 599 second, C<s2>. Returns true or false. 600 601 =for apidoc Am|bool|strLE|char* s1|char* s2 602 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or 603 equal to the second, C<s2>. Returns true or false. 604 605 =for apidoc Am|bool|strGT|char* s1|char* s2 606 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than 607 the second, C<s2>. Returns true or false. 608 609 =for apidoc Am|bool|strGE|char* s1|char* s2 610 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than 611 or equal to the second, C<s2>. Returns true or false. 612 613 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len 614 Test two C<NUL>-terminated strings to see if they are different. The C<len> 615 parameter indicates the number of bytes to compare. Returns true or false. (A 616 wrapper for C<strncmp>). 617 618 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len 619 Test two C<NUL>-terminated strings to see if they are equal. The C<len> 620 parameter indicates the number of bytes to compare. Returns true or false. (A 621 wrapper for C<strncmp>). 622 623 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len 624 Test two buffers (which may contain embedded C<NUL> characters, to see if they 625 are equal. The C<len> parameter indicates the number of bytes to compare. 626 Returns true or false. It is undefined behavior if either of the buffers 627 doesn't contain at least C<len> bytes. 628 629 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2" 630 Like L</memEQ>, but the second string is a literal enclosed in double quotes, 631 C<l1> gives the number of bytes in C<s1>. 632 Returns true or false. 633 634 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len 635 Test two buffers (which may contain embedded C<NUL> characters, to see if they 636 are not equal. The C<len> parameter indicates the number of bytes to compare. 637 Returns true or false. It is undefined behavior if either of the buffers 638 doesn't contain at least C<len> bytes. 639 640 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2" 641 Like L</memNE>, but the second string is a literal enclosed in double quotes, 642 C<l1> gives the number of bytes in C<s1>. 643 Returns true or false. 644 645 =for apidoc Am|bool|memCHRs|"list"|char c 646 Returns the position of the first occurrence of the byte C<c> in the literal 647 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are 648 treated as unsigned char. Thus this macro can be used to determine if C<c> is 649 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c> 650 is C<NUL> (and the set doesn't include C<NUL>). 651 652 =cut 653 654 New macros should use the following conventions for their names (which are 655 based on the underlying C library functions): 656 657 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s? 658 659 Each has two main parameters, string-like operands that are compared 660 against each other, as specified by the macro name. Some macros may 661 additionally have one or potentially even two length parameters. If a length 662 parameter applies to both string parameters, it will be positioned third; 663 otherwise any length parameter immediately follows the string parameter it 664 applies to. 665 666 If the prefix to the name is 'str', the string parameter is a pointer to a C 667 language string. Such a string does not contain embedded NUL bytes; its 668 length may be unknown, but can be calculated by C<strlen()>, since it is 669 terminated by a NUL, which isn't included in its length. 670 671 The optional 'n' following 'str' means that there is a third parameter, 672 giving the maximum number of bytes to look at in each string. Even if both 673 strings are longer than the length parameter, those extra bytes will be 674 unexamined. 675 676 The 's' suffix means that the 2nd byte string parameter is a literal C 677 double-quoted string. Its length will automatically be calculated by the 678 macro, so no length parameter will ever be needed for it. 679 680 If the prefix is 'mem', the string parameters don't have to be C strings; 681 they may contain embedded NUL bytes, do not necessarily have a terminating 682 NUL, and their lengths can be known only through other means, which in 683 practice are additional parameter(s) passed to the function. All 'mem' 684 functions have at least one length parameter. Barring any 'l' or 's' suffix, 685 there is a single length parameter, in position 3, which applies to both 686 string parameters. The 's' suffix means, as described above, that the 2nd 687 string is a literal double-quoted C string (hence its length is calculated by 688 the macro, and the length parameter to the function applies just to the first 689 string parameter, and hence is positioned just after it). An 'l' suffix 690 means that the 2nd string parameter has its own length parameter, and the 691 signature will look like memFOOl(s1, l1, s2, l2). 692 693 BEGIN (and END) are for testing if the 2nd string is an initial (or final) 694 substring of the 1st string. 'P' if present indicates that the substring 695 must be a "proper" one in tha mathematical sense that the first one must be 696 strictly larger than the 2nd. 697 698 */ 699 700 701 #define strNE(s1,s2) (strcmp(s1,s2) != 0) 702 #define strEQ(s1,s2) (strcmp(s1,s2) == 0) 703 #define strLT(s1,s2) (strcmp(s1,s2) < 0) 704 #define strLE(s1,s2) (strcmp(s1,s2) <= 0) 705 #define strGT(s1,s2) (strcmp(s1,s2) > 0) 706 #define strGE(s1,s2) (strcmp(s1,s2) >= 0) 707 708 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0) 709 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0) 710 711 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0) 712 #define memNE(s1,s2,l) (! memEQ(s1,s2,l)) 713 714 /* memEQ and memNE where second comparand is a string constant */ 715 #define memEQs(s1, l, s2) \ 716 (((sizeof(s2)-1) == (l)) && memEQ((s1), ASSERT_IS_LITERAL(s2), (sizeof(s2)-1))) 717 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2)) 718 719 /* Keep these private until we decide it was a good idea */ 720 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX) 721 722 #define strBEGINs(s1,s2) (strncmp(s1,ASSERT_IS_LITERAL(s2), sizeof(s2)-1) == 0) 723 724 #define memBEGINs(s1, l, s2) \ 725 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ 726 && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) 727 #define memBEGINPs(s1, l, s2) \ 728 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \ 729 && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) 730 #define memENDs(s1, l, s2) \ 731 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \ 732 && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) 733 #define memENDPs(s1, l, s2) \ 734 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \ 735 && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1)) 736 #endif /* End of making macros private */ 737 738 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0) 739 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0) 740 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0) 741 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0) 742 743 #define memCHRs(s1,c) ((const char *) memchr(ASSERT_IS_LITERAL(s1) , c, sizeof(s1)-1)) 744 745 /* 746 * Character classes. 747 * 748 * Unfortunately, the introduction of locales means that we 749 * can't trust isupper(), etc. to tell the truth. And when 750 * it comes to /\w+/ with tainting enabled, we *must* be able 751 * to trust our character classes. 752 * 753 * Therefore, the default tests in the text of Perl will be independent of 754 * locale. Any code that wants to depend on the current locale will use the 755 * macros that contain _LC in their names 756 */ 757 758 #ifdef USE_LOCALE_CTYPE 759 # ifndef CTYPE256 760 # define CTYPE256 761 # endif 762 #endif 763 764 /* 765 766 =head1 Character classification 767 This section is about functions (really macros) that classify characters 768 into types, such as punctuation versus alphabetic, etc. Most of these are 769 analogous to regular expression character classes. (See 770 L<perlrecharclass/POSIX Character Classes>.) There are several variants for 771 each class. (Not all macros have all variants; each item below lists the 772 ones valid for it.) None are affected by C<use bytes>, and only the ones 773 with C<LC> in the name are affected by the current locale. 774 775 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value, 776 treating it as a code point, and returns a boolean as to whether or not the 777 character represented by it is (or on non-ASCII platforms, corresponds to) an 778 ASCII character in the named class based on platform, Unicode, and Perl rules. 779 If the input is a number that doesn't fit in an octet, FALSE is returned. 780 781 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function 782 with no suffix C<"_A">. This variant is used to emphasize by its name that 783 only ASCII-range characters can return TRUE. 784 785 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set 786 onto the platform. That is, the code points that are ASCII are unaffected, 787 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated 788 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return 789 true when called with the code point 0xDF, which is a word character in both 790 ASCII and EBCDIC (though it represents different characters in each). 791 If the input is a number that doesn't fit in an octet, FALSE is returned. 792 (Perl's documentation uses a colloquial definition of Latin-1, to include all 793 code points below 256.) 794 795 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for 796 inputs below 256, but if the code point is larger than 255, Unicode rules are 797 used to determine if it is in the character class. For example, 798 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A 799 WITH MACRON in Unicode, and is a word character. 800 801 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>, 802 but are used for UTF-8 encoded strings. The two forms are different names for 803 the same thing. Each call to one of these classifies the first character of 804 the string starting at C<p>. The second parameter, C<e>, points to anywhere in 805 the string beyond the first character, up to one byte past the end of the 806 entire string. Although both variants are identical, the suffix C<_safe> in 807 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>, 808 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in 809 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in 810 some way, the program may croak, or the function may return FALSE, at the 811 discretion of the implementation, and subject to change in future releases. 812 813 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants, 814 but the result is based on the current locale, which is what C<LC> in the name 815 stands for. If Perl can determine that the current locale is a UTF-8 locale, 816 it uses the published Unicode rules; otherwise, it uses the C library function 817 that gives the named classification. For example, C<isDIGIT_LC()> when not in 818 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always 819 returned if the input won't fit into an octet. On some platforms where the C 820 library function is known to be defective, Perl changes its result to follow 821 the POSIX standard's rules. 822 823 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less 824 than 256, but for larger ones it returns the Unicode classification of the code 825 point. 826 827 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like 828 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms 829 are different names for the same thing. Each call to one of these classifies 830 the first character of the string starting at C<p>. The second parameter, 831 C<e>, points to anywhere in the string beyond the first character, up to one 832 byte past the end of the entire string. Although both variants are identical, 833 the suffix C<_safe> in one name emphasizes that it will not attempt to read 834 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this 835 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input 836 character is malformed in some way, the program may croak, or the function may 837 return FALSE, at the discretion of the implementation, and subject to change in 838 future releases. 839 840 =for apidoc Am|bool|isALPHA|UV ch 841 =for apidoc_item ||isALPHA_A|UV ch 842 =for apidoc_item ||isALPHA_LC|UV ch 843 =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end 844 =for apidoc_item ||isALPHA_LC_uvchr|UV ch 845 =for apidoc_item ||isALPHA_L1|UV ch 846 =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end 847 =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end 848 =for apidoc_item ||isALPHA_uvchr|UV ch 849 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>, 850 analogous to C<m/[[:alpha:]]/>. 851 See the L<top of this section|/Character classification> for an explanation of 852 the variants. 853 854 =cut 855 856 Here and below, we add the prototypes of these macros for downstream programs 857 that would be interested in them, such as Devel::PPPort 858 859 =for apidoc Am|bool|isALPHANUMERIC|UV ch 860 =for apidoc_item ||isALPHANUMERIC_A|UV ch 861 =for apidoc_item ||isALPHANUMERIC_LC|UV ch 862 =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end 863 =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch 864 =for apidoc_item ||isALPHANUMERIC_L1|UV ch 865 =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end 866 =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end 867 =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch 868 Returns a boolean indicating whether the specified character is one of 869 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>. 870 See the L<top of this section|/Character classification> for an explanation of 871 the variants. 872 873 =for apidoc Am|bool|isALNUMC|UV ch 874 =for apidoc_item ||isALNUMC_A|UV ch 875 =for apidoc_item ||isALNUMC_LC|UV ch 876 =for apidoc_item ||isALNUMC_LC_uvchr|UV ch 877 =for apidoc_item ||isALNUMC_L1|UV ch 878 These are discouraged, backward compatibility macros for L</C<isALPHANUMERIC>>. 879 That is, each returns a boolean indicating whether the specified character is 880 one of C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>. 881 882 The C<C> suffix in the names was meant to indicate that they correspond to the 883 C language L<C<isalnum(3)>>. 884 885 =for apidoc Am|bool|isASCII|UV ch 886 =for apidoc_item ||isASCII_A|UV ch 887 =for apidoc_item ||isASCII_LC|UV ch 888 =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end 889 =for apidoc_item ||isASCII_LC_uvchr|UV ch 890 =for apidoc_item ||isASCII_L1|UV ch 891 =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end 892 =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end 893 =for apidoc_item ||isASCII_uvchr|UV ch 894 Returns a boolean indicating whether the specified character is one of the 128 895 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>. 896 On non-ASCII platforms, it returns TRUE iff this 897 character corresponds to an ASCII character. Variants C<isASCII_A()> and 898 C<isASCII_L1()> are identical to C<isASCII()>. 899 See the L<top of this section|/Character classification> for an explanation of 900 the variants. 901 Note, however, that some platforms do not have the C library routine 902 C<isascii()>. In these cases, the variants whose names contain C<LC> are the 903 same as the corresponding ones without. 904 905 Also note, that because all ASCII characters are UTF-8 invariant (meaning they 906 have the exact same representation (always a single byte) whether encoded in 907 UTF-8 or not), C<isASCII> will give the correct results when called with any 908 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and 909 C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8. 910 911 =for apidoc Am|bool|isBLANK|UV ch 912 =for apidoc_item ||isBLANK_A|UV ch 913 =for apidoc_item ||isBLANK_LC|UV ch 914 =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end 915 =for apidoc_item ||isBLANK_LC_uvchr|UV ch 916 =for apidoc_item ||isBLANK_L1|UV ch 917 =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end 918 =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end 919 =for apidoc_item ||isBLANK_uvchr|UV ch 920 Returns a boolean indicating whether the specified character is a 921 character considered to be a blank, analogous to C<m/[[:blank:]]/>. 922 See the L<top of this section|/Character classification> for an explanation of 923 the variants. 924 Note, 925 however, that some platforms do not have the C library routine 926 C<isblank()>. In these cases, the variants whose names contain C<LC> are 927 the same as the corresponding ones without. 928 929 =for apidoc Am|bool|isCNTRL|UV ch 930 =for apidoc_item ||isCNTRL_A|UV ch 931 =for apidoc_item ||isCNTRL_LC|UV ch 932 =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end 933 =for apidoc_item ||isCNTRL_LC_uvchr|UV ch 934 =for apidoc_item ||isCNTRL_L1|UV ch 935 =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end 936 =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end 937 =for apidoc_item ||isCNTRL_uvchr|UV ch 938 939 Returns a boolean indicating whether the specified character is a 940 control character, analogous to C<m/[[:cntrl:]]/>. 941 See the L<top of this section|/Character classification> for an explanation of 942 the variants. 943 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant. 944 945 =for apidoc Am|bool|isDIGIT|UV ch 946 =for apidoc_item ||isDIGIT_A|UV ch 947 =for apidoc_item ||isDIGIT_LC|UV ch 948 =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end 949 =for apidoc_item ||isDIGIT_LC_uvchr|UV ch 950 =for apidoc_item ||isDIGIT_L1|UV ch 951 =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end 952 =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end 953 =for apidoc_item ||isDIGIT_uvchr|UV ch 954 955 Returns a boolean indicating whether the specified character is a 956 digit, analogous to C<m/[[:digit:]]/>. 957 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>. 958 See the L<top of this section|/Character classification> for an explanation of 959 the variants. 960 961 =for apidoc Am|bool|isGRAPH|UV ch 962 =for apidoc_item ||isGRAPH_A|UV ch 963 =for apidoc_item ||isGRAPH_LC|UV ch 964 =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end 965 =for apidoc_item ||isGRAPH_LC_uvchr|UV ch 966 =for apidoc_item ||isGRAPH_L1|UV ch 967 =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end 968 =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end 969 =for apidoc_item ||isGRAPH_uvchr|UV ch 970 Returns a boolean indicating whether the specified character is a 971 graphic character, analogous to C<m/[[:graph:]]/>. 972 See the L<top of this section|/Character classification> for an explanation of 973 the variants. 974 975 =for apidoc Am|bool|isLOWER|UV ch 976 =for apidoc_item ||isLOWER_A|UV ch 977 =for apidoc_item ||isLOWER_LC|UV ch 978 =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end 979 =for apidoc_item ||isLOWER_LC_uvchr|UV ch 980 =for apidoc_item ||isLOWER_L1|UV ch 981 =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end 982 =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end 983 =for apidoc_item ||isLOWER_uvchr|UV ch 984 Returns a boolean indicating whether the specified character is a 985 lowercase character, analogous to C<m/[[:lower:]]/>. 986 See the L<top of this section|/Character classification> for an explanation of 987 the variants 988 989 =for apidoc Am|bool|isOCTAL|UV ch 990 =for apidoc_item ||isOCTAL_A|UV ch 991 =for apidoc_item ||isOCTAL_L1|UV ch 992 Returns a boolean indicating whether the specified character is an 993 octal digit, [0-7]. 994 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to 995 C<isOCTAL>. 996 997 =for apidoc Am|bool|isPUNCT|UV ch 998 =for apidoc_item ||isPUNCT_A|UV ch 999 =for apidoc_item ||isPUNCT_LC|UV ch 1000 =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end 1001 =for apidoc_item ||isPUNCT_LC_uvchr|UV ch 1002 =for apidoc_item ||isPUNCT_L1|UV ch 1003 =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end 1004 =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end 1005 =for apidoc_item ||isPUNCT_uvchr|UV ch 1006 Returns a boolean indicating whether the specified character is a 1007 punctuation character, analogous to C<m/[[:punct:]]/>. 1008 Note that the definition of what is punctuation isn't as 1009 straightforward as one might desire. See L<perlrecharclass/POSIX Character 1010 Classes> for details. 1011 See the L<top of this section|/Character classification> for an explanation of 1012 the variants. 1013 1014 =for apidoc Am|bool|isSPACE|UV ch 1015 =for apidoc_item ||isSPACE_A|UV ch 1016 =for apidoc_item ||isSPACE_LC|UV ch 1017 =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end 1018 =for apidoc_item ||isSPACE_LC_uvchr|UV ch 1019 =for apidoc_item ||isSPACE_L1|UV ch 1020 =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end 1021 =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end 1022 =for apidoc_item ||isSPACE_uvchr|UV ch 1023 Returns a boolean indicating whether the specified character is a 1024 whitespace character. This is analogous 1025 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18 1026 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the 1027 locale forms of this macro (the ones with C<LC> in their names) matched 1028 precisely what C<m/[[:space:]]/> does. In those releases, the only difference, 1029 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab. 1030 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.) 1031 See the L<top of this section|/Character classification> for an explanation of 1032 the variants. 1033 1034 =for apidoc Am|bool|isPSXSPC|UV ch 1035 =for apidoc_item ||isPSXSPC_A|UV ch 1036 =for apidoc_item ||isPSXSPC_LC|UV ch 1037 =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end 1038 =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch 1039 =for apidoc_item ||isPSXSPC_L1|UV ch 1040 =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end 1041 =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end 1042 =for apidoc_item ||isPSXSPC_uvchr|UV ch 1043 (short for Posix Space) 1044 Starting in 5.18, this is identical in all its forms to the 1045 corresponding C<isSPACE()> macros. 1046 The locale forms of this macro are identical to their corresponding 1047 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the 1048 non-locale forms differ from their C<isSPACE()> forms only in that the 1049 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do. 1050 Otherwise they are identical. Thus this macro is analogous to what 1051 C<m/[[:space:]]/> matches in a regular expression. 1052 See the L<top of this section|/Character classification> for an explanation of 1053 the variants. 1054 1055 =for apidoc Am|bool|isUPPER|UV ch 1056 =for apidoc_item ||isUPPER_A|UV ch 1057 =for apidoc_item ||isUPPER_LC|UV ch 1058 =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end 1059 =for apidoc_item ||isUPPER_LC_uvchr|UV ch 1060 =for apidoc_item ||isUPPER_L1|UV ch 1061 =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end 1062 =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end 1063 =for apidoc_item ||isUPPER_uvchr|UV ch 1064 Returns a boolean indicating whether the specified character is an 1065 uppercase character, analogous to C<m/[[:upper:]]/>. 1066 See the L<top of this section|/Character classification> for an explanation of 1067 the variants. 1068 1069 =for apidoc Am|bool|isPRINT|UV ch 1070 =for apidoc_item ||isPRINT_A|UV ch 1071 =for apidoc_item ||isPRINT_LC|UV ch 1072 =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end 1073 =for apidoc_item ||isPRINT_LC_uvchr|UV ch 1074 =for apidoc_item ||isPRINT_L1|UV ch 1075 =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end 1076 =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end 1077 =for apidoc_item ||isPRINT_uvchr|UV ch 1078 Returns a boolean indicating whether the specified character is a 1079 printable character, analogous to C<m/[[:print:]]/>. 1080 See the L<top of this section|/Character classification> for an explanation of 1081 the variants. 1082 1083 =for apidoc Am|bool|isWORDCHAR|UV ch 1084 =for apidoc_item ||isWORDCHAR_A|UV ch 1085 =for apidoc_item ||isWORDCHAR_LC|UV ch 1086 =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end 1087 =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch 1088 =for apidoc_item ||isWORDCHAR_L1|UV ch 1089 =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end 1090 =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end 1091 =for apidoc_item ||isWORDCHAR_uvchr|UV ch 1092 Returns a boolean indicating whether the specified character is a character 1093 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match 1094 in a regular expression. A word character is an alphabetic character, a 1095 decimal digit, a connecting punctuation character (such as an underscore), or 1096 a "mark" character that attaches to one of those (like some sort of accent). 1097 1098 See the L<top of this section|/Character classification> for an explanation of 1099 the variants. 1100 1101 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>, 1102 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and 1103 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally 1104 include the platform's native underscore. 1105 1106 =for apidoc Am|bool|isALNUM |UV ch 1107 =for apidoc_item ||isALNUM_A |UV ch 1108 =for apidoc_item ||isALNUM_LC |UV ch 1109 =for apidoc_item ||isALNUM_LC_uvchr|UV ch 1110 These are each a synonym for their respectively named L</C<isWORDCHAR>> 1111 variant. 1112 1113 They are provided for backward compatibility, even though a word character 1114 includes more than the standard C language meaning of alphanumeric. 1115 To get the C language definition, use the corresponding L</C<isALPHANUMERIC>> 1116 variant. 1117 1118 =for apidoc Am|bool|isXDIGIT|UV ch 1119 =for apidoc_item ||isXDIGIT_A|UV ch 1120 =for apidoc_item ||isXDIGIT_LC|UV ch 1121 =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end 1122 =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch 1123 =for apidoc_item ||isXDIGIT_L1|UV ch 1124 =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end 1125 =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end 1126 =for apidoc_item ||isXDIGIT_uvchr|UV ch 1127 Returns a boolean indicating whether the specified character is a hexadecimal 1128 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()> 1129 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>. 1130 See the L<top of this section|/Character classification> for an explanation of 1131 the variants. 1132 1133 =for apidoc Am|bool|isIDFIRST|UV ch 1134 =for apidoc_item ||isIDFIRST_A|UV ch 1135 =for apidoc_item ||isIDFIRST_LC|UV ch 1136 =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end 1137 =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch 1138 =for apidoc_item ||isIDFIRST_L1|UV ch 1139 =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end 1140 =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end 1141 =for apidoc_item ||isIDFIRST_uvchr|UV ch 1142 Returns a boolean indicating whether the specified character can be the first 1143 character of an identifier. This is very close to, but not quite the same as 1144 the official Unicode property C<XID_Start>. The difference is that this 1145 returns true only if the input character also matches L</isWORDCHAR>. 1146 See the L<top of this section|/Character classification> for an explanation of 1147 the variants. 1148 1149 =for apidoc Am|bool|isIDCONT|UV ch 1150 =for apidoc_item ||isIDCONT_A|UV ch 1151 =for apidoc_item ||isIDCONT_LC|UV ch 1152 =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end 1153 =for apidoc_item ||isIDCONT_LC_uvchr|UV ch 1154 =for apidoc_item ||isIDCONT_L1|UV ch 1155 =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end 1156 =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end 1157 =for apidoc_item ||isIDCONT_uvchr|UV ch 1158 Returns a boolean indicating whether the specified character can be the 1159 second or succeeding character of an identifier. This is very close to, but 1160 not quite the same as the official Unicode property C<XID_Continue>. The 1161 difference is that this returns true only if the input character also matches 1162 L</isWORDCHAR>. See the L<top of this section|/Character classification> for 1163 an explanation of the variants. 1164 1165 =for apidoc_section $numeric 1166 1167 =for apidoc Am|U8|READ_XDIGIT|char str* 1168 Returns the value of an ASCII-range hex digit and advances the string pointer. 1169 Behaviour is only well defined when isXDIGIT(*str) is true. 1170 1171 =head1 Character case changing 1172 Perl uses "full" Unicode case mappings. This means that converting a single 1173 character to another case may result in a sequence of more than one character. 1174 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two 1175 character sequence C<SS>. This presents some complications The lowercase of 1176 all characters in the range 0..255 is a single character, and thus 1177 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't 1178 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has 1179 an API that does allow every possible legal result to be returned.) Likewise 1180 no other function that is crippled by not being able to give the correct 1181 results for the full range of possible inputs has been implemented here. 1182 1183 =for apidoc Am|UV|toUPPER|UV cp 1184 =for apidoc_item |UV|toUPPER_A|UV cp 1185 =for apidoc_item |UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp 1186 =for apidoc_item |UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp 1187 =for apidoc_item |UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp 1188 1189 These all return the uppercase of a character. The differences are what domain 1190 they operate on, and whether the input is specified as a code point (those 1191 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter 1192 case, the code point to use is the first one in the buffer of UTF-8 encoded 1193 code points, delineated by the arguments S<C<p .. e - 1>>. 1194 1195 C<toUPPER> and C<toUPPER_A> are synonyms of each other. They return the 1196 uppercase of any lowercase ASCII-range code point. All other inputs are 1197 returned unchanged. Since these are macros, the input type may be any integral 1198 one, and the output will occupy the same number of bits as the input. 1199 1200 There is no C<toUPPER_L1> nor C<toUPPER_LATIN1> as the uppercase of some code 1201 points in the 0..255 range is above that range or consists of multiple 1202 characters. Instead use C<toUPPER_uvchr>. 1203 1204 C<toUPPER_uvchr> returns the uppercase of any Unicode code point. The return 1205 value is identical to that of C<toUPPER_A> for input code points in the ASCII 1206 range. The uppercase of the vast majority of Unicode code points is the same 1207 as the code point itself. For these, and for code points above the legal 1208 Unicode maximum, this returns the input code point unchanged. It additionally 1209 stores the UTF-8 of the result into the buffer beginning at C<s>, and its 1210 length in bytes into C<*lenp>. The caller must have made C<s> large enough to 1211 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. 1212 1213 NOTE: the uppercase of a code point may be more than one code point. The 1214 return value of this function is only the first of these. The entire uppercase 1215 is returned in C<s>. To determine if the result is more than a single code 1216 point, you can do something like this: 1217 1218 uc = toUPPER_uvchr(cp, s, &len); 1219 if (len > UTF8SKIP(s)) { is multiple code points } 1220 else { is a single code point } 1221 1222 C<toUPPER_utf8> and C<toUPPER_utf8_safe> are synonyms of each other. The only 1223 difference between these and C<toUPPER_uvchr> is that the source for these is 1224 encoded in UTF-8, instead of being a code point. It is passed as a buffer 1225 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> 1226 buffer may certainly contain more than one code point; but only the first one 1227 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is 1228 malformed in some way, the program may croak, or the function may return the 1229 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to 1230 change in future releases. 1231 1232 =for apidoc Am|UV|toFOLD|UV cp 1233 =for apidoc_item |UV|toFOLD_A|UV cp 1234 =for apidoc_item |UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp 1235 =for apidoc_item |UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp 1236 =for apidoc_item |UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp 1237 1238 These all return the foldcase of a character. "foldcase" is an internal case 1239 for C</i> pattern matching. If the foldcase of character A and the foldcase of 1240 character B are the same, they match caselessly; otherwise they don't. 1241 1242 The differences in the forms are what domain they operate on, and whether the 1243 input is specified as a code point (those forms with a C<cp> parameter) or as a 1244 UTF-8 string (the others). In the latter case, the code point to use is the 1245 first one in the buffer of UTF-8 encoded code points, delineated by the 1246 arguments S<C<p .. e - 1>>. 1247 1248 C<toFOLD> and C<toFOLD_A> are synonyms of each other. They return the 1249 foldcase of any ASCII-range code point. In this range, the foldcase is 1250 identical to the lowercase. All other inputs are returned unchanged. Since 1251 these are macros, the input type may be any integral one, and the output will 1252 occupy the same number of bits as the input. 1253 1254 There is no C<toFOLD_L1> nor C<toFOLD_LATIN1> as the foldcase of some code 1255 points in the 0..255 range is above that range or consists of multiple 1256 characters. Instead use C<toFOLD_uvchr>. 1257 1258 C<toFOLD_uvchr> returns the foldcase of any Unicode code point. The return 1259 value is identical to that of C<toFOLD_A> for input code points in the ASCII 1260 range. The foldcase of the vast majority of Unicode code points is the same 1261 as the code point itself. For these, and for code points above the legal 1262 Unicode maximum, this returns the input code point unchanged. It additionally 1263 stores the UTF-8 of the result into the buffer beginning at C<s>, and its 1264 length in bytes into C<*lenp>. The caller must have made C<s> large enough to 1265 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. 1266 1267 NOTE: the foldcase of a code point may be more than one code point. The 1268 return value of this function is only the first of these. The entire foldcase 1269 is returned in C<s>. To determine if the result is more than a single code 1270 point, you can do something like this: 1271 1272 uc = toFOLD_uvchr(cp, s, &len); 1273 if (len > UTF8SKIP(s)) { is multiple code points } 1274 else { is a single code point } 1275 1276 C<toFOLD_utf8> and C<toFOLD_utf8_safe> are synonyms of each other. The only 1277 difference between these and C<toFOLD_uvchr> is that the source for these is 1278 encoded in UTF-8, instead of being a code point. It is passed as a buffer 1279 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> 1280 buffer may certainly contain more than one code point; but only the first one 1281 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is 1282 malformed in some way, the program may croak, or the function may return the 1283 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to 1284 change in future releases. 1285 1286 =for apidoc Am|UV|toLOWER|UV cp 1287 =for apidoc_item |UV|toLOWER_A|UV cp 1288 =for apidoc_item |UV|toLOWER_LATIN1|UV cp 1289 =for apidoc_item |UV|toLOWER_LC|UV cp 1290 =for apidoc_item |UV|toLOWER_L1|UV cp 1291 =for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp 1292 =for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp 1293 =for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp 1294 1295 These all return the lowercase of a character. The differences are what domain 1296 they operate on, and whether the input is specified as a code point (those 1297 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter 1298 case, the code point to use is the first one in the buffer of UTF-8 encoded 1299 code points, delineated by the arguments S<C<p .. e - 1>>. 1300 1301 C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the 1302 lowercase of any uppercase ASCII-range code point. All other inputs are 1303 returned unchanged. Since these are macros, the input type may be any integral 1304 one, and the output will occupy the same number of bits as the input. 1305 1306 C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave 1307 identically as C<toLOWER> for ASCII-range input. But additionally will return 1308 the lowercase of any uppercase code point in the entire 0..255 range, assuming 1309 a Latin-1 encoding (or the EBCDIC equivalent on such platforms). 1310 1311 C<toLOWER_LC> returns the lowercase of the input code point according to the 1312 rules of the current POSIX locale. Input code points outside the range 0..255 1313 are returned unchanged. 1314 1315 C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return 1316 value is identical to that of C<toLOWER_L1> for input code points in the 0..255 1317 range. The lowercase of the vast majority of Unicode code points is the same 1318 as the code point itself. For these, and for code points above the legal 1319 Unicode maximum, this returns the input code point unchanged. It additionally 1320 stores the UTF-8 of the result into the buffer beginning at C<s>, and its 1321 length in bytes into C<*lenp>. The caller must have made C<s> large enough to 1322 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. 1323 1324 NOTE: the lowercase of a code point may be more than one code point. The 1325 return value of this function is only the first of these. The entire lowercase 1326 is returned in C<s>. To determine if the result is more than a single code 1327 point, you can do something like this: 1328 1329 uc = toLOWER_uvchr(cp, s, &len); 1330 if (len > UTF8SKIP(s)) { is multiple code points } 1331 else { is a single code point } 1332 1333 C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only 1334 difference between these and C<toLOWER_uvchr> is that the source for these is 1335 encoded in UTF-8, instead of being a code point. It is passed as a buffer 1336 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> 1337 buffer may certainly contain more than one code point; but only the first one 1338 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is 1339 malformed in some way, the program may croak, or the function may return the 1340 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to 1341 change in future releases. 1342 1343 =for apidoc Am|UV|toTITLE|UV cp 1344 =for apidoc_item |UV|toTITLE_A|UV cp 1345 =for apidoc_item |UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp 1346 =for apidoc_item |UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp 1347 =for apidoc_item |UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp 1348 1349 These all return the titlecase of a character. The differences are what domain 1350 they operate on, and whether the input is specified as a code point (those 1351 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter 1352 case, the code point to use is the first one in the buffer of UTF-8 encoded 1353 code points, delineated by the arguments S<C<p .. e - 1>>. 1354 1355 C<toTITLE> and C<toTITLE_A> are synonyms of each other. They return the 1356 titlecase of any lowercase ASCII-range code point. In this range, the 1357 titlecase is identical to the uppercase. All other inputs are returned 1358 unchanged. Since these are macros, the input type may be any integral one, and 1359 the output will occupy the same number of bits as the input. 1360 1361 There is no C<toTITLE_L1> nor C<toTITLE_LATIN1> as the titlecase of some code 1362 points in the 0..255 range is above that range or consists of multiple 1363 characters. Instead use C<toTITLE_uvchr>. 1364 1365 C<toTITLE_uvchr> returns the titlecase of any Unicode code point. The return 1366 value is identical to that of C<toTITLE_A> for input code points in the ASCII 1367 range. The titlecase of the vast majority of Unicode code points is the same 1368 as the code point itself. For these, and for code points above the legal 1369 Unicode maximum, this returns the input code point unchanged. It additionally 1370 stores the UTF-8 of the result into the buffer beginning at C<s>, and its 1371 length in bytes into C<*lenp>. The caller must have made C<s> large enough to 1372 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow. 1373 1374 NOTE: the titlecase of a code point may be more than one code point. The 1375 return value of this function is only the first of these. The entire titlecase 1376 is returned in C<s>. To determine if the result is more than a single code 1377 point, you can do something like this: 1378 1379 uc = toTITLE_uvchr(cp, s, &len); 1380 if (len > UTF8SKIP(s)) { is multiple code points } 1381 else { is a single code point } 1382 1383 C<toTITLE_utf8> and C<toTITLE_utf8_safe> are synonyms of each other. The only 1384 difference between these and C<toTITLE_uvchr> is that the source for these is 1385 encoded in UTF-8, instead of being a code point. It is passed as a buffer 1386 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p> 1387 buffer may certainly contain more than one code point; but only the first one 1388 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is 1389 malformed in some way, the program may croak, or the function may return the 1390 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to 1391 change in future releases. 1392 1393 =cut 1394 1395 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names 1396 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change, 1397 and aren't general purpose as they don't work on U+DF, and assert against that. 1398 and isCASED_LC, as it really is more of an internal thing. 1399 1400 Note that these macros are repeated in Devel::PPPort, so should also be 1401 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc 1402 1403 */ 1404 1405 /* 1406 void below because that's the best fit, and works for Devel::PPPort 1407 =for apidoc_section $integer 1408 =for apidoc AyT||WIDEST_UTYPE 1409 1410 Yields the widest unsigned integer type on the platform, currently either 1411 C<U32> or C<U64>. This can be used in declarations such as 1412 1413 WIDEST_UTYPE my_uv; 1414 1415 or casts 1416 1417 my_uv = (WIDEST_UTYPE) val; 1418 1419 =cut 1420 1421 */ 1422 #define WIDEST_UTYPE PERL_UINTMAX_T 1423 1424 /* Where there could be some confusion, use this as a static assert in macros 1425 * to make sure that a parameter isn't a pointer. But some compilers can't 1426 * handle this. The only one known so far that doesn't is gcc 3.3.6; the check 1427 * below isn't thorough for such an old compiler, so may have to be revised if 1428 * experience so dictates. */ 1429 #if ! PERL_IS_GCC || PERL_GCC_VERSION_GT(3,3,6) 1430 # define ASSERT_NOT_PTR(x) ((x) | 0) 1431 #else 1432 # define ASSERT_NOT_PTR(x) (x) 1433 #endif 1434 1435 /* Likewise, this is effectively a static assert to be used to guarantee the 1436 * parameter is a pointer 1437 * 1438 * NOT suitable for void* 1439 */ 1440 #define ASSERT_IS_PTR(x) (__ASSERT_(sizeof(*(x))) (x)) 1441 1442 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in 1443 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no 1444 * bits of information are lost even on a 64-bit machine, but to get the 1445 * compiler to optimize it out if possible. This is because Configure makes 1446 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the 1447 * sizeof() guarantees that this evaluates to a constant true at compile time. 1448 * 1449 * For Coverity, be always true, because otherwise Coverity thinks 1450 * it finds several expressions that are always true, independent 1451 * of operands. Well, they are, but that is kind of the point. 1452 */ 1453 #ifndef __COVERITY__ 1454 /* The '| 0' part in ASSERT_NOT_PTR ensures a compiler error if c is not 1455 * integer (like e.g., a pointer) */ 1456 # define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \ 1457 || (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) >> 8) == 0) 1458 #else 1459 # define FITS_IN_8_BITS(c) (1) 1460 #endif 1461 1462 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative 1463 * Written this way so that after optimization, only one conditional test is 1464 * needed. (The NV casts stop any warnings about comparison always being true 1465 * if called with an unsigned. The cast preserves the sign, which is all we 1466 * care about.) */ 1467 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \ 1468 __ASSERT_((NV) (n) >= 0) \ 1469 withinCOUNT_KNOWN_VALID_((c), (l), (n))) 1470 1471 /* For internal use only, this can be used in places where it is known that the 1472 * parameters to withinCOUNT() are valid, to avoid the asserts. For example, 1473 * inRANGE() below, calls this several times, but does all the necessary 1474 * asserts itself, once. The reason that this is necessary is that the 1475 * duplicate asserts were exceeding the internal limits of some compilers */ 1476 #define withinCOUNT_KNOWN_VALID_(c, l, n) \ 1477 ((((WIDEST_UTYPE) (c)) - ASSERT_NOT_PTR(l)) \ 1478 <= ((WIDEST_UTYPE) ASSERT_NOT_PTR(n))) 1479 1480 /* Returns true if c is in the range l..u, where 'l' is non-negative 1481 * Written this way so that after optimization, only one conditional test is 1482 * needed. */ 1483 #define inRANGE(c, l, u) (__ASSERT_((NV) (l) >= 0) __ASSERT_((u) >= (l)) \ 1484 ( (sizeof(c) == sizeof(U8)) ? inRANGE_helper_(U8, (c), (l), ((u))) \ 1485 : (sizeof(c) == sizeof(U16)) ? inRANGE_helper_(U16,(c), (l), ((u))) \ 1486 : (sizeof(c) == sizeof(U32)) ? inRANGE_helper_(U32,(c), (l), ((u))) \ 1487 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \ 1488 inRANGE_helper_(WIDEST_UTYPE,(c), (l), ((u)))))) 1489 1490 /* For internal use, this is used by machine-generated code which generates 1491 * known valid calls, with a known sizeof(). This avoids the extra code and 1492 * asserts that were exceeding internal limits of some compilers. */ 1493 #define inRANGE_helper_(cast, c, l, u) \ 1494 withinCOUNT_KNOWN_VALID_(((cast) (c)), (l), ((u) - (l))) 1495 1496 #ifdef EBCDIC 1497 # ifndef _ALL_SOURCE 1498 /* The native libc isascii() et.al. functions return the wrong results 1499 * on at least z/OS unless this is defined. */ 1500 # error _ALL_SOURCE should probably be defined 1501 # endif 1502 #else 1503 /* There is a simple definition of ASCII for ASCII platforms. But the 1504 * EBCDIC one isn't so simple, so is defined using table look-up like the 1505 * other macros below. 1506 * 1507 * The cast here is used instead of '(c) >= 0', because some compilers emit 1508 * a warning that that test is always true when the parameter is an 1509 * unsigned type. khw supposes that it could be written as 1510 * && ((c) == '\0' || (c) > 0) 1511 * to avoid the message, but the cast will likely avoid extra branches even 1512 * with stupid compilers. */ 1513 # define isASCII(c) (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) < 128) 1514 #endif 1515 1516 /* Take the eight possible bit patterns of the lower 3 bits and you get the 1517 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits 1518 * can be ignored. If the rest match '0', we have an octal */ 1519 #define isOCTAL_A(c) ((((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) & ~7) == '0') 1520 1521 #ifdef H_PERL /* If have access to perl.h, lookup in its table */ 1522 1523 /* Character class numbers. For internal core Perl use only. The ones less 1524 * than 32 are used in PL_charclass[] and the ones up through the one that 1525 * corresponds to <HIGHEST_REGCOMP_DOT_H_SYNC_> are used by regcomp.h and 1526 * related files. PL_charclass ones use names used in l1_char_class_tab.h but 1527 * their actual definitions are here. If that file has a name not used here, 1528 * it won't compile. 1529 * 1530 * The first group of these is ordered in what I (khw) estimate to be the 1531 * frequency of their use. This gives a slight edge to exiting a loop earlier 1532 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a 1533 * real Posix character class, and some (small) inefficiencies in regular 1534 * expression handling would be introduced by putting it in the middle of those 1535 * that are. Also, cntrl and ascii come after the others as it may be useful 1536 * to group these which have no members that match above Latin1, (or above 1537 * ASCII in the latter case) */ 1538 1539 # define CC_WORDCHAR_ 0 /* \w and [:word:] */ 1540 # define CC_DIGIT_ 1 /* \d and [:digit:] */ 1541 # define CC_ALPHA_ 2 /* [:alpha:] */ 1542 # define CC_LOWER_ 3 /* [:lower:] */ 1543 # define CC_UPPER_ 4 /* [:upper:] */ 1544 # define CC_PUNCT_ 5 /* [:punct:] */ 1545 # define CC_PRINT_ 6 /* [:print:] */ 1546 # define CC_ALPHANUMERIC_ 7 /* [:alnum:] */ 1547 # define CC_GRAPH_ 8 /* [:graph:] */ 1548 # define CC_CASED_ 9 /* [:lower:] or [:upper:] under /i */ 1549 # define CC_SPACE_ 10 /* \s, [:space:] */ 1550 # define CC_BLANK_ 11 /* [:blank:] */ 1551 # define CC_XDIGIT_ 12 /* [:xdigit:] */ 1552 # define CC_CNTRL_ 13 /* [:cntrl:] */ 1553 # define CC_ASCII_ 14 /* [:ascii:] */ 1554 # define CC_VERTSPACE_ 15 /* \v */ 1555 1556 # define HIGHEST_REGCOMP_DOT_H_SYNC_ CC_VERTSPACE_ 1557 1558 /* The members of the third group below do not need to be coordinated with data 1559 * structures in regcomp.[ch] and regexec.c. */ 1560 # define CC_IDFIRST_ 16 1561 # define CC_CHARNAME_CONT_ 17 1562 # define CC_NONLATIN1_FOLD_ 18 1563 # define CC_NONLATIN1_SIMPLE_FOLD_ 19 1564 # define CC_QUOTEMETA_ 20 1565 # define CC_NON_FINAL_FOLD_ 21 1566 # define CC_IS_IN_SOME_FOLD_ 22 1567 # define CC_BINDIGIT_ 23 1568 # define CC_OCTDIGIT_ 24 1569 # define CC_MNEMONIC_CNTRL_ 25 1570 1571 /* Unused: 26-31 1572 * If more bits are needed, one could add a second word for non-64bit 1573 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd 1574 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it 1575 * is used only for optimization (as of this writing), and differs in the 1576 * Latin1 range from the ALPHA bit only in two relatively unimportant 1577 * characters: the masculine and feminine ordinal indicators, so removing it 1578 * would just cause /i regexes which match them to run less efficiently. 1579 * Similarly the EBCDIC-only bits are used just for speed, and could be 1580 * replaced by other means */ 1581 1582 #if defined(PERL_CORE) || defined(PERL_EXT) 1583 /* An enum version of the character class numbers, to help compilers 1584 * optimize */ 1585 typedef enum { 1586 CC_ENUM_ALPHA_ = CC_ALPHA_, 1587 CC_ENUM_ALPHANUMERIC_ = CC_ALPHANUMERIC_, 1588 CC_ENUM_ASCII_ = CC_ASCII_, 1589 CC_ENUM_BLANK_ = CC_BLANK_, 1590 CC_ENUM_CASED_ = CC_CASED_, 1591 CC_ENUM_CNTRL_ = CC_CNTRL_, 1592 CC_ENUM_DIGIT_ = CC_DIGIT_, 1593 CC_ENUM_GRAPH_ = CC_GRAPH_, 1594 CC_ENUM_LOWER_ = CC_LOWER_, 1595 CC_ENUM_PRINT_ = CC_PRINT_, 1596 CC_ENUM_PUNCT_ = CC_PUNCT_, 1597 CC_ENUM_SPACE_ = CC_SPACE_, 1598 CC_ENUM_UPPER_ = CC_UPPER_, 1599 CC_ENUM_VERTSPACE_ = CC_VERTSPACE_, 1600 CC_ENUM_WORDCHAR_ = CC_WORDCHAR_, 1601 CC_ENUM_XDIGIT_ = CC_XDIGIT_ 1602 } char_class_number_; 1603 #endif 1604 1605 #define POSIX_CC_COUNT (HIGHEST_REGCOMP_DOT_H_SYNC_ + 1) 1606 1607 START_EXTERN_C 1608 # ifdef DOINIT 1609 EXTCONST U32 PL_charclass[] = { 1610 # include "l1_char_class_tab.h" 1611 }; 1612 1613 # else /* ! DOINIT */ 1614 EXTCONST U32 PL_charclass[]; 1615 # endif 1616 END_EXTERN_C 1617 1618 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */ 1619 # define CC_mask_(classnum) (1U << (classnum)) 1620 1621 /* For internal core Perl use only: the base macro for defining macros like 1622 * isALPHA */ 1623 # define generic_isCC_(c, classnum) cBOOL(FITS_IN_8_BITS(c) \ 1624 && (PL_charclass[(U8) (c)] & CC_mask_(classnum))) 1625 1626 /* The mask for the _A versions of the macros; it just adds in the bit for 1627 * ASCII. */ 1628 # define CC_mask_A_(classnum) (CC_mask_(classnum) | CC_mask_(CC_ASCII_)) 1629 1630 /* For internal core Perl use only: the base macro for defining macros like 1631 * isALPHA_A. The foo_A version makes sure that both the desired bit and 1632 * the ASCII bit are present */ 1633 # define generic_isCC_A_(c, classnum) (FITS_IN_8_BITS(c) \ 1634 && ((PL_charclass[(U8) (c)] & CC_mask_A_(classnum)) \ 1635 == CC_mask_A_(classnum))) 1636 1637 /* On ASCII platforms certain classes form a single range. It's faster to 1638 * special case these. isDIGIT is a single range on all platforms */ 1639 # ifdef EBCDIC 1640 # define isALPHA_A(c) generic_isCC_A_(c, CC_ALPHA_) 1641 # define isGRAPH_A(c) generic_isCC_A_(c, CC_GRAPH_) 1642 # define isLOWER_A(c) generic_isCC_A_(c, CC_LOWER_) 1643 # define isPRINT_A(c) generic_isCC_A_(c, CC_PRINT_) 1644 # define isUPPER_A(c) generic_isCC_A_(c, CC_UPPER_) 1645 # else 1646 /* By folding the upper and lowercase, we can use a single range */ 1647 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z') 1648 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e) 1649 # define isLOWER_A(c) inRANGE(c, 'a', 'z') 1650 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e) 1651 # define isUPPER_A(c) inRANGE(c, 'A', 'Z') 1652 # endif 1653 # define isALPHANUMERIC_A(c) generic_isCC_A_(c, CC_ALPHANUMERIC_) 1654 # define isBLANK_A(c) generic_isCC_A_(c, CC_BLANK_) 1655 # define isCNTRL_A(c) generic_isCC_A_(c, CC_CNTRL_) 1656 # define isDIGIT_A(c) inRANGE(c, '0', '9') 1657 # define isPUNCT_A(c) generic_isCC_A_(c, CC_PUNCT_) 1658 # define isSPACE_A(c) generic_isCC_A_(c, CC_SPACE_) 1659 # define isWORDCHAR_A(c) generic_isCC_A_(c, CC_WORDCHAR_) 1660 # define isXDIGIT_A(c) generic_isCC_(c, CC_XDIGIT_) /* No non-ASCII xdigits */ 1661 # define isIDFIRST_A(c) generic_isCC_A_(c, CC_IDFIRST_) 1662 # define isALPHA_L1(c) generic_isCC_(c, CC_ALPHA_) 1663 # define isALPHANUMERIC_L1(c) generic_isCC_(c, CC_ALPHANUMERIC_) 1664 # define isBLANK_L1(c) generic_isCC_(c, CC_BLANK_) 1665 1666 /* continuation character for legal NAME in \N{NAME} */ 1667 # define isCHARNAME_CONT(c) generic_isCC_(c, CC_CHARNAME_CONT_) 1668 1669 # define isCNTRL_L1(c) generic_isCC_(c, CC_CNTRL_) 1670 # define isGRAPH_L1(c) generic_isCC_(c, CC_GRAPH_) 1671 # define isLOWER_L1(c) generic_isCC_(c, CC_LOWER_) 1672 # define isPRINT_L1(c) generic_isCC_(c, CC_PRINT_) 1673 # define isPSXSPC_L1(c) isSPACE_L1(c) 1674 # define isPUNCT_L1(c) generic_isCC_(c, CC_PUNCT_) 1675 # define isSPACE_L1(c) generic_isCC_(c, CC_SPACE_) 1676 # define isUPPER_L1(c) generic_isCC_(c, CC_UPPER_) 1677 # define isWORDCHAR_L1(c) generic_isCC_(c, CC_WORDCHAR_) 1678 # define isIDFIRST_L1(c) generic_isCC_(c, CC_IDFIRST_) 1679 1680 # ifdef EBCDIC 1681 # define isASCII(c) generic_isCC_(c, CC_ASCII_) 1682 # endif 1683 1684 /* Participates in a single-character fold with a character above 255 */ 1685 # if defined(PERL_IN_REGCOMP_ANY) || defined(PERL_IN_REGEXEC_C) 1686 # define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(c) \ 1687 (( ! cBOOL(FITS_IN_8_BITS(c))) \ 1688 || (PL_charclass[(U8) (c)] & CC_mask_(CC_NONLATIN1_SIMPLE_FOLD_))) 1689 1690 # define IS_NON_FINAL_FOLD(c) generic_isCC_(c, CC_NON_FINAL_FOLD_) 1691 # define IS_IN_SOME_FOLD_L1(c) generic_isCC_(c, CC_IS_IN_SOME_FOLD_) 1692 # endif 1693 1694 /* Like the above, but also can be part of a multi-char fold */ 1695 # define HAS_NONLATIN1_FOLD_CLOSURE(c) \ 1696 ( (! cBOOL(FITS_IN_8_BITS(c))) \ 1697 || (PL_charclass[(U8) (c)] & CC_mask_(CC_NONLATIN1_FOLD_))) 1698 1699 # define _isQUOTEMETA(c) generic_isCC_(c, CC_QUOTEMETA_) 1700 1701 /* is c a control character for which we have a mnemonic? */ 1702 # if defined(PERL_CORE) || defined(PERL_EXT) 1703 # define isMNEMONIC_CNTRL(c) generic_isCC_(c, CC_MNEMONIC_CNTRL_) 1704 # endif 1705 #else /* else we don't have perl.h H_PERL */ 1706 1707 /* If we don't have perl.h, we are compiling a utility program. Below we 1708 * hard-code various macro definitions that wouldn't otherwise be available 1709 * to it. Most are coded based on first principles. These are written to 1710 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */ 1711 # define isDIGIT_A(c) inRANGE(c, '0', '9') 1712 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t') 1713 # define isSPACE_A(c) (isBLANK_A(c) \ 1714 || (c) == '\n' \ 1715 || (c) == '\r' \ 1716 || (c) == '\v' \ 1717 || (c) == '\f') 1718 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for 1719 * uppercase. The tests for those aren't necessary on ASCII, but hurt only 1720 * performance (if optimization isn't on), and allow the same code to be 1721 * used for both platform types */ 1722 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \ 1723 || inRANGE((c), 'j', 'r') \ 1724 || inRANGE((c), 's', 'z') 1725 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \ 1726 || inRANGE((c), 'J', 'R') \ 1727 || inRANGE((c), 'S', 'Z') 1728 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c)) 1729 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c)) 1730 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_') 1731 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_') 1732 # define isXDIGIT_A(c) ( isDIGIT_A(c) \ 1733 || inRANGE((c), 'a', 'f') \ 1734 || inRANGE((c), 'A', 'F') 1735 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \ 1736 || (c) == '#' || (c) == '$' || (c) == '%' \ 1737 || (c) == '&' || (c) == '\'' || (c) == '(' \ 1738 || (c) == ')' || (c) == '*' || (c) == '+' \ 1739 || (c) == ',' || (c) == '.' || (c) == '/' \ 1740 || (c) == ':' || (c) == ';' || (c) == '<' \ 1741 || (c) == '=' || (c) == '>' || (c) == '?' \ 1742 || (c) == '@' || (c) == '[' || (c) == '\\' \ 1743 || (c) == ']' || (c) == '^' || (c) == '_' \ 1744 || (c) == '`' || (c) == '{' || (c) == '|' \ 1745 || (c) == '}' || (c) == '~') 1746 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c)) 1747 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ') 1748 1749 # ifdef EBCDIC 1750 /* The below is accurate for the 3 EBCDIC code pages traditionally 1751 * supported by perl. The only difference between them in the controls 1752 * is the position of \n, and that is represented symbolically below */ 1753 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \ 1754 || (c) == '\f' || (c) == '\n' || (c) == '\r' \ 1755 || (c) == '\t' || (c) == '\v' \ 1756 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \ 1757 || (c) == 7F /* U+7F DEL */ \ 1758 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \ 1759 DC[1-3] */ \ 1760 || (c) == 0x18 /* U+18 CAN */ \ 1761 || (c) == 0x19 /* U+19 EOM */ \ 1762 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \ 1763 || (c) == 0x26 /* U+17 ETB */ \ 1764 || (c) == 0x27 /* U+1B ESC */ \ 1765 || (c) == 0x2D /* U+05 ENQ */ \ 1766 || (c) == 0x2E /* U+06 ACK */ \ 1767 || (c) == 0x32 /* U+16 SYN */ \ 1768 || (c) == 0x37 /* U+04 EOT */ \ 1769 || (c) == 0x3C /* U+14 DC4 */ \ 1770 || (c) == 0x3D /* U+15 NAK */ \ 1771 || (c) == 0x3F)/* U+1A SUB */ 1772 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c)) 1773 # else /* isASCII is already defined for ASCII platforms, so can use that to 1774 define isCNTRL */ 1775 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c)) 1776 # endif 1777 1778 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them 1779 * during debugging, and it seems best to keep them. We may be called 1780 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't 1781 * do anything anyway, so make it not a problem */ 1782 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1) 1783 # define NATIVE_TO_LATIN1(ch) (ch) 1784 # endif 1785 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c)) 1786 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c)) 1787 # define isBLANK_L1(c) (isBLANK_A(c) \ 1788 || (FITS_IN_8_BITS(c) \ 1789 && NATIVE_TO_LATIN1((U8) c) == 0xA0)) 1790 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c))) 1791 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c))) 1792 # define isLOWER_L1(c) (isLOWER_A(c) \ 1793 || (FITS_IN_8_BITS(c) \ 1794 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \ 1795 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \ 1796 || NATIVE_TO_LATIN1((U8) c) == 0xAA \ 1797 || NATIVE_TO_LATIN1((U8) c) == 0xBA \ 1798 || NATIVE_TO_LATIN1((U8) c) == 0xB5))) 1799 # define isPRINT_L1(c) (isPRINT_A(c) \ 1800 || (FITS_IN_8_BITS(c) \ 1801 && NATIVE_TO_LATIN1((U8) c) >= 0xA0)) 1802 # define isPUNCT_L1(c) (isPUNCT_A(c) \ 1803 || (FITS_IN_8_BITS(c) \ 1804 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \ 1805 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \ 1806 || NATIVE_TO_LATIN1((U8) c) == 0xAB \ 1807 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \ 1808 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \ 1809 || NATIVE_TO_LATIN1((U8) c) == 0xBB \ 1810 || NATIVE_TO_LATIN1((U8) c) == 0xBF))) 1811 # define isSPACE_L1(c) (isSPACE_A(c) \ 1812 || (FITS_IN_8_BITS(c) \ 1813 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \ 1814 || NATIVE_TO_LATIN1((U8) c) == 0xA0))) 1815 # define isUPPER_L1(c) (isUPPER_A(c) \ 1816 || (FITS_IN_8_BITS(c) \ 1817 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \ 1818 0xC0, 0xDE) \ 1819 && NATIVE_TO_LATIN1((U8) c) != 0xD7))) 1820 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c)) 1821 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_') 1822 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \ 1823 || isBLANK_L1(c) \ 1824 || (c) == '-' \ 1825 || (c) == '(' \ 1826 || (c) == ')') 1827 /* The following are not fully accurate in the above-ASCII range. I (khw) 1828 * don't think it's necessary to be so for the purposes where this gets 1829 * compiled */ 1830 # define isQUOTEMETA_(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c)) 1831 1832 /* Many of the macros later in this file are defined in terms of these. By 1833 * implementing them with a function, which converts the class number into 1834 * a call to the desired macro, all of the later ones work. However, that 1835 * function won't be actually defined when building a utility program (no 1836 * perl.h), and so a compiler error will be generated if one is attempted 1837 * to be used. And the above-Latin1 code points require Unicode tables to 1838 * be present, something unlikely to be the case when bootstrapping */ 1839 # define generic_isCC_(c, classnum) \ 1840 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE)) 1841 # define generic_isCC_A_(c, classnum) \ 1842 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE)) 1843 #endif /* End of no perl.h H_PERL */ 1844 1845 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c) 1846 #define isALPHA(c) isALPHA_A(c) 1847 #define isASCII_A(c) isASCII(c) 1848 #define isASCII_L1(c) isASCII(c) 1849 #define isBLANK(c) isBLANK_A(c) 1850 #define isCNTRL(c) isCNTRL_A(c) 1851 #define isDIGIT(c) isDIGIT_A(c) 1852 #define isGRAPH(c) isGRAPH_A(c) 1853 #define isIDFIRST(c) isIDFIRST_A(c) 1854 #define isLOWER(c) isLOWER_A(c) 1855 #define isPRINT(c) isPRINT_A(c) 1856 #define isPSXSPC_A(c) isSPACE_A(c) 1857 #define isPSXSPC(c) isPSXSPC_A(c) 1858 #define isPSXSPC_L1(c) isSPACE_L1(c) 1859 #define isPUNCT(c) isPUNCT_A(c) 1860 #define isSPACE(c) isSPACE_A(c) 1861 #define isUPPER(c) isUPPER_A(c) 1862 #define isWORDCHAR(c) isWORDCHAR_A(c) 1863 #define isXDIGIT(c) isXDIGIT_A(c) 1864 1865 /* ASCII casing. These could also be written as 1866 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c)) 1867 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c)) 1868 which uses table lookup and mask instead of subtraction. (This would 1869 work because the _MOD does not apply in the ASCII range). 1870 1871 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII 1872 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are 1873 there non-ASCII invariants, and all of them are controls.) */ 1874 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c)) 1875 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c)) 1876 1877 /* In the ASCII range, these are equivalent to what they're here defined to be. 1878 * But by creating these definitions, other code doesn't have to be aware of 1879 * this detail. Actually this works for all UTF-8 invariants, not just the 1880 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */ 1881 #define toFOLD(c) toLOWER(c) 1882 #define toTITLE(c) toUPPER(c) 1883 1884 #define toLOWER_A(c) toLOWER(c) 1885 #define toUPPER_A(c) toUPPER(c) 1886 #define toFOLD_A(c) toFOLD(c) 1887 #define toTITLE_A(c) toTITLE(c) 1888 1889 /* Use table lookup for speed; returns the input itself if is out-of-range */ 1890 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \ 1891 ? (c) \ 1892 : PL_latin1_lc[ (U8) (c) ]) 1893 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */ 1894 1895 /* Modified uc. Is correct uc except for three non-ascii chars which are 1896 * all mapped to one of them, and these need special handling; returns the 1897 * input itself if is out-of-range */ 1898 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \ 1899 ? (c) \ 1900 : PL_mod_latin1_uc[ (U8) (c) ]) 1901 #ifdef USE_LOCALE_CTYPE 1902 # define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale 1903 # define IN_UTF8_TURKIC_LOCALE PL_in_utf8_turkic_locale 1904 #else 1905 # define IN_UTF8_CTYPE_LOCALE false 1906 # define IN_UTF8_TURKIC_LOCALE false 1907 #endif 1908 1909 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */ 1910 1911 /* For internal core Perl use only: the base macro for defining macros like 1912 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point 1913 * (0-255) to check. In a UTF-8 locale, the result is the same as calling 1914 * isFOO_L1(); 'classnum' is something like CC_UPPER_, which gives the class 1915 * number for doing this. For non-UTF-8 locales, the code to actually do the 1916 * test this is passed in 'non_utf8'. If 'c' is above 255, 0 is returned. For 1917 * accessing the full range of possible code points under locale rules, use the 1918 * macros based on generic_LC_uvchr_ instead of this. */ 1919 #define generic_LC_base_(c, classnum, non_utf8_func) \ 1920 (! FITS_IN_8_BITS(c) \ 1921 ? 0 \ 1922 : IN_UTF8_CTYPE_LOCALE \ 1923 ? cBOOL(PL_charclass[(U8) (c)] & CC_mask_(classnum)) \ 1924 : cBOOL(non_utf8_func(c))) 1925 1926 /* A helper macro for defining macros like isALPHA_LC. On systems without 1927 * proper locales, these reduce to, e.g., isALPHA_A */ 1928 #ifdef CTYPE256 1929 # define generic_LC_(c, classnum, non_utf8_func) \ 1930 generic_LC_base_(c, classnum, non_utf8_func) 1931 #else 1932 # define generic_LC_(c, classnum, non_utf8_func) \ 1933 generic_isCC_A_(c, classnum) 1934 #endif 1935 1936 /* Below are the definitions for the locale-sensitive character classification 1937 * macros whose input domain is a byte, and the locale isn't UTF-8. These are 1938 * as close as possible to the bare versions on the platform and still yield 1939 * POSIX Standard-compliant results. 1940 * 1941 * There is currently only one place these definitions should be used, in 1942 * certain function calls like Perl_iswordchar_() in inline.h. 1943 * 1944 * Most likely you want to use the macros a ways below with names like 1945 * isALPHA_LC(). Rarely, you may want isU8_ALPHA_LC(), somewhat below. 1946 * 1947 * The first two aren't in C89, so the fallback is to use the non-locale 1948 * sensitive versions; these are the same for all platforms */ 1949 #if defined(HAS_ISASCII) 1950 # define is_posix_ASCII(c) isascii((U8) (c)) 1951 #else 1952 # define is_posix_ASCII(c) isASCII(c) 1953 #endif 1954 1955 #if defined(HAS_ISBLANK) 1956 # define is_posix_BLANK(c) isblank((U8) (c)) 1957 #else 1958 # define is_posix_BLANK(c) isBLANK(c) 1959 #endif 1960 1961 /* The next few are the same in all platforms. */ 1962 #define is_posix_CNTRL(c) iscntrl((U8) (c)) 1963 #define is_posix_IDFIRST(c) (UNLIKELY((c) == '_') || is_posix_ALPHA(c)) 1964 #define is_posix_SPACE(c) isspace((U8) (c)) 1965 #define is_posix_WORDCHAR(c) (UNLIKELY((c) == '_') || is_posix_ALPHANUMERIC(c)) 1966 1967 /* The base-level case changing macros are also the same in all platforms */ 1968 #define to_posix_LOWER(c) tolower((U8) (c)) 1969 #define to_posix_UPPER(c) toupper((U8) (c)) 1970 #define to_posix_FOLD(c) to_posix_LOWER(c) 1971 1972 #ifdef WIN32 1973 1974 /* The Windows functions don't bother to follow the POSIX standard, which for 1975 * example says that something can't both be a printable and a control. But 1976 * Windows treats \t as both a control and a printable, and does such things as 1977 * making superscripts into both digits and punctuation. These #defines tame 1978 * these flaws by assuming that the definitions of controls (and the other few 1979 * ones defined above) are correct, and then making sure that other definitions 1980 * don't have weirdnesses, by adding a check that \w and its subsets aren't 1981 * ispunct(), and things that are \W, like ispunct(), arent't controls. Not 1982 * all possible weirdnesses are checked for, just ones that were detected on 1983 * actual Microsoft code pages */ 1984 # define is_posix_ALPHA(c) \ 1985 (isalpha((U8) (c)) && ! is_posix_PUNCT(c)) 1986 # define is_posix_ALPHANUMERIC(c) \ 1987 (isalnum((U8) (c)) && ! is_posix_PUNCT(c)) 1988 # define is_posix_CASED(c) \ 1989 ((isupper((U8) (c)) || islower((U8) (c))) && ! is_posix_PUNCT(c)) 1990 # define is_posix_DIGIT(c) \ 1991 (isdigit((U8) (c)) && ! is_posix_PUNCT(c)) 1992 # define is_posix_GRAPH(c) \ 1993 (isgraph((U8) (c)) && ! is_posix_CNTRL(c)) 1994 # define is_posix_LOWER(c) \ 1995 (islower((U8) (c)) && ! is_posix_PUNCT(c)) 1996 # define is_posix_PRINT(c) \ 1997 (isprint((U8) (c)) && ! is_posix_CNTRL(c)) 1998 # define is_posix_PUNCT(c) \ 1999 (ispunct((U8) (c)) && ! is_posix_CNTRL(c)) 2000 # define is_posix_UPPER(c) \ 2001 (isupper((U8) (c)) && ! is_posix_PUNCT(c)) 2002 # define is_posix_XDIGIT(c) \ 2003 (isxdigit((U8) (c)) && ! is_posix_PUNCT(c)) 2004 #else 2005 2006 /* For all other platforms, as far as we know, isdigit(), etc. work sanely 2007 * enough */ 2008 # define is_posix_ALPHA(c) isalpha((U8) (c)) 2009 # define is_posix_ALPHANUMERIC(c) isalnum((U8) (c)) 2010 # define is_posix_CASED(c) (islower((U8) (c)) || isupper((U8) (c))) 2011 # define is_posix_DIGIT(c) isdigit((U8) (c)) 2012 2013 /* ... But it seems that IBM products treat NBSP as both a space and a 2014 * graphic; these are the two platforms that we have active test beds for. 2015 */ 2016 # if defined(OS390) || defined(_AIX) 2017 # define is_posix_GRAPH(c) (isgraph((U8) (c)) && ! isspace((U8) (c))) 2018 # else 2019 # define is_posix_GRAPH(c) isgraph((U8) (c)) 2020 # endif 2021 # define is_posix_LOWER(c) islower((U8) (c)) 2022 # define is_posix_PRINT(c) isprint((U8) (c)) 2023 # define is_posix_PUNCT(c) ispunct((U8) (c)) 2024 # define is_posix_UPPER(c) isupper((U8) (c)) 2025 # define is_posix_XDIGIT(c) isxdigit((U8) (c)) 2026 #endif 2027 2028 /* Below is the next level up, which currently expands to nothing more 2029 * than the previous layer. These are the macros to use if you really need 2030 * something whose input domain is a byte, and the locale isn't UTF-8; that is, 2031 * where you normally would have to use things like bare isalnum(). 2032 * 2033 * But most likely you should instead use the layer defined further below which 2034 * has names like isALPHA_LC. They deal with larger-than-byte inputs, and 2035 * UTF-8 locales. 2036 * 2037 * (Note, proper general operation of the bare libc functions requires you to 2038 * cast to U8. These do that for you automatically.) */ 2039 2040 # define WRAP_U8_LC_(c, classnum, posix) posix(c) 2041 2042 #define isU8_ALPHANUMERIC_LC(c) \ 2043 WRAP_U8_LC_((c), CC_ALPHANUMERIC_, is_posix_ALPHANUMERIC) 2044 #define isU8_ALPHA_LC(c) WRAP_U8_LC_((c), CC_ALPHA_, is_posix_ALPHA) 2045 #define isU8_ASCII_LC(c) WRAP_U8_LC_((c), CC_ASCII_, is_posix_ASCII) 2046 #define isU8_BLANK_LC(c) WRAP_U8_LC_((c), CC_BLANK_, is_posix_BLANK) 2047 #define isU8_CASED_LC(c) WRAP_U8_LC_((c), CC_CASED_, is_posix_CASED) 2048 #define isU8_CNTRL_LC(c) WRAP_U8_LC_((c), CC_CNTRL_, is_posix_CNTRL) 2049 #define isU8_DIGIT_LC(c) WRAP_U8_LC_((c), CC_DIGIT_, is_posix_DIGIT) 2050 #define isU8_GRAPH_LC(c) WRAP_U8_LC_((c), CC_GRAPH_, is_posix_GRAPH) 2051 #define isU8_IDFIRST_LC(c) WRAP_U8_LC_((c), CC_IDFIRST_, is_posix_IDFIRST) 2052 #define isU8_LOWER_LC(c) WRAP_U8_LC_((c), CC_LOWER_, is_posix_LOWER) 2053 #define isU8_PRINT_LC(c) WRAP_U8_LC_((c), CC_PRINT_, is_posix_PRINT) 2054 #define isU8_PUNCT_LC(c) WRAP_U8_LC_((c), CC_PUNCT_, is_posix_PUNCT) 2055 #define isU8_SPACE_LC(c) WRAP_U8_LC_((c), CC_SPACE_, is_posix_SPACE) 2056 #define isU8_UPPER_LC(c) WRAP_U8_LC_((c), CC_UPPER_, is_posix_UPPER) 2057 #define isU8_WORDCHAR_LC(c) WRAP_U8_LC_((c), CC_WORDCHAR_, is_posix_WORDCHAR) 2058 #define isU8_XDIGIT_LC(c) WRAP_U8_LC_((c), CC_XDIGIT_, is_posix_XDIGIT) 2059 2060 #define toU8_LOWER_LC(c) WRAP_U8_LC_((c), CC_TOLOWER_, to_posix_LOWER) 2061 #define toU8_UPPER_LC(c) WRAP_U8_LC_((c), CC_TOUPPER_, to_posix_UPPER) 2062 #define toU8_FOLD_LC(c) toU8_LOWER_LC(c) 2063 2064 /* The definitions below use the ones above to create versions in which the 2065 * input domain isn't restricted to bytes (though always returning false if the 2066 * input doesn't fit in a byte), and to behave properly should the locale be 2067 * UTF-8. These are the documented ones, suitable for general use (though 2068 * toUPPER_LC and toFOLD_LC aren't documented because they need special 2069 * handling to deal with SHARP S expanding to two characters). */ 2070 2071 #define isASCII_LC(c) (FITS_IN_8_BITS(c) && isU8_ASCII_LC(c)) 2072 #define isALPHA_LC(c) generic_LC_(c, CC_ALPHA_, isU8_ALPHA_LC) 2073 #define isALPHANUMERIC_LC(c) \ 2074 generic_LC_(c, CC_ALPHANUMERIC_, isU8_ALPHANUMERIC_LC) 2075 #define isBLANK_LC(c) generic_LC_(c, CC_BLANK_, isU8_BLANK_LC) 2076 #define isCASED_LC(c) generic_LC_(c, CC_CASED_, isU8_CASED_LC) 2077 #define isCNTRL_LC(c) generic_LC_(c, CC_CNTRL_, isU8_CNTRL_LC) 2078 #define isDIGIT_LC(c) generic_LC_(c, CC_DIGIT_, isU8_DIGIT_LC) 2079 #define isGRAPH_LC(c) generic_LC_(c, CC_GRAPH_, isU8_GRAPH_LC) 2080 #define isIDFIRST_LC(c) generic_LC_(c, CC_IDFIRST_, isU8_IDFIRST_LC) 2081 #define isLOWER_LC(c) generic_LC_(c, CC_LOWER_, isU8_LOWER_LC) 2082 #define isPRINT_LC(c) generic_LC_(c, CC_PRINT_, isU8_PRINT_LC) 2083 #define isPUNCT_LC(c) generic_LC_(c, CC_PUNCT_, isU8_PUNCT_LC) 2084 #define isSPACE_LC(c) generic_LC_(c, CC_SPACE_, isU8_SPACE_LC) 2085 #define isUPPER_LC(c) generic_LC_(c, CC_UPPER_, isU8_UPPER_LC) 2086 #define isWORDCHAR_LC(c) generic_LC_(c, CC_WORDCHAR_, isU8_WORDCHAR_LC) 2087 #define isXDIGIT_LC(c) generic_LC_(c, CC_XDIGIT_, isU8_XDIGIT_LC) 2088 2089 #ifndef CTYPE256 2090 # define toLOWER_LC(c) toLOWER_A(c) 2091 # define toUPPER_LC(c) toUPPER_A(c) 2092 # define toFOLD_LC(c) toFOLD_A(c) 2093 #else 2094 2095 /* In the next three macros, the reason for using the PL_latin arrays is in 2096 * case the system function is defective; it ensures uniform results that 2097 * conform to the Unicode standard. */ 2098 2099 /* This does not handle the anomalies in UTF-8 Turkic locales. */ 2100 # define toLOWER_LC(c) ((! FITS_IN_8_BITS(c)) \ 2101 ? (c) \ 2102 : ((IN_UTF8_CTYPE_LOCALE) \ 2103 ? PL_latin1_lc[ (U8) (c) ] \ 2104 : ((U8) toU8_LOWER_LC(c)))) 2105 2106 /* In this macro, note that the result can be larger than a byte in a UTF-8 2107 * locale. It returns a single value, so can't adequately return the upper 2108 * case of LATIN SMALL LETTER SHARP S in a UTF-8 locale (which should be a 2109 * string of two values "SS"); instead it asserts against that under 2110 * DEBUGGING, and otherwise returns its input. It does not handle the 2111 * anomalies in UTF-8 Turkic locales. */ 2112 # define toUPPER_LC(c) \ 2113 ((! FITS_IN_8_BITS(c)) \ 2114 ? (c) \ 2115 : ((! IN_UTF8_CTYPE_LOCALE) \ 2116 ? ((U8) toU8_UPPER_LC(c)) \ 2117 : (UNLIKELY(((U8)(c)) == MICRO_SIGN) \ 2118 ? GREEK_CAPITAL_LETTER_MU \ 2119 : ((UNLIKELY(((U8) (c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \ 2120 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \ 2121 : (UNLIKELY(((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \ 2122 ? (__ASSERT_(0) (c)) /* Fail on Sharp S in DEBUGGING */ \ 2123 : PL_mod_latin1_uc[ (U8) (c) ])))))) 2124 2125 /* In this macro, note that the result can be larger than a byte in a UTF-8 2126 * locale. It returns a single value, so can't adequately return the fold case 2127 * of LATIN SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of 2128 * two values "ss"); instead it asserts against that under DEBUGGING, and 2129 * otherwise returns its input. It does not handle the anomalies in UTF-8 2130 * Turkic locales */ 2131 # define toFOLD_LC(c) \ 2132 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \ 2133 ? GREEK_SMALL_LETTER_MU \ 2134 : (__ASSERT_( ! IN_UTF8_CTYPE_LOCALE \ 2135 || LIKELY((c) != LATIN_SMALL_LETTER_SHARP_S)) \ 2136 toLOWER_LC(c))) 2137 #endif 2138 2139 #define isIDCONT(c) isWORDCHAR(c) 2140 #define isIDCONT_A(c) isWORDCHAR_A(c) 2141 #define isIDCONT_L1(c) isWORDCHAR_L1(c) 2142 #define isIDCONT_LC(c) isWORDCHAR_LC(c) 2143 #define isPSXSPC_LC(c) isSPACE_LC(c) 2144 2145 /* For internal core Perl use only: the base macros for defining macros like 2146 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class 2147 * number defined earlier in this file. generic_uvchr_() is used for POSIX 2148 * classes where there is a macro or function 'above_latin1' that takes the 2149 * single argument 'c' and returns the desired value. These exist for those 2150 * classes which have simple definitions, avoiding the overhead of an inversion 2151 * list binary search. generic_invlist_uvchr_() can be used 2152 * for classes where that overhead is faster than a direct lookup. 2153 * generic_uvchr_() won't compile if 'c' isn't unsigned, as it won't match the 2154 * 'above_latin1' prototype. generic_isCC_() macro does bounds checking, so 2155 * have duplicate checks here, so could create versions of the macros that 2156 * don't, but experiments show that gcc optimizes them out anyway. */ 2157 2158 /* Note that all ignore 'use bytes' */ 2159 #define generic_uvchr_(classnum, above_latin1, c) ((c) < 256 \ 2160 ? generic_isCC_(c, classnum) \ 2161 : above_latin1(c)) 2162 #define generic_invlist_uvchr_(classnum, c) ((c) < 256 \ 2163 ? generic_isCC_(c, classnum) \ 2164 : _is_uni_FOO(classnum, c)) 2165 #define isALPHA_uvchr(c) generic_invlist_uvchr_(CC_ALPHA_, c) 2166 #define isALPHANUMERIC_uvchr(c) generic_invlist_uvchr_(CC_ALPHANUMERIC_, c) 2167 #define isASCII_uvchr(c) isASCII(c) 2168 #define isBLANK_uvchr(c) generic_uvchr_(CC_BLANK_, is_HORIZWS_cp_high, c) 2169 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */ 2170 #define isDIGIT_uvchr(c) generic_invlist_uvchr_(CC_DIGIT_, c) 2171 #define isGRAPH_uvchr(c) generic_invlist_uvchr_(CC_GRAPH_, c) 2172 #define isIDCONT_uvchr(c) \ 2173 generic_uvchr_(CC_WORDCHAR_, _is_uni_perl_idcont, c) 2174 #define isIDFIRST_uvchr(c) \ 2175 generic_uvchr_(CC_IDFIRST_, _is_uni_perl_idstart, c) 2176 #define isLOWER_uvchr(c) generic_invlist_uvchr_(CC_LOWER_, c) 2177 #define isPRINT_uvchr(c) generic_invlist_uvchr_(CC_PRINT_, c) 2178 2179 #define isPUNCT_uvchr(c) generic_invlist_uvchr_(CC_PUNCT_, c) 2180 #define isSPACE_uvchr(c) generic_uvchr_(CC_SPACE_, is_XPERLSPACE_cp_high, c) 2181 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c) 2182 2183 #define isUPPER_uvchr(c) generic_invlist_uvchr_(CC_UPPER_, c) 2184 #define isVERTWS_uvchr(c) generic_uvchr_(CC_VERTSPACE_, is_VERTWS_cp_high, c) 2185 #define isWORDCHAR_uvchr(c) generic_invlist_uvchr_(CC_WORDCHAR_, c) 2186 #define isXDIGIT_uvchr(c) generic_uvchr_(CC_XDIGIT_, is_XDIGIT_cp_high, c) 2187 2188 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l) 2189 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l) 2190 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l) 2191 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l) 2192 2193 /* For backwards compatibility, even though '_uni' should mean official Unicode 2194 * code points, in Perl it means native for those below 256 */ 2195 #define isALPHA_uni(c) isALPHA_uvchr(c) 2196 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c) 2197 #define isASCII_uni(c) isASCII_uvchr(c) 2198 #define isBLANK_uni(c) isBLANK_uvchr(c) 2199 #define isCNTRL_uni(c) isCNTRL_uvchr(c) 2200 #define isDIGIT_uni(c) isDIGIT_uvchr(c) 2201 #define isGRAPH_uni(c) isGRAPH_uvchr(c) 2202 #define isIDCONT_uni(c) isIDCONT_uvchr(c) 2203 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c) 2204 #define isLOWER_uni(c) isLOWER_uvchr(c) 2205 #define isPRINT_uni(c) isPRINT_uvchr(c) 2206 #define isPUNCT_uni(c) isPUNCT_uvchr(c) 2207 #define isSPACE_uni(c) isSPACE_uvchr(c) 2208 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c) 2209 #define isUPPER_uni(c) isUPPER_uvchr(c) 2210 #define isVERTWS_uni(c) isVERTWS_uvchr(c) 2211 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c) 2212 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c) 2213 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l) 2214 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l) 2215 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l) 2216 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l) 2217 2218 /* For internal core Perl use only: the base macros for defining macros like 2219 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code 2220 * point, not just 0-255. Like generic_uvchr_, there are two versions, one for 2221 * simple class definitions; the other for more complex. These are like 2222 * generic_uvchr_, so see it for more info. */ 2223 #define generic_LC_uvchr_(latin1, above_latin1, c) \ 2224 (c < 256 ? latin1(c) : above_latin1(c)) 2225 #define generic_LC_invlist_uvchr_(latin1, classnum, c) \ 2226 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c)) 2227 2228 #define isALPHA_LC_uvchr(c) generic_LC_invlist_uvchr_(isALPHA_LC, CC_ALPHA_, c) 2229 #define isALPHANUMERIC_LC_uvchr(c) generic_LC_invlist_uvchr_(isALPHANUMERIC_LC, \ 2230 CC_ALPHANUMERIC_, c) 2231 #define isASCII_LC_uvchr(c) isASCII_LC(c) 2232 #define isBLANK_LC_uvchr(c) generic_LC_uvchr_(isBLANK_LC, \ 2233 is_HORIZWS_cp_high, c) 2234 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0) 2235 #define isDIGIT_LC_uvchr(c) generic_LC_invlist_uvchr_(isDIGIT_LC, CC_DIGIT_, c) 2236 #define isGRAPH_LC_uvchr(c) generic_LC_invlist_uvchr_(isGRAPH_LC, CC_GRAPH_, c) 2237 #define isIDCONT_LC_uvchr(c) generic_LC_uvchr_(isIDCONT_LC, \ 2238 _is_uni_perl_idcont, c) 2239 #define isIDFIRST_LC_uvchr(c) generic_LC_uvchr_(isIDFIRST_LC, \ 2240 _is_uni_perl_idstart, c) 2241 #define isLOWER_LC_uvchr(c) generic_LC_invlist_uvchr_(isLOWER_LC, CC_LOWER_, c) 2242 #define isPRINT_LC_uvchr(c) generic_LC_invlist_uvchr_(isPRINT_LC, CC_PRINT_, c) 2243 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c) 2244 #define isPUNCT_LC_uvchr(c) generic_LC_invlist_uvchr_(isPUNCT_LC, CC_PUNCT_, c) 2245 #define isSPACE_LC_uvchr(c) generic_LC_uvchr_(isSPACE_LC, \ 2246 is_XPERLSPACE_cp_high, c) 2247 #define isUPPER_LC_uvchr(c) generic_LC_invlist_uvchr_(isUPPER_LC, CC_UPPER_, c) 2248 #define isWORDCHAR_LC_uvchr(c) generic_LC_invlist_uvchr_(isWORDCHAR_LC, \ 2249 CC_WORDCHAR_, c) 2250 #define isXDIGIT_LC_uvchr(c) generic_LC_uvchr_(isXDIGIT_LC, \ 2251 is_XDIGIT_cp_high, c) 2252 2253 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c)) 2254 2255 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but 2256 * they don't otherwise go out of their way to look for malformed UTF-8. If 2257 * they can return accurate results without knowing if the input is otherwise 2258 * malformed, they do so. For example isASCII is accurate in spite of any 2259 * non-length malformations because it looks only at a single byte. Likewise 2260 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8 2261 * variant ones return FALSE. But, if the input has to be well-formed in order 2262 * for the results to be accurate, the macros will test and if malformed will 2263 * call a routine to die 2264 * 2265 * Except for toke.c, the macros do assume that e > p, asserting that on 2266 * DEBUGGING builds. Much code that calls these depends on this being true, 2267 * for other reasons. toke.c is treated specially as using the regular 2268 * assertion breaks it in many ways. All strings that these operate on there 2269 * are supposed to have an extra NUL character at the end, so that *e = \0. A 2270 * bunch of code in toke.c assumes that this is true, so the assertion allows 2271 * for that */ 2272 #ifdef PERL_IN_TOKE_C 2273 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0')) 2274 #else 2275 # define _utf8_safe_assert(p,e) ((e) > (p)) 2276 #endif 2277 2278 #define generic_utf8_safe_(classnum, p, e, above_latin1) \ 2279 ((! _utf8_safe_assert(p, e)) \ 2280 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\ 2281 : (UTF8_IS_INVARIANT(*(p))) \ 2282 ? generic_isCC_(*(p), classnum) \ 2283 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ 2284 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ 2285 ? generic_isCC_(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \ 2286 classnum) \ 2287 : (_force_out_malformed_utf8_message( \ 2288 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ 2289 : above_latin1)) 2290 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value. 2291 * 'above_latin1' can be a macro */ 2292 #define generic_func_utf8_safe_(classnum, above_latin1, p, e) \ 2293 generic_utf8_safe_(classnum, p, e, above_latin1(p, e)) 2294 #define generic_non_invlist_utf8_safe_(classnum, above_latin1, p, e) \ 2295 generic_utf8_safe_(classnum, p, e, \ 2296 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ 2297 ? (_force_out_malformed_utf8_message( \ 2298 (U8 *) (p), (U8 *) (e), 0, 1), 0) \ 2299 : above_latin1(p))) 2300 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an 2301 * 'above_latin1' parameter */ 2302 #define generic_invlist_utf8_safe_(classnum, p, e) \ 2303 generic_utf8_safe_(classnum, p, e, _is_utf8_FOO(classnum, p, e)) 2304 2305 /* Like the above, but should be used only when it is known that there are no 2306 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the 2307 * class is TRUE for. Hence it can skip the tests for this range. 2308 * 'above_latin1' should include its arguments */ 2309 #define generic_utf8_safe_no_upper_latin1_(classnum, p, e, above_latin1) \ 2310 (__ASSERT_(_utf8_safe_assert(p, e)) \ 2311 (isASCII(*(p))) \ 2312 ? generic_isCC_(*(p), classnum) \ 2313 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \ 2314 ? 0 /* Note that doesn't check validity for latin1 */ \ 2315 : above_latin1) 2316 2317 2318 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e) 2319 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e) 2320 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e) 2321 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e) 2322 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e) 2323 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e) 2324 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e) 2325 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e) 2326 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e) 2327 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e) 2328 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e) 2329 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e) 2330 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e) 2331 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e) 2332 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e) 2333 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e) 2334 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e) 2335 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e) 2336 2337 #define isALPHA_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_ALPHA_, p, e) 2338 #define isALPHANUMERIC_utf8_safe(p, e) \ 2339 generic_invlist_utf8_safe_(CC_ALPHANUMERIC_, p, e) 2340 #define isASCII_utf8_safe(p, e) \ 2341 /* Because ASCII is invariant under utf8, the non-utf8 macro \ 2342 * works */ \ 2343 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p))) 2344 #define isBLANK_utf8_safe(p, e) \ 2345 generic_non_invlist_utf8_safe_(CC_BLANK_, is_HORIZWS_high, p, e) 2346 2347 #ifdef EBCDIC 2348 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this 2349 * more efficient macro instead of the more general one */ 2350 # define isCNTRL_utf8_safe(p, e) \ 2351 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p))) 2352 #else 2353 # define isCNTRL_utf8_safe(p, e) generic_utf8_safe_(CC_CNTRL_, p, e, 0) 2354 #endif 2355 2356 #define isDIGIT_utf8_safe(p, e) \ 2357 generic_utf8_safe_no_upper_latin1_(CC_DIGIT_, p, e, \ 2358 _is_utf8_FOO(CC_DIGIT_, p, e)) 2359 #define isGRAPH_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_GRAPH_, p, e) 2360 #define isIDCONT_utf8_safe(p, e) generic_func_utf8_safe_(CC_WORDCHAR_, \ 2361 _is_utf8_perl_idcont, p, e) 2362 2363 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that 2364 * IDFIRST is an alnum. See 2365 * https://github.com/Perl/perl5/issues/10275 for more detail than you 2366 * ever wanted to know about. (In the ASCII range, there isn't a difference.) 2367 * This used to be not the XID version, but we decided to go with the more 2368 * modern Unicode definition */ 2369 #define isIDFIRST_utf8_safe(p, e) \ 2370 generic_func_utf8_safe_(CC_IDFIRST_, \ 2371 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e)) 2372 2373 #define isLOWER_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_LOWER_, p, e) 2374 #define isPRINT_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_PRINT_, p, e) 2375 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e) 2376 #define isPUNCT_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_PUNCT_, p, e) 2377 #define isSPACE_utf8_safe(p, e) \ 2378 generic_non_invlist_utf8_safe_(CC_SPACE_, is_XPERLSPACE_high, p, e) 2379 #define isUPPER_utf8_safe(p, e) generic_invlist_utf8_safe_(CC_UPPER_, p, e) 2380 #define isVERTWS_utf8_safe(p, e) \ 2381 generic_non_invlist_utf8_safe_(CC_VERTSPACE_, is_VERTWS_high, p, e) 2382 #define isWORDCHAR_utf8_safe(p, e) \ 2383 generic_invlist_utf8_safe_(CC_WORDCHAR_, p, e) 2384 #define isXDIGIT_utf8_safe(p, e) \ 2385 generic_utf8_safe_no_upper_latin1_(CC_XDIGIT_, p, e, \ 2386 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ 2387 ? (_force_out_malformed_utf8_message( \ 2388 (U8 *) (p), (U8 *) (e), 0, 1), 0) \ 2389 : is_XDIGIT_high(p))) 2390 2391 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l) 2392 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l) 2393 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l) 2394 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l) 2395 2396 /* For internal core use only, subject to change */ 2397 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f) 2398 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f) 2399 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f) 2400 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f) 2401 2402 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL) 2403 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0) 2404 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0) 2405 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0) 2406 2407 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e) 2408 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e) 2409 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e) 2410 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e) 2411 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e) 2412 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e) 2413 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e) 2414 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e) 2415 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e) 2416 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e) 2417 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e) 2418 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e) 2419 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e) 2420 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e) 2421 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e) 2422 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e) 2423 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e) 2424 2425 /* For internal core Perl use only: the base macros for defining macros like 2426 * isALPHA_LC_utf8_safe. These are like generic_utf8_, but if the first code 2427 * point in 'p' is within the 0-255 range, it uses locale rules from the 2428 * passed-in 'macro' parameter */ 2429 #define generic_LC_utf8_safe_(macro, p, e, above_latin1) \ 2430 (__ASSERT_(_utf8_safe_assert(p, e)) \ 2431 (UTF8_IS_INVARIANT(*(p))) \ 2432 ? macro(*(p)) \ 2433 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \ 2434 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \ 2435 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \ 2436 : (_force_out_malformed_utf8_message( \ 2437 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \ 2438 : above_latin1)) 2439 2440 #define generic_LC_invlist_utf8_safe_(macro, classnum, p, e) \ 2441 generic_LC_utf8_safe_(macro, p, e, \ 2442 _is_utf8_FOO(classnum, p, e)) 2443 2444 #define generic_LC_func_utf8_safe_(macro, above_latin1, p, e) \ 2445 generic_LC_utf8_safe_(macro, p, e, above_latin1(p, e)) 2446 2447 #define generic_LC_non_invlist_utf8_safe_(classnum, above_latin1, p, e) \ 2448 generic_LC_utf8_safe_(classnum, p, e, \ 2449 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \ 2450 ? (_force_out_malformed_utf8_message( \ 2451 (U8 *) (p), (U8 *) (e), 0, 1), 0) \ 2452 : above_latin1(p))) 2453 2454 #define isALPHANUMERIC_LC_utf8_safe(p, e) \ 2455 generic_LC_invlist_utf8_safe_(isALPHANUMERIC_LC, \ 2456 CC_ALPHANUMERIC_, p, e) 2457 #define isALPHA_LC_utf8_safe(p, e) \ 2458 generic_LC_invlist_utf8_safe_(isALPHA_LC, CC_ALPHA_, p, e) 2459 #define isASCII_LC_utf8_safe(p, e) \ 2460 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p))) 2461 #define isBLANK_LC_utf8_safe(p, e) \ 2462 generic_LC_non_invlist_utf8_safe_(isBLANK_LC, is_HORIZWS_high, p, e) 2463 #define isCNTRL_LC_utf8_safe(p, e) \ 2464 generic_LC_utf8_safe_(isCNTRL_LC, p, e, 0) 2465 #define isDIGIT_LC_utf8_safe(p, e) \ 2466 generic_LC_invlist_utf8_safe_(isDIGIT_LC, CC_DIGIT_, p, e) 2467 #define isGRAPH_LC_utf8_safe(p, e) \ 2468 generic_LC_invlist_utf8_safe_(isGRAPH_LC, CC_GRAPH_, p, e) 2469 #define isIDCONT_LC_utf8_safe(p, e) \ 2470 generic_LC_func_utf8_safe_(isIDCONT_LC, \ 2471 _is_utf8_perl_idcont, p, e) 2472 #define isIDFIRST_LC_utf8_safe(p, e) \ 2473 generic_LC_func_utf8_safe_(isIDFIRST_LC, \ 2474 _is_utf8_perl_idstart, p, e) 2475 #define isLOWER_LC_utf8_safe(p, e) \ 2476 generic_LC_invlist_utf8_safe_(isLOWER_LC, CC_LOWER_, p, e) 2477 #define isPRINT_LC_utf8_safe(p, e) \ 2478 generic_LC_invlist_utf8_safe_(isPRINT_LC, CC_PRINT_, p, e) 2479 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e) 2480 #define isPUNCT_LC_utf8_safe(p, e) \ 2481 generic_LC_invlist_utf8_safe_(isPUNCT_LC, CC_PUNCT_, p, e) 2482 #define isSPACE_LC_utf8_safe(p, e) \ 2483 generic_LC_non_invlist_utf8_safe_(isSPACE_LC, is_XPERLSPACE_high, p, e) 2484 #define isUPPER_LC_utf8_safe(p, e) \ 2485 generic_LC_invlist_utf8_safe_(isUPPER_LC, CC_UPPER_, p, e) 2486 #define isWORDCHAR_LC_utf8_safe(p, e) \ 2487 generic_LC_invlist_utf8_safe_(isWORDCHAR_LC, CC_WORDCHAR_, p, e) 2488 #define isXDIGIT_LC_utf8_safe(p, e) \ 2489 generic_LC_non_invlist_utf8_safe_(isXDIGIT_LC, is_XDIGIT_high, p, e) 2490 2491 /* Macros for backwards compatibility and for completeness when the ASCII and 2492 * Latin1 values are identical */ 2493 #define isALPHAU(c) isALPHA_L1(c) 2494 #define isDIGIT_L1(c) isDIGIT_A(c) 2495 #define isOCTAL(c) isOCTAL_A(c) 2496 #define isOCTAL_L1(c) isOCTAL_A(c) 2497 #define isXDIGIT_L1(c) isXDIGIT_A(c) 2498 #define isALNUM(c) isWORDCHAR(c) 2499 #define isALNUM_A(c) isALNUM(c) 2500 #define isALNUMU(c) isWORDCHAR_L1(c) 2501 #define isALNUM_LC(c) isWORDCHAR_LC(c) 2502 #define isALNUM_uni(c) isWORDCHAR_uni(c) 2503 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c) 2504 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e) 2505 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e) 2506 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e) 2507 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e) 2508 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */ 2509 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c) 2510 #define isALNUMC(c) isALPHANUMERIC(c) 2511 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c) 2512 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c) 2513 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c) 2514 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e) 2515 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e) 2516 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e) 2517 2518 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII, 2519 * except that they don't necessarily mean the same characters, e.g. CTRL-D is 2520 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC. 2521 * '?' is special-cased on EBCDIC to APC, which is the control there that is 2522 * the outlier from the block that contains the other controls, just like 2523 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0 2524 * block. If it weren't special cased, it would yield a non-control. 2525 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D, 2526 * etc. */ 2527 #ifndef EBCDIC 2528 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64) 2529 #else 2530 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \ 2531 ((isPRINT_A(c)) \ 2532 ? (UNLIKELY((c) == '?') \ 2533 ? QUESTION_MARK_CTRL \ 2534 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \ 2535 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \ 2536 ? '?' \ 2537 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64))))) 2538 #endif 2539 2540 /* 2541 =for apidoc Ay||line_t 2542 The typedef to use to declare variables that are to hold line numbers. 2543 2544 =cut 2545 2546 Line numbers are unsigned, 32 bits. 2547 */ 2548 typedef U32 line_t; 2549 #define LINE_Tf U32uf 2550 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */ 2551 2552 /* Helpful alias for version prescan */ 2553 #define is_LAX_VERSION(a,b) \ 2554 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL)) 2555 2556 #define is_STRICT_VERSION(a,b) \ 2557 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL)) 2558 2559 #define BADVERSION(a,b,c) \ 2560 if (b) { \ 2561 *b = c; \ 2562 } \ 2563 return a; 2564 2565 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or 2566 * a-f) to its numeric value without using any branches. The input is 2567 * validated only by an assert() in DEBUGGING builds. 2568 * 2569 * It works by right shifting and isolating the bit that is 0 for the digits, 2570 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones 2571 * position, and then to the eights position. Both are added together to form 2572 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the 2573 * final four bits of the input to form the correct value. */ 2574 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \ 2575 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \ 2576 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \ 2577 + ((c) & 0xF)) /* 0-9 if input valid hex digit */ 2578 2579 /* The argument is a string pointer, which is advanced. */ 2580 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1))) 2581 2582 /* Converts a character known to represent an octal digit (0-7) to its numeric 2583 * value. The input is validated only by an assert() in DEBUGGING builds. In 2584 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */ 2585 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c))) 2586 2587 /* Efficiently returns a boolean as to if two native characters are equivalent 2588 * case-insensitively. At least one of the characters must be one of [A-Za-z]; 2589 * the ALPHA in the name is to remind you of that. This is asserted() in 2590 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro 2591 * works (on valid input) for both non- and UTF-8-encoded bytes. 2592 * 2593 * When one of the inputs is a compile-time constant and gets folded by the 2594 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII 2595 * machines, 'A' and 'a' differ by a single bit; the same with the upper and 2596 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they 2597 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an 2598 * exclusive 'or' to find that bit and then inverts it to form a mask, with 2599 * just a single 0, in the bit position where the upper- and lowercase differ. 2600 * */ 2601 #define isALPHA_FOLD_EQ(c1, c2) \ 2602 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \ 2603 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a'))) 2604 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2))) 2605 2606 /* 2607 =for apidoc_section $memory 2608 2609 =for apidoc Am|void|Newx|void* ptr|int nitems|type 2610 =for apidoc_item |void*|safemalloc|size_t size 2611 2612 The XSUB-writer's interface to the C C<malloc> function. 2613 2614 Memory obtained by this should B<ONLY> be freed with L</"Safefree">. 2615 2616 In 5.9.3, Newx() and friends replace the older New() API, and drops 2617 the first parameter, I<x>, a debug aid which allowed callers to identify 2618 themselves. This aid has been superseded by a new build option, 2619 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still 2620 there for use in XS modules supporting older perls. 2621 2622 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast 2623 The XSUB-writer's interface to the C C<malloc> function, with 2624 cast. See also C<L</Newx>>. 2625 2626 Memory obtained by this should B<ONLY> be freed with L</"Safefree">. 2627 2628 =for apidoc Am|void|Newxz|void* ptr|int nitems|type 2629 =for apidoc_item |void*|safecalloc|size_t nitems|size_t item_size 2630 2631 The XSUB-writer's interface to the C C<malloc> function. The allocated 2632 memory is zeroed with C<memzero>. See also C<L</Newx>>. 2633 2634 Memory obtained by this should B<ONLY> be freed with L</"Safefree">. 2635 2636 =for apidoc Am|void|Renew|void* ptr|int nitems|type 2637 =for apidoc_item |void*|saferealloc|void *ptr|size_t size 2638 2639 The XSUB-writer's interface to the C C<realloc> function. 2640 2641 Memory obtained by this should B<ONLY> be freed with L</"Safefree">. 2642 2643 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast 2644 The XSUB-writer's interface to the C C<realloc> function, with 2645 cast. 2646 2647 Memory obtained by this should B<ONLY> be freed with L</"Safefree">. 2648 2649 =for apidoc Am|void|Safefree|void* ptr 2650 The XSUB-writer's interface to the C C<free> function. 2651 2652 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends. 2653 2654 =for apidoc_section $string 2655 =for apidoc Am|void |Move |void* src|void* dest|int nitems|type 2656 =for apidoc_item |void *|MoveD|void* src|void* dest|int nitems|type 2657 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the 2658 source, C<dest> is the destination, C<nitems> is the number of items, and 2659 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>. 2660 2661 C<MoveD> is like C<Move> but returns C<dest>. Useful 2662 for encouraging compilers to tail-call 2663 optimise. 2664 2665 =for apidoc Am|void |Copy |void* src|void* dest|int nitems|type 2666 =for apidoc_item |void *|CopyD|void* src|void* dest|int nitems|type 2667 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the 2668 source, C<dest> is the destination, C<nitems> is the number of items, and 2669 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>. 2670 2671 C<CopyD> is like C<Copy> but returns C<dest>. Useful 2672 for encouraging compilers to tail-call 2673 optimise. 2674 2675 =for apidoc Am|void |NewCopy |void* src|void* dest|int nitems|type 2676 Combines Newx() and Copy() into a single macro. Dest will be allocated 2677 using Newx() and then src will be copied into it. 2678 2679 =for apidoc Am|void |Zero |void* dest|int nitems|type 2680 =for apidoc_item |void *|ZeroD|void* dest|int nitems|type 2681 2682 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the 2683 destination, C<nitems> is the number of items, and C<type> is the type. 2684 2685 C<ZeroD> is like C<Zero> but returns C<dest>. Useful 2686 for encouraging compilers to tail-call 2687 optimise. 2688 2689 =for apidoc_section $utility 2690 =for apidoc Amu|void|StructCopy|type *src|type *dest|type 2691 This is an architecture-independent macro to copy one structure to another. 2692 2693 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte 2694 2695 Fill up memory with a byte pattern (a byte repeated over and over 2696 again) that hopefully catches attempts to access uninitialized memory. 2697 2698 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type 2699 2700 PoisonWith(0xAB) for catching access to allocated but uninitialized memory. 2701 2702 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type 2703 2704 PoisonWith(0xEF) for catching access to freed memory. 2705 2706 =for apidoc Am|void|Poison|void* dest|int nitems|type 2707 2708 PoisonWith(0xEF) for catching access to freed memory. 2709 2710 =cut */ 2711 2712 /* Maintained for backwards-compatibility only. Use newSV() instead. */ 2713 #ifndef PERL_CORE 2714 #define NEWSV(x,len) newSV(len) 2715 #endif 2716 2717 #define MEM_SIZE_MAX ((MEM_SIZE)-1) 2718 2719 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1)) 2720 2721 #ifdef PERL_MALLOC_WRAP 2722 2723 /* This expression will be constant-folded at compile time. It checks 2724 * whether or not the type of the count n is so small (e.g. U8 or U16, or 2725 * U32 on 64-bit systems) that there's no way a wrap-around could occur. 2726 * As well as avoiding the need for a run-time check in some cases, it's 2727 * designed to avoid compiler warnings like: 2728 * comparison is always false due to limited range of data type 2729 * It's mathematically equivalent to 2730 * max(n) * sizeof(t) > MEM_SIZE_MAX 2731 */ 2732 2733 2734 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \ 2735 ( sizeof(MEM_SIZE) < sizeof(n) \ 2736 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n)))) 2737 2738 /* This is written in a slightly odd way to avoid various spurious 2739 * compiler warnings. We *want* to write the expression as 2740 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C) 2741 * (for some compile-time constant C), but even when the LHS 2742 * constant-folds to false at compile-time, g++ insists on emitting 2743 * warnings about the RHS (e.g. "comparison is always false"), so instead 2744 * we write it as 2745 * 2746 * (cond ? n : X) > C 2747 * 2748 * where X is a constant with X > C always false. Choosing a value for X 2749 * is tricky. If 0, some compilers will complain about 0 > C always being 2750 * false; if 1, Coverity complains when n happens to be the constant value 2751 * '1', that cond ? 1 : 1 has the same value on both branches; so use C 2752 * for X and hope that nothing else whines. 2753 */ 2754 2755 # define _MEM_WRAP_WILL_WRAP(n,t) \ 2756 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \ 2757 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t)) 2758 2759 # define MEM_WRAP_CHECK(n,t) \ 2760 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ 2761 && (croak_memory_wrap(),0)) 2762 2763 # define MEM_WRAP_CHECK_1(n,t,a) \ 2764 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ 2765 && (Perl_croak_nocontext("%s",(a)),0)) 2766 2767 /* "a" arg must be a string literal */ 2768 # define MEM_WRAP_CHECK_s(n,t,a) \ 2769 ( (void) (UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \ 2770 && (Perl_croak_nocontext(ASSERT_IS_LITERAL(a)), 0))) 2771 2772 # define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), 2773 2774 # define PERL_STRLEN_ROUNDUP(n) ((void)(((n) > MEM_SIZE_MAX - 2 * PERL_STRLEN_ROUNDUP_QUANTUM) ? (croak_memory_wrap(),0) : 0), _PERL_STRLEN_ROUNDUP_UNCHECKED(n)) 2775 #else 2776 2777 # define MEM_WRAP_CHECK(n,t) 2778 # define MEM_WRAP_CHECK_1(n,t,a) 2779 # define MEM_WRAP_CHECK_s(n,t,a) 2780 # define MEM_WRAP_CHECK_(n,t) 2781 2782 # define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n) 2783 2784 #endif 2785 2786 #ifdef PERL_MEM_LOG 2787 /* 2788 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s 2789 * go through functions, which are handy for debugging breakpoints, but 2790 * which more importantly get the immediate calling environment (file and 2791 * line number, and C function name if available) passed in. This info can 2792 * then be used for logging the calls, for which one gets a sample 2793 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined. 2794 * 2795 * Known problems: 2796 * - not all memory allocs get logged, only those 2797 * that go through Newx() and derivatives (while all 2798 * Safefrees do get logged) 2799 * - __FILE__ and __LINE__ do not work everywhere 2800 * - __func__ or __FUNCTION__ even less so 2801 * - I think more goes on after the perlio frees but 2802 * the thing is that STDERR gets closed (as do all 2803 * the file descriptors) 2804 * - no deeper calling stack than the caller of the Newx() 2805 * or the kind, but do I look like a C reflection/introspection 2806 * utility to you? 2807 * - the function prototypes for the logging functions 2808 * probably should maybe be somewhere else than handy.h 2809 * - one could consider inlining (macrofying) the logging 2810 * for speed, but I am too lazy 2811 * - one could imagine recording the allocations in a hash, 2812 * (keyed by the allocation address?), and maintain that 2813 * through reallocs and frees, but how to do that without 2814 * any News() happening...? 2815 * - lots of -Ddefines to get useful/controllable output 2816 * - lots of ENV reads 2817 */ 2818 2819 # ifdef PERL_CORE 2820 # ifndef PERL_MEM_LOG_NOIMPL 2821 enum mem_log_type { 2822 MLT_ALLOC, 2823 MLT_REALLOC, 2824 MLT_FREE, 2825 MLT_NEW_SV, 2826 MLT_DEL_SV 2827 }; 2828 # endif 2829 # endif 2830 2831 #endif 2832 2833 #ifdef PERL_MEM_LOG 2834 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__) 2835 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__) 2836 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__) 2837 #endif 2838 2839 #ifndef MEM_LOG_ALLOC 2840 #define MEM_LOG_ALLOC(n,t,a) (a) 2841 #endif 2842 #ifndef MEM_LOG_REALLOC 2843 #define MEM_LOG_REALLOC(n,t,v,a) (a) 2844 #endif 2845 #ifndef MEM_LOG_FREE 2846 #define MEM_LOG_FREE(a) (a) 2847 #endif 2848 2849 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) 2850 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t)))))) 2851 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t))))) 2852 2853 #ifndef PERL_CORE 2854 /* pre 5.9.x compatibility */ 2855 #define New(x,v,n,t) Newx(v,n,t) 2856 #define Newc(x,v,n,t,c) Newxc(v,n,t,c) 2857 #define Newz(x,v,n,t) Newxz(v,n,t) 2858 #endif 2859 2860 #define Renew(v,n,t) \ 2861 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) 2862 #define Renewc(v,n,t,c) \ 2863 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t)))))) 2864 2865 #ifdef PERL_POISON 2866 #define Safefree(d) \ 2867 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0) 2868 #else 2869 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d))) 2870 #endif 2871 2872 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) * 2873 * as it handles cases like constant string arguments without throwing warnings * 2874 * the cast is required, as is the inequality check, to avoid warnings */ 2875 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 ) 2876 2877 2878 #define Move(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memmove((char*)(d),(const char*)(s), (n) * sizeof(t))) 2879 #define Copy(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), (void)memcpy((char*)(d),(const char*)(s), (n) * sizeof(t))) 2880 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t))) 2881 2882 /* Like above, but returns a pointer to 'd' */ 2883 #define MoveD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memmove((char*)(d),(const char*)(s), (n) * sizeof(t))) 2884 #define CopyD(s,d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), perl_assert_ptr(s), memcpy((char*)(d),(const char*)(s), (n) * sizeof(t))) 2885 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t))) 2886 2887 #define NewCopy(s,d,n,t) STMT_START { \ 2888 Newx(d,n,t); \ 2889 Copy(s,d,n,t); \ 2890 } STMT_END 2891 2892 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t))) 2893 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB) 2894 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF) 2895 #define Poison(d,n,t) PoisonFree(d,n,t) 2896 2897 #ifdef PERL_POISON 2898 # define PERL_POISON_EXPR(x) x 2899 #else 2900 # define PERL_POISON_EXPR(x) 2901 #endif 2902 2903 /* Shallow copy */ 2904 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s))) 2905 2906 /* 2907 =for apidoc_section $utility 2908 2909 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a 2910 2911 Returns the number of elements in the input C array (so you want your 2912 zero-based indices to be less than but not equal to). 2913 2914 =for apidoc Am|void *|C_ARRAY_END|void *a 2915 2916 Returns a pointer to one element past the final element of the input C array. 2917 2918 =cut 2919 2920 C_ARRAY_END is one past the last: half-open/half-closed range, not 2921 last-inclusive range. 2922 */ 2923 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0])) 2924 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a)) 2925 2926 #if defined(PERL_CORE) || defined(PERL_EXT_RE_BUILD) 2927 /* strlen() of a literal string constant. Restricting this to core, in part 2928 * because it can generate compiler warnings about comparing unlike signs */ 2929 # define STRLENs(s) (sizeof("" s "") - 1) 2930 #endif 2931 2932 #ifdef NEED_VA_COPY 2933 # ifdef va_copy 2934 # define Perl_va_copy(s, d) va_copy(d, s) 2935 # elif defined(__va_copy) 2936 # define Perl_va_copy(s, d) __va_copy(d, s) 2937 # else 2938 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list) 2939 # endif 2940 #endif 2941 2942 /* convenience debug macros */ 2943 #ifdef USE_ITHREADS 2944 #define pTHX_FORMAT "Perl interpreter: 0x%p" 2945 #define pTHX__FORMAT ", Perl interpreter: 0x%p" 2946 #define pTHX_VALUE_ (void *)my_perl, 2947 #define pTHX_VALUE (void *)my_perl 2948 #define pTHX__VALUE_ ,(void *)my_perl, 2949 #define pTHX__VALUE ,(void *)my_perl 2950 #else 2951 #define pTHX_FORMAT 2952 #define pTHX__FORMAT 2953 #define pTHX_VALUE_ 2954 #define pTHX_VALUE 2955 #define pTHX__VALUE_ 2956 #define pTHX__VALUE 2957 #endif /* USE_ITHREADS */ 2958 2959 /* 2960 Perl_deprecate was not part of the public API, and did not have a deprecate() 2961 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor 2962 CPAN::Unpack show any users outside the core. 2963 2964 =for apidoc_section $warning 2965 =for apidoc Cdm||deprecate|U32 category|"message" 2966 Wrapper around Perl_ck_warner_d() to produce a deprecated warning in the 2967 given category with an appropriate message. The C<message> argument must 2968 be a C string. The string " is deprecated" will automatically be added 2969 to the end of the C<message>. 2970 2971 =for apidoc Cdm||deprecate_disappears_in|U32 category|"when"|"message" 2972 Wrapper around Perl_ck_warner_d() to produce a deprecated warning in the 2973 given category with an appropriate message that the construct referred 2974 to by the message will disappear in a specific release. The C<when> and 2975 C<message> arguments must be a C string. The C<when> string is expected 2976 to be of the form "5.40", with no minor element in the version. The actual 2977 message output will be the result of the following expression C<message 2978 " is deprecated, and will disappear in Perl " when> which is why C<message> 2979 and C<when> must be literal C strings. 2980 2981 =for apidoc Cdm||deprecate_fatal_in|U32 category|"when"|"message" 2982 Wrapper around Perl_ck_warner_d() to produce a deprecated warning in the 2983 given category with an appropriate message that the construct referred 2984 to by the message will become fatal in a specific release. The C<when> 2985 and C<message> arguments must be a C string. The C<when> string is expected 2986 to be of the form "5.40", with no minor element in the version. The actual 2987 message output will be the result of the following expression C<message " is 2988 deprecated, and will become fatal in Perl " when> which is why C<message> 2989 and C<when> must be literal C strings. 2990 2991 =cut 2992 */ 2993 2994 #ifdef PERL_CORE 2995 # define deprecate(category,message) \ 2996 Perl_ck_warner_d(aTHX_ packWARN(category), \ 2997 message " is deprecated") 2998 2999 # define deprecate_disappears_in(category,when,message) \ 3000 Perl_ck_warner_d(aTHX_ packWARN(category), \ 3001 message " is deprecated, and will disappear in Perl " when) 3002 3003 # define deprecate_fatal_in(category,when,message) \ 3004 Perl_ck_warner_d(aTHX_ packWARN(category), \ 3005 message " is deprecated, and will become fatal in Perl " when) 3006 #endif 3007 3008 /* Internal macros to deal with gids and uids */ 3009 #ifdef PERL_CORE 3010 3011 # if Uid_t_size > IVSIZE 3012 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid)) 3013 # define SvUID(sv) SvNV(sv) 3014 # elif Uid_t_sign <= 0 3015 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid)) 3016 # define SvUID(sv) SvIV(sv) 3017 # else 3018 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid)) 3019 # define SvUID(sv) SvUV(sv) 3020 # endif /* Uid_t_size */ 3021 3022 # if Gid_t_size > IVSIZE 3023 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid)) 3024 # define SvGID(sv) SvNV(sv) 3025 # elif Gid_t_sign <= 0 3026 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid)) 3027 # define SvGID(sv) SvIV(sv) 3028 # else 3029 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid)) 3030 # define SvGID(sv) SvUV(sv) 3031 # endif /* Gid_t_size */ 3032 3033 #endif 3034 3035 /* These are simple Marsaglia XOR-SHIFT RNG's for 64 and 32 bits. These 3036 * RNG's are of reasonable quality, very fast, and have the interesting 3037 * property that provided 'x' is non-zero they create a cycle of 2^32-1 3038 * or 2^64-1 "random" like numbers, with the exception of 0. Thus they 3039 * are very useful when you want an integer to "dance" in a random way, 3040 * but you also never want it to become 0 and thus false. 3041 * 3042 * Obviously they leave x unchanged if it starts out as 0. 3043 * 3044 * We have two variants just because that can be helpful in certain 3045 * places. There is no advantage to either, they are equally bad as each 3046 * other as far RNG's go. Sufficiently random for many purposes, but 3047 * insufficiently random for serious use as they fail important tests in 3048 * the Test01 BigCrush RNG test suite by L’Ecuyer and Simard. (Note 3049 * that Drand48 also fails BigCrush). The main point is they produce 3050 * different sequences and in places where we want some randomlike 3051 * behavior they are cheap and easy. 3052 * 3053 * Marsaglia was one of the early researchers into RNG testing and wrote 3054 * the Diehard RNG test suite, which after his death become the 3055 * Dieharder RNG suite, and was generally supplanted by the Test01 suite 3056 * by L'Ecruyer and associates. 3057 * 3058 * There are dozens of shift parameters that create a pseudo random ring 3059 * of integers 1..2^N-1, if you need a different sequence just read the 3060 * paper and select a set of parameters. In fact, simply reversing the 3061 * shift order from L/R/L to R/L/R should result in another valid 3062 * example, but read the paper before you do that. 3063 * 3064 * PDF of the original paper: 3065 * https://www.jstatsoft.org/article/download/v008i14/916 3066 * Wikipedia: 3067 * https://en.wikipedia.org/wiki/Xorshift 3068 * Criticism: 3069 * https://www.iro.umontreal.ca/~lecuyer/myftp/papers/xorshift.pdf 3070 * Test01: 3071 * http://simul.iro.umontreal.ca/testu01/tu01.html 3072 * Diehard: 3073 * https://en.wikipedia.org/wiki/Diehard_tests 3074 * Dieharder: 3075 * https://webhome.phy.duke.edu/~rgb/General/rand_rate/rand_rate.abs 3076 * 3077 */ 3078 3079 /* 32 bit version */ 3080 #define PERL_XORSHIFT32_A(x) \ 3081 STMT_START { \ 3082 (x) ^= ((x) << 13); \ 3083 (x) ^= ((x) >> 17); \ 3084 (x) ^= ((x) << 5); \ 3085 } STMT_END 3086 3087 /* 64 bit version */ 3088 #define PERL_XORSHIFT64_A(x) \ 3089 STMT_START { \ 3090 (x) ^= ((x) << 13); \ 3091 (x) ^= ((x) >> 7); \ 3092 (x) ^= ((x) << 17); \ 3093 } STMT_END 3094 3095 /* 32 bit version */ 3096 #define PERL_XORSHIFT32_B(x) \ 3097 STMT_START { \ 3098 (x) ^= ((x) << 5); \ 3099 (x) ^= ((x) >> 27); \ 3100 (x) ^= ((x) << 8); \ 3101 } STMT_END 3102 3103 /* 64 bit version - currently this is unused, 3104 * it is provided here to complement the 32 bit _B 3105 * variant which IS used. */ 3106 #define PERL_XORSHIFT64_B(x) \ 3107 STMT_START { \ 3108 (x) ^= ((x) << 15); \ 3109 (x) ^= ((x) >> 49); \ 3110 (x) ^= ((x) << 26); \ 3111 } STMT_END 3112 3113 3114 #endif /* PERL_HANDY_H_ */ 3115 3116 /* 3117 * ex: set ts=8 sts=4 sw=4 et: 3118 */ 3119