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
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.
11 /* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
14 #ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
18 # define Null(type) ((type)NULL)
21 =for apidoc_section $string
22 =for apidoc AmnU||Nullch
23 Null character pointer. (No longer available when C<PERL_CORE> is
26 =for apidoc_section $SV
27 =for apidoc AmnU||Nullsv
28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
32 Below are signatures of functions from config.h which can't easily be gleaned
33 from it, and are very unlikely to change
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
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
45 =for apidoc_section $string
46 =for apidoc Amu|token|CAT2|token x|token y
47 =for apidoc Amu|string|STRINGIFY|token x
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
57 # define Nullch Null(char*)
58 # define Nullfp Null(PerlIO*)
59 # define Nullsv Null(SV*)
72 =for apidoc_section $SV
73 =for apidoc Am|void *|MUTABLE_PTR|void * p
74 =for apidoc_item |AV *|MUTABLE_AV|AV * p
75 =for apidoc_item |CV *|MUTABLE_CV|CV * p
76 =for apidoc_item |GV *|MUTABLE_GV|GV * p
77 =for apidoc_item |HV *|MUTABLE_HV|HV * p
78 =for apidoc_item |IO *|MUTABLE_IO|IO * p
79 =for apidoc_item |SV *|MUTABLE_SV|SV * p
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;
86 AV *av1 = (AV*)sv; <== BAD: the const has been silently
88 AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
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.
96 #if defined(PERL_USE_GCC_BRACE_GROUPS)
97 # define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
99 # define MUTABLE_PTR(p) ((void *) (p))
102 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
103 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
104 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
105 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
106 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
107 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
109 #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
110 # include <stdbool.h>
116 /* bool is built-in for g++-2.6.3 and later, which might be used
117 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
118 be sure _G_config.h will be included before this file. _G_config.h
119 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
120 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
121 g++ can be identified by __GNUG__.
122 Andy Dougherty February 2000
124 #ifdef __GNUG__ /* GNU g++ has bool built-in */
125 # ifndef PERL_BOOL_AS_CHAR
141 =for apidoc_section $casting
142 =for apidoc Am|bool|cBOOL|bool expr
144 Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
145 if C<bool> is defined as C<char>, for example, then the cast from C<int> is
146 implementation-defined.
148 C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
152 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
154 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
155 * XXX Should really be a Configure probe, with HAS__FUNCTION__
156 * and FUNCTION__ as results.
157 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
158 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
159 # define FUNCTION__ __func__
160 #elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
161 # define FUNCTION__ ""
163 # define FUNCTION__ __FUNCTION__ /* Common extension. */
166 /* XXX A note on the perl source internal type system. The
167 original intent was that I32 be *exactly* 32 bits.
169 Currently, we only guarantee that I32 is *at least* 32 bits.
170 Specifically, if int is 64 bits, then so is I32. (This is the case
171 for the Cray.) This has the advantage of meshing nicely with
172 standard library calls (where we pass an I32 and the library is
173 expecting an int), but the disadvantage that an I32 is not 32 bits.
174 Andy Dougherty August 1996
176 There is no guarantee that there is *any* integral type with
177 exactly 32 bits. It is perfectly legal for a system to have
178 sizeof(short) == sizeof(int) == sizeof(long) == 8.
180 Similarly, there is no guarantee that I16 and U16 have exactly 16
183 For dealing with issues that may arise from various 32/64-bit
184 systems, we will ask Configure to check out
186 SHORTSIZE == sizeof(short)
187 INTSIZE == sizeof(int)
188 LONGSIZE == sizeof(long)
189 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
190 PTRSIZE == sizeof(void *)
191 DOUBLESIZE == sizeof(double)
192 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
196 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
197 # include <inttypes.h>
198 # ifdef INT32_MIN_BROKEN
200 # define INT32_MIN (-2147483647-1)
202 # ifdef INT64_MIN_BROKEN
204 # define INT64_MIN (-9223372036854775807LL-1)
220 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
221 Please search CHAR_MAX in perl.h for further details. */
223 # define U8_MAX UINT8_MAX
225 # define U8_MAX PERL_UCHAR_MAX
228 # define U8_MIN UINT8_MIN
230 # define U8_MIN PERL_UCHAR_MIN
234 # define I16_MAX INT16_MAX
236 # define I16_MAX PERL_SHORT_MAX
239 # define I16_MIN INT16_MIN
241 # define I16_MIN PERL_SHORT_MIN
244 # define U16_MAX UINT16_MAX
246 # define U16_MAX PERL_USHORT_MAX
249 # define U16_MIN UINT16_MIN
251 # define U16_MIN PERL_USHORT_MIN
255 # define I32_MAX INT32_MAX
257 # define I32_MAX PERL_INT_MAX
259 # define I32_MAX PERL_LONG_MAX
262 # define I32_MIN INT32_MIN
264 # define I32_MIN PERL_INT_MIN
266 # define I32_MIN PERL_LONG_MIN
269 # ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
270 # define U32_MAX UINT_MAX
272 # define U32_MAX 4294967295U
275 # define U32_MAX PERL_UINT_MAX
277 # define U32_MAX PERL_ULONG_MAX
280 # define U32_MIN UINT32_MIN
282 # define U32_MIN PERL_UINT_MIN
284 # define U32_MIN PERL_ULONG_MIN
287 /* These C99 typedefs are useful sometimes for, say, loop variables whose
288 * maximum values are small, but for which speed trumps size. If we have a C99
289 * compiler, use that. Otherwise, a plain 'int' should be good enough.
291 * Restrict these to core for now until we are more certain this is a good
293 #if defined(PERL_CORE) || defined(PERL_EXT)
295 typedef int_fast8_t PERL_INT_FAST8_T;
296 typedef uint_fast8_t PERL_UINT_FAST8_T;
297 typedef int_fast16_t PERL_INT_FAST16_T;
298 typedef uint_fast16_t PERL_UINT_FAST16_T;
300 typedef int PERL_INT_FAST8_T;
301 typedef unsigned int PERL_UINT_FAST8_T;
302 typedef int PERL_INT_FAST16_T;
303 typedef unsigned int PERL_UINT_FAST16_T;
307 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
308 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
309 * required to represent any possible unsigned number containing N bits.
310 * TYPE_DIGITS gives the number of decimal digits required to represent any
311 * possible unsigned number of type T. */
312 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
313 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
314 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
316 /* Unused by core; should be deprecated */
317 #define Ctl(ch) ((ch) & 037)
319 #if defined(PERL_CORE) || defined(PERL_EXT)
321 # define MIN(a,b) ((a) < (b) ? (a) : (b))
324 # define MAX(a,b) ((a) > (b) ? (a) : (b))
328 /* Returns a boolean as to whether the input unsigned number is a power of 2
329 * (2**0, 2**1, etc). In other words if it has just a single bit set.
330 * If not, subtracting 1 would leave the uppermost bit set, so the & would
332 #if defined(PERL_CORE) || defined(PERL_EXT)
333 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
336 /* Returns a mask with the lowest n bits set */
337 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
339 /* The largest unsigned number that will fit into n bits */
340 #define nBIT_UMAX(n) nBIT_MASK(n)
343 =for apidoc_section $directives
344 =for apidoc Am||__ASSERT_|bool expr
346 This is a helper macro to avoid preprocessor issues, replaced by nothing
347 unless under DEBUGGING, where it expands to an assert of its argument,
348 followed by a comma (hence the comma operator). If we just used a straight
349 assert(), we would get a comma with nothing before it when not DEBUGGING.
353 We also use empty definition under Coverity since the __ASSERT__
354 checks often check for things that Really Cannot Happen, and Coverity
355 detects that and gets all excited. */
357 #if defined(DEBUGGING) && !defined(__COVERITY__) \
358 && ! defined(PERL_SMALL_MACRO_BUFFER)
359 # define __ASSERT_(statement) assert(statement),
361 # define __ASSERT_(statement)
365 =for apidoc_section $SV
367 =for apidoc Ama|SV*|newSVpvs|"literal string"
368 Like C<newSVpvn>, but takes a literal string instead of a
371 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
372 Like C<newSVpvn_flags>, but takes a literal string instead of
373 a string/length pair.
375 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
376 Like C<newSVpvn_share>, but takes a literal string instead of
377 a string/length pair and omits the hash parameter.
379 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
380 Like C<sv_catpvn_flags>, but takes a literal string instead
381 of a string/length pair.
383 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
384 Like C<sv_catpvn_nomg>, but takes a literal string instead of
385 a string/length pair.
387 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
388 Like C<sv_catpvn>, but takes a literal string instead of a
391 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
392 Like C<sv_catpvn_mg>, but takes a literal string instead of a
395 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
396 Like C<sv_setpvn>, but takes a literal string instead of a
399 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
400 Like C<sv_setpvn_mg>, but takes a literal string instead of a
403 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
404 Like C<sv_setref_pvn>, but takes a literal string instead of
405 a string/length pair.
407 =for apidoc_section $string
409 =for apidoc Ama|char*|savepvs|"literal string"
410 Like C<savepvn>, but takes a literal string instead of a
413 =for apidoc Ama|char*|savesharedpvs|"literal string"
414 A version of C<savepvs()> which allocates the duplicate string in memory
415 which is shared between threads.
417 =for apidoc_section $GV
419 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
420 Like C<gv_stashpvn>, but takes a literal string instead of a
423 =for apidoc_section $HV
425 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
426 Like C<hv_fetch>, but takes a literal string instead of a
429 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
430 Like C<hv_store>, but takes a literal string instead of a
432 and omits the hash parameter.
434 =for apidoc_section $lexer
436 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
438 Like L</lex_stuff_pvn>, but takes a literal string instead of
439 a string/length pair.
445 =for apidoc_section $string
447 =for apidoc Amu|pair|STR_WITH_LEN|"literal string"
449 Returns two comma separated tokens of the input literal string, and its length.
450 This is convenience macro which helps out in some API calls.
451 Note that it can't be used as an argument to macros or functions that under
452 some configurations might be macros, which means that it requires the full
453 Perl_xxx(aTHX_ ...) form for any API calls where it's used.
458 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
460 /* STR_WITH_LEN() shortcuts */
461 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
462 #define newSVpvs_flags(str,flags) \
463 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
464 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
465 #define sv_catpvs_flags(sv, str, flags) \
466 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
467 #define sv_catpvs_nomg(sv, str) \
468 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
469 #define sv_catpvs(sv, str) \
470 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
471 #define sv_catpvs_mg(sv, str) \
472 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
473 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
474 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
475 #define sv_setref_pvs(rv, classname, str) \
476 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
477 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
478 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
479 #define gv_stashpvs(str, create) \
480 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
482 #define gv_fetchpvs(namebeg, flags, sv_type) \
483 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type)
484 #define gv_fetchpvn gv_fetchpvn_flags
485 #define sv_catxmlpvs(dsv, str, utf8) \
486 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
489 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
491 #define get_cvs(str, flags) \
492 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
494 /* internal helpers */
496 #ifndef PERL_VERSION_MAJOR
497 # define PERL_VERSION_MAJOR PERL_REVISION
499 # undef PERL_REVISION /* We don't want code to be using these */
501 #ifndef PERL_VERSION_MINOR
502 # define PERL_VERSION_MINOR PERL_VERSION
506 #ifndef PERL_VERSION_PATCH
507 # define PERL_VERSION_PATCH PERL_SUBVERSION
509 # undef PERL_SUBVERSION
512 #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
513 /* '10*' leaves room for things like alpha, beta, releases */ \
514 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
515 #define PERL_DECIMAL_VERSION_ \
516 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
520 =for apidoc_section $versioning
521 =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
522 =for apidoc_item PERL_VERSION_NE
523 =for apidoc_item PERL_VERSION_LT
524 =for apidoc_item PERL_VERSION_LE
525 =for apidoc_item PERL_VERSION_GT
526 =for apidoc_item PERL_VERSION_GE
528 Returns whether or not the perl currently being compiled has the specified
529 relationship to the perl given by the parameters. For example,
531 #if PERL_VERSION_GT(5,24,2)
532 code that will only be compiled on perls after v5.24.2
537 Note that this is usable in making compile-time decisions
539 You may use the special value '*' for the final number to mean ALL possible
542 #if PERL_VERSION_EQ(5,31,'*')
544 means all perls in the 5.31 series. And
546 #if PERL_VERSION_NE(5,24,'*')
548 means all perls EXCEPT 5.24 ones. And
550 #if PERL_VERSION_LE(5,9,'*')
554 #if PERL_VERSION_LT(5,10,0)
556 This means you don't have to think so much when converting from the existing
557 deprecated C<PERL_VERSION> to using this macro:
559 #if PERL_VERSION <= 9
563 #if PERL_VERSION_LE(5,9,'*')
568 /* N.B. These don't work if the patch version is 42 or 92, as those are what
569 * '*' is in ASCII and EBCDIC respectively */
570 # define PERL_VERSION_EQ(j,n,p) \
572 ? ( (j) == PERL_VERSION_MAJOR \
573 && (n) == PERL_VERSION_MINOR) \
574 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
575 # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
577 # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
578 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
580 (((p) == '*') ? 0 : p)))
581 # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
583 # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
584 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
585 (((p) == '*') ? ((n)+1) : (n)), \
586 (((p) == '*') ? 0 : p)))
587 # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
590 =for apidoc_section $string
592 =for apidoc Am|bool|strNE|char* s1|char* s2
593 Test two C<NUL>-terminated strings to see if they are different. Returns true
596 =for apidoc Am|bool|strEQ|char* s1|char* s2
597 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
600 =for apidoc Am|bool|strLT|char* s1|char* s2
601 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
602 second, C<s2>. Returns true or false.
604 =for apidoc Am|bool|strLE|char* s1|char* s2
605 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
606 equal to the second, C<s2>. Returns true or false.
608 =for apidoc Am|bool|strGT|char* s1|char* s2
609 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
610 the second, C<s2>. Returns true or false.
612 =for apidoc Am|bool|strGE|char* s1|char* s2
613 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
614 or equal to the second, C<s2>. Returns true or false.
616 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
617 Test two C<NUL>-terminated strings to see if they are different. The C<len>
618 parameter indicates the number of bytes to compare. Returns true or false. (A
619 wrapper for C<strncmp>).
621 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
622 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
623 parameter indicates the number of bytes to compare. Returns true or false. (A
624 wrapper for C<strncmp>).
626 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
627 Test two buffers (which may contain embedded C<NUL> characters, to see if they
628 are equal. The C<len> parameter indicates the number of bytes to compare.
629 Returns zero if equal, or non-zero if non-equal.
631 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
632 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
633 C<l1> gives the number of bytes in C<s1>.
634 Returns zero if equal, or non-zero if non-equal.
636 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
637 Test two buffers (which may contain embedded C<NUL> characters, to see if they
638 are not equal. The C<len> parameter indicates the number of bytes to compare.
639 Returns zero if non-equal, or non-zero if equal.
641 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
642 Like L</memNE>, but the second string is a literal enclosed in double quotes,
643 C<l1> gives the number of bytes in C<s1>.
644 Returns zero if non-equal, or zero if non-equal.
646 =for apidoc Am|bool|memCHRs|"list"|char c
647 Returns the position of the first occurence of the byte C<c> in the literal
648 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
649 treated as unsigned char. Thus this macro can be used to determine if C<c> is
650 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
651 is C<NUL> (and the set doesn't include C<NUL>).
655 New macros should use the following conventions for their names (which are
656 based on the underlying C library functions):
658 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
660 Each has two main parameters, string-like operands that are compared
661 against each other, as specified by the macro name. Some macros may
662 additionally have one or potentially even two length parameters. If a length
663 parameter applies to both string parameters, it will be positioned third;
664 otherwise any length parameter immediately follows the string parameter it
667 If the prefix to the name is 'str', the string parameter is a pointer to a C
668 language string. Such a string does not contain embedded NUL bytes; its
669 length may be unknown, but can be calculated by C<strlen()>, since it is
670 terminated by a NUL, which isn't included in its length.
672 The optional 'n' following 'str' means that there is a third parameter,
673 giving the maximum number of bytes to look at in each string. Even if both
674 strings are longer than the length parameter, those extra bytes will be
677 The 's' suffix means that the 2nd byte string parameter is a literal C
678 double-quoted string. Its length will automatically be calculated by the
679 macro, so no length parameter will ever be needed for it.
681 If the prefix is 'mem', the string parameters don't have to be C strings;
682 they may contain embedded NUL bytes, do not necessarily have a terminating
683 NUL, and their lengths can be known only through other means, which in
684 practice are additional parameter(s) passed to the function. All 'mem'
685 functions have at least one length parameter. Barring any 'l' or 's' suffix,
686 there is a single length parameter, in position 3, which applies to both
687 string parameters. The 's' suffix means, as described above, that the 2nd
688 string is a literal double-quoted C string (hence its length is calculated by
689 the macro, and the length parameter to the function applies just to the first
690 string parameter, and hence is positioned just after it). An 'l' suffix
691 means that the 2nd string parameter has its own length parameter, and the
692 signature will look like memFOOl(s1, l1, s2, l2).
694 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
695 substring of the 1st string. 'P' if present indicates that the substring
696 must be a "proper" one in tha mathematical sense that the first one must be
697 strictly larger than the 2nd.
702 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
703 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
704 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
705 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
706 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
707 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
709 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
710 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
712 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
713 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
715 /* memEQ and memNE where second comparand is a string constant */
716 #define memEQs(s1, l, s2) \
717 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
718 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
720 /* Keep these private until we decide it was a good idea */
721 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
723 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
725 #define memBEGINs(s1, l, s2) \
726 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
727 && memEQ(s1, "" s2 "", sizeof(s2)-1))
728 #define memBEGINPs(s1, l, s2) \
729 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
730 && memEQ(s1, "" s2 "", sizeof(s2)-1))
731 #define memENDs(s1, l, s2) \
732 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
733 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
734 #define memENDPs(s1, l, s2) \
735 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
736 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
737 #endif /* End of making macros private */
739 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
740 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
741 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
742 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
744 #define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
749 * Unfortunately, the introduction of locales means that we
750 * can't trust isupper(), etc. to tell the truth. And when
751 * it comes to /\w+/ with tainting enabled, we *must* be able
752 * to trust our character classes.
754 * Therefore, the default tests in the text of Perl will be
755 * independent of locale. Any code that wants to depend on
756 * the current locale will use the tests that begin with "lc".
759 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
767 =head1 Character classification
768 This section is about functions (really macros) that classify characters
769 into types, such as punctuation versus alphabetic, etc. Most of these are
770 analogous to regular expression character classes. (See
771 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
772 each class. (Not all macros have all variants; each item below lists the
773 ones valid for it.) None are affected by C<use bytes>, and only the ones
774 with C<LC> in the name are affected by the current locale.
776 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
777 treating it as a code point, and returns a boolean as to whether or not the
778 character represented by it is (or on non-ASCII platforms, corresponds to) an
779 ASCII character in the named class based on platform, Unicode, and Perl rules.
780 If the input is a number that doesn't fit in an octet, FALSE is returned.
782 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
783 with no suffix C<"_A">. This variant is used to emphasize by its name that
784 only ASCII-range characters can return TRUE.
786 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
787 onto the platform. That is, the code points that are ASCII are unaffected,
788 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
789 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
790 true when called with the code point 0xDF, which is a word character in both
791 ASCII and EBCDIC (though it represents different characters in each).
792 If the input is a number that doesn't fit in an octet, FALSE is returned.
793 (Perl's documentation uses a colloquial definition of Latin-1, to include all
794 code points below 256.)
796 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
797 inputs below 256, but if the code point is larger than 255, Unicode rules are
798 used to determine if it is in the character class. For example,
799 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
800 WITH MACRON in Unicode, and is a word character.
802 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
803 but are used for UTF-8 encoded strings. The two forms are different names for
804 the same thing. Each call to one of these classifies the first character of
805 the string starting at C<p>. The second parameter, C<e>, points to anywhere in
806 the string beyond the first character, up to one byte past the end of the
807 entire string. Although both variants are identical, the suffix C<_safe> in
808 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
809 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
810 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
811 some way, the program may croak, or the function may return FALSE, at the
812 discretion of the implementation, and subject to change in future releases.
814 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
815 but the result is based on the current locale, which is what C<LC> in the name
816 stands for. If Perl can determine that the current locale is a UTF-8 locale,
817 it uses the published Unicode rules; otherwise, it uses the C library function
818 that gives the named classification. For example, C<isDIGIT_LC()> when not in
819 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
820 returned if the input won't fit into an octet. On some platforms where the C
821 library function is known to be defective, Perl changes its result to follow
822 the POSIX standard's rules.
824 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
825 than 256, but for larger ones it returns the Unicode classification of the code
828 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
829 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
830 are different names for the same thing. Each call to one of these classifies
831 the first character of the string starting at C<p>. The second parameter,
832 C<e>, points to anywhere in the string beyond the first character, up to one
833 byte past the end of the entire string. Although both variants are identical,
834 the suffix C<_safe> in one name emphasizes that it will not attempt to read
835 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
836 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
837 character is malformed in some way, the program may croak, or the function may
838 return FALSE, at the discretion of the implementation, and subject to change in
841 =for apidoc Am|bool|isALPHA|UV ch
842 =for apidoc_item ||isALPHA_A|UV ch
843 =for apidoc_item ||isALPHA_L1|UV ch
844 =for apidoc_item ||isALPHA_uvchr|UV ch
845 =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
846 =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
847 =for apidoc_item ||isALPHA_LC|UV ch
848 =for apidoc_item ||isALPHA_LC_uvchr|UV ch
849 =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
850 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
851 analogous to C<m/[[:alpha:]]/>.
852 See the L<top of this section|/Character classification> for an explanation of
857 Here and below, we add the prototypes of these macros for downstream programs
858 that would be interested in them, such as Devel::PPPort
860 =for apidoc Am|bool|isALPHANUMERIC|UV ch
861 =for apidoc_item ||isALPHANUMERIC_A|UV ch
862 =for apidoc_item ||isALPHANUMERIC_L1|UV ch
863 =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
864 =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
865 =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
866 =for apidoc_item ||isALPHANUMERIC_LC|UV ch
867 =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
868 =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
869 =for apidoc_item ||isALNUMC|UV ch
870 =for apidoc_item ||isALNUMC_A|UV ch
871 =for apidoc_item ||isALNUMC_L1|UV ch
872 =for apidoc_item ||isALNUMC_LC|UV ch
873 =for apidoc_item ||isALNUMC_LC_uvchr|UV ch
874 Returns a boolean indicating whether the specified character is one of
875 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
876 See the L<top of this section|/Character classification> for an explanation of
879 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
880 this corresponds to the C language alphanumeric definition). Also
881 there are the variants
882 C<isALNUMC_A>, C<isALNUMC_L1>
883 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
885 =for apidoc Am|bool|isASCII|UV ch
886 =for apidoc_item ||isASCII_A|UV ch
887 =for apidoc_item ||isASCII_L1|UV ch
888 =for apidoc_item ||isASCII_uvchr|UV ch
889 =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
890 =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
891 =for apidoc_item ||isASCII_LC|UV ch
892 =for apidoc_item ||isASCII_LC_uvchr|UV ch
893 =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
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
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.
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.
911 =for apidoc Am|bool|isBLANK|UV ch
912 =for apidoc_item ||isBLANK_A|UV ch
913 =for apidoc_item ||isBLANK_L1|UV ch
914 =for apidoc_item ||isBLANK_uvchr|UV ch
915 =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
916 =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
917 =for apidoc_item ||isBLANK_LC|UV ch
918 =for apidoc_item ||isBLANK_LC_uvchr|UV ch
919 =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
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
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.
929 =for apidoc Am|bool|isCNTRL|UV ch
930 =for apidoc_item ||isCNTRL_A|UV ch
931 =for apidoc_item ||isCNTRL_L1|UV ch
932 =for apidoc_item ||isCNTRL_uvchr|UV ch
933 =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
934 =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
935 =for apidoc_item ||isCNTRL_LC|UV ch
936 =for apidoc_item ||isCNTRL_LC_uvchr|UV ch
937 =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
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
943 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
945 =for apidoc Am|bool|isDIGIT|UV ch
946 =for apidoc_item ||isDIGIT_A|UV ch
947 =for apidoc_item ||isDIGIT_L1|UV ch
948 =for apidoc_item ||isDIGIT_uvchr|UV ch
949 =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
950 =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
951 =for apidoc_item ||isDIGIT_LC|UV ch
952 =for apidoc_item ||isDIGIT_LC_uvchr|UV ch
953 =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
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
961 =for apidoc Am|bool|isGRAPH|UV ch
962 =for apidoc_item ||isGRAPH_A|UV ch
963 =for apidoc_item ||isGRAPH_L1|UV ch
964 =for apidoc_item ||isGRAPH_uvchr|UV ch
965 =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
966 =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
967 =for apidoc_item ||isGRAPH_LC|UV ch
968 =for apidoc_item ||isGRAPH_LC_uvchr|UV ch
969 =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
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
975 =for apidoc Am|bool|isLOWER|UV ch
976 =for apidoc_item ||isLOWER_A|UV ch
977 =for apidoc_item ||isLOWER_L1|UV ch
978 =for apidoc_item ||isLOWER_uvchr|UV ch
979 =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
980 =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
981 =for apidoc_item ||isLOWER_LC|UV ch
982 =for apidoc_item ||isLOWER_LC_uvchr|UV ch
983 =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
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
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
994 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
997 =for apidoc Am|bool|isPUNCT|UV ch
998 =for apidoc_item ||isPUNCT_A|UV ch
999 =for apidoc_item ||isPUNCT_L1|UV ch
1000 =for apidoc_item ||isPUNCT_uvchr|UV ch
1001 =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
1002 =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
1003 =for apidoc_item ||isPUNCT_LC|UV ch
1004 =for apidoc_item ||isPUNCT_LC_uvchr|UV ch
1005 =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
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
1014 =for apidoc Am|bool|isSPACE|UV ch
1015 =for apidoc_item ||isSPACE_A|UV ch
1016 =for apidoc_item ||isSPACE_L1|UV ch
1017 =for apidoc_item ||isSPACE_uvchr|UV ch
1018 =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
1019 =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
1020 =for apidoc_item ||isSPACE_LC|UV ch
1021 =for apidoc_item ||isSPACE_LC_uvchr|UV ch
1022 =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
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
1034 =for apidoc Am|bool|isPSXSPC|UV ch
1035 =for apidoc_item ||isPSXSPC_A|UV ch
1036 =for apidoc_item ||isPSXSPC_L1|UV ch
1037 =for apidoc_item ||isPSXSPC_uvchr|UV ch
1038 =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1039 =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
1040 =for apidoc_item ||isPSXSPC_LC|UV ch
1041 =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
1042 =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
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
1055 =for apidoc Am|bool|isUPPER|UV ch
1056 =for apidoc_item ||isUPPER_A|UV ch
1057 =for apidoc_item ||isUPPER_L1|UV ch
1058 =for apidoc_item ||isUPPER_uvchr|UV ch
1059 =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1060 =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
1061 =for apidoc_item ||isUPPER_LC|UV ch
1062 =for apidoc_item ||isUPPER_LC_uvchr|UV ch
1063 =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
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
1069 =for apidoc Am|bool|isPRINT|UV ch
1070 =for apidoc_item ||isPRINT_A|UV ch
1071 =for apidoc_item ||isPRINT_L1|UV ch
1072 =for apidoc_item ||isPRINT_uvchr|UV ch
1073 =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1074 =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
1075 =for apidoc_item ||isPRINT_LC|UV ch
1076 =for apidoc_item ||isPRINT_LC_uvchr|UV ch
1077 =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
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
1083 =for apidoc Am|bool|isWORDCHAR|UV ch
1084 =for apidoc_item ||isWORDCHAR_A|UV ch
1085 =for apidoc_item ||isWORDCHAR_L1|UV ch
1086 =for apidoc_item ||isWORDCHAR_uvchr|UV ch
1087 =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1088 =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
1089 =for apidoc_item ||isWORDCHAR_LC|UV ch
1090 =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
1091 =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1092 =for apidoc_item ||isALNUM|UV ch
1093 =for apidoc_item ||isALNUM_A|UV ch
1094 =for apidoc_item ||isALNUM_LC|UV ch
1095 =for apidoc_item ||isALNUM_LC_uvchr|UV ch
1096 Returns a boolean indicating whether the specified character is a character
1097 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1098 in a regular expression. A word character is an alphabetic character, a
1099 decimal digit, a connecting punctuation character (such as an underscore), or
1100 a "mark" character that attaches to one of those (like some sort of accent).
1101 C<isALNUM()> is a synonym provided for backward compatibility, even though a
1102 word character includes more than the standard C language meaning of
1104 See the L<top of this section|/Character classification> for an explanation of
1106 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1107 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1108 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1109 include the platform's native underscore.
1111 =for apidoc Am|bool|isXDIGIT|UV ch
1112 =for apidoc_item ||isXDIGIT_A|UV ch
1113 =for apidoc_item ||isXDIGIT_L1|UV ch
1114 =for apidoc_item ||isXDIGIT_uvchr|UV ch
1115 =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1116 =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
1117 =for apidoc_item ||isXDIGIT_LC|UV ch
1118 =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
1119 =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1120 Returns a boolean indicating whether the specified character is a hexadecimal
1121 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1122 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1123 See the L<top of this section|/Character classification> for an explanation of
1126 =for apidoc Am|bool|isIDFIRST|UV ch
1127 =for apidoc_item ||isIDFIRST_A|UV ch
1128 =for apidoc_item ||isIDFIRST_L1|UV ch
1129 =for apidoc_item ||isIDFIRST_uvchr|UV ch
1130 =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1131 =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
1132 =for apidoc_item ||isIDFIRST_LC|UV ch
1133 =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
1134 =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1135 Returns a boolean indicating whether the specified character can be the first
1136 character of an identifier. This is very close to, but not quite the same as
1137 the official Unicode property C<XID_Start>. The difference is that this
1138 returns true only if the input character also matches L</isWORDCHAR>.
1139 See the L<top of this section|/Character classification> for an explanation of
1142 =for apidoc Am|bool|isIDCONT|UV ch
1143 =for apidoc_item ||isIDCONT_A|UV ch
1144 =for apidoc_item ||isIDCONT_L1|UV ch
1145 =for apidoc_item ||isIDCONT_uvchr|UV ch
1146 =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1147 =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
1148 =for apidoc_item ||isIDCONT_LC|UV ch
1149 =for apidoc_item ||isIDCONT_LC_uvchr|UV ch
1150 =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1151 Returns a boolean indicating whether the specified character can be the
1152 second or succeeding character of an identifier. This is very close to, but
1153 not quite the same as the official Unicode property C<XID_Continue>. The
1154 difference is that this returns true only if the input character also matches
1155 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1156 an explanation of the variants.
1158 =for apidoc_section $numeric
1160 =for apidoc Am|U8|READ_XDIGIT|char str*
1161 Returns the value of an ASCII-range hex digit and advances the string pointer.
1162 Behaviour is only well defined when isXDIGIT(*str) is true.
1164 =head1 Character case changing
1165 Perl uses "full" Unicode case mappings. This means that converting a single
1166 character to another case may result in a sequence of more than one character.
1167 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1168 character sequence C<SS>. This presents some complications The lowercase of
1169 all characters in the range 0..255 is a single character, and thus
1170 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1171 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1172 an API that does allow every possible legal result to be returned.) Likewise
1173 no other function that is crippled by not being able to give the correct
1174 results for the full range of possible inputs has been implemented here.
1176 =for apidoc Am|U8|toUPPER|int ch
1177 Converts the specified character to uppercase. If the input is anything but an
1178 ASCII lowercase character, that input character itself is returned. Variant
1179 C<toUPPER_A> is equivalent.
1181 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1182 Converts the code point C<cp> to its uppercase version, and
1183 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1184 point is interpreted as native if less than 256; otherwise as Unicode. Note
1185 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1186 bytes since the uppercase version may be longer than the original character.
1188 The first code point of the uppercased version is returned
1189 (but note, as explained at L<the top of this section|/Character case
1190 changing>, that there may be more.)
1192 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1193 =for apidoc_item toUPPER_utf8_safe
1194 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1195 extending no further than S<C<e - 1>> to its uppercase version, and
1196 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1197 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1198 bytes since the uppercase version may be longer than the original character.
1200 The first code point of the uppercased version is returned
1201 (but note, as explained at L<the top of this section|/Character case
1202 changing>, that there may be more).
1204 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1205 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1206 the UTF-8 for the input character is malformed in some way, the program may
1207 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1208 of the implementation, and subject to change in future releases.
1210 C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1212 =for apidoc Am|U8|toFOLD|U8 ch
1213 Converts the specified character to foldcase. If the input is anything but an
1214 ASCII uppercase character, that input character itself is returned. Variant
1215 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
1216 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
1218 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1219 Converts the code point C<cp> to its foldcase version, and
1220 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1221 point is interpreted as native if less than 256; otherwise as Unicode. Note
1222 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1223 bytes since the foldcase version may be longer than the original character.
1225 The first code point of the foldcased version is returned
1226 (but note, as explained at L<the top of this section|/Character case
1227 changing>, that there may be more).
1229 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1230 =for apidoc_item toFOLD_utf8_safe
1231 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1232 extending no further than S<C<e - 1>> to its foldcase version, and
1233 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1234 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1235 bytes since the foldcase version may be longer than the original character.
1237 The first code point of the foldcased version is returned
1238 (but note, as explained at L<the top of this section|/Character case
1239 changing>, that there may be more).
1242 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1243 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1244 input character is malformed in some way, the program may croak, or the
1245 function may return the REPLACEMENT CHARACTER, at the discretion of the
1246 implementation, and subject to change in future releases.
1248 C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1250 =for apidoc Am|UV|toLOWER|UV cp
1251 =for apidoc_item |UV|toLOWER_A|UV cp
1252 =for apidoc_item |UV|toLOWER_L1|UV cp
1253 =for apidoc_item |UV|toLOWER_LATIN1|UV cp
1254 =for apidoc_item |UV|toLOWER_LC|UV cp
1255 =for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1256 =for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1257 =for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1259 These all return the lowercase of a character. The differences are what domain
1260 they operate on, and whether the input is specified as a code point (those
1261 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1262 case, the code point to use is the first one in the buffer of UTF-8 encoded
1263 code points, delineated by the arguments S<C<p .. e - 1>>.
1265 C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the
1266 lowercase of any uppercase ASCII-range code point. All other inputs are
1267 returned unchanged. Since these are macros, the input type may be any integral
1268 one, and the output will occupy the same number of bits as the input.
1270 C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave
1271 identically as C<toLOWER> for ASCII-range input. But additionally will return
1272 the lowercase of any uppercase code point in the entire 0..255 range, assuming
1273 a Latin-1 encoding (or the EBCDIC equivalent on such platforms).
1275 C<toLOWER_LC> returns the lowercase of the input code point according to the
1276 rules of the current POSIX locale. Input code points outside the range 0..255
1277 are returned unchanged.
1279 C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return
1280 value is identical to that of C<toLOWER_L1> for input code points in the 0..255
1281 range. The lowercase of the vast majority of Unicode code points is the same
1282 as the code point itself. For these, and for code points above the legal
1283 Unicode maximum, this returns the input code point unchanged. It additionally
1284 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1285 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1286 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1288 NOTE: the lowercase of a code point may be more than one code point. The
1289 return value of this function is only the first of these. The entire lowercase
1290 is returned in C<s>. To determine if the result is more than a single code
1291 point, you can do something like this:
1293 uc = toLOWER_uvchr(cp, s, &len);
1294 if (len > UTF8SKIP(s)) { is multiple code points }
1295 else { is a single code point }
1297 C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only
1298 difference between these and C<toLOWER_uvchr> is that the source for these is
1299 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1300 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1301 buffer may certainly contain more than one code point; but only the first one
1302 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1303 malformed in some way, the program may croak, or the function may return the
1304 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1305 change in future releases.
1307 =for apidoc Am|U8|toTITLE|U8 ch
1308 Converts the specified character to titlecase. If the input is anything but an
1309 ASCII lowercase character, that input character itself is returned. Variant
1310 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1311 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1312 not a concept used in locale handling, so there is no functionality for that.)
1314 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1315 Converts the code point C<cp> to its titlecase version, and
1316 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1317 point is interpreted as native if less than 256; otherwise as Unicode. Note
1318 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1319 bytes since the titlecase version may be longer than the original character.
1321 The first code point of the titlecased version is returned
1322 (but note, as explained at L<the top of this section|/Character case
1323 changing>, that there may be more).
1325 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1326 =for apidoc_item toTITLE_utf8_safe
1327 Convert the first UTF-8 encoded character in the sequence starting at C<p> and
1328 extending no further than S<C<e - 1>> to its titlecase version, and
1329 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1330 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1331 bytes since the titlecase version may be longer than the original character.
1333 The first code point of the titlecased version is returned
1334 (but note, as explained at L<the top of this section|/Character case
1335 changing>, that there may be more).
1338 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1339 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1340 input character is malformed in some way, the program may croak, or the
1341 function may return the REPLACEMENT CHARACTER, at the discretion of the
1342 implementation, and subject to change in future releases.
1344 C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1348 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1349 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1350 and aren't general purpose as they don't work on U+DF, and assert against that.
1352 Note that these macros are repeated in Devel::PPPort, so should also be
1353 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1358 void below because that's the best fit, and works for Devel::PPPort
1359 =for apidoc_section $integer
1360 =for apidoc AyT||WIDEST_UTYPE
1362 Yields the widest unsigned integer type on the platform, currently either
1363 C<U32> or C<U64>. This can be used in declarations such as
1369 my_uv = (WIDEST_UTYPE) val;
1375 # define WIDEST_UTYPE U64
1377 # define WIDEST_UTYPE U32
1380 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1381 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1382 * bits of information are lost even on a 64-bit machine, but to get the
1383 * compiler to optimize it out if possible. This is because Configure makes
1384 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1385 * sizeof() guarantees that this evaluates to a constant true at compile time.
1387 * For Coverity, be always true, because otherwise Coverity thinks
1388 * it finds several expressions that are always true, independent
1389 * of operands. Well, they are, but that is kind of the point.
1391 #ifndef __COVERITY__
1392 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1394 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1395 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1397 #define FITS_IN_8_BITS(c) (1)
1400 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
1401 * Written this way so that after optimization, only one conditional test is
1402 * needed. (The NV casts stop any warnings about comparison always being true
1403 * if called with an unsigned. The cast preserves the sign, which is all we
1405 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1406 __ASSERT_((NV) (n) >= 0) \
1407 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
1409 /* Returns true if c is in the range l..u, where 'l' is non-negative
1410 * Written this way so that after optimization, only one conditional test is
1412 #define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
1413 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1414 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1415 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1416 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
1419 # ifndef _ALL_SOURCE
1420 /* The native libc isascii() et.al. functions return the wrong results
1421 * on at least z/OS unless this is defined. */
1422 # error _ALL_SOURCE should probably be defined
1425 /* There is a simple definition of ASCII for ASCII platforms. But the
1426 * EBCDIC one isn't so simple, so is defined using table look-up like the
1427 * other macros below.
1429 * The cast here is used instead of '(c) >= 0', because some compilers emit
1430 * a warning that that test is always true when the parameter is an
1431 * unsigned type. khw supposes that it could be written as
1432 * && ((c) == '\0' || (c) > 0)
1433 * to avoid the message, but the cast will likely avoid extra branches even
1434 * with stupid compilers.
1436 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1438 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1441 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1442 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1443 * can be ignored. If the rest match '0', we have an octal */
1444 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1446 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1448 /* Character class numbers. For internal core Perl use only. The ones less
1449 * than 32 are used in PL_charclass[] and the ones up through the one that
1450 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1451 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1452 * their actual definitions are here. If that file has a name not used here,
1455 * The first group of these is ordered in what I (khw) estimate to be the
1456 * frequency of their use. This gives a slight edge to exiting a loop earlier
1457 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1458 * real Posix character class, and some (small) inefficiencies in regular
1459 * expression handling would be introduced by putting it in the middle of those
1460 * that are. Also, cntrl and ascii come after the others as it may be useful
1461 * to group these which have no members that match above Latin1, (or above
1462 * ASCII in the latter case) */
1464 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1465 # define _CC_DIGIT 1 /* \d and [:digit:] */
1466 # define _CC_ALPHA 2 /* [:alpha:] */
1467 # define _CC_LOWER 3 /* [:lower:] */
1468 # define _CC_UPPER 4 /* [:upper:] */
1469 # define _CC_PUNCT 5 /* [:punct:] */
1470 # define _CC_PRINT 6 /* [:print:] */
1471 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1472 # define _CC_GRAPH 8 /* [:graph:] */
1473 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1474 # define _CC_SPACE 10 /* \s, [:space:] */
1475 # define _CC_BLANK 11 /* [:blank:] */
1476 # define _CC_XDIGIT 12 /* [:xdigit:] */
1477 # define _CC_CNTRL 13 /* [:cntrl:] */
1478 # define _CC_ASCII 14 /* [:ascii:] */
1479 # define _CC_VERTSPACE 15 /* \v */
1481 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1483 /* The members of the third group below do not need to be coordinated with data
1484 * structures in regcomp.[ch] and regexec.c. */
1485 # define _CC_IDFIRST 16
1486 # define _CC_CHARNAME_CONT 17
1487 # define _CC_NONLATIN1_FOLD 18
1488 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1489 # define _CC_QUOTEMETA 20
1490 # define _CC_NON_FINAL_FOLD 21
1491 # define _CC_IS_IN_SOME_FOLD 22
1492 # define _CC_BINDIGIT 23
1493 # define _CC_OCTDIGIT 24
1494 # define _CC_MNEMONIC_CNTRL 25
1496 /* This next group is only used on EBCDIC platforms, so theoretically could be
1497 * shared with something entirely different that's only on ASCII platforms */
1498 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1500 * If more bits are needed, one could add a second word for non-64bit
1501 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1502 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1503 * is used only for optimization (as of this writing), and differs in the
1504 * Latin1 range from the ALPHA bit only in two relatively unimportant
1505 * characters: the masculine and feminine ordinal indicators, so removing it
1506 * would just cause /i regexes which match them to run less efficiently.
1507 * Similarly the EBCDIC-only bits are used just for speed, and could be
1508 * replaced by other means */
1510 #if defined(PERL_CORE) || defined(PERL_EXT)
1511 /* An enum version of the character class numbers, to help compilers
1514 _CC_ENUM_ALPHA = _CC_ALPHA,
1515 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1516 _CC_ENUM_ASCII = _CC_ASCII,
1517 _CC_ENUM_BLANK = _CC_BLANK,
1518 _CC_ENUM_CASED = _CC_CASED,
1519 _CC_ENUM_CNTRL = _CC_CNTRL,
1520 _CC_ENUM_DIGIT = _CC_DIGIT,
1521 _CC_ENUM_GRAPH = _CC_GRAPH,
1522 _CC_ENUM_LOWER = _CC_LOWER,
1523 _CC_ENUM_PRINT = _CC_PRINT,
1524 _CC_ENUM_PUNCT = _CC_PUNCT,
1525 _CC_ENUM_SPACE = _CC_SPACE,
1526 _CC_ENUM_UPPER = _CC_UPPER,
1527 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1528 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1529 _CC_ENUM_XDIGIT = _CC_XDIGIT
1530 } _char_class_number;
1533 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1537 EXTCONST U32 PL_charclass[] = {
1538 # include "l1_char_class_tab.h"
1541 # else /* ! DOINIT */
1542 EXTCONST U32 PL_charclass[];
1546 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1547 # define _CC_mask(classnum) (1U << (classnum))
1549 /* For internal core Perl use only: the base macro for defining macros like
1551 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1552 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1554 /* The mask for the _A versions of the macros; it just adds in the bit for
1556 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1558 /* For internal core Perl use only: the base macro for defining macros like
1559 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1560 * the ASCII bit are present */
1561 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1562 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1563 == _CC_mask_A(classnum)))
1565 /* On ASCII platforms certain classes form a single range. It's faster to
1566 * special case these. isDIGIT is a single range on all platforms */
1568 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1569 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1570 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1571 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1572 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1574 /* By folding the upper and lowercase, we can use a single range */
1575 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1576 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1577 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1578 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1579 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1581 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1582 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1583 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1584 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1585 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1586 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1587 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1588 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1590 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1591 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1592 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1593 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1595 /* continuation character for legal NAME in \N{NAME} */
1596 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1598 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1599 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1600 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1601 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1602 # define isPSXSPC_L1(c) isSPACE_L1(c)
1603 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1604 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1605 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1606 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1607 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1610 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1613 /* Participates in a single-character fold with a character above 255 */
1614 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1616 /* Like the above, but also can be part of a multi-char fold */
1617 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
1619 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1620 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1621 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1622 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1623 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1625 /* is c a control character for which we have a mnemonic? */
1626 # if defined(PERL_CORE) || defined(PERL_EXT)
1627 # define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1629 #else /* else we don't have perl.h H_PERL */
1631 /* If we don't have perl.h, we are compiling a utility program. Below we
1632 * hard-code various macro definitions that wouldn't otherwise be available
1633 * to it. Most are coded based on first principles. These are written to
1634 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1635 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1636 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1637 # define isSPACE_A(c) (isBLANK_A(c) \
1642 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1643 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1644 * performance (if optimization isn't on), and allow the same code to be
1645 * used for both platform types */
1646 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1647 || inRANGE((c), 'j', 'r') \
1648 || inRANGE((c), 's', 'z')
1649 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1650 || inRANGE((c), 'J', 'R') \
1651 || inRANGE((c), 'S', 'Z')
1652 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1653 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1654 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1655 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1656 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1657 || inRANGE((c), 'a', 'f') \
1658 || inRANGE((c), 'A', 'F')
1659 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1660 || (c) == '#' || (c) == '$' || (c) == '%' \
1661 || (c) == '&' || (c) == '\'' || (c) == '(' \
1662 || (c) == ')' || (c) == '*' || (c) == '+' \
1663 || (c) == ',' || (c) == '.' || (c) == '/' \
1664 || (c) == ':' || (c) == ';' || (c) == '<' \
1665 || (c) == '=' || (c) == '>' || (c) == '?' \
1666 || (c) == '@' || (c) == '[' || (c) == '\\' \
1667 || (c) == ']' || (c) == '^' || (c) == '_' \
1668 || (c) == '`' || (c) == '{' || (c) == '|' \
1669 || (c) == '}' || (c) == '~')
1670 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1671 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1674 /* The below is accurate for the 3 EBCDIC code pages traditionally
1675 * supported by perl. The only difference between them in the controls
1676 * is the position of \n, and that is represented symbolically below */
1677 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1678 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1679 || (c) == '\t' || (c) == '\v' \
1680 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1681 || (c) == 7F /* U+7F DEL */ \
1682 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1684 || (c) == 0x18 /* U+18 CAN */ \
1685 || (c) == 0x19 /* U+19 EOM */ \
1686 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1687 || (c) == 0x26 /* U+17 ETB */ \
1688 || (c) == 0x27 /* U+1B ESC */ \
1689 || (c) == 0x2D /* U+05 ENQ */ \
1690 || (c) == 0x2E /* U+06 ACK */ \
1691 || (c) == 0x32 /* U+16 SYN */ \
1692 || (c) == 0x37 /* U+04 EOT */ \
1693 || (c) == 0x3C /* U+14 DC4 */ \
1694 || (c) == 0x3D /* U+15 NAK */ \
1695 || (c) == 0x3F)/* U+1A SUB */
1696 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1697 # else /* isASCII is already defined for ASCII platforms, so can use that to
1699 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1702 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1703 * during debugging, and it seems best to keep them. We may be called
1704 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1705 * do anything anyway, so make it not a problem */
1706 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1707 # define NATIVE_TO_LATIN1(ch) (ch)
1709 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1710 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1711 # define isBLANK_L1(c) (isBLANK_A(c) \
1712 || (FITS_IN_8_BITS(c) \
1713 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1714 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1715 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1716 # define isLOWER_L1(c) (isLOWER_A(c) \
1717 || (FITS_IN_8_BITS(c) \
1718 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1719 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1720 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1721 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1722 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1723 # define isPRINT_L1(c) (isPRINT_A(c) \
1724 || (FITS_IN_8_BITS(c) \
1725 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1726 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1727 || (FITS_IN_8_BITS(c) \
1728 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1729 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1730 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1731 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1732 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1733 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1734 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1735 # define isSPACE_L1(c) (isSPACE_A(c) \
1736 || (FITS_IN_8_BITS(c) \
1737 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1738 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1739 # define isUPPER_L1(c) (isUPPER_A(c) \
1740 || (FITS_IN_8_BITS(c) \
1741 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1743 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1744 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1745 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1746 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1751 /* The following are not fully accurate in the above-ASCII range. I (khw)
1752 * don't think it's necessary to be so for the purposes where this gets
1754 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1755 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1757 /* And these aren't accurate at all. They are useful only for above
1758 * Latin1, which utilities and bootstrapping don't deal with */
1759 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1760 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1761 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1763 /* Many of the macros later in this file are defined in terms of these. By
1764 * implementing them with a function, which converts the class number into
1765 * a call to the desired macro, all of the later ones work. However, that
1766 * function won't be actually defined when building a utility program (no
1767 * perl.h), and so a compiler error will be generated if one is attempted
1768 * to be used. And the above-Latin1 code points require Unicode tables to
1769 * be present, something unlikely to be the case when bootstrapping */
1770 # define _generic_isCC(c, classnum) \
1771 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1772 # define _generic_isCC_A(c, classnum) \
1773 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1774 #endif /* End of no perl.h H_PERL */
1776 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1777 #define isALPHA(c) isALPHA_A(c)
1778 #define isASCII_A(c) isASCII(c)
1779 #define isASCII_L1(c) isASCII(c)
1780 #define isBLANK(c) isBLANK_A(c)
1781 #define isCNTRL(c) isCNTRL_A(c)
1782 #define isDIGIT(c) isDIGIT_A(c)
1783 #define isGRAPH(c) isGRAPH_A(c)
1784 #define isIDFIRST(c) isIDFIRST_A(c)
1785 #define isLOWER(c) isLOWER_A(c)
1786 #define isPRINT(c) isPRINT_A(c)
1787 #define isPSXSPC_A(c) isSPACE_A(c)
1788 #define isPSXSPC(c) isPSXSPC_A(c)
1789 #define isPSXSPC_L1(c) isSPACE_L1(c)
1790 #define isPUNCT(c) isPUNCT_A(c)
1791 #define isSPACE(c) isSPACE_A(c)
1792 #define isUPPER(c) isUPPER_A(c)
1793 #define isWORDCHAR(c) isWORDCHAR_A(c)
1794 #define isXDIGIT(c) isXDIGIT_A(c)
1796 /* ASCII casing. These could also be written as
1797 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1798 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1799 which uses table lookup and mask instead of subtraction. (This would
1800 work because the _MOD does not apply in the ASCII range).
1802 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1803 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1804 there non-ASCII invariants, and all of them are controls.) */
1805 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1806 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1808 /* In the ASCII range, these are equivalent to what they're here defined to be.
1809 * But by creating these definitions, other code doesn't have to be aware of
1810 * this detail. Actually this works for all UTF-8 invariants, not just the
1811 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1812 #define toFOLD(c) toLOWER(c)
1813 #define toTITLE(c) toUPPER(c)
1815 #define toLOWER_A(c) toLOWER(c)
1816 #define toUPPER_A(c) toUPPER(c)
1817 #define toFOLD_A(c) toFOLD(c)
1818 #define toTITLE_A(c) toTITLE(c)
1820 /* Use table lookup for speed; returns the input itself if is out-of-range */
1821 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1823 : PL_latin1_lc[ (U8) (c) ])
1824 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1826 /* Modified uc. Is correct uc except for three non-ascii chars which are
1827 * all mapped to one of them, and these need special handling; returns the
1828 * input itself if is out-of-range */
1829 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1831 : PL_mod_latin1_uc[ (U8) (c) ])
1832 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1834 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1836 /* For internal core Perl use only: the base macro for defining macros like
1837 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1838 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1839 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1840 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1841 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1842 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1843 * code points under locale rules, use the macros based on _generic_LC_uvchr
1844 * instead of this. */
1845 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1846 (! FITS_IN_8_BITS(c) \
1848 : IN_UTF8_CTYPE_LOCALE \
1849 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1852 /* For internal core Perl use only: a helper macro for defining macros like
1853 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1854 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1855 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1856 * _generic_LC_base for more info */
1857 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1858 _generic_LC_base(c,utf8_locale_classnum, \
1859 non_utf8_func( (_LC_CAST) (c)))
1861 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1862 * 'c' is the platform's native underscore character */
1863 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1864 _generic_LC_base(c, utf8_locale_classnum, \
1865 (non_utf8_func( (_LC_CAST) (c)) \
1866 || (char)(c) == '_'))
1868 /* These next three are also for internal core Perl use only: case-change
1869 * helper macros. The reason for using the PL_latin arrays is in case the
1870 * system function is defective; it ensures uniform results that conform to the
1871 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1872 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1874 : (IN_UTF8_CTYPE_LOCALE) \
1875 ? PL_latin1_lc[ (U8) (c) ] \
1876 : (cast)function((cast)(c)))
1878 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1879 * returns a single value, so can't adequately return the upper case of LATIN
1880 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1881 * values "SS"); instead it asserts against that under DEBUGGING, and
1882 * otherwise returns its input. It does not handle the anomalies in UTF-8
1883 * Turkic locales. */
1884 #define _generic_toUPPER_LC(c, function, cast) \
1885 (! FITS_IN_8_BITS(c) \
1887 : ((! IN_UTF8_CTYPE_LOCALE) \
1888 ? (cast)function((cast)(c)) \
1889 : ((((U8)(c)) == MICRO_SIGN) \
1890 ? GREEK_CAPITAL_LETTER_MU \
1891 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1892 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1893 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1894 ? (__ASSERT_(0) (c)) \
1895 : PL_mod_latin1_uc[ (U8) (c) ])))))
1897 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1898 * returns a single value, so can't adequately return the fold case of LATIN
1899 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1900 * values "ss"); instead it asserts against that under DEBUGGING, and
1901 * otherwise returns its input. It does not handle the anomalies in UTF-8
1903 #define _generic_toFOLD_LC(c, function, cast) \
1904 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1905 ? GREEK_SMALL_LETTER_MU \
1906 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1907 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1908 _generic_toLOWER_LC(c, function, cast)))
1910 /* Use the libc versions for these if available. */
1911 #if defined(HAS_ISASCII)
1912 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1914 # define isASCII_LC(c) isASCII(c)
1917 #if defined(HAS_ISBLANK)
1918 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1919 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1920 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1926 /* The Windows functions don't bother to follow the POSIX standard, which
1927 * for example says that something can't both be a printable and a control.
1928 * But Windows treats the \t control as a printable, and does such things
1929 * as making superscripts into both digits and punctuation. This tames
1930 * these flaws by assuming that the definitions of both controls and space
1931 * are correct, and then making sure that other definitions don't have
1932 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1933 * Not all possible weirdnesses are checked for, just the ones that were
1934 * detected on actual Microsoft code pages */
1936 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1937 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1939 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1940 && isALPHANUMERIC_LC(c))
1941 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1943 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1944 isALPHANUMERIC_LC(c))
1945 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1946 # define isIDFIRST_LC(c) (((c) == '_') \
1947 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1948 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1949 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1950 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1951 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1952 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1953 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1954 && isALPHANUMERIC_LC(c))
1956 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1957 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1958 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1960 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1961 /* For most other platforms */
1963 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1964 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1965 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1966 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1967 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1968 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1969 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1970 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1971 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1972 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1973 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1974 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1975 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1978 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1979 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1980 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1982 #else /* The final fallback position */
1984 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1985 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1986 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1987 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1988 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1989 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1990 # define isLOWER_LC(c) (isascii(c) && islower(c))
1991 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1992 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1993 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1994 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1995 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1996 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1998 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1999 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
2000 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
2004 #define isIDCONT(c) isWORDCHAR(c)
2005 #define isIDCONT_A(c) isWORDCHAR_A(c)
2006 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
2007 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
2008 #define isPSXSPC_LC(c) isSPACE_LC(c)
2010 /* For internal core Perl use only: the base macros for defining macros like
2011 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
2012 * number defined earlier in this file. _generic_uvchr() is used for POSIX
2013 * classes where there is a macro or function 'above_latin1' that takes the
2014 * single argument 'c' and returns the desired value. These exist for those
2015 * classes which have simple definitions, avoiding the overhead of an inversion
2016 * list binary search. _generic_invlist_uvchr() can be used
2017 * for classes where that overhead is faster than a direct lookup.
2018 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
2019 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2020 * have duplicate checks here, so could create versions of the macros that
2021 * don't, but experiments show that gcc optimizes them out anyway. */
2023 /* Note that all ignore 'use bytes' */
2024 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2025 ? _generic_isCC(c, classnum) \
2027 #define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
2028 ? _generic_isCC(c, classnum) \
2029 : _is_uni_FOO(classnum, c))
2030 #define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2031 #define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
2032 #define isASCII_uvchr(c) isASCII(c)
2033 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2034 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2035 #define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2036 #define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
2037 #define isIDCONT_uvchr(c) \
2038 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2039 #define isIDFIRST_uvchr(c) \
2040 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2041 #define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2042 #define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
2044 #define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
2045 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2046 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2048 #define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
2049 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2050 #define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
2051 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2053 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2054 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2055 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2056 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2058 /* For backwards compatibility, even though '_uni' should mean official Unicode
2059 * code points, in Perl it means native for those below 256 */
2060 #define isALPHA_uni(c) isALPHA_uvchr(c)
2061 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2062 #define isASCII_uni(c) isASCII_uvchr(c)
2063 #define isBLANK_uni(c) isBLANK_uvchr(c)
2064 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
2065 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
2066 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
2067 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
2068 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2069 #define isLOWER_uni(c) isLOWER_uvchr(c)
2070 #define isPRINT_uni(c) isPRINT_uvchr(c)
2071 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
2072 #define isSPACE_uni(c) isSPACE_uvchr(c)
2073 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2074 #define isUPPER_uni(c) isUPPER_uvchr(c)
2075 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
2076 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2077 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2078 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2079 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2080 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2081 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2083 /* For internal core Perl use only: the base macros for defining macros like
2084 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2085 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2086 * simple class definitions; the other for more complex. These are like
2087 * _generic_uvchr, so see it for more info. */
2088 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2089 (c < 256 ? latin1(c) : above_latin1(c))
2090 #define _generic_LC_invlist_uvchr(latin1, classnum, c) \
2091 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2093 #define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2094 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
2095 _CC_ALPHANUMERIC, c)
2096 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2097 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2098 is_HORIZWS_cp_high, c)
2099 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2100 #define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2101 #define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2102 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2103 _is_uni_perl_idcont, c)
2104 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2105 _is_uni_perl_idstart, c)
2106 #define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2107 #define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
2108 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2109 #define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2110 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2111 is_XPERLSPACE_cp_high, c)
2112 #define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2113 #define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
2115 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2116 is_XDIGIT_cp_high, c)
2118 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2120 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2121 * they don't otherwise go out of their way to look for malformed UTF-8. If
2122 * they can return accurate results without knowing if the input is otherwise
2123 * malformed, they do so. For example isASCII is accurate in spite of any
2124 * non-length malformations because it looks only at a single byte. Likewise
2125 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2126 * variant ones return FALSE. But, if the input has to be well-formed in order
2127 * for the results to be accurate, the macros will test and if malformed will
2128 * call a routine to die
2130 * Except for toke.c, the macros do assume that e > p, asserting that on
2131 * DEBUGGING builds. Much code that calls these depends on this being true,
2132 * for other reasons. toke.c is treated specially as using the regular
2133 * assertion breaks it in many ways. All strings that these operate on there
2134 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2135 * bunch of code in toke.c assumes that this is true, so the assertion allows
2137 #ifdef PERL_IN_TOKE_C
2138 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2140 # define _utf8_safe_assert(p,e) ((e) > (p))
2143 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2144 ((! _utf8_safe_assert(p, e)) \
2145 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2146 : (UTF8_IS_INVARIANT(*(p))) \
2147 ? _generic_isCC(*(p), classnum) \
2148 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2149 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2150 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2152 : (_force_out_malformed_utf8_message( \
2153 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2155 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2156 * 'above_latin1' can be a macro */
2157 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2158 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2159 #define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2160 _generic_utf8_safe(classnum, p, e, \
2161 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2162 ? (_force_out_malformed_utf8_message( \
2163 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2165 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2166 * 'above_latin1' parameter */
2167 #define _generic_invlist_utf8_safe(classnum, p, e) \
2168 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
2170 /* Like the above, but should be used only when it is known that there are no
2171 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2172 * class is TRUE for. Hence it can skip the tests for this range.
2173 * 'above_latin1' should include its arguments */
2174 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2175 (__ASSERT_(_utf8_safe_assert(p, e)) \
2176 (UTF8_IS_INVARIANT(*(p))) \
2177 ? _generic_isCC(*(p), classnum) \
2178 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2179 ? 0 /* Note that doesn't check validity for latin1 */ \
2183 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2184 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2185 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2186 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2187 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2188 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2189 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2190 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2191 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2192 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2193 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2194 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2195 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2196 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2197 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2198 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2199 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2200 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
2202 #define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
2203 #define isALPHANUMERIC_utf8_safe(p, e) \
2204 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
2205 #define isASCII_utf8_safe(p, e) \
2206 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2208 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2209 #define isBLANK_utf8_safe(p, e) \
2210 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2213 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2214 * more efficient macro instead of the more general one */
2215 # define isCNTRL_utf8_safe(p, e) \
2216 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2218 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2221 #define isDIGIT_utf8_safe(p, e) \
2222 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2223 _is_utf8_FOO(_CC_DIGIT, p, e))
2224 #define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
2225 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2226 _is_utf8_perl_idcont, p, e)
2228 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2229 * IDFIRST is an alnum. See
2230 * https://github.com/Perl/perl5/issues/10275 for more detail than you
2231 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2232 * This used to be not the XID version, but we decided to go with the more
2233 * modern Unicode definition */
2234 #define isIDFIRST_utf8_safe(p, e) \
2235 _generic_func_utf8_safe(_CC_IDFIRST, \
2236 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
2238 #define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2239 #define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
2240 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2241 #define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
2242 #define isSPACE_utf8_safe(p, e) \
2243 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2244 #define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
2245 #define isVERTWS_utf8_safe(p, e) \
2246 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2247 #define isWORDCHAR_utf8_safe(p, e) \
2248 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
2249 #define isXDIGIT_utf8_safe(p, e) \
2250 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2251 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2252 ? (_force_out_malformed_utf8_message( \
2253 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2254 : is_XDIGIT_high(p)))
2256 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2257 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2258 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2259 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2261 /* For internal core use only, subject to change */
2262 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2263 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2264 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2265 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
2267 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2268 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2269 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2270 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2272 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2273 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2274 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2275 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2276 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2277 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2278 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2279 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2280 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2281 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2282 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2283 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2284 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2285 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2286 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2287 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2288 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
2290 /* For internal core Perl use only: the base macros for defining macros like
2291 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2292 * point in 'p' is within the 0-255 range, it uses locale rules from the
2293 * passed-in 'macro' parameter */
2294 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2295 (__ASSERT_(_utf8_safe_assert(p, e)) \
2296 (UTF8_IS_INVARIANT(*(p))) \
2298 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2299 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2300 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2301 : (_force_out_malformed_utf8_message( \
2302 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2305 #define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2306 _generic_LC_utf8_safe(macro, p, e, \
2307 _is_utf8_FOO(classnum, p, e))
2309 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2310 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2312 #define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2313 _generic_LC_utf8_safe(classnum, p, e, \
2314 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2315 ? (_force_out_malformed_utf8_message( \
2316 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2319 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2320 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
2321 _CC_ALPHANUMERIC, p, e)
2322 #define isALPHA_LC_utf8_safe(p, e) \
2323 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2324 #define isASCII_LC_utf8_safe(p, e) \
2325 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2326 #define isBLANK_LC_utf8_safe(p, e) \
2327 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2328 #define isCNTRL_LC_utf8_safe(p, e) \
2329 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2330 #define isDIGIT_LC_utf8_safe(p, e) \
2331 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2332 #define isGRAPH_LC_utf8_safe(p, e) \
2333 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2334 #define isIDCONT_LC_utf8_safe(p, e) \
2335 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2336 _is_utf8_perl_idcont, p, e)
2337 #define isIDFIRST_LC_utf8_safe(p, e) \
2338 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2339 _is_utf8_perl_idstart, p, e)
2340 #define isLOWER_LC_utf8_safe(p, e) \
2341 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2342 #define isPRINT_LC_utf8_safe(p, e) \
2343 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2344 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2345 #define isPUNCT_LC_utf8_safe(p, e) \
2346 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2347 #define isSPACE_LC_utf8_safe(p, e) \
2348 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2349 #define isUPPER_LC_utf8_safe(p, e) \
2350 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2351 #define isWORDCHAR_LC_utf8_safe(p, e) \
2352 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2353 #define isXDIGIT_LC_utf8_safe(p, e) \
2354 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2356 /* Macros for backwards compatibility and for completeness when the ASCII and
2357 * Latin1 values are identical */
2358 #define isALPHAU(c) isALPHA_L1(c)
2359 #define isDIGIT_L1(c) isDIGIT_A(c)
2360 #define isOCTAL(c) isOCTAL_A(c)
2361 #define isOCTAL_L1(c) isOCTAL_A(c)
2362 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2363 #define isALNUM(c) isWORDCHAR(c)
2364 #define isALNUM_A(c) isALNUM(c)
2365 #define isALNUMU(c) isWORDCHAR_L1(c)
2366 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2367 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2368 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2369 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
2370 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
2371 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
2372 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
2373 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2374 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2375 #define isALNUMC(c) isALPHANUMERIC(c)
2376 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2377 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2378 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2379 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
2380 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2381 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
2383 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2384 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2385 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2386 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2387 * the outlier from the block that contains the other controls, just like
2388 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2389 * block. If it weren't special cased, it would yield a non-control.
2390 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2393 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2395 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2397 ? (UNLIKELY((c) == '?') \
2398 ? QUESTION_MARK_CTRL \
2399 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2400 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2402 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2405 /* Line numbers are unsigned, 32 bits. */
2407 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2409 /* Helpful alias for version prescan */
2410 #define is_LAX_VERSION(a,b) \
2411 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2413 #define is_STRICT_VERSION(a,b) \
2414 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2416 #define BADVERSION(a,b,c) \
2422 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2423 * a-f) to its numeric value without using any branches. The input is
2424 * validated only by an assert() in DEBUGGING builds.
2426 * It works by right shifting and isolating the bit that is 0 for the digits,
2427 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2428 * position, and then to the eights position. Both are added together to form
2429 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2430 * final four bits of the input to form the correct value. */
2431 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2432 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2433 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2434 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2436 /* The argument is a string pointer, which is advanced. */
2437 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
2439 /* Converts a character known to represent an octal digit (0-7) to its numeric
2440 * value. The input is validated only by an assert() in DEBUGGING builds. In
2441 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2442 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2444 /* Efficiently returns a boolean as to if two native characters are equivalent
2445 * case-insensitively. At least one of the characters must be one of [A-Za-z];
2446 * the ALPHA in the name is to remind you of that. This is asserted() in
2447 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2448 * works (on valid input) for both non- and UTF-8-encoded bytes.
2450 * When one of the inputs is a compile-time constant and gets folded by the
2451 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2452 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2453 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2454 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2455 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2456 * just a single 0, in the bit position where the upper- and lowercase differ.
2458 #define isALPHA_FOLD_EQ(c1, c2) \
2459 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2460 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2461 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2464 =for apidoc_section $memory
2466 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2467 The XSUB-writer's interface to the C C<malloc> function.
2469 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2471 In 5.9.3, Newx() and friends replace the older New() API, and drops
2472 the first parameter, I<x>, a debug aid which allowed callers to identify
2473 themselves. This aid has been superseded by a new build option,
2474 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2475 there for use in XS modules supporting older perls.
2477 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2478 The XSUB-writer's interface to the C C<malloc> function, with
2479 cast. See also C<L</Newx>>.
2481 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2483 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2484 The XSUB-writer's interface to the C C<malloc> function. The allocated
2485 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2487 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2489 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2490 The XSUB-writer's interface to the C C<realloc> function.
2492 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2494 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2495 The XSUB-writer's interface to the C C<realloc> function, with
2498 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2500 =for apidoc Am|void|Safefree|void* ptr
2501 The XSUB-writer's interface to the C C<free> function.
2503 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2505 =for apidoc_section $string
2506 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2507 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2508 source, C<dest> is the destination, C<nitems> is the number of items, and
2509 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2511 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2512 Like C<Move> but returns C<dest>. Useful
2513 for encouraging compilers to tail-call
2516 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2517 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2518 source, C<dest> is the destination, C<nitems> is the number of items, and
2519 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2521 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2523 Like C<Copy> but returns C<dest>. Useful
2524 for encouraging compilers to tail-call
2527 =for apidoc Am|void|Zero|void* dest|int nitems|type
2529 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2530 destination, C<nitems> is the number of items, and C<type> is the type.
2532 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2534 Like C<Zero> but returns dest. Useful
2535 for encouraging compilers to tail-call
2538 =for apidoc_section $utility
2539 =for apidoc Amu|void|StructCopy|type *src|type *dest|type
2540 This is an architecture-independent macro to copy one structure to another.
2542 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2544 Fill up memory with a byte pattern (a byte repeated over and over
2545 again) that hopefully catches attempts to access uninitialized memory.
2547 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2549 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2551 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2553 PoisonWith(0xEF) for catching access to freed memory.
2555 =for apidoc Am|void|Poison|void* dest|int nitems|type
2557 PoisonWith(0xEF) for catching access to freed memory.
2561 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2563 #define NEWSV(x,len) newSV(len)
2566 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2568 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2570 #ifdef PERL_MALLOC_WRAP
2572 /* This expression will be constant-folded at compile time. It checks
2573 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2574 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2575 * As well as avoiding the need for a run-time check in some cases, it's
2576 * designed to avoid compiler warnings like:
2577 * comparison is always false due to limited range of data type
2578 * It's mathematically equivalent to
2579 * max(n) * sizeof(t) > MEM_SIZE_MAX
2582 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2583 ( sizeof(MEM_SIZE) < sizeof(n) \
2584 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2586 /* This is written in a slightly odd way to avoid various spurious
2587 * compiler warnings. We *want* to write the expression as
2588 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2589 * (for some compile-time constant C), but even when the LHS
2590 * constant-folds to false at compile-time, g++ insists on emitting
2591 * warnings about the RHS (e.g. "comparison is always false"), so instead
2594 * (cond ? n : X) > C
2596 * where X is a constant with X > C always false. Choosing a value for X
2597 * is tricky. If 0, some compilers will complain about 0 > C always being
2598 * false; if 1, Coverity complains when n happens to be the constant value
2599 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2600 * for X and hope that nothing else whines.
2603 # define _MEM_WRAP_WILL_WRAP(n,t) \
2604 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2605 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2607 # define MEM_WRAP_CHECK(n,t) \
2608 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2609 && (croak_memory_wrap(),0))
2611 # define MEM_WRAP_CHECK_1(n,t,a) \
2612 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2613 && (Perl_croak_nocontext("%s",(a)),0))
2615 /* "a" arg must be a string literal */
2616 # define MEM_WRAP_CHECK_s(n,t,a) \
2617 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2618 && (Perl_croak_nocontext("" a ""),0))
2620 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2622 #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))
2625 #define MEM_WRAP_CHECK(n,t)
2626 #define MEM_WRAP_CHECK_1(n,t,a)
2627 #define MEM_WRAP_CHECK_s(n,t,a)
2628 #define MEM_WRAP_CHECK_(n,t)
2630 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2636 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2637 * go through functions, which are handy for debugging breakpoints, but
2638 * which more importantly get the immediate calling environment (file and
2639 * line number, and C function name if available) passed in. This info can
2640 * then be used for logging the calls, for which one gets a sample
2641 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2644 * - not all memory allocs get logged, only those
2645 * that go through Newx() and derivatives (while all
2646 * Safefrees do get logged)
2647 * - __FILE__ and __LINE__ do not work everywhere
2648 * - __func__ or __FUNCTION__ even less so
2649 * - I think more goes on after the perlio frees but
2650 * the thing is that STDERR gets closed (as do all
2651 * the file descriptors)
2652 * - no deeper calling stack than the caller of the Newx()
2653 * or the kind, but do I look like a C reflection/introspection
2655 * - the function prototypes for the logging functions
2656 * probably should maybe be somewhere else than handy.h
2657 * - one could consider inlining (macrofying) the logging
2658 * for speed, but I am too lazy
2659 * - one could imagine recording the allocations in a hash,
2660 * (keyed by the allocation address?), and maintain that
2661 * through reallocs and frees, but how to do that without
2662 * any News() happening...?
2663 * - lots of -Ddefines to get useful/controllable output
2664 * - lots of ENV reads
2668 # ifndef PERL_MEM_LOG_NOIMPL
2677 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2678 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2679 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2686 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2687 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2688 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2691 #ifndef MEM_LOG_ALLOC
2692 #define MEM_LOG_ALLOC(n,t,a) (a)
2694 #ifndef MEM_LOG_REALLOC
2695 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2697 #ifndef MEM_LOG_FREE
2698 #define MEM_LOG_FREE(a) (a)
2701 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2702 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2703 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2706 /* pre 5.9.x compatibility */
2707 #define New(x,v,n,t) Newx(v,n,t)
2708 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2709 #define Newz(x,v,n,t) Newxz(v,n,t)
2712 #define Renew(v,n,t) \
2713 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2714 #define Renewc(v,n,t,c) \
2715 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2718 #define Safefree(d) \
2719 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2721 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2724 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2725 * as it handles cases like constant string arguments without throwing warnings *
2726 * the cast is required, as is the inequality check, to avoid warnings */
2727 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2730 #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)))
2731 #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)))
2732 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2734 /* Like above, but returns a pointer to 'd' */
2735 #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)))
2736 #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)))
2737 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2739 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2740 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2741 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2742 #define Poison(d,n,t) PoisonFree(d,n,t)
2745 # define PERL_POISON_EXPR(x) x
2747 # define PERL_POISON_EXPR(x)
2750 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2753 =for apidoc_section $utility
2755 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2757 Returns the number of elements in the input C array (so you want your
2758 zero-based indices to be less than but not equal to).
2760 =for apidoc Am|void *|C_ARRAY_END|void *a
2762 Returns a pointer to one element past the final element of the input C array.
2766 C_ARRAY_END is one past the last: half-open/half-closed range, not
2767 last-inclusive range.
2769 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2770 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2774 # define Perl_va_copy(s, d) va_copy(d, s)
2775 # elif defined(__va_copy)
2776 # define Perl_va_copy(s, d) __va_copy(d, s)
2778 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2782 /* convenience debug macros */
2784 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2785 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2786 #define pTHX_VALUE_ (void *)my_perl,
2787 #define pTHX_VALUE (void *)my_perl
2788 #define pTHX__VALUE_ ,(void *)my_perl,
2789 #define pTHX__VALUE ,(void *)my_perl
2792 #define pTHX__FORMAT
2795 #define pTHX__VALUE_
2797 #endif /* USE_ITHREADS */
2799 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2800 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2801 CPAN::Unpack show any users outside the core. */
2803 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2804 "Use of " s " is deprecated")
2805 # define deprecate_disappears_in(when,message) \
2806 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2807 message ", and will disappear in Perl " when)
2808 # define deprecate_fatal_in(when,message) \
2809 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2810 message ". Its use will be fatal in Perl " when)
2813 /* Internal macros to deal with gids and uids */
2816 # if Uid_t_size > IVSIZE
2817 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2818 # define SvUID(sv) SvNV(sv)
2819 # elif Uid_t_sign <= 0
2820 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2821 # define SvUID(sv) SvIV(sv)
2823 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2824 # define SvUID(sv) SvUV(sv)
2825 # endif /* Uid_t_size */
2827 # if Gid_t_size > IVSIZE
2828 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2829 # define SvGID(sv) SvNV(sv)
2830 # elif Gid_t_sign <= 0
2831 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2832 # define SvGID(sv) SvIV(sv)
2834 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2835 # define SvGID(sv) SvUV(sv)
2836 # endif /* Gid_t_size */
2840 #endif /* PERL_HANDY_H_ */
2843 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob