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 Handling
22 =for apidoc AmnU||Nullch
23 Null character pointer. (No longer available when C<PERL_CORE> is
26 =for apidoc_section SV Handling
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 configuration values
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 Handling
46 =for apidoc Amu|token|CAT2|token x|token y
47 =for apidoc Amu|string|STRINGIFY|token x
49 =for apidoc_section Numeric Functions
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 Handling
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 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
222 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
223 Please search CHAR_MAX in perl.h for further details. */
224 #define U8_MAX UINT8_MAX
225 #define U8_MIN UINT8_MIN
227 #define I16_MAX INT16_MAX
228 #define I16_MIN INT16_MIN
229 #define U16_MAX UINT16_MAX
230 #define U16_MIN UINT16_MIN
232 #define I32_MAX INT32_MAX
233 #define I32_MIN INT32_MIN
234 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
235 # define U32_MAX UINT32_MAX
237 # define U32_MAX 4294967295U
239 #define U32_MIN UINT32_MIN
243 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
244 Please search CHAR_MAX in perl.h for further details. */
245 #define U8_MAX PERL_UCHAR_MAX
246 #define U8_MIN PERL_UCHAR_MIN
248 #define I16_MAX PERL_SHORT_MAX
249 #define I16_MIN PERL_SHORT_MIN
250 #define U16_MAX PERL_USHORT_MAX
251 #define U16_MIN PERL_USHORT_MIN
254 # define I32_MAX PERL_INT_MAX
255 # define I32_MIN PERL_INT_MIN
256 # define U32_MAX PERL_UINT_MAX
257 # define U32_MIN PERL_UINT_MIN
259 # define I32_MAX PERL_LONG_MAX
260 # define I32_MIN PERL_LONG_MIN
261 # define U32_MAX PERL_ULONG_MAX
262 # define U32_MIN PERL_ULONG_MIN
267 /* These C99 typedefs are useful sometimes for, say, loop variables whose
268 * maximum values are small, but for which speed trumps size. If we have a C99
269 * compiler, use that. Otherwise, a plain 'int' should be good enough.
271 * Restrict these to core for now until we are more certain this is a good
273 #if defined(PERL_CORE) || defined(PERL_EXT)
275 typedef int_fast8_t PERL_INT_FAST8_T;
276 typedef uint_fast8_t PERL_UINT_FAST8_T;
277 typedef int_fast16_t PERL_INT_FAST16_T;
278 typedef uint_fast16_t PERL_UINT_FAST16_T;
280 typedef int PERL_INT_FAST8_T;
281 typedef unsigned int PERL_UINT_FAST8_T;
282 typedef int PERL_INT_FAST16_T;
283 typedef unsigned int PERL_UINT_FAST16_T;
287 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
288 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
289 * required to represent any possible unsigned number containing N bits.
290 * TYPE_DIGITS gives the number of decimal digits required to represent any
291 * possible unsigned number of type T. */
292 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
293 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
294 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
296 /* Unused by core; should be deprecated */
297 #define Ctl(ch) ((ch) & 037)
299 #if defined(PERL_CORE) || defined(PERL_EXT)
301 # define MIN(a,b) ((a) < (b) ? (a) : (b))
304 # define MAX(a,b) ((a) > (b) ? (a) : (b))
308 /* Returns a boolean as to whether the input unsigned number is a power of 2
309 * (2**0, 2**1, etc). In other words if it has just a single bit set.
310 * If not, subtracting 1 would leave the uppermost bit set, so the & would
312 #if defined(PERL_CORE) || defined(PERL_EXT)
313 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
316 /* Returns a mask with the lowest n bits set */
317 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
319 /* The largest unsigned number that will fit into n bits */
320 #define nBIT_UMAX(n) nBIT_MASK(n)
323 =for apidoc_section Compiler directives
324 =for apidoc Am||__ASSERT_|bool expr
326 This is a helper macro to avoid preprocessor issues, replaced by nothing
327 unless under DEBUGGING, where it expands to an assert of its argument,
328 followed by a comma (hence the comma operator). If we just used a straight
329 assert(), we would get a comma with nothing before it when not DEBUGGING.
333 We also use empty definition under Coverity since the __ASSERT__
334 checks often check for things that Really Cannot Happen, and Coverity
335 detects that and gets all excited. */
337 #if defined(DEBUGGING) && !defined(__COVERITY__) \
338 && ! defined(PERL_SMALL_MACRO_BUFFER)
339 # define __ASSERT_(statement) assert(statement),
341 # define __ASSERT_(statement)
345 =for apidoc_section SV Handling
347 =for apidoc Ama|SV*|newSVpvs|"literal string"
348 Like C<newSVpvn>, but takes a literal string instead of a
351 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
352 Like C<newSVpvn_flags>, but takes a literal string instead of
353 a string/length pair.
355 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
356 Like C<newSVpvn_share>, but takes a literal string instead of
357 a string/length pair and omits the hash parameter.
359 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
360 Like C<sv_catpvn_flags>, but takes a literal string instead
361 of a string/length pair.
363 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
364 Like C<sv_catpvn_nomg>, but takes a literal string instead of
365 a string/length pair.
367 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
368 Like C<sv_catpvn>, but takes a literal string instead of a
371 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
372 Like C<sv_catpvn_mg>, but takes a literal string instead of a
375 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
376 Like C<sv_setpvn>, but takes a literal string instead of a
379 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
380 Like C<sv_setpvn_mg>, but takes a literal string instead of a
383 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
384 Like C<sv_setref_pvn>, but takes a literal string instead of
385 a string/length pair.
387 =for apidoc_section String Handling
389 =for apidoc Ama|char*|savepvs|"literal string"
390 Like C<savepvn>, but takes a literal string instead of a
393 =for apidoc Ama|char*|savesharedpvs|"literal string"
394 A version of C<savepvs()> which allocates the duplicate string in memory
395 which is shared between threads.
397 =for apidoc_section GV Handling
399 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
400 Like C<gv_stashpvn>, but takes a literal string instead of a
403 =for apidoc_section HV Handling
405 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
406 Like C<hv_fetch>, but takes a literal string instead of a
409 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
410 Like C<hv_store>, but takes a literal string instead of a
412 and omits the hash parameter.
414 =for apidoc_section Lexer interface
416 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
418 Like L</lex_stuff_pvn>, but takes a literal string instead of
419 a string/length pair.
425 =for apidoc_section String Handling
427 =for apidoc Amu|pair|STR_WITH_LEN|"literal string"
429 Returns two comma separated tokens of the input literal string, and its length.
430 This is convenience macro which helps out in some API calls.
431 Note that it can't be used as an argument to macros or functions that under
432 some configurations might be macros, which means that it requires the full
433 Perl_xxx(aTHX_ ...) form for any API calls where it's used.
438 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
440 /* STR_WITH_LEN() shortcuts */
441 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
442 #define newSVpvs_flags(str,flags) \
443 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
444 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
445 #define sv_catpvs_flags(sv, str, flags) \
446 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
447 #define sv_catpvs_nomg(sv, str) \
448 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
449 #define sv_catpvs(sv, str) \
450 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
451 #define sv_catpvs_mg(sv, str) \
452 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
453 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
454 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
455 #define sv_setref_pvs(rv, classname, str) \
456 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
457 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
458 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
459 #define gv_stashpvs(str, create) \
460 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
462 #define gv_fetchpvs(namebeg, flags, sv_type) \
463 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), flags, sv_type)
464 #define gv_fetchpvn gv_fetchpvn_flags
465 #define sv_catxmlpvs(dsv, str, utf8) \
466 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
469 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
471 #define get_cvs(str, flags) \
472 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
474 /* internal helpers */
476 #ifndef PERL_VERSION_MAJOR
477 # define PERL_VERSION_MAJOR PERL_REVISION
479 # undef PERL_REVISION /* We don't want code to be using these */
481 #ifndef PERL_VERSION_MINOR
482 # define PERL_VERSION_MINOR PERL_VERSION
486 #ifndef PERL_VERSION_PATCH
487 # define PERL_VERSION_PATCH PERL_SUBVERSION
489 # undef PERL_SUBVERSION
492 #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
493 /* '10*' leaves room for things like alpha, beta, releases */ \
494 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
495 #define PERL_DECIMAL_VERSION_ \
496 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
500 =for apidoc_section Versioning
501 =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
502 =for apidoc_item PERL_VERSION_NE
503 =for apidoc_item PERL_VERSION_LT
504 =for apidoc_item PERL_VERSION_LE
505 =for apidoc_item PERL_VERSION_GT
506 =for apidoc_item PERL_VERSION_GE
508 Returns whether or not the perl currently being compiled has the specified
509 relationship to the perl given by the parameters. For example,
511 #if PERL_VERSION_GT(5,24,2)
512 code that will only be compiled on perls after v5.24.2
517 Note that this is usable in making compile-time decisions
519 You may use the special value '*' for the final number to mean ALL possible
522 #if PERL_VERSION_EQ(5,31,'*')
524 means all perls in the 5.31 series. And
526 #if PERL_VERSION_NE(5,24,'*')
528 means all perls EXCEPT 5.24 ones. And
530 #if PERL_VERSION_LE(5,9,'*')
534 #if PERL_VERSION_LT(5,10,0)
536 This means you don't have to think so much when converting from the existing
537 deprecated C<PERL_VERSION> to using this macro:
539 #if PERL_VERSION <= 9
543 #if PERL_VERSION_LE(5,9,'*')
548 /* N.B. These don't work if the patch version is 42 or 92, as those are what
549 * '*' is in ASCII and EBCDIC respectively */
550 # define PERL_VERSION_EQ(j,n,p) \
552 ? ( (j) == PERL_VERSION_MAJOR \
553 && (n) == PERL_VERSION_MINOR) \
554 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
555 # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
557 # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
558 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
560 (((p) == '*') ? 0 : p)))
561 # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
563 # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
564 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
565 (((p) == '*') ? ((n)+1) : (n)), \
566 (((p) == '*') ? 0 : p)))
567 # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
570 =for apidoc_section String Handling
572 =for apidoc Am|bool|strNE|char* s1|char* s2
573 Test two C<NUL>-terminated strings to see if they are different. Returns true
576 =for apidoc Am|bool|strEQ|char* s1|char* s2
577 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
580 =for apidoc Am|bool|strLT|char* s1|char* s2
581 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
582 second, C<s2>. Returns true or false.
584 =for apidoc Am|bool|strLE|char* s1|char* s2
585 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
586 equal to the second, C<s2>. Returns true or false.
588 =for apidoc Am|bool|strGT|char* s1|char* s2
589 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
590 the second, C<s2>. Returns true or false.
592 =for apidoc Am|bool|strGE|char* s1|char* s2
593 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
594 or equal to the second, C<s2>. Returns true or false.
596 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
597 Test two C<NUL>-terminated strings to see if they are different. The C<len>
598 parameter indicates the number of bytes to compare. Returns true or false. (A
599 wrapper for C<strncmp>).
601 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
602 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
603 parameter indicates the number of bytes to compare. Returns true or false. (A
604 wrapper for C<strncmp>).
606 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
607 Test two buffers (which may contain embedded C<NUL> characters, to see if they
608 are equal. The C<len> parameter indicates the number of bytes to compare.
609 Returns zero if equal, or non-zero if non-equal.
611 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
612 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
613 C<l1> gives the number of bytes in C<s1>.
614 Returns zero if equal, or non-zero if non-equal.
616 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
617 Test two buffers (which may contain embedded C<NUL> characters, to see if they
618 are not equal. The C<len> parameter indicates the number of bytes to compare.
619 Returns zero if non-equal, or non-zero if equal.
621 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
622 Like L</memNE>, but the second string is a literal enclosed in double quotes,
623 C<l1> gives the number of bytes in C<s1>.
624 Returns zero if non-equal, or zero if non-equal.
626 =for apidoc Am|bool|memCHRs|"list"|char c
627 Returns the position of the first occurence of the byte C<c> in the literal
628 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
629 treated as unsigned char. Thus this macro can be used to determine if C<c> is
630 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
631 is C<NUL> (and the set doesn't include C<NUL>).
635 New macros should use the following conventions for their names (which are
636 based on the underlying C library functions):
638 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
640 Each has two main parameters, string-like operands that are compared
641 against each other, as specified by the macro name. Some macros may
642 additionally have one or potentially even two length parameters. If a length
643 parameter applies to both string parameters, it will be positioned third;
644 otherwise any length parameter immediately follows the string parameter it
647 If the prefix to the name is 'str', the string parameter is a pointer to a C
648 language string. Such a string does not contain embedded NUL bytes; its
649 length may be unknown, but can be calculated by C<strlen()>, since it is
650 terminated by a NUL, which isn't included in its length.
652 The optional 'n' following 'str' means that there is a third parameter,
653 giving the maximum number of bytes to look at in each string. Even if both
654 strings are longer than the length parameter, those extra bytes will be
657 The 's' suffix means that the 2nd byte string parameter is a literal C
658 double-quoted string. Its length will automatically be calculated by the
659 macro, so no length parameter will ever be needed for it.
661 If the prefix is 'mem', the string parameters don't have to be C strings;
662 they may contain embedded NUL bytes, do not necessarily have a terminating
663 NUL, and their lengths can be known only through other means, which in
664 practice are additional parameter(s) passed to the function. All 'mem'
665 functions have at least one length parameter. Barring any 'l' or 's' suffix,
666 there is a single length parameter, in position 3, which applies to both
667 string parameters. The 's' suffix means, as described above, that the 2nd
668 string is a literal double-quoted C string (hence its length is calculated by
669 the macro, and the length parameter to the function applies just to the first
670 string parameter, and hence is positioned just after it). An 'l' suffix
671 means that the 2nd string parameter has its own length parameter, and the
672 signature will look like memFOOl(s1, l1, s2, l2).
674 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
675 substring of the 1st string. 'P' if present indicates that the substring
676 must be a "proper" one in tha mathematical sense that the first one must be
677 strictly larger than the 2nd.
682 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
683 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
684 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
685 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
686 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
687 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
689 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
690 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
692 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
693 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
695 /* memEQ and memNE where second comparand is a string constant */
696 #define memEQs(s1, l, s2) \
697 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
698 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
700 /* Keep these private until we decide it was a good idea */
701 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
703 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
705 #define memBEGINs(s1, l, s2) \
706 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
707 && memEQ(s1, "" s2 "", sizeof(s2)-1))
708 #define memBEGINPs(s1, l, s2) \
709 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
710 && memEQ(s1, "" s2 "", sizeof(s2)-1))
711 #define memENDs(s1, l, s2) \
712 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
713 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
714 #define memENDPs(s1, l, s2) \
715 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
716 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
717 #endif /* End of making macros private */
719 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
720 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
721 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
722 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
724 #define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
729 * Unfortunately, the introduction of locales means that we
730 * can't trust isupper(), etc. to tell the truth. And when
731 * it comes to /\w+/ with tainting enabled, we *must* be able
732 * to trust our character classes.
734 * Therefore, the default tests in the text of Perl will be
735 * independent of locale. Any code that wants to depend on
736 * the current locale will use the tests that begin with "lc".
739 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
747 =head1 Character classification
748 This section is about functions (really macros) that classify characters
749 into types, such as punctuation versus alphabetic, etc. Most of these are
750 analogous to regular expression character classes. (See
751 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
752 each class. (Not all macros have all variants; each item below lists the
753 ones valid for it.) None are affected by C<use bytes>, and only the ones
754 with C<LC> in the name are affected by the current locale.
756 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
757 treating it as a code point, and returns a boolean as to whether or not the
758 character represented by it is (or on non-ASCII platforms, corresponds to) an
759 ASCII character in the named class based on platform, Unicode, and Perl rules.
760 If the input is a number that doesn't fit in an octet, FALSE is returned.
762 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
763 with no suffix C<"_A">. This variant is used to emphasize by its name that
764 only ASCII-range characters can return TRUE.
766 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
767 onto the platform. That is, the code points that are ASCII are unaffected,
768 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
769 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
770 true when called with the code point 0xDF, which is a word character in both
771 ASCII and EBCDIC (though it represents different characters in each).
772 If the input is a number that doesn't fit in an octet, FALSE is returned.
773 (Perl's documentation uses a colloquial definition of Latin-1, to include all
774 code points below 256.)
776 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
777 inputs below 256, but if the code point is larger than 255, Unicode rules are
778 used to determine if it is in the character class. For example,
779 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
780 WITH MACRON in Unicode, and is a word character.
782 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
783 but are used for UTF-8 encoded strings. The two forms are different names for
784 the same thing. Each call to one of these classifies the first character of
785 the string starting at C<p>. The second parameter, C<e>, points to anywhere in
786 the string beyond the first character, up to one byte past the end of the
787 entire string. Although both variants are identical, the suffix C<_safe> in
788 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
789 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
790 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
791 some way, the program may croak, or the function may return FALSE, at the
792 discretion of the implementation, and subject to change in future releases.
794 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
795 but the result is based on the current locale, which is what C<LC> in the name
796 stands for. If Perl can determine that the current locale is a UTF-8 locale,
797 it uses the published Unicode rules; otherwise, it uses the C library function
798 that gives the named classification. For example, C<isDIGIT_LC()> when not in
799 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
800 returned if the input won't fit into an octet. On some platforms where the C
801 library function is known to be defective, Perl changes its result to follow
802 the POSIX standard's rules.
804 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
805 than 256, but for larger ones it returns the Unicode classification of the code
808 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
809 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
810 are different names for the same thing. Each call to one of these classifies
811 the first character of the string starting at C<p>. The second parameter,
812 C<e>, points to anywhere in the string beyond the first character, up to one
813 byte past the end of the entire string. Although both variants are identical,
814 the suffix C<_safe> in one name emphasizes that it will not attempt to read
815 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
816 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
817 character is malformed in some way, the program may croak, or the function may
818 return FALSE, at the discretion of the implementation, and subject to change in
821 =for apidoc Am|bool|isALPHA|UV ch
822 =for apidoc_item ||isALPHA_A|UV ch
823 =for apidoc_item ||isALPHA_L1|UV ch
824 =for apidoc_item ||isALPHA_uvchr|UV ch
825 =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
826 =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
827 =for apidoc_item ||isALPHA_LC|UV ch
828 =for apidoc_item ||isALPHA_LC_uvchr|UV ch
829 =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
830 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
831 analogous to C<m/[[:alpha:]]/>.
832 See the L<top of this section|/Character classification> for an explanation of
837 Here and below, we add the protoypes of these macros for downstream programs
838 that would be interested in them, such as Devel::PPPort
840 =for apidoc Am|bool|isALPHANUMERIC|UV ch
841 =for apidoc_item ||isALPHANUMERIC_A|UV ch
842 =for apidoc_item ||isALPHANUMERIC_L1|UV ch
843 =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
844 =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
845 =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
846 =for apidoc_item ||isALPHANUMERIC_LC|UV ch
847 =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
848 =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
849 =for apidoc_item ||isALNUMC|UV ch
850 =for apidoc_item ||isALNUMC_A|UV ch
851 =for apidoc_item ||isALNUMC_L1|UV ch
852 =for apidoc_item ||isALNUMC_LC|UV ch
853 =for apidoc_item ||isALNUMC_LC_uvchr|UV ch
854 Returns a boolean indicating whether the specified character is one of
855 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
856 See the L<top of this section|/Character classification> for an explanation of
859 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
860 this corresponds to the C language alphanumeric definition). Also
861 there are the variants
862 C<isALNUMC_A>, C<isALNUMC_L1>
863 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
865 =for apidoc Am|bool|isASCII|UV ch
866 =for apidoc_item ||isASCII_A|UV ch
867 =for apidoc_item ||isASCII_L1|UV ch
868 =for apidoc_item ||isASCII_uvchr|UV ch
869 =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
870 =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
871 =for apidoc_item ||isASCII_LC|UV ch
872 =for apidoc_item ||isASCII_LC_uvchr|UV ch
873 =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
874 Returns a boolean indicating whether the specified character is one of the 128
875 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
876 On non-ASCII platforms, it returns TRUE iff this
877 character corresponds to an ASCII character. Variants C<isASCII_A()> and
878 C<isASCII_L1()> are identical to C<isASCII()>.
879 See the L<top of this section|/Character classification> for an explanation of
881 Note, however, that some platforms do not have the C library routine
882 C<isascii()>. In these cases, the variants whose names contain C<LC> are the
883 same as the corresponding ones without.
885 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
886 have the exact same representation (always a single byte) whether encoded in
887 UTF-8 or not), C<isASCII> will give the correct results when called with any
888 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
889 C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
891 =for apidoc Am|bool|isBLANK|UV ch
892 =for apidoc_item ||isBLANK_A|UV ch
893 =for apidoc_item ||isBLANK_L1|UV ch
894 =for apidoc_item ||isBLANK_uvchr|UV ch
895 =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
896 =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
897 =for apidoc_item ||isBLANK_LC|UV ch
898 =for apidoc_item ||isBLANK_LC_uvchr|UV ch
899 =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
900 Returns a boolean indicating whether the specified character is a
901 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
902 See the L<top of this section|/Character classification> for an explanation of
905 however, that some platforms do not have the C library routine
906 C<isblank()>. In these cases, the variants whose names contain C<LC> are
907 the same as the corresponding ones without.
909 =for apidoc Am|bool|isCNTRL|UV ch
910 =for apidoc_item ||isCNTRL_A|UV ch
911 =for apidoc_item ||isCNTRL_L1|UV ch
912 =for apidoc_item ||isCNTRL_uvchr|UV ch
913 =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
914 =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
915 =for apidoc_item ||isCNTRL_LC|UV ch
916 =for apidoc_item ||isCNTRL_LC_uvchr|UV ch
917 =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
919 Returns a boolean indicating whether the specified character is a
920 control character, analogous to C<m/[[:cntrl:]]/>.
921 See the L<top of this section|/Character classification> for an explanation of
923 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
925 =for apidoc Am|bool|isDIGIT|UV ch
926 =for apidoc_item ||isDIGIT_A|UV ch
927 =for apidoc_item ||isDIGIT_L1|UV ch
928 =for apidoc_item ||isDIGIT_uvchr|UV ch
929 =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
930 =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
931 =for apidoc_item ||isDIGIT_LC|UV ch
932 =for apidoc_item ||isDIGIT_LC_uvchr|UV ch
933 =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
935 Returns a boolean indicating whether the specified character is a
936 digit, analogous to C<m/[[:digit:]]/>.
937 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
938 See the L<top of this section|/Character classification> for an explanation of
941 =for apidoc Am|bool|isGRAPH|UV ch
942 =for apidoc_item ||isGRAPH_A|UV ch
943 =for apidoc_item ||isGRAPH_L1|UV ch
944 =for apidoc_item ||isGRAPH_uvchr|UV ch
945 =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
946 =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
947 =for apidoc_item ||isGRAPH_LC|UV ch
948 =for apidoc_item ||isGRAPH_LC_uvchr|UV ch
949 =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
950 Returns a boolean indicating whether the specified character is a
951 graphic character, analogous to C<m/[[:graph:]]/>.
952 See the L<top of this section|/Character classification> for an explanation of
955 =for apidoc Am|bool|isLOWER|UV ch
956 =for apidoc_item ||isLOWER_A|UV ch
957 =for apidoc_item ||isLOWER_L1|UV ch
958 =for apidoc_item ||isLOWER_uvchr|UV ch
959 =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
960 =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
961 =for apidoc_item ||isLOWER_LC|UV ch
962 =for apidoc_item ||isLOWER_LC_uvchr|UV ch
963 =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
964 Returns a boolean indicating whether the specified character is a
965 lowercase character, analogous to C<m/[[:lower:]]/>.
966 See the L<top of this section|/Character classification> for an explanation of
969 =for apidoc Am|bool|isOCTAL|UV ch
970 =for apidoc_item ||isOCTAL_A|UV ch
971 =for apidoc_item ||isOCTAL_L1|UV ch
972 Returns a boolean indicating whether the specified character is an
974 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
977 =for apidoc Am|bool|isPUNCT|UV ch
978 =for apidoc_item ||isPUNCT_A|UV ch
979 =for apidoc_item ||isPUNCT_L1|UV ch
980 =for apidoc_item ||isPUNCT_uvchr|UV ch
981 =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
982 =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
983 =for apidoc_item ||isPUNCT_LC|UV ch
984 =for apidoc_item ||isPUNCT_LC_uvchr|UV ch
985 =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
986 Returns a boolean indicating whether the specified character is a
987 punctuation character, analogous to C<m/[[:punct:]]/>.
988 Note that the definition of what is punctuation isn't as
989 straightforward as one might desire. See L<perlrecharclass/POSIX Character
990 Classes> for details.
991 See the L<top of this section|/Character classification> for an explanation of
994 =for apidoc Am|bool|isSPACE|UV ch
995 =for apidoc_item ||isSPACE_A|UV ch
996 =for apidoc_item ||isSPACE_L1|UV ch
997 =for apidoc_item ||isSPACE_uvchr|UV ch
998 =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
999 =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
1000 =for apidoc_item ||isSPACE_LC|UV ch
1001 =for apidoc_item ||isSPACE_LC_uvchr|UV ch
1002 =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
1003 Returns a boolean indicating whether the specified character is a
1004 whitespace character. This is analogous
1005 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
1006 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
1007 locale forms of this macro (the ones with C<LC> in their names) matched
1008 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1009 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1010 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
1011 See the L<top of this section|/Character classification> for an explanation of
1014 =for apidoc Am|bool|isPSXSPC|UV ch
1015 =for apidoc_item ||isPSXSPC_A|UV ch
1016 =for apidoc_item ||isPSXSPC_L1|UV ch
1017 =for apidoc_item ||isPSXSPC_uvchr|UV ch
1018 =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1019 =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
1020 =for apidoc_item ||isPSXSPC_LC|UV ch
1021 =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
1022 =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
1023 (short for Posix Space)
1024 Starting in 5.18, this is identical in all its forms to the
1025 corresponding C<isSPACE()> macros.
1026 The locale forms of this macro are identical to their corresponding
1027 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1028 non-locale forms differ from their C<isSPACE()> forms only in that the
1029 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1030 Otherwise they are identical. Thus this macro is analogous to what
1031 C<m/[[:space:]]/> matches in a regular expression.
1032 See the L<top of this section|/Character classification> for an explanation of
1035 =for apidoc Am|bool|isUPPER|UV ch
1036 =for apidoc_item ||isUPPER_A|UV ch
1037 =for apidoc_item ||isUPPER_L1|UV ch
1038 =for apidoc_item ||isUPPER_uvchr|UV ch
1039 =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1040 =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
1041 =for apidoc_item ||isUPPER_LC|UV ch
1042 =for apidoc_item ||isUPPER_LC_uvchr|UV ch
1043 =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
1044 Returns a boolean indicating whether the specified character is an
1045 uppercase character, analogous to C<m/[[:upper:]]/>.
1046 See the L<top of this section|/Character classification> for an explanation of
1049 =for apidoc Am|bool|isPRINT|UV ch
1050 =for apidoc_item ||isPRINT_A|UV ch
1051 =for apidoc_item ||isPRINT_L1|UV ch
1052 =for apidoc_item ||isPRINT_uvchr|UV ch
1053 =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1054 =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
1055 =for apidoc_item ||isPRINT_LC|UV ch
1056 =for apidoc_item ||isPRINT_LC_uvchr|UV ch
1057 =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
1058 Returns a boolean indicating whether the specified character is a
1059 printable character, analogous to C<m/[[:print:]]/>.
1060 See the L<top of this section|/Character classification> for an explanation of
1063 =for apidoc Am|bool|isWORDCHAR|UV ch
1064 =for apidoc_item ||isWORDCHAR_A|UV ch
1065 =for apidoc_item ||isWORDCHAR_L1|UV ch
1066 =for apidoc_item ||isWORDCHAR_uvchr|UV ch
1067 =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1068 =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
1069 =for apidoc_item ||isWORDCHAR_LC|UV ch
1070 =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
1071 =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1072 =for apidoc_item ||isALNUM|UV ch
1073 =for apidoc_item ||isALNUM_A|UV ch
1074 =for apidoc_item ||isALNUM_LC|UV ch
1075 =for apidoc_item ||isALNUM_LC_uvchr|UV ch
1076 Returns a boolean indicating whether the specified character is a character
1077 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1078 in a regular expression. A word character is an alphabetic character, a
1079 decimal digit, a connecting punctuation character (such as an underscore), or
1080 a "mark" character that attaches to one of those (like some sort of accent).
1081 C<isALNUM()> is a synonym provided for backward compatibility, even though a
1082 word character includes more than the standard C language meaning of
1084 See the L<top of this section|/Character classification> for an explanation of
1086 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1087 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1088 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1089 include the platform's native underscore.
1091 =for apidoc Am|bool|isXDIGIT|UV ch
1092 =for apidoc_item ||isXDIGIT_A|UV ch
1093 =for apidoc_item ||isXDIGIT_L1|UV ch
1094 =for apidoc_item ||isXDIGIT_uvchr|UV ch
1095 =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1096 =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
1097 =for apidoc_item ||isXDIGIT_LC|UV ch
1098 =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
1099 =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1100 Returns a boolean indicating whether the specified character is a hexadecimal
1101 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1102 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1103 See the L<top of this section|/Character classification> for an explanation of
1106 =for apidoc Am|bool|isIDFIRST|UV ch
1107 =for apidoc_item ||isIDFIRST_A|UV ch
1108 =for apidoc_item ||isIDFIRST_L1|UV ch
1109 =for apidoc_item ||isIDFIRST_uvchr|UV ch
1110 =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1111 =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
1112 =for apidoc_item ||isIDFIRST_LC|UV ch
1113 =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
1114 =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1115 Returns a boolean indicating whether the specified character can be the first
1116 character of an identifier. This is very close to, but not quite the same as
1117 the official Unicode property C<XID_Start>. The difference is that this
1118 returns true only if the input character also matches L</isWORDCHAR>.
1119 See the L<top of this section|/Character classification> for an explanation of
1122 =for apidoc Am|bool|isIDCONT|UV ch
1123 =for apidoc_item ||isIDCONT_A|UV ch
1124 =for apidoc_item ||isIDCONT_L1|UV ch
1125 =for apidoc_item ||isIDCONT_uvchr|UV ch
1126 =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1127 =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
1128 =for apidoc_item ||isIDCONT_LC|UV ch
1129 =for apidoc_item ||isIDCONT_LC_uvchr|UV ch
1130 =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1131 Returns a boolean indicating whether the specified character can be the
1132 second or succeeding character of an identifier. This is very close to, but
1133 not quite the same as the official Unicode property C<XID_Continue>. The
1134 difference is that this returns true only if the input character also matches
1135 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1136 an explanation of the variants.
1138 =for apidoc_section Numeric Functions
1140 =for apidoc Am|U8|READ_XDIGIT|char str*
1141 Returns the value of an ASCII-range hex digit and advances the string pointer.
1142 Behaviour is only well defined when isXDIGIT(*str) is true.
1144 =head1 Character case changing
1145 Perl uses "full" Unicode case mappings. This means that converting a single
1146 character to another case may result in a sequence of more than one character.
1147 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1148 character sequence C<SS>. This presents some complications The lowercase of
1149 all characters in the range 0..255 is a single character, and thus
1150 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1151 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1152 an API that does allow every possible legal result to be returned.) Likewise
1153 no other function that is crippled by not being able to give the correct
1154 results for the full range of possible inputs has been implemented here.
1156 =for apidoc Am|U8|toUPPER|int ch
1157 Converts the specified character to uppercase. If the input is anything but an
1158 ASCII lowercase character, that input character itself is returned. Variant
1159 C<toUPPER_A> is equivalent.
1161 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1162 Converts the code point C<cp> to its uppercase version, and
1163 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1164 point is interpreted as native if less than 256; otherwise as Unicode. Note
1165 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1166 bytes since the uppercase version may be longer than the original character.
1168 The first code point of the uppercased version is returned
1169 (but note, as explained at L<the top of this section|/Character case
1170 changing>, that there may be more.)
1172 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1173 =for apidoc_item toUPPER_utf8_safe
1174 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1175 extending no further than S<C<e - 1>> to its uppercase version, and
1176 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1177 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1178 bytes since the uppercase version may be longer than the original character.
1180 The first code point of the uppercased version is returned
1181 (but note, as explained at L<the top of this section|/Character case
1182 changing>, that there may be more).
1184 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1185 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1186 the UTF-8 for the input character is malformed in some way, the program may
1187 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1188 of the implementation, and subject to change in future releases.
1190 C<toUPPER_utf8_safe> is now just a different spelling of plain C<toUPPER_utf8>
1192 =for apidoc Am|U8|toFOLD|U8 ch
1193 Converts the specified character to foldcase. If the input is anything but an
1194 ASCII uppercase character, that input character itself is returned. Variant
1195 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
1196 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
1198 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1199 Converts the code point C<cp> to its foldcase version, and
1200 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1201 point is interpreted as native if less than 256; otherwise as Unicode. Note
1202 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1203 bytes since the foldcase version may be longer than the original character.
1205 The first code point of the foldcased 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 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1210 =for apidoc_item toFOLD_utf8_safe
1211 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1212 extending no further than S<C<e - 1>> to its foldcase version, and
1213 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1214 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1215 bytes since the foldcase version may be longer than the original character.
1217 The first code point of the foldcased version is returned
1218 (but note, as explained at L<the top of this section|/Character case
1219 changing>, that there may be more).
1222 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1223 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1224 input character is malformed in some way, the program may croak, or the
1225 function may return the REPLACEMENT CHARACTER, at the discretion of the
1226 implementation, and subject to change in future releases.
1228 C<toFOLD_utf8_safe> is now just a different spelling of plain C<toFOLD_utf8>
1230 =for apidoc Am|U8|toLOWER|U8 ch
1231 Converts the specified character to lowercase. If the input is anything but an
1232 ASCII uppercase character, that input character itself is returned. Variant
1233 C<toLOWER_A> is equivalent.
1235 =for apidoc Am|U8|toLOWER_L1|U8 ch
1236 Converts the specified Latin1 character to lowercase. The results are
1237 undefined if the input doesn't fit in a byte.
1239 =for apidoc Am|U8|toLOWER_LC|U8 ch
1240 Converts the specified character to lowercase using the current locale's rules,
1241 if possible; otherwise returns the input character itself.
1243 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1244 Converts the code point C<cp> to its lowercase version, and
1245 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1246 point is interpreted as native if less than 256; otherwise as Unicode. Note
1247 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1248 bytes since the lowercase version may be longer than the original character.
1250 The first code point of the lowercased version is returned
1251 (but note, as explained at L<the top of this section|/Character case
1252 changing>, that there may be more).
1254 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1255 =for apidoc_item toLOWER_utf8_safe
1256 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1257 extending no further than S<C<e - 1>> to its lowercase version, and
1258 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1259 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1260 bytes since the lowercase version may be longer than the original character.
1262 The first code point of the lowercased version is returned
1263 (but note, as explained at L<the top of this section|/Character case
1264 changing>, that there may be more).
1265 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1266 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1267 the UTF-8 for the input character is malformed in some way, the program may
1268 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1269 of the implementation, and subject to change in future releases.
1271 C<toLOWER_utf8_safe> is now just a different spelling of plain C<toLOWER_utf8>
1273 =for apidoc Am|U8|toTITLE|U8 ch
1274 Converts the specified character to titlecase. If the input is anything but an
1275 ASCII lowercase character, that input character itself is returned. Variant
1276 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1277 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1278 not a concept used in locale handling, so there is no functionality for that.)
1280 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1281 Converts the code point C<cp> to its titlecase version, and
1282 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1283 point is interpreted as native if less than 256; otherwise as Unicode. Note
1284 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1285 bytes since the titlecase version may be longer than the original character.
1287 The first code point of the titlecased version is returned
1288 (but note, as explained at L<the top of this section|/Character case
1289 changing>, that there may be more).
1291 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1292 =for apidoc_item toTITLE_utf8_safe
1293 Convert the first UTF-8 encoded character in the sequence starting at C<p> and
1294 extending no further than S<C<e - 1>> to its titlecase version, and
1295 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1296 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1297 bytes since the titlecase version may be longer than the original character.
1299 The first code point of the titlecased version is returned
1300 (but note, as explained at L<the top of this section|/Character case
1301 changing>, that there may be more).
1304 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1305 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1306 input character is malformed in some way, the program may croak, or the
1307 function may return the REPLACEMENT CHARACTER, at the discretion of the
1308 implementation, and subject to change in future releases.
1310 C<toTITLE_utf8_safe> is now just a different spelling of plain C<toTITLE_utf8>
1314 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1315 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1316 and aren't general purpose as they don't work on U+DF, and assert against that.
1318 Note that these macros are repeated in Devel::PPPort, so should also be
1319 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1324 void below because that's the best fit, and works for Devel::PPPort
1325 =for apidoc_section Integer configuration values
1326 =for apidoc AmnU|void|WIDEST_UTYPE
1328 Yields the widest unsigned integer type on the platform, currently either
1329 C<U32> or C<U64>. This can be used in declarations such as
1335 my_uv = (WIDEST_UTYPE) val;
1341 # define WIDEST_UTYPE U64
1343 # define WIDEST_UTYPE U32
1346 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1347 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1348 * bits of information are lost even on a 64-bit machine, but to get the
1349 * compiler to optimize it out if possible. This is because Configure makes
1350 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1351 * sizeof() guarantees that this evaluates to a constant true at compile time.
1353 * For Coverity, be always true, because otherwise Coverity thinks
1354 * it finds several expressions that are always true, independent
1355 * of operands. Well, they are, but that is kind of the point.
1357 #ifndef __COVERITY__
1358 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1360 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1361 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1363 #define FITS_IN_8_BITS(c) (1)
1366 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
1367 * Written this way so that after optimization, only one conditional test is
1368 * needed. (The NV casts stop any warnings about comparison always being true
1369 * if called with an unsigned. The cast preserves the sign, which is all we
1371 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1372 __ASSERT_((NV) (n) >= 0) \
1373 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
1375 /* Returns true if c is in the range l..u, where 'l' is non-negative
1376 * Written this way so that after optimization, only one conditional test is
1378 #define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
1379 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1380 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1381 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1382 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
1385 # ifndef _ALL_SOURCE
1386 /* The native libc isascii() et.al. functions return the wrong results
1387 * on at least z/OS unless this is defined. */
1388 # error _ALL_SOURCE should probably be defined
1391 /* There is a simple definition of ASCII for ASCII platforms. But the
1392 * EBCDIC one isn't so simple, so is defined using table look-up like the
1393 * other macros below.
1395 * The cast here is used instead of '(c) >= 0', because some compilers emit
1396 * a warning that that test is always true when the parameter is an
1397 * unsigned type. khw supposes that it could be written as
1398 * && ((c) == '\0' || (c) > 0)
1399 * to avoid the message, but the cast will likely avoid extra branches even
1400 * with stupid compilers.
1402 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1404 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1407 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1408 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1409 * can be ignored. If the rest match '0', we have an octal */
1410 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1412 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1414 /* Character class numbers. For internal core Perl use only. The ones less
1415 * than 32 are used in PL_charclass[] and the ones up through the one that
1416 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1417 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1418 * their actual definitions are here. If that file has a name not used here,
1421 * The first group of these is ordered in what I (khw) estimate to be the
1422 * frequency of their use. This gives a slight edge to exiting a loop earlier
1423 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1424 * real Posix character class, and some (small) inefficiencies in regular
1425 * expression handling would be introduced by putting it in the middle of those
1426 * that are. Also, cntrl and ascii come after the others as it may be useful
1427 * to group these which have no members that match above Latin1, (or above
1428 * ASCII in the latter case) */
1430 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1431 # define _CC_DIGIT 1 /* \d and [:digit:] */
1432 # define _CC_ALPHA 2 /* [:alpha:] */
1433 # define _CC_LOWER 3 /* [:lower:] */
1434 # define _CC_UPPER 4 /* [:upper:] */
1435 # define _CC_PUNCT 5 /* [:punct:] */
1436 # define _CC_PRINT 6 /* [:print:] */
1437 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1438 # define _CC_GRAPH 8 /* [:graph:] */
1439 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1440 # define _CC_SPACE 10 /* \s, [:space:] */
1441 # define _CC_BLANK 11 /* [:blank:] */
1442 # define _CC_XDIGIT 12 /* [:xdigit:] */
1443 # define _CC_CNTRL 13 /* [:cntrl:] */
1444 # define _CC_ASCII 14 /* [:ascii:] */
1445 # define _CC_VERTSPACE 15 /* \v */
1447 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1449 /* The members of the third group below do not need to be coordinated with data
1450 * structures in regcomp.[ch] and regexec.c. */
1451 # define _CC_IDFIRST 16
1452 # define _CC_CHARNAME_CONT 17
1453 # define _CC_NONLATIN1_FOLD 18
1454 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1455 # define _CC_QUOTEMETA 20
1456 # define _CC_NON_FINAL_FOLD 21
1457 # define _CC_IS_IN_SOME_FOLD 22
1458 # define _CC_BINDIGIT 23
1459 # define _CC_OCTDIGIT 24
1460 # define _CC_MNEMONIC_CNTRL 25
1462 /* This next group is only used on EBCDIC platforms, so theoretically could be
1463 * shared with something entirely different that's only on ASCII platforms */
1464 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1466 * If more bits are needed, one could add a second word for non-64bit
1467 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1468 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1469 * is used only for optimization (as of this writing), and differs in the
1470 * Latin1 range from the ALPHA bit only in two relatively unimportant
1471 * characters: the masculine and feminine ordinal indicators, so removing it
1472 * would just cause /i regexes which match them to run less efficiently.
1473 * Similarly the EBCDIC-only bits are used just for speed, and could be
1474 * replaced by other means */
1476 #if defined(PERL_CORE) || defined(PERL_EXT)
1477 /* An enum version of the character class numbers, to help compilers
1480 _CC_ENUM_ALPHA = _CC_ALPHA,
1481 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1482 _CC_ENUM_ASCII = _CC_ASCII,
1483 _CC_ENUM_BLANK = _CC_BLANK,
1484 _CC_ENUM_CASED = _CC_CASED,
1485 _CC_ENUM_CNTRL = _CC_CNTRL,
1486 _CC_ENUM_DIGIT = _CC_DIGIT,
1487 _CC_ENUM_GRAPH = _CC_GRAPH,
1488 _CC_ENUM_LOWER = _CC_LOWER,
1489 _CC_ENUM_PRINT = _CC_PRINT,
1490 _CC_ENUM_PUNCT = _CC_PUNCT,
1491 _CC_ENUM_SPACE = _CC_SPACE,
1492 _CC_ENUM_UPPER = _CC_UPPER,
1493 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1494 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1495 _CC_ENUM_XDIGIT = _CC_XDIGIT
1496 } _char_class_number;
1499 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1503 EXTCONST U32 PL_charclass[] = {
1504 # include "l1_char_class_tab.h"
1507 # else /* ! DOINIT */
1508 EXTCONST U32 PL_charclass[];
1512 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1513 # define _CC_mask(classnum) (1U << (classnum))
1515 /* For internal core Perl use only: the base macro for defining macros like
1517 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1518 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1520 /* The mask for the _A versions of the macros; it just adds in the bit for
1522 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1524 /* For internal core Perl use only: the base macro for defining macros like
1525 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1526 * the ASCII bit are present */
1527 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1528 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1529 == _CC_mask_A(classnum)))
1531 /* On ASCII platforms certain classes form a single range. It's faster to
1532 * special case these. isDIGIT is a single range on all platforms */
1534 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1535 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1536 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1537 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1538 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1540 /* By folding the upper and lowercase, we can use a single range */
1541 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1542 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1543 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1544 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1545 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1547 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1548 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1549 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1550 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1551 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1552 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1553 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1554 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1556 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1557 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1558 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1559 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1561 /* continuation character for legal NAME in \N{NAME} */
1562 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1564 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1565 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1566 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1567 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1568 # define isPSXSPC_L1(c) isSPACE_L1(c)
1569 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1570 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1571 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1572 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1573 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1576 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1579 /* Participates in a single-character fold with a character above 255 */
1580 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1582 /* Like the above, but also can be part of a multi-char fold */
1583 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
1585 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1586 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1587 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1588 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1589 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1591 /* is c a control character for which we have a mnemonic? */
1592 # if defined(PERL_CORE) || defined(PERL_EXT)
1593 # define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1595 #else /* else we don't have perl.h H_PERL */
1597 /* If we don't have perl.h, we are compiling a utility program. Below we
1598 * hard-code various macro definitions that wouldn't otherwise be available
1599 * to it. Most are coded based on first principles. These are written to
1600 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1601 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1602 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1603 # define isSPACE_A(c) (isBLANK_A(c) \
1608 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1609 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1610 * performance (if optimization isn't on), and allow the same code to be
1611 * used for both platform types */
1612 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1613 || inRANGE((c), 'j', 'r') \
1614 || inRANGE((c), 's', 'z')
1615 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1616 || inRANGE((c), 'J', 'R') \
1617 || inRANGE((c), 'S', 'Z')
1618 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1619 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1620 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1621 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1622 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1623 || inRANGE((c), 'a', 'f') \
1624 || inRANGE((c), 'A', 'F')
1625 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1626 || (c) == '#' || (c) == '$' || (c) == '%' \
1627 || (c) == '&' || (c) == '\'' || (c) == '(' \
1628 || (c) == ')' || (c) == '*' || (c) == '+' \
1629 || (c) == ',' || (c) == '.' || (c) == '/' \
1630 || (c) == ':' || (c) == ';' || (c) == '<' \
1631 || (c) == '=' || (c) == '>' || (c) == '?' \
1632 || (c) == '@' || (c) == '[' || (c) == '\\' \
1633 || (c) == ']' || (c) == '^' || (c) == '_' \
1634 || (c) == '`' || (c) == '{' || (c) == '|' \
1635 || (c) == '}' || (c) == '~')
1636 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1637 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1640 /* The below is accurate for the 3 EBCDIC code pages traditionally
1641 * supported by perl. The only difference between them in the controls
1642 * is the position of \n, and that is represented symbolically below */
1643 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1644 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1645 || (c) == '\t' || (c) == '\v' \
1646 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1647 || (c) == 7F /* U+7F DEL */ \
1648 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1650 || (c) == 0x18 /* U+18 CAN */ \
1651 || (c) == 0x19 /* U+19 EOM */ \
1652 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1653 || (c) == 0x26 /* U+17 ETB */ \
1654 || (c) == 0x27 /* U+1B ESC */ \
1655 || (c) == 0x2D /* U+05 ENQ */ \
1656 || (c) == 0x2E /* U+06 ACK */ \
1657 || (c) == 0x32 /* U+16 SYN */ \
1658 || (c) == 0x37 /* U+04 EOT */ \
1659 || (c) == 0x3C /* U+14 DC4 */ \
1660 || (c) == 0x3D /* U+15 NAK */ \
1661 || (c) == 0x3F)/* U+1A SUB */
1662 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1663 # else /* isASCII is already defined for ASCII platforms, so can use that to
1665 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1668 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1669 * during debugging, and it seems best to keep them. We may be called
1670 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1671 * do anything anyway, so make it not a problem */
1672 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1673 # define NATIVE_TO_LATIN1(ch) (ch)
1675 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1676 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1677 # define isBLANK_L1(c) (isBLANK_A(c) \
1678 || (FITS_IN_8_BITS(c) \
1679 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1680 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1681 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1682 # define isLOWER_L1(c) (isLOWER_A(c) \
1683 || (FITS_IN_8_BITS(c) \
1684 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1685 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1686 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1687 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1688 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1689 # define isPRINT_L1(c) (isPRINT_A(c) \
1690 || (FITS_IN_8_BITS(c) \
1691 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1692 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1693 || (FITS_IN_8_BITS(c) \
1694 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1695 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1696 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1697 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1698 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1699 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1700 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1701 # define isSPACE_L1(c) (isSPACE_A(c) \
1702 || (FITS_IN_8_BITS(c) \
1703 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1704 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1705 # define isUPPER_L1(c) (isUPPER_A(c) \
1706 || (FITS_IN_8_BITS(c) \
1707 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1709 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1710 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1711 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1712 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1717 /* The following are not fully accurate in the above-ASCII range. I (khw)
1718 * don't think it's necessary to be so for the purposes where this gets
1720 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1721 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1723 /* And these aren't accurate at all. They are useful only for above
1724 * Latin1, which utilities and bootstrapping don't deal with */
1725 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1726 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1727 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1729 /* Many of the macros later in this file are defined in terms of these. By
1730 * implementing them with a function, which converts the class number into
1731 * a call to the desired macro, all of the later ones work. However, that
1732 * function won't be actually defined when building a utility program (no
1733 * perl.h), and so a compiler error will be generated if one is attempted
1734 * to be used. And the above-Latin1 code points require Unicode tables to
1735 * be present, something unlikely to be the case when bootstrapping */
1736 # define _generic_isCC(c, classnum) \
1737 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1738 # define _generic_isCC_A(c, classnum) \
1739 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1740 #endif /* End of no perl.h H_PERL */
1742 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1743 #define isALPHA(c) isALPHA_A(c)
1744 #define isASCII_A(c) isASCII(c)
1745 #define isASCII_L1(c) isASCII(c)
1746 #define isBLANK(c) isBLANK_A(c)
1747 #define isCNTRL(c) isCNTRL_A(c)
1748 #define isDIGIT(c) isDIGIT_A(c)
1749 #define isGRAPH(c) isGRAPH_A(c)
1750 #define isIDFIRST(c) isIDFIRST_A(c)
1751 #define isLOWER(c) isLOWER_A(c)
1752 #define isPRINT(c) isPRINT_A(c)
1753 #define isPSXSPC_A(c) isSPACE_A(c)
1754 #define isPSXSPC(c) isPSXSPC_A(c)
1755 #define isPSXSPC_L1(c) isSPACE_L1(c)
1756 #define isPUNCT(c) isPUNCT_A(c)
1757 #define isSPACE(c) isSPACE_A(c)
1758 #define isUPPER(c) isUPPER_A(c)
1759 #define isWORDCHAR(c) isWORDCHAR_A(c)
1760 #define isXDIGIT(c) isXDIGIT_A(c)
1762 /* ASCII casing. These could also be written as
1763 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1764 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1765 which uses table lookup and mask instead of subtraction. (This would
1766 work because the _MOD does not apply in the ASCII range).
1768 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1769 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1770 there non-ASCII invariants, and all of them are controls.) */
1771 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1772 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1774 /* In the ASCII range, these are equivalent to what they're here defined to be.
1775 * But by creating these definitions, other code doesn't have to be aware of
1776 * this detail. Actually this works for all UTF-8 invariants, not just the
1777 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1778 #define toFOLD(c) toLOWER(c)
1779 #define toTITLE(c) toUPPER(c)
1781 #define toLOWER_A(c) toLOWER(c)
1782 #define toUPPER_A(c) toUPPER(c)
1783 #define toFOLD_A(c) toFOLD(c)
1784 #define toTITLE_A(c) toTITLE(c)
1786 /* Use table lookup for speed; returns the input itself if is out-of-range */
1787 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1789 : PL_latin1_lc[ (U8) (c) ])
1790 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1792 /* Modified uc. Is correct uc except for three non-ascii chars which are
1793 * all mapped to one of them, and these need special handling; returns the
1794 * input itself if is out-of-range */
1795 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1797 : PL_mod_latin1_uc[ (U8) (c) ])
1798 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1800 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1802 /* For internal core Perl use only: the base macro for defining macros like
1803 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1804 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1805 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1806 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1807 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1808 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1809 * code points under locale rules, use the macros based on _generic_LC_uvchr
1810 * instead of this. */
1811 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1812 (! FITS_IN_8_BITS(c) \
1814 : IN_UTF8_CTYPE_LOCALE \
1815 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1818 /* For internal core Perl use only: a helper macro for defining macros like
1819 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1820 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1821 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1822 * _generic_LC_base for more info */
1823 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1824 _generic_LC_base(c,utf8_locale_classnum, \
1825 non_utf8_func( (_LC_CAST) (c)))
1827 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1828 * 'c' is the platform's native underscore character */
1829 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1830 _generic_LC_base(c, utf8_locale_classnum, \
1831 (non_utf8_func( (_LC_CAST) (c)) \
1832 || (char)(c) == '_'))
1834 /* These next three are also for internal core Perl use only: case-change
1835 * helper macros. The reason for using the PL_latin arrays is in case the
1836 * system function is defective; it ensures uniform results that conform to the
1837 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1838 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1840 : (IN_UTF8_CTYPE_LOCALE) \
1841 ? PL_latin1_lc[ (U8) (c) ] \
1842 : (cast)function((cast)(c)))
1844 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1845 * returns a single value, so can't adequately return the upper case of LATIN
1846 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1847 * values "SS"); instead it asserts against that under DEBUGGING, and
1848 * otherwise returns its input. It does not handle the anomalies in UTF-8
1849 * Turkic locales. */
1850 #define _generic_toUPPER_LC(c, function, cast) \
1851 (! FITS_IN_8_BITS(c) \
1853 : ((! IN_UTF8_CTYPE_LOCALE) \
1854 ? (cast)function((cast)(c)) \
1855 : ((((U8)(c)) == MICRO_SIGN) \
1856 ? GREEK_CAPITAL_LETTER_MU \
1857 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1858 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1859 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1860 ? (__ASSERT_(0) (c)) \
1861 : PL_mod_latin1_uc[ (U8) (c) ])))))
1863 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1864 * returns a single value, so can't adequately return the fold case of LATIN
1865 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1866 * values "ss"); instead it asserts against that under DEBUGGING, and
1867 * otherwise returns its input. It does not handle the anomalies in UTF-8
1869 #define _generic_toFOLD_LC(c, function, cast) \
1870 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1871 ? GREEK_SMALL_LETTER_MU \
1872 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1873 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1874 _generic_toLOWER_LC(c, function, cast)))
1876 /* Use the libc versions for these if available. */
1877 #if defined(HAS_ISASCII)
1878 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1880 # define isASCII_LC(c) isASCII(c)
1883 #if defined(HAS_ISBLANK)
1884 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1885 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1886 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1892 /* The Windows functions don't bother to follow the POSIX standard, which
1893 * for example says that something can't both be a printable and a control.
1894 * But Windows treats the \t control as a printable, and does such things
1895 * as making superscripts into both digits and punctuation. This tames
1896 * these flaws by assuming that the definitions of both controls and space
1897 * are correct, and then making sure that other definitions don't have
1898 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1899 * Not all possible weirdnesses are checked for, just the ones that were
1900 * detected on actual Microsoft code pages */
1902 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1903 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1905 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1906 && isALPHANUMERIC_LC(c))
1907 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1909 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1910 isALPHANUMERIC_LC(c))
1911 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1912 # define isIDFIRST_LC(c) (((c) == '_') \
1913 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1914 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1915 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1916 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1917 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1918 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1919 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1920 && isALPHANUMERIC_LC(c))
1922 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1923 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1924 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1926 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1927 /* For most other platforms */
1929 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1930 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1931 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1932 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1933 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1934 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1935 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1936 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1937 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1938 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1939 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1940 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1941 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1944 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1945 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1946 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1948 #else /* The final fallback position */
1950 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1951 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1952 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1953 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1954 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1955 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1956 # define isLOWER_LC(c) (isascii(c) && islower(c))
1957 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1958 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1959 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1960 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1961 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1962 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1964 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1965 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1966 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1970 #define isIDCONT(c) isWORDCHAR(c)
1971 #define isIDCONT_A(c) isWORDCHAR_A(c)
1972 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1973 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1974 #define isPSXSPC_LC(c) isSPACE_LC(c)
1976 /* For internal core Perl use only: the base macros for defining macros like
1977 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1978 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1979 * classes where there is a macro or function 'above_latin1' that takes the
1980 * single argument 'c' and returns the desired value. These exist for those
1981 * classes which have simple definitions, avoiding the overhead of an inversion
1982 * list binary search. _generic_invlist_uvchr() can be used
1983 * for classes where that overhead is faster than a direct lookup.
1984 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
1985 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
1986 * have duplicate checks here, so could create versions of the macros that
1987 * don't, but experiments show that gcc optimizes them out anyway. */
1989 /* Note that all ignore 'use bytes' */
1990 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
1991 ? _generic_isCC(c, classnum) \
1993 #define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
1994 ? _generic_isCC(c, classnum) \
1995 : _is_uni_FOO(classnum, c))
1996 #define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
1997 #define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
1998 #define isASCII_uvchr(c) isASCII(c)
1999 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2000 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2001 #define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2002 #define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
2003 #define isIDCONT_uvchr(c) \
2004 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2005 #define isIDFIRST_uvchr(c) \
2006 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2007 #define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2008 #define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
2010 #define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
2011 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2012 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2014 #define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
2015 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2016 #define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
2017 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2019 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2020 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2021 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2022 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2024 /* For backwards compatibility, even though '_uni' should mean official Unicode
2025 * code points, in Perl it means native for those below 256 */
2026 #define isALPHA_uni(c) isALPHA_uvchr(c)
2027 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2028 #define isASCII_uni(c) isASCII_uvchr(c)
2029 #define isBLANK_uni(c) isBLANK_uvchr(c)
2030 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
2031 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
2032 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
2033 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
2034 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2035 #define isLOWER_uni(c) isLOWER_uvchr(c)
2036 #define isPRINT_uni(c) isPRINT_uvchr(c)
2037 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
2038 #define isSPACE_uni(c) isSPACE_uvchr(c)
2039 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2040 #define isUPPER_uni(c) isUPPER_uvchr(c)
2041 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
2042 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2043 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2044 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2045 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2046 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2047 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2049 /* For internal core Perl use only: the base macros for defining macros like
2050 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2051 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2052 * simple class definitions; the other for more complex. These are like
2053 * _generic_uvchr, so see it for more info. */
2054 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2055 (c < 256 ? latin1(c) : above_latin1(c))
2056 #define _generic_LC_invlist_uvchr(latin1, classnum, c) \
2057 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2059 #define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2060 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
2061 _CC_ALPHANUMERIC, c)
2062 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2063 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2064 is_HORIZWS_cp_high, c)
2065 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2066 #define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2067 #define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2068 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2069 _is_uni_perl_idcont, c)
2070 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2071 _is_uni_perl_idstart, c)
2072 #define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2073 #define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
2074 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2075 #define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2076 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2077 is_XPERLSPACE_cp_high, c)
2078 #define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2079 #define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
2081 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2082 is_XDIGIT_cp_high, c)
2084 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2086 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2087 * they don't otherwise go out of their way to look for malformed UTF-8. If
2088 * they can return accurate results without knowing if the input is otherwise
2089 * malformed, they do so. For example isASCII is accurate in spite of any
2090 * non-length malformations because it looks only at a single byte. Likewise
2091 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2092 * variant ones return FALSE. But, if the input has to be well-formed in order
2093 * for the results to be accurate, the macros will test and if malformed will
2094 * call a routine to die
2096 * Except for toke.c, the macros do assume that e > p, asserting that on
2097 * DEBUGGING builds. Much code that calls these depends on this being true,
2098 * for other reasons. toke.c is treated specially as using the regular
2099 * assertion breaks it in many ways. All strings that these operate on there
2100 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2101 * bunch of code in toke.c assumes that this is true, so the assertion allows
2103 #ifdef PERL_IN_TOKE_C
2104 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2106 # define _utf8_safe_assert(p,e) ((e) > (p))
2109 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2110 ((! _utf8_safe_assert(p, e)) \
2111 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2112 : (UTF8_IS_INVARIANT(*(p))) \
2113 ? _generic_isCC(*(p), classnum) \
2114 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2115 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2116 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2118 : (_force_out_malformed_utf8_message( \
2119 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2121 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2122 * 'above_latin1' can be a macro */
2123 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2124 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2125 #define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2126 _generic_utf8_safe(classnum, p, e, \
2127 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2128 ? (_force_out_malformed_utf8_message( \
2129 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2131 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2132 * 'above_latin1' parameter */
2133 #define _generic_invlist_utf8_safe(classnum, p, e) \
2134 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
2136 /* Like the above, but should be used only when it is known that there are no
2137 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2138 * class is TRUE for. Hence it can skip the tests for this range.
2139 * 'above_latin1' should include its arguments */
2140 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2141 (__ASSERT_(_utf8_safe_assert(p, e)) \
2142 (UTF8_IS_INVARIANT(*(p))) \
2143 ? _generic_isCC(*(p), classnum) \
2144 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2145 ? 0 /* Note that doesn't check validity for latin1 */ \
2149 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2150 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2151 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2152 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2153 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2154 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2155 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2156 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2157 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2158 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2159 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2160 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2161 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2162 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2163 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2164 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2165 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2166 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
2168 #define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
2169 #define isALPHANUMERIC_utf8_safe(p, e) \
2170 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
2171 #define isASCII_utf8_safe(p, e) \
2172 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2174 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2175 #define isBLANK_utf8_safe(p, e) \
2176 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2179 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2180 * more efficient macro instead of the more general one */
2181 # define isCNTRL_utf8_safe(p, e) \
2182 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2184 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2187 #define isDIGIT_utf8_safe(p, e) \
2188 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2189 _is_utf8_FOO(_CC_DIGIT, p, e))
2190 #define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
2191 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2192 _is_utf8_perl_idcont, p, e)
2194 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2195 * IDFIRST is an alnum. See
2196 * https://github.com/Perl/perl5/issues/10275 for more detail than you
2197 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2198 * This used to be not the XID version, but we decided to go with the more
2199 * modern Unicode definition */
2200 #define isIDFIRST_utf8_safe(p, e) \
2201 _generic_func_utf8_safe(_CC_IDFIRST, \
2202 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
2204 #define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2205 #define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
2206 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2207 #define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
2208 #define isSPACE_utf8_safe(p, e) \
2209 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2210 #define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
2211 #define isVERTWS_utf8_safe(p, e) \
2212 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2213 #define isWORDCHAR_utf8_safe(p, e) \
2214 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
2215 #define isXDIGIT_utf8_safe(p, e) \
2216 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2217 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2218 ? (_force_out_malformed_utf8_message( \
2219 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2220 : is_XDIGIT_high(p)))
2222 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2223 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2224 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2225 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2227 /* For internal core use only, subject to change */
2228 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2229 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2230 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2231 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
2233 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2234 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2235 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2236 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2238 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2239 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2240 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2241 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2242 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2243 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2244 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2245 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2246 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2247 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2248 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2249 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2250 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2251 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2252 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2253 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2254 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
2256 /* For internal core Perl use only: the base macros for defining macros like
2257 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2258 * point in 'p' is within the 0-255 range, it uses locale rules from the
2259 * passed-in 'macro' parameter */
2260 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2261 (__ASSERT_(_utf8_safe_assert(p, e)) \
2262 (UTF8_IS_INVARIANT(*(p))) \
2264 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2265 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2266 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2267 : (_force_out_malformed_utf8_message( \
2268 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2271 #define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2272 _generic_LC_utf8_safe(macro, p, e, \
2273 _is_utf8_FOO(classnum, p, e))
2275 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2276 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2278 #define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2279 _generic_LC_utf8_safe(classnum, p, e, \
2280 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2281 ? (_force_out_malformed_utf8_message( \
2282 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2285 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2286 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
2287 _CC_ALPHANUMERIC, p, e)
2288 #define isALPHA_LC_utf8_safe(p, e) \
2289 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2290 #define isASCII_LC_utf8_safe(p, e) \
2291 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2292 #define isBLANK_LC_utf8_safe(p, e) \
2293 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2294 #define isCNTRL_LC_utf8_safe(p, e) \
2295 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2296 #define isDIGIT_LC_utf8_safe(p, e) \
2297 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2298 #define isGRAPH_LC_utf8_safe(p, e) \
2299 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2300 #define isIDCONT_LC_utf8_safe(p, e) \
2301 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2302 _is_utf8_perl_idcont, p, e)
2303 #define isIDFIRST_LC_utf8_safe(p, e) \
2304 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2305 _is_utf8_perl_idstart, p, e)
2306 #define isLOWER_LC_utf8_safe(p, e) \
2307 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2308 #define isPRINT_LC_utf8_safe(p, e) \
2309 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2310 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2311 #define isPUNCT_LC_utf8_safe(p, e) \
2312 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2313 #define isSPACE_LC_utf8_safe(p, e) \
2314 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2315 #define isUPPER_LC_utf8_safe(p, e) \
2316 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2317 #define isWORDCHAR_LC_utf8_safe(p, e) \
2318 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2319 #define isXDIGIT_LC_utf8_safe(p, e) \
2320 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2322 /* Macros for backwards compatibility and for completeness when the ASCII and
2323 * Latin1 values are identical */
2324 #define isALPHAU(c) isALPHA_L1(c)
2325 #define isDIGIT_L1(c) isDIGIT_A(c)
2326 #define isOCTAL(c) isOCTAL_A(c)
2327 #define isOCTAL_L1(c) isOCTAL_A(c)
2328 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2329 #define isALNUM(c) isWORDCHAR(c)
2330 #define isALNUM_A(c) isALNUM(c)
2331 #define isALNUMU(c) isWORDCHAR_L1(c)
2332 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2333 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2334 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2335 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
2336 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
2337 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
2338 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
2339 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2340 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2341 #define isALNUMC(c) isALPHANUMERIC(c)
2342 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2343 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2344 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2345 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
2346 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2347 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
2349 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2350 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2351 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2352 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2353 * the outlier from the block that contains the other controls, just like
2354 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2355 * block. If it weren't special cased, it would yield a non-control.
2356 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2359 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2361 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2363 ? (UNLIKELY((c) == '?') \
2364 ? QUESTION_MARK_CTRL \
2365 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2366 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2368 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2371 /* Line numbers are unsigned, 32 bits. */
2373 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2375 /* Helpful alias for version prescan */
2376 #define is_LAX_VERSION(a,b) \
2377 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2379 #define is_STRICT_VERSION(a,b) \
2380 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2382 #define BADVERSION(a,b,c) \
2388 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2389 * a-f) to its numeric value without using any branches. The input is
2390 * validated only by an assert() in DEBUGGING builds.
2392 * It works by right shifting and isolating the bit that is 0 for the digits,
2393 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2394 * position, and then to the eights position. Both are added together to form
2395 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2396 * final four bits of the input to form the correct value. */
2397 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2398 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2399 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2400 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2402 /* The argument is a string pointer, which is advanced. */
2403 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
2405 /* Converts a character known to represent an octal digit (0-7) to its numeric
2406 * value. The input is validated only by an assert() in DEBUGGING builds. In
2407 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2408 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2410 /* Efficiently returns a boolean as to if two native characters are equivalent
2411 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2412 * the ALPHA in the name is to remind you of that. This is asserted() in
2413 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2414 * works (on valid input) for both non- and UTF-8-encoded bytes.
2416 * When one of the inputs is a compile-time constant and gets folded by the
2417 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2418 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2419 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2420 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2421 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2422 * just a single 0, in the bit position where the upper- and lowercase differ.
2424 #define isALPHA_FOLD_EQ(c1, c2) \
2425 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2426 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2427 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2430 =for apidoc_section Memory Management
2432 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2433 The XSUB-writer's interface to the C C<malloc> function.
2435 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2437 In 5.9.3, Newx() and friends replace the older New() API, and drops
2438 the first parameter, I<x>, a debug aid which allowed callers to identify
2439 themselves. This aid has been superseded by a new build option,
2440 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2441 there for use in XS modules supporting older perls.
2443 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2444 The XSUB-writer's interface to the C C<malloc> function, with
2445 cast. See also C<L</Newx>>.
2447 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2449 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2450 The XSUB-writer's interface to the C C<malloc> function. The allocated
2451 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2453 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2455 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2456 The XSUB-writer's interface to the C C<realloc> function.
2458 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2460 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2461 The XSUB-writer's interface to the C C<realloc> function, with
2464 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2466 =for apidoc Am|void|Safefree|void* ptr
2467 The XSUB-writer's interface to the C C<free> function.
2469 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2471 =for apidoc_section String Handling
2472 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2473 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2474 source, C<dest> is the destination, C<nitems> is the number of items, and
2475 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2477 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2478 Like C<Move> but returns C<dest>. Useful
2479 for encouraging compilers to tail-call
2482 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2483 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2484 source, C<dest> is the destination, C<nitems> is the number of items, and
2485 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2487 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2489 Like C<Copy> but returns C<dest>. Useful
2490 for encouraging compilers to tail-call
2493 =for apidoc Am|void|Zero|void* dest|int nitems|type
2495 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2496 destination, C<nitems> is the number of items, and C<type> is the type.
2498 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2500 Like C<Zero> but returns dest. Useful
2501 for encouraging compilers to tail-call
2504 =for apidoc_section Utility Functions
2505 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2506 This is an architecture-independent macro to copy one structure to another.
2508 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2510 Fill up memory with a byte pattern (a byte repeated over and over
2511 again) that hopefully catches attempts to access uninitialized memory.
2513 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2515 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2517 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2519 PoisonWith(0xEF) for catching access to freed memory.
2521 =for apidoc Am|void|Poison|void* dest|int nitems|type
2523 PoisonWith(0xEF) for catching access to freed memory.
2527 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2529 #define NEWSV(x,len) newSV(len)
2532 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2534 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2536 #ifdef PERL_MALLOC_WRAP
2538 /* This expression will be constant-folded at compile time. It checks
2539 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2540 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2541 * As well as avoiding the need for a run-time check in some cases, it's
2542 * designed to avoid compiler warnings like:
2543 * comparison is always false due to limited range of data type
2544 * It's mathematically equivalent to
2545 * max(n) * sizeof(t) > MEM_SIZE_MAX
2548 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2549 ( sizeof(MEM_SIZE) < sizeof(n) \
2550 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2552 /* This is written in a slightly odd way to avoid various spurious
2553 * compiler warnings. We *want* to write the expression as
2554 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2555 * (for some compile-time constant C), but even when the LHS
2556 * constant-folds to false at compile-time, g++ insists on emitting
2557 * warnings about the RHS (e.g. "comparison is always false"), so instead
2560 * (cond ? n : X) > C
2562 * where X is a constant with X > C always false. Choosing a value for X
2563 * is tricky. If 0, some compilers will complain about 0 > C always being
2564 * false; if 1, Coverity complains when n happens to be the constant value
2565 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2566 * for X and hope that nothing else whines.
2569 # define _MEM_WRAP_WILL_WRAP(n,t) \
2570 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2571 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2573 # define MEM_WRAP_CHECK(n,t) \
2574 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2575 && (croak_memory_wrap(),0))
2577 # define MEM_WRAP_CHECK_1(n,t,a) \
2578 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2579 && (Perl_croak_nocontext("%s",(a)),0))
2581 /* "a" arg must be a string literal */
2582 # define MEM_WRAP_CHECK_s(n,t,a) \
2583 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2584 && (Perl_croak_nocontext("" a ""),0))
2586 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2588 #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))
2591 #define MEM_WRAP_CHECK(n,t)
2592 #define MEM_WRAP_CHECK_1(n,t,a)
2593 #define MEM_WRAP_CHECK_s(n,t,a)
2594 #define MEM_WRAP_CHECK_(n,t)
2596 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2602 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2603 * go through functions, which are handy for debugging breakpoints, but
2604 * which more importantly get the immediate calling environment (file and
2605 * line number, and C function name if available) passed in. This info can
2606 * then be used for logging the calls, for which one gets a sample
2607 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2610 * - not all memory allocs get logged, only those
2611 * that go through Newx() and derivatives (while all
2612 * Safefrees do get logged)
2613 * - __FILE__ and __LINE__ do not work everywhere
2614 * - __func__ or __FUNCTION__ even less so
2615 * - I think more goes on after the perlio frees but
2616 * the thing is that STDERR gets closed (as do all
2617 * the file descriptors)
2618 * - no deeper calling stack than the caller of the Newx()
2619 * or the kind, but do I look like a C reflection/introspection
2621 * - the function prototypes for the logging functions
2622 * probably should maybe be somewhere else than handy.h
2623 * - one could consider inlining (macrofying) the logging
2624 * for speed, but I am too lazy
2625 * - one could imagine recording the allocations in a hash,
2626 * (keyed by the allocation address?), and maintain that
2627 * through reallocs and frees, but how to do that without
2628 * any News() happening...?
2629 * - lots of -Ddefines to get useful/controllable output
2630 * - lots of ENV reads
2634 # ifndef PERL_MEM_LOG_NOIMPL
2643 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2644 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2645 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2652 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2653 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2654 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2657 #ifndef MEM_LOG_ALLOC
2658 #define MEM_LOG_ALLOC(n,t,a) (a)
2660 #ifndef MEM_LOG_REALLOC
2661 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2663 #ifndef MEM_LOG_FREE
2664 #define MEM_LOG_FREE(a) (a)
2667 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2668 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2669 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2672 /* pre 5.9.x compatibility */
2673 #define New(x,v,n,t) Newx(v,n,t)
2674 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2675 #define Newz(x,v,n,t) Newxz(v,n,t)
2678 #define Renew(v,n,t) \
2679 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2680 #define Renewc(v,n,t,c) \
2681 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2684 #define Safefree(d) \
2685 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2687 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2690 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2691 * as it handles cases like constant string arguments without throwing warnings *
2692 * the cast is required, as is the inequality check, to avoid warnings */
2693 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2696 #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)))
2697 #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)))
2698 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2700 /* Like above, but returns a pointer to 'd' */
2701 #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)))
2702 #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)))
2703 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2705 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2706 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2707 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2708 #define Poison(d,n,t) PoisonFree(d,n,t)
2711 # define PERL_POISON_EXPR(x) x
2713 # define PERL_POISON_EXPR(x)
2716 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2719 =for apidoc_section Utility Functions
2721 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2723 Returns the number of elements in the input C array (so you want your
2724 zero-based indices to be less than but not equal to).
2726 =for apidoc Am|void *|C_ARRAY_END|void *a
2728 Returns a pointer to one element past the final element of the input C array.
2732 C_ARRAY_END is one past the last: half-open/half-closed range, not
2733 last-inclusive range.
2735 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2736 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2740 # define Perl_va_copy(s, d) va_copy(d, s)
2741 # elif defined(__va_copy)
2742 # define Perl_va_copy(s, d) __va_copy(d, s)
2744 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2748 /* convenience debug macros */
2750 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2751 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2752 #define pTHX_VALUE_ (void *)my_perl,
2753 #define pTHX_VALUE (void *)my_perl
2754 #define pTHX__VALUE_ ,(void *)my_perl,
2755 #define pTHX__VALUE ,(void *)my_perl
2758 #define pTHX__FORMAT
2761 #define pTHX__VALUE_
2763 #endif /* USE_ITHREADS */
2765 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2766 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2767 CPAN::Unpack show any users outside the core. */
2769 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2770 "Use of " s " is deprecated")
2771 # define deprecate_disappears_in(when,message) \
2772 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2773 message ", and will disappear in Perl " when)
2774 # define deprecate_fatal_in(when,message) \
2775 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2776 message ". Its use will be fatal in Perl " when)
2779 /* Internal macros to deal with gids and uids */
2782 # if Uid_t_size > IVSIZE
2783 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2784 # define SvUID(sv) SvNV(sv)
2785 # elif Uid_t_sign <= 0
2786 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2787 # define SvUID(sv) SvIV(sv)
2789 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2790 # define SvUID(sv) SvUV(sv)
2791 # endif /* Uid_t_size */
2793 # if Gid_t_size > IVSIZE
2794 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2795 # define SvGID(sv) SvNV(sv)
2796 # elif Gid_t_sign <= 0
2797 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2798 # define SvGID(sv) SvIV(sv)
2800 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2801 # define SvGID(sv) SvUV(sv)
2802 # endif /* Gid_t_size */
2806 #endif /* PERL_HANDY_H_ */
2809 * ex: set ts=8 sts=4 sw=4 et: