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
288 =for apidoc_section $integer
289 =for apidoc Ay|| PERL_INT_FAST8_T
290 =for apidoc_item PERL_INT_FAST16_T
291 =for apidoc_item PERL_UINT_FAST8_T
292 =for apidoc_item PERL_UINT_FAST16_T
294 These are equivalent to the correspondingly-named C99 typedefs on platforms
295 that have those; they evaluate to C<int> and C<unsigned int> on platforms that
296 don't, so that you can portably take advantage of this C99 feature.
301 typedef int_fast8_t PERL_INT_FAST8_T;
302 typedef uint_fast8_t PERL_UINT_FAST8_T;
303 typedef int_fast16_t PERL_INT_FAST16_T;
304 typedef uint_fast16_t PERL_UINT_FAST16_T;
306 typedef int PERL_INT_FAST8_T;
307 typedef unsigned int PERL_UINT_FAST8_T;
308 typedef int PERL_INT_FAST16_T;
309 typedef unsigned int PERL_UINT_FAST16_T;
312 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
313 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
314 * required to represent any possible unsigned number containing N bits.
315 * TYPE_DIGITS gives the number of decimal digits required to represent any
316 * possible unsigned number of type T. */
317 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
318 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
319 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
321 /* Unused by core; should be deprecated */
322 #define Ctl(ch) ((ch) & 037)
324 #if defined(PERL_CORE) || defined(PERL_EXT)
326 # define MIN(a,b) ((a) < (b) ? (a) : (b))
329 # define MAX(a,b) ((a) > (b) ? (a) : (b))
333 /* Returns a boolean as to whether the input unsigned number is a power of 2
334 * (2**0, 2**1, etc). In other words if it has just a single bit set.
335 * If not, subtracting 1 would leave the uppermost bit set, so the & would
337 #if defined(PERL_CORE) || defined(PERL_EXT)
338 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
341 /* Returns a mask with the lowest n bits set */
342 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
344 /* The largest unsigned number that will fit into n bits */
345 #define nBIT_UMAX(n) nBIT_MASK(n)
348 =for apidoc_section $directives
349 =for apidoc Am||__ASSERT_|bool expr
351 This is a helper macro to avoid preprocessor issues, replaced by nothing
352 unless under DEBUGGING, where it expands to an assert of its argument,
353 followed by a comma (hence the comma operator). If we just used a straight
354 assert(), we would get a comma with nothing before it when not DEBUGGING.
358 We also use empty definition under Coverity since the __ASSERT__
359 checks often check for things that Really Cannot Happen, and Coverity
360 detects that and gets all excited. */
362 #if defined(DEBUGGING) && !defined(__COVERITY__) \
363 && ! defined(PERL_SMALL_MACRO_BUFFER)
364 # define __ASSERT_(statement) assert(statement),
366 # define __ASSERT_(statement)
370 =for apidoc_section $SV
372 =for apidoc Ama|SV*|newSVpvs|"literal string"
373 Like C<newSVpvn>, but takes a literal string instead of a
376 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
377 Like C<newSVpvn_flags>, but takes a literal string instead of
378 a string/length pair.
380 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
381 Like C<newSVpvn_share>, but takes a literal string instead of
382 a string/length pair and omits the hash parameter.
384 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
385 Like C<sv_catpvn_flags>, but takes a literal string instead
386 of a string/length pair.
388 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
389 Like C<sv_catpvn_nomg>, but takes a literal string instead of
390 a string/length pair.
392 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
393 Like C<sv_catpvn>, but takes a literal string instead of a
396 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
397 Like C<sv_catpvn_mg>, but takes a literal string instead of a
400 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
401 Like C<sv_setpvn>, but takes a literal string instead of a
404 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
405 Like C<sv_setpvn_mg>, but takes a literal string instead of a
408 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
409 Like C<sv_setref_pvn>, but takes a literal string instead of
410 a string/length pair.
412 =for apidoc_section $string
414 =for apidoc Ama|char*|savepvs|"literal string"
415 Like C<savepvn>, but takes a literal string instead of a
418 =for apidoc Ama|char*|savesharedpvs|"literal string"
419 A version of C<savepvs()> which allocates the duplicate string in memory
420 which is shared between threads.
422 =for apidoc_section $GV
424 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
425 Like C<gv_stashpvn>, but takes a literal string instead of a
428 =for apidoc_section $HV
430 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
431 Like C<hv_fetch>, but takes a literal string instead of a
434 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
435 Like C<hv_store>, but takes a literal string instead of a
437 and omits the hash parameter.
439 =for apidoc_section $lexer
441 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
443 Like L</lex_stuff_pvn>, but takes a literal string instead of
444 a string/length pair.
450 =for apidoc_section $string
452 =for apidoc Amu|pair|STR_WITH_LEN|"literal string"
454 Returns two comma separated tokens of the input literal string, and its length.
455 This is convenience macro which helps out in some API calls.
456 Note that it can't be used as an argument to macros or functions that under
457 some configurations might be macros, which means that it requires the full
458 Perl_xxx(aTHX_ ...) form for any API calls where it's used.
463 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
465 /* STR_WITH_LEN() shortcuts */
466 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
467 #define newSVpvs_flags(str,flags) \
468 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
469 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
470 #define sv_catpvs_flags(sv, str, flags) \
471 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
472 #define sv_catpvs_nomg(sv, str) \
473 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
474 #define sv_catpvs(sv, str) \
475 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
476 #define sv_catpvs_mg(sv, str) \
477 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
478 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
479 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
480 #define sv_setref_pvs(rv, classname, str) \
481 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
482 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
483 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
484 #define gv_stashpvs(str, create) \
485 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
487 #define gv_fetchpvs(namebeg, flags, sv_type) \
488 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type)
489 #define gv_fetchpvn gv_fetchpvn_flags
490 #define sv_catxmlpvs(dsv, str, utf8) \
491 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
494 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
496 #define get_cvs(str, flags) \
497 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
499 /* internal helpers */
501 #ifndef PERL_VERSION_MAJOR
502 # define PERL_VERSION_MAJOR PERL_REVISION
504 # undef PERL_REVISION /* We don't want code to be using these */
506 #ifndef PERL_VERSION_MINOR
507 # define PERL_VERSION_MINOR PERL_VERSION
511 #ifndef PERL_VERSION_PATCH
512 # define PERL_VERSION_PATCH PERL_SUBVERSION
514 # undef PERL_SUBVERSION
517 #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
518 /* '10*' leaves room for things like alpha, beta, releases */ \
519 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
520 #define PERL_DECIMAL_VERSION_ \
521 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
525 =for apidoc_section $versioning
526 =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
527 =for apidoc_item PERL_VERSION_NE
528 =for apidoc_item PERL_VERSION_LT
529 =for apidoc_item PERL_VERSION_LE
530 =for apidoc_item PERL_VERSION_GT
531 =for apidoc_item PERL_VERSION_GE
533 Returns whether or not the perl currently being compiled has the specified
534 relationship to the perl given by the parameters. For example,
536 #if PERL_VERSION_GT(5,24,2)
537 code that will only be compiled on perls after v5.24.2
542 Note that this is usable in making compile-time decisions
544 You may use the special value '*' for the final number to mean ALL possible
547 #if PERL_VERSION_EQ(5,31,'*')
549 means all perls in the 5.31 series. And
551 #if PERL_VERSION_NE(5,24,'*')
553 means all perls EXCEPT 5.24 ones. And
555 #if PERL_VERSION_LE(5,9,'*')
559 #if PERL_VERSION_LT(5,10,0)
561 This means you don't have to think so much when converting from the existing
562 deprecated C<PERL_VERSION> to using this macro:
564 #if PERL_VERSION <= 9
568 #if PERL_VERSION_LE(5,9,'*')
573 /* N.B. These don't work if the patch version is 42 or 92, as those are what
574 * '*' is in ASCII and EBCDIC respectively */
575 # define PERL_VERSION_EQ(j,n,p) \
577 ? ( (j) == PERL_VERSION_MAJOR \
578 && (n) == PERL_VERSION_MINOR) \
579 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
580 # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
582 # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
583 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
585 (((p) == '*') ? 0 : p)))
586 # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
588 # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
589 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
590 (((p) == '*') ? ((n)+1) : (n)), \
591 (((p) == '*') ? 0 : p)))
592 # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
595 =for apidoc_section $string
597 =for apidoc Am|bool|strNE|char* s1|char* s2
598 Test two C<NUL>-terminated strings to see if they are different. Returns true
601 =for apidoc Am|bool|strEQ|char* s1|char* s2
602 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
605 =for apidoc Am|bool|strLT|char* s1|char* s2
606 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
607 second, C<s2>. Returns true or false.
609 =for apidoc Am|bool|strLE|char* s1|char* s2
610 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
611 equal to the second, C<s2>. Returns true or false.
613 =for apidoc Am|bool|strGT|char* s1|char* s2
614 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
615 the second, C<s2>. Returns true or false.
617 =for apidoc Am|bool|strGE|char* s1|char* s2
618 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
619 or equal to the second, C<s2>. Returns true or false.
621 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
622 Test two C<NUL>-terminated strings to see if they are different. 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|strnEQ|char* s1|char* s2|STRLEN len
627 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
628 parameter indicates the number of bytes to compare. Returns true or false. (A
629 wrapper for C<strncmp>).
631 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
632 Test two buffers (which may contain embedded C<NUL> characters, to see if they
633 are equal. The C<len> parameter indicates the number of bytes to compare.
634 Returns zero if equal, or non-zero if non-equal.
636 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
637 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
638 C<l1> gives the number of bytes in C<s1>.
639 Returns zero if equal, or non-zero if non-equal.
641 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
642 Test two buffers (which may contain embedded C<NUL> characters, to see if they
643 are not equal. The C<len> parameter indicates the number of bytes to compare.
644 Returns zero if non-equal, or non-zero if equal.
646 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
647 Like L</memNE>, but the second string is a literal enclosed in double quotes,
648 C<l1> gives the number of bytes in C<s1>.
649 Returns zero if non-equal, or zero if non-equal.
651 =for apidoc Am|bool|memCHRs|"list"|char c
652 Returns the position of the first occurence of the byte C<c> in the literal
653 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
654 treated as unsigned char. Thus this macro can be used to determine if C<c> is
655 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
656 is C<NUL> (and the set doesn't include C<NUL>).
660 New macros should use the following conventions for their names (which are
661 based on the underlying C library functions):
663 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
665 Each has two main parameters, string-like operands that are compared
666 against each other, as specified by the macro name. Some macros may
667 additionally have one or potentially even two length parameters. If a length
668 parameter applies to both string parameters, it will be positioned third;
669 otherwise any length parameter immediately follows the string parameter it
672 If the prefix to the name is 'str', the string parameter is a pointer to a C
673 language string. Such a string does not contain embedded NUL bytes; its
674 length may be unknown, but can be calculated by C<strlen()>, since it is
675 terminated by a NUL, which isn't included in its length.
677 The optional 'n' following 'str' means that there is a third parameter,
678 giving the maximum number of bytes to look at in each string. Even if both
679 strings are longer than the length parameter, those extra bytes will be
682 The 's' suffix means that the 2nd byte string parameter is a literal C
683 double-quoted string. Its length will automatically be calculated by the
684 macro, so no length parameter will ever be needed for it.
686 If the prefix is 'mem', the string parameters don't have to be C strings;
687 they may contain embedded NUL bytes, do not necessarily have a terminating
688 NUL, and their lengths can be known only through other means, which in
689 practice are additional parameter(s) passed to the function. All 'mem'
690 functions have at least one length parameter. Barring any 'l' or 's' suffix,
691 there is a single length parameter, in position 3, which applies to both
692 string parameters. The 's' suffix means, as described above, that the 2nd
693 string is a literal double-quoted C string (hence its length is calculated by
694 the macro, and the length parameter to the function applies just to the first
695 string parameter, and hence is positioned just after it). An 'l' suffix
696 means that the 2nd string parameter has its own length parameter, and the
697 signature will look like memFOOl(s1, l1, s2, l2).
699 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
700 substring of the 1st string. 'P' if present indicates that the substring
701 must be a "proper" one in tha mathematical sense that the first one must be
702 strictly larger than the 2nd.
707 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
708 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
709 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
710 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
711 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
712 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
714 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
715 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
717 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
718 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
720 /* memEQ and memNE where second comparand is a string constant */
721 #define memEQs(s1, l, s2) \
722 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
723 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
725 /* Keep these private until we decide it was a good idea */
726 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
728 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
730 #define memBEGINs(s1, l, s2) \
731 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
732 && memEQ(s1, "" s2 "", sizeof(s2)-1))
733 #define memBEGINPs(s1, l, s2) \
734 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
735 && memEQ(s1, "" s2 "", sizeof(s2)-1))
736 #define memENDs(s1, l, s2) \
737 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
738 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
739 #define memENDPs(s1, l, s2) \
740 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
741 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
742 #endif /* End of making macros private */
744 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
745 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
746 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
747 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
749 #define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
754 * Unfortunately, the introduction of locales means that we
755 * can't trust isupper(), etc. to tell the truth. And when
756 * it comes to /\w+/ with tainting enabled, we *must* be able
757 * to trust our character classes.
759 * Therefore, the default tests in the text of Perl will be
760 * independent of locale. Any code that wants to depend on
761 * the current locale will use the tests that begin with "lc".
764 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
772 =head1 Character classification
773 This section is about functions (really macros) that classify characters
774 into types, such as punctuation versus alphabetic, etc. Most of these are
775 analogous to regular expression character classes. (See
776 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
777 each class. (Not all macros have all variants; each item below lists the
778 ones valid for it.) None are affected by C<use bytes>, and only the ones
779 with C<LC> in the name are affected by the current locale.
781 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
782 treating it as a code point, and returns a boolean as to whether or not the
783 character represented by it is (or on non-ASCII platforms, corresponds to) an
784 ASCII character in the named class based on platform, Unicode, and Perl rules.
785 If the input is a number that doesn't fit in an octet, FALSE is returned.
787 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
788 with no suffix C<"_A">. This variant is used to emphasize by its name that
789 only ASCII-range characters can return TRUE.
791 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
792 onto the platform. That is, the code points that are ASCII are unaffected,
793 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
794 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
795 true when called with the code point 0xDF, which is a word character in both
796 ASCII and EBCDIC (though it represents different characters in each).
797 If the input is a number that doesn't fit in an octet, FALSE is returned.
798 (Perl's documentation uses a colloquial definition of Latin-1, to include all
799 code points below 256.)
801 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
802 inputs below 256, but if the code point is larger than 255, Unicode rules are
803 used to determine if it is in the character class. For example,
804 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
805 WITH MACRON in Unicode, and is a word character.
807 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
808 but are used for UTF-8 encoded strings. The two forms are different names for
809 the same thing. Each call to one of these classifies the first character of
810 the string starting at C<p>. The second parameter, C<e>, points to anywhere in
811 the string beyond the first character, up to one byte past the end of the
812 entire string. Although both variants are identical, the suffix C<_safe> in
813 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
814 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
815 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
816 some way, the program may croak, or the function may return FALSE, at the
817 discretion of the implementation, and subject to change in future releases.
819 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
820 but the result is based on the current locale, which is what C<LC> in the name
821 stands for. If Perl can determine that the current locale is a UTF-8 locale,
822 it uses the published Unicode rules; otherwise, it uses the C library function
823 that gives the named classification. For example, C<isDIGIT_LC()> when not in
824 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
825 returned if the input won't fit into an octet. On some platforms where the C
826 library function is known to be defective, Perl changes its result to follow
827 the POSIX standard's rules.
829 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
830 than 256, but for larger ones it returns the Unicode classification of the code
833 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
834 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
835 are different names for the same thing. Each call to one of these classifies
836 the first character of the string starting at C<p>. The second parameter,
837 C<e>, points to anywhere in the string beyond the first character, up to one
838 byte past the end of the entire string. Although both variants are identical,
839 the suffix C<_safe> in one name emphasizes that it will not attempt to read
840 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
841 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
842 character is malformed in some way, the program may croak, or the function may
843 return FALSE, at the discretion of the implementation, and subject to change in
846 =for apidoc Am|bool|isALPHA|UV ch
847 =for apidoc_item ||isALPHA_A|UV ch
848 =for apidoc_item ||isALPHA_L1|UV ch
849 =for apidoc_item ||isALPHA_uvchr|UV ch
850 =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
851 =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
852 =for apidoc_item ||isALPHA_LC|UV ch
853 =for apidoc_item ||isALPHA_LC_uvchr|UV ch
854 =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
855 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
856 analogous to C<m/[[:alpha:]]/>.
857 See the L<top of this section|/Character classification> for an explanation of
862 Here and below, we add the prototypes of these macros for downstream programs
863 that would be interested in them, such as Devel::PPPort
865 =for apidoc Am|bool|isALPHANUMERIC|UV ch
866 =for apidoc_item ||isALPHANUMERIC_A|UV ch
867 =for apidoc_item ||isALPHANUMERIC_L1|UV ch
868 =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
869 =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
870 =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
871 =for apidoc_item ||isALPHANUMERIC_LC|UV ch
872 =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
873 =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
874 =for apidoc_item ||isALNUMC|UV ch
875 =for apidoc_item ||isALNUMC_A|UV ch
876 =for apidoc_item ||isALNUMC_L1|UV ch
877 =for apidoc_item ||isALNUMC_LC|UV ch
878 =for apidoc_item ||isALNUMC_LC_uvchr|UV ch
879 Returns a boolean indicating whether the specified character is one of
880 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
881 See the L<top of this section|/Character classification> for an explanation of
884 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
885 this corresponds to the C language alphanumeric definition). Also
886 there are the variants
887 C<isALNUMC_A>, C<isALNUMC_L1>
888 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
890 =for apidoc Am|bool|isASCII|UV ch
891 =for apidoc_item ||isASCII_A|UV ch
892 =for apidoc_item ||isASCII_L1|UV ch
893 =for apidoc_item ||isASCII_uvchr|UV ch
894 =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
895 =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
896 =for apidoc_item ||isASCII_LC|UV ch
897 =for apidoc_item ||isASCII_LC_uvchr|UV ch
898 =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
899 Returns a boolean indicating whether the specified character is one of the 128
900 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
901 On non-ASCII platforms, it returns TRUE iff this
902 character corresponds to an ASCII character. Variants C<isASCII_A()> and
903 C<isASCII_L1()> are identical to C<isASCII()>.
904 See the L<top of this section|/Character classification> for an explanation of
906 Note, however, that some platforms do not have the C library routine
907 C<isascii()>. In these cases, the variants whose names contain C<LC> are the
908 same as the corresponding ones without.
910 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
911 have the exact same representation (always a single byte) whether encoded in
912 UTF-8 or not), C<isASCII> will give the correct results when called with any
913 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
914 C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
916 =for apidoc Am|bool|isBLANK|UV ch
917 =for apidoc_item ||isBLANK_A|UV ch
918 =for apidoc_item ||isBLANK_L1|UV ch
919 =for apidoc_item ||isBLANK_uvchr|UV ch
920 =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
921 =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
922 =for apidoc_item ||isBLANK_LC|UV ch
923 =for apidoc_item ||isBLANK_LC_uvchr|UV ch
924 =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
925 Returns a boolean indicating whether the specified character is a
926 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
927 See the L<top of this section|/Character classification> for an explanation of
930 however, that some platforms do not have the C library routine
931 C<isblank()>. In these cases, the variants whose names contain C<LC> are
932 the same as the corresponding ones without.
934 =for apidoc Am|bool|isCNTRL|UV ch
935 =for apidoc_item ||isCNTRL_A|UV ch
936 =for apidoc_item ||isCNTRL_L1|UV ch
937 =for apidoc_item ||isCNTRL_uvchr|UV ch
938 =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
939 =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
940 =for apidoc_item ||isCNTRL_LC|UV ch
941 =for apidoc_item ||isCNTRL_LC_uvchr|UV ch
942 =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
944 Returns a boolean indicating whether the specified character is a
945 control character, analogous to C<m/[[:cntrl:]]/>.
946 See the L<top of this section|/Character classification> for an explanation of
948 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
950 =for apidoc Am|bool|isDIGIT|UV ch
951 =for apidoc_item ||isDIGIT_A|UV ch
952 =for apidoc_item ||isDIGIT_L1|UV ch
953 =for apidoc_item ||isDIGIT_uvchr|UV ch
954 =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
955 =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
956 =for apidoc_item ||isDIGIT_LC|UV ch
957 =for apidoc_item ||isDIGIT_LC_uvchr|UV ch
958 =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
960 Returns a boolean indicating whether the specified character is a
961 digit, analogous to C<m/[[:digit:]]/>.
962 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
963 See the L<top of this section|/Character classification> for an explanation of
966 =for apidoc Am|bool|isGRAPH|UV ch
967 =for apidoc_item ||isGRAPH_A|UV ch
968 =for apidoc_item ||isGRAPH_L1|UV ch
969 =for apidoc_item ||isGRAPH_uvchr|UV ch
970 =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
971 =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
972 =for apidoc_item ||isGRAPH_LC|UV ch
973 =for apidoc_item ||isGRAPH_LC_uvchr|UV ch
974 =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
975 Returns a boolean indicating whether the specified character is a
976 graphic character, analogous to C<m/[[:graph:]]/>.
977 See the L<top of this section|/Character classification> for an explanation of
980 =for apidoc Am|bool|isLOWER|UV ch
981 =for apidoc_item ||isLOWER_A|UV ch
982 =for apidoc_item ||isLOWER_L1|UV ch
983 =for apidoc_item ||isLOWER_uvchr|UV ch
984 =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
985 =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
986 =for apidoc_item ||isLOWER_LC|UV ch
987 =for apidoc_item ||isLOWER_LC_uvchr|UV ch
988 =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
989 Returns a boolean indicating whether the specified character is a
990 lowercase character, analogous to C<m/[[:lower:]]/>.
991 See the L<top of this section|/Character classification> for an explanation of
994 =for apidoc Am|bool|isOCTAL|UV ch
995 =for apidoc_item ||isOCTAL_A|UV ch
996 =for apidoc_item ||isOCTAL_L1|UV ch
997 Returns a boolean indicating whether the specified character is an
999 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
1002 =for apidoc Am|bool|isPUNCT|UV ch
1003 =for apidoc_item ||isPUNCT_A|UV ch
1004 =for apidoc_item ||isPUNCT_L1|UV ch
1005 =for apidoc_item ||isPUNCT_uvchr|UV ch
1006 =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
1007 =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
1008 =for apidoc_item ||isPUNCT_LC|UV ch
1009 =for apidoc_item ||isPUNCT_LC_uvchr|UV ch
1010 =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
1011 Returns a boolean indicating whether the specified character is a
1012 punctuation character, analogous to C<m/[[:punct:]]/>.
1013 Note that the definition of what is punctuation isn't as
1014 straightforward as one might desire. See L<perlrecharclass/POSIX Character
1015 Classes> for details.
1016 See the L<top of this section|/Character classification> for an explanation of
1019 =for apidoc Am|bool|isSPACE|UV ch
1020 =for apidoc_item ||isSPACE_A|UV ch
1021 =for apidoc_item ||isSPACE_L1|UV ch
1022 =for apidoc_item ||isSPACE_uvchr|UV ch
1023 =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
1024 =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
1025 =for apidoc_item ||isSPACE_LC|UV ch
1026 =for apidoc_item ||isSPACE_LC_uvchr|UV ch
1027 =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
1028 Returns a boolean indicating whether the specified character is a
1029 whitespace character. This is analogous
1030 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
1031 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
1032 locale forms of this macro (the ones with C<LC> in their names) matched
1033 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1034 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1035 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
1036 See the L<top of this section|/Character classification> for an explanation of
1039 =for apidoc Am|bool|isPSXSPC|UV ch
1040 =for apidoc_item ||isPSXSPC_A|UV ch
1041 =for apidoc_item ||isPSXSPC_L1|UV ch
1042 =for apidoc_item ||isPSXSPC_uvchr|UV ch
1043 =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1044 =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
1045 =for apidoc_item ||isPSXSPC_LC|UV ch
1046 =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
1047 =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
1048 (short for Posix Space)
1049 Starting in 5.18, this is identical in all its forms to the
1050 corresponding C<isSPACE()> macros.
1051 The locale forms of this macro are identical to their corresponding
1052 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1053 non-locale forms differ from their C<isSPACE()> forms only in that the
1054 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1055 Otherwise they are identical. Thus this macro is analogous to what
1056 C<m/[[:space:]]/> matches in a regular expression.
1057 See the L<top of this section|/Character classification> for an explanation of
1060 =for apidoc Am|bool|isUPPER|UV ch
1061 =for apidoc_item ||isUPPER_A|UV ch
1062 =for apidoc_item ||isUPPER_L1|UV ch
1063 =for apidoc_item ||isUPPER_uvchr|UV ch
1064 =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1065 =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
1066 =for apidoc_item ||isUPPER_LC|UV ch
1067 =for apidoc_item ||isUPPER_LC_uvchr|UV ch
1068 =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
1069 Returns a boolean indicating whether the specified character is an
1070 uppercase character, analogous to C<m/[[:upper:]]/>.
1071 See the L<top of this section|/Character classification> for an explanation of
1074 =for apidoc Am|bool|isPRINT|UV ch
1075 =for apidoc_item ||isPRINT_A|UV ch
1076 =for apidoc_item ||isPRINT_L1|UV ch
1077 =for apidoc_item ||isPRINT_uvchr|UV ch
1078 =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1079 =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
1080 =for apidoc_item ||isPRINT_LC|UV ch
1081 =for apidoc_item ||isPRINT_LC_uvchr|UV ch
1082 =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
1083 Returns a boolean indicating whether the specified character is a
1084 printable character, analogous to C<m/[[:print:]]/>.
1085 See the L<top of this section|/Character classification> for an explanation of
1088 =for apidoc Am|bool|isWORDCHAR|UV ch
1089 =for apidoc_item ||isWORDCHAR_A|UV ch
1090 =for apidoc_item ||isWORDCHAR_L1|UV ch
1091 =for apidoc_item ||isWORDCHAR_uvchr|UV ch
1092 =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1093 =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
1094 =for apidoc_item ||isWORDCHAR_LC|UV ch
1095 =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
1096 =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1097 =for apidoc_item ||isALNUM|UV ch
1098 =for apidoc_item ||isALNUM_A|UV ch
1099 =for apidoc_item ||isALNUM_LC|UV ch
1100 =for apidoc_item ||isALNUM_LC_uvchr|UV ch
1101 Returns a boolean indicating whether the specified character is a character
1102 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1103 in a regular expression. A word character is an alphabetic character, a
1104 decimal digit, a connecting punctuation character (such as an underscore), or
1105 a "mark" character that attaches to one of those (like some sort of accent).
1106 C<isALNUM()> is a synonym provided for backward compatibility, even though a
1107 word character includes more than the standard C language meaning of
1109 See the L<top of this section|/Character classification> for an explanation of
1111 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1112 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1113 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1114 include the platform's native underscore.
1116 =for apidoc Am|bool|isXDIGIT|UV ch
1117 =for apidoc_item ||isXDIGIT_A|UV ch
1118 =for apidoc_item ||isXDIGIT_L1|UV ch
1119 =for apidoc_item ||isXDIGIT_uvchr|UV ch
1120 =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1121 =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
1122 =for apidoc_item ||isXDIGIT_LC|UV ch
1123 =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
1124 =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1125 Returns a boolean indicating whether the specified character is a hexadecimal
1126 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1127 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1128 See the L<top of this section|/Character classification> for an explanation of
1131 =for apidoc Am|bool|isIDFIRST|UV ch
1132 =for apidoc_item ||isIDFIRST_A|UV ch
1133 =for apidoc_item ||isIDFIRST_L1|UV ch
1134 =for apidoc_item ||isIDFIRST_uvchr|UV ch
1135 =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1136 =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
1137 =for apidoc_item ||isIDFIRST_LC|UV ch
1138 =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
1139 =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1140 Returns a boolean indicating whether the specified character can be the first
1141 character of an identifier. This is very close to, but not quite the same as
1142 the official Unicode property C<XID_Start>. The difference is that this
1143 returns true only if the input character also matches L</isWORDCHAR>.
1144 See the L<top of this section|/Character classification> for an explanation of
1147 =for apidoc Am|bool|isIDCONT|UV ch
1148 =for apidoc_item ||isIDCONT_A|UV ch
1149 =for apidoc_item ||isIDCONT_L1|UV ch
1150 =for apidoc_item ||isIDCONT_uvchr|UV ch
1151 =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1152 =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
1153 =for apidoc_item ||isIDCONT_LC|UV ch
1154 =for apidoc_item ||isIDCONT_LC_uvchr|UV ch
1155 =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1156 Returns a boolean indicating whether the specified character can be the
1157 second or succeeding character of an identifier. This is very close to, but
1158 not quite the same as the official Unicode property C<XID_Continue>. The
1159 difference is that this returns true only if the input character also matches
1160 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1161 an explanation of the variants.
1163 =for apidoc_section $numeric
1165 =for apidoc Am|U8|READ_XDIGIT|char str*
1166 Returns the value of an ASCII-range hex digit and advances the string pointer.
1167 Behaviour is only well defined when isXDIGIT(*str) is true.
1169 =head1 Character case changing
1170 Perl uses "full" Unicode case mappings. This means that converting a single
1171 character to another case may result in a sequence of more than one character.
1172 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1173 character sequence C<SS>. This presents some complications The lowercase of
1174 all characters in the range 0..255 is a single character, and thus
1175 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1176 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1177 an API that does allow every possible legal result to be returned.) Likewise
1178 no other function that is crippled by not being able to give the correct
1179 results for the full range of possible inputs has been implemented here.
1181 =for apidoc Am|U8|toUPPER|int ch
1182 Converts the specified character to uppercase. If the input is anything but an
1183 ASCII lowercase character, that input character itself is returned. Variant
1184 C<toUPPER_A> is equivalent.
1186 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1187 Converts the code point C<cp> to its uppercase version, and
1188 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1189 point is interpreted as native if less than 256; otherwise as Unicode. Note
1190 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1191 bytes since the uppercase version may be longer than the original character.
1193 The first code point of the uppercased version is returned
1194 (but note, as explained at L<the top of this section|/Character case
1195 changing>, that there may be more.)
1197 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1198 =for apidoc_item toUPPER_utf8_safe
1199 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1200 extending no further than S<C<e - 1>> to its uppercase version, and
1201 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1202 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1203 bytes since the uppercase version may be longer than the original character.
1205 The first code point of the uppercased version is returned
1206 (but note, as explained at L<the top of this section|/Character case
1207 changing>, that there may be more).
1209 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1210 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1211 the UTF-8 for the input character is malformed in some way, the program may
1212 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1213 of the implementation, and subject to change in future releases.
1215 C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1217 =for apidoc Am|U8|toFOLD|U8 ch
1218 Converts the specified character to foldcase. If the input is anything but an
1219 ASCII uppercase character, that input character itself is returned. Variant
1220 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
1221 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
1223 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1224 Converts the code point C<cp> to its foldcase version, and
1225 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1226 point is interpreted as native if less than 256; otherwise as Unicode. Note
1227 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1228 bytes since the foldcase version may be longer than the original character.
1230 The first code point of the foldcased version is returned
1231 (but note, as explained at L<the top of this section|/Character case
1232 changing>, that there may be more).
1234 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1235 =for apidoc_item toFOLD_utf8_safe
1236 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1237 extending no further than S<C<e - 1>> to its foldcase version, and
1238 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1239 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1240 bytes since the foldcase version may be longer than the original character.
1242 The first code point of the foldcased version is returned
1243 (but note, as explained at L<the top of this section|/Character case
1244 changing>, that there may be more).
1247 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1248 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1249 input character is malformed in some way, the program may croak, or the
1250 function may return the REPLACEMENT CHARACTER, at the discretion of the
1251 implementation, and subject to change in future releases.
1253 C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1255 =for apidoc Am|UV|toLOWER|UV cp
1256 =for apidoc_item |UV|toLOWER_A|UV cp
1257 =for apidoc_item |UV|toLOWER_L1|UV cp
1258 =for apidoc_item |UV|toLOWER_LATIN1|UV cp
1259 =for apidoc_item |UV|toLOWER_LC|UV cp
1260 =for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1261 =for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1262 =for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1264 These all return the lowercase of a character. The differences are what domain
1265 they operate on, and whether the input is specified as a code point (those
1266 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1267 case, the code point to use is the first one in the buffer of UTF-8 encoded
1268 code points, delineated by the arguments S<C<p .. e - 1>>.
1270 C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the
1271 lowercase of any uppercase ASCII-range code point. All other inputs are
1272 returned unchanged. Since these are macros, the input type may be any integral
1273 one, and the output will occupy the same number of bits as the input.
1275 C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave
1276 identically as C<toLOWER> for ASCII-range input. But additionally will return
1277 the lowercase of any uppercase code point in the entire 0..255 range, assuming
1278 a Latin-1 encoding (or the EBCDIC equivalent on such platforms).
1280 C<toLOWER_LC> returns the lowercase of the input code point according to the
1281 rules of the current POSIX locale. Input code points outside the range 0..255
1282 are returned unchanged.
1284 C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return
1285 value is identical to that of C<toLOWER_L1> for input code points in the 0..255
1286 range. The lowercase of the vast majority of Unicode code points is the same
1287 as the code point itself. For these, and for code points above the legal
1288 Unicode maximum, this returns the input code point unchanged. It additionally
1289 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1290 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1291 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1293 NOTE: the lowercase of a code point may be more than one code point. The
1294 return value of this function is only the first of these. The entire lowercase
1295 is returned in C<s>. To determine if the result is more than a single code
1296 point, you can do something like this:
1298 uc = toLOWER_uvchr(cp, s, &len);
1299 if (len > UTF8SKIP(s)) { is multiple code points }
1300 else { is a single code point }
1302 C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only
1303 difference between these and C<toLOWER_uvchr> is that the source for these is
1304 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1305 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1306 buffer may certainly contain more than one code point; but only the first one
1307 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1308 malformed in some way, the program may croak, or the function may return the
1309 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1310 change in future releases.
1312 =for apidoc Am|U8|toTITLE|U8 ch
1313 Converts the specified character to titlecase. If the input is anything but an
1314 ASCII lowercase character, that input character itself is returned. Variant
1315 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1316 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1317 not a concept used in locale handling, so there is no functionality for that.)
1319 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1320 Converts the code point C<cp> to its titlecase version, and
1321 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1322 point is interpreted as native if less than 256; otherwise as Unicode. Note
1323 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1324 bytes since the titlecase version may be longer than the original character.
1326 The first code point of the titlecased version is returned
1327 (but note, as explained at L<the top of this section|/Character case
1328 changing>, that there may be more).
1330 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1331 =for apidoc_item toTITLE_utf8_safe
1332 Convert the first UTF-8 encoded character in the sequence starting at C<p> and
1333 extending no further than S<C<e - 1>> to its titlecase version, and
1334 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1335 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1336 bytes since the titlecase version may be longer than the original character.
1338 The first code point of the titlecased version is returned
1339 (but note, as explained at L<the top of this section|/Character case
1340 changing>, that there may be more).
1343 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1344 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1345 input character is malformed in some way, the program may croak, or the
1346 function may return the REPLACEMENT CHARACTER, at the discretion of the
1347 implementation, and subject to change in future releases.
1349 C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1353 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1354 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1355 and aren't general purpose as they don't work on U+DF, and assert against that.
1357 Note that these macros are repeated in Devel::PPPort, so should also be
1358 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1363 void below because that's the best fit, and works for Devel::PPPort
1364 =for apidoc_section $integer
1365 =for apidoc AyT||WIDEST_UTYPE
1367 Yields the widest unsigned integer type on the platform, currently either
1368 C<U32> or C<U64>. This can be used in declarations such as
1374 my_uv = (WIDEST_UTYPE) val;
1380 # define WIDEST_UTYPE U64
1382 # define WIDEST_UTYPE U32
1385 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1386 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1387 * bits of information are lost even on a 64-bit machine, but to get the
1388 * compiler to optimize it out if possible. This is because Configure makes
1389 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1390 * sizeof() guarantees that this evaluates to a constant true at compile time.
1392 * For Coverity, be always true, because otherwise Coverity thinks
1393 * it finds several expressions that are always true, independent
1394 * of operands. Well, they are, but that is kind of the point.
1396 #ifndef __COVERITY__
1397 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1399 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1400 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1402 #define FITS_IN_8_BITS(c) (1)
1405 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
1406 * Written this way so that after optimization, only one conditional test is
1407 * needed. (The NV casts stop any warnings about comparison always being true
1408 * if called with an unsigned. The cast preserves the sign, which is all we
1410 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1411 __ASSERT_((NV) (n) >= 0) \
1412 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
1414 /* Returns true if c is in the range l..u, where 'l' is non-negative
1415 * Written this way so that after optimization, only one conditional test is
1417 #define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
1418 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1419 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1420 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1421 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
1424 # ifndef _ALL_SOURCE
1425 /* The native libc isascii() et.al. functions return the wrong results
1426 * on at least z/OS unless this is defined. */
1427 # error _ALL_SOURCE should probably be defined
1430 /* There is a simple definition of ASCII for ASCII platforms. But the
1431 * EBCDIC one isn't so simple, so is defined using table look-up like the
1432 * other macros below.
1434 * The cast here is used instead of '(c) >= 0', because some compilers emit
1435 * a warning that that test is always true when the parameter is an
1436 * unsigned type. khw supposes that it could be written as
1437 * && ((c) == '\0' || (c) > 0)
1438 * to avoid the message, but the cast will likely avoid extra branches even
1439 * with stupid compilers.
1441 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1443 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1446 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1447 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1448 * can be ignored. If the rest match '0', we have an octal */
1449 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1451 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1453 /* Character class numbers. For internal core Perl use only. The ones less
1454 * than 32 are used in PL_charclass[] and the ones up through the one that
1455 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1456 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1457 * their actual definitions are here. If that file has a name not used here,
1460 * The first group of these is ordered in what I (khw) estimate to be the
1461 * frequency of their use. This gives a slight edge to exiting a loop earlier
1462 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1463 * real Posix character class, and some (small) inefficiencies in regular
1464 * expression handling would be introduced by putting it in the middle of those
1465 * that are. Also, cntrl and ascii come after the others as it may be useful
1466 * to group these which have no members that match above Latin1, (or above
1467 * ASCII in the latter case) */
1469 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1470 # define _CC_DIGIT 1 /* \d and [:digit:] */
1471 # define _CC_ALPHA 2 /* [:alpha:] */
1472 # define _CC_LOWER 3 /* [:lower:] */
1473 # define _CC_UPPER 4 /* [:upper:] */
1474 # define _CC_PUNCT 5 /* [:punct:] */
1475 # define _CC_PRINT 6 /* [:print:] */
1476 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1477 # define _CC_GRAPH 8 /* [:graph:] */
1478 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1479 # define _CC_SPACE 10 /* \s, [:space:] */
1480 # define _CC_BLANK 11 /* [:blank:] */
1481 # define _CC_XDIGIT 12 /* [:xdigit:] */
1482 # define _CC_CNTRL 13 /* [:cntrl:] */
1483 # define _CC_ASCII 14 /* [:ascii:] */
1484 # define _CC_VERTSPACE 15 /* \v */
1486 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1488 /* The members of the third group below do not need to be coordinated with data
1489 * structures in regcomp.[ch] and regexec.c. */
1490 # define _CC_IDFIRST 16
1491 # define _CC_CHARNAME_CONT 17
1492 # define _CC_NONLATIN1_FOLD 18
1493 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1494 # define _CC_QUOTEMETA 20
1495 # define _CC_NON_FINAL_FOLD 21
1496 # define _CC_IS_IN_SOME_FOLD 22
1497 # define _CC_BINDIGIT 23
1498 # define _CC_OCTDIGIT 24
1499 # define _CC_MNEMONIC_CNTRL 25
1501 /* This next group is only used on EBCDIC platforms, so theoretically could be
1502 * shared with something entirely different that's only on ASCII platforms */
1503 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1505 * If more bits are needed, one could add a second word for non-64bit
1506 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1507 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1508 * is used only for optimization (as of this writing), and differs in the
1509 * Latin1 range from the ALPHA bit only in two relatively unimportant
1510 * characters: the masculine and feminine ordinal indicators, so removing it
1511 * would just cause /i regexes which match them to run less efficiently.
1512 * Similarly the EBCDIC-only bits are used just for speed, and could be
1513 * replaced by other means */
1515 #if defined(PERL_CORE) || defined(PERL_EXT)
1516 /* An enum version of the character class numbers, to help compilers
1519 _CC_ENUM_ALPHA = _CC_ALPHA,
1520 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1521 _CC_ENUM_ASCII = _CC_ASCII,
1522 _CC_ENUM_BLANK = _CC_BLANK,
1523 _CC_ENUM_CASED = _CC_CASED,
1524 _CC_ENUM_CNTRL = _CC_CNTRL,
1525 _CC_ENUM_DIGIT = _CC_DIGIT,
1526 _CC_ENUM_GRAPH = _CC_GRAPH,
1527 _CC_ENUM_LOWER = _CC_LOWER,
1528 _CC_ENUM_PRINT = _CC_PRINT,
1529 _CC_ENUM_PUNCT = _CC_PUNCT,
1530 _CC_ENUM_SPACE = _CC_SPACE,
1531 _CC_ENUM_UPPER = _CC_UPPER,
1532 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1533 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1534 _CC_ENUM_XDIGIT = _CC_XDIGIT
1535 } _char_class_number;
1538 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1542 EXTCONST U32 PL_charclass[] = {
1543 # include "l1_char_class_tab.h"
1546 # else /* ! DOINIT */
1547 EXTCONST U32 PL_charclass[];
1551 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1552 # define _CC_mask(classnum) (1U << (classnum))
1554 /* For internal core Perl use only: the base macro for defining macros like
1556 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1557 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1559 /* The mask for the _A versions of the macros; it just adds in the bit for
1561 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1563 /* For internal core Perl use only: the base macro for defining macros like
1564 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1565 * the ASCII bit are present */
1566 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1567 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1568 == _CC_mask_A(classnum)))
1570 /* On ASCII platforms certain classes form a single range. It's faster to
1571 * special case these. isDIGIT is a single range on all platforms */
1573 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1574 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1575 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1576 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1577 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1579 /* By folding the upper and lowercase, we can use a single range */
1580 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1581 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1582 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1583 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1584 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1586 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1587 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1588 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1589 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1590 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1591 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1592 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1593 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1595 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1596 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1597 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1598 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1600 /* continuation character for legal NAME in \N{NAME} */
1601 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1603 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1604 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1605 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1606 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1607 # define isPSXSPC_L1(c) isSPACE_L1(c)
1608 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1609 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1610 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1611 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1612 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1615 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1618 /* Participates in a single-character fold with a character above 255 */
1619 # if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
1620 # define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(c) \
1621 (( ! cBOOL(FITS_IN_8_BITS(c))) \
1622 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1624 # define IS_NON_FINAL_FOLD(c) _generic_isCC(c, _CC_NON_FINAL_FOLD)
1625 # define IS_IN_SOME_FOLD_L1(c) _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1628 /* Like the above, but also can be part of a multi-char fold */
1629 # define HAS_NONLATIN1_FOLD_CLOSURE(c) \
1630 ( (! cBOOL(FITS_IN_8_BITS(c))) \
1631 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
1633 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1635 /* is c a control character for which we have a mnemonic? */
1636 # if defined(PERL_CORE) || defined(PERL_EXT)
1637 # define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1639 #else /* else we don't have perl.h H_PERL */
1641 /* If we don't have perl.h, we are compiling a utility program. Below we
1642 * hard-code various macro definitions that wouldn't otherwise be available
1643 * to it. Most are coded based on first principles. These are written to
1644 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1645 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1646 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1647 # define isSPACE_A(c) (isBLANK_A(c) \
1652 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1653 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1654 * performance (if optimization isn't on), and allow the same code to be
1655 * used for both platform types */
1656 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1657 || inRANGE((c), 'j', 'r') \
1658 || inRANGE((c), 's', 'z')
1659 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1660 || inRANGE((c), 'J', 'R') \
1661 || inRANGE((c), 'S', 'Z')
1662 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1663 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1664 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1665 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1666 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1667 || inRANGE((c), 'a', 'f') \
1668 || inRANGE((c), 'A', 'F')
1669 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1670 || (c) == '#' || (c) == '$' || (c) == '%' \
1671 || (c) == '&' || (c) == '\'' || (c) == '(' \
1672 || (c) == ')' || (c) == '*' || (c) == '+' \
1673 || (c) == ',' || (c) == '.' || (c) == '/' \
1674 || (c) == ':' || (c) == ';' || (c) == '<' \
1675 || (c) == '=' || (c) == '>' || (c) == '?' \
1676 || (c) == '@' || (c) == '[' || (c) == '\\' \
1677 || (c) == ']' || (c) == '^' || (c) == '_' \
1678 || (c) == '`' || (c) == '{' || (c) == '|' \
1679 || (c) == '}' || (c) == '~')
1680 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1681 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1684 /* The below is accurate for the 3 EBCDIC code pages traditionally
1685 * supported by perl. The only difference between them in the controls
1686 * is the position of \n, and that is represented symbolically below */
1687 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1688 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1689 || (c) == '\t' || (c) == '\v' \
1690 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1691 || (c) == 7F /* U+7F DEL */ \
1692 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1694 || (c) == 0x18 /* U+18 CAN */ \
1695 || (c) == 0x19 /* U+19 EOM */ \
1696 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1697 || (c) == 0x26 /* U+17 ETB */ \
1698 || (c) == 0x27 /* U+1B ESC */ \
1699 || (c) == 0x2D /* U+05 ENQ */ \
1700 || (c) == 0x2E /* U+06 ACK */ \
1701 || (c) == 0x32 /* U+16 SYN */ \
1702 || (c) == 0x37 /* U+04 EOT */ \
1703 || (c) == 0x3C /* U+14 DC4 */ \
1704 || (c) == 0x3D /* U+15 NAK */ \
1705 || (c) == 0x3F)/* U+1A SUB */
1706 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1707 # else /* isASCII is already defined for ASCII platforms, so can use that to
1709 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1712 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1713 * during debugging, and it seems best to keep them. We may be called
1714 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1715 * do anything anyway, so make it not a problem */
1716 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1717 # define NATIVE_TO_LATIN1(ch) (ch)
1719 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1720 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1721 # define isBLANK_L1(c) (isBLANK_A(c) \
1722 || (FITS_IN_8_BITS(c) \
1723 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1724 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1725 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1726 # define isLOWER_L1(c) (isLOWER_A(c) \
1727 || (FITS_IN_8_BITS(c) \
1728 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1729 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1730 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1731 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1732 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1733 # define isPRINT_L1(c) (isPRINT_A(c) \
1734 || (FITS_IN_8_BITS(c) \
1735 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1736 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1737 || (FITS_IN_8_BITS(c) \
1738 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1739 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1740 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1741 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1742 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1743 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1744 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1745 # define isSPACE_L1(c) (isSPACE_A(c) \
1746 || (FITS_IN_8_BITS(c) \
1747 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1748 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1749 # define isUPPER_L1(c) (isUPPER_A(c) \
1750 || (FITS_IN_8_BITS(c) \
1751 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1753 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1754 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1755 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1756 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1761 /* The following are not fully accurate in the above-ASCII range. I (khw)
1762 * don't think it's necessary to be so for the purposes where this gets
1764 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1765 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1767 /* And these aren't accurate at all. They are useful only for above
1768 * Latin1, which utilities and bootstrapping don't deal with */
1769 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1770 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1771 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1773 /* Many of the macros later in this file are defined in terms of these. By
1774 * implementing them with a function, which converts the class number into
1775 * a call to the desired macro, all of the later ones work. However, that
1776 * function won't be actually defined when building a utility program (no
1777 * perl.h), and so a compiler error will be generated if one is attempted
1778 * to be used. And the above-Latin1 code points require Unicode tables to
1779 * be present, something unlikely to be the case when bootstrapping */
1780 # define _generic_isCC(c, classnum) \
1781 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1782 # define _generic_isCC_A(c, classnum) \
1783 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1784 #endif /* End of no perl.h H_PERL */
1786 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1787 #define isALPHA(c) isALPHA_A(c)
1788 #define isASCII_A(c) isASCII(c)
1789 #define isASCII_L1(c) isASCII(c)
1790 #define isBLANK(c) isBLANK_A(c)
1791 #define isCNTRL(c) isCNTRL_A(c)
1792 #define isDIGIT(c) isDIGIT_A(c)
1793 #define isGRAPH(c) isGRAPH_A(c)
1794 #define isIDFIRST(c) isIDFIRST_A(c)
1795 #define isLOWER(c) isLOWER_A(c)
1796 #define isPRINT(c) isPRINT_A(c)
1797 #define isPSXSPC_A(c) isSPACE_A(c)
1798 #define isPSXSPC(c) isPSXSPC_A(c)
1799 #define isPSXSPC_L1(c) isSPACE_L1(c)
1800 #define isPUNCT(c) isPUNCT_A(c)
1801 #define isSPACE(c) isSPACE_A(c)
1802 #define isUPPER(c) isUPPER_A(c)
1803 #define isWORDCHAR(c) isWORDCHAR_A(c)
1804 #define isXDIGIT(c) isXDIGIT_A(c)
1806 /* ASCII casing. These could also be written as
1807 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1808 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1809 which uses table lookup and mask instead of subtraction. (This would
1810 work because the _MOD does not apply in the ASCII range).
1812 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1813 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1814 there non-ASCII invariants, and all of them are controls.) */
1815 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1816 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1818 /* In the ASCII range, these are equivalent to what they're here defined to be.
1819 * But by creating these definitions, other code doesn't have to be aware of
1820 * this detail. Actually this works for all UTF-8 invariants, not just the
1821 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1822 #define toFOLD(c) toLOWER(c)
1823 #define toTITLE(c) toUPPER(c)
1825 #define toLOWER_A(c) toLOWER(c)
1826 #define toUPPER_A(c) toUPPER(c)
1827 #define toFOLD_A(c) toFOLD(c)
1828 #define toTITLE_A(c) toTITLE(c)
1830 /* Use table lookup for speed; returns the input itself if is out-of-range */
1831 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1833 : PL_latin1_lc[ (U8) (c) ])
1834 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1836 /* Modified uc. Is correct uc except for three non-ascii chars which are
1837 * all mapped to one of them, and these need special handling; returns the
1838 * input itself if is out-of-range */
1839 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1841 : PL_mod_latin1_uc[ (U8) (c) ])
1842 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1844 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1846 /* For internal core Perl use only: the base macro for defining macros like
1847 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1848 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1849 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1850 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1851 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1852 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1853 * code points under locale rules, use the macros based on _generic_LC_uvchr
1854 * instead of this. */
1855 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1856 (! FITS_IN_8_BITS(c) \
1858 : IN_UTF8_CTYPE_LOCALE \
1859 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1862 /* For internal core Perl use only: a helper macro for defining macros like
1863 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1864 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1865 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1866 * _generic_LC_base for more info */
1867 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1868 _generic_LC_base(c,utf8_locale_classnum, \
1869 non_utf8_func( (_LC_CAST) (c)))
1871 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1872 * 'c' is the platform's native underscore character */
1873 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1874 _generic_LC_base(c, utf8_locale_classnum, \
1875 (non_utf8_func( (_LC_CAST) (c)) \
1876 || (char)(c) == '_'))
1878 /* These next three are also for internal core Perl use only: case-change
1879 * helper macros. The reason for using the PL_latin arrays is in case the
1880 * system function is defective; it ensures uniform results that conform to the
1881 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1882 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1884 : (IN_UTF8_CTYPE_LOCALE) \
1885 ? PL_latin1_lc[ (U8) (c) ] \
1886 : (cast)function((cast)(c)))
1888 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1889 * returns a single value, so can't adequately return the upper case of LATIN
1890 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1891 * values "SS"); instead it asserts against that under DEBUGGING, and
1892 * otherwise returns its input. It does not handle the anomalies in UTF-8
1893 * Turkic locales. */
1894 #define _generic_toUPPER_LC(c, function, cast) \
1895 (! FITS_IN_8_BITS(c) \
1897 : ((! IN_UTF8_CTYPE_LOCALE) \
1898 ? (cast)function((cast)(c)) \
1899 : ((((U8)(c)) == MICRO_SIGN) \
1900 ? GREEK_CAPITAL_LETTER_MU \
1901 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1902 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1903 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1904 ? (__ASSERT_(0) (c)) \
1905 : PL_mod_latin1_uc[ (U8) (c) ])))))
1907 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1908 * returns a single value, so can't adequately return the fold case of LATIN
1909 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1910 * values "ss"); instead it asserts against that under DEBUGGING, and
1911 * otherwise returns its input. It does not handle the anomalies in UTF-8
1913 #define _generic_toFOLD_LC(c, function, cast) \
1914 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1915 ? GREEK_SMALL_LETTER_MU \
1916 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1917 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1918 _generic_toLOWER_LC(c, function, cast)))
1920 /* Use the libc versions for these if available. */
1921 #if defined(HAS_ISASCII)
1922 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1924 # define isASCII_LC(c) isASCII(c)
1927 #if defined(HAS_ISBLANK)
1928 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1929 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1930 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1936 /* The Windows functions don't bother to follow the POSIX standard, which
1937 * for example says that something can't both be a printable and a control.
1938 * But Windows treats the \t control as a printable, and does such things
1939 * as making superscripts into both digits and punctuation. This tames
1940 * these flaws by assuming that the definitions of both controls and space
1941 * are correct, and then making sure that other definitions don't have
1942 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1943 * Not all possible weirdnesses are checked for, just the ones that were
1944 * detected on actual Microsoft code pages */
1946 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1947 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1949 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1950 && isALPHANUMERIC_LC(c))
1951 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1953 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1954 isALPHANUMERIC_LC(c))
1955 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1956 # define isIDFIRST_LC(c) (((c) == '_') \
1957 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1958 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1959 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1960 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1961 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1962 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1963 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1964 && isALPHANUMERIC_LC(c))
1966 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1967 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1968 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1970 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1971 /* For most other platforms */
1973 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1974 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1975 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1976 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1977 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1978 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1979 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1980 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1981 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1982 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1983 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1984 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1985 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1988 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1989 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1990 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1992 #else /* The final fallback position */
1994 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1995 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1996 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1997 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1998 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1999 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
2000 # define isLOWER_LC(c) (isascii(c) && islower(c))
2001 # define isPRINT_LC(c) (isascii(c) && isprint(c))
2002 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
2003 # define isSPACE_LC(c) (isascii(c) && isspace(c))
2004 # define isUPPER_LC(c) (isascii(c) && isupper(c))
2005 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
2006 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
2008 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
2009 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
2010 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
2014 #define isIDCONT(c) isWORDCHAR(c)
2015 #define isIDCONT_A(c) isWORDCHAR_A(c)
2016 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
2017 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
2018 #define isPSXSPC_LC(c) isSPACE_LC(c)
2020 /* For internal core Perl use only: the base macros for defining macros like
2021 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
2022 * number defined earlier in this file. _generic_uvchr() is used for POSIX
2023 * classes where there is a macro or function 'above_latin1' that takes the
2024 * single argument 'c' and returns the desired value. These exist for those
2025 * classes which have simple definitions, avoiding the overhead of an inversion
2026 * list binary search. _generic_invlist_uvchr() can be used
2027 * for classes where that overhead is faster than a direct lookup.
2028 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
2029 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2030 * have duplicate checks here, so could create versions of the macros that
2031 * don't, but experiments show that gcc optimizes them out anyway. */
2033 /* Note that all ignore 'use bytes' */
2034 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2035 ? _generic_isCC(c, classnum) \
2037 #define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
2038 ? _generic_isCC(c, classnum) \
2039 : _is_uni_FOO(classnum, c))
2040 #define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2041 #define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
2042 #define isASCII_uvchr(c) isASCII(c)
2043 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2044 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2045 #define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2046 #define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
2047 #define isIDCONT_uvchr(c) \
2048 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2049 #define isIDFIRST_uvchr(c) \
2050 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2051 #define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2052 #define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
2054 #define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
2055 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2056 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2058 #define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
2059 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2060 #define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
2061 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2063 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2064 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2065 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2066 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2068 /* For backwards compatibility, even though '_uni' should mean official Unicode
2069 * code points, in Perl it means native for those below 256 */
2070 #define isALPHA_uni(c) isALPHA_uvchr(c)
2071 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2072 #define isASCII_uni(c) isASCII_uvchr(c)
2073 #define isBLANK_uni(c) isBLANK_uvchr(c)
2074 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
2075 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
2076 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
2077 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
2078 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2079 #define isLOWER_uni(c) isLOWER_uvchr(c)
2080 #define isPRINT_uni(c) isPRINT_uvchr(c)
2081 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
2082 #define isSPACE_uni(c) isSPACE_uvchr(c)
2083 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2084 #define isUPPER_uni(c) isUPPER_uvchr(c)
2085 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
2086 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2087 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2088 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2089 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2090 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2091 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2093 /* For internal core Perl use only: the base macros for defining macros like
2094 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2095 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2096 * simple class definitions; the other for more complex. These are like
2097 * _generic_uvchr, so see it for more info. */
2098 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2099 (c < 256 ? latin1(c) : above_latin1(c))
2100 #define _generic_LC_invlist_uvchr(latin1, classnum, c) \
2101 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2103 #define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2104 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
2105 _CC_ALPHANUMERIC, c)
2106 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2107 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2108 is_HORIZWS_cp_high, c)
2109 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2110 #define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2111 #define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2112 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2113 _is_uni_perl_idcont, c)
2114 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2115 _is_uni_perl_idstart, c)
2116 #define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2117 #define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
2118 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2119 #define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2120 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2121 is_XPERLSPACE_cp_high, c)
2122 #define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2123 #define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
2125 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2126 is_XDIGIT_cp_high, c)
2128 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2130 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2131 * they don't otherwise go out of their way to look for malformed UTF-8. If
2132 * they can return accurate results without knowing if the input is otherwise
2133 * malformed, they do so. For example isASCII is accurate in spite of any
2134 * non-length malformations because it looks only at a single byte. Likewise
2135 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2136 * variant ones return FALSE. But, if the input has to be well-formed in order
2137 * for the results to be accurate, the macros will test and if malformed will
2138 * call a routine to die
2140 * Except for toke.c, the macros do assume that e > p, asserting that on
2141 * DEBUGGING builds. Much code that calls these depends on this being true,
2142 * for other reasons. toke.c is treated specially as using the regular
2143 * assertion breaks it in many ways. All strings that these operate on there
2144 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2145 * bunch of code in toke.c assumes that this is true, so the assertion allows
2147 #ifdef PERL_IN_TOKE_C
2148 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2150 # define _utf8_safe_assert(p,e) ((e) > (p))
2153 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2154 ((! _utf8_safe_assert(p, e)) \
2155 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2156 : (UTF8_IS_INVARIANT(*(p))) \
2157 ? _generic_isCC(*(p), classnum) \
2158 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2159 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2160 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2162 : (_force_out_malformed_utf8_message( \
2163 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2165 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2166 * 'above_latin1' can be a macro */
2167 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2168 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2169 #define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2170 _generic_utf8_safe(classnum, p, e, \
2171 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2172 ? (_force_out_malformed_utf8_message( \
2173 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2175 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2176 * 'above_latin1' parameter */
2177 #define _generic_invlist_utf8_safe(classnum, p, e) \
2178 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
2180 /* Like the above, but should be used only when it is known that there are no
2181 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2182 * class is TRUE for. Hence it can skip the tests for this range.
2183 * 'above_latin1' should include its arguments */
2184 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2185 (__ASSERT_(_utf8_safe_assert(p, e)) \
2186 (UTF8_IS_INVARIANT(*(p))) \
2187 ? _generic_isCC(*(p), classnum) \
2188 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2189 ? 0 /* Note that doesn't check validity for latin1 */ \
2193 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2194 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2195 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2196 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2197 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2198 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2199 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2200 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2201 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2202 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2203 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2204 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2205 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2206 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2207 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2208 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2209 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2210 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
2212 #define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
2213 #define isALPHANUMERIC_utf8_safe(p, e) \
2214 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
2215 #define isASCII_utf8_safe(p, e) \
2216 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2218 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2219 #define isBLANK_utf8_safe(p, e) \
2220 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2223 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2224 * more efficient macro instead of the more general one */
2225 # define isCNTRL_utf8_safe(p, e) \
2226 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2228 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2231 #define isDIGIT_utf8_safe(p, e) \
2232 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2233 _is_utf8_FOO(_CC_DIGIT, p, e))
2234 #define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
2235 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2236 _is_utf8_perl_idcont, p, e)
2238 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2239 * IDFIRST is an alnum. See
2240 * https://github.com/Perl/perl5/issues/10275 for more detail than you
2241 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2242 * This used to be not the XID version, but we decided to go with the more
2243 * modern Unicode definition */
2244 #define isIDFIRST_utf8_safe(p, e) \
2245 _generic_func_utf8_safe(_CC_IDFIRST, \
2246 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
2248 #define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2249 #define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
2250 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2251 #define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
2252 #define isSPACE_utf8_safe(p, e) \
2253 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2254 #define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
2255 #define isVERTWS_utf8_safe(p, e) \
2256 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2257 #define isWORDCHAR_utf8_safe(p, e) \
2258 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
2259 #define isXDIGIT_utf8_safe(p, e) \
2260 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2261 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2262 ? (_force_out_malformed_utf8_message( \
2263 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2264 : is_XDIGIT_high(p)))
2266 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2267 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2268 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2269 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2271 /* For internal core use only, subject to change */
2272 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2273 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2274 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2275 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
2277 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2278 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2279 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2280 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2282 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2283 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2284 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2285 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2286 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2287 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2288 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2289 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2290 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2291 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2292 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2293 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2294 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2295 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2296 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2297 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2298 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
2300 /* For internal core Perl use only: the base macros for defining macros like
2301 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2302 * point in 'p' is within the 0-255 range, it uses locale rules from the
2303 * passed-in 'macro' parameter */
2304 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2305 (__ASSERT_(_utf8_safe_assert(p, e)) \
2306 (UTF8_IS_INVARIANT(*(p))) \
2308 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2309 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2310 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2311 : (_force_out_malformed_utf8_message( \
2312 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2315 #define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2316 _generic_LC_utf8_safe(macro, p, e, \
2317 _is_utf8_FOO(classnum, p, e))
2319 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2320 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2322 #define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2323 _generic_LC_utf8_safe(classnum, p, e, \
2324 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2325 ? (_force_out_malformed_utf8_message( \
2326 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2329 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2330 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
2331 _CC_ALPHANUMERIC, p, e)
2332 #define isALPHA_LC_utf8_safe(p, e) \
2333 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2334 #define isASCII_LC_utf8_safe(p, e) \
2335 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2336 #define isBLANK_LC_utf8_safe(p, e) \
2337 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2338 #define isCNTRL_LC_utf8_safe(p, e) \
2339 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2340 #define isDIGIT_LC_utf8_safe(p, e) \
2341 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2342 #define isGRAPH_LC_utf8_safe(p, e) \
2343 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2344 #define isIDCONT_LC_utf8_safe(p, e) \
2345 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2346 _is_utf8_perl_idcont, p, e)
2347 #define isIDFIRST_LC_utf8_safe(p, e) \
2348 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2349 _is_utf8_perl_idstart, p, e)
2350 #define isLOWER_LC_utf8_safe(p, e) \
2351 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2352 #define isPRINT_LC_utf8_safe(p, e) \
2353 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2354 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2355 #define isPUNCT_LC_utf8_safe(p, e) \
2356 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2357 #define isSPACE_LC_utf8_safe(p, e) \
2358 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2359 #define isUPPER_LC_utf8_safe(p, e) \
2360 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2361 #define isWORDCHAR_LC_utf8_safe(p, e) \
2362 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2363 #define isXDIGIT_LC_utf8_safe(p, e) \
2364 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2366 /* Macros for backwards compatibility and for completeness when the ASCII and
2367 * Latin1 values are identical */
2368 #define isALPHAU(c) isALPHA_L1(c)
2369 #define isDIGIT_L1(c) isDIGIT_A(c)
2370 #define isOCTAL(c) isOCTAL_A(c)
2371 #define isOCTAL_L1(c) isOCTAL_A(c)
2372 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2373 #define isALNUM(c) isWORDCHAR(c)
2374 #define isALNUM_A(c) isALNUM(c)
2375 #define isALNUMU(c) isWORDCHAR_L1(c)
2376 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2377 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2378 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2379 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
2380 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
2381 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
2382 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
2383 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2384 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2385 #define isALNUMC(c) isALPHANUMERIC(c)
2386 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2387 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2388 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2389 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
2390 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2391 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
2393 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2394 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2395 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2396 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2397 * the outlier from the block that contains the other controls, just like
2398 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2399 * block. If it weren't special cased, it would yield a non-control.
2400 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2403 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2405 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2407 ? (UNLIKELY((c) == '?') \
2408 ? QUESTION_MARK_CTRL \
2409 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2410 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2412 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2415 /* Line numbers are unsigned, 32 bits. */
2417 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2419 /* Helpful alias for version prescan */
2420 #define is_LAX_VERSION(a,b) \
2421 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2423 #define is_STRICT_VERSION(a,b) \
2424 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2426 #define BADVERSION(a,b,c) \
2432 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2433 * a-f) to its numeric value without using any branches. The input is
2434 * validated only by an assert() in DEBUGGING builds.
2436 * It works by right shifting and isolating the bit that is 0 for the digits,
2437 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2438 * position, and then to the eights position. Both are added together to form
2439 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2440 * final four bits of the input to form the correct value. */
2441 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2442 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2443 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2444 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2446 /* The argument is a string pointer, which is advanced. */
2447 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
2449 /* Converts a character known to represent an octal digit (0-7) to its numeric
2450 * value. The input is validated only by an assert() in DEBUGGING builds. In
2451 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2452 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2454 /* Efficiently returns a boolean as to if two native characters are equivalent
2455 * case-insensitively. At least one of the characters must be one of [A-Za-z];
2456 * the ALPHA in the name is to remind you of that. This is asserted() in
2457 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2458 * works (on valid input) for both non- and UTF-8-encoded bytes.
2460 * When one of the inputs is a compile-time constant and gets folded by the
2461 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2462 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2463 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2464 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2465 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2466 * just a single 0, in the bit position where the upper- and lowercase differ.
2468 #define isALPHA_FOLD_EQ(c1, c2) \
2469 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2470 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2471 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2474 =for apidoc_section $memory
2476 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2477 The XSUB-writer's interface to the C C<malloc> function.
2479 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2481 In 5.9.3, Newx() and friends replace the older New() API, and drops
2482 the first parameter, I<x>, a debug aid which allowed callers to identify
2483 themselves. This aid has been superseded by a new build option,
2484 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2485 there for use in XS modules supporting older perls.
2487 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2488 The XSUB-writer's interface to the C C<malloc> function, with
2489 cast. See also C<L</Newx>>.
2491 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2493 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2494 The XSUB-writer's interface to the C C<malloc> function. The allocated
2495 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2497 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2499 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2500 The XSUB-writer's interface to the C C<realloc> function.
2502 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2504 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2505 The XSUB-writer's interface to the C C<realloc> function, with
2508 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2510 =for apidoc Am|void|Safefree|void* ptr
2511 The XSUB-writer's interface to the C C<free> function.
2513 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2515 =for apidoc_section $string
2516 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2517 The XSUB-writer's interface to the C C<memmove> 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. Can do overlapping moves. See also C<L</Copy>>.
2521 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2522 Like C<Move> but returns C<dest>. Useful
2523 for encouraging compilers to tail-call
2526 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2527 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2528 source, C<dest> is the destination, C<nitems> is the number of items, and
2529 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2531 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2533 Like C<Copy> but returns C<dest>. Useful
2534 for encouraging compilers to tail-call
2537 =for apidoc Am|void|Zero|void* dest|int nitems|type
2539 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2540 destination, C<nitems> is the number of items, and C<type> is the type.
2542 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2544 Like C<Zero> but returns dest. Useful
2545 for encouraging compilers to tail-call
2548 =for apidoc_section $utility
2549 =for apidoc Amu|void|StructCopy|type *src|type *dest|type
2550 This is an architecture-independent macro to copy one structure to another.
2552 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2554 Fill up memory with a byte pattern (a byte repeated over and over
2555 again) that hopefully catches attempts to access uninitialized memory.
2557 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2559 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2561 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2563 PoisonWith(0xEF) for catching access to freed memory.
2565 =for apidoc Am|void|Poison|void* dest|int nitems|type
2567 PoisonWith(0xEF) for catching access to freed memory.
2571 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2573 #define NEWSV(x,len) newSV(len)
2576 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2578 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2580 #ifdef PERL_MALLOC_WRAP
2582 /* This expression will be constant-folded at compile time. It checks
2583 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2584 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2585 * As well as avoiding the need for a run-time check in some cases, it's
2586 * designed to avoid compiler warnings like:
2587 * comparison is always false due to limited range of data type
2588 * It's mathematically equivalent to
2589 * max(n) * sizeof(t) > MEM_SIZE_MAX
2592 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2593 ( sizeof(MEM_SIZE) < sizeof(n) \
2594 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2596 /* This is written in a slightly odd way to avoid various spurious
2597 * compiler warnings. We *want* to write the expression as
2598 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2599 * (for some compile-time constant C), but even when the LHS
2600 * constant-folds to false at compile-time, g++ insists on emitting
2601 * warnings about the RHS (e.g. "comparison is always false"), so instead
2604 * (cond ? n : X) > C
2606 * where X is a constant with X > C always false. Choosing a value for X
2607 * is tricky. If 0, some compilers will complain about 0 > C always being
2608 * false; if 1, Coverity complains when n happens to be the constant value
2609 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2610 * for X and hope that nothing else whines.
2613 # define _MEM_WRAP_WILL_WRAP(n,t) \
2614 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2615 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2617 # define MEM_WRAP_CHECK(n,t) \
2618 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2619 && (croak_memory_wrap(),0))
2621 # define MEM_WRAP_CHECK_1(n,t,a) \
2622 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2623 && (Perl_croak_nocontext("%s",(a)),0))
2625 /* "a" arg must be a string literal */
2626 # define MEM_WRAP_CHECK_s(n,t,a) \
2627 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2628 && (Perl_croak_nocontext("" a ""),0))
2630 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2632 #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))
2635 #define MEM_WRAP_CHECK(n,t)
2636 #define MEM_WRAP_CHECK_1(n,t,a)
2637 #define MEM_WRAP_CHECK_s(n,t,a)
2638 #define MEM_WRAP_CHECK_(n,t)
2640 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2646 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2647 * go through functions, which are handy for debugging breakpoints, but
2648 * which more importantly get the immediate calling environment (file and
2649 * line number, and C function name if available) passed in. This info can
2650 * then be used for logging the calls, for which one gets a sample
2651 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2654 * - not all memory allocs get logged, only those
2655 * that go through Newx() and derivatives (while all
2656 * Safefrees do get logged)
2657 * - __FILE__ and __LINE__ do not work everywhere
2658 * - __func__ or __FUNCTION__ even less so
2659 * - I think more goes on after the perlio frees but
2660 * the thing is that STDERR gets closed (as do all
2661 * the file descriptors)
2662 * - no deeper calling stack than the caller of the Newx()
2663 * or the kind, but do I look like a C reflection/introspection
2665 * - the function prototypes for the logging functions
2666 * probably should maybe be somewhere else than handy.h
2667 * - one could consider inlining (macrofying) the logging
2668 * for speed, but I am too lazy
2669 * - one could imagine recording the allocations in a hash,
2670 * (keyed by the allocation address?), and maintain that
2671 * through reallocs and frees, but how to do that without
2672 * any News() happening...?
2673 * - lots of -Ddefines to get useful/controllable output
2674 * - lots of ENV reads
2678 # ifndef PERL_MEM_LOG_NOIMPL
2687 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2688 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2689 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2696 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2697 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2698 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2701 #ifndef MEM_LOG_ALLOC
2702 #define MEM_LOG_ALLOC(n,t,a) (a)
2704 #ifndef MEM_LOG_REALLOC
2705 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2707 #ifndef MEM_LOG_FREE
2708 #define MEM_LOG_FREE(a) (a)
2711 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2712 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2713 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2716 /* pre 5.9.x compatibility */
2717 #define New(x,v,n,t) Newx(v,n,t)
2718 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2719 #define Newz(x,v,n,t) Newxz(v,n,t)
2722 #define Renew(v,n,t) \
2723 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2724 #define Renewc(v,n,t,c) \
2725 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2728 #define Safefree(d) \
2729 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2731 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2734 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2735 * as it handles cases like constant string arguments without throwing warnings *
2736 * the cast is required, as is the inequality check, to avoid warnings */
2737 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2740 #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)))
2741 #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)))
2742 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2744 /* Like above, but returns a pointer to 'd' */
2745 #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)))
2746 #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)))
2747 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2749 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2750 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2751 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2752 #define Poison(d,n,t) PoisonFree(d,n,t)
2755 # define PERL_POISON_EXPR(x) x
2757 # define PERL_POISON_EXPR(x)
2760 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2763 =for apidoc_section $utility
2765 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2767 Returns the number of elements in the input C array (so you want your
2768 zero-based indices to be less than but not equal to).
2770 =for apidoc Am|void *|C_ARRAY_END|void *a
2772 Returns a pointer to one element past the final element of the input C array.
2776 C_ARRAY_END is one past the last: half-open/half-closed range, not
2777 last-inclusive range.
2779 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2780 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2784 # define Perl_va_copy(s, d) va_copy(d, s)
2785 # elif defined(__va_copy)
2786 # define Perl_va_copy(s, d) __va_copy(d, s)
2788 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2792 /* convenience debug macros */
2794 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2795 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2796 #define pTHX_VALUE_ (void *)my_perl,
2797 #define pTHX_VALUE (void *)my_perl
2798 #define pTHX__VALUE_ ,(void *)my_perl,
2799 #define pTHX__VALUE ,(void *)my_perl
2802 #define pTHX__FORMAT
2805 #define pTHX__VALUE_
2807 #endif /* USE_ITHREADS */
2809 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2810 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2811 CPAN::Unpack show any users outside the core. */
2813 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2814 "Use of " s " is deprecated")
2815 # define deprecate_disappears_in(when,message) \
2816 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2817 message ", and will disappear in Perl " when)
2818 # define deprecate_fatal_in(when,message) \
2819 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2820 message ". Its use will be fatal in Perl " when)
2823 /* Internal macros to deal with gids and uids */
2826 # if Uid_t_size > IVSIZE
2827 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2828 # define SvUID(sv) SvNV(sv)
2829 # elif Uid_t_sign <= 0
2830 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2831 # define SvUID(sv) SvIV(sv)
2833 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2834 # define SvUID(sv) SvUV(sv)
2835 # endif /* Uid_t_size */
2837 # if Gid_t_size > IVSIZE
2838 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2839 # define SvGID(sv) SvNV(sv)
2840 # elif Gid_t_sign <= 0
2841 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2842 # define SvGID(sv) SvIV(sv)
2844 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2845 # define SvGID(sv) SvUV(sv)
2846 # endif /* Gid_t_size */
2850 #endif /* PERL_HANDY_H_ */
2853 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob