3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000,
4 * 2001, 2002, 2004, 2005, 2006, 2007, 2008, 2012 by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
11 /* IMPORTANT NOTE: Everything whose name begins with an underscore is for
12 * internal core Perl use only. */
14 #ifndef PERL_HANDY_H_ /* Guard against nested #inclusion */
18 # define Null(type) ((type)NULL)
21 =for apidoc_section $string
22 =for apidoc AmnU||Nullch
23 Null character pointer. (No longer available when C<PERL_CORE> is
26 =for apidoc_section $SV
27 =for apidoc AmnU||Nullsv
28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
32 Below are signatures of functions from config.h which can't easily be gleaned
33 from it, and are very unlikely to change
35 =for apidoc_section $signals
36 =for apidoc Am|int|Sigsetjmp|jmp_buf env|int savesigs
37 =for apidoc Am|void|Siglongjmp|jmp_buf env|int val
39 =for apidoc_section $filesystem
40 =for apidoc Am|void *|FILE_ptr|FILE * f
41 =for apidoc Am|Size_t|FILE_cnt|FILE * f
42 =for apidoc Am|void *|FILE_base|FILE * f
43 =for apidoc Am|Size_t|FILE_bufsiz|FILE *f
45 =for apidoc_section $string
46 =for apidoc Amu|token|CAT2|token x|token y
47 =for apidoc Amu|string|STRINGIFY|token x
49 =for apidoc_section $numeric
50 =for apidoc Am|double|Drand01
51 =for apidoc Am|void|seedDrand01|Rand_seed_t x
52 =for apidoc Am|char *|Gconvert|double x|Size_t n|bool t|char * b
57 # define Nullch Null(char*)
58 # define Nullfp Null(PerlIO*)
59 # define Nullsv Null(SV*)
72 =for apidoc_section $SV
73 =for apidoc Am|void *|MUTABLE_PTR|void * p
74 =for apidoc_item |AV *|MUTABLE_AV|AV * p
75 =for apidoc_item |CV *|MUTABLE_CV|CV * p
76 =for apidoc_item |GV *|MUTABLE_GV|GV * p
77 =for apidoc_item |HV *|MUTABLE_HV|HV * p
78 =for apidoc_item |IO *|MUTABLE_IO|IO * p
79 =for apidoc_item |SV *|MUTABLE_SV|SV * p
81 The C<MUTABLE_I<*>>() macros cast pointers to the types shown, in such a way
82 (compiler permitting) that casting away const-ness will give a warning;
86 AV *av1 = (AV*)sv; <== BAD: the const has been silently
88 AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
90 C<MUTABLE_PTR> is the base macro used to derive new casts. The other
91 already-built-in ones return pointers to what their names indicate.
96 #if defined(PERL_USE_GCC_BRACE_GROUPS)
97 # define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
99 # define MUTABLE_PTR(p) ((void *) (p))
102 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
103 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
104 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
105 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
106 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
107 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
110 # include <stdbool.h>
114 =for apidoc_section $casting
115 =for apidoc Am|bool|cBOOL|bool expr
117 Cast-to-bool. When Perl was able to be compiled on pre-C99 compilers, a
118 C<(bool)> cast didn't necessarily do the right thing, so this macro was
119 created (and made somewhat complicated to work around bugs in old
120 compilers). Now, many years later, and C99 is used, this is no longer
121 required, but is kept for backwards compatibility.
125 #define cBOOL(cbool) ((bool) (cbool))
127 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
128 * XXX Should really be a Configure probe, with HAS__FUNCTION__
129 * and FUNCTION__ as results.
130 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
131 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
132 # define FUNCTION__ __func__
133 #elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
134 # define FUNCTION__ ""
136 # define FUNCTION__ __FUNCTION__ /* Common extension. */
139 /* XXX A note on the perl source internal type system. The
140 original intent was that I32 be *exactly* 32 bits.
142 Currently, we only guarantee that I32 is *at least* 32 bits.
143 Specifically, if int is 64 bits, then so is I32. (This is the case
144 for the Cray.) This has the advantage of meshing nicely with
145 standard library calls (where we pass an I32 and the library is
146 expecting an int), but the disadvantage that an I32 is not 32 bits.
147 Andy Dougherty August 1996
149 There is no guarantee that there is *any* integral type with
150 exactly 32 bits. It is perfectly legal for a system to have
151 sizeof(short) == sizeof(int) == sizeof(long) == 8.
153 Similarly, there is no guarantee that I16 and U16 have exactly 16
156 For dealing with issues that may arise from various 32/64-bit
157 systems, we will ask Configure to check out
159 SHORTSIZE == sizeof(short)
160 INTSIZE == sizeof(int)
161 LONGSIZE == sizeof(long)
162 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
163 PTRSIZE == sizeof(void *)
164 DOUBLESIZE == sizeof(double)
165 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
169 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
170 # include <inttypes.h>
171 # ifdef INT32_MIN_BROKEN
173 # define INT32_MIN (-2147483647-1)
175 # ifdef INT64_MIN_BROKEN
177 # define INT64_MIN (-9223372036854775807LL-1)
193 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
194 Please search CHAR_MAX in perl.h for further details. */
196 # define U8_MAX UINT8_MAX
198 # define U8_MAX PERL_UCHAR_MAX
201 # define U8_MIN UINT8_MIN
203 # define U8_MIN PERL_UCHAR_MIN
207 # define I16_MAX INT16_MAX
209 # define I16_MAX PERL_SHORT_MAX
212 # define I16_MIN INT16_MIN
214 # define I16_MIN PERL_SHORT_MIN
217 # define U16_MAX UINT16_MAX
219 # define U16_MAX PERL_USHORT_MAX
222 # define U16_MIN UINT16_MIN
224 # define U16_MIN PERL_USHORT_MIN
228 # define I32_MAX INT32_MAX
230 # define I32_MAX PERL_INT_MAX
232 # define I32_MAX PERL_LONG_MAX
235 # define I32_MIN INT32_MIN
237 # define I32_MIN PERL_INT_MIN
239 # define I32_MIN PERL_LONG_MIN
242 # ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
243 # define U32_MAX UINT_MAX
245 # define U32_MAX 4294967295U
248 # define U32_MAX PERL_UINT_MAX
250 # define U32_MAX PERL_ULONG_MAX
253 # define U32_MIN UINT32_MIN
255 # define U32_MIN PERL_UINT_MIN
257 # define U32_MIN PERL_ULONG_MIN
261 =for apidoc_section $integer
262 =for apidoc Ay|| PERL_INT_FAST8_T
263 =for apidoc_item PERL_INT_FAST16_T
264 =for apidoc_item PERL_UINT_FAST8_T
265 =for apidoc_item PERL_UINT_FAST16_T
267 These are equivalent to the correspondingly-named C99 typedefs on platforms
268 that have those; they evaluate to C<int> and C<unsigned int> on platforms that
269 don't, so that you can portably take advantage of this C99 feature.
274 typedef int_fast8_t PERL_INT_FAST8_T;
275 typedef uint_fast8_t PERL_UINT_FAST8_T;
276 typedef int_fast16_t PERL_INT_FAST16_T;
277 typedef uint_fast16_t PERL_UINT_FAST16_T;
279 typedef int PERL_INT_FAST8_T;
280 typedef unsigned int PERL_UINT_FAST8_T;
281 typedef int PERL_INT_FAST16_T;
282 typedef unsigned int PERL_UINT_FAST16_T;
285 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
286 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
287 * required to represent any possible unsigned number containing N bits.
288 * TYPE_DIGITS gives the number of decimal digits required to represent any
289 * possible unsigned number of type T. */
290 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
291 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
292 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
294 /* Unused by core; should be deprecated */
295 #define Ctl(ch) ((ch) & 037)
297 #if defined(PERL_CORE) || defined(PERL_EXT)
299 # define MIN(a,b) ((a) < (b) ? (a) : (b))
302 # define MAX(a,b) ((a) > (b) ? (a) : (b))
306 /* Returns a boolean as to whether the input unsigned number is a power of 2
307 * (2**0, 2**1, etc). In other words if it has just a single bit set.
308 * If not, subtracting 1 would leave the uppermost bit set, so the & would
310 #if defined(PERL_CORE) || defined(PERL_EXT)
311 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
314 /* Returns a mask with the lowest n bits set */
315 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
317 /* The largest unsigned number that will fit into n bits */
318 #define nBIT_UMAX(n) nBIT_MASK(n)
321 =for apidoc_section $directives
322 =for apidoc Am||__ASSERT_|bool expr
324 This is a helper macro to avoid preprocessor issues, replaced by nothing
325 unless under DEBUGGING, where it expands to an assert of its argument,
326 followed by a comma (hence the comma operator). If we just used a straight
327 assert(), we would get a comma with nothing before it when not DEBUGGING.
331 We also use empty definition under Coverity since the __ASSERT_
332 checks often check for things that Really Cannot Happen, and Coverity
333 detects that and gets all excited. */
335 #if defined(DEBUGGING) && !defined(__COVERITY__) \
336 && ! defined(PERL_SMALL_MACRO_BUFFER)
337 # define __ASSERT_(statement) assert(statement),
339 # define __ASSERT_(statement)
343 =for apidoc_section $SV
345 =for apidoc Ama|SV*|newSVpvs|"literal string"
346 Like C<newSVpvn>, but takes a literal string instead of a
349 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
350 Like C<newSVpvn_flags>, but takes a literal string instead of
351 a string/length pair.
353 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
354 Like C<newSVpvn_share>, but takes a literal string instead of
355 a string/length pair and omits the hash parameter.
357 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
358 Like C<sv_catpvn_flags>, but takes a literal string instead
359 of a string/length pair.
361 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
362 Like C<sv_catpvn_nomg>, but takes a literal string instead of
363 a string/length pair.
365 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
366 Like C<sv_catpvn>, but takes a literal string instead of a
369 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
370 Like C<sv_catpvn_mg>, but takes a literal string instead of a
373 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
374 Like C<sv_setpvn>, but takes a literal string instead of a
377 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
378 Like C<sv_setpvn_mg>, but takes a literal string instead of a
381 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
382 Like C<sv_setref_pvn>, but takes a literal string instead of
383 a string/length pair.
385 =for apidoc_section $string
387 =for apidoc Ama|char*|savepvs|"literal string"
388 Like C<savepvn>, but takes a literal string instead of a
391 =for apidoc Ama|char*|savesharedpvs|"literal string"
392 A version of C<savepvs()> which allocates the duplicate string in memory
393 which is shared between threads.
395 =for apidoc_section $GV
397 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
398 Like C<gv_stashpvn>, but takes a literal string instead of a
401 =for apidoc_section $HV
403 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
404 Like C<hv_fetch>, but takes a literal string instead of a
407 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
408 Like C<hv_store>, but takes a literal string instead of a
410 and omits the hash parameter.
412 =for apidoc_section $lexer
414 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
416 Like L</lex_stuff_pvn>, but takes a literal string instead of
417 a string/length pair.
422 #define ASSERT_IS_LITERAL(s) ("" s "")
425 =for apidoc_section $string
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) ASSERT_IS_LITERAL(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
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 true or false. It is undefined behavior if either of the buffers
610 doesn't contain at least C<len> bytes.
612 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
613 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
614 C<l1> gives the number of bytes in C<s1>.
615 Returns true or false.
617 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
618 Test two buffers (which may contain embedded C<NUL> characters, to see if they
619 are not equal. The C<len> parameter indicates the number of bytes to compare.
620 Returns true or false. It is undefined behavior if either of the buffers
621 doesn't contain at least C<len> bytes.
623 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
624 Like L</memNE>, but the second string is a literal enclosed in double quotes,
625 C<l1> gives the number of bytes in C<s1>.
626 Returns true or false.
628 =for apidoc Am|bool|memCHRs|"list"|char c
629 Returns the position of the first occurence of the byte C<c> in the literal
630 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
631 treated as unsigned char. Thus this macro can be used to determine if C<c> is
632 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
633 is C<NUL> (and the set doesn't include C<NUL>).
637 New macros should use the following conventions for their names (which are
638 based on the underlying C library functions):
640 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
642 Each has two main parameters, string-like operands that are compared
643 against each other, as specified by the macro name. Some macros may
644 additionally have one or potentially even two length parameters. If a length
645 parameter applies to both string parameters, it will be positioned third;
646 otherwise any length parameter immediately follows the string parameter it
649 If the prefix to the name is 'str', the string parameter is a pointer to a C
650 language string. Such a string does not contain embedded NUL bytes; its
651 length may be unknown, but can be calculated by C<strlen()>, since it is
652 terminated by a NUL, which isn't included in its length.
654 The optional 'n' following 'str' means that there is a third parameter,
655 giving the maximum number of bytes to look at in each string. Even if both
656 strings are longer than the length parameter, those extra bytes will be
659 The 's' suffix means that the 2nd byte string parameter is a literal C
660 double-quoted string. Its length will automatically be calculated by the
661 macro, so no length parameter will ever be needed for it.
663 If the prefix is 'mem', the string parameters don't have to be C strings;
664 they may contain embedded NUL bytes, do not necessarily have a terminating
665 NUL, and their lengths can be known only through other means, which in
666 practice are additional parameter(s) passed to the function. All 'mem'
667 functions have at least one length parameter. Barring any 'l' or 's' suffix,
668 there is a single length parameter, in position 3, which applies to both
669 string parameters. The 's' suffix means, as described above, that the 2nd
670 string is a literal double-quoted C string (hence its length is calculated by
671 the macro, and the length parameter to the function applies just to the first
672 string parameter, and hence is positioned just after it). An 'l' suffix
673 means that the 2nd string parameter has its own length parameter, and the
674 signature will look like memFOOl(s1, l1, s2, l2).
676 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
677 substring of the 1st string. 'P' if present indicates that the substring
678 must be a "proper" one in tha mathematical sense that the first one must be
679 strictly larger than the 2nd.
684 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
685 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
686 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
687 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
688 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
689 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
691 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
692 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
694 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
695 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
697 /* memEQ and memNE where second comparand is a string constant */
698 #define memEQs(s1, l, s2) \
699 (((sizeof(s2)-1) == (l)) && memEQ((s1), ASSERT_IS_LITERAL(s2), (sizeof(s2)-1)))
700 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
702 /* Keep these private until we decide it was a good idea */
703 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
705 #define strBEGINs(s1,s2) (strncmp(s1,ASSERT_IS_LITERAL(s2), sizeof(s2)-1) == 0)
707 #define memBEGINs(s1, l, s2) \
708 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
709 && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1))
710 #define memBEGINPs(s1, l, s2) \
711 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
712 && memEQ(s1, ASSERT_IS_LITERAL(s2), sizeof(s2)-1))
713 #define memENDs(s1, l, s2) \
714 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
715 && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1))
716 #define memENDPs(s1, l, s2) \
717 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
718 && memEQ(s1 + (l) - (sizeof(s2) - 1), ASSERT_IS_LITERAL(s2), sizeof(s2)-1))
719 #endif /* End of making macros private */
721 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
722 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
723 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
724 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
726 #define memCHRs(s1,c) ((const char *) memchr(ASSERT_IS_LITERAL(s1) , c, sizeof(s1)-1))
731 * Unfortunately, the introduction of locales means that we
732 * can't trust isupper(), etc. to tell the truth. And when
733 * it comes to /\w+/ with tainting enabled, we *must* be able
734 * to trust our character classes.
736 * Therefore, the default tests in the text of Perl will be
737 * independent of locale. Any code that wants to depend on
738 * the current locale will use the tests that begin with "lc".
749 =head1 Character classification
750 This section is about functions (really macros) that classify characters
751 into types, such as punctuation versus alphabetic, etc. Most of these are
752 analogous to regular expression character classes. (See
753 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
754 each class. (Not all macros have all variants; each item below lists the
755 ones valid for it.) None are affected by C<use bytes>, and only the ones
756 with C<LC> in the name are affected by the current locale.
758 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
759 treating it as a code point, and returns a boolean as to whether or not the
760 character represented by it is (or on non-ASCII platforms, corresponds to) an
761 ASCII character in the named class based on platform, Unicode, and Perl rules.
762 If the input is a number that doesn't fit in an octet, FALSE is returned.
764 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
765 with no suffix C<"_A">. This variant is used to emphasize by its name that
766 only ASCII-range characters can return TRUE.
768 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
769 onto the platform. That is, the code points that are ASCII are unaffected,
770 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
771 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
772 true when called with the code point 0xDF, which is a word character in both
773 ASCII and EBCDIC (though it represents different characters in each).
774 If the input is a number that doesn't fit in an octet, FALSE is returned.
775 (Perl's documentation uses a colloquial definition of Latin-1, to include all
776 code points below 256.)
778 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
779 inputs below 256, but if the code point is larger than 255, Unicode rules are
780 used to determine if it is in the character class. For example,
781 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
782 WITH MACRON in Unicode, and is a word character.
784 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
785 but are used for UTF-8 encoded strings. The two forms are different names for
786 the same thing. Each call to one of these classifies the first character of
787 the string starting at C<p>. The second parameter, C<e>, points to anywhere in
788 the string beyond the first character, up to one byte past the end of the
789 entire string. Although both variants are identical, the suffix C<_safe> in
790 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
791 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
792 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
793 some way, the program may croak, or the function may return FALSE, at the
794 discretion of the implementation, and subject to change in future releases.
796 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
797 but the result is based on the current locale, which is what C<LC> in the name
798 stands for. If Perl can determine that the current locale is a UTF-8 locale,
799 it uses the published Unicode rules; otherwise, it uses the C library function
800 that gives the named classification. For example, C<isDIGIT_LC()> when not in
801 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
802 returned if the input won't fit into an octet. On some platforms where the C
803 library function is known to be defective, Perl changes its result to follow
804 the POSIX standard's rules.
806 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
807 than 256, but for larger ones it returns the Unicode classification of the code
810 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
811 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
812 are different names for the same thing. Each call to one of these classifies
813 the first character of the string starting at C<p>. The second parameter,
814 C<e>, points to anywhere in the string beyond the first character, up to one
815 byte past the end of the entire string. Although both variants are identical,
816 the suffix C<_safe> in one name emphasizes that it will not attempt to read
817 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
818 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
819 character is malformed in some way, the program may croak, or the function may
820 return FALSE, at the discretion of the implementation, and subject to change in
823 =for apidoc Am|bool|isALPHA|UV ch
824 =for apidoc_item ||isALPHA_A|UV ch
825 =for apidoc_item ||isALPHA_L1|UV ch
826 =for apidoc_item ||isALPHA_uvchr|UV ch
827 =for apidoc_item ||isALPHA_utf8_safe|U8 * s|U8 * end
828 =for apidoc_item ||isALPHA_utf8|U8 * s|U8 * end
829 =for apidoc_item ||isALPHA_LC|UV ch
830 =for apidoc_item ||isALPHA_LC_uvchr|UV ch
831 =for apidoc_item ||isALPHA_LC_utf8_safe|U8 * s| U8 *end
832 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
833 analogous to C<m/[[:alpha:]]/>.
834 See the L<top of this section|/Character classification> for an explanation of
839 Here and below, we add the prototypes of these macros for downstream programs
840 that would be interested in them, such as Devel::PPPort
842 =for apidoc Am|bool|isALPHANUMERIC|UV ch
843 =for apidoc_item ||isALPHANUMERIC_A|UV ch
844 =for apidoc_item ||isALPHANUMERIC_L1|UV ch
845 =for apidoc_item ||isALPHANUMERIC_uvchr|UV ch
846 =for apidoc_item ||isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
847 =for apidoc_item ||isALPHANUMERIC_utf8|U8 * s|U8 * end
848 =for apidoc_item ||isALPHANUMERIC_LC|UV ch
849 =for apidoc_item ||isALPHANUMERIC_LC_uvchr|UV ch
850 =for apidoc_item ||isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
851 =for apidoc_item ||isALNUMC|UV ch
852 =for apidoc_item ||isALNUMC_A|UV ch
853 =for apidoc_item ||isALNUMC_L1|UV ch
854 =for apidoc_item ||isALNUMC_LC|UV ch
855 =for apidoc_item ||isALNUMC_LC_uvchr|UV ch
856 Returns a boolean indicating whether the specified character is one of
857 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
858 See the L<top of this section|/Character classification> for an explanation of
861 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
862 this corresponds to the C language alphanumeric definition). Also
863 there are the variants
864 C<isALNUMC_A>, C<isALNUMC_L1>
865 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
867 =for apidoc Am|bool|isASCII|UV ch
868 =for apidoc_item ||isASCII_A|UV ch
869 =for apidoc_item ||isASCII_L1|UV ch
870 =for apidoc_item ||isASCII_uvchr|UV ch
871 =for apidoc_item ||isASCII_utf8_safe|U8 * s|U8 * end
872 =for apidoc_item ||isASCII_utf8|U8 * s|U8 * end
873 =for apidoc_item ||isASCII_LC|UV ch
874 =for apidoc_item ||isASCII_LC_uvchr|UV ch
875 =for apidoc_item ||isASCII_LC_utf8_safe|U8 * s| U8 *end
876 Returns a boolean indicating whether the specified character is one of the 128
877 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
878 On non-ASCII platforms, it returns TRUE iff this
879 character corresponds to an ASCII character. Variants C<isASCII_A()> and
880 C<isASCII_L1()> are identical to C<isASCII()>.
881 See the L<top of this section|/Character classification> for an explanation of
883 Note, however, that some platforms do not have the C library routine
884 C<isascii()>. In these cases, the variants whose names contain C<LC> are the
885 same as the corresponding ones without.
887 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
888 have the exact same representation (always a single byte) whether encoded in
889 UTF-8 or not), C<isASCII> will give the correct results when called with any
890 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
891 C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
893 =for apidoc Am|bool|isBLANK|UV ch
894 =for apidoc_item ||isBLANK_A|UV ch
895 =for apidoc_item ||isBLANK_L1|UV ch
896 =for apidoc_item ||isBLANK_uvchr|UV ch
897 =for apidoc_item ||isBLANK_utf8_safe|U8 * s|U8 * end
898 =for apidoc_item ||isBLANK_utf8|U8 * s|U8 * end
899 =for apidoc_item ||isBLANK_LC|UV ch
900 =for apidoc_item ||isBLANK_LC_uvchr|UV ch
901 =for apidoc_item ||isBLANK_LC_utf8_safe|U8 * s| U8 *end
902 Returns a boolean indicating whether the specified character is a
903 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
904 See the L<top of this section|/Character classification> for an explanation of
907 however, that some platforms do not have the C library routine
908 C<isblank()>. In these cases, the variants whose names contain C<LC> are
909 the same as the corresponding ones without.
911 =for apidoc Am|bool|isCNTRL|UV ch
912 =for apidoc_item ||isCNTRL_A|UV ch
913 =for apidoc_item ||isCNTRL_L1|UV ch
914 =for apidoc_item ||isCNTRL_uvchr|UV ch
915 =for apidoc_item ||isCNTRL_utf8_safe|U8 * s|U8 * end
916 =for apidoc_item ||isCNTRL_utf8|U8 * s|U8 * end
917 =for apidoc_item ||isCNTRL_LC|UV ch
918 =for apidoc_item ||isCNTRL_LC_uvchr|UV ch
919 =for apidoc_item ||isCNTRL_LC_utf8_safe|U8 * s| U8 *end
921 Returns a boolean indicating whether the specified character is a
922 control character, analogous to C<m/[[:cntrl:]]/>.
923 See the L<top of this section|/Character classification> for an explanation of
925 On EBCDIC platforms, you almost always want to use the C<isCNTRL_L1> variant.
927 =for apidoc Am|bool|isDIGIT|UV ch
928 =for apidoc_item ||isDIGIT_A|UV ch
929 =for apidoc_item ||isDIGIT_L1|UV ch
930 =for apidoc_item ||isDIGIT_uvchr|UV ch
931 =for apidoc_item ||isDIGIT_utf8_safe|U8 * s|U8 * end
932 =for apidoc_item ||isDIGIT_utf8|U8 * s|U8 * end
933 =for apidoc_item ||isDIGIT_LC|UV ch
934 =for apidoc_item ||isDIGIT_LC_uvchr|UV ch
935 =for apidoc_item ||isDIGIT_LC_utf8_safe|U8 * s| U8 *end
937 Returns a boolean indicating whether the specified character is a
938 digit, analogous to C<m/[[:digit:]]/>.
939 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
940 See the L<top of this section|/Character classification> for an explanation of
943 =for apidoc Am|bool|isGRAPH|UV ch
944 =for apidoc_item ||isGRAPH_A|UV ch
945 =for apidoc_item ||isGRAPH_L1|UV ch
946 =for apidoc_item ||isGRAPH_uvchr|UV ch
947 =for apidoc_item ||isGRAPH_utf8_safe|U8 * s|U8 * end
948 =for apidoc_item ||isGRAPH_utf8|U8 * s|U8 * end
949 =for apidoc_item ||isGRAPH_LC|UV ch
950 =for apidoc_item ||isGRAPH_LC_uvchr|UV ch
951 =for apidoc_item ||isGRAPH_LC_utf8_safe|U8 * s| U8 *end
952 Returns a boolean indicating whether the specified character is a
953 graphic character, analogous to C<m/[[:graph:]]/>.
954 See the L<top of this section|/Character classification> for an explanation of
957 =for apidoc Am|bool|isLOWER|UV ch
958 =for apidoc_item ||isLOWER_A|UV ch
959 =for apidoc_item ||isLOWER_L1|UV ch
960 =for apidoc_item ||isLOWER_uvchr|UV ch
961 =for apidoc_item ||isLOWER_utf8_safe|U8 * s|U8 * end
962 =for apidoc_item ||isLOWER_utf8|U8 * s|U8 * end
963 =for apidoc_item ||isLOWER_LC|UV ch
964 =for apidoc_item ||isLOWER_LC_uvchr|UV ch
965 =for apidoc_item ||isLOWER_LC_utf8_safe|U8 * s| U8 *end
966 Returns a boolean indicating whether the specified character is a
967 lowercase character, analogous to C<m/[[:lower:]]/>.
968 See the L<top of this section|/Character classification> for an explanation of
971 =for apidoc Am|bool|isOCTAL|UV ch
972 =for apidoc_item ||isOCTAL_A|UV ch
973 =for apidoc_item ||isOCTAL_L1|UV ch
974 Returns a boolean indicating whether the specified character is an
976 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
979 =for apidoc Am|bool|isPUNCT|UV ch
980 =for apidoc_item ||isPUNCT_A|UV ch
981 =for apidoc_item ||isPUNCT_L1|UV ch
982 =for apidoc_item ||isPUNCT_uvchr|UV ch
983 =for apidoc_item ||isPUNCT_utf8_safe|U8 * s|U8 * end
984 =for apidoc_item ||isPUNCT_utf8|U8 * s|U8 * end
985 =for apidoc_item ||isPUNCT_LC|UV ch
986 =for apidoc_item ||isPUNCT_LC_uvchr|UV ch
987 =for apidoc_item ||isPUNCT_LC_utf8_safe|U8 * s| U8 *end
988 Returns a boolean indicating whether the specified character is a
989 punctuation character, analogous to C<m/[[:punct:]]/>.
990 Note that the definition of what is punctuation isn't as
991 straightforward as one might desire. See L<perlrecharclass/POSIX Character
992 Classes> for details.
993 See the L<top of this section|/Character classification> for an explanation of
996 =for apidoc Am|bool|isSPACE|UV ch
997 =for apidoc_item ||isSPACE_A|UV ch
998 =for apidoc_item ||isSPACE_L1|UV ch
999 =for apidoc_item ||isSPACE_uvchr|UV ch
1000 =for apidoc_item ||isSPACE_utf8_safe|U8 * s|U8 * end
1001 =for apidoc_item ||isSPACE_utf8|U8 * s|U8 * end
1002 =for apidoc_item ||isSPACE_LC|UV ch
1003 =for apidoc_item ||isSPACE_LC_uvchr|UV ch
1004 =for apidoc_item ||isSPACE_LC_utf8_safe|U8 * s| U8 *end
1005 Returns a boolean indicating whether the specified character is a
1006 whitespace character. This is analogous
1007 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
1008 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
1009 locale forms of this macro (the ones with C<LC> in their names) matched
1010 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
1011 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
1012 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
1013 See the L<top of this section|/Character classification> for an explanation of
1016 =for apidoc Am|bool|isPSXSPC|UV ch
1017 =for apidoc_item ||isPSXSPC_A|UV ch
1018 =for apidoc_item ||isPSXSPC_L1|UV ch
1019 =for apidoc_item ||isPSXSPC_uvchr|UV ch
1020 =for apidoc_item ||isPSXSPC_utf8_safe|U8 * s|U8 * end
1021 =for apidoc_item ||isPSXSPC_utf8|U8 * s|U8 * end
1022 =for apidoc_item ||isPSXSPC_LC|UV ch
1023 =for apidoc_item ||isPSXSPC_LC_uvchr|UV ch
1024 =for apidoc_item ||isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
1025 (short for Posix Space)
1026 Starting in 5.18, this is identical in all its forms to the
1027 corresponding C<isSPACE()> macros.
1028 The locale forms of this macro are identical to their corresponding
1029 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1030 non-locale forms differ from their C<isSPACE()> forms only in that the
1031 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1032 Otherwise they are identical. Thus this macro is analogous to what
1033 C<m/[[:space:]]/> matches in a regular expression.
1034 See the L<top of this section|/Character classification> for an explanation of
1037 =for apidoc Am|bool|isUPPER|UV ch
1038 =for apidoc_item ||isUPPER_A|UV ch
1039 =for apidoc_item ||isUPPER_L1|UV ch
1040 =for apidoc_item ||isUPPER_uvchr|UV ch
1041 =for apidoc_item ||isUPPER_utf8_safe|U8 * s|U8 * end
1042 =for apidoc_item ||isUPPER_utf8|U8 * s|U8 * end
1043 =for apidoc_item ||isUPPER_LC|UV ch
1044 =for apidoc_item ||isUPPER_LC_uvchr|UV ch
1045 =for apidoc_item ||isUPPER_LC_utf8_safe|U8 * s| U8 *end
1046 Returns a boolean indicating whether the specified character is an
1047 uppercase character, analogous to C<m/[[:upper:]]/>.
1048 See the L<top of this section|/Character classification> for an explanation of
1051 =for apidoc Am|bool|isPRINT|UV ch
1052 =for apidoc_item ||isPRINT_A|UV ch
1053 =for apidoc_item ||isPRINT_L1|UV ch
1054 =for apidoc_item ||isPRINT_uvchr|UV ch
1055 =for apidoc_item ||isPRINT_utf8_safe|U8 * s|U8 * end
1056 =for apidoc_item ||isPRINT_utf8|U8 * s|U8 * end
1057 =for apidoc_item ||isPRINT_LC|UV ch
1058 =for apidoc_item ||isPRINT_LC_uvchr|UV ch
1059 =for apidoc_item ||isPRINT_LC_utf8_safe|U8 * s| U8 *end
1060 Returns a boolean indicating whether the specified character is a
1061 printable character, analogous to C<m/[[:print:]]/>.
1062 See the L<top of this section|/Character classification> for an explanation of
1065 =for apidoc Am|bool|isWORDCHAR|UV ch
1066 =for apidoc_item ||isWORDCHAR_A|UV ch
1067 =for apidoc_item ||isWORDCHAR_L1|UV ch
1068 =for apidoc_item ||isWORDCHAR_uvchr|UV ch
1069 =for apidoc_item ||isWORDCHAR_utf8_safe|U8 * s|U8 * end
1070 =for apidoc_item ||isWORDCHAR_utf8|U8 * s|U8 * end
1071 =for apidoc_item ||isWORDCHAR_LC|UV ch
1072 =for apidoc_item ||isWORDCHAR_LC_uvchr|UV ch
1073 =for apidoc_item ||isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1074 =for apidoc_item ||isALNUM|UV ch
1075 =for apidoc_item ||isALNUM_A|UV ch
1076 =for apidoc_item ||isALNUM_LC|UV ch
1077 =for apidoc_item ||isALNUM_LC_uvchr|UV ch
1078 Returns a boolean indicating whether the specified character is a character
1079 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1080 in a regular expression. A word character is an alphabetic character, a
1081 decimal digit, a connecting punctuation character (such as an underscore), or
1082 a "mark" character that attaches to one of those (like some sort of accent).
1083 C<isALNUM()> is a synonym provided for backward compatibility, even though a
1084 word character includes more than the standard C language meaning of
1086 See the L<top of this section|/Character classification> for an explanation of
1088 C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1089 C<isWORDCHAR_LC>, C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and
1090 C<isWORDCHAR_LC_utf8_safe> are also as described there, but additionally
1091 include the platform's native underscore.
1093 =for apidoc Am|bool|isXDIGIT|UV ch
1094 =for apidoc_item ||isXDIGIT_A|UV ch
1095 =for apidoc_item ||isXDIGIT_L1|UV ch
1096 =for apidoc_item ||isXDIGIT_uvchr|UV ch
1097 =for apidoc_item ||isXDIGIT_utf8_safe|U8 * s|U8 * end
1098 =for apidoc_item ||isXDIGIT_utf8|U8 * s|U8 * end
1099 =for apidoc_item ||isXDIGIT_LC|UV ch
1100 =for apidoc_item ||isXDIGIT_LC_uvchr|UV ch
1101 =for apidoc_item ||isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1102 Returns a boolean indicating whether the specified character is a hexadecimal
1103 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1104 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1105 See the L<top of this section|/Character classification> for an explanation of
1108 =for apidoc Am|bool|isIDFIRST|UV ch
1109 =for apidoc_item ||isIDFIRST_A|UV ch
1110 =for apidoc_item ||isIDFIRST_L1|UV ch
1111 =for apidoc_item ||isIDFIRST_uvchr|UV ch
1112 =for apidoc_item ||isIDFIRST_utf8_safe|U8 * s|U8 * end
1113 =for apidoc_item ||isIDFIRST_utf8|U8 * s|U8 * end
1114 =for apidoc_item ||isIDFIRST_LC|UV ch
1115 =for apidoc_item ||isIDFIRST_LC_uvchr|UV ch
1116 =for apidoc_item ||isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1117 Returns a boolean indicating whether the specified character can be the first
1118 character of an identifier. This is very close to, but not quite the same as
1119 the official Unicode property C<XID_Start>. The difference is that this
1120 returns true only if the input character also matches L</isWORDCHAR>.
1121 See the L<top of this section|/Character classification> for an explanation of
1124 =for apidoc Am|bool|isIDCONT|UV ch
1125 =for apidoc_item ||isIDCONT_A|UV ch
1126 =for apidoc_item ||isIDCONT_L1|UV ch
1127 =for apidoc_item ||isIDCONT_uvchr|UV ch
1128 =for apidoc_item ||isIDCONT_utf8_safe|U8 * s|U8 * end
1129 =for apidoc_item ||isIDCONT_utf8|U8 * s|U8 * end
1130 =for apidoc_item ||isIDCONT_LC|UV ch
1131 =for apidoc_item ||isIDCONT_LC_uvchr|UV ch
1132 =for apidoc_item ||isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1133 Returns a boolean indicating whether the specified character can be the
1134 second or succeeding character of an identifier. This is very close to, but
1135 not quite the same as the official Unicode property C<XID_Continue>. The
1136 difference is that this returns true only if the input character also matches
1137 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1138 an explanation of the variants.
1140 =for apidoc_section $numeric
1142 =for apidoc Am|U8|READ_XDIGIT|char str*
1143 Returns the value of an ASCII-range hex digit and advances the string pointer.
1144 Behaviour is only well defined when isXDIGIT(*str) is true.
1146 =head1 Character case changing
1147 Perl uses "full" Unicode case mappings. This means that converting a single
1148 character to another case may result in a sequence of more than one character.
1149 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1150 character sequence C<SS>. This presents some complications The lowercase of
1151 all characters in the range 0..255 is a single character, and thus
1152 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1153 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1154 an API that does allow every possible legal result to be returned.) Likewise
1155 no other function that is crippled by not being able to give the correct
1156 results for the full range of possible inputs has been implemented here.
1158 =for apidoc Am|UV|toUPPER|UV cp
1159 =for apidoc_item |UV|toUPPER_A|UV cp
1160 =for apidoc_item |UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1161 =for apidoc_item |UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1162 =for apidoc_item |UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1164 These all return the uppercase of a character. The differences are what domain
1165 they operate on, and whether the input is specified as a code point (those
1166 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1167 case, the code point to use is the first one in the buffer of UTF-8 encoded
1168 code points, delineated by the arguments S<C<p .. e - 1>>.
1170 C<toUPPER> and C<toUPPER_A> are synonyms of each other. They return the
1171 uppercase of any lowercase ASCII-range code point. All other inputs are
1172 returned unchanged. Since these are macros, the input type may be any integral
1173 one, and the output will occupy the same number of bits as the input.
1175 There is no C<toUPPER_L1> nor C<toUPPER_LATIN1> as the uppercase of some code
1176 points in the 0..255 range is above that range or consists of multiple
1177 characters. Instead use C<toUPPER_uvchr>.
1179 C<toUPPER_uvchr> returns the uppercase of any Unicode code point. The return
1180 value is identical to that of C<toUPPER_A> for input code points in the ASCII
1181 range. The uppercase of the vast majority of Unicode code points is the same
1182 as the code point itself. For these, and for code points above the legal
1183 Unicode maximum, this returns the input code point unchanged. It additionally
1184 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1185 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1186 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1188 NOTE: the uppercase of a code point may be more than one code point. The
1189 return value of this function is only the first of these. The entire uppercase
1190 is returned in C<s>. To determine if the result is more than a single code
1191 point, you can do something like this:
1193 uc = toUPPER_uvchr(cp, s, &len);
1194 if (len > UTF8SKIP(s)) { is multiple code points }
1195 else { is a single code point }
1197 C<toUPPER_utf8> and C<toUPPER_utf8_safe> are synonyms of each other. The only
1198 difference between these and C<toUPPER_uvchr> is that the source for these is
1199 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1200 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1201 buffer may certainly contain more than one code point; but only the first one
1202 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1203 malformed in some way, the program may croak, or the function may return the
1204 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1205 change in future releases.
1207 =for apidoc Am|UV|toFOLD|UV cp
1208 =for apidoc_item |UV|toFOLD_A|UV cp
1209 =for apidoc_item |UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1210 =for apidoc_item |UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1211 =for apidoc_item |UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1213 These all return the foldcase of a character. "foldcase" is an internal case
1214 for C</i> pattern matching. If the foldcase of character A and the foldcase of
1215 character B are the same, they match caselessly; otherwise they don't.
1217 The differences in the forms are what domain they operate on, and whether the
1218 input is specified as a code point (those forms with a C<cp> parameter) or as a
1219 UTF-8 string (the others). In the latter case, the code point to use is the
1220 first one in the buffer of UTF-8 encoded code points, delineated by the
1221 arguments S<C<p .. e - 1>>.
1223 C<toFOLD> and C<toFOLD_A> are synonyms of each other. They return the
1224 foldcase of any ASCII-range code point. In this range, the foldcase is
1225 identical to the lowercase. All other inputs are returned unchanged. Since
1226 these are macros, the input type may be any integral one, and the output will
1227 occupy the same number of bits as the input.
1229 There is no C<toFOLD_L1> nor C<toFOLD_LATIN1> as the foldcase of some code
1230 points in the 0..255 range is above that range or consists of multiple
1231 characters. Instead use C<toFOLD_uvchr>.
1233 C<toFOLD_uvchr> returns the foldcase of any Unicode code point. The return
1234 value is identical to that of C<toFOLD_A> for input code points in the ASCII
1235 range. The foldcase of the vast majority of Unicode code points is the same
1236 as the code point itself. For these, and for code points above the legal
1237 Unicode maximum, this returns the input code point unchanged. It additionally
1238 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1239 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1240 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1242 NOTE: the foldcase of a code point may be more than one code point. The
1243 return value of this function is only the first of these. The entire foldcase
1244 is returned in C<s>. To determine if the result is more than a single code
1245 point, you can do something like this:
1247 uc = toFOLD_uvchr(cp, s, &len);
1248 if (len > UTF8SKIP(s)) { is multiple code points }
1249 else { is a single code point }
1251 C<toFOLD_utf8> and C<toFOLD_utf8_safe> are synonyms of each other. The only
1252 difference between these and C<toFOLD_uvchr> is that the source for these is
1253 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1254 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1255 buffer may certainly contain more than one code point; but only the first one
1256 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1257 malformed in some way, the program may croak, or the function may return the
1258 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1259 change in future releases.
1261 =for apidoc Am|UV|toLOWER|UV cp
1262 =for apidoc_item |UV|toLOWER_A|UV cp
1263 =for apidoc_item |UV|toLOWER_L1|UV cp
1264 =for apidoc_item |UV|toLOWER_LATIN1|UV cp
1265 =for apidoc_item |UV|toLOWER_LC|UV cp
1266 =for apidoc_item |UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1267 =for apidoc_item |UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1268 =for apidoc_item |UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1270 These all return the lowercase of a character. The differences are what domain
1271 they operate on, and whether the input is specified as a code point (those
1272 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1273 case, the code point to use is the first one in the buffer of UTF-8 encoded
1274 code points, delineated by the arguments S<C<p .. e - 1>>.
1276 C<toLOWER> and C<toLOWER_A> are synonyms of each other. They return the
1277 lowercase of any uppercase ASCII-range code point. All other inputs are
1278 returned unchanged. Since these are macros, the input type may be any integral
1279 one, and the output will occupy the same number of bits as the input.
1281 C<toLOWER_L1> and C<toLOWER_LATIN1> are synonyms of each other. They behave
1282 identically as C<toLOWER> for ASCII-range input. But additionally will return
1283 the lowercase of any uppercase code point in the entire 0..255 range, assuming
1284 a Latin-1 encoding (or the EBCDIC equivalent on such platforms).
1286 C<toLOWER_LC> returns the lowercase of the input code point according to the
1287 rules of the current POSIX locale. Input code points outside the range 0..255
1288 are returned unchanged.
1290 C<toLOWER_uvchr> returns the lowercase of any Unicode code point. The return
1291 value is identical to that of C<toLOWER_L1> for input code points in the 0..255
1292 range. The lowercase of the vast majority of Unicode code points is the same
1293 as the code point itself. For these, and for code points above the legal
1294 Unicode maximum, this returns the input code point unchanged. It additionally
1295 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1296 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1297 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1299 NOTE: the lowercase of a code point may be more than one code point. The
1300 return value of this function is only the first of these. The entire lowercase
1301 is returned in C<s>. To determine if the result is more than a single code
1302 point, you can do something like this:
1304 uc = toLOWER_uvchr(cp, s, &len);
1305 if (len > UTF8SKIP(s)) { is multiple code points }
1306 else { is a single code point }
1308 C<toLOWER_utf8> and C<toLOWER_utf8_safe> are synonyms of each other. The only
1309 difference between these and C<toLOWER_uvchr> is that the source for these is
1310 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1311 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1312 buffer may certainly contain more than one code point; but only the first one
1313 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1314 malformed in some way, the program may croak, or the function may return the
1315 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1316 change in future releases.
1318 =for apidoc Am|UV|toTITLE|UV cp
1319 =for apidoc_item |UV|toTITLE_A|UV cp
1320 =for apidoc_item |UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1321 =for apidoc_item |UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1322 =for apidoc_item |UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1324 These all return the titlecase of a character. The differences are what domain
1325 they operate on, and whether the input is specified as a code point (those
1326 forms with a C<cp> parameter) or as a UTF-8 string (the others). In the latter
1327 case, the code point to use is the first one in the buffer of UTF-8 encoded
1328 code points, delineated by the arguments S<C<p .. e - 1>>.
1330 C<toTITLE> and C<toTITLE_A> are synonyms of each other. They return the
1331 titlecase of any lowercase ASCII-range code point. In this range, the
1332 titlecase is identical to the uppercase. All other inputs are returned
1333 unchanged. Since these are macros, the input type may be any integral one, and
1334 the output will occupy the same number of bits as the input.
1336 There is no C<toTITLE_L1> nor C<toTITLE_LATIN1> as the titlecase of some code
1337 points in the 0..255 range is above that range or consists of multiple
1338 characters. Instead use C<toTITLE_uvchr>.
1340 C<toTITLE_uvchr> returns the titlecase of any Unicode code point. The return
1341 value is identical to that of C<toTITLE_A> for input code points in the ASCII
1342 range. The titlecase of the vast majority of Unicode code points is the same
1343 as the code point itself. For these, and for code points above the legal
1344 Unicode maximum, this returns the input code point unchanged. It additionally
1345 stores the UTF-8 of the result into the buffer beginning at C<s>, and its
1346 length in bytes into C<*lenp>. The caller must have made C<s> large enough to
1347 contain at least C<UTF8_MAXBYTES_CASE+1> bytes to avoid possible overflow.
1349 NOTE: the titlecase of a code point may be more than one code point. The
1350 return value of this function is only the first of these. The entire titlecase
1351 is returned in C<s>. To determine if the result is more than a single code
1352 point, you can do something like this:
1354 uc = toTITLE_uvchr(cp, s, &len);
1355 if (len > UTF8SKIP(s)) { is multiple code points }
1356 else { is a single code point }
1358 C<toTITLE_utf8> and C<toTITLE_utf8_safe> are synonyms of each other. The only
1359 difference between these and C<toTITLE_uvchr> is that the source for these is
1360 encoded in UTF-8, instead of being a code point. It is passed as a buffer
1361 starting at C<p>, with C<e> pointing to one byte beyond its end. The C<p>
1362 buffer may certainly contain more than one code point; but only the first one
1363 (up through S<C<e - 1>>) is examined. If the UTF-8 for the input character is
1364 malformed in some way, the program may croak, or the function may return the
1365 REPLACEMENT CHARACTER, at the discretion of the implementation, and subject to
1366 change in future releases.
1370 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1371 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1372 and aren't general purpose as they don't work on U+DF, and assert against that.
1374 Note that these macros are repeated in Devel::PPPort, so should also be
1375 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1380 void below because that's the best fit, and works for Devel::PPPort
1381 =for apidoc_section $integer
1382 =for apidoc AyT||WIDEST_UTYPE
1384 Yields the widest unsigned integer type on the platform, currently either
1385 C<U32> or C<U64>. This can be used in declarations such as
1391 my_uv = (WIDEST_UTYPE) val;
1397 # define WIDEST_UTYPE U64
1399 # define WIDEST_UTYPE U32
1402 /* Where there could be some confusion, use this as a static assert in macros
1403 * to make sure that a parameter isn't a pointer. But some compilers can't
1404 * handle this. The only one known so far that doesn't is gcc 3.3.6; the check
1405 * below isn't thorough for such an old compiler, so may have to be revised if
1406 * experience so dictates. */
1407 #if ! PERL_IS_GCC || PERL_GCC_VERSION_GT(3,3,6)
1408 # define ASSERT_NOT_PTR(x) ((x) | 0)
1410 # define ASSERT_NOT_PTR(x) (x)
1413 /* Likewise, this is effectively a static assert to be used to guarantee the
1414 * parameter is a pointer */
1415 #define ASSERT_IS_PTR(x) (__ASSERT_(sizeof(*(x))) (x))
1417 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1418 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1419 * bits of information are lost even on a 64-bit machine, but to get the
1420 * compiler to optimize it out if possible. This is because Configure makes
1421 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1422 * sizeof() guarantees that this evaluates to a constant true at compile time.
1424 * For Coverity, be always true, because otherwise Coverity thinks
1425 * it finds several expressions that are always true, independent
1426 * of operands. Well, they are, but that is kind of the point.
1428 #ifndef __COVERITY__
1429 /* The '| 0' part in ASSERT_NOT_PTR ensures a compiler error if c is not
1430 * integer (like e.g., a pointer) */
1431 # define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1432 || (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) >> 8) == 0)
1434 # define FITS_IN_8_BITS(c) (1)
1437 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
1438 * Written this way so that after optimization, only one conditional test is
1439 * needed. (The NV casts stop any warnings about comparison always being true
1440 * if called with an unsigned. The cast preserves the sign, which is all we
1442 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1443 __ASSERT_((NV) (n) >= 0) \
1444 withinCOUNT_KNOWN_VALID_((c), (l), (n)))
1446 /* For internal use only, this can be used in places where it is known that the
1447 * parameters to withinCOUNT() are valid, to avoid the asserts. For example,
1448 * inRANGE() below, calls this several times, but does all the necessary
1449 * asserts itself, once. The reason that this is necessary is that the
1450 * duplicate asserts were exceeding the internal limits of some compilers */
1451 #define withinCOUNT_KNOWN_VALID_(c, l, n) \
1452 ((((WIDEST_UTYPE) (c)) - ASSERT_NOT_PTR(l)) \
1453 <= ((WIDEST_UTYPE) ASSERT_NOT_PTR(n)))
1455 /* Returns true if c is in the range l..u, where 'l' is non-negative
1456 * Written this way so that after optimization, only one conditional test is
1458 #define inRANGE(c, l, u) (__ASSERT_((NV) (l) >= 0) __ASSERT_((u) >= (l)) \
1459 ( (sizeof(c) == sizeof(U8)) ? inRANGE_helper_(U8, (c), (l), ((u))) \
1460 : (sizeof(c) == sizeof(U16)) ? inRANGE_helper_(U16,(c), (l), ((u))) \
1461 : (sizeof(c) == sizeof(U32)) ? inRANGE_helper_(U32,(c), (l), ((u))) \
1462 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1463 inRANGE_helper_(WIDEST_UTYPE,(c), (l), ((u))))))
1465 /* For internal use, this is used by machine-generated code which generates
1466 * known valid calls, with a known sizeof(). This avoids the extra code and
1467 * asserts that were exceeding internal limits of some compilers. */
1468 #define inRANGE_helper_(cast, c, l, u) \
1469 withinCOUNT_KNOWN_VALID_(((cast) (c)), (l), ((u) - (l)))
1472 # ifndef _ALL_SOURCE
1473 /* The native libc isascii() et.al. functions return the wrong results
1474 * on at least z/OS unless this is defined. */
1475 # error _ALL_SOURCE should probably be defined
1478 /* There is a simple definition of ASCII for ASCII platforms. But the
1479 * EBCDIC one isn't so simple, so is defined using table look-up like the
1480 * other macros below.
1482 * The cast here is used instead of '(c) >= 0', because some compilers emit
1483 * a warning that that test is always true when the parameter is an
1484 * unsigned type. khw supposes that it could be written as
1485 * && ((c) == '\0' || (c) > 0)
1486 * to avoid the message, but the cast will likely avoid extra branches even
1487 * with stupid compilers. */
1488 # define isASCII(c) (((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) < 128)
1491 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1492 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1493 * can be ignored. If the rest match '0', we have an octal */
1494 #define isOCTAL_A(c) ((((WIDEST_UTYPE) ASSERT_NOT_PTR(c)) & ~7) == '0')
1496 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1498 /* Character class numbers. For internal core Perl use only. The ones less
1499 * than 32 are used in PL_charclass[] and the ones up through the one that
1500 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1501 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1502 * their actual definitions are here. If that file has a name not used here,
1505 * The first group of these is ordered in what I (khw) estimate to be the
1506 * frequency of their use. This gives a slight edge to exiting a loop earlier
1507 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1508 * real Posix character class, and some (small) inefficiencies in regular
1509 * expression handling would be introduced by putting it in the middle of those
1510 * that are. Also, cntrl and ascii come after the others as it may be useful
1511 * to group these which have no members that match above Latin1, (or above
1512 * ASCII in the latter case) */
1514 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1515 # define _CC_DIGIT 1 /* \d and [:digit:] */
1516 # define _CC_ALPHA 2 /* [:alpha:] */
1517 # define _CC_LOWER 3 /* [:lower:] */
1518 # define _CC_UPPER 4 /* [:upper:] */
1519 # define _CC_PUNCT 5 /* [:punct:] */
1520 # define _CC_PRINT 6 /* [:print:] */
1521 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1522 # define _CC_GRAPH 8 /* [:graph:] */
1523 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1524 # define _CC_SPACE 10 /* \s, [:space:] */
1525 # define _CC_BLANK 11 /* [:blank:] */
1526 # define _CC_XDIGIT 12 /* [:xdigit:] */
1527 # define _CC_CNTRL 13 /* [:cntrl:] */
1528 # define _CC_ASCII 14 /* [:ascii:] */
1529 # define _CC_VERTSPACE 15 /* \v */
1531 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1533 /* The members of the third group below do not need to be coordinated with data
1534 * structures in regcomp.[ch] and regexec.c. */
1535 # define _CC_IDFIRST 16
1536 # define _CC_CHARNAME_CONT 17
1537 # define _CC_NONLATIN1_FOLD 18
1538 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1539 # define _CC_QUOTEMETA 20
1540 # define _CC_NON_FINAL_FOLD 21
1541 # define _CC_IS_IN_SOME_FOLD 22
1542 # define _CC_BINDIGIT 23
1543 # define _CC_OCTDIGIT 24
1544 # define _CC_MNEMONIC_CNTRL 25
1547 * If more bits are needed, one could add a second word for non-64bit
1548 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1549 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1550 * is used only for optimization (as of this writing), and differs in the
1551 * Latin1 range from the ALPHA bit only in two relatively unimportant
1552 * characters: the masculine and feminine ordinal indicators, so removing it
1553 * would just cause /i regexes which match them to run less efficiently.
1554 * Similarly the EBCDIC-only bits are used just for speed, and could be
1555 * replaced by other means */
1557 #if defined(PERL_CORE) || defined(PERL_EXT)
1558 /* An enum version of the character class numbers, to help compilers
1561 _CC_ENUM_ALPHA = _CC_ALPHA,
1562 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1563 _CC_ENUM_ASCII = _CC_ASCII,
1564 _CC_ENUM_BLANK = _CC_BLANK,
1565 _CC_ENUM_CASED = _CC_CASED,
1566 _CC_ENUM_CNTRL = _CC_CNTRL,
1567 _CC_ENUM_DIGIT = _CC_DIGIT,
1568 _CC_ENUM_GRAPH = _CC_GRAPH,
1569 _CC_ENUM_LOWER = _CC_LOWER,
1570 _CC_ENUM_PRINT = _CC_PRINT,
1571 _CC_ENUM_PUNCT = _CC_PUNCT,
1572 _CC_ENUM_SPACE = _CC_SPACE,
1573 _CC_ENUM_UPPER = _CC_UPPER,
1574 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1575 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1576 _CC_ENUM_XDIGIT = _CC_XDIGIT
1577 } _char_class_number;
1580 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1584 EXTCONST U32 PL_charclass[] = {
1585 # include "l1_char_class_tab.h"
1588 # else /* ! DOINIT */
1589 EXTCONST U32 PL_charclass[];
1593 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1594 # define _CC_mask(classnum) (1U << (classnum))
1596 /* For internal core Perl use only: the base macro for defining macros like
1598 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1599 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1601 /* The mask for the _A versions of the macros; it just adds in the bit for
1603 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1605 /* For internal core Perl use only: the base macro for defining macros like
1606 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1607 * the ASCII bit are present */
1608 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1609 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1610 == _CC_mask_A(classnum)))
1612 /* On ASCII platforms certain classes form a single range. It's faster to
1613 * special case these. isDIGIT is a single range on all platforms */
1615 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1616 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1617 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1618 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1619 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1621 /* By folding the upper and lowercase, we can use a single range */
1622 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1623 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1624 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1625 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1626 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1628 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1629 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1630 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1631 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1632 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1633 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1634 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1635 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1637 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1638 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1639 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1640 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1642 /* continuation character for legal NAME in \N{NAME} */
1643 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1645 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1646 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1647 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1648 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1649 # define isPSXSPC_L1(c) isSPACE_L1(c)
1650 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1651 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1652 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1653 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1654 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1657 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1660 /* Participates in a single-character fold with a character above 255 */
1661 # if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
1662 # define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(c) \
1663 (( ! cBOOL(FITS_IN_8_BITS(c))) \
1664 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1666 # define IS_NON_FINAL_FOLD(c) _generic_isCC(c, _CC_NON_FINAL_FOLD)
1667 # define IS_IN_SOME_FOLD_L1(c) _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1670 /* Like the above, but also can be part of a multi-char fold */
1671 # define HAS_NONLATIN1_FOLD_CLOSURE(c) \
1672 ( (! cBOOL(FITS_IN_8_BITS(c))) \
1673 || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
1675 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1677 /* is c a control character for which we have a mnemonic? */
1678 # if defined(PERL_CORE) || defined(PERL_EXT)
1679 # define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1681 #else /* else we don't have perl.h H_PERL */
1683 /* If we don't have perl.h, we are compiling a utility program. Below we
1684 * hard-code various macro definitions that wouldn't otherwise be available
1685 * to it. Most are coded based on first principles. These are written to
1686 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1687 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1688 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1689 # define isSPACE_A(c) (isBLANK_A(c) \
1694 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1695 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1696 * performance (if optimization isn't on), and allow the same code to be
1697 * used for both platform types */
1698 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1699 || inRANGE((c), 'j', 'r') \
1700 || inRANGE((c), 's', 'z')
1701 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1702 || inRANGE((c), 'J', 'R') \
1703 || inRANGE((c), 'S', 'Z')
1704 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1705 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1706 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1707 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1708 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1709 || inRANGE((c), 'a', 'f') \
1710 || inRANGE((c), 'A', 'F')
1711 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1712 || (c) == '#' || (c) == '$' || (c) == '%' \
1713 || (c) == '&' || (c) == '\'' || (c) == '(' \
1714 || (c) == ')' || (c) == '*' || (c) == '+' \
1715 || (c) == ',' || (c) == '.' || (c) == '/' \
1716 || (c) == ':' || (c) == ';' || (c) == '<' \
1717 || (c) == '=' || (c) == '>' || (c) == '?' \
1718 || (c) == '@' || (c) == '[' || (c) == '\\' \
1719 || (c) == ']' || (c) == '^' || (c) == '_' \
1720 || (c) == '`' || (c) == '{' || (c) == '|' \
1721 || (c) == '}' || (c) == '~')
1722 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1723 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1726 /* The below is accurate for the 3 EBCDIC code pages traditionally
1727 * supported by perl. The only difference between them in the controls
1728 * is the position of \n, and that is represented symbolically below */
1729 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1730 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1731 || (c) == '\t' || (c) == '\v' \
1732 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1733 || (c) == 7F /* U+7F DEL */ \
1734 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1736 || (c) == 0x18 /* U+18 CAN */ \
1737 || (c) == 0x19 /* U+19 EOM */ \
1738 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1739 || (c) == 0x26 /* U+17 ETB */ \
1740 || (c) == 0x27 /* U+1B ESC */ \
1741 || (c) == 0x2D /* U+05 ENQ */ \
1742 || (c) == 0x2E /* U+06 ACK */ \
1743 || (c) == 0x32 /* U+16 SYN */ \
1744 || (c) == 0x37 /* U+04 EOT */ \
1745 || (c) == 0x3C /* U+14 DC4 */ \
1746 || (c) == 0x3D /* U+15 NAK */ \
1747 || (c) == 0x3F)/* U+1A SUB */
1748 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1749 # else /* isASCII is already defined for ASCII platforms, so can use that to
1751 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1754 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1755 * during debugging, and it seems best to keep them. We may be called
1756 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1757 * do anything anyway, so make it not a problem */
1758 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1759 # define NATIVE_TO_LATIN1(ch) (ch)
1761 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1762 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1763 # define isBLANK_L1(c) (isBLANK_A(c) \
1764 || (FITS_IN_8_BITS(c) \
1765 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1766 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1767 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1768 # define isLOWER_L1(c) (isLOWER_A(c) \
1769 || (FITS_IN_8_BITS(c) \
1770 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1771 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1772 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1773 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1774 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1775 # define isPRINT_L1(c) (isPRINT_A(c) \
1776 || (FITS_IN_8_BITS(c) \
1777 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1778 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1779 || (FITS_IN_8_BITS(c) \
1780 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1781 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1782 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1783 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1784 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1785 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1786 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1787 # define isSPACE_L1(c) (isSPACE_A(c) \
1788 || (FITS_IN_8_BITS(c) \
1789 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1790 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1791 # define isUPPER_L1(c) (isUPPER_A(c) \
1792 || (FITS_IN_8_BITS(c) \
1793 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1795 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1796 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1797 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1798 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1803 /* The following are not fully accurate in the above-ASCII range. I (khw)
1804 * don't think it's necessary to be so for the purposes where this gets
1806 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1807 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1809 /* And these aren't accurate at all. They are useful only for above
1810 * Latin1, which utilities and bootstrapping don't deal with */
1811 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1812 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1813 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1815 /* Many of the macros later in this file are defined in terms of these. By
1816 * implementing them with a function, which converts the class number into
1817 * a call to the desired macro, all of the later ones work. However, that
1818 * function won't be actually defined when building a utility program (no
1819 * perl.h), and so a compiler error will be generated if one is attempted
1820 * to be used. And the above-Latin1 code points require Unicode tables to
1821 * be present, something unlikely to be the case when bootstrapping */
1822 # define _generic_isCC(c, classnum) \
1823 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1824 # define _generic_isCC_A(c, classnum) \
1825 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1826 #endif /* End of no perl.h H_PERL */
1828 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1829 #define isALPHA(c) isALPHA_A(c)
1830 #define isASCII_A(c) isASCII(c)
1831 #define isASCII_L1(c) isASCII(c)
1832 #define isBLANK(c) isBLANK_A(c)
1833 #define isCNTRL(c) isCNTRL_A(c)
1834 #define isDIGIT(c) isDIGIT_A(c)
1835 #define isGRAPH(c) isGRAPH_A(c)
1836 #define isIDFIRST(c) isIDFIRST_A(c)
1837 #define isLOWER(c) isLOWER_A(c)
1838 #define isPRINT(c) isPRINT_A(c)
1839 #define isPSXSPC_A(c) isSPACE_A(c)
1840 #define isPSXSPC(c) isPSXSPC_A(c)
1841 #define isPSXSPC_L1(c) isSPACE_L1(c)
1842 #define isPUNCT(c) isPUNCT_A(c)
1843 #define isSPACE(c) isSPACE_A(c)
1844 #define isUPPER(c) isUPPER_A(c)
1845 #define isWORDCHAR(c) isWORDCHAR_A(c)
1846 #define isXDIGIT(c) isXDIGIT_A(c)
1848 /* ASCII casing. These could also be written as
1849 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1850 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1851 which uses table lookup and mask instead of subtraction. (This would
1852 work because the _MOD does not apply in the ASCII range).
1854 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1855 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1856 there non-ASCII invariants, and all of them are controls.) */
1857 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1858 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1860 /* In the ASCII range, these are equivalent to what they're here defined to be.
1861 * But by creating these definitions, other code doesn't have to be aware of
1862 * this detail. Actually this works for all UTF-8 invariants, not just the
1863 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1864 #define toFOLD(c) toLOWER(c)
1865 #define toTITLE(c) toUPPER(c)
1867 #define toLOWER_A(c) toLOWER(c)
1868 #define toUPPER_A(c) toUPPER(c)
1869 #define toFOLD_A(c) toFOLD(c)
1870 #define toTITLE_A(c) toTITLE(c)
1872 /* Use table lookup for speed; returns the input itself if is out-of-range */
1873 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1875 : PL_latin1_lc[ (U8) (c) ])
1876 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1878 /* Modified uc. Is correct uc except for three non-ascii chars which are
1879 * all mapped to one of them, and these need special handling; returns the
1880 * input itself if is out-of-range */
1881 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1883 : PL_mod_latin1_uc[ (U8) (c) ])
1884 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1886 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1888 /* For internal core Perl use only: the base macro for defining macros like
1889 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1890 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1891 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1892 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1893 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1894 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1895 * code points under locale rules, use the macros based on _generic_LC_uvchr
1896 * instead of this. */
1897 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1898 (! FITS_IN_8_BITS(c) \
1900 : IN_UTF8_CTYPE_LOCALE \
1901 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1904 /* For internal core Perl use only: a helper macro for defining macros like
1905 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1906 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1907 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1908 * _generic_LC_base for more info */
1909 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1910 _generic_LC_base(c,utf8_locale_classnum, \
1911 non_utf8_func( (_LC_CAST) (c)))
1913 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1914 * 'c' is the platform's native underscore character */
1915 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1916 _generic_LC_base(c, utf8_locale_classnum, \
1917 (non_utf8_func( (_LC_CAST) (c)) \
1918 || (char)(c) == '_'))
1920 /* These next three are also for internal core Perl use only: case-change
1921 * helper macros. The reason for using the PL_latin arrays is in case the
1922 * system function is defective; it ensures uniform results that conform to the
1923 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1924 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1926 : (IN_UTF8_CTYPE_LOCALE) \
1927 ? PL_latin1_lc[ (U8) (c) ] \
1928 : (cast)function((cast)(c)))
1930 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1931 * returns a single value, so can't adequately return the upper case of LATIN
1932 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1933 * values "SS"); instead it asserts against that under DEBUGGING, and
1934 * otherwise returns its input. It does not handle the anomalies in UTF-8
1935 * Turkic locales. */
1936 #define _generic_toUPPER_LC(c, function, cast) \
1937 (! FITS_IN_8_BITS(c) \
1939 : ((! IN_UTF8_CTYPE_LOCALE) \
1940 ? (cast)function((cast)(c)) \
1941 : ((((U8)(c)) == MICRO_SIGN) \
1942 ? GREEK_CAPITAL_LETTER_MU \
1943 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1944 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1945 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1946 ? (__ASSERT_(0) (c)) \
1947 : PL_mod_latin1_uc[ (U8) (c) ])))))
1949 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1950 * returns a single value, so can't adequately return the fold case of LATIN
1951 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1952 * values "ss"); instead it asserts against that under DEBUGGING, and
1953 * otherwise returns its input. It does not handle the anomalies in UTF-8
1955 #define _generic_toFOLD_LC(c, function, cast) \
1956 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1957 ? GREEK_SMALL_LETTER_MU \
1958 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1959 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1960 _generic_toLOWER_LC(c, function, cast)))
1962 /* Use the libc versions for these if available. */
1963 #if defined(HAS_ISASCII)
1964 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1966 # define isASCII_LC(c) isASCII(c)
1969 #if defined(HAS_ISBLANK)
1970 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1971 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1972 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1978 /* The Windows functions don't bother to follow the POSIX standard, which
1979 * for example says that something can't both be a printable and a control.
1980 * But Windows treats the \t control as a printable, and does such things
1981 * as making superscripts into both digits and punctuation. This tames
1982 * these flaws by assuming that the definitions of both controls and space
1983 * are correct, and then making sure that other definitions don't have
1984 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1985 * Not all possible weirdnesses are checked for, just the ones that were
1986 * detected on actual Microsoft code pages */
1988 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1989 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1991 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1992 && isALPHANUMERIC_LC(c))
1993 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1995 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1996 isALPHANUMERIC_LC(c))
1997 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1998 # define isIDFIRST_LC(c) (((c) == '_') \
1999 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
2000 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
2001 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
2002 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
2003 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
2004 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
2005 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
2006 && isALPHANUMERIC_LC(c))
2008 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
2009 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
2010 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
2012 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
2013 /* For most other platforms */
2015 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
2016 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
2017 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
2018 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
2019 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
2020 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
2021 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
2022 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
2023 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
2024 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
2025 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
2026 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
2027 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
2030 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
2031 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
2032 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
2034 #else /* The final fallback position */
2036 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
2037 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
2038 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
2039 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
2040 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
2041 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
2042 # define isLOWER_LC(c) (isascii(c) && islower(c))
2043 # define isPRINT_LC(c) (isascii(c) && isprint(c))
2044 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
2045 # define isSPACE_LC(c) (isascii(c) && isspace(c))
2046 # define isUPPER_LC(c) (isascii(c) && isupper(c))
2047 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
2048 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
2050 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
2051 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
2052 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
2056 #define isIDCONT(c) isWORDCHAR(c)
2057 #define isIDCONT_A(c) isWORDCHAR_A(c)
2058 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
2059 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
2060 #define isPSXSPC_LC(c) isSPACE_LC(c)
2062 /* For internal core Perl use only: the base macros for defining macros like
2063 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
2064 * number defined earlier in this file. _generic_uvchr() is used for POSIX
2065 * classes where there is a macro or function 'above_latin1' that takes the
2066 * single argument 'c' and returns the desired value. These exist for those
2067 * classes which have simple definitions, avoiding the overhead of an inversion
2068 * list binary search. _generic_invlist_uvchr() can be used
2069 * for classes where that overhead is faster than a direct lookup.
2070 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
2071 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2072 * have duplicate checks here, so could create versions of the macros that
2073 * don't, but experiments show that gcc optimizes them out anyway. */
2075 /* Note that all ignore 'use bytes' */
2076 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2077 ? _generic_isCC(c, classnum) \
2079 #define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
2080 ? _generic_isCC(c, classnum) \
2081 : _is_uni_FOO(classnum, c))
2082 #define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2083 #define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
2084 #define isASCII_uvchr(c) isASCII(c)
2085 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2086 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2087 #define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2088 #define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
2089 #define isIDCONT_uvchr(c) \
2090 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2091 #define isIDFIRST_uvchr(c) \
2092 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2093 #define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2094 #define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
2096 #define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
2097 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2098 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2100 #define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
2101 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2102 #define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
2103 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2105 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2106 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2107 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2108 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2110 /* For backwards compatibility, even though '_uni' should mean official Unicode
2111 * code points, in Perl it means native for those below 256 */
2112 #define isALPHA_uni(c) isALPHA_uvchr(c)
2113 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2114 #define isASCII_uni(c) isASCII_uvchr(c)
2115 #define isBLANK_uni(c) isBLANK_uvchr(c)
2116 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
2117 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
2118 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
2119 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
2120 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2121 #define isLOWER_uni(c) isLOWER_uvchr(c)
2122 #define isPRINT_uni(c) isPRINT_uvchr(c)
2123 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
2124 #define isSPACE_uni(c) isSPACE_uvchr(c)
2125 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2126 #define isUPPER_uni(c) isUPPER_uvchr(c)
2127 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
2128 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2129 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2130 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2131 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2132 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2133 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2135 /* For internal core Perl use only: the base macros for defining macros like
2136 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2137 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2138 * simple class definitions; the other for more complex. These are like
2139 * _generic_uvchr, so see it for more info. */
2140 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2141 (c < 256 ? latin1(c) : above_latin1(c))
2142 #define _generic_LC_invlist_uvchr(latin1, classnum, c) \
2143 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2145 #define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2146 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
2147 _CC_ALPHANUMERIC, c)
2148 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2149 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2150 is_HORIZWS_cp_high, c)
2151 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2152 #define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2153 #define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2154 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2155 _is_uni_perl_idcont, c)
2156 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2157 _is_uni_perl_idstart, c)
2158 #define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2159 #define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
2160 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2161 #define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2162 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2163 is_XPERLSPACE_cp_high, c)
2164 #define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2165 #define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
2167 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2168 is_XDIGIT_cp_high, c)
2170 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2172 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2173 * they don't otherwise go out of their way to look for malformed UTF-8. If
2174 * they can return accurate results without knowing if the input is otherwise
2175 * malformed, they do so. For example isASCII is accurate in spite of any
2176 * non-length malformations because it looks only at a single byte. Likewise
2177 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2178 * variant ones return FALSE. But, if the input has to be well-formed in order
2179 * for the results to be accurate, the macros will test and if malformed will
2180 * call a routine to die
2182 * Except for toke.c, the macros do assume that e > p, asserting that on
2183 * DEBUGGING builds. Much code that calls these depends on this being true,
2184 * for other reasons. toke.c is treated specially as using the regular
2185 * assertion breaks it in many ways. All strings that these operate on there
2186 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2187 * bunch of code in toke.c assumes that this is true, so the assertion allows
2189 #ifdef PERL_IN_TOKE_C
2190 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2192 # define _utf8_safe_assert(p,e) ((e) > (p))
2195 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2196 ((! _utf8_safe_assert(p, e)) \
2197 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2198 : (UTF8_IS_INVARIANT(*(p))) \
2199 ? _generic_isCC(*(p), classnum) \
2200 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2201 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2202 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2204 : (_force_out_malformed_utf8_message( \
2205 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2207 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2208 * 'above_latin1' can be a macro */
2209 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2210 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2211 #define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2212 _generic_utf8_safe(classnum, p, e, \
2213 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2214 ? (_force_out_malformed_utf8_message( \
2215 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2217 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2218 * 'above_latin1' parameter */
2219 #define _generic_invlist_utf8_safe(classnum, p, e) \
2220 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
2222 /* Like the above, but should be used only when it is known that there are no
2223 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2224 * class is TRUE for. Hence it can skip the tests for this range.
2225 * 'above_latin1' should include its arguments */
2226 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2227 (__ASSERT_(_utf8_safe_assert(p, e)) \
2229 ? _generic_isCC(*(p), classnum) \
2230 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2231 ? 0 /* Note that doesn't check validity for latin1 */ \
2235 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2236 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2237 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2238 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2239 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2240 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2241 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2242 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2243 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2244 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2245 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2246 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2247 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2248 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2249 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2250 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2251 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2252 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
2254 #define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
2255 #define isALPHANUMERIC_utf8_safe(p, e) \
2256 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
2257 #define isASCII_utf8_safe(p, e) \
2258 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2260 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2261 #define isBLANK_utf8_safe(p, e) \
2262 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2265 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2266 * more efficient macro instead of the more general one */
2267 # define isCNTRL_utf8_safe(p, e) \
2268 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2270 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2273 #define isDIGIT_utf8_safe(p, e) \
2274 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2275 _is_utf8_FOO(_CC_DIGIT, p, e))
2276 #define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
2277 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2278 _is_utf8_perl_idcont, p, e)
2280 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2281 * IDFIRST is an alnum. See
2282 * https://github.com/Perl/perl5/issues/10275 for more detail than you
2283 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2284 * This used to be not the XID version, but we decided to go with the more
2285 * modern Unicode definition */
2286 #define isIDFIRST_utf8_safe(p, e) \
2287 _generic_func_utf8_safe(_CC_IDFIRST, \
2288 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
2290 #define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2291 #define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
2292 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2293 #define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
2294 #define isSPACE_utf8_safe(p, e) \
2295 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2296 #define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
2297 #define isVERTWS_utf8_safe(p, e) \
2298 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2299 #define isWORDCHAR_utf8_safe(p, e) \
2300 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
2301 #define isXDIGIT_utf8_safe(p, e) \
2302 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2303 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2304 ? (_force_out_malformed_utf8_message( \
2305 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2306 : is_XDIGIT_high(p)))
2308 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2309 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2310 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2311 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2313 /* For internal core use only, subject to change */
2314 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2315 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2316 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2317 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
2319 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2320 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2321 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2322 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2324 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2325 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2326 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2327 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2328 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2329 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2330 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2331 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2332 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2333 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2334 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2335 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2336 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2337 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2338 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2339 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2340 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
2342 /* For internal core Perl use only: the base macros for defining macros like
2343 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2344 * point in 'p' is within the 0-255 range, it uses locale rules from the
2345 * passed-in 'macro' parameter */
2346 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2347 (__ASSERT_(_utf8_safe_assert(p, e)) \
2348 (UTF8_IS_INVARIANT(*(p))) \
2350 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2351 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2352 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2353 : (_force_out_malformed_utf8_message( \
2354 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2357 #define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2358 _generic_LC_utf8_safe(macro, p, e, \
2359 _is_utf8_FOO(classnum, p, e))
2361 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2362 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2364 #define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2365 _generic_LC_utf8_safe(classnum, p, e, \
2366 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2367 ? (_force_out_malformed_utf8_message( \
2368 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2371 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2372 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
2373 _CC_ALPHANUMERIC, p, e)
2374 #define isALPHA_LC_utf8_safe(p, e) \
2375 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2376 #define isASCII_LC_utf8_safe(p, e) \
2377 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2378 #define isBLANK_LC_utf8_safe(p, e) \
2379 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2380 #define isCNTRL_LC_utf8_safe(p, e) \
2381 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2382 #define isDIGIT_LC_utf8_safe(p, e) \
2383 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2384 #define isGRAPH_LC_utf8_safe(p, e) \
2385 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2386 #define isIDCONT_LC_utf8_safe(p, e) \
2387 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2388 _is_utf8_perl_idcont, p, e)
2389 #define isIDFIRST_LC_utf8_safe(p, e) \
2390 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2391 _is_utf8_perl_idstart, p, e)
2392 #define isLOWER_LC_utf8_safe(p, e) \
2393 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2394 #define isPRINT_LC_utf8_safe(p, e) \
2395 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2396 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2397 #define isPUNCT_LC_utf8_safe(p, e) \
2398 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2399 #define isSPACE_LC_utf8_safe(p, e) \
2400 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2401 #define isUPPER_LC_utf8_safe(p, e) \
2402 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2403 #define isWORDCHAR_LC_utf8_safe(p, e) \
2404 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2405 #define isXDIGIT_LC_utf8_safe(p, e) \
2406 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2408 /* Macros for backwards compatibility and for completeness when the ASCII and
2409 * Latin1 values are identical */
2410 #define isALPHAU(c) isALPHA_L1(c)
2411 #define isDIGIT_L1(c) isDIGIT_A(c)
2412 #define isOCTAL(c) isOCTAL_A(c)
2413 #define isOCTAL_L1(c) isOCTAL_A(c)
2414 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2415 #define isALNUM(c) isWORDCHAR(c)
2416 #define isALNUM_A(c) isALNUM(c)
2417 #define isALNUMU(c) isWORDCHAR_L1(c)
2418 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2419 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2420 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2421 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
2422 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
2423 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
2424 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
2425 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2426 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2427 #define isALNUMC(c) isALPHANUMERIC(c)
2428 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2429 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2430 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2431 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
2432 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2433 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
2435 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2436 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2437 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2438 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2439 * the outlier from the block that contains the other controls, just like
2440 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2441 * block. If it weren't special cased, it would yield a non-control.
2442 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2445 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2447 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2449 ? (UNLIKELY((c) == '?') \
2450 ? QUESTION_MARK_CTRL \
2451 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2452 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2454 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2458 =for apidoc Ay||line_t
2459 The typedef to use to declare variables that are to hold line numbers.
2463 Line numbers are unsigned, 32 bits.
2466 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2468 /* Helpful alias for version prescan */
2469 #define is_LAX_VERSION(a,b) \
2470 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2472 #define is_STRICT_VERSION(a,b) \
2473 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2475 #define BADVERSION(a,b,c) \
2481 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2482 * a-f) to its numeric value without using any branches. The input is
2483 * validated only by an assert() in DEBUGGING builds.
2485 * It works by right shifting and isolating the bit that is 0 for the digits,
2486 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2487 * position, and then to the eights position. Both are added together to form
2488 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2489 * final four bits of the input to form the correct value. */
2490 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2491 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2492 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2493 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2495 /* The argument is a string pointer, which is advanced. */
2496 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
2498 /* Converts a character known to represent an octal digit (0-7) to its numeric
2499 * value. The input is validated only by an assert() in DEBUGGING builds. In
2500 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2501 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2503 /* Efficiently returns a boolean as to if two native characters are equivalent
2504 * case-insensitively. At least one of the characters must be one of [A-Za-z];
2505 * the ALPHA in the name is to remind you of that. This is asserted() in
2506 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2507 * works (on valid input) for both non- and UTF-8-encoded bytes.
2509 * When one of the inputs is a compile-time constant and gets folded by the
2510 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2511 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2512 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2513 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2514 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2515 * just a single 0, in the bit position where the upper- and lowercase differ.
2517 #define isALPHA_FOLD_EQ(c1, c2) \
2518 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2519 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2520 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2523 =for apidoc_section $memory
2525 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2526 The XSUB-writer's interface to the C C<malloc> function.
2528 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2530 In 5.9.3, Newx() and friends replace the older New() API, and drops
2531 the first parameter, I<x>, a debug aid which allowed callers to identify
2532 themselves. This aid has been superseded by a new build option,
2533 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2534 there for use in XS modules supporting older perls.
2536 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2537 The XSUB-writer's interface to the C C<malloc> function, with
2538 cast. See also C<L</Newx>>.
2540 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2542 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2543 The XSUB-writer's interface to the C C<malloc> function. The allocated
2544 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2546 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2548 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2549 The XSUB-writer's interface to the C C<realloc> function.
2551 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2553 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2554 The XSUB-writer's interface to the C C<realloc> function, with
2557 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2559 =for apidoc Am|void|Safefree|void* ptr
2560 The XSUB-writer's interface to the C C<free> function.
2562 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2564 =for apidoc_section $string
2565 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2566 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2567 source, C<dest> is the destination, C<nitems> is the number of items, and
2568 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2570 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2571 Like C<Move> but returns C<dest>. Useful
2572 for encouraging compilers to tail-call
2575 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2576 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2577 source, C<dest> is the destination, C<nitems> is the number of items, and
2578 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2580 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2582 Like C<Copy> but returns C<dest>. Useful
2583 for encouraging compilers to tail-call
2586 =for apidoc Am|void|Zero|void* dest|int nitems|type
2588 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2589 destination, C<nitems> is the number of items, and C<type> is the type.
2591 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2593 Like C<Zero> but returns dest. Useful
2594 for encouraging compilers to tail-call
2597 =for apidoc_section $utility
2598 =for apidoc Amu|void|StructCopy|type *src|type *dest|type
2599 This is an architecture-independent macro to copy one structure to another.
2601 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2603 Fill up memory with a byte pattern (a byte repeated over and over
2604 again) that hopefully catches attempts to access uninitialized memory.
2606 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2608 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2610 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2612 PoisonWith(0xEF) for catching access to freed memory.
2614 =for apidoc Am|void|Poison|void* dest|int nitems|type
2616 PoisonWith(0xEF) for catching access to freed memory.
2620 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2622 #define NEWSV(x,len) newSV(len)
2625 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2627 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2629 #ifdef PERL_MALLOC_WRAP
2631 /* This expression will be constant-folded at compile time. It checks
2632 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2633 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2634 * As well as avoiding the need for a run-time check in some cases, it's
2635 * designed to avoid compiler warnings like:
2636 * comparison is always false due to limited range of data type
2637 * It's mathematically equivalent to
2638 * max(n) * sizeof(t) > MEM_SIZE_MAX
2641 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2642 ( sizeof(MEM_SIZE) < sizeof(n) \
2643 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2645 /* This is written in a slightly odd way to avoid various spurious
2646 * compiler warnings. We *want* to write the expression as
2647 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2648 * (for some compile-time constant C), but even when the LHS
2649 * constant-folds to false at compile-time, g++ insists on emitting
2650 * warnings about the RHS (e.g. "comparison is always false"), so instead
2653 * (cond ? n : X) > C
2655 * where X is a constant with X > C always false. Choosing a value for X
2656 * is tricky. If 0, some compilers will complain about 0 > C always being
2657 * false; if 1, Coverity complains when n happens to be the constant value
2658 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2659 * for X and hope that nothing else whines.
2662 # define _MEM_WRAP_WILL_WRAP(n,t) \
2663 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2664 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2666 # define MEM_WRAP_CHECK(n,t) \
2667 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2668 && (croak_memory_wrap(),0))
2670 # define MEM_WRAP_CHECK_1(n,t,a) \
2671 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2672 && (Perl_croak_nocontext("%s",(a)),0))
2674 /* "a" arg must be a string literal */
2675 # define MEM_WRAP_CHECK_s(n,t,a) \
2676 ( (void) (UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2677 && (Perl_croak_nocontext(ASSERT_IS_LITERAL(a)), 0)))
2679 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2681 #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))
2684 #define MEM_WRAP_CHECK(n,t)
2685 #define MEM_WRAP_CHECK_1(n,t,a)
2686 #define MEM_WRAP_CHECK_s(n,t,a)
2687 #define MEM_WRAP_CHECK_(n,t)
2689 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2695 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2696 * go through functions, which are handy for debugging breakpoints, but
2697 * which more importantly get the immediate calling environment (file and
2698 * line number, and C function name if available) passed in. This info can
2699 * then be used for logging the calls, for which one gets a sample
2700 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2703 * - not all memory allocs get logged, only those
2704 * that go through Newx() and derivatives (while all
2705 * Safefrees do get logged)
2706 * - __FILE__ and __LINE__ do not work everywhere
2707 * - __func__ or __FUNCTION__ even less so
2708 * - I think more goes on after the perlio frees but
2709 * the thing is that STDERR gets closed (as do all
2710 * the file descriptors)
2711 * - no deeper calling stack than the caller of the Newx()
2712 * or the kind, but do I look like a C reflection/introspection
2714 * - the function prototypes for the logging functions
2715 * probably should maybe be somewhere else than handy.h
2716 * - one could consider inlining (macrofying) the logging
2717 * for speed, but I am too lazy
2718 * - one could imagine recording the allocations in a hash,
2719 * (keyed by the allocation address?), and maintain that
2720 * through reallocs and frees, but how to do that without
2721 * any News() happening...?
2722 * - lots of -Ddefines to get useful/controllable output
2723 * - lots of ENV reads
2727 # ifndef PERL_MEM_LOG_NOIMPL
2736 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2737 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2738 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2745 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2746 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2747 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2750 #ifndef MEM_LOG_ALLOC
2751 #define MEM_LOG_ALLOC(n,t,a) (a)
2753 #ifndef MEM_LOG_REALLOC
2754 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2756 #ifndef MEM_LOG_FREE
2757 #define MEM_LOG_FREE(a) (a)
2760 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2761 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2762 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2765 /* pre 5.9.x compatibility */
2766 #define New(x,v,n,t) Newx(v,n,t)
2767 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2768 #define Newz(x,v,n,t) Newxz(v,n,t)
2771 #define Renew(v,n,t) \
2772 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2773 #define Renewc(v,n,t,c) \
2774 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2777 #define Safefree(d) \
2778 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2780 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2783 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2784 * as it handles cases like constant string arguments without throwing warnings *
2785 * the cast is required, as is the inequality check, to avoid warnings */
2786 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2789 #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)))
2790 #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)))
2791 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2793 /* Like above, but returns a pointer to 'd' */
2794 #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)))
2795 #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)))
2796 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2798 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2799 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2800 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2801 #define Poison(d,n,t) PoisonFree(d,n,t)
2804 # define PERL_POISON_EXPR(x) x
2806 # define PERL_POISON_EXPR(x)
2810 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2813 =for apidoc_section $utility
2815 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2817 Returns the number of elements in the input C array (so you want your
2818 zero-based indices to be less than but not equal to).
2820 =for apidoc Am|void *|C_ARRAY_END|void *a
2822 Returns a pointer to one element past the final element of the input C array.
2826 C_ARRAY_END is one past the last: half-open/half-closed range, not
2827 last-inclusive range.
2829 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2830 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2834 # define Perl_va_copy(s, d) va_copy(d, s)
2835 # elif defined(__va_copy)
2836 # define Perl_va_copy(s, d) __va_copy(d, s)
2838 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2842 /* convenience debug macros */
2844 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2845 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2846 #define pTHX_VALUE_ (void *)my_perl,
2847 #define pTHX_VALUE (void *)my_perl
2848 #define pTHX__VALUE_ ,(void *)my_perl,
2849 #define pTHX__VALUE ,(void *)my_perl
2852 #define pTHX__FORMAT
2855 #define pTHX__VALUE_
2857 #endif /* USE_ITHREADS */
2859 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2860 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2861 CPAN::Unpack show any users outside the core. */
2863 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2864 "Use of " s " is deprecated")
2865 # define deprecate_disappears_in(when,message) \
2866 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2867 message " is deprecated, and will disappear in Perl " when)
2868 # define deprecate_fatal_in(when,message) \
2869 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2870 message " is deprecated, and will become fatal in Perl " when)
2873 /* Internal macros to deal with gids and uids */
2876 # if Uid_t_size > IVSIZE
2877 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2878 # define SvUID(sv) SvNV(sv)
2879 # elif Uid_t_sign <= 0
2880 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2881 # define SvUID(sv) SvIV(sv)
2883 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2884 # define SvUID(sv) SvUV(sv)
2885 # endif /* Uid_t_size */
2887 # if Gid_t_size > IVSIZE
2888 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2889 # define SvGID(sv) SvNV(sv)
2890 # elif Gid_t_sign <= 0
2891 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2892 # define SvGID(sv) SvIV(sv)
2894 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2895 # define SvGID(sv) SvUV(sv)
2896 # endif /* Gid_t_size */
2900 #endif /* PERL_HANDY_H_ */
2903 * ex: set ts=8 sts=4 sw=4 et: