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)
23 =for apidoc AmnU||Nullch
24 Null character pointer. (No longer available when C<PERL_CORE> is
27 =for apidoc AmnU||Nullsv
28 Null SV pointer. (No longer available when C<PERL_CORE> is defined.)
33 # define Nullch Null(char*)
34 # define Nullfp Null(PerlIO*)
35 # define Nullsv Null(SV*)
47 /* The MUTABLE_*() macros cast pointers to the types shown, in such a way
48 * (compiler permitting) that casting away const-ness will give a warning;
52 * AV *av1 = (AV*)sv; <== BAD: the const has been silently cast away
53 * AV *av2 = MUTABLE_AV(sv); <== GOOD: it may warn
56 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
57 # define MUTABLE_PTR(p) ({ void *p_ = (p); p_; })
59 # define MUTABLE_PTR(p) ((void *) (p))
62 #define MUTABLE_AV(p) ((AV *)MUTABLE_PTR(p))
63 #define MUTABLE_CV(p) ((CV *)MUTABLE_PTR(p))
64 #define MUTABLE_GV(p) ((GV *)MUTABLE_PTR(p))
65 #define MUTABLE_HV(p) ((HV *)MUTABLE_PTR(p))
66 #define MUTABLE_IO(p) ((IO *)MUTABLE_PTR(p))
67 #define MUTABLE_SV(p) ((SV *)MUTABLE_PTR(p))
69 #if defined(I_STDBOOL) && !defined(PERL_BOOL_AS_CHAR)
76 /* bool is built-in for g++-2.6.3 and later, which might be used
77 for extensions. <_G_config.h> defines _G_HAVE_BOOL, but we can't
78 be sure _G_config.h will be included before this file. _G_config.h
79 also defines _G_HAVE_BOOL for both gcc and g++, but only g++
80 actually has bool. Hence, _G_HAVE_BOOL is pretty useless for us.
81 g++ can be identified by __GNUG__.
82 Andy Dougherty February 2000
84 #ifdef __GNUG__ /* GNU g++ has bool built-in */
85 # ifndef PERL_BOOL_AS_CHAR
101 =for apidoc Am|bool|cBOOL|bool expr
103 Cast-to-bool. A simple S<C<(bool) I<expr>>> cast may not do the right thing:
104 if C<bool> is defined as C<char>, for example, then the cast from C<int> is
105 implementation-defined.
107 C<(bool)!!(cbool)> in a ternary triggers a bug in xlc on AIX
111 #define cBOOL(cbool) ((cbool) ? (bool)1 : (bool)0)
113 /* Try to figure out __func__ or __FUNCTION__ equivalent, if any.
114 * XXX Should really be a Configure probe, with HAS__FUNCTION__
115 * and FUNCTION__ as results.
116 * XXX Similarly, a Configure probe for __FILE__ and __LINE__ is needed. */
117 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || (defined(__SUNPRO_C)) /* C99 or close enough. */
118 # define FUNCTION__ __func__
119 #elif (defined(__DECC_VER)) /* Tru64 or VMS, and strict C89 being used, but not modern enough cc (in Tur64, -c99 not known, only -std1). */
120 # define FUNCTION__ ""
122 # define FUNCTION__ __FUNCTION__ /* Common extension. */
125 /* XXX A note on the perl source internal type system. The
126 original intent was that I32 be *exactly* 32 bits.
128 Currently, we only guarantee that I32 is *at least* 32 bits.
129 Specifically, if int is 64 bits, then so is I32. (This is the case
130 for the Cray.) This has the advantage of meshing nicely with
131 standard library calls (where we pass an I32 and the library is
132 expecting an int), but the disadvantage that an I32 is not 32 bits.
133 Andy Dougherty August 1996
135 There is no guarantee that there is *any* integral type with
136 exactly 32 bits. It is perfectly legal for a system to have
137 sizeof(short) == sizeof(int) == sizeof(long) == 8.
139 Similarly, there is no guarantee that I16 and U16 have exactly 16
142 For dealing with issues that may arise from various 32/64-bit
143 systems, we will ask Configure to check out
145 SHORTSIZE == sizeof(short)
146 INTSIZE == sizeof(int)
147 LONGSIZE == sizeof(long)
148 LONGLONGSIZE == sizeof(long long) (if HAS_LONG_LONG)
149 PTRSIZE == sizeof(void *)
150 DOUBLESIZE == sizeof(double)
151 LONG_DOUBLESIZE == sizeof(long double) (if HAS_LONG_DOUBLE).
155 #ifdef I_INTTYPES /* e.g. Linux has int64_t without <inttypes.h> */
156 # include <inttypes.h>
157 # ifdef INT32_MIN_BROKEN
159 # define INT32_MIN (-2147483647-1)
161 # ifdef INT64_MIN_BROKEN
163 # define INT64_MIN (-9223372036854775807LL-1)
179 #if defined(UINT8_MAX) && defined(INT16_MAX) && defined(INT32_MAX)
181 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
182 Please search CHAR_MAX in perl.h for further details. */
183 #define U8_MAX UINT8_MAX
184 #define U8_MIN UINT8_MIN
186 #define I16_MAX INT16_MAX
187 #define I16_MIN INT16_MIN
188 #define U16_MAX UINT16_MAX
189 #define U16_MIN UINT16_MIN
191 #define I32_MAX INT32_MAX
192 #define I32_MIN INT32_MIN
193 #ifndef UINT32_MAX_BROKEN /* e.g. HP-UX with gcc messes this up */
194 # define U32_MAX UINT32_MAX
196 # define U32_MAX 4294967295U
198 #define U32_MIN UINT32_MIN
202 /* I8_MAX and I8_MIN constants are not defined, as I8 is an ambiguous type.
203 Please search CHAR_MAX in perl.h for further details. */
204 #define U8_MAX PERL_UCHAR_MAX
205 #define U8_MIN PERL_UCHAR_MIN
207 #define I16_MAX PERL_SHORT_MAX
208 #define I16_MIN PERL_SHORT_MIN
209 #define U16_MAX PERL_USHORT_MAX
210 #define U16_MIN PERL_USHORT_MIN
213 # define I32_MAX PERL_INT_MAX
214 # define I32_MIN PERL_INT_MIN
215 # define U32_MAX PERL_UINT_MAX
216 # define U32_MIN PERL_UINT_MIN
218 # define I32_MAX PERL_LONG_MAX
219 # define I32_MIN PERL_LONG_MIN
220 # define U32_MAX PERL_ULONG_MAX
221 # define U32_MIN PERL_ULONG_MIN
226 /* These C99 typedefs are useful sometimes for, say, loop variables whose
227 * maximum values are small, but for which speed trumps size. If we have a C99
228 * compiler, use that. Otherwise, a plain 'int' should be good enough.
230 * Restrict these to core for now until we are more certain this is a good
232 #if defined(PERL_CORE) || defined(PERL_EXT)
234 typedef int_fast8_t PERL_INT_FAST8_T;
235 typedef uint_fast8_t PERL_UINT_FAST8_T;
236 typedef int_fast16_t PERL_INT_FAST16_T;
237 typedef uint_fast16_t PERL_UINT_FAST16_T;
239 typedef int PERL_INT_FAST8_T;
240 typedef unsigned int PERL_UINT_FAST8_T;
241 typedef int PERL_INT_FAST16_T;
242 typedef unsigned int PERL_UINT_FAST16_T;
246 /* log(2) (i.e., log base 10 of 2) is pretty close to 0.30103, just in case
247 * anyone is grepping for it. So BIT_DIGITS gives the number of decimal digits
248 * required to represent any possible unsigned number containing N bits.
249 * TYPE_DIGITS gives the number of decimal digits required to represent any
250 * possible unsigned number of type T. */
251 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log10(2) =~ 146/485 */
252 #define TYPE_DIGITS(T) BIT_DIGITS(sizeof(T) * 8)
253 #define TYPE_CHARS(T) (TYPE_DIGITS(T) + 2) /* sign, NUL */
255 /* Unused by core; should be deprecated */
256 #define Ctl(ch) ((ch) & 037)
258 #if defined(PERL_CORE) || defined(PERL_EXT)
260 # define MIN(a,b) ((a) < (b) ? (a) : (b))
263 # define MAX(a,b) ((a) > (b) ? (a) : (b))
267 /* Returns a boolean as to whether the input unsigned number is a power of 2
268 * (2**0, 2**1, etc). In other words if it has just a single bit set.
269 * If not, subtracting 1 would leave the uppermost bit set, so the & would
271 #if defined(PERL_CORE) || defined(PERL_EXT)
272 # define isPOWER_OF_2(n) ((n) && ((n) & ((n)-1)) == 0)
275 /* Returns a mask with the lowest n bits set */
276 #define nBIT_MASK(n) ((UINTMAX_C(1) << (n)) - 1)
278 /* The largest unsigned number that will fit into n bits */
279 #define nBIT_UMAX(n) nBIT_MASK(n)
282 =for apidoc Am|void|__ASSERT_|bool expr
284 This is a helper macro to avoid preprocessor issues, replaced by nothing
285 unless under DEBUGGING, where it expands to an assert of its argument,
286 followed by a comma (hence the comma operator). If we just used a straight
287 assert(), we would get a comma with nothing before it when not DEBUGGING.
291 We also use empty definition under Coverity since the __ASSERT__
292 checks often check for things that Really Cannot Happen, and Coverity
293 detects that and gets all excited. */
295 #if defined(DEBUGGING) && !defined(__COVERITY__) \
296 && ! defined(PERL_SMALL_MACRO_BUFFER)
297 # define __ASSERT_(statement) assert(statement),
299 # define __ASSERT_(statement)
303 =head1 SV Manipulation Functions
305 =for apidoc Ama|SV*|newSVpvs|"literal string"
306 Like C<newSVpvn>, but takes a literal string instead of a
309 =for apidoc Ama|SV*|newSVpvs_flags|"literal string"|U32 flags
310 Like C<newSVpvn_flags>, but takes a literal string instead of
311 a string/length pair.
313 =for apidoc Ama|SV*|newSVpvs_share|"literal string"
314 Like C<newSVpvn_share>, but takes a literal string instead of
315 a string/length pair and omits the hash parameter.
317 =for apidoc Am|void|sv_catpvs_flags|SV* sv|"literal string"|I32 flags
318 Like C<sv_catpvn_flags>, but takes a literal string instead
319 of a string/length pair.
321 =for apidoc Am|void|sv_catpvs_nomg|SV* sv|"literal string"
322 Like C<sv_catpvn_nomg>, but takes a literal string instead of
323 a string/length pair.
325 =for apidoc Am|void|sv_catpvs|SV* sv|"literal string"
326 Like C<sv_catpvn>, but takes a literal string instead of a
329 =for apidoc Am|void|sv_catpvs_mg|SV* sv|"literal string"
330 Like C<sv_catpvn_mg>, but takes a literal string instead of a
333 =for apidoc Am|void|sv_setpvs|SV* sv|"literal string"
334 Like C<sv_setpvn>, but takes a literal string instead of a
337 =for apidoc Am|void|sv_setpvs_mg|SV* sv|"literal string"
338 Like C<sv_setpvn_mg>, but takes a literal string instead of a
341 =for apidoc Am|SV *|sv_setref_pvs|SV *const rv|const char *const classname|"literal string"
342 Like C<sv_setref_pvn>, but takes a literal string instead of
343 a string/length pair.
345 =head1 Memory Management
347 =for apidoc Ama|char*|savepvs|"literal string"
348 Like C<savepvn>, but takes a literal string instead of a
351 =for apidoc Ama|char*|savesharedpvs|"literal string"
352 A version of C<savepvs()> which allocates the duplicate string in memory
353 which is shared between threads.
357 =for apidoc Am|HV*|gv_stashpvs|"name"|I32 create
358 Like C<gv_stashpvn>, but takes a literal string instead of a
361 =head1 Hash Manipulation Functions
363 =for apidoc Am|SV**|hv_fetchs|HV* tb|"key"|I32 lval
364 Like C<hv_fetch>, but takes a literal string instead of a
367 =for apidoc Am|SV**|hv_stores|HV* tb|"key"|SV* val
368 Like C<hv_store>, but takes a literal string instead of a
370 and omits the hash parameter.
372 =head1 Lexer interface
374 =for apidoc Amx|void|lex_stuff_pvs|"pv"|U32 flags
376 Like L</lex_stuff_pvn>, but takes a literal string instead of
377 a string/length pair.
385 =for apidoc Amu|pair|STR_WITH_LEN|"literal string"
387 Returns two comma separated tokens of the input literal string, and its length.
388 This is convenience macro which helps out in some API calls.
389 Note that it can't be used as an argument to macros or functions that under
390 some configurations might be macros, which means that it requires the full
391 Perl_xxx(aTHX_ ...) form for any API calls where it's used.
396 #define STR_WITH_LEN(s) ("" s ""), (sizeof(s)-1)
398 /* STR_WITH_LEN() shortcuts */
399 #define newSVpvs(str) Perl_newSVpvn(aTHX_ STR_WITH_LEN(str))
400 #define newSVpvs_flags(str,flags) \
401 Perl_newSVpvn_flags(aTHX_ STR_WITH_LEN(str), flags)
402 #define newSVpvs_share(str) Perl_newSVpvn_share(aTHX_ STR_WITH_LEN(str), 0)
403 #define sv_catpvs_flags(sv, str, flags) \
404 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), flags)
405 #define sv_catpvs_nomg(sv, str) \
406 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), 0)
407 #define sv_catpvs(sv, str) \
408 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC)
409 #define sv_catpvs_mg(sv, str) \
410 Perl_sv_catpvn_flags(aTHX_ sv, STR_WITH_LEN(str), SV_GMAGIC|SV_SMAGIC)
411 #define sv_setpvs(sv, str) Perl_sv_setpvn(aTHX_ sv, STR_WITH_LEN(str))
412 #define sv_setpvs_mg(sv, str) Perl_sv_setpvn_mg(aTHX_ sv, STR_WITH_LEN(str))
413 #define sv_setref_pvs(rv, classname, str) \
414 Perl_sv_setref_pvn(aTHX_ rv, classname, STR_WITH_LEN(str))
415 #define savepvs(str) Perl_savepvn(aTHX_ STR_WITH_LEN(str))
416 #define savesharedpvs(str) Perl_savesharedpvn(aTHX_ STR_WITH_LEN(str))
417 #define gv_stashpvs(str, create) \
418 Perl_gv_stashpvn(aTHX_ STR_WITH_LEN(str), create)
419 #define gv_fetchpvs(namebeg, add, sv_type) \
420 Perl_gv_fetchpvn_flags(aTHX_ STR_WITH_LEN(namebeg), add, sv_type)
421 #define gv_fetchpvn(namebeg, len, add, sv_type) \
422 Perl_gv_fetchpvn_flags(aTHX_ namebeg, len, add, sv_type)
423 #define sv_catxmlpvs(dsv, str, utf8) \
424 Perl_sv_catxmlpvn(aTHX_ dsv, STR_WITH_LEN(str), utf8)
427 #define lex_stuff_pvs(pv,flags) Perl_lex_stuff_pvn(aTHX_ STR_WITH_LEN(pv), flags)
429 #define get_cvs(str, flags) \
430 Perl_get_cvn_flags(aTHX_ STR_WITH_LEN(str), (flags))
432 /* internal helpers */
434 #ifndef PERL_VERSION_MAJOR
435 # define PERL_VERSION_MAJOR PERL_REVISION
437 # undef PERL_REVISION /* We don't want code to be using these */
439 #ifndef PERL_VERSION_MINOR
440 # define PERL_VERSION_MINOR PERL_VERSION
444 #ifndef PERL_VERSION_PATCH
445 # define PERL_VERSION_PATCH PERL_SUBVERSION
447 # undef PERL_SUBVERSION
450 #define PERL_JNP_TO_DECIMAL_(maJor,miNor,Patch) \
451 /* '10*' leaves room for things like alpha, beta, releases */ \
452 (10 * ((maJor) * 1000000) + ((miNor) * 1000) + (Patch))
453 #define PERL_DECIMAL_VERSION_ \
454 PERL_JNP_TO_DECIMAL_(PERL_VERSION_MAJOR, PERL_VERSION_MINOR, \
458 =for apidoc AmR|bool|PERL_VERSION_EQ|const U8 major|const U8 minor|const U8 patch
460 Returns whether or not the perl currently being compiled has the specified
461 relationship to the perl given by the parameters. For example,
463 #if PERL_VERSION_GT(5,24,2)
464 code that will only be compiled on perls after v5.24.2
469 Note that this is usable in making compile-time decisions
471 The possible comparisons are C<PERL_VERSION_EQ>, C<PERL_VERSION_NE>,
472 C<PERL_VERSION_GE>, C<PERL_VERSION_GT>, C<PERL_VERSION_LE>, and
475 You may use the special value '*' for the final number to mean ALL possible
478 #if PERL_VERSION_EQ(5,31,'*')
480 means all perls in the 5.31 series. And
482 #if PERL_VERSION_NE(5,24,'*')
484 means all perls EXCEPT 5.24 ones. And
486 #if PERL_VERSION_LE(5,9,'*')
490 #if PERL_VERSION_LT(5,10,0)
492 This means you don't have to think so much when converting from the existing
493 deprecated C<PERL_VERSION> to using this macro:
495 #if PERL_VERSION <= 9
499 #if PERL_VERSION_LE(5,9,'*')
501 =for apidoc AmRh|bool|PERL_VERSION_NE|const U8 major|const U8 minor|const U8 patch
502 =for apidoc AmRh|bool|PERL_VERSION_GE|const U8 major|const U8 minor|const U8 patch
503 =for apidoc AmRh|bool|PERL_VERSION_GT|const U8 major|const U8 minor|const U8 patch
504 =for apidoc AmRh|bool|PERL_VERSION_LE|const U8 major|const U8 minor|const U8 patch
505 =for apidoc AmRh|bool|PERL_VERSION_LT|const U8 major|const U8 minor|const U8 patch
510 /* N.B. These don't work if the patch version is 42 or 92, as those are what
511 * '*' is in ASCII and EBCDIC respectively */
512 # define PERL_VERSION_EQ(j,n,p) \
514 ? ( (j) == PERL_VERSION_MAJOR \
515 && (n) == PERL_VERSION_MINOR) \
516 : (PERL_DECIMAL_VERSION_ == PERL_JNP_TO_DECIMAL_(j,n,p)))
517 # define PERL_VERSION_NE(j,n,p) (! PERL_VERSION_EQ(j,n,p))
519 # define PERL_VERSION_LT(j,n,p) /* < '*' effectively means < 0 */ \
520 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
522 (((p) == '*') ? 0 : p)))
523 # define PERL_VERSION_GE(j,n,p) (! PERL_VERSION_LT(j,n,p))
525 # define PERL_VERSION_LE(j,n,p) /* <= '*' effectively means < n+1 */ \
526 (PERL_DECIMAL_VERSION_ < PERL_JNP_TO_DECIMAL_( (j), \
527 (((p) == '*') ? ((n)+1) : (n)), \
528 (((p) == '*') ? 0 : p)))
529 # define PERL_VERSION_GT(j,n,p) (! PERL_VERSION_LE(j,n,p))
532 =head1 Miscellaneous Functions
534 =for apidoc Am|bool|strNE|char* s1|char* s2
535 Test two C<NUL>-terminated strings to see if they are different. Returns true
538 =for apidoc Am|bool|strEQ|char* s1|char* s2
539 Test two C<NUL>-terminated strings to see if they are equal. Returns true or
542 =for apidoc Am|bool|strLT|char* s1|char* s2
543 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than the
544 second, C<s2>. Returns true or false.
546 =for apidoc Am|bool|strLE|char* s1|char* s2
547 Test two C<NUL>-terminated strings to see if the first, C<s1>, is less than or
548 equal to the second, C<s2>. Returns true or false.
550 =for apidoc Am|bool|strGT|char* s1|char* s2
551 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
552 the second, C<s2>. Returns true or false.
554 =for apidoc Am|bool|strGE|char* s1|char* s2
555 Test two C<NUL>-terminated strings to see if the first, C<s1>, is greater than
556 or equal to the second, C<s2>. Returns true or false.
558 =for apidoc Am|bool|strnNE|char* s1|char* s2|STRLEN len
559 Test two C<NUL>-terminated strings to see if they are different. The C<len>
560 parameter indicates the number of bytes to compare. Returns true or false. (A
561 wrapper for C<strncmp>).
563 =for apidoc Am|bool|strnEQ|char* s1|char* s2|STRLEN len
564 Test two C<NUL>-terminated strings to see if they are equal. The C<len>
565 parameter indicates the number of bytes to compare. Returns true or false. (A
566 wrapper for C<strncmp>).
568 =for apidoc Am|bool|memEQ|char* s1|char* s2|STRLEN len
569 Test two buffers (which may contain embedded C<NUL> characters, to see if they
570 are equal. The C<len> parameter indicates the number of bytes to compare.
571 Returns zero if equal, or non-zero if non-equal.
573 =for apidoc Am|bool|memEQs|char* s1|STRLEN l1|"s2"
574 Like L</memEQ>, but the second string is a literal enclosed in double quotes,
575 C<l1> gives the number of bytes in C<s1>.
576 Returns zero if equal, or non-zero if non-equal.
578 =for apidoc Am|bool|memNE|char* s1|char* s2|STRLEN len
579 Test two buffers (which may contain embedded C<NUL> characters, to see if they
580 are not equal. The C<len> parameter indicates the number of bytes to compare.
581 Returns zero if non-equal, or non-zero if equal.
583 =for apidoc Am|bool|memNEs|char* s1|STRLEN l1|"s2"
584 Like L</memNE>, but the second string is a literal enclosed in double quotes,
585 C<l1> gives the number of bytes in C<s1>.
586 Returns zero if non-equal, or zero if non-equal.
588 =for apidoc Am|bool|memCHRs|"list"|char c
589 Returns the position of the first occurence of the byte C<c> in the literal
590 string C<"list">, or NULL if C<c> doesn't appear in C<"list">. All bytes are
591 treated as unsigned char. Thus this macro can be used to determine if C<c> is
592 in a set of particular characters. Unlike L<strchr(3)>, it works even if C<c>
593 is C<NUL> (and the set doesn't include C<NUL>).
597 New macros should use the following conventions for their names (which are
598 based on the underlying C library functions):
600 (mem | str n? ) (EQ | NE | LT | GT | GE | (( BEGIN | END ) P? )) l? s?
602 Each has two main parameters, string-like operands that are compared
603 against each other, as specified by the macro name. Some macros may
604 additionally have one or potentially even two length parameters. If a length
605 parameter applies to both string parameters, it will be positioned third;
606 otherwise any length parameter immediately follows the string parameter it
609 If the prefix to the name is 'str', the string parameter is a pointer to a C
610 language string. Such a string does not contain embedded NUL bytes; its
611 length may be unknown, but can be calculated by C<strlen()>, since it is
612 terminated by a NUL, which isn't included in its length.
614 The optional 'n' following 'str' means that there is a third parameter,
615 giving the maximum number of bytes to look at in each string. Even if both
616 strings are longer than the length parameter, those extra bytes will be
619 The 's' suffix means that the 2nd byte string parameter is a literal C
620 double-quoted string. Its length will automatically be calculated by the
621 macro, so no length parameter will ever be needed for it.
623 If the prefix is 'mem', the string parameters don't have to be C strings;
624 they may contain embedded NUL bytes, do not necessarily have a terminating
625 NUL, and their lengths can be known only through other means, which in
626 practice are additional parameter(s) passed to the function. All 'mem'
627 functions have at least one length parameter. Barring any 'l' or 's' suffix,
628 there is a single length parameter, in position 3, which applies to both
629 string parameters. The 's' suffix means, as described above, that the 2nd
630 string is a literal double-quoted C string (hence its length is calculated by
631 the macro, and the length parameter to the function applies just to the first
632 string parameter, and hence is positioned just after it). An 'l' suffix
633 means that the 2nd string parameter has its own length parameter, and the
634 signature will look like memFOOl(s1, l1, s2, l2).
636 BEGIN (and END) are for testing if the 2nd string is an initial (or final)
637 substring of the 1st string. 'P' if present indicates that the substring
638 must be a "proper" one in tha mathematical sense that the first one must be
639 strictly larger than the 2nd.
644 #define strNE(s1,s2) (strcmp(s1,s2) != 0)
645 #define strEQ(s1,s2) (strcmp(s1,s2) == 0)
646 #define strLT(s1,s2) (strcmp(s1,s2) < 0)
647 #define strLE(s1,s2) (strcmp(s1,s2) <= 0)
648 #define strGT(s1,s2) (strcmp(s1,s2) > 0)
649 #define strGE(s1,s2) (strcmp(s1,s2) >= 0)
651 #define strnNE(s1,s2,l) (strncmp(s1,s2,l) != 0)
652 #define strnEQ(s1,s2,l) (strncmp(s1,s2,l) == 0)
654 #define memEQ(s1,s2,l) (memcmp(((const void *) (s1)), ((const void *) (s2)), l) == 0)
655 #define memNE(s1,s2,l) (! memEQ(s1,s2,l))
657 /* memEQ and memNE where second comparand is a string constant */
658 #define memEQs(s1, l, s2) \
659 (((sizeof(s2)-1) == (l)) && memEQ((s1), ("" s2 ""), (sizeof(s2)-1)))
660 #define memNEs(s1, l, s2) (! memEQs(s1, l, s2))
662 /* Keep these private until we decide it was a good idea */
663 #if defined(PERL_CORE) || defined(PERL_EXT) || defined(PERL_EXT_POSIX)
665 #define strBEGINs(s1,s2) (strncmp(s1,"" s2 "", sizeof(s2)-1) == 0)
667 #define memBEGINs(s1, l, s2) \
668 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
669 && memEQ(s1, "" s2 "", sizeof(s2)-1))
670 #define memBEGINPs(s1, l, s2) \
671 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) - 1 \
672 && memEQ(s1, "" s2 "", sizeof(s2)-1))
673 #define memENDs(s1, l, s2) \
674 ( (Ptrdiff_t) (l) >= (Ptrdiff_t) sizeof(s2) - 1 \
675 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
676 #define memENDPs(s1, l, s2) \
677 ( (Ptrdiff_t) (l) > (Ptrdiff_t) sizeof(s2) \
678 && memEQ(s1 + (l) - (sizeof(s2) - 1), "" s2 "", sizeof(s2)-1))
679 #endif /* End of making macros private */
681 #define memLT(s1,s2,l) (memcmp(s1,s2,l) < 0)
682 #define memLE(s1,s2,l) (memcmp(s1,s2,l) <= 0)
683 #define memGT(s1,s2,l) (memcmp(s1,s2,l) > 0)
684 #define memGE(s1,s2,l) (memcmp(s1,s2,l) >= 0)
686 #define memCHRs(s1,c) ((const char *) memchr("" s1 "" , c, sizeof(s1)-1))
691 * Unfortunately, the introduction of locales means that we
692 * can't trust isupper(), etc. to tell the truth. And when
693 * it comes to /\w+/ with tainting enabled, we *must* be able
694 * to trust our character classes.
696 * Therefore, the default tests in the text of Perl will be
697 * independent of locale. Any code that wants to depend on
698 * the current locale will use the tests that begin with "lc".
701 #ifdef HAS_SETLOCALE /* XXX Is there a better test for this? */
709 =head1 Character classification
710 This section is about functions (really macros) that classify characters
711 into types, such as punctuation versus alphabetic, etc. Most of these are
712 analogous to regular expression character classes. (See
713 L<perlrecharclass/POSIX Character Classes>.) There are several variants for
714 each class. (Not all macros have all variants; each item below lists the
715 ones valid for it.) None are affected by C<use bytes>, and only the ones
716 with C<LC> in the name are affected by the current locale.
718 The base function, e.g., C<isALPHA()>, takes any signed or unsigned value,
719 treating it as a code point, and returns a boolean as to whether or not the
720 character represented by it is (or on non-ASCII platforms, corresponds to) an
721 ASCII character in the named class based on platform, Unicode, and Perl rules.
722 If the input is a number that doesn't fit in an octet, FALSE is returned.
724 Variant C<isI<FOO>_A> (e.g., C<isALPHA_A()>) is identical to the base function
725 with no suffix C<"_A">. This variant is used to emphasize by its name that
726 only ASCII-range characters can return TRUE.
728 Variant C<isI<FOO>_L1> imposes the Latin-1 (or EBCDIC equivalent) character set
729 onto the platform. That is, the code points that are ASCII are unaffected,
730 since ASCII is a subset of Latin-1. But the non-ASCII code points are treated
731 as if they are Latin-1 characters. For example, C<isWORDCHAR_L1()> will return
732 true when called with the code point 0xDF, which is a word character in both
733 ASCII and EBCDIC (though it represents different characters in each).
734 If the input is a number that doesn't fit in an octet, FALSE is returned.
735 (Perl's documentation uses a colloquial definition of Latin-1, to include all
736 code points below 256.)
738 Variant C<isI<FOO>_uvchr> is exactly like the C<isI<FOO>_L1> variant, for
739 inputs below 256, but if the code point is larger than 255, Unicode rules are
740 used to determine if it is in the character class. For example,
741 C<isWORDCHAR_uvchr(0x100)> returns TRUE, since 0x100 is LATIN CAPITAL LETTER A
742 WITH MACRON in Unicode, and is a word character.
744 Variants C<isI<FOO>_utf8> and C<isI<FOO>_utf8_safe> are like C<isI<FOO>_uvchr>,
745 but are used for UTF-8 encoded strings. The two forms are different names for
746 the same thing. Each call to one of these classifies the first character of
747 the string starting at C<p>. The second parameter, C<e>, points to anywhere in
748 the string beyond the first character, up to one byte past the end of the
749 entire string. Although both variants are identical, the suffix C<_safe> in
750 one name emphasizes that it will not attempt to read beyond S<C<e - 1>>,
751 provided that the constraint S<C<s E<lt> e>> is true (this is asserted for in
752 C<-DDEBUGGING> builds). If the UTF-8 for the input character is malformed in
753 some way, the program may croak, or the function may return FALSE, at the
754 discretion of the implementation, and subject to change in future releases.
756 Variant C<isI<FOO>_LC> is like the C<isI<FOO>_A> and C<isI<FOO>_L1> variants,
757 but the result is based on the current locale, which is what C<LC> in the name
758 stands for. If Perl can determine that the current locale is a UTF-8 locale,
759 it uses the published Unicode rules; otherwise, it uses the C library function
760 that gives the named classification. For example, C<isDIGIT_LC()> when not in
761 a UTF-8 locale returns the result of calling C<isdigit()>. FALSE is always
762 returned if the input won't fit into an octet. On some platforms where the C
763 library function is known to be defective, Perl changes its result to follow
764 the POSIX standard's rules.
766 Variant C<isI<FOO>_LC_uvchr> acts exactly like C<isI<FOO>_LC> for inputs less
767 than 256, but for larger ones it returns the Unicode classification of the code
770 Variants C<isI<FOO>_LC_utf8> and C<isI<FOO>_LC_utf8_safe> are like
771 C<isI<FOO>_LC_uvchr>, but are used for UTF-8 encoded strings. The two forms
772 are different names for the same thing. Each call to one of these classifies
773 the first character of the string starting at C<p>. The second parameter,
774 C<e>, points to anywhere in the string beyond the first character, up to one
775 byte past the end of the entire string. Although both variants are identical,
776 the suffix C<_safe> in one name emphasizes that it will not attempt to read
777 beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is true (this
778 is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the input
779 character is malformed in some way, the program may croak, or the function may
780 return FALSE, at the discretion of the implementation, and subject to change in
783 =for apidoc Am|bool|isALPHA|int ch
784 Returns a boolean indicating whether the specified input is one of C<[A-Za-z]>,
785 analogous to C<m/[[:alpha:]]/>.
786 See the L<top of this section|/Character classification> for an explanation of
788 C<isALPHA_A>, C<isALPHA_L1>, C<isALPHA_uvchr>, C<isALPHA_utf8>,
789 C<isALPHA_utf8_safe>, C<isALPHA_LC>, C<isALPHA_LC_uvchr>, C<isALPHA_LC_utf8>,
790 and C<isALPHA_LC_utf8_safe>.
794 Here and below, we add the protoypes of these macros for downstream programs
795 that would be interested in them, such as Devel::PPPort
797 =for apidoc Amh|bool|isALPHA_A|int ch
798 =for apidoc Amh|bool|isALPHA_L1|int ch
799 =for apidoc Amh|bool|isALPHA_uvchr|int ch
800 =for apidoc Amh|bool|isALPHA_utf8_safe|U8 * s|U8 * end
801 =for apidoc Amh|bool|isALPHA_utf8|U8 * s|U8 * end
802 =for apidoc Amh|bool|isALPHA_LC|int ch
803 =for apidoc Amh|bool|isALPHA_LC_uvchr|int ch
804 =for apidoc Amh|bool|isALPHA_LC_utf8_safe|U8 * s| U8 *end
806 =for apidoc Am|bool|isALPHANUMERIC|int ch
807 Returns a boolean indicating whether the specified character is one of
808 C<[A-Za-z0-9]>, analogous to C<m/[[:alnum:]]/>.
809 See the L<top of this section|/Character classification> for an explanation of
811 C<isALPHANUMERIC_A>, C<isALPHANUMERIC_L1>, C<isALPHANUMERIC_uvchr>,
812 C<isALPHANUMERIC_utf8>, C<isALPHANUMERIC_utf8_safe>, C<isALPHANUMERIC_LC>,
813 C<isALPHANUMERIC_LC_uvchr>, C<isALPHANUMERIC_LC_utf8>, and
814 C<isALPHANUMERIC_LC_utf8_safe>.
816 A (discouraged from use) synonym is C<isALNUMC> (where the C<C> suffix means
817 this corresponds to the C language alphanumeric definition). Also
818 there are the variants
819 C<isALNUMC_A>, C<isALNUMC_L1>
820 C<isALNUMC_LC>, and C<isALNUMC_LC_uvchr>.
822 =for apidoc Amh|bool|isALPHANUMERIC_A|int ch
823 =for apidoc Amh|bool|isALPHANUMERIC_L1|int ch
824 =for apidoc Amh|bool|isALPHANUMERIC_uvchr|int ch
825 =for apidoc Amh|bool|isALPHANUMERIC_utf8_safe|U8 * s|U8 * end
826 =for apidoc Amh|bool|isALPHANUMERIC_utf8|U8 * s|U8 * end
827 =for apidoc Amh|bool|isALPHANUMERIC_LC|int ch
828 =for apidoc Amh|bool|isALPHANUMERIC_LC_uvchr|int ch
829 =for apidoc Amh|bool|isALPHANUMERIC_LC_utf8_safe|U8 * s| U8 *end
830 =for apidoc Amh|bool|isALNUMC|int ch
831 =for apidoc Amh|bool|isALNUMC_A|int ch
832 =for apidoc Amh|bool|isALNUMC_L1|int ch
833 =for apidoc Amh|bool|isALNUMC_LC|int ch
834 =for apidoc Amh|bool|isALNUMC_LC_uvchr|int ch
836 =for apidoc Am|bool|isASCII|int ch
837 Returns a boolean indicating whether the specified character is one of the 128
838 characters in the ASCII character set, analogous to C<m/[[:ascii:]]/>.
839 On non-ASCII platforms, it returns TRUE iff this
840 character corresponds to an ASCII character. Variants C<isASCII_A()> and
841 C<isASCII_L1()> are identical to C<isASCII()>.
842 See the L<top of this section|/Character classification> for an explanation of
844 C<isASCII_uvchr>, C<isASCII_utf8>, C<isASCII_utf8_safe>, C<isASCII_LC>,
845 C<isASCII_LC_uvchr>, C<isASCII_LC_utf8>, and C<isASCII_LC_utf8_safe>.
846 Note, however, that some platforms do not have the C library routine
847 C<isascii()>. In these cases, the variants whose names contain C<LC> are the
848 same as the corresponding ones without.
850 =for apidoc Amh|bool|isASCII_A|int ch
851 =for apidoc Amh|bool|isASCII_L1|int ch
852 =for apidoc Amh|bool|isASCII_uvchr|int ch
853 =for apidoc Amh|bool|isASCII_utf8_safe|U8 * s|U8 * end
854 =for apidoc Amh|bool|isASCII_utf8|U8 * s|U8 * end
855 =for apidoc Amh|bool|isASCII_LC|int ch
856 =for apidoc Amh|bool|isASCII_LC_uvchr|int ch
857 =for apidoc Amh|bool|isASCII_LC_utf8_safe|U8 * s| U8 *end
859 Also note, that because all ASCII characters are UTF-8 invariant (meaning they
860 have the exact same representation (always a single byte) whether encoded in
861 UTF-8 or not), C<isASCII> will give the correct results when called with any
862 byte in any string encoded or not in UTF-8. And similarly C<isASCII_utf8> and
863 C<isASCII_utf8_safe> will work properly on any string encoded or not in UTF-8.
865 =for apidoc Am|bool|isBLANK|char ch
866 Returns a boolean indicating whether the specified character is a
867 character considered to be a blank, analogous to C<m/[[:blank:]]/>.
868 See the L<top of this section|/Character classification> for an explanation of
870 C<isBLANK_A>, C<isBLANK_L1>, C<isBLANK_uvchr>, C<isBLANK_utf8>,
871 C<isBLANK_utf8_safe>, C<isBLANK_LC>, C<isBLANK_LC_uvchr>, C<isBLANK_LC_utf8>,
872 and C<isBLANK_LC_utf8_safe>. Note,
873 however, that some platforms do not have the C library routine
874 C<isblank()>. In these cases, the variants whose names contain C<LC> are
875 the same as the corresponding ones without.
877 =for apidoc Amh|bool|isBLANK_A|int ch
878 =for apidoc Amh|bool|isBLANK_L1|int ch
879 =for apidoc Amh|bool|isBLANK_uvchr|int ch
880 =for apidoc Amh|bool|isBLANK_utf8_safe|U8 * s|U8 * end
881 =for apidoc Amh|bool|isBLANK_utf8|U8 * s|U8 * end
882 =for apidoc Amh|bool|isBLANK_LC|int ch
883 =for apidoc Amh|bool|isBLANK_LC_uvchr|int ch
884 =for apidoc Amh|bool|isBLANK_LC_utf8_safe|U8 * s| U8 *end
886 =for apidoc Am|bool|isCNTRL|char ch
887 Returns a boolean indicating whether the specified character is a
888 control character, analogous to C<m/[[:cntrl:]]/>.
889 See the L<top of this section|/Character classification> for an explanation of
891 C<isCNTRL_A>, C<isCNTRL_L1>, C<isCNTRL_uvchr>, C<isCNTRL_utf8>,
892 C<isCNTRL_utf8_safe>, C<isCNTRL_LC>, C<isCNTRL_LC_uvchr>, C<isCNTRL_LC_utf8>
893 and C<isCNTRL_LC_utf8_safe>. On EBCDIC
894 platforms, you almost always want to use the C<isCNTRL_L1> variant.
896 =for apidoc Amh|bool|isCNTRL_A|int ch
897 =for apidoc Amh|bool|isCNTRL_L1|int ch
898 =for apidoc Amh|bool|isCNTRL_uvchr|int ch
899 =for apidoc Amh|bool|isCNTRL_utf8_safe|U8 * s|U8 * end
900 =for apidoc Amh|bool|isCNTRL_utf8|U8 * s|U8 * end
901 =for apidoc Amh|bool|isCNTRL_LC|int ch
902 =for apidoc Amh|bool|isCNTRL_LC_uvchr|int ch
903 =for apidoc Amh|bool|isCNTRL_LC_utf8_safe|U8 * s| U8 *end
905 =for apidoc Am|bool|isDIGIT|char ch
906 Returns a boolean indicating whether the specified character is a
907 digit, analogous to C<m/[[:digit:]]/>.
908 Variants C<isDIGIT_A> and C<isDIGIT_L1> are identical to C<isDIGIT>.
909 See the L<top of this section|/Character classification> for an explanation of
911 C<isDIGIT_uvchr>, C<isDIGIT_utf8>, C<isDIGIT_utf8_safe>, C<isDIGIT_LC>,
912 C<isDIGIT_LC_uvchr>, C<isDIGIT_LC_utf8>, and C<isDIGIT_LC_utf8_safe>.
914 =for apidoc Amh|bool|isDIGIT_A|int ch
915 =for apidoc Amh|bool|isDIGIT_L1|int ch
916 =for apidoc Amh|bool|isDIGIT_uvchr|int ch
917 =for apidoc Amh|bool|isDIGIT_utf8_safe|U8 * s|U8 * end
918 =for apidoc Amh|bool|isDIGIT_utf8|U8 * s|U8 * end
919 =for apidoc Amh|bool|isDIGIT_LC|int ch
920 =for apidoc Amh|bool|isDIGIT_LC_uvchr|int ch
921 =for apidoc Amh|bool|isDIGIT_LC_utf8_safe|U8 * s| U8 *end
923 =for apidoc Am|bool|isGRAPH|char ch
924 Returns a boolean indicating whether the specified character is a
925 graphic character, analogous to C<m/[[:graph:]]/>.
926 See the L<top of this section|/Character classification> for an explanation of
927 variants C<isGRAPH_A>, C<isGRAPH_L1>, C<isGRAPH_uvchr>, C<isGRAPH_utf8>,
928 C<isGRAPH_utf8_safe>, C<isGRAPH_LC>, C<isGRAPH_LC_uvchr>,
929 C<isGRAPH_LC_utf8_safe>, and C<isGRAPH_LC_utf8_safe>.
931 =for apidoc Amh|bool|isGRAPH_A|int ch
932 =for apidoc Amh|bool|isGRAPH_L1|int ch
933 =for apidoc Amh|bool|isGRAPH_uvchr|int ch
934 =for apidoc Amh|bool|isGRAPH_utf8_safe|U8 * s|U8 * end
935 =for apidoc Amh|bool|isGRAPH_utf8|U8 * s|U8 * end
936 =for apidoc Amh|bool|isGRAPH_LC|int ch
937 =for apidoc Amh|bool|isGRAPH_LC_uvchr|int ch
938 =for apidoc Amh|bool|isGRAPH_LC_utf8_safe|U8 * s| U8 *end
940 =for apidoc Am|bool|isLOWER|char ch
941 Returns a boolean indicating whether the specified character is a
942 lowercase character, analogous to C<m/[[:lower:]]/>.
943 See the L<top of this section|/Character classification> for an explanation of
945 C<isLOWER_A>, C<isLOWER_L1>, C<isLOWER_uvchr>, C<isLOWER_utf8>,
946 C<isLOWER_utf8_safe>, C<isLOWER_LC>, C<isLOWER_LC_uvchr>, C<isLOWER_LC_utf8>,
947 and C<isLOWER_LC_utf8_safe>.
949 =for apidoc Amh|bool|isLOWER_A|int ch
950 =for apidoc Amh|bool|isLOWER_L1|int ch
951 =for apidoc Amh|bool|isLOWER_uvchr|int ch
952 =for apidoc Amh|bool|isLOWER_utf8_safe|U8 * s|U8 * end
953 =for apidoc Amh|bool|isLOWER_utf8|U8 * s|U8 * end
954 =for apidoc Amh|bool|isLOWER_LC|int ch
955 =for apidoc Amh|bool|isLOWER_LC_uvchr|int ch
956 =for apidoc Amh|bool|isLOWER_LC_utf8_safe|U8 * s| U8 *end
958 =for apidoc Am|bool|isOCTAL|char ch
959 Returns a boolean indicating whether the specified character is an
961 The only two variants are C<isOCTAL_A> and C<isOCTAL_L1>; each is identical to
964 =for apidoc Amh|bool|isOCTAL_A|int ch
965 =for apidoc Amh|bool|isOCTAL_L1|int ch
967 =for apidoc Am|bool|isPUNCT|char ch
968 Returns a boolean indicating whether the specified character is a
969 punctuation character, analogous to C<m/[[:punct:]]/>.
970 Note that the definition of what is punctuation isn't as
971 straightforward as one might desire. See L<perlrecharclass/POSIX Character
972 Classes> for details.
973 See the L<top of this section|/Character classification> for an explanation of
974 variants C<isPUNCT_A>, C<isPUNCT_L1>, C<isPUNCT_uvchr>, C<isPUNCT_utf8>,
975 C<isPUNCT_utf8_safe>, C<isPUNCT_LC>, C<isPUNCT_LC_uvchr>, C<isPUNCT_LC_utf8>,
976 and C<isPUNCT_LC_utf8_safe>.
978 =for apidoc Amh|bool|isPUNCT_A|int ch
979 =for apidoc Amh|bool|isPUNCT_L1|int ch
980 =for apidoc Amh|bool|isPUNCT_uvchr|int ch
981 =for apidoc Amh|bool|isPUNCT_utf8_safe|U8 * s|U8 * end
982 =for apidoc Amh|bool|isPUNCT_utf8|U8 * s|U8 * end
983 =for apidoc Amh|bool|isPUNCT_LC|int ch
984 =for apidoc Amh|bool|isPUNCT_LC_uvchr|int ch
985 =for apidoc Amh|bool|isPUNCT_LC_utf8_safe|U8 * s| U8 *end
987 =for apidoc Am|bool|isSPACE|char ch
988 Returns a boolean indicating whether the specified character is a
989 whitespace character. This is analogous
990 to what C<m/\s/> matches in a regular expression. Starting in Perl 5.18
991 this also matches what C<m/[[:space:]]/> does. Prior to 5.18, only the
992 locale forms of this macro (the ones with C<LC> in their names) matched
993 precisely what C<m/[[:space:]]/> does. In those releases, the only difference,
994 in the non-locale variants, was that C<isSPACE()> did not match a vertical tab.
995 (See L</isPSXSPC> for a macro that matches a vertical tab in all releases.)
996 See the L<top of this section|/Character classification> for an explanation of
998 C<isSPACE_A>, C<isSPACE_L1>, C<isSPACE_uvchr>, C<isSPACE_utf8>,
999 C<isSPACE_utf8_safe>, C<isSPACE_LC>, C<isSPACE_LC_uvchr>, C<isSPACE_LC_utf8>,
1000 and C<isSPACE_LC_utf8_safe>.
1002 =for apidoc Amh|bool|isSPACE_A|int ch
1003 =for apidoc Amh|bool|isSPACE_L1|int ch
1004 =for apidoc Amh|bool|isSPACE_uvchr|int ch
1005 =for apidoc Amh|bool|isSPACE_utf8_safe|U8 * s|U8 * end
1006 =for apidoc Amh|bool|isSPACE_utf8|U8 * s|U8 * end
1007 =for apidoc Amh|bool|isSPACE_LC|int ch
1008 =for apidoc Amh|bool|isSPACE_LC_uvchr|int ch
1009 =for apidoc Amh|bool|isSPACE_LC_utf8_safe|U8 * s| U8 *end
1011 =for apidoc Am|bool|isPSXSPC|char ch
1012 (short for Posix Space)
1013 Starting in 5.18, this is identical in all its forms to the
1014 corresponding C<isSPACE()> macros.
1015 The locale forms of this macro are identical to their corresponding
1016 C<isSPACE()> forms in all Perl releases. In releases prior to 5.18, the
1017 non-locale forms differ from their C<isSPACE()> forms only in that the
1018 C<isSPACE()> forms don't match a Vertical Tab, and the C<isPSXSPC()> forms do.
1019 Otherwise they are identical. Thus this macro is analogous to what
1020 C<m/[[:space:]]/> matches in a regular expression.
1021 See the L<top of this section|/Character classification> for an explanation of
1022 variants C<isPSXSPC_A>, C<isPSXSPC_L1>, C<isPSXSPC_uvchr>, C<isPSXSPC_utf8>,
1023 C<isPSXSPC_utf8_safe>, C<isPSXSPC_LC>, C<isPSXSPC_LC_uvchr>,
1024 C<isPSXSPC_LC_utf8>, and C<isPSXSPC_LC_utf8_safe>.
1026 =for apidoc Amh|bool|isPSXSPC_A|int ch
1027 =for apidoc Amh|bool|isPSXSPC_L1|int ch
1028 =for apidoc Amh|bool|isPSXSPC_uvchr|int ch
1029 =for apidoc Amh|bool|isPSXSPC_utf8_safe|U8 * s|U8 * end
1030 =for apidoc Amh|bool|isPSXSPC_utf8|U8 * s|U8 * end
1031 =for apidoc Amh|bool|isPSXSPC_LC|int ch
1032 =for apidoc Amh|bool|isPSXSPC_LC_uvchr|int ch
1033 =for apidoc Amh|bool|isPSXSPC_LC_utf8_safe|U8 * s| U8 *end
1035 =for apidoc Am|bool|isUPPER|char ch
1036 Returns a boolean indicating whether the specified character is an
1037 uppercase character, analogous to C<m/[[:upper:]]/>.
1038 See the L<top of this section|/Character classification> for an explanation of
1039 variants C<isUPPER_A>, C<isUPPER_L1>, C<isUPPER_uvchr>, C<isUPPER_utf8>,
1040 C<isUPPER_utf8_safe>, C<isUPPER_LC>, C<isUPPER_LC_uvchr>, C<isUPPER_LC_utf8>,
1041 and C<isUPPER_LC_utf8_safe>.
1043 =for apidoc Amh|bool|isUPPER_A|int ch
1044 =for apidoc Amh|bool|isUPPER_L1|int ch
1045 =for apidoc Amh|bool|isUPPER_uvchr|int ch
1046 =for apidoc Amh|bool|isUPPER_utf8_safe|U8 * s|U8 * end
1047 =for apidoc Amh|bool|isUPPER_utf8|U8 * s|U8 * end
1048 =for apidoc Amh|bool|isUPPER_LC|int ch
1049 =for apidoc Amh|bool|isUPPER_LC_uvchr|int ch
1050 =for apidoc Amh|bool|isUPPER_LC_utf8_safe|U8 * s| U8 *end
1052 =for apidoc Am|bool|isPRINT|char ch
1053 Returns a boolean indicating whether the specified character is a
1054 printable character, analogous to C<m/[[:print:]]/>.
1055 See the L<top of this section|/Character classification> for an explanation of
1057 C<isPRINT_A>, C<isPRINT_L1>, C<isPRINT_uvchr>, C<isPRINT_utf8>,
1058 C<isPRINT_utf8_safe>, C<isPRINT_LC>, C<isPRINT_LC_uvchr>, C<isPRINT_LC_utf8>,
1059 and C<isPRINT_LC_utf8_safe>.
1061 =for apidoc Amh|bool|isPRINT_A|int ch
1062 =for apidoc Amh|bool|isPRINT_L1|int ch
1063 =for apidoc Amh|bool|isPRINT_uvchr|int ch
1064 =for apidoc Amh|bool|isPRINT_utf8_safe|U8 * s|U8 * end
1065 =for apidoc Amh|bool|isPRINT_utf8|U8 * s|U8 * end
1066 =for apidoc Amh|bool|isPRINT_LC|int ch
1067 =for apidoc Amh|bool|isPRINT_LC_uvchr|int ch
1068 =for apidoc Amh|bool|isPRINT_LC_utf8_safe|U8 * s| U8 *end
1070 =for apidoc Am|bool|isWORDCHAR|char ch
1071 Returns a boolean indicating whether the specified character is a character
1072 that is a word character, analogous to what C<m/\w/> and C<m/[[:word:]]/> match
1073 in a regular expression. A word character is an alphabetic character, a
1074 decimal digit, a connecting punctuation character (such as an underscore), or
1075 a "mark" character that attaches to one of those (like some sort of accent).
1076 C<isALNUM()> is a synonym provided for backward compatibility, even though a
1077 word character includes more than the standard C language meaning of
1079 See the L<top of this section|/Character classification> for an explanation of
1080 variants C<isWORDCHAR_A>, C<isWORDCHAR_L1>, C<isWORDCHAR_uvchr>,
1081 C<isWORDCHAR_utf8>, and C<isWORDCHAR_utf8_safe>. C<isWORDCHAR_LC>,
1082 C<isWORDCHAR_LC_uvchr>, C<isWORDCHAR_LC_utf8>, and C<isWORDCHAR_LC_utf8_safe>
1083 are also as described there, but additionally include the platform's native
1086 =for apidoc Amh|bool|isWORDCHAR_A|int ch
1087 =for apidoc Amh|bool|isWORDCHAR_L1|int ch
1088 =for apidoc Amh|bool|isWORDCHAR_uvchr|int ch
1089 =for apidoc Amh|bool|isWORDCHAR_utf8_safe|U8 * s|U8 * end
1090 =for apidoc Amh|bool|isWORDCHAR_utf8|U8 * s|U8 * end
1091 =for apidoc Amh|bool|isWORDCHAR_LC|int ch
1092 =for apidoc Amh|bool|isWORDCHAR_LC_uvchr|int ch
1093 =for apidoc Amh|bool|isWORDCHAR_LC_utf8_safe|U8 * s| U8 *end
1094 =for apidoc Amh|bool|isALNUM|int ch
1095 =for apidoc Amh|bool|isALNUM_A|int ch
1096 =for apidoc Amh|bool|isALNUM_LC|int ch
1097 =for apidoc Amh|bool|isALNUM_LC_uvchr|int ch
1099 =for apidoc Am|bool|isXDIGIT|char ch
1100 Returns a boolean indicating whether the specified character is a hexadecimal
1101 digit. In the ASCII range these are C<[0-9A-Fa-f]>. Variants C<isXDIGIT_A()>
1102 and C<isXDIGIT_L1()> are identical to C<isXDIGIT()>.
1103 See the L<top of this section|/Character classification> for an explanation of
1105 C<isXDIGIT_uvchr>, C<isXDIGIT_utf8>, C<isXDIGIT_utf8_safe>, C<isXDIGIT_LC>,
1106 C<isXDIGIT_LC_uvchr>, C<isXDIGIT_LC_utf8>, and C<isXDIGIT_LC_utf8_safe>.
1108 =for apidoc Amh|bool|isXDIGIT_A|int ch
1109 =for apidoc Amh|bool|isXDIGIT_L1|int ch
1110 =for apidoc Amh|bool|isXDIGIT_uvchr|int ch
1111 =for apidoc Amh|bool|isXDIGIT_utf8_safe|U8 * s|U8 * end
1112 =for apidoc Amh|bool|isXDIGIT_utf8|U8 * s|U8 * end
1113 =for apidoc Amh|bool|isXDIGIT_LC|int ch
1114 =for apidoc Amh|bool|isXDIGIT_LC_uvchr|int ch
1115 =for apidoc Amh|bool|isXDIGIT_LC_utf8_safe|U8 * s| U8 *end
1117 =for apidoc Am|bool|isIDFIRST|char ch
1118 Returns a boolean indicating whether the specified character can be the first
1119 character of an identifier. This is very close to, but not quite the same as
1120 the official Unicode property C<XID_Start>. The difference is that this
1121 returns true only if the input character also matches L</isWORDCHAR>.
1122 See the L<top of this section|/Character classification> for an explanation of
1124 C<isIDFIRST_A>, C<isIDFIRST_L1>, C<isIDFIRST_uvchr>, C<isIDFIRST_utf8>,
1125 C<isIDFIRST_utf8_safe>, C<isIDFIRST_LC>, C<isIDFIRST_LC_uvchr>,
1126 C<isIDFIRST_LC_utf8>, and C<isIDFIRST_LC_utf8_safe>.
1128 =for apidoc Amh|bool|isIDFIRST_A|int ch
1129 =for apidoc Amh|bool|isIDFIRST_L1|int ch
1130 =for apidoc Amh|bool|isIDFIRST_uvchr|int ch
1131 =for apidoc Amh|bool|isIDFIRST_utf8_safe|U8 * s|U8 * end
1132 =for apidoc Amh|bool|isIDFIRST_utf8|U8 * s|U8 * end
1133 =for apidoc Amh|bool|isIDFIRST_LC|int ch
1134 =for apidoc Amh|bool|isIDFIRST_LC_uvchr|int ch
1135 =for apidoc Amh|bool|isIDFIRST_LC_utf8_safe|U8 * s| U8 *end
1137 =for apidoc Am|bool|isIDCONT|char ch
1138 Returns a boolean indicating whether the specified character can be the
1139 second or succeeding character of an identifier. This is very close to, but
1140 not quite the same as the official Unicode property C<XID_Continue>. The
1141 difference is that this returns true only if the input character also matches
1142 L</isWORDCHAR>. See the L<top of this section|/Character classification> for
1143 an explanation of variants C<isIDCONT_A>, C<isIDCONT_L1>, C<isIDCONT_uvchr>,
1144 C<isIDCONT_utf8>, C<isIDCONT_utf8_safe>, C<isIDCONT_LC>, C<isIDCONT_LC_uvchr>,
1145 C<isIDCONT_LC_utf8>, and C<isIDCONT_LC_utf8_safe>.
1147 =for apidoc Amh|bool|isIDCONT_A|int ch
1148 =for apidoc Amh|bool|isIDCONT_L1|int ch
1149 =for apidoc Amh|bool|isIDCONT_uvchr|int ch
1150 =for apidoc Amh|bool|isIDCONT_utf8_safe|U8 * s|U8 * end
1151 =for apidoc Amh|bool|isIDCONT_utf8|U8 * s|U8 * end
1152 =for apidoc Amh|bool|isIDCONT_LC|int ch
1153 =for apidoc Amh|bool|isIDCONT_LC_uvchr|int ch
1154 =for apidoc Amh|bool|isIDCONT_LC_utf8_safe|U8 * s| U8 *end
1156 =head1 Miscellaneous Functions
1158 =for apidoc Am|U8|READ_XDIGIT|char str*
1159 Returns the value of an ASCII-range hex digit and advances the string pointer.
1160 Behaviour is only well defined when isXDIGIT(*str) is true.
1162 =head1 Character case changing
1163 Perl uses "full" Unicode case mappings. This means that converting a single
1164 character to another case may result in a sequence of more than one character.
1165 For example, the uppercase of C<E<223>> (LATIN SMALL LETTER SHARP S) is the two
1166 character sequence C<SS>. This presents some complications The lowercase of
1167 all characters in the range 0..255 is a single character, and thus
1168 C<L</toLOWER_L1>> is furnished. But, C<toUPPER_L1> can't exist, as it couldn't
1169 return a valid result for all legal inputs. Instead C<L</toUPPER_uvchr>> has
1170 an API that does allow every possible legal result to be returned.) Likewise
1171 no other function that is crippled by not being able to give the correct
1172 results for the full range of possible inputs has been implemented here.
1174 =for apidoc Am|U8|toUPPER|int ch
1175 Converts the specified character to uppercase. If the input is anything but an
1176 ASCII lowercase character, that input character itself is returned. Variant
1177 C<toUPPER_A> is equivalent.
1179 =for apidoc Am|UV|toUPPER_uvchr|UV cp|U8* s|STRLEN* lenp
1180 Converts the code point C<cp> to its uppercase version, and
1181 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1182 point is interpreted as native if less than 256; otherwise as Unicode. Note
1183 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1184 bytes since the uppercase version may be longer than the original character.
1186 The first code point of the uppercased version is returned
1187 (but note, as explained at L<the top of this section|/Character case
1188 changing>, that there may be more.)
1190 =for apidoc Am|UV|toUPPER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1191 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1192 extending no further than S<C<e - 1>> to its uppercase version, and
1193 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1194 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1195 bytes since the uppercase version may be longer than the original character.
1197 The first code point of the uppercased version is returned
1198 (but note, as explained at L<the top of this section|/Character case
1199 changing>, that there may be more).
1201 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1202 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1203 the UTF-8 for the input character is malformed in some way, the program may
1204 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1205 of the implementation, and subject to change in future releases.
1207 =for apidoc Am|UV|toUPPER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1208 Same as L</toUPPER_utf8>.
1210 =for apidoc Am|U8|toFOLD|U8 ch
1211 Converts the specified character to foldcase. If the input is anything but an
1212 ASCII uppercase character, that input character itself is returned. Variant
1213 C<toFOLD_A> is equivalent. (There is no equivalent C<to_FOLD_L1> for the full
1214 Latin1 range, as the full generality of L</toFOLD_uvchr> is needed there.)
1216 =for apidoc Am|UV|toFOLD_uvchr|UV cp|U8* s|STRLEN* lenp
1217 Converts the code point C<cp> to its foldcase version, and
1218 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1219 point is interpreted as native if less than 256; otherwise as Unicode. Note
1220 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1221 bytes since the foldcase version may be longer than the original character.
1223 The first code point of the foldcased version is returned
1224 (but note, as explained at L<the top of this section|/Character case
1225 changing>, that there may be more).
1227 =for apidoc Am|UV|toFOLD_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1228 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1229 extending no further than S<C<e - 1>> to its foldcase version, and
1230 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1231 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1232 bytes since the foldcase version may be longer than the original character.
1234 The first code point of the foldcased version is returned
1235 (but note, as explained at L<the top of this section|/Character case
1236 changing>, that there may be more).
1239 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1240 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1241 input character is malformed in some way, the program may croak, or the
1242 function may return the REPLACEMENT CHARACTER, at the discretion of the
1243 implementation, and subject to change in future releases.
1245 =for apidoc Am|UV|toFOLD_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1246 Same as L</toFOLD_utf8>.
1248 =for apidoc Am|U8|toLOWER|U8 ch
1249 Converts the specified character to lowercase. If the input is anything but an
1250 ASCII uppercase character, that input character itself is returned. Variant
1251 C<toLOWER_A> is equivalent.
1253 =for apidoc Am|U8|toLOWER_L1|U8 ch
1254 Converts the specified Latin1 character to lowercase. The results are
1255 undefined if the input doesn't fit in a byte.
1257 =for apidoc Am|U8|toLOWER_LC|U8 ch
1258 Converts the specified character to lowercase using the current locale's rules,
1259 if possible; otherwise returns the input character itself.
1261 =for apidoc Am|UV|toLOWER_uvchr|UV cp|U8* s|STRLEN* lenp
1262 Converts the code point C<cp> to its lowercase version, and
1263 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1264 point is interpreted as native if less than 256; otherwise as Unicode. Note
1265 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1266 bytes since the lowercase version may be longer than the original character.
1268 The first code point of the lowercased version is returned
1269 (but note, as explained at L<the top of this section|/Character case
1270 changing>, that there may be more).
1272 =for apidoc Am|UV|toLOWER_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1273 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1274 extending no further than S<C<e - 1>> to its lowercase version, and
1275 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1276 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1277 bytes since the lowercase version may be longer than the original character.
1279 The first code point of the lowercased version is returned
1280 (but note, as explained at L<the top of this section|/Character case
1281 changing>, that there may be more).
1282 It will not attempt to read beyond S<C<e - 1>>, provided that the constraint
1283 S<C<s E<lt> e>> is true (this is asserted for in C<-DDEBUGGING> builds). If
1284 the UTF-8 for the input character is malformed in some way, the program may
1285 croak, or the function may return the REPLACEMENT CHARACTER, at the discretion
1286 of the implementation, and subject to change in future releases.
1288 =for apidoc Am|UV|toLOWER_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1289 Same as L</toLOWER_utf8>.
1291 =for apidoc Am|U8|toTITLE|U8 ch
1292 Converts the specified character to titlecase. If the input is anything but an
1293 ASCII lowercase character, that input character itself is returned. Variant
1294 C<toTITLE_A> is equivalent. (There is no C<toTITLE_L1> for the full Latin1
1295 range, as the full generality of L</toTITLE_uvchr> is needed there. Titlecase is
1296 not a concept used in locale handling, so there is no functionality for that.)
1298 =for apidoc Am|UV|toTITLE_uvchr|UV cp|U8* s|STRLEN* lenp
1299 Converts the code point C<cp> to its titlecase version, and
1300 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. The code
1301 point is interpreted as native if less than 256; otherwise as Unicode. Note
1302 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1303 bytes since the titlecase version may be longer than the original character.
1305 The first code point of the titlecased version is returned
1306 (but note, as explained at L<the top of this section|/Character case
1307 changing>, that there may be more).
1309 =for apidoc Am|UV|toTITLE_utf8|U8* p|U8* e|U8* s|STRLEN* lenp
1310 Converts the first UTF-8 encoded character in the sequence starting at C<p> and
1311 extending no further than S<C<e - 1>> to its titlecase version, and
1312 stores that in UTF-8 in C<s>, and its length in bytes in C<lenp>. Note
1313 that the buffer pointed to by C<s> needs to be at least C<UTF8_MAXBYTES_CASE+1>
1314 bytes since the titlecase version may be longer than the original character.
1316 The first code point of the titlecased version is returned
1317 (but note, as explained at L<the top of this section|/Character case
1318 changing>, that there may be more).
1321 to read beyond S<C<e - 1>>, provided that the constraint S<C<s E<lt> e>> is
1322 true (this is asserted for in C<-DDEBUGGING> builds). If the UTF-8 for the
1323 input character is malformed in some way, the program may croak, or the
1324 function may return the REPLACEMENT CHARACTER, at the discretion of the
1325 implementation, and subject to change in future releases.
1327 =for apidoc Am|UV|toTITLE_utf8_safe|U8* p|U8* e|U8* s|STRLEN* lenp
1328 Same as L</toTITLE_utf8>.
1332 XXX Still undocumented isVERTWS_uvchr and _utf8; it's unclear what their names
1333 really should be. Also toUPPER_LC and toFOLD_LC, which are subject to change,
1334 and aren't general purpose as they don't work on U+DF, and assert against that.
1336 Note that these macros are repeated in Devel::PPPort, so should also be
1337 patched there. The file as of this writing is cpan/Devel-PPPort/parts/inc/misc
1342 void below because that's the best fit, and works for Devel::PPPort
1343 =for apidoc AmnU|void|WIDEST_UTYPE
1345 Yields the widest unsigned integer type on the platform, currently either
1346 C<U32> or C<64>. This can be used in declarations such as
1352 my_uv = (WIDEST_UTYPE) val;
1358 # define WIDEST_UTYPE U64
1360 # define WIDEST_UTYPE U32
1363 /* FITS_IN_8_BITS(c) returns true if c doesn't have a bit set other than in
1364 * the lower 8. It is designed to be hopefully bomb-proof, making sure that no
1365 * bits of information are lost even on a 64-bit machine, but to get the
1366 * compiler to optimize it out if possible. This is because Configure makes
1367 * sure that the machine has an 8-bit byte, so if c is stored in a byte, the
1368 * sizeof() guarantees that this evaluates to a constant true at compile time.
1370 * For Coverity, be always true, because otherwise Coverity thinks
1371 * it finds several expressions that are always true, independent
1372 * of operands. Well, they are, but that is kind of the point.
1374 #ifndef __COVERITY__
1375 /* The '| 0' part ensures a compiler error if c is not integer (like e.g., a
1377 #define FITS_IN_8_BITS(c) ( (sizeof(c) == 1) \
1378 || !(((WIDEST_UTYPE)((c) | 0)) & ~0xFF))
1380 #define FITS_IN_8_BITS(c) (1)
1383 /* Returns true if l <= c <= (l + n), where 'l' and 'n' are non-negative
1384 * Written this way so that after optimization, only one conditional test is
1385 * needed. (The NV casts stop any warnings about comparison always being true
1386 * if called with an unsigned. The cast preserves the sign, which is all we
1388 #define withinCOUNT(c, l, n) (__ASSERT_((NV) (l) >= 0) \
1389 __ASSERT_((NV) (n) >= 0) \
1390 (((WIDEST_UTYPE) (((c)) - ((l) | 0))) <= (((WIDEST_UTYPE) ((n) | 0)))))
1392 /* Returns true if c is in the range l..u, where 'l' is non-negative
1393 * Written this way so that after optimization, only one conditional test is
1395 #define inRANGE(c, l, u) (__ASSERT_((u) >= (l)) \
1396 ( (sizeof(c) == sizeof(U8)) ? withinCOUNT(((U8) (c)), (l), ((u) - (l))) \
1397 : (sizeof(c) == sizeof(U32)) ? withinCOUNT(((U32) (c)), (l), ((u) - (l))) \
1398 : (__ASSERT_(sizeof(c) == sizeof(WIDEST_UTYPE)) \
1399 withinCOUNT(((WIDEST_UTYPE) (c)), (l), ((u) - (l))))))
1402 # ifndef _ALL_SOURCE
1403 /* The native libc isascii() et.al. functions return the wrong results
1404 * on at least z/OS unless this is defined. */
1405 # error _ALL_SOURCE should probably be defined
1408 /* There is a simple definition of ASCII for ASCII platforms. But the
1409 * EBCDIC one isn't so simple, so is defined using table look-up like the
1410 * other macros below.
1412 * The cast here is used instead of '(c) >= 0', because some compilers emit
1413 * a warning that that test is always true when the parameter is an
1414 * unsigned type. khw supposes that it could be written as
1415 * && ((c) == '\0' || (c) > 0)
1416 * to avoid the message, but the cast will likely avoid extra branches even
1417 * with stupid compilers.
1419 * The '| 0' part ensures a compiler error if c is not integer (like e.g.,
1421 # define isASCII(c) ((WIDEST_UTYPE)((c) | 0) < 128)
1424 /* Take the eight possible bit patterns of the lower 3 bits and you get the
1425 * lower 3 bits of the 8 octal digits, in both ASCII and EBCDIC, so those bits
1426 * can be ignored. If the rest match '0', we have an octal */
1427 #define isOCTAL_A(c) (((WIDEST_UTYPE)((c) | 0) & ~7) == '0')
1429 #ifdef H_PERL /* If have access to perl.h, lookup in its table */
1431 /* Character class numbers. For internal core Perl use only. The ones less
1432 * than 32 are used in PL_charclass[] and the ones up through the one that
1433 * corresponds to <_HIGHEST_REGCOMP_DOT_H_SYNC> are used by regcomp.h and
1434 * related files. PL_charclass ones use names used in l1_char_class_tab.h but
1435 * their actual definitions are here. If that file has a name not used here,
1438 * The first group of these is ordered in what I (khw) estimate to be the
1439 * frequency of their use. This gives a slight edge to exiting a loop earlier
1440 * (in reginclass() in regexec.c). Except \v should be last, as it isn't a
1441 * real Posix character class, and some (small) inefficiencies in regular
1442 * expression handling would be introduced by putting it in the middle of those
1443 * that are. Also, cntrl and ascii come after the others as it may be useful
1444 * to group these which have no members that match above Latin1, (or above
1445 * ASCII in the latter case) */
1447 # define _CC_WORDCHAR 0 /* \w and [:word:] */
1448 # define _CC_DIGIT 1 /* \d and [:digit:] */
1449 # define _CC_ALPHA 2 /* [:alpha:] */
1450 # define _CC_LOWER 3 /* [:lower:] */
1451 # define _CC_UPPER 4 /* [:upper:] */
1452 # define _CC_PUNCT 5 /* [:punct:] */
1453 # define _CC_PRINT 6 /* [:print:] */
1454 # define _CC_ALPHANUMERIC 7 /* [:alnum:] */
1455 # define _CC_GRAPH 8 /* [:graph:] */
1456 # define _CC_CASED 9 /* [:lower:] or [:upper:] under /i */
1457 # define _CC_SPACE 10 /* \s, [:space:] */
1458 # define _CC_BLANK 11 /* [:blank:] */
1459 # define _CC_XDIGIT 12 /* [:xdigit:] */
1460 # define _CC_CNTRL 13 /* [:cntrl:] */
1461 # define _CC_ASCII 14 /* [:ascii:] */
1462 # define _CC_VERTSPACE 15 /* \v */
1464 # define _HIGHEST_REGCOMP_DOT_H_SYNC _CC_VERTSPACE
1466 /* The members of the third group below do not need to be coordinated with data
1467 * structures in regcomp.[ch] and regexec.c. */
1468 # define _CC_IDFIRST 16
1469 # define _CC_CHARNAME_CONT 17
1470 # define _CC_NONLATIN1_FOLD 18
1471 # define _CC_NONLATIN1_SIMPLE_FOLD 19
1472 # define _CC_QUOTEMETA 20
1473 # define _CC_NON_FINAL_FOLD 21
1474 # define _CC_IS_IN_SOME_FOLD 22
1475 # define _CC_BINDIGIT 23
1476 # define _CC_OCTDIGIT 24
1477 # define _CC_MNEMONIC_CNTRL 25
1479 /* This next group is only used on EBCDIC platforms, so theoretically could be
1480 * shared with something entirely different that's only on ASCII platforms */
1481 # define _CC_UTF8_START_BYTE_IS_FOR_AT_LEAST_SURROGATE 31
1483 * If more bits are needed, one could add a second word for non-64bit
1484 * QUAD_IS_INT systems, using some #ifdefs to distinguish between having a 2nd
1485 * word or not. The IS_IN_SOME_FOLD bit is the most easily expendable, as it
1486 * is used only for optimization (as of this writing), and differs in the
1487 * Latin1 range from the ALPHA bit only in two relatively unimportant
1488 * characters: the masculine and feminine ordinal indicators, so removing it
1489 * would just cause /i regexes which match them to run less efficiently.
1490 * Similarly the EBCDIC-only bits are used just for speed, and could be
1491 * replaced by other means */
1493 #if defined(PERL_CORE) || defined(PERL_EXT)
1494 /* An enum version of the character class numbers, to help compilers
1497 _CC_ENUM_ALPHA = _CC_ALPHA,
1498 _CC_ENUM_ALPHANUMERIC = _CC_ALPHANUMERIC,
1499 _CC_ENUM_ASCII = _CC_ASCII,
1500 _CC_ENUM_BLANK = _CC_BLANK,
1501 _CC_ENUM_CASED = _CC_CASED,
1502 _CC_ENUM_CNTRL = _CC_CNTRL,
1503 _CC_ENUM_DIGIT = _CC_DIGIT,
1504 _CC_ENUM_GRAPH = _CC_GRAPH,
1505 _CC_ENUM_LOWER = _CC_LOWER,
1506 _CC_ENUM_PRINT = _CC_PRINT,
1507 _CC_ENUM_PUNCT = _CC_PUNCT,
1508 _CC_ENUM_SPACE = _CC_SPACE,
1509 _CC_ENUM_UPPER = _CC_UPPER,
1510 _CC_ENUM_VERTSPACE = _CC_VERTSPACE,
1511 _CC_ENUM_WORDCHAR = _CC_WORDCHAR,
1512 _CC_ENUM_XDIGIT = _CC_XDIGIT
1513 } _char_class_number;
1516 #define POSIX_CC_COUNT (_HIGHEST_REGCOMP_DOT_H_SYNC + 1)
1520 EXTCONST U32 PL_charclass[] = {
1521 # include "l1_char_class_tab.h"
1524 # else /* ! DOINIT */
1525 EXTCONST U32 PL_charclass[];
1529 /* The 1U keeps Solaris from griping when shifting sets the uppermost bit */
1530 # define _CC_mask(classnum) (1U << (classnum))
1532 /* For internal core Perl use only: the base macro for defining macros like
1534 # define _generic_isCC(c, classnum) cBOOL(FITS_IN_8_BITS(c) \
1535 && (PL_charclass[(U8) (c)] & _CC_mask(classnum)))
1537 /* The mask for the _A versions of the macros; it just adds in the bit for
1539 # define _CC_mask_A(classnum) (_CC_mask(classnum) | _CC_mask(_CC_ASCII))
1541 /* For internal core Perl use only: the base macro for defining macros like
1542 * isALPHA_A. The foo_A version makes sure that both the desired bit and
1543 * the ASCII bit are present */
1544 # define _generic_isCC_A(c, classnum) (FITS_IN_8_BITS(c) \
1545 && ((PL_charclass[(U8) (c)] & _CC_mask_A(classnum)) \
1546 == _CC_mask_A(classnum)))
1548 /* On ASCII platforms certain classes form a single range. It's faster to
1549 * special case these. isDIGIT is a single range on all platforms */
1551 # define isALPHA_A(c) _generic_isCC_A(c, _CC_ALPHA)
1552 # define isGRAPH_A(c) _generic_isCC_A(c, _CC_GRAPH)
1553 # define isLOWER_A(c) _generic_isCC_A(c, _CC_LOWER)
1554 # define isPRINT_A(c) _generic_isCC_A(c, _CC_PRINT)
1555 # define isUPPER_A(c) _generic_isCC_A(c, _CC_UPPER)
1557 /* By folding the upper and lowercase, we can use a single range */
1558 # define isALPHA_A(c) inRANGE((~('A' ^ 'a') & (c)), 'A', 'Z')
1559 # define isGRAPH_A(c) inRANGE(c, ' ' + 1, 0x7e)
1560 # define isLOWER_A(c) inRANGE(c, 'a', 'z')
1561 # define isPRINT_A(c) inRANGE(c, ' ', 0x7e)
1562 # define isUPPER_A(c) inRANGE(c, 'A', 'Z')
1564 # define isALPHANUMERIC_A(c) _generic_isCC_A(c, _CC_ALPHANUMERIC)
1565 # define isBLANK_A(c) _generic_isCC_A(c, _CC_BLANK)
1566 # define isCNTRL_A(c) _generic_isCC_A(c, _CC_CNTRL)
1567 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1568 # define isPUNCT_A(c) _generic_isCC_A(c, _CC_PUNCT)
1569 # define isSPACE_A(c) _generic_isCC_A(c, _CC_SPACE)
1570 # define isWORDCHAR_A(c) _generic_isCC_A(c, _CC_WORDCHAR)
1571 # define isXDIGIT_A(c) _generic_isCC(c, _CC_XDIGIT) /* No non-ASCII xdigits
1573 # define isIDFIRST_A(c) _generic_isCC_A(c, _CC_IDFIRST)
1574 # define isALPHA_L1(c) _generic_isCC(c, _CC_ALPHA)
1575 # define isALPHANUMERIC_L1(c) _generic_isCC(c, _CC_ALPHANUMERIC)
1576 # define isBLANK_L1(c) _generic_isCC(c, _CC_BLANK)
1578 /* continuation character for legal NAME in \N{NAME} */
1579 # define isCHARNAME_CONT(c) _generic_isCC(c, _CC_CHARNAME_CONT)
1581 # define isCNTRL_L1(c) _generic_isCC(c, _CC_CNTRL)
1582 # define isGRAPH_L1(c) _generic_isCC(c, _CC_GRAPH)
1583 # define isLOWER_L1(c) _generic_isCC(c, _CC_LOWER)
1584 # define isPRINT_L1(c) _generic_isCC(c, _CC_PRINT)
1585 # define isPSXSPC_L1(c) isSPACE_L1(c)
1586 # define isPUNCT_L1(c) _generic_isCC(c, _CC_PUNCT)
1587 # define isSPACE_L1(c) _generic_isCC(c, _CC_SPACE)
1588 # define isUPPER_L1(c) _generic_isCC(c, _CC_UPPER)
1589 # define isWORDCHAR_L1(c) _generic_isCC(c, _CC_WORDCHAR)
1590 # define isIDFIRST_L1(c) _generic_isCC(c, _CC_IDFIRST)
1593 # define isASCII(c) _generic_isCC(c, _CC_ASCII)
1596 /* Participates in a single-character fold with a character above 255 */
1597 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_SIMPLE_FOLD)))
1599 /* Like the above, but also can be part of a multi-char fold */
1600 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) ((! cBOOL(FITS_IN_8_BITS(c))) || (PL_charclass[(U8) (c)] & _CC_mask(_CC_NONLATIN1_FOLD)))
1602 # define _isQUOTEMETA(c) _generic_isCC(c, _CC_QUOTEMETA)
1603 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1604 _generic_isCC(c, _CC_NON_FINAL_FOLD)
1605 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) \
1606 _generic_isCC(c, _CC_IS_IN_SOME_FOLD)
1608 /* is c a control character for which we have a mnemonic? */
1609 # if defined(PERL_CORE) || defined(PERL_EXT)
1610 # define isMNEMONIC_CNTRL(c) _generic_isCC(c, _CC_MNEMONIC_CNTRL)
1612 #else /* else we don't have perl.h H_PERL */
1614 /* If we don't have perl.h, we are compiling a utility program. Below we
1615 * hard-code various macro definitions that wouldn't otherwise be available
1616 * to it. Most are coded based on first principles. These are written to
1617 * avoid EBCDIC vs. ASCII #ifdef's as much as possible. */
1618 # define isDIGIT_A(c) inRANGE(c, '0', '9')
1619 # define isBLANK_A(c) ((c) == ' ' || (c) == '\t')
1620 # define isSPACE_A(c) (isBLANK_A(c) \
1625 /* On EBCDIC, there are gaps between 'i' and 'j'; 'r' and 's'. Same for
1626 * uppercase. The tests for those aren't necessary on ASCII, but hurt only
1627 * performance (if optimization isn't on), and allow the same code to be
1628 * used for both platform types */
1629 # define isLOWER_A(c) inRANGE((c), 'a', 'i') \
1630 || inRANGE((c), 'j', 'r') \
1631 || inRANGE((c), 's', 'z')
1632 # define isUPPER_A(c) inRANGE((c), 'A', 'I') \
1633 || inRANGE((c), 'J', 'R') \
1634 || inRANGE((c), 'S', 'Z')
1635 # define isALPHA_A(c) (isUPPER_A(c) || isLOWER_A(c))
1636 # define isALPHANUMERIC_A(c) (isALPHA_A(c) || isDIGIT_A(c))
1637 # define isWORDCHAR_A(c) (isALPHANUMERIC_A(c) || (c) == '_')
1638 # define isIDFIRST_A(c) (isALPHA_A(c) || (c) == '_')
1639 # define isXDIGIT_A(c) ( isDIGIT_A(c) \
1640 || inRANGE((c), 'a', 'f') \
1641 || inRANGE((c), 'A', 'F')
1642 # define isPUNCT_A(c) ((c) == '-' || (c) == '!' || (c) == '"' \
1643 || (c) == '#' || (c) == '$' || (c) == '%' \
1644 || (c) == '&' || (c) == '\'' || (c) == '(' \
1645 || (c) == ')' || (c) == '*' || (c) == '+' \
1646 || (c) == ',' || (c) == '.' || (c) == '/' \
1647 || (c) == ':' || (c) == ';' || (c) == '<' \
1648 || (c) == '=' || (c) == '>' || (c) == '?' \
1649 || (c) == '@' || (c) == '[' || (c) == '\\' \
1650 || (c) == ']' || (c) == '^' || (c) == '_' \
1651 || (c) == '`' || (c) == '{' || (c) == '|' \
1652 || (c) == '}' || (c) == '~')
1653 # define isGRAPH_A(c) (isALPHANUMERIC_A(c) || isPUNCT_A(c))
1654 # define isPRINT_A(c) (isGRAPH_A(c) || (c) == ' ')
1657 /* The below is accurate for the 3 EBCDIC code pages traditionally
1658 * supported by perl. The only difference between them in the controls
1659 * is the position of \n, and that is represented symbolically below */
1660 # define isCNTRL_A(c) ((c) == '\0' || (c) == '\a' || (c) == '\b' \
1661 || (c) == '\f' || (c) == '\n' || (c) == '\r' \
1662 || (c) == '\t' || (c) == '\v' \
1663 || inRANGE((c), 1, 3) /* SOH, STX, ETX */ \
1664 || (c) == 7F /* U+7F DEL */ \
1665 || inRANGE((c), 0x0E, 0x13) /* SO SI DLE \
1667 || (c) == 0x18 /* U+18 CAN */ \
1668 || (c) == 0x19 /* U+19 EOM */ \
1669 || inRANGE((c), 0x1C, 0x1F) /* [FGRU]S */ \
1670 || (c) == 0x26 /* U+17 ETB */ \
1671 || (c) == 0x27 /* U+1B ESC */ \
1672 || (c) == 0x2D /* U+05 ENQ */ \
1673 || (c) == 0x2E /* U+06 ACK */ \
1674 || (c) == 0x32 /* U+16 SYN */ \
1675 || (c) == 0x37 /* U+04 EOT */ \
1676 || (c) == 0x3C /* U+14 DC4 */ \
1677 || (c) == 0x3D /* U+15 NAK */ \
1678 || (c) == 0x3F)/* U+1A SUB */
1679 # define isASCII(c) (isCNTRL_A(c) || isPRINT_A(c))
1680 # else /* isASCII is already defined for ASCII platforms, so can use that to
1682 # define isCNTRL_A(c) (isASCII(c) && ! isPRINT_A(c))
1685 /* The _L1 macros may be unnecessary for the utilities; I (khw) added them
1686 * during debugging, and it seems best to keep them. We may be called
1687 * without NATIVE_TO_LATIN1 being defined. On ASCII platforms, it doesn't
1688 * do anything anyway, so make it not a problem */
1689 # if ! defined(EBCDIC) && ! defined(NATIVE_TO_LATIN1)
1690 # define NATIVE_TO_LATIN1(ch) (ch)
1692 # define isALPHA_L1(c) (isUPPER_L1(c) || isLOWER_L1(c))
1693 # define isALPHANUMERIC_L1(c) (isALPHA_L1(c) || isDIGIT_A(c))
1694 # define isBLANK_L1(c) (isBLANK_A(c) \
1695 || (FITS_IN_8_BITS(c) \
1696 && NATIVE_TO_LATIN1((U8) c) == 0xA0))
1697 # define isCNTRL_L1(c) (FITS_IN_8_BITS(c) && (! isPRINT_L1(c)))
1698 # define isGRAPH_L1(c) (isPRINT_L1(c) && (! isBLANK_L1(c)))
1699 # define isLOWER_L1(c) (isLOWER_A(c) \
1700 || (FITS_IN_8_BITS(c) \
1701 && (( NATIVE_TO_LATIN1((U8) c) >= 0xDF \
1702 && NATIVE_TO_LATIN1((U8) c) != 0xF7) \
1703 || NATIVE_TO_LATIN1((U8) c) == 0xAA \
1704 || NATIVE_TO_LATIN1((U8) c) == 0xBA \
1705 || NATIVE_TO_LATIN1((U8) c) == 0xB5)))
1706 # define isPRINT_L1(c) (isPRINT_A(c) \
1707 || (FITS_IN_8_BITS(c) \
1708 && NATIVE_TO_LATIN1((U8) c) >= 0xA0))
1709 # define isPUNCT_L1(c) (isPUNCT_A(c) \
1710 || (FITS_IN_8_BITS(c) \
1711 && ( NATIVE_TO_LATIN1((U8) c) == 0xA1 \
1712 || NATIVE_TO_LATIN1((U8) c) == 0xA7 \
1713 || NATIVE_TO_LATIN1((U8) c) == 0xAB \
1714 || NATIVE_TO_LATIN1((U8) c) == 0xB6 \
1715 || NATIVE_TO_LATIN1((U8) c) == 0xB7 \
1716 || NATIVE_TO_LATIN1((U8) c) == 0xBB \
1717 || NATIVE_TO_LATIN1((U8) c) == 0xBF)))
1718 # define isSPACE_L1(c) (isSPACE_A(c) \
1719 || (FITS_IN_8_BITS(c) \
1720 && ( NATIVE_TO_LATIN1((U8) c) == 0x85 \
1721 || NATIVE_TO_LATIN1((U8) c) == 0xA0)))
1722 # define isUPPER_L1(c) (isUPPER_A(c) \
1723 || (FITS_IN_8_BITS(c) \
1724 && ( IN_RANGE(NATIVE_TO_LATIN1((U8) c), \
1726 && NATIVE_TO_LATIN1((U8) c) != 0xD7)))
1727 # define isWORDCHAR_L1(c) (isIDFIRST_L1(c) || isDIGIT_A(c))
1728 # define isIDFIRST_L1(c) (isALPHA_L1(c) || NATIVE_TO_LATIN1(c) == '_')
1729 # define isCHARNAME_CONT(c) (isWORDCHAR_L1(c) \
1734 /* The following are not fully accurate in the above-ASCII range. I (khw)
1735 * don't think it's necessary to be so for the purposes where this gets
1737 # define _isQUOTEMETA(c) (FITS_IN_8_BITS(c) && ! isWORDCHAR_L1(c))
1738 # define _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) isALPHA_L1(c)
1740 /* And these aren't accurate at all. They are useful only for above
1741 * Latin1, which utilities and bootstrapping don't deal with */
1742 # define _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c) 0
1743 # define _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1744 # define _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(c) 0
1746 /* Many of the macros later in this file are defined in terms of these. By
1747 * implementing them with a function, which converts the class number into
1748 * a call to the desired macro, all of the later ones work. However, that
1749 * function won't be actually defined when building a utility program (no
1750 * perl.h), and so a compiler error will be generated if one is attempted
1751 * to be used. And the above-Latin1 code points require Unicode tables to
1752 * be present, something unlikely to be the case when bootstrapping */
1753 # define _generic_isCC(c, classnum) \
1754 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), TRUE))
1755 # define _generic_isCC_A(c, classnum) \
1756 (FITS_IN_8_BITS(c) && S_bootstrap_ctype((U8) (c), (classnum), FALSE))
1757 #endif /* End of no perl.h H_PERL */
1759 #define isALPHANUMERIC(c) isALPHANUMERIC_A(c)
1760 #define isALPHA(c) isALPHA_A(c)
1761 #define isASCII_A(c) isASCII(c)
1762 #define isASCII_L1(c) isASCII(c)
1763 #define isBLANK(c) isBLANK_A(c)
1764 #define isCNTRL(c) isCNTRL_A(c)
1765 #define isDIGIT(c) isDIGIT_A(c)
1766 #define isGRAPH(c) isGRAPH_A(c)
1767 #define isIDFIRST(c) isIDFIRST_A(c)
1768 #define isLOWER(c) isLOWER_A(c)
1769 #define isPRINT(c) isPRINT_A(c)
1770 #define isPSXSPC_A(c) isSPACE_A(c)
1771 #define isPSXSPC(c) isPSXSPC_A(c)
1772 #define isPSXSPC_L1(c) isSPACE_L1(c)
1773 #define isPUNCT(c) isPUNCT_A(c)
1774 #define isSPACE(c) isSPACE_A(c)
1775 #define isUPPER(c) isUPPER_A(c)
1776 #define isWORDCHAR(c) isWORDCHAR_A(c)
1777 #define isXDIGIT(c) isXDIGIT_A(c)
1779 /* ASCII casing. These could also be written as
1780 #define toLOWER(c) (isASCII(c) ? toLOWER_LATIN1(c) : (c))
1781 #define toUPPER(c) (isASCII(c) ? toUPPER_LATIN1_MOD(c) : (c))
1782 which uses table lookup and mask instead of subtraction. (This would
1783 work because the _MOD does not apply in the ASCII range).
1785 These actually are UTF-8 invariant casing, not just ASCII, as any non-ASCII
1786 UTF-8 invariants are neither upper nor lower. (Only on EBCDIC platforms are
1787 there non-ASCII invariants, and all of them are controls.) */
1788 #define toLOWER(c) (isUPPER(c) ? (U8)((c) + ('a' - 'A')) : (c))
1789 #define toUPPER(c) (isLOWER(c) ? (U8)((c) - ('a' - 'A')) : (c))
1791 /* In the ASCII range, these are equivalent to what they're here defined to be.
1792 * But by creating these definitions, other code doesn't have to be aware of
1793 * this detail. Actually this works for all UTF-8 invariants, not just the
1794 * ASCII range. (EBCDIC platforms can have non-ASCII invariants.) */
1795 #define toFOLD(c) toLOWER(c)
1796 #define toTITLE(c) toUPPER(c)
1798 #define toLOWER_A(c) toLOWER(c)
1799 #define toUPPER_A(c) toUPPER(c)
1800 #define toFOLD_A(c) toFOLD(c)
1801 #define toTITLE_A(c) toTITLE(c)
1803 /* Use table lookup for speed; returns the input itself if is out-of-range */
1804 #define toLOWER_LATIN1(c) ((! FITS_IN_8_BITS(c)) \
1806 : PL_latin1_lc[ (U8) (c) ])
1807 #define toLOWER_L1(c) toLOWER_LATIN1(c) /* Synonym for consistency */
1809 /* Modified uc. Is correct uc except for three non-ascii chars which are
1810 * all mapped to one of them, and these need special handling; returns the
1811 * input itself if is out-of-range */
1812 #define toUPPER_LATIN1_MOD(c) ((! FITS_IN_8_BITS(c)) \
1814 : PL_mod_latin1_uc[ (U8) (c) ])
1815 #define IN_UTF8_CTYPE_LOCALE PL_in_utf8_CTYPE_locale
1817 /* Use foo_LC_uvchr() instead of these for beyond the Latin1 range */
1819 /* For internal core Perl use only: the base macro for defining macros like
1820 * isALPHA_LC, which uses the current LC_CTYPE locale. 'c' is the code point
1821 * (0-255) to check. In a UTF-8 locale, the result is the same as calling
1822 * isFOO_L1(); the 'utf8_locale_classnum' parameter is something like
1823 * _CC_UPPER, which gives the class number for doing this. For non-UTF-8
1824 * locales, the code to actually do the test this is passed in 'non_utf8'. If
1825 * 'c' is above 255, 0 is returned. For accessing the full range of possible
1826 * code points under locale rules, use the macros based on _generic_LC_uvchr
1827 * instead of this. */
1828 #define _generic_LC_base(c, utf8_locale_classnum, non_utf8) \
1829 (! FITS_IN_8_BITS(c) \
1831 : IN_UTF8_CTYPE_LOCALE \
1832 ? cBOOL(PL_charclass[(U8) (c)] & _CC_mask(utf8_locale_classnum)) \
1835 /* For internal core Perl use only: a helper macro for defining macros like
1836 * isALPHA_LC. 'c' is the code point (0-255) to check. The function name to
1837 * actually do this test is passed in 'non_utf8_func', which is called on 'c',
1838 * casting 'c' to the macro _LC_CAST, which should not be parenthesized. See
1839 * _generic_LC_base for more info */
1840 #define _generic_LC(c, utf8_locale_classnum, non_utf8_func) \
1841 _generic_LC_base(c,utf8_locale_classnum, \
1842 non_utf8_func( (_LC_CAST) (c)))
1844 /* For internal core Perl use only: like _generic_LC, but also returns TRUE if
1845 * 'c' is the platform's native underscore character */
1846 #define _generic_LC_underscore(c,utf8_locale_classnum,non_utf8_func) \
1847 _generic_LC_base(c, utf8_locale_classnum, \
1848 (non_utf8_func( (_LC_CAST) (c)) \
1849 || (char)(c) == '_'))
1851 /* These next three are also for internal core Perl use only: case-change
1852 * helper macros. The reason for using the PL_latin arrays is in case the
1853 * system function is defective; it ensures uniform results that conform to the
1854 * Unicod standard. It does not handle the anomalies in UTF-8 Turkic locales */
1855 #define _generic_toLOWER_LC(c, function, cast) (! FITS_IN_8_BITS(c) \
1857 : (IN_UTF8_CTYPE_LOCALE) \
1858 ? PL_latin1_lc[ (U8) (c) ] \
1859 : (cast)function((cast)(c)))
1861 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1862 * returns a single value, so can't adequately return the upper case of LATIN
1863 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1864 * values "SS"); instead it asserts against that under DEBUGGING, and
1865 * otherwise returns its input. It does not handle the anomalies in UTF-8
1866 * Turkic locales. */
1867 #define _generic_toUPPER_LC(c, function, cast) \
1868 (! FITS_IN_8_BITS(c) \
1870 : ((! IN_UTF8_CTYPE_LOCALE) \
1871 ? (cast)function((cast)(c)) \
1872 : ((((U8)(c)) == MICRO_SIGN) \
1873 ? GREEK_CAPITAL_LETTER_MU \
1874 : ((((U8)(c)) == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS) \
1875 ? LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS \
1876 : ((((U8)(c)) == LATIN_SMALL_LETTER_SHARP_S) \
1877 ? (__ASSERT_(0) (c)) \
1878 : PL_mod_latin1_uc[ (U8) (c) ])))))
1880 /* Note that the result can be larger than a byte in a UTF-8 locale. It
1881 * returns a single value, so can't adequately return the fold case of LATIN
1882 * SMALL LETTER SHARP S in a UTF-8 locale (which should be a string of two
1883 * values "ss"); instead it asserts against that under DEBUGGING, and
1884 * otherwise returns its input. It does not handle the anomalies in UTF-8
1886 #define _generic_toFOLD_LC(c, function, cast) \
1887 ((UNLIKELY((c) == MICRO_SIGN) && IN_UTF8_CTYPE_LOCALE) \
1888 ? GREEK_SMALL_LETTER_MU \
1889 : (__ASSERT_(! IN_UTF8_CTYPE_LOCALE \
1890 || (c) != LATIN_SMALL_LETTER_SHARP_S) \
1891 _generic_toLOWER_LC(c, function, cast)))
1893 /* Use the libc versions for these if available. */
1894 #if defined(HAS_ISASCII)
1895 # define isASCII_LC(c) (FITS_IN_8_BITS(c) && isascii( (U8) (c)))
1897 # define isASCII_LC(c) isASCII(c)
1900 #if defined(HAS_ISBLANK)
1901 # define isBLANK_LC(c) _generic_LC(c, _CC_BLANK, isblank)
1902 #else /* Unlike isASCII, varies if in a UTF-8 locale */
1903 # define isBLANK_LC(c) ((IN_UTF8_CTYPE_LOCALE) ? isBLANK_L1(c) : isBLANK(c))
1909 /* The Windows functions don't bother to follow the POSIX standard, which
1910 * for example says that something can't both be a printable and a control.
1911 * But Windows treats the \t control as a printable, and does such things
1912 * as making superscripts into both digits and punctuation. This tames
1913 * these flaws by assuming that the definitions of both controls and space
1914 * are correct, and then making sure that other definitions don't have
1915 * weirdnesses, by making sure that isalnum() isn't also ispunct(), etc.
1916 * Not all possible weirdnesses are checked for, just the ones that were
1917 * detected on actual Microsoft code pages */
1919 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1920 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1922 # define isALPHA_LC(c) (_generic_LC(c, _CC_ALPHA, isalpha) \
1923 && isALPHANUMERIC_LC(c))
1924 # define isALPHANUMERIC_LC(c) (_generic_LC(c, _CC_ALPHANUMERIC, isalnum) && \
1926 # define isDIGIT_LC(c) (_generic_LC(c, _CC_DIGIT, isdigit) && \
1927 isALPHANUMERIC_LC(c))
1928 # define isGRAPH_LC(c) (_generic_LC(c, _CC_GRAPH, isgraph) && isPRINT_LC(c))
1929 # define isIDFIRST_LC(c) (((c) == '_') \
1930 || (_generic_LC(c, _CC_IDFIRST, isalpha) && ! isPUNCT_LC(c)))
1931 # define isLOWER_LC(c) (_generic_LC(c, _CC_LOWER, islower) && isALPHA_LC(c))
1932 # define isPRINT_LC(c) (_generic_LC(c, _CC_PRINT, isprint) && ! isCNTRL_LC(c))
1933 # define isPUNCT_LC(c) (_generic_LC(c, _CC_PUNCT, ispunct) && ! isCNTRL_LC(c))
1934 # define isUPPER_LC(c) (_generic_LC(c, _CC_UPPER, isupper) && isALPHA_LC(c))
1935 # define isWORDCHAR_LC(c) (((c) == '_') || isALPHANUMERIC_LC(c))
1936 # define isXDIGIT_LC(c) (_generic_LC(c, _CC_XDIGIT, isxdigit) \
1937 && isALPHANUMERIC_LC(c))
1939 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1940 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1941 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1943 #elif defined(CTYPE256) || (!defined(isascii) && !defined(HAS_ISASCII))
1944 /* For most other platforms */
1946 # define isALPHA_LC(c) _generic_LC(c, _CC_ALPHA, isalpha)
1947 # define isALPHANUMERIC_LC(c) _generic_LC(c, _CC_ALPHANUMERIC, isalnum)
1948 # define isCNTRL_LC(c) _generic_LC(c, _CC_CNTRL, iscntrl)
1949 # define isDIGIT_LC(c) _generic_LC(c, _CC_DIGIT, isdigit)
1950 # define isGRAPH_LC(c) _generic_LC(c, _CC_GRAPH, isgraph)
1951 # define isIDFIRST_LC(c) _generic_LC_underscore(c, _CC_IDFIRST, isalpha)
1952 # define isLOWER_LC(c) _generic_LC(c, _CC_LOWER, islower)
1953 # define isPRINT_LC(c) _generic_LC(c, _CC_PRINT, isprint)
1954 # define isPUNCT_LC(c) _generic_LC(c, _CC_PUNCT, ispunct)
1955 # define isSPACE_LC(c) _generic_LC(c, _CC_SPACE, isspace)
1956 # define isUPPER_LC(c) _generic_LC(c, _CC_UPPER, isupper)
1957 # define isWORDCHAR_LC(c) _generic_LC_underscore(c, _CC_WORDCHAR, isalnum)
1958 # define isXDIGIT_LC(c) _generic_LC(c, _CC_XDIGIT, isxdigit)
1961 # define toLOWER_LC(c) _generic_toLOWER_LC((c), tolower, U8)
1962 # define toUPPER_LC(c) _generic_toUPPER_LC((c), toupper, U8)
1963 # define toFOLD_LC(c) _generic_toFOLD_LC((c), tolower, U8)
1965 #else /* The final fallback position */
1967 # define isALPHA_LC(c) (isascii(c) && isalpha(c))
1968 # define isALPHANUMERIC_LC(c) (isascii(c) && isalnum(c))
1969 # define isCNTRL_LC(c) (isascii(c) && iscntrl(c))
1970 # define isDIGIT_LC(c) (isascii(c) && isdigit(c))
1971 # define isGRAPH_LC(c) (isascii(c) && isgraph(c))
1972 # define isIDFIRST_LC(c) (isascii(c) && (isalpha(c) || (c) == '_'))
1973 # define isLOWER_LC(c) (isascii(c) && islower(c))
1974 # define isPRINT_LC(c) (isascii(c) && isprint(c))
1975 # define isPUNCT_LC(c) (isascii(c) && ispunct(c))
1976 # define isSPACE_LC(c) (isascii(c) && isspace(c))
1977 # define isUPPER_LC(c) (isascii(c) && isupper(c))
1978 # define isWORDCHAR_LC(c) (isascii(c) && (isalnum(c) || (c) == '_'))
1979 # define isXDIGIT_LC(c) (isascii(c) && isxdigit(c))
1981 # define toLOWER_LC(c) (isascii(c) ? tolower(c) : (c))
1982 # define toUPPER_LC(c) (isascii(c) ? toupper(c) : (c))
1983 # define toFOLD_LC(c) (isascii(c) ? tolower(c) : (c))
1987 #define isIDCONT(c) isWORDCHAR(c)
1988 #define isIDCONT_A(c) isWORDCHAR_A(c)
1989 #define isIDCONT_L1(c) isWORDCHAR_L1(c)
1990 #define isIDCONT_LC(c) isWORDCHAR_LC(c)
1991 #define isPSXSPC_LC(c) isSPACE_LC(c)
1993 /* For internal core Perl use only: the base macros for defining macros like
1994 * isALPHA_uvchr. 'c' is the code point to check. 'classnum' is the POSIX class
1995 * number defined earlier in this file. _generic_uvchr() is used for POSIX
1996 * classes where there is a macro or function 'above_latin1' that takes the
1997 * single argument 'c' and returns the desired value. These exist for those
1998 * classes which have simple definitions, avoiding the overhead of an inversion
1999 * list binary search. _generic_invlist_uvchr() can be used
2000 * for classes where that overhead is faster than a direct lookup.
2001 * _generic_uvchr() won't compile if 'c' isn't unsigned, as it won't match the
2002 * 'above_latin1' prototype. _generic_isCC() macro does bounds checking, so
2003 * have duplicate checks here, so could create versions of the macros that
2004 * don't, but experiments show that gcc optimizes them out anyway. */
2006 /* Note that all ignore 'use bytes' */
2007 #define _generic_uvchr(classnum, above_latin1, c) ((c) < 256 \
2008 ? _generic_isCC(c, classnum) \
2010 #define _generic_invlist_uvchr(classnum, c) ((c) < 256 \
2011 ? _generic_isCC(c, classnum) \
2012 : _is_uni_FOO(classnum, c))
2013 #define isALPHA_uvchr(c) _generic_invlist_uvchr(_CC_ALPHA, c)
2014 #define isALPHANUMERIC_uvchr(c) _generic_invlist_uvchr(_CC_ALPHANUMERIC, c)
2015 #define isASCII_uvchr(c) isASCII(c)
2016 #define isBLANK_uvchr(c) _generic_uvchr(_CC_BLANK, is_HORIZWS_cp_high, c)
2017 #define isCNTRL_uvchr(c) isCNTRL_L1(c) /* All controls are in Latin1 */
2018 #define isDIGIT_uvchr(c) _generic_invlist_uvchr(_CC_DIGIT, c)
2019 #define isGRAPH_uvchr(c) _generic_invlist_uvchr(_CC_GRAPH, c)
2020 #define isIDCONT_uvchr(c) \
2021 _generic_uvchr(_CC_WORDCHAR, _is_uni_perl_idcont, c)
2022 #define isIDFIRST_uvchr(c) \
2023 _generic_uvchr(_CC_IDFIRST, _is_uni_perl_idstart, c)
2024 #define isLOWER_uvchr(c) _generic_invlist_uvchr(_CC_LOWER, c)
2025 #define isPRINT_uvchr(c) _generic_invlist_uvchr(_CC_PRINT, c)
2027 #define isPUNCT_uvchr(c) _generic_invlist_uvchr(_CC_PUNCT, c)
2028 #define isSPACE_uvchr(c) _generic_uvchr(_CC_SPACE, is_XPERLSPACE_cp_high, c)
2029 #define isPSXSPC_uvchr(c) isSPACE_uvchr(c)
2031 #define isUPPER_uvchr(c) _generic_invlist_uvchr(_CC_UPPER, c)
2032 #define isVERTWS_uvchr(c) _generic_uvchr(_CC_VERTSPACE, is_VERTWS_cp_high, c)
2033 #define isWORDCHAR_uvchr(c) _generic_invlist_uvchr(_CC_WORDCHAR, c)
2034 #define isXDIGIT_uvchr(c) _generic_uvchr(_CC_XDIGIT, is_XDIGIT_cp_high, c)
2036 #define toFOLD_uvchr(c,s,l) to_uni_fold(c,s,l)
2037 #define toLOWER_uvchr(c,s,l) to_uni_lower(c,s,l)
2038 #define toTITLE_uvchr(c,s,l) to_uni_title(c,s,l)
2039 #define toUPPER_uvchr(c,s,l) to_uni_upper(c,s,l)
2041 /* For backwards compatibility, even though '_uni' should mean official Unicode
2042 * code points, in Perl it means native for those below 256 */
2043 #define isALPHA_uni(c) isALPHA_uvchr(c)
2044 #define isALPHANUMERIC_uni(c) isALPHANUMERIC_uvchr(c)
2045 #define isASCII_uni(c) isASCII_uvchr(c)
2046 #define isBLANK_uni(c) isBLANK_uvchr(c)
2047 #define isCNTRL_uni(c) isCNTRL_uvchr(c)
2048 #define isDIGIT_uni(c) isDIGIT_uvchr(c)
2049 #define isGRAPH_uni(c) isGRAPH_uvchr(c)
2050 #define isIDCONT_uni(c) isIDCONT_uvchr(c)
2051 #define isIDFIRST_uni(c) isIDFIRST_uvchr(c)
2052 #define isLOWER_uni(c) isLOWER_uvchr(c)
2053 #define isPRINT_uni(c) isPRINT_uvchr(c)
2054 #define isPUNCT_uni(c) isPUNCT_uvchr(c)
2055 #define isSPACE_uni(c) isSPACE_uvchr(c)
2056 #define isPSXSPC_uni(c) isPSXSPC_uvchr(c)
2057 #define isUPPER_uni(c) isUPPER_uvchr(c)
2058 #define isVERTWS_uni(c) isVERTWS_uvchr(c)
2059 #define isWORDCHAR_uni(c) isWORDCHAR_uvchr(c)
2060 #define isXDIGIT_uni(c) isXDIGIT_uvchr(c)
2061 #define toFOLD_uni(c,s,l) toFOLD_uvchr(c,s,l)
2062 #define toLOWER_uni(c,s,l) toLOWER_uvchr(c,s,l)
2063 #define toTITLE_uni(c,s,l) toTITLE_uvchr(c,s,l)
2064 #define toUPPER_uni(c,s,l) toUPPER_uvchr(c,s,l)
2066 /* For internal core Perl use only: the base macros for defining macros like
2067 * isALPHA_LC_uvchr. These are like isALPHA_LC, but the input can be any code
2068 * point, not just 0-255. Like _generic_uvchr, there are two versions, one for
2069 * simple class definitions; the other for more complex. These are like
2070 * _generic_uvchr, so see it for more info. */
2071 #define _generic_LC_uvchr(latin1, above_latin1, c) \
2072 (c < 256 ? latin1(c) : above_latin1(c))
2073 #define _generic_LC_invlist_uvchr(latin1, classnum, c) \
2074 (c < 256 ? latin1(c) : _is_uni_FOO(classnum, c))
2076 #define isALPHA_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHA_LC, _CC_ALPHA, c)
2077 #define isALPHANUMERIC_LC_uvchr(c) _generic_LC_invlist_uvchr(isALPHANUMERIC_LC, \
2078 _CC_ALPHANUMERIC, c)
2079 #define isASCII_LC_uvchr(c) isASCII_LC(c)
2080 #define isBLANK_LC_uvchr(c) _generic_LC_uvchr(isBLANK_LC, \
2081 is_HORIZWS_cp_high, c)
2082 #define isCNTRL_LC_uvchr(c) (c < 256 ? isCNTRL_LC(c) : 0)
2083 #define isDIGIT_LC_uvchr(c) _generic_LC_invlist_uvchr(isDIGIT_LC, _CC_DIGIT, c)
2084 #define isGRAPH_LC_uvchr(c) _generic_LC_invlist_uvchr(isGRAPH_LC, _CC_GRAPH, c)
2085 #define isIDCONT_LC_uvchr(c) _generic_LC_uvchr(isIDCONT_LC, \
2086 _is_uni_perl_idcont, c)
2087 #define isIDFIRST_LC_uvchr(c) _generic_LC_uvchr(isIDFIRST_LC, \
2088 _is_uni_perl_idstart, c)
2089 #define isLOWER_LC_uvchr(c) _generic_LC_invlist_uvchr(isLOWER_LC, _CC_LOWER, c)
2090 #define isPRINT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPRINT_LC, _CC_PRINT, c)
2091 #define isPSXSPC_LC_uvchr(c) isSPACE_LC_uvchr(c)
2092 #define isPUNCT_LC_uvchr(c) _generic_LC_invlist_uvchr(isPUNCT_LC, _CC_PUNCT, c)
2093 #define isSPACE_LC_uvchr(c) _generic_LC_uvchr(isSPACE_LC, \
2094 is_XPERLSPACE_cp_high, c)
2095 #define isUPPER_LC_uvchr(c) _generic_LC_invlist_uvchr(isUPPER_LC, _CC_UPPER, c)
2096 #define isWORDCHAR_LC_uvchr(c) _generic_LC_invlist_uvchr(isWORDCHAR_LC, \
2098 #define isXDIGIT_LC_uvchr(c) _generic_LC_uvchr(isXDIGIT_LC, \
2099 is_XDIGIT_cp_high, c)
2101 #define isBLANK_LC_uni(c) isBLANK_LC_uvchr(UNI_TO_NATIVE(c))
2103 /* The "_safe" macros make sure that we don't attempt to read beyond 'e', but
2104 * they don't otherwise go out of their way to look for malformed UTF-8. If
2105 * they can return accurate results without knowing if the input is otherwise
2106 * malformed, they do so. For example isASCII is accurate in spite of any
2107 * non-length malformations because it looks only at a single byte. Likewise
2108 * isDIGIT looks just at the first byte for code points 0-255, as all UTF-8
2109 * variant ones return FALSE. But, if the input has to be well-formed in order
2110 * for the results to be accurate, the macros will test and if malformed will
2111 * call a routine to die
2113 * Except for toke.c, the macros do assume that e > p, asserting that on
2114 * DEBUGGING builds. Much code that calls these depends on this being true,
2115 * for other reasons. toke.c is treated specially as using the regular
2116 * assertion breaks it in many ways. All strings that these operate on there
2117 * are supposed to have an extra NUL character at the end, so that *e = \0. A
2118 * bunch of code in toke.c assumes that this is true, so the assertion allows
2120 #ifdef PERL_IN_TOKE_C
2121 # define _utf8_safe_assert(p,e) ((e) > (p) || ((e) == (p) && *(p) == '\0'))
2123 # define _utf8_safe_assert(p,e) ((e) > (p))
2126 #define _generic_utf8_safe(classnum, p, e, above_latin1) \
2127 ((! _utf8_safe_assert(p, e)) \
2128 ? (_force_out_malformed_utf8_message((U8 *) (p), (U8 *) (e), 0, 1), 0)\
2129 : (UTF8_IS_INVARIANT(*(p))) \
2130 ? _generic_isCC(*(p), classnum) \
2131 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2132 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2133 ? _generic_isCC(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1 )), \
2135 : (_force_out_malformed_utf8_message( \
2136 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2138 /* Like the above, but calls 'above_latin1(p)' to get the utf8 value.
2139 * 'above_latin1' can be a macro */
2140 #define _generic_func_utf8_safe(classnum, above_latin1, p, e) \
2141 _generic_utf8_safe(classnum, p, e, above_latin1(p, e))
2142 #define _generic_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2143 _generic_utf8_safe(classnum, p, e, \
2144 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2145 ? (_force_out_malformed_utf8_message( \
2146 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2148 /* Like the above, but passes classnum to _isFOO_utf8(), instead of having an
2149 * 'above_latin1' parameter */
2150 #define _generic_invlist_utf8_safe(classnum, p, e) \
2151 _generic_utf8_safe(classnum, p, e, _is_utf8_FOO(classnum, p, e))
2153 /* Like the above, but should be used only when it is known that there are no
2154 * characters in the upper-Latin1 range (128-255 on ASCII platforms) which the
2155 * class is TRUE for. Hence it can skip the tests for this range.
2156 * 'above_latin1' should include its arguments */
2157 #define _generic_utf8_safe_no_upper_latin1(classnum, p, e, above_latin1) \
2158 (__ASSERT_(_utf8_safe_assert(p, e)) \
2159 (UTF8_IS_INVARIANT(*(p))) \
2160 ? _generic_isCC(*(p), classnum) \
2161 : (UTF8_IS_DOWNGRADEABLE_START(*(p))) \
2162 ? 0 /* Note that doesn't check validity for latin1 */ \
2166 #define isALPHA_utf8(p, e) isALPHA_utf8_safe(p, e)
2167 #define isALPHANUMERIC_utf8(p, e) isALPHANUMERIC_utf8_safe(p, e)
2168 #define isASCII_utf8(p, e) isASCII_utf8_safe(p, e)
2169 #define isBLANK_utf8(p, e) isBLANK_utf8_safe(p, e)
2170 #define isCNTRL_utf8(p, e) isCNTRL_utf8_safe(p, e)
2171 #define isDIGIT_utf8(p, e) isDIGIT_utf8_safe(p, e)
2172 #define isGRAPH_utf8(p, e) isGRAPH_utf8_safe(p, e)
2173 #define isIDCONT_utf8(p, e) isIDCONT_utf8_safe(p, e)
2174 #define isIDFIRST_utf8(p, e) isIDFIRST_utf8_safe(p, e)
2175 #define isLOWER_utf8(p, e) isLOWER_utf8_safe(p, e)
2176 #define isPRINT_utf8(p, e) isPRINT_utf8_safe(p, e)
2177 #define isPSXSPC_utf8(p, e) isPSXSPC_utf8_safe(p, e)
2178 #define isPUNCT_utf8(p, e) isPUNCT_utf8_safe(p, e)
2179 #define isSPACE_utf8(p, e) isSPACE_utf8_safe(p, e)
2180 #define isUPPER_utf8(p, e) isUPPER_utf8_safe(p, e)
2181 #define isVERTWS_utf8(p, e) isVERTWS_utf8_safe(p, e)
2182 #define isWORDCHAR_utf8(p, e) isWORDCHAR_utf8_safe(p, e)
2183 #define isXDIGIT_utf8(p, e) isXDIGIT_utf8_safe(p, e)
2185 #define isALPHA_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_ALPHA, p, e)
2186 #define isALPHANUMERIC_utf8_safe(p, e) \
2187 _generic_invlist_utf8_safe(_CC_ALPHANUMERIC, p, e)
2188 #define isASCII_utf8_safe(p, e) \
2189 /* Because ASCII is invariant under utf8, the non-utf8 macro \
2191 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII(*(p)))
2192 #define isBLANK_utf8_safe(p, e) \
2193 _generic_non_invlist_utf8_safe(_CC_BLANK, is_HORIZWS_high, p, e)
2196 /* Because all controls are UTF-8 invariants in EBCDIC, we can use this
2197 * more efficient macro instead of the more general one */
2198 # define isCNTRL_utf8_safe(p, e) \
2199 (__ASSERT_(_utf8_safe_assert(p, e)) isCNTRL_L1(*(p)))
2201 # define isCNTRL_utf8_safe(p, e) _generic_utf8_safe(_CC_CNTRL, p, e, 0)
2204 #define isDIGIT_utf8_safe(p, e) \
2205 _generic_utf8_safe_no_upper_latin1(_CC_DIGIT, p, e, \
2206 _is_utf8_FOO(_CC_DIGIT, p, e))
2207 #define isGRAPH_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_GRAPH, p, e)
2208 #define isIDCONT_utf8_safe(p, e) _generic_func_utf8_safe(_CC_WORDCHAR, \
2209 _is_utf8_perl_idcont, p, e)
2211 /* To prevent S_scan_word in toke.c from hanging, we have to make sure that
2212 * IDFIRST is an alnum. See
2213 * https://github.com/Perl/perl5/issues/10275 for more detail than you
2214 * ever wanted to know about. (In the ASCII range, there isn't a difference.)
2215 * This used to be not the XID version, but we decided to go with the more
2216 * modern Unicode definition */
2217 #define isIDFIRST_utf8_safe(p, e) \
2218 _generic_func_utf8_safe(_CC_IDFIRST, \
2219 _is_utf8_perl_idstart, (U8 *) (p), (U8 *) (e))
2221 #define isLOWER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_LOWER, p, e)
2222 #define isPRINT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PRINT, p, e)
2223 #define isPSXSPC_utf8_safe(p, e) isSPACE_utf8_safe(p, e)
2224 #define isPUNCT_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_PUNCT, p, e)
2225 #define isSPACE_utf8_safe(p, e) \
2226 _generic_non_invlist_utf8_safe(_CC_SPACE, is_XPERLSPACE_high, p, e)
2227 #define isUPPER_utf8_safe(p, e) _generic_invlist_utf8_safe(_CC_UPPER, p, e)
2228 #define isVERTWS_utf8_safe(p, e) \
2229 _generic_non_invlist_utf8_safe(_CC_VERTSPACE, is_VERTWS_high, p, e)
2230 #define isWORDCHAR_utf8_safe(p, e) \
2231 _generic_invlist_utf8_safe(_CC_WORDCHAR, p, e)
2232 #define isXDIGIT_utf8_safe(p, e) \
2233 _generic_utf8_safe_no_upper_latin1(_CC_XDIGIT, p, e, \
2234 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2235 ? (_force_out_malformed_utf8_message( \
2236 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2237 : is_XDIGIT_high(p)))
2239 #define toFOLD_utf8(p,e,s,l) toFOLD_utf8_safe(p,e,s,l)
2240 #define toLOWER_utf8(p,e,s,l) toLOWER_utf8_safe(p,e,s,l)
2241 #define toTITLE_utf8(p,e,s,l) toTITLE_utf8_safe(p,e,s,l)
2242 #define toUPPER_utf8(p,e,s,l) toUPPER_utf8_safe(p,e,s,l)
2244 /* For internal core use only, subject to change */
2245 #define _toFOLD_utf8_flags(p,e,s,l,f) _to_utf8_fold_flags (p,e,s,l,f)
2246 #define _toLOWER_utf8_flags(p,e,s,l,f) _to_utf8_lower_flags(p,e,s,l,f)
2247 #define _toTITLE_utf8_flags(p,e,s,l,f) _to_utf8_title_flags(p,e,s,l,f)
2248 #define _toUPPER_utf8_flags(p,e,s,l,f) _to_utf8_upper_flags(p,e,s,l,f)
2250 #define toFOLD_utf8_safe(p,e,s,l) _toFOLD_utf8_flags(p,e,s,l, FOLD_FLAGS_FULL)
2251 #define toLOWER_utf8_safe(p,e,s,l) _toLOWER_utf8_flags(p,e,s,l, 0)
2252 #define toTITLE_utf8_safe(p,e,s,l) _toTITLE_utf8_flags(p,e,s,l, 0)
2253 #define toUPPER_utf8_safe(p,e,s,l) _toUPPER_utf8_flags(p,e,s,l, 0)
2255 #define isALPHA_LC_utf8(p, e) isALPHA_LC_utf8_safe(p, e)
2256 #define isALPHANUMERIC_LC_utf8(p, e) isALPHANUMERIC_LC_utf8_safe(p, e)
2257 #define isASCII_LC_utf8(p, e) isASCII_LC_utf8_safe(p, e)
2258 #define isBLANK_LC_utf8(p, e) isBLANK_LC_utf8_safe(p, e)
2259 #define isCNTRL_LC_utf8(p, e) isCNTRL_LC_utf8_safe(p, e)
2260 #define isDIGIT_LC_utf8(p, e) isDIGIT_LC_utf8_safe(p, e)
2261 #define isGRAPH_LC_utf8(p, e) isGRAPH_LC_utf8_safe(p, e)
2262 #define isIDCONT_LC_utf8(p, e) isIDCONT_LC_utf8_safe(p, e)
2263 #define isIDFIRST_LC_utf8(p, e) isIDFIRST_LC_utf8_safe(p, e)
2264 #define isLOWER_LC_utf8(p, e) isLOWER_LC_utf8_safe(p, e)
2265 #define isPRINT_LC_utf8(p, e) isPRINT_LC_utf8_safe(p, e)
2266 #define isPSXSPC_LC_utf8(p, e) isPSXSPC_LC_utf8_safe(p, e)
2267 #define isPUNCT_LC_utf8(p, e) isPUNCT_LC_utf8_safe(p, e)
2268 #define isSPACE_LC_utf8(p, e) isSPACE_LC_utf8_safe(p, e)
2269 #define isUPPER_LC_utf8(p, e) isUPPER_LC_utf8_safe(p, e)
2270 #define isWORDCHAR_LC_utf8(p, e) isWORDCHAR_LC_utf8_safe(p, e)
2271 #define isXDIGIT_LC_utf8(p, e) isXDIGIT_LC_utf8_safe(p, e)
2273 /* For internal core Perl use only: the base macros for defining macros like
2274 * isALPHA_LC_utf8_safe. These are like _generic_utf8, but if the first code
2275 * point in 'p' is within the 0-255 range, it uses locale rules from the
2276 * passed-in 'macro' parameter */
2277 #define _generic_LC_utf8_safe(macro, p, e, above_latin1) \
2278 (__ASSERT_(_utf8_safe_assert(p, e)) \
2279 (UTF8_IS_INVARIANT(*(p))) \
2281 : (UTF8_IS_DOWNGRADEABLE_START(*(p)) \
2282 ? ((LIKELY((e) - (p) > 1 && UTF8_IS_CONTINUATION(*((p)+1)))) \
2283 ? macro(EIGHT_BIT_UTF8_TO_NATIVE(*(p), *((p)+1))) \
2284 : (_force_out_malformed_utf8_message( \
2285 (U8 *) (p), (U8 *) (e), 0, 1), 0)) \
2288 #define _generic_LC_invlist_utf8_safe(macro, classnum, p, e) \
2289 _generic_LC_utf8_safe(macro, p, e, \
2290 _is_utf8_FOO(classnum, p, e))
2292 #define _generic_LC_func_utf8_safe(macro, above_latin1, p, e) \
2293 _generic_LC_utf8_safe(macro, p, e, above_latin1(p, e))
2295 #define _generic_LC_non_invlist_utf8_safe(classnum, above_latin1, p, e) \
2296 _generic_LC_utf8_safe(classnum, p, e, \
2297 (UNLIKELY((e) - (p) < UTF8SKIP(p)) \
2298 ? (_force_out_malformed_utf8_message( \
2299 (U8 *) (p), (U8 *) (e), 0, 1), 0) \
2302 #define isALPHANUMERIC_LC_utf8_safe(p, e) \
2303 _generic_LC_invlist_utf8_safe(isALPHANUMERIC_LC, \
2304 _CC_ALPHANUMERIC, p, e)
2305 #define isALPHA_LC_utf8_safe(p, e) \
2306 _generic_LC_invlist_utf8_safe(isALPHA_LC, _CC_ALPHA, p, e)
2307 #define isASCII_LC_utf8_safe(p, e) \
2308 (__ASSERT_(_utf8_safe_assert(p, e)) isASCII_LC(*(p)))
2309 #define isBLANK_LC_utf8_safe(p, e) \
2310 _generic_LC_non_invlist_utf8_safe(isBLANK_LC, is_HORIZWS_high, p, e)
2311 #define isCNTRL_LC_utf8_safe(p, e) \
2312 _generic_LC_utf8_safe(isCNTRL_LC, p, e, 0)
2313 #define isDIGIT_LC_utf8_safe(p, e) \
2314 _generic_LC_invlist_utf8_safe(isDIGIT_LC, _CC_DIGIT, p, e)
2315 #define isGRAPH_LC_utf8_safe(p, e) \
2316 _generic_LC_invlist_utf8_safe(isGRAPH_LC, _CC_GRAPH, p, e)
2317 #define isIDCONT_LC_utf8_safe(p, e) \
2318 _generic_LC_func_utf8_safe(isIDCONT_LC, \
2319 _is_utf8_perl_idcont, p, e)
2320 #define isIDFIRST_LC_utf8_safe(p, e) \
2321 _generic_LC_func_utf8_safe(isIDFIRST_LC, \
2322 _is_utf8_perl_idstart, p, e)
2323 #define isLOWER_LC_utf8_safe(p, e) \
2324 _generic_LC_invlist_utf8_safe(isLOWER_LC, _CC_LOWER, p, e)
2325 #define isPRINT_LC_utf8_safe(p, e) \
2326 _generic_LC_invlist_utf8_safe(isPRINT_LC, _CC_PRINT, p, e)
2327 #define isPSXSPC_LC_utf8_safe(p, e) isSPACE_LC_utf8_safe(p, e)
2328 #define isPUNCT_LC_utf8_safe(p, e) \
2329 _generic_LC_invlist_utf8_safe(isPUNCT_LC, _CC_PUNCT, p, e)
2330 #define isSPACE_LC_utf8_safe(p, e) \
2331 _generic_LC_non_invlist_utf8_safe(isSPACE_LC, is_XPERLSPACE_high, p, e)
2332 #define isUPPER_LC_utf8_safe(p, e) \
2333 _generic_LC_invlist_utf8_safe(isUPPER_LC, _CC_UPPER, p, e)
2334 #define isWORDCHAR_LC_utf8_safe(p, e) \
2335 _generic_LC_invlist_utf8_safe(isWORDCHAR_LC, _CC_WORDCHAR, p, e)
2336 #define isXDIGIT_LC_utf8_safe(p, e) \
2337 _generic_LC_non_invlist_utf8_safe(isXDIGIT_LC, is_XDIGIT_high, p, e)
2339 /* Macros for backwards compatibility and for completeness when the ASCII and
2340 * Latin1 values are identical */
2341 #define isALPHAU(c) isALPHA_L1(c)
2342 #define isDIGIT_L1(c) isDIGIT_A(c)
2343 #define isOCTAL(c) isOCTAL_A(c)
2344 #define isOCTAL_L1(c) isOCTAL_A(c)
2345 #define isXDIGIT_L1(c) isXDIGIT_A(c)
2346 #define isALNUM(c) isWORDCHAR(c)
2347 #define isALNUM_A(c) isALNUM(c)
2348 #define isALNUMU(c) isWORDCHAR_L1(c)
2349 #define isALNUM_LC(c) isWORDCHAR_LC(c)
2350 #define isALNUM_uni(c) isWORDCHAR_uni(c)
2351 #define isALNUM_LC_uvchr(c) isWORDCHAR_LC_uvchr(c)
2352 #define isALNUM_utf8(p,e) isWORDCHAR_utf8(p,e)
2353 #define isALNUM_utf8_safe(p,e) isWORDCHAR_utf8_safe(p,e)
2354 #define isALNUM_LC_utf8(p,e)isWORDCHAR_LC_utf8(p,e)
2355 #define isALNUM_LC_utf8_safe(p,e)isWORDCHAR_LC_utf8_safe(p,e)
2356 #define isALNUMC_A(c) isALPHANUMERIC_A(c) /* Mnemonic: "C's alnum" */
2357 #define isALNUMC_L1(c) isALPHANUMERIC_L1(c)
2358 #define isALNUMC(c) isALPHANUMERIC(c)
2359 #define isALNUMC_LC(c) isALPHANUMERIC_LC(c)
2360 #define isALNUMC_uni(c) isALPHANUMERIC_uni(c)
2361 #define isALNUMC_LC_uvchr(c) isALPHANUMERIC_LC_uvchr(c)
2362 #define isALNUMC_utf8(p,e) isALPHANUMERIC_utf8(p,e)
2363 #define isALNUMC_utf8_safe(p,e) isALPHANUMERIC_utf8_safe(p,e)
2364 #define isALNUMC_LC_utf8_safe(p,e) isALPHANUMERIC_LC_utf8_safe(p,e)
2366 /* On EBCDIC platforms, CTRL-@ is 0, CTRL-A is 1, etc, just like on ASCII,
2367 * except that they don't necessarily mean the same characters, e.g. CTRL-D is
2368 * 4 on both systems, but that is EOT on ASCII; ST on EBCDIC.
2369 * '?' is special-cased on EBCDIC to APC, which is the control there that is
2370 * the outlier from the block that contains the other controls, just like
2371 * toCTRL('?') on ASCII yields DEL, the control that is the outlier from the C0
2372 * block. If it weren't special cased, it would yield a non-control.
2373 * The conversion works both ways, so toCTRL('D') is 4, and toCTRL(4) is D,
2376 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) toUPPER(((U8)(c))) ^ 64)
2378 # define toCTRL(c) (__ASSERT_(FITS_IN_8_BITS(c)) \
2380 ? (UNLIKELY((c) == '?') \
2381 ? QUESTION_MARK_CTRL \
2382 : (NATIVE_TO_LATIN1(toUPPER((U8) (c))) ^ 64)) \
2383 : (UNLIKELY((c) == QUESTION_MARK_CTRL) \
2385 : (LATIN1_TO_NATIVE(((U8) (c)) ^ 64)))))
2388 /* Line numbers are unsigned, 32 bits. */
2390 #define NOLINE ((line_t) 4294967295UL) /* = FFFFFFFF */
2392 /* Helpful alias for version prescan */
2393 #define is_LAX_VERSION(a,b) \
2394 (a != Perl_prescan_version(aTHX_ a, FALSE, b, NULL, NULL, NULL, NULL))
2396 #define is_STRICT_VERSION(a,b) \
2397 (a != Perl_prescan_version(aTHX_ a, TRUE, b, NULL, NULL, NULL, NULL))
2399 #define BADVERSION(a,b,c) \
2405 /* Converts a character KNOWN to represent a hexadecimal digit (0-9, A-F, or
2406 * a-f) to its numeric value without using any branches. The input is
2407 * validated only by an assert() in DEBUGGING builds.
2409 * It works by right shifting and isolating the bit that is 0 for the digits,
2410 * and 1 for at least the alphas A-F, a-f. The bit is shifted to the ones
2411 * position, and then to the eights position. Both are added together to form
2412 * 0 if the input is '0'-'9' and to form 9 if alpha. This is added to the
2413 * final four bits of the input to form the correct value. */
2414 #define XDIGIT_VALUE(c) (__ASSERT_(isXDIGIT(c)) \
2415 ((NATIVE_TO_LATIN1(c) >> 6) & 1) /* 1 if alpha; 0 if not */ \
2416 + ((NATIVE_TO_LATIN1(c) >> 3) & 8) /* 8 if alpha; 0 if not */ \
2417 + ((c) & 0xF)) /* 0-9 if input valid hex digit */
2419 /* The argument is a string pointer, which is advanced. */
2420 #define READ_XDIGIT(s) ((s)++, XDIGIT_VALUE(*((s) - 1)))
2422 /* Converts a character known to represent an octal digit (0-7) to its numeric
2423 * value. The input is validated only by an assert() in DEBUGGING builds. In
2424 * both ASCII and EBCDIC the last 3 bits of the octal digits range from 0-7. */
2425 #define OCTAL_VALUE(c) (__ASSERT_(isOCTAL(c)) (7 & (c)))
2427 /* Efficiently returns a boolean as to if two native characters are equivalent
2428 * case-insenstively. At least one of the characters must be one of [A-Za-z];
2429 * the ALPHA in the name is to remind you of that. This is asserted() in
2430 * DEBUGGING builds. Because [A-Za-z] are invariant under UTF-8, this macro
2431 * works (on valid input) for both non- and UTF-8-encoded bytes.
2433 * When one of the inputs is a compile-time constant and gets folded by the
2434 * compiler, this reduces to an AND and a TEST. On both EBCDIC and ASCII
2435 * machines, 'A' and 'a' differ by a single bit; the same with the upper and
2436 * lower case of all other ASCII-range alphabetics. On ASCII platforms, they
2437 * are 32 apart; on EBCDIC, they are 64. At compile time, this uses an
2438 * exclusive 'or' to find that bit and then inverts it to form a mask, with
2439 * just a single 0, in the bit position where the upper- and lowercase differ.
2441 #define isALPHA_FOLD_EQ(c1, c2) \
2442 (__ASSERT_(isALPHA_A(c1) || isALPHA_A(c2)) \
2443 ((c1) & ~('A' ^ 'a')) == ((c2) & ~('A' ^ 'a')))
2444 #define isALPHA_FOLD_NE(c1, c2) (! isALPHA_FOLD_EQ((c1), (c2)))
2447 =head1 Memory Management
2449 =for apidoc Am|void|Newx|void* ptr|int nitems|type
2450 The XSUB-writer's interface to the C C<malloc> function.
2452 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2454 In 5.9.3, Newx() and friends replace the older New() API, and drops
2455 the first parameter, I<x>, a debug aid which allowed callers to identify
2456 themselves. This aid has been superseded by a new build option,
2457 PERL_MEM_LOG (see L<perlhacktips/PERL_MEM_LOG>). The older API is still
2458 there for use in XS modules supporting older perls.
2460 =for apidoc Am|void|Newxc|void* ptr|int nitems|type|cast
2461 The XSUB-writer's interface to the C C<malloc> function, with
2462 cast. See also C<L</Newx>>.
2464 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2466 =for apidoc Am|void|Newxz|void* ptr|int nitems|type
2467 The XSUB-writer's interface to the C C<malloc> function. The allocated
2468 memory is zeroed with C<memzero>. See also C<L</Newx>>.
2470 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2472 =for apidoc Am|void|Renew|void* ptr|int nitems|type
2473 The XSUB-writer's interface to the C C<realloc> function.
2475 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2477 =for apidoc Am|void|Renewc|void* ptr|int nitems|type|cast
2478 The XSUB-writer's interface to the C C<realloc> function, with
2481 Memory obtained by this should B<ONLY> be freed with L</"Safefree">.
2483 =for apidoc Am|void|Safefree|void* ptr
2484 The XSUB-writer's interface to the C C<free> function.
2486 This should B<ONLY> be used on memory obtained using L</"Newx"> and friends.
2488 =for apidoc Am|void|Move|void* src|void* dest|int nitems|type
2489 The XSUB-writer's interface to the C C<memmove> function. The C<src> is the
2490 source, C<dest> is the destination, C<nitems> is the number of items, and
2491 C<type> is the type. Can do overlapping moves. See also C<L</Copy>>.
2493 =for apidoc Am|void *|MoveD|void* src|void* dest|int nitems|type
2494 Like C<Move> but returns C<dest>. Useful
2495 for encouraging compilers to tail-call
2498 =for apidoc Am|void|Copy|void* src|void* dest|int nitems|type
2499 The XSUB-writer's interface to the C C<memcpy> function. The C<src> is the
2500 source, C<dest> is the destination, C<nitems> is the number of items, and
2501 C<type> is the type. May fail on overlapping copies. See also C<L</Move>>.
2503 =for apidoc Am|void *|CopyD|void* src|void* dest|int nitems|type
2505 Like C<Copy> but returns C<dest>. Useful
2506 for encouraging compilers to tail-call
2509 =for apidoc Am|void|Zero|void* dest|int nitems|type
2511 The XSUB-writer's interface to the C C<memzero> function. The C<dest> is the
2512 destination, C<nitems> is the number of items, and C<type> is the type.
2514 =for apidoc Am|void *|ZeroD|void* dest|int nitems|type
2516 Like C<Zero> but returns dest. Useful
2517 for encouraging compilers to tail-call
2520 =for apidoc Am|void|StructCopy|type *src|type *dest|type
2521 This is an architecture-independent macro to copy one structure to another.
2523 =for apidoc Am|void|PoisonWith|void* dest|int nitems|type|U8 byte
2525 Fill up memory with a byte pattern (a byte repeated over and over
2526 again) that hopefully catches attempts to access uninitialized memory.
2528 =for apidoc Am|void|PoisonNew|void* dest|int nitems|type
2530 PoisonWith(0xAB) for catching access to allocated but uninitialized memory.
2532 =for apidoc Am|void|PoisonFree|void* dest|int nitems|type
2534 PoisonWith(0xEF) for catching access to freed memory.
2536 =for apidoc Am|void|Poison|void* dest|int nitems|type
2538 PoisonWith(0xEF) for catching access to freed memory.
2542 /* Maintained for backwards-compatibility only. Use newSV() instead. */
2544 #define NEWSV(x,len) newSV(len)
2547 #define MEM_SIZE_MAX ((MEM_SIZE)-1)
2549 #define _PERL_STRLEN_ROUNDUP_UNCHECKED(n) (((n) - 1 + PERL_STRLEN_ROUNDUP_QUANTUM) & ~((MEM_SIZE)PERL_STRLEN_ROUNDUP_QUANTUM - 1))
2551 #ifdef PERL_MALLOC_WRAP
2553 /* This expression will be constant-folded at compile time. It checks
2554 * whether or not the type of the count n is so small (e.g. U8 or U16, or
2555 * U32 on 64-bit systems) that there's no way a wrap-around could occur.
2556 * As well as avoiding the need for a run-time check in some cases, it's
2557 * designed to avoid compiler warnings like:
2558 * comparison is always false due to limited range of data type
2559 * It's mathematically equivalent to
2560 * max(n) * sizeof(t) > MEM_SIZE_MAX
2563 # define _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) \
2564 ( sizeof(MEM_SIZE) < sizeof(n) \
2565 || sizeof(t) > ((MEM_SIZE)1 << 8*(sizeof(MEM_SIZE) - sizeof(n))))
2567 /* This is written in a slightly odd way to avoid various spurious
2568 * compiler warnings. We *want* to write the expression as
2569 * _MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) && (n > C)
2570 * (for some compile-time constant C), but even when the LHS
2571 * constant-folds to false at compile-time, g++ insists on emitting
2572 * warnings about the RHS (e.g. "comparison is always false"), so instead
2575 * (cond ? n : X) > C
2577 * where X is a constant with X > C always false. Choosing a value for X
2578 * is tricky. If 0, some compilers will complain about 0 > C always being
2579 * false; if 1, Coverity complains when n happens to be the constant value
2580 * '1', that cond ? 1 : 1 has the same value on both branches; so use C
2581 * for X and hope that nothing else whines.
2584 # define _MEM_WRAP_WILL_WRAP(n,t) \
2585 ((_MEM_WRAP_NEEDS_RUNTIME_CHECK(n,t) ? (MEM_SIZE)(n) : \
2586 MEM_SIZE_MAX/sizeof(t)) > MEM_SIZE_MAX/sizeof(t))
2588 # define MEM_WRAP_CHECK(n,t) \
2589 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2590 && (croak_memory_wrap(),0))
2592 # define MEM_WRAP_CHECK_1(n,t,a) \
2593 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2594 && (Perl_croak_nocontext("%s",(a)),0))
2596 /* "a" arg must be a string literal */
2597 # define MEM_WRAP_CHECK_s(n,t,a) \
2598 (void)(UNLIKELY(_MEM_WRAP_WILL_WRAP(n,t)) \
2599 && (Perl_croak_nocontext("" a ""),0))
2601 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
2603 #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))
2606 #define MEM_WRAP_CHECK(n,t)
2607 #define MEM_WRAP_CHECK_1(n,t,a)
2608 #define MEM_WRAP_CHECK_s(n,t,a)
2609 #define MEM_WRAP_CHECK_(n,t)
2611 #define PERL_STRLEN_ROUNDUP(n) _PERL_STRLEN_ROUNDUP_UNCHECKED(n)
2617 * If PERL_MEM_LOG is defined, all Newx()s, Renew()s, and Safefree()s
2618 * go through functions, which are handy for debugging breakpoints, but
2619 * which more importantly get the immediate calling environment (file and
2620 * line number, and C function name if available) passed in. This info can
2621 * then be used for logging the calls, for which one gets a sample
2622 * implementation unless -DPERL_MEM_LOG_NOIMPL is also defined.
2625 * - not all memory allocs get logged, only those
2626 * that go through Newx() and derivatives (while all
2627 * Safefrees do get logged)
2628 * - __FILE__ and __LINE__ do not work everywhere
2629 * - __func__ or __FUNCTION__ even less so
2630 * - I think more goes on after the perlio frees but
2631 * the thing is that STDERR gets closed (as do all
2632 * the file descriptors)
2633 * - no deeper calling stack than the caller of the Newx()
2634 * or the kind, but do I look like a C reflection/introspection
2636 * - the function prototypes for the logging functions
2637 * probably should maybe be somewhere else than handy.h
2638 * - one could consider inlining (macrofying) the logging
2639 * for speed, but I am too lazy
2640 * - one could imagine recording the allocations in a hash,
2641 * (keyed by the allocation address?), and maintain that
2642 * through reallocs and frees, but how to do that without
2643 * any News() happening...?
2644 * - lots of -Ddefines to get useful/controllable output
2645 * - lots of ENV reads
2649 # ifndef PERL_MEM_LOG_NOIMPL
2658 # if defined(PERL_IN_SV_C) /* those are only used in sv.c */
2659 void Perl_mem_log_new_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2660 void Perl_mem_log_del_sv(const SV *sv, const char *filename, const int linenumber, const char *funcname);
2667 #define MEM_LOG_ALLOC(n,t,a) Perl_mem_log_alloc(n,sizeof(t),STRINGIFY(t),a,__FILE__,__LINE__,FUNCTION__)
2668 #define MEM_LOG_REALLOC(n,t,v,a) Perl_mem_log_realloc(n,sizeof(t),STRINGIFY(t),v,a,__FILE__,__LINE__,FUNCTION__)
2669 #define MEM_LOG_FREE(a) Perl_mem_log_free(a,__FILE__,__LINE__,FUNCTION__)
2672 #ifndef MEM_LOG_ALLOC
2673 #define MEM_LOG_ALLOC(n,t,a) (a)
2675 #ifndef MEM_LOG_REALLOC
2676 #define MEM_LOG_REALLOC(n,t,v,a) (a)
2678 #ifndef MEM_LOG_FREE
2679 #define MEM_LOG_FREE(a) (a)
2682 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2683 #define Newxc(v,n,t,c) (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_ALLOC(n,t,safemalloc((MEM_SIZE)((n)*sizeof(t))))))
2684 #define Newxz(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_ALLOC(n,t,safecalloc((n),sizeof(t)))))
2687 /* pre 5.9.x compatibility */
2688 #define New(x,v,n,t) Newx(v,n,t)
2689 #define Newc(x,v,n,t,c) Newxc(v,n,t,c)
2690 #define Newz(x,v,n,t) Newxz(v,n,t)
2693 #define Renew(v,n,t) \
2694 (v = (MEM_WRAP_CHECK_(n,t) (t*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2695 #define Renewc(v,n,t,c) \
2696 (v = (MEM_WRAP_CHECK_(n,t) (c*)MEM_LOG_REALLOC(n,t,v,saferealloc((Malloc_t)(v),(MEM_SIZE)((n)*sizeof(t))))))
2699 #define Safefree(d) \
2700 ((d) ? (void)(safefree(MEM_LOG_FREE((Malloc_t)(d))), Poison(&(d), 1, Malloc_t)) : (void) 0)
2702 #define Safefree(d) safefree(MEM_LOG_FREE((Malloc_t)(d)))
2705 /* assert that a valid ptr has been supplied - use this instead of assert(ptr) *
2706 * as it handles cases like constant string arguments without throwing warnings *
2707 * the cast is required, as is the inequality check, to avoid warnings */
2708 #define perl_assert_ptr(p) assert( ((void*)(p)) != 0 )
2711 #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)))
2712 #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)))
2713 #define Zero(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), (void)memzero((char*)(d), (n) * sizeof(t)))
2715 /* Like above, but returns a pointer to 'd' */
2716 #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)))
2717 #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)))
2718 #define ZeroD(d,n,t) (MEM_WRAP_CHECK_(n,t) perl_assert_ptr(d), memzero((char*)(d), (n) * sizeof(t)))
2720 #define PoisonWith(d,n,t,b) (MEM_WRAP_CHECK_(n,t) (void)memset((char*)(d), (U8)(b), (n) * sizeof(t)))
2721 #define PoisonNew(d,n,t) PoisonWith(d,n,t,0xAB)
2722 #define PoisonFree(d,n,t) PoisonWith(d,n,t,0xEF)
2723 #define Poison(d,n,t) PoisonFree(d,n,t)
2726 # define PERL_POISON_EXPR(x) x
2728 # define PERL_POISON_EXPR(x)
2731 #define StructCopy(s,d,t) (*((t*)(d)) = *((t*)(s)))
2736 =for apidoc Am|STRLEN|C_ARRAY_LENGTH|void *a
2738 Returns the number of elements in the input C array (so you want your
2739 zero-based indices to be less than but not equal to).
2741 =for apidoc Am|void *|C_ARRAY_END|void *a
2743 Returns a pointer to one element past the final element of the input C array.
2747 C_ARRAY_END is one past the last: half-open/half-closed range, not
2748 last-inclusive range.
2750 #define C_ARRAY_LENGTH(a) (sizeof(a)/sizeof((a)[0]))
2751 #define C_ARRAY_END(a) ((a) + C_ARRAY_LENGTH(a))
2755 # define Perl_va_copy(s, d) va_copy(d, s)
2756 # elif defined(__va_copy)
2757 # define Perl_va_copy(s, d) __va_copy(d, s)
2759 # define Perl_va_copy(s, d) Copy(s, d, 1, va_list)
2763 /* convenience debug macros */
2765 #define pTHX_FORMAT "Perl interpreter: 0x%p"
2766 #define pTHX__FORMAT ", Perl interpreter: 0x%p"
2767 #define pTHX_VALUE_ (void *)my_perl,
2768 #define pTHX_VALUE (void *)my_perl
2769 #define pTHX__VALUE_ ,(void *)my_perl,
2770 #define pTHX__VALUE ,(void *)my_perl
2773 #define pTHX__FORMAT
2776 #define pTHX__VALUE_
2778 #endif /* USE_ITHREADS */
2780 /* Perl_deprecate was not part of the public API, and did not have a deprecate()
2781 shortcut macro defined without -DPERL_CORE. Neither codesearch.google.com nor
2782 CPAN::Unpack show any users outside the core. */
2784 # define deprecate(s) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2785 "Use of " s " is deprecated")
2786 # define deprecate_disappears_in(when,message) \
2787 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2788 message ", and will disappear in Perl " when)
2789 # define deprecate_fatal_in(when,message) \
2790 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
2791 message ". Its use will be fatal in Perl " when)
2794 /* Internal macros to deal with gids and uids */
2797 # if Uid_t_size > IVSIZE
2798 # define sv_setuid(sv, uid) sv_setnv((sv), (NV)(uid))
2799 # define SvUID(sv) SvNV(sv)
2800 # elif Uid_t_sign <= 0
2801 # define sv_setuid(sv, uid) sv_setiv((sv), (IV)(uid))
2802 # define SvUID(sv) SvIV(sv)
2804 # define sv_setuid(sv, uid) sv_setuv((sv), (UV)(uid))
2805 # define SvUID(sv) SvUV(sv)
2806 # endif /* Uid_t_size */
2808 # if Gid_t_size > IVSIZE
2809 # define sv_setgid(sv, gid) sv_setnv((sv), (NV)(gid))
2810 # define SvGID(sv) SvNV(sv)
2811 # elif Gid_t_sign <= 0
2812 # define sv_setgid(sv, gid) sv_setiv((sv), (IV)(gid))
2813 # define SvGID(sv) SvIV(sv)
2815 # define sv_setgid(sv, gid) sv_setuv((sv), (UV)(gid))
2816 # define SvGID(sv) SvUV(sv)
2817 # endif /* Gid_t_size */
2821 #endif /* PERL_HANDY_H_ */
2824 * ex: set ts=8 sts=4 sw=4 et: